High Frequency Oscillatory Ventilation - A Decade of Progress

Preface

High-frequency oscillatory ventilation for adult acute respiratorydistress syndrome: A decade of progress

Stephen Derdak, DO

It has been a decade since the lastCritical Care Medicine Supplement onhigh-frequency oscillatory ventilation(HFOV) was published (1). That issue rep-resented a primarily European perspec-tive on the status of basic and clinicalHFOV research at that time. The empha-sis of the 1994 Supplement was on theo-retical concepts of HFOV mechanics andresearch involving small animal modelsand human neonatal clinical applica-tions. Indeed, it was stated that “...theefficiency of high-frequency oscillatoryventilation for large animals and adulthumans has not yet been established” (2).In part, this limitation of using HFOV inlarger animal models and adult humanswas thought to be related to technicaldeficiencies of the available devices (3, 4).In the decade that followed, importantadvances in understanding the heteroge-neity of acute respiratory distress syn-drome (ARDS) (e.g., using computed to-mography), ventilator-induced lunginjury, and the potential effects of venti-lator-induced lung injury in contributingto multiple-organ dysfunction syndromeled to the development of “lung-protec-

tive” conventional ventilation strategiesemphasizing use of reduced tidal volumesand inspiratory plateau pressures. Thepublication of the ARDSNet study, dem-onstrating reduced mortality using a lowtidal volume, low inspiratory plateaupressure protocol, has set a new standardof care for patients with ARDS (5). Animportant ancillary finding in this studywas that better oxygenation does not al-ways equal a better lung or better sur-vival. Indeed, survivors in the low tidalvolume group sometimes had worse ini-tial oxygenation responses than nonsur-vivors in the high tidal volume arm. Theconcept that better short-term oxygen-ation does not necessarily lead to bettersurvival resurfaced in clinical trials ofprone positioning and inhaled nitric ox-ide, in which early improvements in ox-ygenation did not translate to mortalitybenefits (6, 7).

In addition to lung-protective me-chanical ventilation strategies for pa-tients with ARDS, progress has beenmade in the integration of fluid and he-modynamic management, sedation/anal-gesia/neuromuscular blocker use, phar-macologic adjuncts (e.g., recombinantactivated protein C for septic ARDS), andventilation adjuncts (e.g., prone position-ing, lung recruitment maneuvers).Throughout the past decade, basic andclinical research on the use of alternativemethods of mechanical ventilation for se-vere adult ARDS was ongoing. HFOV, air-way-pressure release ventilation, andhigh-frequency percussive ventilationcontinued to be investigated and used byclinicians, particularly when patientswere thought to be failing conventionalvolume-cycled ventilation (8 –12). De-

spite the increasing utilization of lung-protective conventional low tidal volumestrategies, approximately one third of pa-tients with acute respiratory distress syn-drome still die of or with the disease. Itremains a frustrating disease to treat,particularly in the patient who presentsprimarily with severe oxygenation failurewithout immediately life-threateningmultiple organ failure.

A recent perspective on mechanicalventilation for ARDS has outlined an ap-proach using conventional ventilation(integrated with prone positioning andlung recruitment maneuvers) through-out all phases of the disease (13). I wouldsuggest an alternative approach. Giventhe heterogeneity of what we define asadult ARDS—both between patients andwithin the same patient over time (e.g.,as the disease progresses through exuda-tive, organizational, and fibrocystic stag-es)—a single method of mechanical ven-tilation may not be optimal for everypatient or throughout every phase of anindividual patient’s disease (Fig. 1). Ap-proaches integrating “open-lung” ventila-tor strategies (e.g., HFOV, airway-pres-sure release ventilation) with lung-protective CV strategies should beinvestigated in carefully designed (andpowered) randomized, control trials.

The focus on this Supplement is toreview progress made in the use of HFOVfor adult ARDS. Evolution of clinicalHFOV use in adults over the last decadehas included: earlier intervention (ratherthan last resort use), setting endotrachealtube cuff leaks to facilitate PaCO2 elimi-nation, use of higher frequencies (Hz)and lower oscillatory pressure amplitudes(�P) to facilitate lung protection, allow-

Key Words: high-frequency oscillatory ventilation;acute respiratory distress syndrome; airway-pressurerelease ventilation

From Pulmonary/Critical Care Medicine, WilfordHall Medical Center, San Antonio, TX.

Supported, in part, by SensorMedics Corporation,which provided use of 3100B high-frequency oscilla-tion ventilators for clinical research.

Views expressed in this article are those of the authorand do not represent the official policy of the Departmentof Defense or other departments of the U.S. government.

Copyright © 2005 by the Society of Critical CareMedicine and Lippincott Williams & Wilkins

DOI: 10.1097/01.CCM.0000155787.26548.4C

Extreme remedies are very appropriate for extreme diseases.—Hippocrates, 460–400 BC

Dum spiro spero. [While I breathe, I hope.]—South Carolina state motto

S113Crit Care Med 2005 Vol. 33, No. 3 (Suppl.)

ance of shallow spontaneous breathing(not all patients require paralysis), inte-gration of lung recruitment maneuvers(with the oscillator piston turned off),and combination of “rescue” HFOV withother adjuncts (e.g., prone positioning,inhaled nitric oxide). Additionally, thisSupplement includes reviews on airway-pressure release ventilation and high-frequency percussive ventilation writtenby clinicians who have extensive experi-ence using these modalities. Ventilatorsoffering these modes are widely availableand have an increasing body of basic andclinical literature and increasing propo-nents.

There remain significant challenges indeveloping optimal mechanical ventila-tion strategies (or sequences) for severeARDS. One major issue is defining whatis meant by “severe” ARDS itself and howto identify patients who might be failing agiven approach. When and how duringthe course of ARDS should we declare apatient as failing a given treatment ap-proach? Oncologists can measure when atumor is growing, by computed tomog-raphy, and use this information as a clearindicator that alternative treatment isneeded. In contrast, intensivists stillstruggle with defining treatment failurein patients with ARDS. Dr. Ware has ad-dressed this difficult issue in the Supple-ment, which has major implications inthe selection of patients for future clini-cal trials—whether of a rescue nature orfor early intervention. Dr. Fessler hasoutlined the role of written ventilator al-gorithms (vs. protocols) to optimize clin-

ical management of patients with ARDS.Future protocols for management ofARDS will need to integrate evidence-based fluid and hemodynamic manage-ment, sedation/analgesia/neuromuscularusage, and weaning approaches. Drs.Froese and Kinsella have provided an ex-cellent perspective on the historical evo-lution of HFOV for neonatal and pediatricapplications, along with lessons learned.Clinicians who care for adults can learnmuch from these experiences.

A recent evidence-based, expert reviewsummarized the following strategies aspotential rescue therapies for patientswith severe ARDS who have failed tradi-tional lung-protective approaches: HFOV,airway-pressure release ventilation,prone positioning, and inhaled nitric ox-ide (14). For all these therapies, definitiverecommendations could not be made dueto the lack of adequately designed (orpowered) randomized, controlled trialsshowing mortality benefits. Nevertheless,intensivists who care for critically ill pa-tients with severe ARDS must make treat-ment decisions based on incomplete data.

It is my hope that this Supplementwill provide useful information for clini-cians currently using HFOV (and airway-pressure release ventilation and high-frequency percussive ventilation) asrescue therapy and will stimulate furtherresearch to define its optimal usage androle in treating patients with ARDS. Allthe contributors to this Supplement areto be thanked for sharing their expertise.I would also like to acknowledge the sup-port of the Critical Care Medicine Supple-

ment Series Editor, J. ChristopherFarmer, MD, and the editorial assistanceof Ms. Lynn Retford and Ms. ElizabethNewman.

REFERENCES

1. High-frequency ventilation: Reappraisal andprogress. Crit Care Med 1994; 22(Suppl):S19–S87

2. Lunkenheimer PP, Redmann K, Stroh N, etal: High-frequency oscillation in an adultporcine model. Crit Care Med 1994;22(Suppl):S37–S48

3. Bryan CA: The oscillations of HFO. Am JRespir Crit Care Med 2001; 163:816–817

4. Froese AB: The incremental application oflung-protective high-frequency oscillatoryventilation. Am J Respir Crit Care Med 2002;166:786–787

5. The Acute Respiratory Distress SyndromeNetwork: Ventilation with lower tidal vol-umes as compared with traditional tidal vol-umes for acute lung injury and the acuterespiratory distress syndrome. N Engl J Med2000; 342:1301–1308

6. Gattinoni L, Pesenti A, Taccone P, et al: Ef-fect of prone positioning on the survival ofpatients with acute respiratory failure.N Engl J Med 2001; 345:568–573

7. Taylor RW, Zimmerman JL, Dellinger RP, etal: Low-dose inhaled nitric oxide in patientswith acute lung injury: A randomized con-trolled trial. JAMA 2004; 291:1603–1609

8. Derdak S, Mehta S, Stewart TE, et al: High-frequency oscillatory ventilation for acute re-spiratory distress syndrome in adults: A ran-domized, controlled trial. Am J Respir CritCare Med 2002; 166:801–808

9. Mehta S, Granton J, MacDonald RJ, et al:High-frequency oscillatory ventilation inadults: The Toronto experience. Chest 2004;126:518–527

10. Putensen C, Zech S, Wrigge H, et al: Long-term effects of spontaneous breathing duringventilatory support in patients with acutelung injury. Am J Respir Crit Care Med 2001;164:43–49

11. Varpula T, Jousela I, Niemi R, et al: Com-bined effects of prone positioning and airwaypressure release ventilation on gas exchangein patients with acute lung injury. Acta An-aesthesiol Scand 2003; 47:516–524

12. Velmahos GC, Chan LS, Tatevossian R, et al:High-frequency percussive ventilation im-proves oxygenation in patients with ARDS.Chest 1999; 116:440–446

13. Marini JJ, Gattinoni L: Ventilatory manage-ment of acute respiratory distress syndrome:A consensus of two. Crit Care Med 2004;32:250–255

14. Sevransky JE, Levy MM, Marini JJ: Mechan-ical ventilation in sepsis-induced acute lunginjury/acute respiratory distress syndrome:An evidence-based review. Crit Care Med2004; 32(Suppl):S548–S553

Figure 1. Patients are initially treated with low tidal volume lung-protective conventional ventilation(CV). Some improve with CV and progress to a spontaneous breathing trial (SBT) and extubation.Alternatively, worsening occurs, prompting increased mean airway pressure (mPaw) with CV (e.g., bypositive end-expiratory pressure, or lung recruiting maneuvers [LRMs], or both) and prone position-ing. Patients failing CV may be considered for an early trial of “open-lung” ventilation with high-frequency oscillatory ventilation (HFOV); responders are gradually transitioned back to CV or airway-pressure release ventilation (APRV) for weaning.

S114 Crit Care Med 2005 Vol. 33, No. 3 (Suppl.)

Scientific Review

High-frequency oscillatory ventilation: Lessons from theneonatal/pediatric experience

Alison B. Froese, MD; John P. Kinsella, MD

Our initial reaction in 1977 inToronto to the observationthat high-frequency, small-volume flow oscillations

could “shake” enough CO2 out of thelung to effect adequate CO2 eliminationwas amazement (1). Basic observationalstudies of efficacy (animals to humans)and safety followed rapidly. Some detoursoccurred such as the declaration of 15 Hzas an optimal frequency for high-fre-quency oscillatory ventilation (HFOV),only to discover later it was an artifact ofour particular device. It quickly becameclear that HFOV eliminated CO2 ex-tremely effectively both from normal andabnormal lungs using stroke volumessmaller than the dead space (2, 3). Anexpert in fluid mechanics was recruitedto unravel the mechanisms of gas trans-port (4) while we turned our attention tothe oxygenation problems presentingclinical challenges in the late 1970s. Re-sistant hypoxemia presented problems inthe atelectasis-prone lung of both neo-nates and adults. The 1974 data of

Taghizadeh and Reynolds (5) on the evo-lution of bronchopulmonary dysplasiaduring the treatment of neonatal respira-tory distress syndrome (RDS) had incrim-inated high inflation pressures more thanhigh inspired oxygen fractions. There-fore, the hunt was on for “gentler” formsof ventilatory support in the atelectasis-prone lung, including the use of extracor-poreal membrane oxygenation. Expecta-tions were that the mysteriousmechanisms of gas transport duringHFOV would be the key to gentler venti-lation.

Evolution of the “Open Lung”Concept

In this milieu, early studies were car-ried out in neonates with severe RDS,comparing gas exchange during HFOV tothat achieved with standard ventilatorytherapy (3). Rapidly, it became clear thatthe mean airway pressure (mPaw) appliedduring HFOV had the most powerful in-fluence on oxygenation, rather than finepoints of tidal volume or frequency (Fig.1). In any given baby, one could ventilatewith a relatively low mPaw and high FIO2,or a higher mPaw and low FIO2, with thewhole range of mPaw options being welltolerated hemodynamically. One had tochoose. Further experiments followed.We learned that brief, sustained increasesin mPaw (termed a sustained inflation orrecruitment maneuver) could producerapid, large increases in PaO2 in lungsexhibiting some hysteresis in their pres-sure/volume relationships (6) and thatoscillatory impulses reexpanded atelec-

tatic lungs better than a static pressure ofthe same mean value (7). Consideringthat healthy humans had well-aeratedlungs, and airway closure and atelectasisappeared deleterious, we chose to givealveolar reexpansion top priority. Ourgoal was to use whatever mean pressureswere necessary to achieve and maintainalveolar aeration recognizing that thesmall volume cycles generated at fre-quencies of 10 to 15 Hz provided thenecessary margin of safety to avoid over-distension of more normal areas of lung.

Concurrently high-frequency jet ven-tilation was proving lifesaving in adultand neonatal patients with life-threaten-ing complications such as bronchopleu-ral fistulas or severe pulmonary intersti-tial emphysema (8, 9). Gentler ventilationin this setting meant using low peak andmean ventilator pressures as well as smalltidal volumes. Serious and prolongedconfusion emerged in the application ofhigh-frequency ventilatory techniquesbecause of failure to understand the dif-ference between these two very differentventilatory strategies (low pressure vs.optimized lung volume) and the very dif-ferent pathophysiological processes (es-tablished air leak vs. diffuse atelectasis)they were addressing. To this day, newusers may conclude that HFOV “is notworking” when in fact it would work verywell if the appropriate protocol were be-ing used for the pathophysiological ab-normality of that particular patient.

The next step in neonatology was apreliminary trial in the early 1980s ofHFOV as the primary mode of ventilatormanagement in infants diagnosed as hav-

From the Departments of Anesthesiology, Physiol-ogy, and Pediatrics (ABF), Queen’s University, King-ston, Ontario, Canada; and the Department of Pediat-rics (JPK), Children’s Hospital/University of ColoradoSchool of Medicine, Denver, CO.

Dr. Kinsella is supported in part by NIHUO1HL064857, NIH SCOR P50HL057144, and NIHGCRC MO1-RR00069.

Address requests for reprints to: Alison B. Froese,MD, Department of Anesthesiology, Kingston GeneralHospital, Kingston, ON, Canada, K7L 2V7.


DOI: 10.1097/01.CCM.0000155923.97849.6D

Efforts to minimize ventilator-induced lung injury in adultswith hypoxemic respiratory failure have recently focused on thepotential role of high-frequency oscillatory ventilation (HFOV).However, HFOV has been studied in newborns with hypoxemicrespiratory failure for nearly 3 decades. In this brief review, weattempt to summarize key physiological principles learned from

this cumulative neonatal/pediatric experience with HFOV. (CritCare Med 2005; 33[Suppl.]:S115–S121)

KEY WORDS: mechanical ventilation; high-frequency ventilation;bronchopulmonary dysplasia; inhaled nitric oxide; ventilation-induced lung injury; lung recruitment


ing RDS (10). The HFOV protocol usedlung recruitment maneuvers to reverseatelectasis and rapidly sought a mPawthat would keep those alveoli open. In-fants were randomized to HFOV or to aconventional mechanical ventilation pro-tocol (CMV) considered optimal at thattime. It rapidly became clear that HFOVimproved oxygenation faster than con-ventional ventilation without an apparentincrease in complications. The maximalmean airway pressures used during ven-tilator management proved the samewith both ventilators. However, a cleartactical difference became obvious. Dur-ing CMV, ventilator pressures were in-creased only when driven by deteriora-tion in oxygenation or CO2 elimination,such that maximum pressure values oc-curred on the second or third day. Atinitiation of HFOV, alveolar reexpansionwas given priority such that the maxi-mum mean pressures were reachedwithin 5.2 � 2.5 hrs and an FIO2 of �0.4was achieved by 18.9 hrs. Concurrently,premature baboon trials of HFOV foundprofound differences in outcome at 24hrs, but only if aggressive reexpansion ofatelectasis using increased mPaw was ini-tiated immediately postdelivery (11).

At this point, in June 1984, it was clearthat opening up the lung was feasible,safe, and effective using HFOV with anopen lung strategy in both the premature(baboon and baby) and adult (rabbit)lung. The question remained: was it nec-essary? The trial in babies had changedboth the ventilator and the priority givento the reversal of atelectasis. That ques-tion could only be answered ethically inanimals. A randomized trial of ventilator

strategies in a saline-lavaged rabbitmodel of an atelectasis-prone lung fol-lowed (12). HFOV and CMV were com-pared both using an open lung strategyand while allowing ongoing atelectasis.The results were clear. Using small tidalvolumes at high frequencies was notenough to minimize the progression oflung injury fully. Early optimization ofend-expiratory lung volume was essentialfor optimal results (13). Equivalent lungvolume optimization could not beachieved safely using the conventionalventilator patterns in use at that time.This lung volume-optimizing approachwas subsequently brought to the atten-tion of adult intensivists by Lachmann inan editorial entitled “Open Up the Lungand Keep the Lung Open!” (14).

Machine Versus Mindset

Then came the National Institutes ofHealth-sponsored trial of HFOV in neo-natal RDS, known as the HIFI Trial (15).At the time, some investigators consid-ered this clinical trial imprudent, becausestrategies for optimal HFOV applicationwere not clearly understood for this pop-ulation. Indeed, the potential of a devicethat might allow lung recruitment whileminimizing the adverse effects of cyclicvolutrauma was poorly understood. How-ever, the organizers wanted to subjectthis new ventilatory technique to therigor of a randomized, controlled pro-spective trial before widespread use en-gendered a milieu in which it was felt“unethical” to conduct such a study. Un-fortunately, the majority of human use at

the time of trial design had used thelow-pressure/small tidal volume optionappropriate for “rescue” of established airleak problems. Data on the open lungapproach consisted of 11 babies in King-ston, some premature baboons in Texas,and an assortment of adult dogs and rab-bits. Not surprisingly, the final HIFI pro-tocol tested the low-volume/low-pressurestrategy of HFOV and found no pulmo-nary benefit. In fact, HFOV appeared to bepotentially dangerous for the fragile brainof the premature, with an increased inci-dence of intraventricular hemorrhage inthat population. The latter probably re-flected a lack of mandatory transcutane-ous CO2 monitoring during transition toHFOV, because HFOV proved more effec-tive at CO2 elimination than many newusers expected, and hypocarbia is knownto induce intraventricular hemorrhageand periventricular leukomalacia in thepremature irrespective of ventilator mo-dality (16).

The primary lesson to be learned fromthe HIFI Trial is that it is easier to bringa new machine into an intensive care unitthan it is to change the mindset of thepeople using it. If a “new” device is usedwith an “old” mindset, the outcome maywell be worse.

Recovery occurred slowly over theearly 1990s. Gradually, HFOV regained aplace through cautious pursuit of ba-boon-style open lung HFOV in the neo-natal intensive care unit at Wilford Hall(17–18) plus further animal investiga-tions into the mechanisms of ventilator-induced lung injury (VILI). Evidence ac-cumulated that early optimization oflung volume in the atelectasis-prone lungprolonged the therapeutic efficacy of ex-ogenous surfactant (19), minimized theaccumulation and activation of neutro-phils in the lung (20, 21), decreased cy-tokine release in the lung and circulation(21), decreased lung water (22), and ingeneral was “lung-protective” (23, 24).HFOV moved back into neonatal and pe-diatric intensive care units, albeit withfear of discrete volume recruitment ma-neuvers in case intraventricular hemor-rhage was to recur. Randomized, con-trolled trials resumed using an optimizedlung volume strategy that aimed for earlyreversal of atelectasis by progressively in-creasing mean pressure, lowered FIO2 be-fore mPaw, and introduced techniquessuch as closed suction systems to mini-mize the loss of lung volume throughperiodic disconnects of the ventilator cir-cuit (25). HFOV became viewed as a safer

Figure 1. Plot of the response of oxygenation (as reflected by the arterial to alveolar oxygen tensionratio [a/A]) to changes in mean airway pressure in a neonate with respiratory distress syndromeventilated with high-frequency oscillatory ventilation at a constant frequency over a range of meanpressures. No circulatory instability was evident over this entire range of mean pressure. Oxygenationwas strongly influenced by the mean distending pressure applied to the lungs. (Reproduced withpermission [3].)


means of splinting the lung open by me-chanical application of an appropriatedistending pressure until the lung couldrecover from its disease process(es)enough to achieve intrinsic alveolar sta-bility. Various ventilator strategies wereexplored with the Life-Pulse HFJV aswell, providing further evidence that itwas lung volume optimization that min-imized VILI in the atelectasis-prone lung,no matter which specific high-frequencydevice one used to achieve it (26). Thelesson learned here was that strategytrumps device.

The Changing TherapeuticMilieu

HFOV reemerged during a period ofsubstantial therapeutic advances in neo-natology. Exogenous surfactant adminis-tration became a standard of care. Mater-nal steroid administration hastened lungmaturation when premature labor threat-ened. Early nasal continuous positive air-way pressure plus exogenous surfactantwas enough to stabilize alveoli in manyinfants. With each therapeutic advance,the increment of lung protection addedby HFOV lessened when prospective trialsrandomized “all comers” of a given birth-weight or gestational age. The substantialpulmonary benefits of the Provo trial pub-lished in 1996 (25) became negligible inlater trials such as Thome et al. publishedin 1999 (27). Interestingly, all of the ther-apeutic advances that have improved thepulmonary status of the premature infanthad one feature in common with an opti-mized-volume HFOV strategy. All havecontributed to early achievement of a ho-mogeneously aerated lung.

Impact on ConventionalVentilator Patterns

Throughout this period, perhaps themost profound impact of HFOV was itseffect on conventional ventilator patternsof practice. McCulloch’s study (12)proved decisively that VILI was a realityand small tidal volumes in themselveswere not sufficiently lung-protective.Soon the race was on to see whetherconventional ventilator protocols couldbe modified to achieve the same degree oflung protection. Conventional ventilatortidal volumes became smaller, positiveend-expiratory pressure levels higher,and spontaneous breath synchronizationsteadily improved (28). By the time thelast large neonatal multicenter compari-

son was performed between HFOV andthe “best” CMV protocol in current use,outcome differences were significant sta-tistically but substantially smaller than incomparative trials 10 yrs earlier (29).Nevertheless, a reduction in the numberof expensive ventilator days has beenenough to push some neonatal intensivecare units to use HFOV routinely in thevery-low-birthweight, most vulnerablepopulation of premature babies.

In the current era that includes the useof prenatal steroid therapy to enhance lungmaturation, exogenous surfactant adminis-tration, and modified conventional ventila-tion strategies, pulmonary benefit fromHFOV is only demonstrable in lungs withmoderate to severe disease. Mild diseasecan be managed well using an appropriatelylung-protective strategy with a variety ofventilator modalities. For example, in thetwo largest recent neonatal studies, signif-icant benefit from HFOV was only seen inthe study that restricted entry to very-low-birthweight infants who met set FIO2 andmPaw requirements after surfactant ad-ministration (29). Only 40% of the infantsmeeting the birthweight criteria were ran-domized. At entry, their FIO2 requirementswere 0.60 with mPaw levels of 8 cm H2O. Inthat population, HFOV increased survivalwithout chronic lung disease and infantsmanaged with HFOV were extubated suc-cessfully 1 wk earlier than those receivinglung-protective ventilation at conventionalfrequencies. In the same issue of the NewEngland Journal of Medicine, Johnson etal. (30) found no substantial benefit and noadverse effects of HFOV compared withconventional ventilation when all prema-ture newborns within a given range of ges-tational age were randomized irrespectiveof the degree of parenchymal disease. Themedian FIO2 in these infants was only 0.40at similar mPaw levels at 2 hrs of age aftertreatment with exogenous surfactant.

Follow up of the Provo study has alsotaught us that when a reduction in pul-monary morbidity is achieved in the new-born period, it results in less pulmonarydysfunction 6 yrs later (31). It therefore iscritical that we learn to identify the pop-ulations in which the smaller pressureand volume swings of HFOV will conferbenefit. At present, we conclude that rou-tine use of HFOV is not indicated for allpremature newborns with respiratory fail-ure, but should be considered for subsets ofinfants with moderate to severe disease.

Another important story in the man-agement of neonatal lung disease hasbeen the interplay between HFOV and the

use of inhaled nitric oxide (iNO). Theterm newborn with hypoxemic respira-tory failure often suffers from persistentpulmonary hypertension with critical hy-poxemia associated with parenchymallung disease as well as intense pulmonaryvasoconstriction causing extrapulmonaryright-to-left shunting through persistentfetal channels at the foramen ovale andthe ductus arteriosus (32). Clark et al.(33) showed that HFOV was more effec-tive than conventional ventilation whenused as a rescue therapy in the termnewborn with hypoxemic respiratory fail-ure. However, in the setting of persistentpulmonary hypertension (PPHN), opti-mizing lung recruitment alone is not al-ways sufficient to improve oxygenation.

The advent of therapy with inhalednitric oxide provided a unique opportu-nity to study the relative roles of optimiz-ing lung recruitment and selectively di-lating the pulmonary circulation inpatients with the complex disorder ofPPHN (34, 35). Indeed, these early stud-ies demonstrated that the effects of in-haled NO can be suboptimal when lungvolume is decreased in association withpulmonary parenchymal disease for sev-eral reasons. Atelectasis and airspace dis-ease (pneumonia, pulmonary edema) de-crease the delivery of iNO to its site ofaction in terminal lung units. Both un-derinflation and overdistension of lungtissue have adverse mechanical effects onpulmonary vascular resistance. It istherefore not surprising that ventilatorstrategy might impact iNO effectivenesssignificantly (Fig. 2).

A randomized, multicenter trial (36)demonstrated that treatment with HFOV� iNO was often successful in patientswho failed to respond to HFOV or iNOalone in severe PPHN, with differences inresponses being related to the specificdisease associated with the complex dis-orders of PPHN. For patients with PPHNcomplicated by severe lung disease, re-sponse rates for HFOV � iNO were betterthan HFOV alone or iNO with conven-tional ventilation. In contrast, for pa-tients without significant parenchymallung disease, both iNO and HFOV � iNOwere more effective than HFOV alone.This response to combined treatmentwith HFOV � iNO likely reflects both adecrease in intrapulmonary shunting us-ing a strategy to recruit and maintainaerated lung rather than just hyperventi-lating patients with severe lung diseaseand PPHN, plus augmented NO deliveryto its site of action.


The interplay between lung volumeoptimization with HFOV and pulmonaryvasodilation from iNO has steadilychanged the pattern of practice in manyneonatal units. Over the past decade,there has been a steady decline in thenumber of newborn patients treated withextracorporeal membrane oxygenation(ECMO), likely as a result of the increaseduse of surfactant, HFOV, and iNO. Con-genital diaphragmatic hernia patientsnow make up the largest proportion ofneonatal ECMO patients, with the overallsurvival rate on ECMO decreasing as thesurvivable cases of lung disease have be-come manageable with less invasive tech-niques. One can only wonder whetheriNO in adult ARDS possibly “failed” be-cause lung volume optimization was nota priority at that time? It is possible thatas adult ventilator protocols explore moreaggressive lung volume optimization, ad-junct treatments such as iNO might yielda stronger signal.

Beyond the newborn period, early, fa-vorable experience with HFOV led pedi-atric intensivists to test its role in themanagement of hypoxemic respiratoryfailure in older patients. Arnold et al. (37)randomized patients with diffuse alveolar

disease and/or air leak syndrome to HFOVor CMV, both placing a high priority onlung volume optimization. They demon-strated that HFOV used with an aggres-sive strategy to recruit and maintain“ideal” lung volume decreased the fre-quency of barotrauma in the majority ofpatients randomized to HFOV. Patientswho “failed HFOV” had an 82% mortalityrate when switched back to conventionalventilation. Survivors had lower oxygen-ation indices after 24 hrs of HFOV (26.2)than nonsurvivors (41.4). In parallel withthe neonatal experience, Dobyns et al.(38) demonstrated that HFOV augmentedthe response to iNO in pediatric patientswith hypoxemic respiratory failure.

Why Might High-FrequencyOscillatory Ventilation Be MoreLung-Protective Than CurrentAlternatives?

This question needs to be addressed inits physiological context. For 30 yrs, wehave known that we face a fundamentalproblem of interregional inhomogene-ities of end-expiratory lung volumes, ofairway opening and closing pressures,and of the distribution of a ventilator

breath, even in the normal lung. Theseregional disparities originate from verti-cal gradients of pleural pressure that cre-ate more regional inhomogeneity supinethan prone or upright and are accentu-ated by many disease processes (39). For30 yrs, we have also known that an ap-plied static or quasistatic airway pressurewill inevitably increase regional end-expiratory alveolar volume, airway pa-tency, and share of tidal volume preciselyin the lung regions least needing the help(40). These interregional issues do notshow up in the overall pressure/volumecurves from which concepts of lower andupper inflection points and “safe” plateaupressures have been derived over recentyears of VILI debate. Fundamentally,there must be a family of different pres-sure/volume curves in every diseasedlung, some with perturbations from lo-calized disease, but many simply arisingfrom intrinsic gravity-related differencesin regional chest wall properties. Overthe years of animal experiments in whichalveolar reexpansion was pursued with avariety of protocols, it became clear thatin a mildly atelectasis-prone lung, onecan achieve alveolar reexpansion with ei-ther HFOV or conventional lower-fre-quency options because the pressure andvolume cycles with either device canmaintain end-expiratory lung volumes independent regions without excessive dis-tension of nondependent lung whileeliminating CO2 adequately. As pulmo-nary dysfunction increases, at somepoint, it becomes impossible to keep de-pendent alveoli above their closing pres-sures without reaching potentially dam-aging peak or plateau pressures, evenwith permissive hypercarbia. It is simplya matter of gravity, physics, and regionalinhomogeneities. In a late 1980s rabbitstudy (19) in which we tried to achieve anopen lung with both HFOV and conven-tional ventilation after exogenous surfac-tant therapy, we learned that the startingpositive end-expiratory pressure levelsthat appeared adequate in terms of lungexpansion, as gauged by oxygenation,were clustered around the pressure atwhich derecruitment accelerated on theoverall deflation pressure/volume curveof the individual animals (Fig. 3). Clearly,the derecruitment potential in dependentzones will have been even greater thanthis overall pressure/volume picture. Notsurprisingly, progressive deteriorationoccurred with that particular CMV strat-egy, which also used larger tidal volumesthan are currently recommended. How-

Figure 2. A, Graph of PaO2 vs. age in hours in an infant with congenital diaphragmatic hernia andpersistent pulmonary hypertension. Although the infant responded initially to inhaled nitric oxide(NO) during conventional mechanical ventilation (CMV), the response was not sustained. During thefirst hour of high-frequency oscillatory ventilation (HFOV), oxygenation did not improve but lungreexpansion was evident on chest radiograph. B, At that point, reintroduction of inhaled NO produceda marked and sustained improvement in oxygenation. Observations like this demonstrate the impor-tance of achieving adequate alveolar expansion for the delivery of inhaled nitric oxide to be effective.(Reproduced with permission [32].)


ever, during HFOV, mPaw values provid-ing similar oxygenation and no hemody-namic compromise proved to be severalcm H2O above that derecruitment zone.In these experiments, the small volumecycles of HFOV maintained alveolar aer-ation more effectively with a comfortablemargin of safety at both extremes of lungvolume. As CMV patterns have movedcloser to those of HFOV, similar out-comes have become attainable with ap-proaches such as synchronized intermit-tent mandatory ventilation with earlyvolume optimization in many infants.However, in more severe disease, HFOVstill offers the intrinsic safety provided byeliminating CO2 with the smallest possi-ble excursions of volume and pressurewithin a “safe zone” of alveolar aerationthat becomes narrower as lung inhomo-geneity intensifies (41).

When Should High-FrequencyOscillatory Ventilation BeInstituted?

The current challenge in both neona-tology and pediatrics is how to determine

when the institution of HFOV may offersuperior lung protection. Laboratory stud-ies of the1980s and 1990s demonstratedthat if the lung was going to “need” HFOVto minimize VILI, then it was important toinstitute HFOV before lung injury was tooextensive (42). “Appropriate criteria” re-main variable between institutions. Indica-tors for a ventilator switch tend to targetsome combination of FIO2 requirement andthe pressure needed to maintain adequatelung expansion and CO2 elimination. In thepreterm infant, “acceptable” peak inspira-tory pressures increase with increases inbirthweight or gestational age. In term in-fants, a mPaw �10 to 12 cm H2O with anFIO2 �0.6 and/or a falling aerated lung vol-ume trigger HFOV in some institutions. Inolder pediatric patients, Doctor and Arnold(43) recommend conversion to HFOV ifattempts to pursue an “open lung” strategyat conventional rates result in a peak pres-sure �35 cm H2O (despite permissive hy-percapnia) or mPaw values approach 15 to18 cm H2O and the FIO2 exceeds 0.6. Somepediatric units convert to HFOV at an FIO2

requirement of 0.5 (Arnold, personal com-munication). Many of these recommenda-tions come from institutions with exten-sive HFOV experience where it has simplyproven easier to maintain lung recruit-ment with less barotrauma using HFOVthan with larger volume cycles at lowerrates in an intensive care unit settingwhere many clinicians contribute to ven-tilator management decisions over thecourse of treatment.

The ideal criteria for possible conver-sion to HFOV would include an accurate,sensitive measure of both atelectasis andoverdistension that could be performedin the intensive care unit while using avariety of ventilator patterns. (See thechapter on electrical impedance tomog-raphy elsewhere in this supplement.)This would help define the point in thecontinuum of care at which HFOV mightprove more lung-protective than alterna-tive lower-frequency options. Unless pro-spective trials can be randomized bysome objective measure(s) of lung re-cruitability, trial “noise” may obscurereal potential benefits.

”Defensible” Mean AirwayPressure And Hertz Settings?

One cannot really quote establishedrecommendations for maximal mPaw orHz settings from the neonatal or pediat-ric literature. One can, however, com-ment that brief exposure to mPaw levels

in the 40s have been used without hemo-dynamic compromise in neonates andmore extensively in pediatric patients toinitiate lung recruitment in the presenceof severe parenchymal lung disease. Briefperiods of oscillation at mPaw levels inthe low 50s have been used in lungs re-sistive to reexpansion (as evident by per-sistent infiltrates on radiograph), fol-lowed by a return to lower maintenancepressures in the mid 40s in pediatric pa-tients (Arnold, personal communication)using the SensorMedics 3100A (Sensor-Medics, Yorba Linda, CA). One cannotarbitrarily limit the maximum mPaw tosome sensible-sounding level because ofthe extreme variability in the mechanicalproperties of the lungs of patients pre-senting for treatment. For example, pa-tients with high intraabdominal pres-sures or restrictive chest walls may needrelatively high mPaw levels, much ofwhich will be dissipated across the chestwall, not the lung. Conversely, there is nopoint in maintaining high pressures thatare compromising hemodynamics in theabsence of demonstrable benefit to lungaeration and/or gas exchange. Settingsmust be tailored to each individual. Onemust be particularly alert to the possibil-ity that a rapid change in pulmonary me-chanics from diuresis or an effective re-cruitment maneuver may require a quickdecrease in maintenance mPaw to avoidlung overdistension and impaired venousreturn.

As a general rule, frequencies of10–15 Hz are used in neonates with dif-fuse alveolar disease. Although used ini-tially rather arbitrarily, this happens tomatch the frequency range subsequentlypredicted to provide the best alveolar ex-pansion with the least overdistension inthe model calculations of Venegas et al.(44) for an atelectasis-prone infant lung.In pediatric practice, Arnold recommendsstarting at 10 Hz with a high power set-ting, decreasing frequency only if patientsize or lung impedance necessitates it(personal communication). The rationalefor this is twofold. Because the theoreti-cal advantages of HFOV are thought toarise from its ability to support gas trans-port with small volume cycles, it makessense to use the smallest stroke volumethat achieves gas transport goals. Second,the degree of filtering of the HFOV pres-sure swing along an endotracheal tube isfrequency-dependent and tube diameter-dependent, with the greatest percent re-duction in pressure swing occurring athigher frequencies and in narrower

Figure 3. Individual and mean pressure/volumecurves obtained after the administration of exog-enous surfactant to rabbits previously made sur-factant deficient by repeated saline lavage. Themark on the deflation limb of each pressure/volume curve indicates the positive end-expira-tory pressure (PEEP) level during the first hourof conventional ventilation that achieved the pro-tocol’s oxygenation target. The goal was toachieve early optimization of lung volume byusing sufficient PEEP to prevent derecruitment.Subsequent analysis of these curves revealed thatalthough the oxygenation data suggested PEEPlevels were adequate, the end-expiratory lung vol-umes were in fact below critical closing pressuresin most animals. Light lines show the pressure/volume relationship of individual animals. Thedark line depicts the average pressure/volumerelationship. (Previously unpublished data fromthe CMV/HiVol group of animals [19].)


tubes. Therefore, it is prudent to oscillateat the highest frequency that supportsgas transport until more is known aboutthe details of pressure transmission intothe lungs.

Achieving Adequate LungRecruitment

To optimize alveolar aeration, it is es-sential that one get atelectatic alveoliopen and then find the maintenancemPaw that will keep them open. Alveolarreexpansion can be achieved either bybrief increases in mean pressure—termed sustained inflations or recruit-ment maneuvers (6, 7)—or by incremen-tal 1–2 cm H2O increases in mPaw untilsatisfactory oxygenation is achieved at anFIO2 of �0.6 or evidence of overinflationbecomes evident on chest radiograph(i.e., flattened diaphragms and/or �9ribs, posteriorly, of lung expansion in anewborn or child) (26, 45). The latterapproach to recruitment has been mostcommonly used in neonatal and pediatricpractice. Careful studies by Thome et al.(46) using SF6 washout to measure meanlung volume on HFOV with changes inmPaw have demonstrated the expectedstrong dependence of mean lung volumeon mPaw but with enough interindi-vidual variations in slope to make it im-possible to predict mean lung volumesolely from mPaw. When HFOV is firstinstituted, a dynamic, interactive processof lung recruitment must be given firstpriority, guided by oxygenation response,clinical observation (i.e., is the patientsuddenly “shaking” more at constant set-tings of power and frequency?), hemody-namic status, and some index of adequacyof lung volume such as chest radiographor the newer emerging technologies. Anoxygenation response can be detectedquickly by holding the FIO2 constant atlevels that produce an oxygen saturationaround 90% while pursuing one’s re-cruitment protocol. Lung “opening” willthen be reflected in saturation changes.Once the lung has been adequately re-cruited and the FIO2 reduced to levels of0.5 or 0.6 (the exact target used variesfrom center to center), reductions inmPaw are pursued slowly and cautiously,aiming to hold the lung above its zone ofderecruitment. If derecruitment occurs,it is reversed with the initial recruitmentapproach and mPaw returned to 2 cmH2O above the pressure at which atelec-tasis developed. The most common errorsamong new users of HFOV are 1) inade-

quate initial lung recruitment, 2) prema-ture reductions in the maintenance mPawbefore recovery of intrinsic alveolar stabil-ity, and 3) failure to decrease the mPawwhen a marked improvement in oxygen-ation occurs. A marked improvement inoxygenation often signals a change in thelung’s pressure/volume relationship, inwhich case one must decrease the mPawappropriately to avoid hyperinflation withresultant hemodynamic compromise.HFOV is basic respiratory physiology in ac-tion. Although recruitment pressures mayseem frighteningly high at times, both an-imal experiments (47) and clinical experi-ence have shown that brief periods of over-distension are less injurious to the lungthan prolonged underrecruitment duringHFOV. Alveolar recruitment will also de-crease the overall impedance of the lung,such that a larger stroke volume will bedelivered for any given power and fre-quency combination. Therefore, when fac-ing CO2 elimination problems, one’s firstresponse should be to verify that the lung isadequately reexpanded rather than just de-creasing frequency to increase stroke vol-ume or increasing power.

Hindrances to High-FrequencyOscillatory VentilationDevelopment

An unfortunate consequence of theearly neonatal HIFI Trial was an inevita-ble freeze on ventilator development.Only one supplier persisted after the neg-ative trial outcome, and it becamechained to its existing design. Even mi-nor modifications faced much morestringent and expensive U.S. Food andDrug Administration requirements thanmajor design modifications to low-frequency ventilators. This means thatHFOV machines have had to remain rel-atively large, awkward, and noisy com-pared with the progressively slimmer,more svelte conventional devices nowgracing our intensive care units. In addi-tion, potentially confusing features suchas the interplay between frequency anddelivered volume have not been correctedin this restrictive environment. Althoughsmall neonatal patients can breathe com-fortably on their HFOV circuit, paralysishas been felt necessary in most largerpatients in whom a cough or sigh cantrigger numerous alarms, with resultantloss of the benefits of maintaining someelement of spontaneous respiration dur-ing ventilator support. The latter is a sig-

nificant disadvantage that needs rectifica-tion.

So, what lessons have we learned overnearly 3 decades in neonatology?

Lessons Learned

1. A premature trial can kill a goodtechnique (almost).

2. Mindsets are harder to change thanmachines.

3. Ventilator strategy must be drivenby the patient’s pathophysiology.

4. Mild pulmonary dysfunction can bemanaged equally well with both low-and high-frequency lung-protectivestrategies.

5. In moderate to severe disease, it iseasier to “stay in the safe zone” oflung expansion using HFOV thanwhen using a lower-frequency device.

6. The definition of appropriate criteriafor transitioning to HFOV is a criticalneed. In mild disease, it is not needed.In end-stage disease, it will be useless.

7. Alveolar recruitment is fundamentalto lung protection with HFOV. Re-cruitment strategies must be tai-lored to each individual patient.Techniques with better resolutionthan a chest radiograph are neededto guide lung volume optimization.

8. Maintenance mean airway pressuresduring HFOV must be much higherthan the positive end-expiratorypressure levels commonly used dur-ing CMV to prevent derecruitment.Fortunately, this can be done safely.This means that users must becomecomfortable with mean pressuresmuch higher than their “comfortzone” with positive end-expiratorypressure for HFOV to be lung-protective.

REFERENCES

1. Bryan AC: The oscillations of high-frequencyoscillation. Am J Respir Crit Care Med 2001;163:816–817

2. Butler WJ, Bohn DJ, Bryan AC, et al: Venti-lation by high-frequency oscillation in hu-mans. Anesth Analg 1980; 59:577–584

3. Marchak BE, Thompson WK, Duffty P, et al:Treatment of RDS by high-frequency oscilla-tory ventilation: a preliminary report. J Pe-diatr 1981; 99:287–292

4. Fredberg JJ: Augmented diffusion in the air-ways can support pulmonary gas exchange.J Appl Physiol 1980; 49:232–238


5. Taghizadeh A, Reynolds EOR: Pathogenesisof bronchopulmonary dysplasia following hy-aline membrane disease. Am J Pathol 1976;82:241–264

6. Kolton M, Cattran CB, Kent G, et al: Oxygen-ation during high-frequency ventilationcompared with conventional mechanicalventilation in two models of lung injury.Anesth Analg 1982; 61:323–332

7. Byford LJ, Finkler JH, Froese AB: Lung vol-ume recruitment during high-frequency os-cillation in atelectasis-prone rabbits. J ApplPhysiol 1988; 64:1607–1614

8. Carlon GC, Kahn RC, Howland WS, et al:Clinical experience with high frequency jetventilation. Crit Care Med 1981; 9:1–6

9. Pokora T, Bing D, Mammel M, et al: Neonatalhigh-frequency jet ventilation. Pediatrics1983; 72:27–32

10. Froese AB, Butler PO, Fletcher WA, et al:High-frequency oscillatory ventilation inpremature infants with respiratory failure: apreliminary report. Anesth Analg 1987; 66:814–824

11. Meredith KS, deLemos RA, Coalson JJ, et al:Role of lung injury in the pathogenesis ofhyaline membrane disease in premature ba-boons. J Appl Physiol 1989; 66:2150–2158

12. McCulloch PR, Forkert PG, Froese AB: Lungvolume maintenance prevents lung injuryduring high frequency oscillatory ventilationin surfactant-deficient rabbits. Am Rev Re-spir Dis 1988; 137:1185–1192

13. Froese AB: Role of lung volume in lung in-jury: HFO in the atelectasis-prone lung. ActaAnaesthesiol Scand 1989; 33(suppl 90):126–130

14. Lachmann B: Open up the lung and keep thelung open. Intensive Care Med 1992; 18:319–321

15. The HIFI Study Group: High frequency os-cillatory ventilation compared with conven-tional mechanical ventilation in the treat-ment of respiratory failure in preterminfants. N Engl J Med 1989; 320:88–93

16. Greisen G, Munck H, Lou H: Severe hypocar-bia in preterm infants and neurodevelop-ment deficit. Acta Paediatr Scand 1987; 76:401–404

17. Schwendeman CA, Clark RH, Yoder BA, et al:Frequency of chronic lung disease in infantswith severe respiratory failure treated withhigh-frequency ventilation and/or extracor-poreal membrane oxygenation. Crit CareMed 1992; 20:372–377

18. Clark RH, Gerstmann DR, Null DM, et al:Prospective randomized comparison of highfrequency oscillatory and conventional ven-tilation in respiratory distress syndrome. Pe-diatrics 1992; 89:5–12

19. Froese AB, McCulloch PR, Sugiura M, et al:Optimizing alveolar expansion prolongs theeffectiveness of exogenous surfactant therapyin the adult rabbit. Am Rev Respir Dis 1993;148:569–577

20. Sugiura M, McCulloch PR, Wren S, et al:Ventilator pattern influences neutrophil in-flux and activation in atelectasis-prone rabbitlung. J Appl Physiol 1994; 77:1355–1365

21. Imai Y, Kawano T, Miyasaka K, et al: Inflam-matory chemical mediators during conven-tional ventilation and during high frequencyoscillatory ventilation. Am J Respir Crit CareMed 1994; 150:1550–1554

22. Tamura M, Kawano T, Fitz-James I, et al:High-frequency oscillatory ventilation andpulmonary extravascular water. AnesthAnalg 1985; 64:1041–1046

23. HiFO Study Group: Randomized study ofhigh-frequency oscillatory ventilation in in-fants with severe respiratory distress syn-drome. J Pediatr 1993; 122:609–619

24. Jackson JC, Truog WE, Standaert TA, et al:Reduction in lung injury after combined sur-factant and high-frequency ventilation. Am JRespir Crit Care Med 1994; 150:534–539

25. Gerstmann DR, Minton SD, Stoddard RA, etal: The Provo multicenter early high-frequency oscillatory ventilation trial: im-proved pulmonary and clinical outcome inrespiratory distress syndrome. Pediatrics1996; 98:1044–1057

26. Gerstmann DR, deLemos RA, Clark RH:High-frequency ventilation: issues of strat-egy. Clin Perinatol 1991; 18:563–580

27. Thome U, Kössel H, Lipowsky G, et al: Ran-domized comparison of high-frequency ven-tilation with high-rate intermittent positivepressure ventilation in preterm infants withrespiratory failure. J Pediatr 1999; 135:39–46

28. Bernstein G, Mannino FL, Heldt GP, et al:Randomized multicenter trial comparingsynchronized and conventional intermittentmandatory ventilation in neonates. J Pediatr1996; 128:453–463

29. Courtney SE, Durand DJ, Asselin JM, et al:High-frequency oscillatory ventilation versusconventional mechanical ventilation forvery-low-birth-weight infants. N Engl J Med2002; 347:643–652

30. Johnson AH, Peacock JL, Greenough A, et al:High-frequency oscillatory ventilation forthe prevention of chronic lung disease ofprematurity. N Engl J Med 2002; 347:633–642

31. Gerstmann DR, Wood K, Miller A, et al:Childhood outcome after early high-fre-quency oscillatory ventilation for neonatalrespiratory distress syndrome. Pediatrics2001; 108:617–623

32. Kinsella JP, Abman SH: Recent develop-ments in the pathophysiology and treatmentof persistent pulmonary hypertension of thenewborn. J Pediatr 1995; 126:853–864

33. Clark RH, Yoder BA, Sell MS: Prospective,randomized comparison of high-frequencyoscillation and conventional ventilation incandidates for extracorporeal membrane ox-ygenation. J Pediatr 1994; 124:447–454

34. Roberts JD, Polaner DM, Lang P, et al: In-haled nitric oxide in persistent pulmonaryhypertension of the newborn. Lancet 1992;340:818–819

35. Kinsella JP, Neish SR, Shaffer E, et al: Low-dose inhalational nitric oxide in persistentpulmonary hypertension of the newborn.Lancet 1992; 340:819–820

36. Kinsella JP, Truog WE, Walsh WF, et al:Randomized multicenter trial of inhaled ni-tric oxide and high frequency oscillatory ven-tilation in severe persistent pulmonary hy-pertension of the newborn. J Pediatr 1997;131:55–62

37. Arnold JH, Hanson JH, Toro-Figuero LO, etal: Prospective, randomized comparison ofhigh-frequency oscillatory ventilation andconventional mechanical ventilation in pedi-atric respiratory failure. Crit Care Med 1994;22:1530–1539

38. Dobyns EL, Anas NG, Fortenberry JD, et al:Interactive effects of high-frequency oscilla-tory ventilation and inhaled nitric oxide inacute hypoxemic respiratory failure in pedi-atrics. Crit Care Med 2002; 30:2425–2429

39. Pelosi P, D’Andrea L, Vitale G, et al: Verticalgradient of regional lung inflation in adultrespiratory distress syndrome. Am J RespirCrit Care Med 1994; 149:8–13

40. Froese AB, Bryan AC: Effects of anesthesiaand paralysis on diaphragmatic mechanics inman. Anesthesiology 1974; 41:242–255

41. Roupie E, Dambrosio M, Servillo G, et al:Titration of tidal volume and induced hyper-capnia in acute respiratory distress syn-drome. Am J Respir Crit Care Med 1995;152:121–128

42. deLemos RA, Coalson JJ, de Lemos JA, et al:Rescue ventilation with high frequency oscil-lation in premature baboons with hyalinemembrane disease. Pediatr Pulmonol 1992;12:29–36

43. Doctor A, Arnold J: Mechanical support ofacute lung injury: options for strategic ven-tilation. New Horizons 1997; 7:359–373

44. Venegas JG, Fredberg JJ: Understanding thepressure cost of ventilation: why does high-frequency ventilation work? Crit Care Med1994; 22(suppl 9):S49–S57

45. Arnold JH: High-frequency ventilation in thepediatric intensive care unit. Pediatr CritCare Med 2000; 1:93–99

46. Thome U, Topfer A, Schaller P, et al: Effectsof mean airway pressure on lung volumeduring high-frequency oscillatory ventilationof preterm infants. Am J Respir Crit CareMed 1998; 157:1213–1218

47. Bond DM, Froese AB: Volume recruitmentmaneuvers are less deleterious than persis-tent low lung volumes in the atelectasisprone rabbit during high-frequency oscilla-tion. Crit Care Med 1993; 21:402–412


Dynamic alveolar mechanics and ventilator-induced lung injury

David Carney, MD; Joseph DiRocco, MD; Gary Nieman, BA

Mechanical ventilation iscritical to survival of mostpatients with acute respira-tory distress syndrome

(ARDS) (1). Improper use of mechanicalventilation in the context of ARDS maycause a secondary, ventilator-inducedlung injury (VILI) (2). VILI has been es-timated to increase mortality in ARDSpatients by 3,900 to 35,000 patients peryear (2). A great deal of research has beendirected toward understanding the mech-anism of VILI (3, 4) and toward the de-velopment of clinically applicable mea-surements that can be used to guide theclinician in adjusting mechanical ventila-tion to minimize VILI (5–8). There arethree basic mechanisms of VILI: vo-lutrauma, atelectrauma, and biotrauma.Volutrauma is caused by alveolar overex-pansion secondary to high lung volumewith or without high pressure. Atelec-trauma is an alveolar shear-stress injurythat occurs with repetitive alveolar re-cruitment– derecruitment (R/D). Bio-trauma is the injury to alveoli secondaryto inflammation in which cytokines arereleased in response to mechanical inju-

ries sustained by the alveolus (9). A con-sistent feature in all of these mechanismsis that the injury manifests itself predom-inantly at the level of the alveolus oralveolar ducts. However, the relative con-tribution of each mechanism is not cur-rently known. Defining the mechanics ofnormal alveolar ventilation and thepathologic changes in alveolar mechanicsinduced by ARDS and improper mechan-ical ventilation will help determine themechanism of VILI.

The study of dynamic alveolar me-chanics examines the behavior of alveoliduring ventilation in the normal and ab-normal lung. Alveolar mechanics must beunderstood to develop effective, protec-tive ventilation strategies that “normal-ize” alveolar behavior in the injured lungand reduce VILI. Other components thatplay a key role in both normal and abnor-mal dynamic alveolar inflation, includingpulmonary surfactant, the elastin/colla-gen supportive framework, or the three-dimensional architecture of the alveolus,have been recently reviewed (10) and falloutside the scope of this review.

The ideal investigative tool in thestudy of dynamic alveolar mechanics isone that can measure the three-dimen-sional changes that occur in the alveolusand alveolar duct continuously through-out tidal ventilation. Because this idealtechnique does not exist, the field has

focused on either dynamic measurementof populations of alveoli (e.g., serial com-puted tomographic images or pressure-volume curves) or by static evaluation ofindividual alveoli (e.g., histology). Ourlaboratory has chosen to evaluate thechanges in the orthogonal projection ofsubpleural alveoli by in vivo microscopywith lung inflation and deflation duringventilation. This technique has providedus with a unique insight into dynamicalveolar mechanics.

Alveolar Mechanics in theUninjured Lung

The behavior of alveoli during ventila-tion in the normal lung remains uncer-tain. The classic theory of alveolar expan-sion and contraction in a “balloon-like,”isotropic fashion may be an oversimplifi-cation. Alveoli may exist in a binary stateof inflation in which changes in lungvolume alter the ratio of inflated to de-flated alveoli and differences in the size ofindividual alveoli are negligible. Ventila-tion may occur primarily with changes inthe size of the alveolar duct or conforma-tional changes as a result of alveolar sep-tal folding. This latter behavior is similarto the crumpling and uncrumpling of apaper bag (11, 12).

There are several studies to supportthe theory of septal folding. Stacks of“heavy” thickened septa and capillaries

From Upstate Medical University, Department ofSurgery, Syracuse, NY.


DOI: 10.1097/01.CCM.0000155928.95341.BC

Objectives: To review the mechanism of dynamic alveolarmechanics (i.e., the dynamic change in alveolar size and shapeduring ventilation) in both the normal and acutely injured lung; toinvestigate the alteration in alveolar mechanics secondary to acutelung injury as a mechanism of ventilator-induced lung injury (VILI);and to examine the hypothesis that the reduced morbidity andmortality associated with protective strategies of mechanical venti-lation is related to the normalization of alveolar mechanics.

Data Extraction and Synthesis: This review is based on originalpublished articles and review papers dealing with the mechanismof lung volume change at the alveolar level and the role of alteredalveolar mechanics as a mechanism of VILI. In addition, data fromour laboratory directly visualizing dynamic alveolar mechanics isreviewed and related to the literature.

Conclusions: The mechanism of alveolar inflation in normallungs is unclear. Nonetheless, normal alveoli are very stableand change size very little with ventilation. Acute lung injurycauses marked destabilization of individual alveoli. Alveolarinstability causes pulmonary damage and is believed to be amajor component in the mechanism of VILI. Ventilator strate-gies that reduce alveolar instability may potentially reduce themorbidity and mortality associated with VILI. (Crit Care Med2005; 33[Suppl.]:S122–S128)

KEY WORDS: alveolar mechanics; dynamic alveolar inflation;alveolar recruitment/de-recruitment; ventilator-induced lung in-jury; acute respiratory distress syndrome; ventilator-induced lunginjury; lung mechanics


have been seen by electron microscopy(11). The pulmonary epithelium has beenshown to fold back on adjacent epithe-lium (12). Furthermore, the physicalcharacteristics of the air–liquid interfaceprovide sufficient “slackness” to allowseptal folding (13).

To determine whether the alveolus ex-pands like a balloon, the alveolar surfacearea was measured at various lung vol-umes. Initial studies demonstrated thatalveolar surface area changed by the twothirds power of lung volume, suggestingthat alveoli change volume isotropically(14–18). Others have shown that alveoliexpand anisotropically (19) or combinedisotropic expansion of the alveolus andalveolar ducts (17). These data suggestthat alveolar expansion is complex withmultiple potential mechanisms of dy-namic alveolar expansion.

It has also been suggested that thelung changes volume by “normal” alveo-lar recruitment and derecruitment (R/D).This postulate suggests that alveoli donot change size during ventilation (otherthan by total collapse or rapid inflation)such that there are more alveoli open atinspiration as compared with expiration(20–24). Smaldone et al. demonstratedthat alveoli shrink in size as the numberof alveoli grows (19). They concluded thatthe normal lung changes volume by al-veolar R/D. This finding was confirmed bythe histologic measurement of alveolarsize from lungs fixed at various volumes(20). In vivo studies observing subpleuralalveoli as the lung was inflated from re-sidual volume (RV) to 80% total lungcapacity (TLC) supported the hypothesisthat the lung changes volume by alveolarR/D (20). A recent morphometric studyby Escolar (24) also demonstrated littlechange in alveoli size with lung volumechange with a large change in alveolarnumber. This group further postulatedthat the hysteresis between the inflationand deflation limb of the static P-V curveis the result of the number of open alve-oli. Mathematical interrogation of a sim-ulated P-V curve created from data ini-tially obtained from excised human lungssupports the theory that changes in thenumber of inflated alveoli are responsiblefor the hysteresis in the P-V curve (23).

In an attempt to overcome erroneousinterpretation based on indirect mea-sures of alveolar change, our laboratoryuses in vivo microscopy to directly ob-serve and analyze dynamic alveolar infla-tion during tidal ventilation, as well aslung inflation to near-TLC (25–30). In

vivo microscopy is the only techniquethat allows real-time analysis of dynamicalveolar mechanics in the living animal.Our work demonstrated that alveolar di-ameter changes minimally during tidalventilation, regardless of the size of thetidal volume (28) (Fig. 1, A and B). Thesedata support the hypothesis that the lungdoes not change volume by isotropic, bal-

loon-like expansion and contraction ofalveoli. However, the two-dimensionallimitation of this technique does not per-mit determination of other possiblemechanisms of lung volume change suchas change in alveolar duct volume or sep-tal folding. Regardless of the mechanismof dynamic alveolar inflation in the nor-mal lung, it is clear that the normal al-

Figure 1. In vivo photomicrographs of the same normal (A and B) and acutely injured lung (C and D).Alveoli at peak inspiration (A) and end expiration (B) in the normal lung are very stable with littlechange in size during tidal ventilation (dots). High positive inspiratory pressure (PIP) and low PEEPinjurious ventilation causes a ventilator-induced lung injury resulting in alveolar instability. Injuredalveoli at peak inspiration (C) are inflated (dots) and totally collapse (arrows) end expiration (D),demonstrating severe instability during tidal ventilation.

Figure 2. Alveolar stability assessed by subtracting the area of the alveolus at inspiration (I) from thatat expiration (E) using computer image analysis. The higher the I-E�, the more unstable the alveoli.Normal alveoli (control) are very stable. Tween lavage deactivates surfactant and causes alveolarinstability (5 mins). Without additional positive end-expiratory pressure (PEEP), alveoli remainsignificantly unstable for 180 mins (TWEEN). Increasing PEEP (TWEEN � PEEP) rapidly stabilizesalveoli. (Reproduced with permission from Am J Respir Crit Care Med (30)).


veolus is very stable with little movementduring tidal ventilation. In fact, our datahas demonstrated that alveoli are stablewith tidal volumes even as high as 30mL/kg and peak airway pressures exceed-ing 50 cm H2O (28).

The bulk of available evidence sup-ports the theory that normal alveoli arevery stable and undergo relatively smallchanges in size during ventilation unlessthey totally collapse or reexpand (i.e.,“normal” R/D). This collapse and expan-sion likely occurs by the folding and un-folding of alveolar septa. The contribu-tion of changes in the size of alveolarducts remains uncertain.

Alteration in Alveolar Mechanicsin Acute Lung Injury

Although the majority of evidence, in-cluding data from our laboratory, sug-gests that normal alveoli do not changesize appreciably during ventilation, largechanges in alveolar size and widespreadalveolar R/D appear predominant in acutelung injury. Taskar et al. caused alveolarR/D by ventilating with negative end-expiratory pressure (NEEP) in both nor-mal (31) and surfactant-deactivated lungs(32). NEEP was associated with transientchanges in compliance and gas exchangein the normal lung but not histologicinjury (31). On the other hand, NEEP inlungs after surfactant-deactivationcaused severe histologic damage in addi-tion to anticipated changes in complianceand gas exchange (32). NEEP-inducedR/D caused lung injury only if the surfac-tant system was comprised. Tremblaydemonstrated that ventilating normal ex-cised lungs with very high peak inspira-tory pressure (PIP) and low positive end-expiratory pressure (PEEP) caused injurydemonstrated by release of inflammatorycytokines (33). Although dynamic alveo-lar stability was not directly measured,the mode of ventilation used in this studyshould theoretically promote alveolarR/D.

The slope of the pressure/volumecurve during tidal ventilation is thoughtto reflect the pattern of alveolar behavior.Grasso et al. (34) hypothesized that alinear curve indicates normal aerated al-veoli, an increase in slope (increasingcompliance) indicated tidal alveolar re-cruitment, and that a decrease in slope(decreasing compliance) identified alveo-lar overinflation. They corroborated thistheory with evidence from computed to-mography scans demonstrating a corre-

lation between the slope of the inspira-tion curve and the lung volume duringtidal recruitment or hyperinflation. Neu-mann et al. (35) caused lung injury byoleic acid, saline lavage, and endotoxinand assessed the temporal properties ofalveolar R/D with computed tomographyscans. They found that alveolar R/D oc-

curred very rapidly in all three injuriesand that the etiology of the injury im-pacts the amount of alveolar collapse andreopening with tidal ventilation. To-gether these studies suggest that dy-namic alveolar inflation is altered inacute lung injury during tidal ventilationwith the predominant mode of lung vol-

Figure 3. Confocal images of subpleural alveoli (top, control and injury). Red nuclei (propidium iodide[PI]) mark the injured cells (top, injury). Bar graph, mean number of PI-positive cells per alveolus;mL/kg, tidal volume; ZEEP, 0 positive end-expiratory pressure; PEEP, positive end-expiratory pressure.(Reproduced with permission from Am J Respir Crit Care Med (38).)

Figure 4. Left panel, rat lung ventilated with high tidal volume–low airway pressure for 20 mins. A typeII epithelial cell (PII) is intact, whereas a type I epithelial cell is injured (arrows). The basementmembrane is denuded (arrows) and lined with cell debris and fibrinous deposits (hyaline membranes).AS, alveolar space, i.e., interstitial edema; ca, capillary lumen (original magnification 7,100�). Rightpanel, high-pressure ventilation plus positive end-expiratory pressure. Type I cells are intact (arrows)with the only pathology being endothelial blebs. En, endothelial cell; PII, original magnification,7,100�). (Reproduced with permission from Am J Respir Crit Care Med (39).)


ume change in the injured lung beingwidespread alveolar R/D.

In stark contrast to our findings thatnormal alveoli are extremely stable, alve-oli in surfactant–deactivation models ofARDS exhibit widespread alveolar R/D oc-curring with each breath (Fig. 1, C and D)(25–30). These studies yield direct visualevidence that dynamic alveolar inflationis dramatically altered in acute lung in-jury.

Dynamic Alveolar Mechanicsand Ventilator-Induced LungInjury

The unstable alveoli that open and col-lapse with each breath can cause a signif-icant shear stress-induced lung injury(36). Ventilator settings that are associ-ated with alveolar R/D (high PIP and lowPEEP) can cause VILI in normal lungs(33). A reduction in tidal volume thatpresumably reduces alveolar R/D hasbeen shown to reduce mortality in pa-tients with ARDS (1, 5).

We have directly correlated alterationsin dynamic alveolar mechanics and VILI(30). Altered dynamic alveolar mechanicswere identified after surfactant–deactiva-tion by in vivo microscopy. In the controlgroup, PEEP was not elevated after sur-factant–deactivation, and alveoli remainunstable for nearly 4 hrs (Fig. 2). PEEPwas elevated immediately after surfac-tant–deactivation in the treatment groupto a level that stabilized all alveoli. Lungfunction, tissue and plasma cytokine andprotease levels, and histologic assessmentof lung tissue were measured. The stabi-lization of alveoli with PEEP significantlyimproved lung function and histologicevidence of lung injury. Interleukin-6,but not tumor necrosis factor-�, was el-evated in both plasma and bronchoalveo-lar lavage (BAL) in the group with unsta-ble alveoli but not in the high-PEEPgroup with stable alveoli. Although thenumber of neutrophils was similarly ele-vated, the neutrophil-released proteasesclosely associated with biotrauma (elas-tase and collagenase) were not elevated ineither group. Given that the low-PEEPgroup had physiological and histologicinjury, yet were devoid of significantchange in tissue-damaging proteases, wefeel these data suggest that abnormal me-chanical forces are the initial mechanismof VILI rather than biotrauma (30). We donot mean to discount the possibility thatpersistent cytokine and protease imbal-ance may be injurious. However, this ap-

Figure 5. Laser confocal images of normal (left) and edematous (right) subpleural alveoli in a rat.Normal air-filled alveoli on the left and edema (solid white) filled alveoli on the right. (Reproducedwith permission from Am J Respir Crit Care Med (41).)

Figure 6. Theoretical stresses imparted on epithelial cells during airway reopening. A, a collapsedcompliant airway is forced open by a finger of air. Circles show how the stresses of reopening mightaffect epithelial cells. B, a fluid-filled narrow channel is cleared by a finger of air. Circles show how thestresses of fluid clearance might affect epithelial cells. (Reproduced with permission from Am J RespirCrit Care Med (42).)


pears to be secondary to the influence ofdirect mechanical forces.

As a result of the large prevalence ofalveolar R/D, atelectrauma appears to bea critical component of VILI. It is unclearhow alveolar R/D and associated shearstresses translate into histologic andfunctional injury. Shear stress may causegross tearing of the alveolar wall, injuryto the cell membrane, or ultrastructuralinjury (37–39). Hotchkiss et al. (37) usedscanning electron microscopy to examinethe postmortem lungs of a patient whowas subjected to high-pressure mechani-cal ventilation during treatment of ARDS.They found multiple gross disruptions ofthe alveolar wall suggesting that grosstearing of the alveolar wall occurs inVILI. Gajic et al. (38) demonstrated thatventilation with high PIP and low PEEPreversibly damaged pulmonary cell mem-branes (Fig. 3). Transmission electronmicroscopy revealed that injurious venti-lation caused significant ultrastructuraldisruption to both pulmonary epitheliumand endothelium (39). There was wide-spread destruction of epithelial cells lead-

ing to denudation of the basement mem-brane. In addition, multiple gaps wereseen in the capillary endothelium (Fig.4). These data highlight the fact that lunginjury caused by VILI includes gross tear-ing of lung tissue, reversible injury tocellular membranes, and ultrastructuraldamage.

An alternate hypothesis has been pro-posed to explain the mechanism of VILIat the alveolar level. The central theme ofthis postulate is that alveoli are not un-stable in ARDS but rather flooded withedema fluid (Fig. 5) (40, 41). The authorssuggest that if this hypothesis is true, themechanism of alveolar injury could notbe caused by shear stress damage withalveolar collapse and reopening. Bilek etal. (42) explored the mechanism of shearforce injury to pulmonary epithelial cellsduring opening of a collapsed airway ormoving a finger of air through a floodedairway. In either case, a dynamic stresswave was imparted onto the walls of theairway causing severe cellular deforma-tion (Fig. 6). This suggests that regard-less of which hypothesis concerning ab-

normal alveolar mechanics is correct(i.e., unstable alveoli with massive alveo-lar R/D or stable patent alveoli completelyfilled with edema), ventilating injuredlungs would cause a shear stress injury topulmonary epithelial cells.

Mechanical Ventilation, AlveolarMechanics, and Reduction ofVentilator-Induced Lung Injury

Is there any evidence that the mecha-nism of protective mechanical ventilationis the result of stabilizing alveoli? Theimprovement in mortality with reducedtidal volume in the ARDSnet study (1)could be attributed to decreasing alveolarR/D during tidal ventilation and thus sta-bilizing alveoli. We demonstrated severealveolar instability (R/D) in a surfactantdeactivation model of ARDS by direct vi-sualization (25–30). In addition, we con-firmed that decreasing tidal volume sig-nificantly improved alveolar stability (27)and that stabilizing alveoli with PEEPsignificantly reduced pulmonary damage,suggesting that modes of ventilation that

Figure 7. Alveolar number (# Alveoli/Field) and alveolar stability (I-E�) before (before RM), during (during RM), and after a recruitment maneuver (RM)with either 5 (positive end-expiratory pressure [PEEP] 5 after RM) or 10 (PEEP 10 after RM) cm H2O PEEP added. Note that with only 5 cm H2O PEEPafter RM (PEEP 5 after RM) that alveoli recollapse (fall in # alveoli/field) and the alveoli that remain open were unstable (increased I-E�). Adding PEEP(PEEP 10 after RM) prevented both recollapse and instability. (Reproduced with permission from Am J Respir Crit Care Med (29).)


reduce or prevent alveolar instability arelung-protective (30).

Application of PEEP has been shownto be protective in animal models of VILI(30, 39). Less is known about the com-bined effects of a recruitment maneuver(RM) followed by PEEP. We have learnedthat alveoli recruited with an acceptedRM will often recollapse if subsequentlyventilated with “minimal” PEEP (e.g.,similar to that amount of PEEP used be-fore the RM) (29). However, increasingPEEP after the RM prevented alveolarrecollapse and stabilized newly recruitedalveoli (29) (Fig. 7). Theoretically, if ven-tilation that allows alveolar instabilitycauses lung injury, mechanical ventila-tion that stabilizes alveoli should reduceVILI.

High-frequency oscillatory ventilation(HFOV) has been shown to be protectivein both animals (43) and humans (44).Imai et al. (43) compared three protectiveconventional ventilation strategies withHFOV in saline-lavaged rabbits. Saline-lavaged rabbits with nonprotective con-ventional mechanical ventilation (CMV)had a significant decrease in pulmonarycompliance, with increases in neutrophilinfiltration, tumor necrosis factor-� con-centration in the bronchoalveolar fluid,and histologic injury. HFOV attenuatedall of these changes. However, protectiveCMV using low tidal volume and highPEEP only improved oxygenation andpulmonary compliance. It was concludedthat HFOV may be preferable to protec-tive conventional ventilation as a lung

protective strategy (43). In a randomizedclinical trial, it was shown that the 30-daymortality of patients on HFOV was 37%,whereas mortality was 52% in the pa-tients on conventional ventilation (44).This difference in mortality did not reachstatistical significance; however, thestudy was underpowered for this endpoint.

In preliminary studies, our groupcompared alveolar stability with CMV toventilation with HFOV in a rat modelusing a 2.5-internal diameter trachealtube. After lung injury by saline lavage,we adjusted each ventilator to yield sim-ilar blood gases and noted that there wasa dramatic improvement in alveolar sta-bility using HFOV as compared with CMV(Fig. 8). Further investigation is requiredto determine whether improvement inalveolar stability actually reduces mor-bidity associated with alveolar R/D-induced VILI.

Review Summary

The mechanism of dynamic lung vol-ume change at the alveolar level is un-known. Rolf Hubmayr recently statedthat, “It is remarkable how little is knownabout alveolar deformation duringbreathing” (41). This lack of knowledgeconcerning normal alveolar functionmakes it very difficult to assess the patho-physiology induced by abnormal alveolarfunction. The majority of studies supporta hypothesis that alveoli in the acutelyinjured lung become unstable, collapsing

and reopening with each breath (26–30,32–34); however, others postulate thatalveoli are stable and filled with edema(39, 40). Regardless of whether alveoli areunstable or edema-filled, the shear stress-induced injury to pulmonary epithelialcells that occur with mechanical ventila-tion can cause clinically significant cellu-lar injury (42). Ventilatory strategies thatreduce alveolar instability in animals (30,39) and humans (1, 5) have reduced VILI.This suggests that ventilator maneuversthat promote alveolar stability such asPEEP or HFOV may reduce VILI, espe-cially if used early in the course of thedisease.

The more understanding that we gainconcerning both normal and abnormalalveolar mechanics, the better equippedwe will be to understand the mechanismof VILI and how to improve ventilatorstrategies to reduce the negative effects ofmechanical ventilation.

REFERENCES

1. Acute Respiratory Distress syndrome Net-work: Ventilation with lower tidal volumes ascompared with traditional tidal volumes foracute lung injury and the acute respiratorydistress syndrome. N Engl J Med 2000; 342:1301–1308

2. Rubenfeld GD: Epidemiology of acute lunginjury. Crit Care Med 2003; 31:S276–S284

3. Pinhu L, Whitehead T, Evans T, et al: Venti-lator-associated lung injury. Lancet 2003;361:332–340

4. Dreyfuss D, Saumon G: Ventilator-inducedlung injury. Lessons from experimental stud-ies. Am J Respir Crit Care Med 1998; 157:294–323

5. Amato MBP, Barbas CSV, Medeiros DM, et al:Effect of a protective-ventilation strategy onmortality in the acute respiratory distresssyndrome. N Engl J Med 1998; 338:347–354

6. Putensen C, Baum M, Hormann C: Selectingventilator settings according to variables de-rived from the quasi-static pressure/volumerelationship in patients with acute lung in-jury. Anesth Analg 1993; 77:436–447

7. Harris RS, Hess DR, Venegas JG: An objectiveanalysis of the pressure-volume curve in theacute respiratory distress syndrome. Am JRespir Crit Care Med 2000; 161:432–439

8. Victorino JA, Borges JB, Okamato, et al: Im-balances in regional lung ventilation. A vali-dation study on electrical impedance tomog-raphy. Am J Respir Crit Care Med 2004; 169:791–800

9. Slutsky AS: Lung injury caused by mechan-ical ventilation. Chest 1999; 116(suppl):9S–15S

10. D’Angelo E: Local alveolar size and transpul-monary pressure in situ and in isolatedlungs. Respir Physiol 1972; 14:251–266

Figure 8. In vivo photomicrographs of subpleural alveoli in the rat after lung injury by saline lavageventilated with either conventional mechanical ventilation (CMV) or high-frequency oscillatory ven-tilation (HFOV) using a 2.5-internal diameter tracheal tube. With CMV, a group of alveoli are seeninflated during inspiration (dots) but collapse with expiration (arrows). Alveoli are very stable withHFOV during ventilation. The same alveolus is seen with HFOV at inflation and exhalation (dots).


11. Dunnill MS: Effect of lung inflation on alve-olar surface area in the dog. Nature 1967;214:1013–1014

12. Klingele TG, Staub NC: Alveolar shapechanges with volume in isolated, air-filledlobes of cat lung. J Appl Physiol 1970; 28:411–414

13. Kuno K, Staub NC: Acute mechanical effectsof lung volume changes on artificial micro-holes in alveolar walls. J Appl Physiol 1968;24:83–92

14. Storey WF, Staub NC: Ventilation of terminalair units. J Appl Physiol 1962; 17:391–397

15. Gil J, Weibel ER: Morphological study ofpressure-volume hysteresis in rat lungs fixedby vascular perfusion. Respir Physiol 1972;15:190–213

16. Gil J, Bachofen H, Gehr P, et al: Alveolarvolume–surface area relation in air and sa-line-filled lungs fixed by vascular perfusion.J Appl Physiol 1979; 47:990–1001

17. Forest JB: Lung tissue plasticity: morpho-metric analysis of anisotropic strain in liquidfilled lungs. Respir Physiol 1976; 27:223–239

18. Carney DE, Bredenberg CE, Schiller HJ, etal: The mechanism of lung volume changeduring mechanical ventilation. Am J RespirCrit Care Med 1999; 160:1697–1702

19. Smaldone GC, Mitzner W, Itoh H: Role ofalveolar recruitment in lung inflation: influ-ence on pressure-volume hysteresis. J ApplPhysiol 1983; 55:1321–1332

20. Lum H, Huang I, Mitzner W: Morphologicalevidence for alveolar recruitment during in-flation at high transpulmonary pressure.J Appl Physiol 1990; 68:2280–2286

21. Salmon RB, Primiano FP, Saidel GM, et al:Human lung pressure-volume relationships:alveolar collapse and airway closures. J ApplPhysiol 1981; 51:353–362

22. Escolar JD, Escolar MA, Guzman J, et al:Pressure volume curve and alveolar recruit-ment/de-recruitment. A morphometricmodel of the respiratory cycle. Histol His-topathol 2002; 17:383–392

23. Oldmixon EH, Hoppin FG: Alveolar septalfolding and lung inflation history. J ApplPhysiol 1991; 71:2369–2379

24. Gatto LA, Fluck RR, Nieman GF: Alveolarmechanics in the acutely injured lung: roleof alveolar instability in the pathogenesis ofventilator-induced lung injury. Respir Care2004; 49:1045–1055

25. Mead J, Takishima T, Leith D: Stress distri-bution in lungs: a model of pulmonary elas-ticity. J Appl Physiol 1970; 28:596–608

26. Taskar V, John J, Evander E, et al: Healthylungs tolerate repetitive collapse and reopen-ing during short periods of mechanical ven-tilation. Acta Anaesthesiol Scand 1995; 39:370–376

27. Taskar V, John J, Evander E, et al: Surfactantdysfunction makes lungs vulnerable to repet-itive collapse and reexpansion. Am J RespirCrit Care Med 1997; 155:313–320

28. Tremblay L, Valenza F, Riberio SP, et al:Injurious ventilatory strategies increase cy-tokines and c-fos m-RNA expression in anisolated rat lung model. J Clin Invest 1997;99:944–952

29. Neumann P, Berglund JE, Mondejar EF, etal: Dynamics of lung collapse and recruit-ment during prolonged breathing in porcinelung injury. J Appl Physiol 1998; 85:1522–1543

30. Grasso S, Terragni P, Mascia L, et al: Airwaypressure-time curve profile (stress index) de-tects tidal recruitment/hyperinflation in ex-perimental acute lung injury. Crit Care Med2004; 32:1018–1027

31. Nieman GF, Bredenberg CE, Clark WR, et al:Alveolar function following surfactant deac-tivation. J Appl Physiol 1981; 51:895–904

32. Schiller HJ, McCann UG, Carney DE, et al:Altered alveolar mechanics in the acutely in-jured lung. Crit Care Med 2001; 29:1049–1055

33. Steinberg J, Schiller HJ, Halter JM, et al:Tidal volume increases do not affect alveolarmechanics in normal lung but cause alveolarover-distension and exacerbate alveolar in-stability following surfactant deactivation.Crit Care Med 2002; 30:2675–2683

34. McCann UG, Schiller HJ, Carney DE, et al:Visual validation of the mechanical stabiliz-ing effects of positive end-expiratory pressure

at the alveolar level. J Surg Res 2001; 99:335–342

35. Halter JM, Steinberg JM, Schiller HJ, et al:Positive end-expiratory pressure (PEEP) aftera recruitment maneuver prevents both alve-olar collapse and recruitment/derecruit-ment. Am J Respir Crit Care Med 2003; 167:1620–1626

36. Steinberg J, Schiller HJ, Halter JM, et al:Alveolar instability causes early ventilator-induced lung injury independent of neutro-phils. Am J Respir Crit Care Med 2004; 169:57–63

37. Hotchkiss JR, Simonson DA, Marek DJ, et al:Pulmonary microvascular fracture in a pa-tient with acute respiratory distress syn-drome. Crit Care Med 2002; 30:2368–2370

38. Gajic O, Lee J, Doerr CH, et al: Ventilator-induced cell wounding and repair in the in-tact lung. Am J Respir Crit Care Med 2003;167:1057–1063

39. Dreyfuss DP, Soler P, Basset G, et al: Highinflation pressure pulmonary edema: respec-tive effects of high airway pressure, high tidalvolume an positive end-expiratory pressure.Am Rev Respir Dis 1988; 137:1159–1164

40. Wilson TA, Anafi RC, Hubmayr RD: Mechan-ics of edematous lungs. J Appl Physiol 2001;90:2088–2093

41. Hubmayr RD: Perspective on lung injury andrecruitment. A skeptical look at the openingand collapse story. Am J Respir Crit Care Med2002; 165:1647–1653

42. Bilek AM, Dee KC, Gaver DP: Mechanisms ofsurface-tension-induced epithelial cell dam-age in a model of pulmonary airway reopen-ing. J Appl Physiol 2003; 94:770–783

43. Imai Y, Nakagawa S, Ito Y, et al: Comparisonof lung protective and high-frequency oscil-latory ventilation. J Appl Physiol 2001; 91:1836–1844

44. Derdak S, Mehta S, Stewart TE, et al: High-frequency oscillatory ventilation for acute re-spiratory distress syndrome in adults. A ran-domized, controlled trial. Am J Respir CritCare Med 2002; 166:801–808


High-frequency oscillatory ventilation and ventilator-inducedlung injury

Yumiko Imai, MD, PhD; Arthur S. Slutsky, MD

Acute respiratory distress syn-drome (ARDS) is the most se-vere form of acute lung injuryand has a mortality rate of at

least 30%. No effective drugs exist totreat ARDS, and therapy is largely sup-portive with mechanical ventilation. Al-though mechanical ventilation is oftenlifesaving for patients with ARDS, me-chanical ventilation can cause lung in-jury, a concept that has been termed ven-tilator-induced lung injury (VILI). In anattempt to minimize the detrimentalconsequences of VILI, scientists and cli-nicians have been studying different ven-tilatory strategies. One form of ventila-tion that has garnered significant interestis the use of high-frequency oscillatoryventilation (HFOV) in patients with vari-ous forms of respiratory distress syn-drome. This article will review a numberof concepts related to VILI and, specifi-

cally, how HFOV might fit into the clini-cian’s armamentarium for the ventilationof patients with respiratory distress syn-drome.

VILI and HFOV: TheoreticalConsiderations

There are a number of mechanismsthat can lead to development of VILI,including gross air leaks (barotrauma),diffuse alveolar injury due to overdisten-sion (volutrauma), injury due to repeatedcycles of recruitment/derecruitment (at-electrauma) (Fig. 1), and the most subtleform of injury due to the release of me-diators from the lung (biotrauma) (Fig.2) (1–3). Lungs of patients with ARDS areheterogeneously damaged, and hence,mechanical ventilation with normal oreven low tidal volumes can lead to re-gional lung injury. Recruitment/dere-cruitment denotes the situation wherebyalveolar units open during inspirationand collapse again during expiration. Therepetitive cycling in which lung unitsopen and collapse again during expirationresults in high shear stress, which canfurther injure the lungs. Reducing themagnitude of these cyclic fluctuationsand application of higher positive end-expiratory pressure (PEEP) levels can

minimize VILI (4, 5). Based on these con-cepts, various “lung-protective” strate-gies have been developed to minimizeVILI. One strategy uses relatively smalltidal volumes and PEEP titrated to a fewcentimeters of H2O above the lower in-flection point (Pinf) on the pressure–volume curve (6, 7). In 2000, the ARDSNet investigators reported a 9% decreasein absolute mortality of patients withARDS using a lung-protective strategyusing a small tidal volume (6 mL/kg pre-dicted body weight) with a PEEP thataveraged �10 cm H2O (8).

Within this context, HFOV can beviewed as providing alveolar ventilationwith very small tidal volumes and thus,theoretically, could be viewed as provid-ing the optimal lung-protective ventila-tory strategy. HFOV has novel gas ex-change mechanisms that contribute tobetter oxygenation and CO2 removal.Bulk convection and diffusion are themain mechanisms of gas exchange dur-ing conventional mechanical ventilation(CMV), whereas interregional gas mixingbetween units with different time con-stants (pendelluft), convective transportattributable to asymmetry between in-spiratory and expiratory velocity profiles,and longitudinal dispersion due to inter-action between the axial velocity profile

From the Institute of Molecular Biotechnology,Austrian Academy of Science, Vienna, Austria (YI); theInterdepartmental Division of Critical Care Medicineand Division of Respirology, Department of Medicine,University of Toronto, Ontario, Canada (ASS); and theDepartment of Critical Care Medicine, St. Michael’sHospital, Toronto, Ontario, Canada (ASS).


DOI: 10.1097/01.CCM.0000156793.05936.81

Introduction: Although mechanical ventilation is lifesaving forpatients with acute respiratory distress syndrome, it can causeventilator-induced lung injury. To minimize ventilator-inducedlung injury, different ventilatory strategies have been developed. One ofthe strategies is the use of high-frequency oscillatory ventilation (HFOV).

Theoretical Backgrounds of Ventilator-Induced Lung Injury and HFOV:Because of the novel gas exchange mechanisms, HFOV can provideadequate gas exchange using extremely small tidal volumes and main-tain high end-expiratory lung volume without inducing overdistension,which should result in minimization of ventilator-induced lung injury.

Studies of HFOV and Lung Injury: There are convincing clinicaland animal data indicating that HFOV is an ideal lung-protectiveventilatory strategy, particularly in the setting of neonatal respi-ratory failure, if lung volume recruitment is performed.

Clinical Implication of HFOV in Adult Acute Respiratory DistressSyndrome: A recent clinical trial demonstrated early (<16 hrs) im-provement in oxygenation with HFOV and a 30-day mortality of 37%with HFOV vs. 52% with pressure-controlled ventilation (p � .102),suggesting that HFOV is as effective and safe as the conventionalstrategy in adult acute respiratory distress syndrome. Future studiesexamining optimal algorithms of HFOV using clinically relevant ani-mal models, and patients with acute respiratory distress syndrome,are imperative to determine whether the wide-spread application ofHFOV is warranted in adult acute respiratory distress syndrome. (CritCare Med 2005; 33[Suppl.]:S129–S134)

KEY WORDS: high-frequency oscillatory ventilation; ventilator-induced lung injury; acute respiratory distress syndrome; animalmodels


and radial concentration gradient (Tay-lor’s dispersion) also play an importantrole during HFOV (Fig. 3) (9–11). Be-cause of these mechanisms, adequate gasexchange during HFOV is possible with

extremely small tidal volumes, often lessthan the anatomic dead space (1–3 mL/kg).In addition, during HFOV, it is possible tomaintain relatively high end-expiratorylung volume, without inducing overdisten-

sion. HFOV may have a larger margin ofsafety in keeping the lung open within thedesired target range of alveolar overdisten-sion in heterogeneously injured ARDSlungs (12) (Fig. 4).

Figure 1. Overdistension (volutrauma) and recruitment/derecruitment (atelectrauma). Overdistension (volutrauma) develops when inspired air preferablydistributes to the areas with higher compliance (nonatelectatic regions). Recruitment/derecruitment denotes the situation whereby alveolar units openduring inspiration and collapse again during expiration in atelectatic regions. This cycle of repeated opening and collapse results in high shear stress thatcan further injure the lungs (atelectrauma), in particular at end-expiration. Reproduced with permission from Frank JA, Imai Y, Slutsky AS: Pathogenesisof ventilator-induced lung injury. Physiological Basis of Ventilatory Support [Marcel Dekker] 2003.

Figure 2. Postulated mechanisms whereby volutrauma, atelectrauma, and biotrauma caused by mechanical ventilation contribute to multiple organ dysfunctionsyndrome (MODS). The potential importance of biotrauma is not only that it can aggravate ongoing lung injury, but also that it can contribute to the developmentof MODS, possibly through the release of proinflammatory mediators from the lung. Adapted with permission from Slutsky and Tremblay (2).


Venegas and Fredberg (13) developed atheoretical model to determine the optimalventilatory variables in patients with lungdisease. They formulated the problem ofdeveloping an optimal ventilatory strategyby dividing it into two simpler problems: 1)examination of the factors related to pres-sure cost per unit of convective oscillatoryflow and 2) the convective flow cost neces-sary to achieve a unit of alveolar ventila-tion. They obtained simple solutions forthese two functions. Their model includedmodels of gas exchange and lung mechan-ics, including the effects of lung inflationtidal volume and respiratory frequency inalveolar ventilation, nonlinear lung tissuecompliance, and alveolar recruitment andderecruitment. They then determined theputative detrimental effects of high-fre-quency ventilation as a function of the ven-tilatory settings and the pathophysiologicvariables of the patient. Their model pre-dicted that for respiratory distress syn-drome (RDS) in neonates, the selectedPEEP level was critical because detrimentalconsequences were increased at both high

and low values of PEEP. Of interest, in theinfant respiratory distress syndrome pa-tient, the choice of which respiratory fre-quency to use was not as critical for fre-quencies of �10 Hz. The analysissupported the use of high-frequency venti-lation in infant respiratory distress syn-drome and of ensuring adequate end-expiratory pressure. Their model predictedthat the range of “safe” frequency–PEEPcombinations would be relatively narrowand move to higher frequencies as lungcompliance decreases. Also, if similar tidalvolumes and levels of PEEP were applied atconventional frequencies (�50 breaths/min), CO2 clearance would be compro-mised. These data provide a theoreticalbackground for the putative advantages ofHFOV in the treatment of patients withARDS.

HFOV and Studies of LungInjury

There are a number of animal studiesdemonstrating reduced VILI with HFOV. A

number of studies that laid the groundworkfor our current understanding of HFOVand VILI were published in the 1980s andearly 1990s by the Toronto group led byBryan and Froese. Kolton et al. (14) exam-ined CMV and HFOV in models of oleic acidinjury and lung lavage. They found thatwhen HFOV was combined with a sustainedinflation (i.e., recruitment maneuver),there were larger mean lung volumes andimproved oxygenation with HFOV. Theysuggested that this approach of a recruit-ment maneuver and high mean airwaypressures during HFOV could “more fullyexploit the pressure volume hysteresis ofunstable lung units than CMV.” Hamiltonet al. (15) examined oxygenation and lungpathology in rabbits with saline lavage–induced lung injury, ventilated with HFOVor CMV. HFOV provided marked improve-ments in oxygenation over 5 hrs, and mostimportantly, the animals treated withHFOV had markedly attenuated lung injuryas assessed by hyaline membranes. Theyconcluded that “avoidance of low lung vol-ume and large pressure–volume changes

Figure 3. Gas transport mechanisms during high-frequency oscillatory ventilation. Bulk convection and diffusion are the main mechanisms of gas exchange duringconventional mechanical ventilation. Interregional gas mixing between units with different time constants (pendelluft), convective transport attributable toasymmetry between inspiratory and expiratory velocity profiles, and longitudinal dispersion due to interaction between the axial velocity profile and radialconcentration gradient (Taylor’s dispersion) also play a important role during HFO ventilation. Reproduced with permission from Slutsky and Drazen (11).


through the use of HFOV results in reducedpulmonary damage.” This critical conceptin the use of HFOV to mitigate VILI is stillvalid today.

McCulloch et al. (16) addressed theissue of whether the use of high meanairway pressures was important duringHFOV. They ventilated rabbits after lunglavage using three different strategies:HFOV at high mean lung volume, HFOVat low mean lung volume, and CMV at alow mean lung volume. The latter twogroups ventilated at low lung volumeshad lower respiratory system compliance,more hyaline membranes, and more se-vere airway epithelial damage. These datademonstrated that maintenance of an ad-equately high mean lung volume is crit-ical to minimize the lung injury causedby mechanical ventilation, and they alsoemphasized the importance of appropri-ate lung recruitment during HFOV.

In 1987, Delemos et al. (17) used avery clinically relevant model of prema-ture baboons with hyaline membrane dis-ease and reported that HFOV at 10 Hzresulted in decreased pulmonary baro-trauma compared with CMV. Meredith etal. (18) used the same premature baboonmodel and reported that application ofHFOV for 24 hrs protected animals fromlung injury as assessed by gas exchange,lung mechanics, morphologic findings,and measurements of platelet-activatingfactor compared with CMV (peak inspira-tory pressure, 31.4–45.0 cm H2O; PEEP,4.0–6.9 cm H2O). Using saline-lavagedadult rabbits, Matsuoka et al. (19) dem-

onstrated that HFOV led to decreased re-spiratory burst of neutrophil activationcompared with CMV (peak inspiratorypressure, 25 cm H2O; PEEP, 5 cm H2O).Imai et al. (20) used the same model andfound that HFOV, as opposed to CMV(peak inspiratory pressure, 25 cm H2O;PEEP, 5 cm H2O), led to decreased pro-duction of platelet-activating factor andthromboxane-A2 in lung. Furthermore,Takata et al. (21) confirmed that only 1 hrof HFOV produced less tumor necrosisfactor-� messenger RNA in intra-alveolarcells compared with CMV (peak inspira-tory pressure, 28 cm H2O; PEEP, 5 cmH2O) in the same model. Very recently,using a surfactant-depleted piglet model,von der Hardt et al. (22) demonstratedthat messenger RNA expression of cyto-kines (interleukin [IL]-1�, IL-6, IL-8, andIL-10), transforming growth factor-�1,endothelin-1, and adhesion molecules (E-selectin, P-selectin, intercellular adhe-sion molecule-1) in lung tissue and IL-8expression in microdissected alveolarmacrophages were highly reduced withHFOV compared with CMV (peak inspira-tory pressure, 38 cm H2O; PEEP, 8 cmH2O), suggesting HFOV can reduce pul-monary inflammatory response.

All of these data from various animalstudies indicate that HFOV can reduceVILI compared with CMV with conven-tional ventilatory strategies (nonprotec-tive ventilatory strategies). However,early clinical trials targeting neonateswere unable to demonstrate the superi-ority of HFOV over CMV, even when CMVwas used with a nonprotective ventilatorystrategy (23). There was no benefit ofHFOV in the High-Frequency Interven-tion (HIFI) study (23), and HFOV wasassociated with an increased prevalenceof air leak, intracranial hemorrhage, andperiventricular leukomalacia. It was sug-gested that this lack of benefit was relatedto the lack of an adequate volume recruit-ment strategy inherent in the protocols(24). Later studies using a volume re-cruitment strategy demonstrated im-proved gas exchange, reductions in baro-trauma, and overall improved outcome inneonatal patients receiving HFOV (25–27). Using volume recruitment maneu-vers, HFOV maintains end-expiratorylung volume higher up on the deflationpressure–volume relationship without in-ducing simultaneous overdistension be-cause of the much smaller tidal volumeused. This should result in minimalstretch injury generated by the pressureamplitude excursions.

Many of the animal studies describedabove were performed before the wide-spread use of protective ventilatory strat-egies using CMV, so a critical question is:does HFOV have a protective advantagefrom VILI over CMV with protective ven-tilatory strategies? To address this issue,we conducted a study using a rabbit-lunglavage model (28). Because there is noconsensus on what constitutes the opti-mal (protective) strategy with CMV, wechose two strategies that have beenshown to decrease mortality in two re-cent randomized, controlled trials: 1) astrategy similar to that used by Amato etal. (6) in which the PEEP was set 2–3 cmH2O greater than the lower Pinf base onthe inflation limb of the pressure–volumecurve (6, 7) and 2) a strategy similar tothe ARDSNet trial (8) using small tidalvolumes and PEEP of �10 cm H2O. Thefirst strategy that used PEEP � Pinf led tohypotension and barotraumas, suggest-ing that a strategy with PEEP above Pinf

may not be always possible during CMVbecause of hemodynamic compromiseand possible barotrauma. In addition, thevalidity of this approach has been ques-tioned because the lung is often not fullyrecruited, even if PEEP � Pinf, and it doesnot take into account the effect of thechest wall on Pinf. The second strategythat was similar to that used in the ARD-SNet trial (8) fulfilled two criteria for anadequate lung-protective strategy: pla-teau pressure of �30 cm H2O and PaO2 of�300 Torr, indicating adequate recruit-ment. We found that HFOV attenuatedthe decrease in pulmonary compliance,lung inflammation assessed by polymor-phonuclear leukocyte infiltration and tu-mor necrosis factor-� concentration inthe alveolar space, and pathologicchanges of the small airways and alveoli,whereas CMV, with a strategy similar tothe ARDSNet trial, only attenuated thedecrease in oxygenation and pulmonarycompliance (Fig. 5). These data suggestthat HFOV may have a larger margin ofsafety in keeping the lung open withinthe desired target range of alveolar over-distension, and thus, HFOV may have ad-vantages over CMV with respect to VILI,even when CMV is used with a protectiveventilatory strategy.

HFOV: Clinical Implications inAdult ARDS

As described above, there are convinc-ing clinical and animal data indicatingthat HFOV can lead to reduced VILI, par-

Figure 4. Pressure–volume curve of a moderatelydiseased lung, such as one with adult acute re-spiratory distress syndrome. Two hazard zonesexist; overdistension and derecruitment and atel-ectasis. High end-expiratory pressures and smalltidal volumes are needed to stay in the “safe”window. High-frequency oscillatory ventilationmay have a larger margin of safety in keeping thelung open within the desired target range ofalveolar overdistension. Reproduced with permis-sion from Froese (12).


ticularly in the setting of neonatal respi-ratory failure. Can HFOV also reduce VILIin adult patients with ARDS? The limita-tions of applying the previous studies toadult patients are that the previous ani-mal studies used higher frequency (e.g.,15 Hz), lower �P (amplitude), andsmaller uncuffed endotracheal tube sizes(e.g., �4 mm) than are typically used inadult patients. Proximal oscillatory pres-sure (�P) transmitted to the distal alveolidepends on multiple variables, includingendotracheal tube diameter, respiratoryfrequency, percentage of inspiratorytime, airway resistance, lung compliance,and gravitational factors (e.g., lower vs.upper lobe). Respiratory mechanics havea significant influence on the intrapul-monary oscillatory pressure (�P) duringHFOV (29). Low lung compliance can re-sult in significant increases in distal os-cillatory pressure transmission, ap-proaching 20–30% of proximal airwaypressure amplitudes. Conversely, in-creases in peripheral airway resistancemay decrease oscillatory pressure trans-mission to the distal alveolar compart-ment while increasing pressure excur-sions in the central airways (e.g.,

trachea). Also, changing the percentageof inspiratory time (e.g., from 33% to50%) may influence oxygenation andventilation by increasing the distal alve-olar pressure and delivered tidal volume.These concepts suggest that developmentof optimal techniques for achieving oxy-genation and ventilation (i.e., frequency,(�P), percentage of inspiratory time,mean airway pressure) are critical for im-proving clinical outcomes with HFOV inadult ARDS.

In this perspective, studies to examinethe optimal techniques of HFOV using aclinically relevant animal model for adultARDS (e.g., acid aspiration–induced orcecal ligation/perforation–induced) usinglarge animals (e.g., pig or sheep) are im-portant for clinical application of HFOVto adult ARDS. Clinicians are now facedwith the challenge of translating the pu-tative advantages of HFOV into a provableeffect on clinical outcome in adult ARDS.Early studies by Fort et al. (30) and Mehtaet al. (31) focused on the use of HFOV asa rescue therapy (n � 17–42), with pa-tients being administered HFOV onlywhen CMV was observed to be failing.They demonstrated that mean airway

pressure could be safely maintained at ahigher level with HFOV and that oxygen-ation improved with HFOV. These reportsalso suggested that early initiation (2days) of HFOV is more likely to result inimproved survival than delayed initiation(�7 days). A recently published clinicaltrial, the Multicenter Oscillatory Ventila-tion For Acute Respiratory Distress Syn-drome Trial (MOAT, n � 148), comparingHFOV with a pressure-controlled ventila-tion strategy (PaO2/FIO2 ratio of �200mm Hg on PEEP of �10 cm H2O) dem-onstrated early (�16 hrs) improvementin PaO2/FIO2 (p � .008) in the HFOVgroup but no significant difference in ox-ygenation index during the initial 72 hrsof treatment (32). Thirty-day mortalitywas 37% in the HFOV group and 52% inthe conventional ventilation group (p �.102). There was no significant differencebetween treatment groups in the preva-lence of barotrauma, hemodynamic in-stability, or mucus plugging. These stud-ies suggest that HFOV is as effective andsafe as the conventional strategy in pa-tients with ARDS. Conclusions

HFOV is a ventilatory technique thatcan provide adequate gas exchange using

Figure 5. a, Numbers of polymorphonuclear leukocyte (PMN) cells in the lung lavage fluid. Values are mean � SD. Numbers of PMN cells were significantlyhigher in the moderate tidal volume and low positive end-expiratory pressure (MVLP) and low tidal volume and high positive end-expiratory pressure(LVHP) groups than in the high-frequency oscillatory ventilation (HFOV) group. **p � .01 compared with HFOV group; n.s., not significant. b, Levels oftumor necrosis factor-� (TNF-�) in the lung lavage fluid before ventilation and at termination of ventilation in the MVLP, LVHP, and HFOV groups. Valuesare mean � SD. *p � .05 compared with HFOV. **p � .01 compared with HFOV. Levels of TNF-� after ventilation were higher in the MVLP and LVHPgroups than in the HFOV group, and no significant differences were seen between MVLP and LVHP groups. Reproduced with permission from Imai et al(28).


very small tidal volumes. This allows oneto ventilate patients at relatively highmean lung volumes, minimizing therisks of volutrauma and of atelectrauma.Animal data are quite convincing thatHFOV is an ideal lung-protective ventila-tory strategy, but there is a paucity ofclinical data supporting this contentionin humans, especially in studies in whicha protective CMV ventilatory strategy isused. Such studies are required before wefully understand the role of HFOV in thetreatment of our most difficult ventilato-ry-dependent patients.

REFERENCES

1. Imai Y, Parodo J, Kajikawa O, et al: Injuriousmechanical ventilation and end-organ epi-thelial cell apoptosis and organ dysfunctionin an experimental model of acute respira-tory distress syndrome. JAMA 2003; 289:2104–2112

2. Slutsky AS, Tremblay LN: Multiple systemorgan failure: Is mechanical ventilation acontributing factor? Am J Respir Crit CareMed 1998; 157(6 Pt 1):1721–1725

3. Tremblay LN, Slutsky AS: Ventilator-inducedinjury: From barotrauma to biotrauma. ProcAssoc Am Physicians 1998; 110:482–488

4. Dreyfuss D, Saumon G: Ventilator-inducedlung injury: Lessons from experimental stud-ies. Am J Respir Crit Care Med 1998; 157:294–323

5. Slutsky AS: Lung injury caused by mechan-ical ventilation. Chest 1999; 116(1 Suppl):9S–15S

6. Amato MB, Barbas CS, Medeiros DM, et al:Effect of a protective-ventilation strategy onmortality in the acute respiratory distresssyndrome [see Comments]. N Engl J Med1998; 338:347–354

7. Ranieri VM, Suter PM, Tortorella C, et al:Effect of mechanical ventilation on inflam-matory mediators in patients with acute re-spiratory distress syndrome: A randomizedcontrolled trial [see Comments]. JAMA 1999;282:54–61

8. Ventilation with lower tidal volumes as com-pared with traditional tidal volumes for acutelung injury and the acute respiratory distresssyndrome: The Acute Respiratory DistressSyndrome Network. N Engl J Med 2000;342:1301–1308

9. Venegas JG, Fredberg JJ: Understanding thepressure cost of ventilation: Why does high-frequency ventilation work? Crit Care Med1994; 22(9 Suppl):S49–S57

10. Chang HK: Mechanisms of gas transport dur-ing ventilation by high-frequency oscillation.J Appl Physiol 1984; 56:553–563

11. Slutsky AS, Drazen JM: Ventilation withsmall tidal volumes. N Engl J Med 2002;347:630–631

12. Froese AB: High-frequency oscillatory venti-lation for adult respiratory distress syn-drome: Let’s get it right this time! Crit CareMed 1997; 25:906–908

13. Venegas JG, Fredberg JJ: Understanding thepressure cost of ventilation: Why does high-frequency ventilation work? Crit Care Med1994; 22(9 Suppl):S49–S57

14. Kolton M, Cattran CB, Kent G, et al: Oxygen-ation during high-frequency ventilationcompared with conventional mechanicalventilation in two models of lung injury.Anesth Analg 1982; 61:323–332

15. Hamilton PP, Onayemi A, Smyth JA, et al:Comparison of conventional and high-frequency ventilation: Oxygenation and lungpathology. J Appl Physiol 1983; 55(1 Pt 1):131–138


17. Delemos RA, Coalson JJ, Gerstmann DR, etal: Ventilatory management of infant ba-boons with hyaline membrane disease: Theuse of high frequency ventilation. PediatrRes 1987; 21:594–602

18. Meredith KS, Delemos RA, Coalson JJ, et al:Role of lung injury in the pathogenesis ofhyaline membrane disease in premature ba-boons. J Appl Physiol 1989; 66:2150–2158

19. Matsuoka T, Kawano T, Miyasaka K: Role ofhigh-frequency ventilation in surfactant-depleted lung injury as measured by granu-locytes. J Appl Physiol 1994; 76:539–544

20. Imai Y, Kawano T, Miyasaka K, et al: Inflam-matory chemical mediators during conven-tional ventilation and during high frequencyoscillatory ventilation. Am J Respir Crit CareMed 1994; 150(6 Pt 1):1550–1554

21. Takata M, Abe J, Tanaka H, et al: Intraalveo-lar expression of tumor necrosis factor-alphagene during conventional and high-fre-quency ventilation. Am J Respir Crit CareMed 1997; 156:272–279

22. von der Hardt K, Kandler MA, Fink L, et al:High frequency oscillatory ventilation sup-presses inflammatory response in lung tissueand microdissected alveolar macrophages insurfactant depleted piglets. Pediatr Res 2004;55:339–346

23. High-frequency oscillatory ventilation com-pared with conventional mechanical ventila-tion in the treatment of respiratory failure inpreterm infants: The HIFI Study Group.N Engl J Med 1989; 320:88–93

24. Bryan AC, Froese AB: Reflections on the HIFItrial. Pediatrics 1991; 87:565–567

25. Clark RH, Yoder BA, Sell MS: Prospective,randomized comparison of high-frequencyoscillation and conventional ventilation incandidates for extracorporeal membrane ox-ygenation. J Pediatr 1994; 124:447–454

26. Gerstmann DR, Minton SD, Stoddard RA, etal: The Provo multicenter early high-frequency oscillatory ventilation trial: Im-proved pulmonary and clinical outcome inrespiratory distress syndrome. Pediatrics1996; 98(6 Pt 1):1044–1057

27. Ogawa Y, Miyasaka K, Kawano T, et al: Amulticenter randomized trial of high fre-quency oscillatory ventilation as comparedwith conventional mechanical ventilation inpreterm infants with respiratory failure.Early Hum Dev 1993; 32:1–10

28. Imai Y, Nakagawa S, Ito Y, et al: Compar-ison of lung protection strategies usingconventional and high-frequency oscilla-tory ventilation. J Appl Physiol 2001; 91:1836 –1844

29. Pillow JJ, Sly PD, Hantos Z, et al: Depen-dence of intrapulmonary pressure ampli-tudes on respiratory mechanics during high-frequency oscillatory ventilation in pretermlambs. Pediatr Res 2002; 52:538–544

30. Fort P, Farmer C, Westerman J, et al: High-frequency oscillatory ventilation for adult re-spiratory distress syndrome: A pilot study.Crit Care Med 1997; 25:937–947

31. Mehta S, Lapinsky SE, Hallett DC, et al:Prospective trial of high-frequency oscilla-tion in adults with acute respiratory distresssyndrome. Crit Care Med 2001; 29:1360–1369



High-frequency oscillatory ventilation: Mechanisms of gasexchange and lung mechanics

J. Jane Pillow, MBBS, FRACP, PhD

T he mechanisms governing gasflow, gas mixing, and pressuretransmission during high-frequency oscillatory ventila-

tion (HFOV) are fundamentally differentto ventilation at more conventional respi-ratory breathing frequencies. They are in-tegrally related to each other through theprevailing mechanical characteristics ofthe intubated respiratory system and fre-quency, symmetry, and magnitude of theapplied pressure waveform. An apprecia-tion of these mechanisms, and their im-plications for gas mixing efficiency, andthe appropriate selection and matching ofventilator settings to the mechanicalproperties of the intubated respiratorysystem represent essential knowledgefoundations for the clinician who usesHFOV to treat respiratory disease in the

adult. This article aims to summarize theknown mechanisms of gas mixing and todiscuss the impact of lung mechanics onpressure transmission, flow generation,and the efficiency of ventilation and howeach of these interact with each otherduring HFOV. A brief overview of thepractical difficulties and progressachieved to date in measuring lung me-chanics during HFOV is given and therelevance of measuring lung mechanicsto optimization of clinical application ofHFOV in acute respiratory distress syn-drome (ARDS) is reviewed.

MECHANISMS OF GASTRANSPORT AND GASEXCHANGE

One of the fundamental principles un-derlying the increased efficiency of HFOVis the altered dynamics of gas flow distri-bution (1), challenging the traditionalconcepts of gas transport during conven-tional ventilation. A number of differentmechanisms have now been identified ashaving a contributory role in promotinggas exchange during HFOV, includingbulk convection, asymmetric velocityprofiles, pendelluft, cardiogenic mixing,

Taylor dispersion and turbulence, molec-ular diffusion, and collateral ventilation.These have been well illustrated previ-ously (2) (see also Fig. 1) and are outlinedsubsequently. It is likely that they are notmutually exclusive and that a combina-tion of the mechanisms augments gastransport during HFOV (3, 4).

Unlike conventional ventilation, bulkconvection plays a relatively small role ingas transport during HFOV, although it islikely to contribute significantly to venti-latory exchange in the most proximal gasexchange units. In an anesthetized dogmodel, Spahn and colleagues (5) showedthat decreasing delivered volume to alevel below the HFO–circuit-related re-breathing volume (as opposed to thelarger anatomic dead space) causes a sud-den rise in the PaCO2. Their findings sug-gest that the efficiency of CO2 eliminationis critically dependent on the net oscilla-tory volume and that bulk convection hasan essential role during HFOV. Turbu-lence in the large airways may also en-hance gas mixing (2).

Asymmetric velocity profiles result innet convective transport of material. Al-though the more central particles arepropelled down the length of the airway,

From the Institute for Child Health Research, andthe School of Women’s and Infants’ Health, Universityof Western Australia, Subiaco, Perth, Australia.

Dr. Pillow is supported by an NHMRC Neil HamiltonFairley Postdoctoral Research Fellowship and receivesresearch funding support from the Women’s and In-fants’ Research Foundation.


DOI: 10.1097/01.CCM.0000155789.52984.B7

Objective: Overview of the mechanisms governing gas trans-port, mechanical factors influencing the transmission of pressureand flow to the lung, and the measurement of lung mechanicsduring high-frequency oscillatory ventilation (HFOV) in acute re-spiratory distress syndrome.

Data Sources and Study Selection: Studies indexed in PubMedillustrating key concepts relevant to the manuscript objectives.Pressure transmission during HFOV in the adult lung was simu-lated using a published theoretical model.

Data Synthesis: Gas transport during HFOV is complex andinvolves a range of different mechanisms, including bulk convec-tion, turbulence, asymmetric velocity profiles, pendelluft, cardio-genic mixing, laminar flow with Taylor dispersion, collateral ven-tilation, and molecular diffusion. Except for molecular diffusion,each mechanism involves generation of convective fluid motion,and is influenced by the mechanical characteristics of the intu-

bated respiratory system and the ventilatory settings. These fac-tors have important consequences for the damping of the oscil-latory pressure waveform and the drop in mean pressure from theairway opening to the lung. New techniques enabling partitioningof airway and tissue properties are being developed for measure-ment of lung mechanics during HFOV.

Conclusions: Awareness of the different mechanisms govern-ing gas transport and the prevailing lung mechanics during HFOVrepresents essential background for the physician planning to usethis mode of ventilation in the adult patient. Monitoring of lungvolume, respiratory mechanics, and ventilation homogeneity mayfacilitate individual optimization of HFOV ventilatory settings inthe future. (Crit Care Med 2005; 33[Suppl.]:S135–S141)

KEY WORDS: resistance; inertance; compliance; oscillometry; gastransport; high-frequency ventilation; barotrauma; tracheal tube;respiratory function


the peripheral particles diffuse radially,promoting axial gas exchange with ex-pired alveolar gas (6). This phenomenonis particularly evident at the airway bifur-cations where there is skewing of theinspiratory profile compared with a moresymmetric expiratory velocity profile (7).The airway bifurcation phenomenonstreams fresh gas toward the alveoli alongthe inner airway walls while “alveolar”gas is streamed away from the alveolialong the outer wall, and hence plays animportant role in the longitudinal con-vective transport mechanisms duringHFOV (8).

Taylor proposed that the longitudinaldispersion of tracer molecules in a diffu-sive process is augmented by radial trans-port mechanisms when laminar flow isapplied in both the absence (6) or pres-

ence of turbulent eddies and secondaryswirling motions (9). Fredberg (10) sub-sequently used a semiempiric analysis topredict that the combination of Taylordispersion and molecular diffusion (aug-mented dispersion) accounts for almostall gas transport during HFOV.

Time-constant inequalities and phaselags between lung regions may set upbulk convective currents recirculating airbetween neighboring lung units (11–13).In vitro (14) and computational (15) lungmodels have shown that gas exchangeduring HFV may be markedly improvedby the interaction of flow between asyn-chronous neighboring airways and hasbeen graphically illustrated with strobo-scopic filming techniques (2, 13). Asym-metries in inertance and compliance ofperipheral airways and lung units are

more important determinants of pendel-luft than are asymmetries in resistance(16).

The superimposition of the rhythmic,strong contractions of the heart may fur-ther promote peripheral gas mixing bypromoting the generation of flow withinneighboring parenchymal regions ratherthan at the airway opening (17). The con-tribution of cardiogenic oscillation dur-ing HFOV has not been quantified, al-though it has been suggested thatcardiogenic mixing may account for up tohalf of the oxygen uptake in the presenceof totally apneic respiration (18). Collat-eral ventilation occurring through non-airway connections between neighboringalveoli has also been proposed as an ad-ditional mechanism of gas transport dur-ing both conventional and HFOV. The

Figure 1. Gas transport mechanisms and pressure damping during high-frequency oscillatory ventilation (HFOV). The major gas-transport mechanismsoperating during HFOV in convection, convection–diffusion, and diffusion zones include: turbulence, bulk convection (direct ventilation of close alveoli),asymmetric inspiratory and expiratory velocity profiles, pendelluft, cardiogenic mixing, laminar flow with Taylor dispersion, collateral ventilation betweenneighboring alveoli, and molecular diffusion (see text for details). The extent to which the oscillatory pressure waveform is damped is influenced by themechanical characteristics of the respiratory system. Atelectatic alveoli will experience higher oscillatory pressures than normally aerated alveoli, whereasincreased peripheral resistance increases the oscillatory pressures transmitted to proximal airways and neighboring alveolar units (adapted with permission(2), © 2005 Massachusetts Medical Society).


relatively high resistance of the collateralchannels to gas flow is likely to limit theextent to which this mechanism contrib-utes to gas mixing during HFOV (19).

Spontaneous mixing of gas particlesarising from Brownian motion contrib-utes to the diffusion of gases in the respi-ratory tract. Gas velocities approximatezero in the alveolar region as a result ofthe very high total cross-sectional area.The dominant mechanism for gas mixingin this zone is molecular diffusion, withnet transport of gas best described byFick’s law (10).

Studies in both theoretical models(10) and in healthy animals and humans(17, 20, 21) have demonstrated that tidalvolume (VT) has a greater effect on gasexchange than frequency (f) duringHFOV. As such, ventilation efficiencyduring HFOV (Q) may be expressed as:

Q � f aVTb [1]

where b � a. The values for a and b in thisequation approximate 1 and 2, respec-tively, although the absolute values maybe influenced by other factors such as theshape and complexity of the oscillatorypressure waveform. The more dominantcontribution of VT to ventilation duringHFOV is the result of the oscillatory re-distribution of gas from central to distalregions where molecular diffusion over-comes Taylor dispersion as the principalinfluence on gas transport (22). The tran-sition frequency (frequency marking thetransition from conventional to high-frequency gas transport mechanisms)varies in proportion to the ratio of met-abolic rate to dead space and hence islower in large animal species comparedwith small animal species and will belower in adults compared with neonates(23). It has been shown that HFV gastransport mechanisms come into play,whereas VT still exceeds airway deadspace volume (VD) (24) and that the tran-sition frequency occurs when alveolarventilation/frequency � 20% of VD and VT

� 1.20 VD (23).

MECHANICAL FACTORSINFLUENCING PRESSURE,FLOW, AND VENTILATIONDURING HIGH-FREQUENCYOSCILLATORY VENTILATION

With the exception of molecular diffu-sion, all of these mechanisms of gastransport in the airways during HFOV aredependent on convective fluid motion

(25). Whereas gas transport is driven byconvective fluid motion, the mechanicsof the respiratory system are also an im-portant consideration as convective fluidmotion is driven by the pressure differ-ences imposed by the chest wall or theventilator (25). In this respect, the im-pedance of the combined ventilator, cir-cuit, tracheal tube, and respiratory sys-tem is an important determinant of theefficiency of ventilation during HFOV.Impedance is a global term that encom-passes the mechanical properties of elas-tance (1/compliance), resistance, and theinertance. Although inertance is essen-tially negligible at conventional breath-ing frequencies, it assumes a muchgreater role at higher frequencies andcannot be ignored during HFOV.

In relatively simple terms, impedancerepresents a mechanical barrier to flowand as impedance increases, higher-pressure swings are required to generatean equivalent flow (and hence also vol-ume delivery to the gas exchanging com-ponent of the lung). As pressure differ-ences that drive flow also distend tissues,one of the major goals of HFV ventilationstrategies is the achievement of adequategas transport with low tidal volumeswhile avoiding pressures that either over-distend (causing barotrauma) or causeairway closure and alveolar collapse (at-electrauma). An understanding of how

the distribution of impedance along thecombined ventilated respiratory systemimpacts on the transmission of the meanand the amplitude of oscillatory pressureswings and flow to various lung compart-ments is vital to the appropriate clinicalapplication of this ventilatory technology.

Ventilator and CircuitConsiderations

Inspiratory–Expiratory Ratio. The se-lection of inspiratory to expiratory timeratio (TI:TE) influences the delivery ofpressure and volume to the lung. Tradi-tionally, clinicians used asymmetric TI:TE

ratios (inspiration shorter than expira-tion) to avoid the development of gastrapping during HFOV. Early studies(26–29) indicating mean pressures in thelung higher than those recorded at theairway opening used low mean airwaypressures that may have precipitated thedevelopment of choke points (30). Usingan optimal volume strategy, airways aresplinted open (31), and providing the in-spiratory and expiratory cycles are ofequal duration and expiration is activerather than passive, there is negligiblechange in mean pressure between theairway opening and the lung (32–34). Incontrast, the use of a TI � 33% actuallyresults in a drop in the mean intrapul-monary pressure as a result of higherflow-dependent tracheal tube (TT) resis-

Figure 2. Effect of frequency and tracheal tubediameter on mean pressure drop. Simulationswere performed using a previously published the-oretical computer model (46) using mechanicalinput parameters that reflect current use of high-frequency oscillatory ventilation in the adult pop-ulation (see Table 1). A square-wave oscillatorypressure waveform with 33% TI was used to drivethe model, including a range of airway openingamplitudes (30–90 cm H2O) at A, three differentfrequencies (3, 5, and 7 Hz) and B, for threetracheal tubes of differing internal diameter (ID)(6.5 mm, 7.5 mm, and 9.0 mm). Ventilator andmodel lung characteristics used for the simula-tions are summarized in Table 1. For any giventidal volume (VT), the magnitude of the differ-ence in mean pressure between the alveolus andthe airway opening (Palv–Pao) increases with in-creasing frequency and decreasing TT internaldiameter.

Table 1. Model parameters for simulation of pres-sure transmission in adult lung during high-frequency oscillation ventilation

ParameterBaseline value

(range)

Frequency, Hz 5 (2–8)T1, % 33Amplitude, cm H2O 70 (30–90)K1, cm H2O � s/La 1.5 (1, 2.4)K2, cm H2O � s2/L2a 3.4 (1.5, 6.8)I, cm H2O � s2/Lb 0.1 (0.133, 0.07)Cg, mL/kg 0.2Rc, cm H2O � s/L 0.2Rptot, cm H2O � s/L 0.5 (0.25–5)Ctot, mL/kg 40 (20–300)

Cg, gas compression in airways; Rc, linearresistance in central airway compartment; Rptot,total resistance of peripheral compartment; Ctot,total compliance of peripheral compartment.

aRohrer tracheal tube constants estimated byextrapolation from published values for smallertubes. Baseline value representative of 7.5 mm IDtracheal tube with values for 6.5 mm and 9.0 mmID tracheal tubes given in brackets; bvalues inbrackets show those in simulations for 6.5 mmand 9.0 mm tracheal tubes, respectively.


tance during inspiration compared withexpiration resulting from higher flowsduring the shortened inspiratory phase(32). The magnitude of the pressure dropincreases with increasing frequency, anddecreasing TT internal diameter (see Fig.2) and relative duration of the inspiratorycomponent (32), and in part explains theneed to increase mean pressure abovethat used with ventilation at more con-ventional rates when initiating HFOVwith TI � 33% of total cycle time. It haspreviously been proposed that the use ofasymmetric TI:TE may exaggerate thenormal asymmetry of inspiratory and ex-piratory velocity profiles as a result of theproportionately higher inspiratory flows,further enhancing the efficiency of gasmixing; however, several studies (35–37)have failed to demonstrate a specific ad-vantage of asymmetric flows on gas mix-ing efficiency. The consequences of TI:TE

for gas mixing or the higher inspiratoryflows associated with asymmetric TI:TE

on induction of shear stress injury hasnot yet been adequately investigated.

Waveform. Ventilators differ in theshape of the pressure waveform deliveredto the airway opening. Sinusoidal wave-forms are normally delivered with equalinspiratory and expiratory cycle durationsand thus have a single dominant fre-quency. More complex waveforms such asthe near-square waves delivered by theSensorMedics 3100B (SensorMedics,Yorba Linda, CA) are by definition com-posed of a fundamental frequency andmultiple higher harmonic frequencies.When asymmetric TI:TE ratios are used,the fundamental frequency (and its cor-

responding harmonics) will be differentduring inspiratory and expiratory oscilla-tory cycles. A comparison of gas-mixingefficiency between sinusoidal and com-plex waveforms has not yet been per-formed; however, it will be difficult to doso in a way that equivalent flow wave-forms are produced if TI:TE ratios otherthan 1:1 are used. Ventilators determinethe TI:TE based on the relative durationsof the respective components of the pres-sure waveform. For the square wave, thisestimate is a reasonably close approxima-tion of inspiratory and expiratory flowcycles, which more accurately define thestart of the inspiratory and expiratory pe-riods. In contrast, a sinusoidal airwayopening pressure waveform constructedwith a 33% TI actually generates aninspiratory flow that accounts for ap-proximately 42% of the total cycle time(unpublished observations). As a conse-quence, the magnitude of the pressuredrop across the TT is less for a ventila-tor delivering sinusoidal pressure wave-forms than for a square waveform ven-tilator (32). Square waveforms areassociated with sudden changes in flowand airway pressures, whereas moregradual and smooth changes are ob-served using sine wave ventilation.There has been a suggestion that squarewave ventilation at conventionalbreathing frequencies is associated withincreased incidence of air leak syn-drome (38); however, the impact ofwaveform shape and the rapidity ofchange in flow and pressure for shearstress and lung injury during HFOV hasnot been assessed.

Frequency

Higher frequencies result in deliveryof lower tidal volumes and also decreasethe magnitude of the alveolar pressureswings (20, 39). What constitutes optimalfrequency in HFOV is an issue often dis-cussed by clinicians. In newborns, fre-quencies between 8 and 15 Hz are mostoften used, whereas lower frequencies aremore often used for adults (40, 41). Giventhe relative importance of VT in deter-mining CO2 exchange (VCO2 � f·VT

2) (10,20, 21, 42), the goal of frequency selec-tion needs to minimize the pressureswings to both proximal and distal lungcompartments while not compromisingthe VT to the extent that insufficient ven-tilation takes place. From a mechanicalpoint of view, in most adult clinical sce-narios, this is likely to occur slightly be-low the resonant frequency near the cor-ner frequency (Fc) of the lung (seesubsequently) (25).

Tracheal Tube. The TT contributesover 50% of the total impedance of therespiratory system and accounts for ap-proximately 90% of the inertance (43).During HFOV, the resistance of the tra-cheal tube is flow-dependent and can bedescribed using Rohrer constants (32,44). As resistance is inversely propor-tional to r4 (where r � radius), smallreductions in the internal diameter of theTT (i.e., from secretions or change in TTsize) decreases (damps), the amplitude ofthe pressure waveform, and reduces re-sultant flow and VT. A decrease in TTinternal diameter also increases the mag-nitude of the drop in mean pressure be-tween the airway opening and the lungparenchyma for any given ventilator am-plitude (45, 46). Whereas these effects arequite marked in narrow neonatal tubes(46), appreciable changes in these param-eters can still be observed in the range ofTT used in the adult population as a re-sult of the significantly higher flows.

Patient Factors

Whereas the traditional teaching ap-proach for HFOV has emphasized exten-sive damping of the pressure waveformbetween the airway opening and the al-veolar compartment (11, 47, 48), it needsto be appreciated that this knowledge waslargely based on measurements made inmodels with highly compliant lungs. Incontrast, theoretical (45, 46), in vitro(49), and in vivo (50, 51) models of lungdisease have demonstrated that damping

Figure 3. Effect of compliance and peripheral resistance on damping of the pressure waveform.Simulations were performed across a range of A, total lung compliance (10–300 cm H2O) and B, totalperipheral resistance (0.5–10 cm H2O s/L) at a frequency of 5 Hz, 33% TI for a 7.5-mm ID TT. Thereis a sharp rise in the magnitude of the oscillatory pressure waveform transmitted to both alveolar andtracheal compartments at low compliance. The amplitude of the tracheal oscillatory pressure wave-form increases with increasing peripheral resistance.


at different points within the airways andalveoli is heavily dependent on the distri-bution, homogeneity, and mechanicalcharacteristics of disease within the re-spiratory system.

HFOV was initially developed for usein the extremely noncompliant and im-mature lung of the premature neonatewith hyaline membrane disease, which istypified by relatively diffuse homoge-neous atelectasis associated with surfac-tant deficiency. The magnitude of pres-sure amplitudes transmitted to the lungin the newborn lung increases exponen-tially with decreasing compliance (46, 49,50). In contrast to the neonate, the mostusual application of HFOV in the adult isfor ARDS. Like hyaline membrane dis-ease, ARDS is characterized mechanicallyby a very poorly compliant lung, usuallyin the absence of altered resistance. Incontrast to hyaline membrane disease,however, the pattern of lung involvementin ARDS may be relatively inhomoge-neous (52, 53), and the cause of the ill-ness may be the result of pulmonary orextrapulmonary causes (54, 55). Modifi-cation of a theoretical lung model usingappropriate ventilator settings (41), tra-cheal tube constants, and respiratory me-chanical parameters (56) for an intubatedadult with ARDS demonstrates the rela-tively higher pressure transmission to theairways and alveolar compartment in thepresence of poor compliance (see Fig.3A).

Ventilation inhomogeneity poses aproblem at conventional ventilation fre-quencies because the distribution of gasis largely controlled by the distribution ofregional lung compliance, and thus het-erogeneous regional expansion and ven-tilation necessarily follow. As ventilationfrequencies approach the resonance fre-quency, however, gas transport is less de-pendent on regional lung compliance(57) and increasingly governed by the re-sistive (26), inertive (48), and branchingangle properties of the central airways(58, 59). Theoretical studies have indi-cated that compliant alveoli are effec-tively spared from excessive oscillatorypressures with the larger alveolar pres-sure swings being directed to the morepoorly compliant compartments (46).

HFOV is also applied in clinical sce-narios other than the purely poorly com-pliant lung. Understanding the impact ofrespiratory mechanics on pressure andflow may help elucidate why such ap-proaches have so far met with variablesuccess. An increase in peripheral airway

resistance (as might be expected to occurif using HFOV in a patient with smallairways disease) will result in a markedincrease in the pressure swings in theairways proximal to the obstruction, de-spite the preservation of relatively smallpressure and volume fluctuations deliv-ered to the alveolar compartment (seeFig. 3B) (46). In such cases, tuning ven-tilator frequency to the corner frequencyof the affected lungs will reduce the like-lihood of excessive pressure excursions inthe proximal airways and limit associatedshear stress (25).

PRESSURE COST OFVENTILATION

Although each of the factors outlinedhere influence the magnitude of pressuretransmission from the airway opening tothe lung, they also have consequences forthe delivery of flow and the magnitude ofresultant tidal volumes. A particularly in-structive way to approach the complexi-ties of these different interactions is tofollow the approach of Venegas and Fred-berg (25) by considering the pressurecost of achieving flow, which can be con-sidered as the pressure cost per unit con-vective flow times the convective flowcost per unit alveolar ventilation. Theirseminal 1994 paper highlighted some im-portant principles for the clinical practiceof HFOV and warrants special attention.The oscillatory pressure cost of flow de-creases rapidly with increasing fre-quency, reaching a minimum at the res-onance frequency:

f0 �1

2��IC[2]

where f0 is the resonance frequency, C isthe compliance, and I is the inertance. Inoverdamped lungs, minimal additionaldamping of the oscillatory pressure isachieved above the corner frequency:

fc �1

2�RC[3]

where fc is the corner frequency and R isthe resistance, suggesting that there isbenefit in increasing ventilation fre-quency up to but not beyond this point.As fc increases, a beneficial decrease inoscillatory pressure continues to beachieved at higher frequencies thanwould be the case in the normal lung.Equally, increasing system resistance willshift the fc to lower frequencies. This con-cept may explain why neonatal conditions

such as meconium aspiration syndromeare most successfully ventilated usinglower frequencies than those used for hy-aline membrane disease.

The adult lung is a relatively over-damped system compared with the neo-natal lung. Using published values forresistance, inertance, and compliance(60), a healthy adult lung would have f0�2.7 Hz and fc �0.32 Hz. Assuming lungcompliance of a patient with ARDS isdecreased to 1/10th the normal value(i.e., approximately 20 mL/cm H2O) (52),f0, and fc for the adult ARDS lung wouldapproximate 8.6 Hz and 3.2 Hz, respec-tively. Whereas the absolute values ofthese parameters will ultimately dependon the relative changes in inertance andresistance, this rough estimation sug-gests values for fc in keeping with currentclinical practice of HFOV in ARDS.

A second and perhaps more importantfactor is the impact of inappropriate pos-itive end-expiratory pressure (PEEP) se-lection. In the healthy lung, pressure costof ventilation increases markedly at highPEEP values, and to a lesser extent alsowhen alveolar recruitment occurs as aresult of low PEEP (25). Venegas andFredberg show that in the poorly compli-ant lung, the selection of PEEP is muchmore critical than the frequency, at leastfor frequencies above the resonant fre-quency of the lung, although frequencychoice remains more important than inthe normal lung. In contrast, pressurecost of ventilation increases dramaticallyabove the corner frequency in the situa-tion in which increased airway resistanceis a dominant feature of disease (25).

MONITORING LUNGMECHANICS AND VENTILATIONINHOMOGENEITY TO GUIDEVENTILATORY PRACTICE

The important interplay between re-spiratory mechanics and the selection ofappropriate ventilation frequency and se-lection of mean airway pressure high-lights the urgent need for bedside assess-ment of lung volume, distribution ofventilation, respiratory mechanics, andoptimal ventilation frequency. Major ad-vances in the noninvasive assessment oflung volume and regional inhomogeneityhave been achieved over the last decade;however, these are the subject of twofurther articles in this supplement (chap-ters 7 and 8) and are not discussed fur-ther here. Progress is also being made inobtaining meaningful measurements of


lung mechanics during HFOV and war-rants some discussion.

Ideally, measurement of lung functionduring HFOV would: 1) provide a com-prehensive bedside assessment of lungfunction across a range of oscillatory fre-quencies during a limited time interval;2) be achievable at (and without compro-mising) the prevailing lung volume; 3)generate detailed information about thefunction and integrity of the respiratorysystem with particular reference to lungvolume and mechanics of the lung paren-chyma; 4) provide measurement parame-ters that are comparable with those ofpatients ventilated at other (lower) fre-quencies; and 5) be applicable to the mea-surement of lung mechanics in both theacute and longer-term follow up of indi-vidual patients.

Until recently, estimation of respira-tory mechanics during HFOV has mostoften been obtained using conventionalpassive mechanics, although this ap-proach provides limited informationabout the lung parenchyma (often theprimary site of pathology in the poorlycompliant lung). In addition, the defla-tion maneuvers associated with suchmeasurements may promote alveolar de-recruitment and have not been founduseful in determining optimum lung vol-ume during HFOV (61). Likewise, themeasurement of dynamic mechanics isalso plagued by problems. Accurate mea-surement of the high instantaneous flowsis difficult to achieve, as are meaningfulmeasurements of dynamic compliance.Compliance is determined by assessingchanges in volume-per-unit change inpressure, but tidal volume becomes in-creasingly independent of changes incompliance with increasing frequency(49). As a result, detecting change in un-derlying lung compliance during HFOVwould require monitoring of intrapulmo-nary pressures. Because the amplitude ofthe oscillatory pressure waveform at theairway opening does not reflect changesin the intrapulmonary pressure ampli-tudes during HFOV, a more invasive ap-proach is required. Using a catheter tipmanometer, Van Genderingen and col-leagues (51) have demonstrated an in-verse relationship between respiratorycompliance and the oscillatory pressureratio (ratio of pressure swings at the dis-tal and proximal ends of an endotrachealtube) and were able to use these measure-ments to define the optimal continuousdistending pressure.

Sipinková et al. (62) obtained mea-surements of oscillatory mechanics dur-ing short bursts of HFOV across a rangeof tidal volumes (5.0, 6.6, and 10 mL) andfrequencies (10, 15, 20, and 25 Hz) inrabbits and demonstrated changes in os-cillatory mechanics from the pressure-flow relationship at the airway openingbefore and after vagotomy in adult rabbitsat each of a range of oscillation frequen-cies and tidal volumes. Drawbacks of thisapproach are the need for different mea-surements at each frequency (potentiallyinducing time-variant results) and therelatively high amplitude of oscillatorysignals required (which may introducenonlinearities).

More recently, the low-frequencyforced oscillation technique (FOT), whichuses a broadband low-amplitude forcingsignal has been used to monitor changesin lung mechanics associated withchanges in lung volume (46, 50) duringHFOV. The low-frequency FOT can meeteach of the criteria listed here, and ani-mal ventilators have been built incorpo-rating similar measurements inter-spersed with normal ventilation at user-defined frequencies. A particularadvantage of this approach is that it facil-itates partitioning of respiratory mechan-ics into airway and parenchymal compo-nents (63). The inclusion of esophagealpressure catheters also enables chest wallimpedance to be measured, a factor thatmay be quite important for ARDS andhave implications for the successful ap-plication of HFOV (54). Gattinoni andcoworkers highlighted the improved re-sponses of patients with chest-wall stiff-ness to increased application of PEEPcompared with those in whom the pri-mary illness effected diminished compli-ance of the lungs (52). Practical limita-tions for clinical introduction of the low-frequency FOT will include the need formeasurements during a relaxed respira-tory pause and the success in achievingleak-free measurements in the presenceof tracheal tube leak by application ofpressure over the carina.

SUMMARY

Whereas some similarities can bedrawn between ventilation at conven-tional breathing rates and HFOV, theadded complexity introduced by thehigher frequencies and the accompany-ing high flows significantly alters the me-chanical balance. Changes in the relativeimportance of compliance, resistance,

and inertance at different frequencies forany given lung condition in determiningthe magnitude of flow and pressure trans-mitted to the distal lung have importantimplications for clinical practice. Clini-cians need to understand the concept ofdrop in mean pressure across the trachealtube, and appreciate how this may bealtered by frequency, internal diameter ofthe tracheal tube, and ventilator ampli-tude. Likewise, an appreciation of lungmechanics in the clinical setting mayguide selection of those patients likely tobenefit from HFOV and appropriate ven-tilator settings (especially frequency) toachieve low tidal volume ventilationwhile avoiding barotrauma to proximalalveolar units. Further development oftools to monitor lung volume, regionalventilation inhomogeneity, and lung me-chanics would establish an objective basisfor optimizing the clinical application ofHFOV in the setting of ARDS.

REFERENCES

1. Slutsky AS, Drazen FM, Ingram RH Jr, et al:Effective pulmonary ventilation with small-volume oscillations at high frequency. Sci-ence 1980; 209:609–671

2. Slutsky AS, Drazen JM: Ventilation withsmall tidal volumes. N Engl J Med 2002;347:630–631


4. Slutsky AS: Mechanisms affecting gas trans-port during high-frequency oscillation. CritCare Med 1984; 12:713–717

5. Spahn DR, Leuthold R, Schmid ER, et al:Significance of bulk convection during high-frequency oscillation. Respir Physiol 1991;84:1–11

6. Taylor GI: Diffusion and mass transport intubes. Proc Roy Soc A 1953; 216:186

7. Haselton FR, Scherer PW: Bronchial bifurca-tions and respiratory mass transport. Science1980; 208:69–71

8. Scherer PW, Haselton FR: Convective ex-change in oscillatory flow through bronchi-al-tree models. J Appl Physiol 1982; 53:1023–1033

9. Taylor GI: The dispersion of matter in tur-bulent flow through a pipe. Proc Roy Soc A1954; 223:446–448

10. Fredberg JJ: Augmented diffusion in the air-ways can support pulmonary gas exchange.J Appl Physiol 1980; 49:232–238

11. Allen JL, Fredberg JJ, Keefe DH, et al: Alve-olar pressure magnitude and asynchronyduring high-frequency oscillations of excisedrabbit lungs. Am Rev Respir Dis 1985; 132:343–349

12. Fredberg JJ, Keefe DH, Glass GM, et al: Al-veolar pressure nonhomogeneity during


small-amplitude high-frequency oscillation.J Appl Physiol 1984; 57:788–800

13. Lehr JL, Butler JP, Westerman PA, et al:Photographic measurement of pleural sur-face motion during lung oscillation. J ApplPhysiol 1985; 59:623–633

14. Ultman JS, Shaw RG, Fabiano DC, et al:Pendelluft and mixing in a single bifurcationlung model during high-frequency oscilla-tion. J Appl Physiol 1988; 65:146–155

15. Tanida Y: Analysis of gas transport in high-frequency ventilation. Front Med Biol Eng1990; 2:181–185

16. High KC, Ultman JS, Karl SR: Mechanicallyinduced pendelluft flow in a model airwaybifurcation during high frequency oscilla-tion. J Biomech Eng 1991; 113:342–347

17. Slutsky AS: Gas mixing by cardiogenic oscil-lations: a theoretical quantitative analysis.J Appl Physiol 1981; 51:1287–1293

18. Slutsky AS, Brown R: Cardiogenic oscilla-tions: a potential mechanism enhancing ox-ygenation during apneic respiration. MedHypotheses 1982; 8:393–400

19. Armengol J, Jones RL, King EG: Collateralventilation during high-frequency oscillationin dogs. J Appl Physiol 1985; 58:173–179

20. Rossing TH, Slutsky AS, Lehr JL, et al: Tidalvolume and frequency dependence of carbondioxide elimination by high-frequency venti-lation. N Engl J Med 1981; 305:1375–1379

21. Jaeger MJ, Kurzweg UH, Banner MJ: Trans-port of gases in high-frequency ventilation.Crit Care Med 1984; 12:708–710

22. Kamm RD, Collins J, Whang J, et al: Gastransport during oscillatory flow in a net-work of branching tubes. J Biomech Eng1984; 106:315–320

23. Venegas JG, Hales CA, Strieder DJ: A generaldimensionless equation of gas transport byhigh-frequency ventilation. J Appl Physiol1986; 60:1025–1030

24. Weinmann GG, Mitzner W, Permutt S: Physio-logical dead space during high-frequency ventila-tion in dogs. J Appl Physiol 1984; 57:881–887

25. Venegas JG, Fredberg JJ: Understanding thepressure cost of ventilation: why does high-frequency ventilation work? Crit Care Med1994; 22:S49–57

26. Simon BA, Weinmann GG, Mitzner W: Meanairway pressure and alveolar pressure duringhigh-frequency ventilation. J Appl Physiol1984; 57:1069–1078

27. Saari AF, Rossing TH, Solway J, et al: Lunginflation during high-frequency ventilation.Am Rev Respir Dis 1984; 129:333–336

28. Gerstmann DR, Fouke JM, Winter DC, et al:Proximal, tracheal, and alveolar pressures dur-ing high-frequency oscillatory ventilation in anormal rabbit model. Pediatr Res 1990; 28:367–373

29. Solway J, Rossing TH, Saari AF, et al: Expira-tory flow limitation and dynamic pulmonaryhyperinflation during high-frequency ventila-tion. J Appl Physiol 1986; 60:2071–2078

30. Bryan AC, Slutsky AS: Long volume duringhigh frequency oscillation. Am Rev RespirDis 1986; 133:928–930

31. Lai-Fook SJ, Hyatt RE, Rodarte JR: Effect ofparenchymal shear modulus and lung vol-ume on bronchial pressure-diameter behav-ior. J Appl Physiol 1978; 44:859–868

32. Pillow JJ, Neil H, Wilkinson MH, et al: Effectof I/E ratio on mean alveolar pressure duringhigh-frequency oscillatory ventilation. J ApplPhysiol 1999; 87:407–414

33. Thome U, Pohlandt F: Effect of the TI/TEratio on mean intratracheal pressure in high-frequency oscillatory ventilation. J ApplPhysiol 1998; 84:1520–1527

34. Hatcher D, Watanabe H, Ashbury T, et al: Me-chanical performance of clinically available,neonatal, high-frequency, oscillatory-type ven-tilators. Crit Care Med 1998; 26:1081–1088

35. Yamada Y, Hales CA, Venegas JG: Inspirato-ry-to-expiratory time ratio and alveolar ven-tilation during high-frequency ventilation indogs. J Appl Physiol 1986; 61:1903–1907

36. Venegas JG, Yamada Y, Custer J, et al: Effectsof respiratory variables on regional gas trans-port during high-frequency ventilation.J Appl Physiol 1988; 64:2108–2118

37. Ghazanshahi SD, Marmarelis VZ, YamashiroSM: Analysis of the gas exchange system dy-namics during high-frequency ventilation.Ann Biomed Eng 1986; 14:525–542

38. Sly PD, Drew JH: Air leak in neonatal respi-ratory distress syndrome. Anaesth IntensiveCare 1984; 12:41–45

39. Slutsky AS, Kamm RD, Rossing TH, et al:Effects of frequency, tidal volume, and lungvolume on CO2 elimination in dogs by highfrequency (2–30 Hz), low tidal volume ven-tilation. J Clin Invest 1981; 68:1475–1484

40. Ritacca FV, Stewart TE: Clinical review:high-frequency oscillatory ventilation inadults—a review of the literature and prac-tical applications. Crit Care 2003; 7:385–390

41. Derdak S: High-frequency oscillatory venti-lation for acute respiratory distress syn-drome in adult patients. Crit Care Med 2003;31:S317–323

42. Venegas JG, Custer J, Kamm RD, et al: Rela-tionship for gas transport during high-frequency ventilation in dogs. J Appl Physiol1985; 59:1539–1547

43. Dorkin HL, Stark AR, Werthammer JW, et al:Respiratory system impedance from 4 to 40 Hzin paralyzed intubated infants with respiratorydisease. J Clin Invest 1983; 72:903–910

44. Gavriely N, Solway J, Loring SH, et al: Pres-sure-flow relationships of endotracheal tubesduring high-frequency ventilation. J ApplPhysiol 1985; 59:3–11

45. van Genderingen HR, Versprille A, Leen-hoven T, et al: Reduction of oscillatory pres-sure along the endotracheal tube is indica-tive for maximal respiratory complianceduring high-frequency oscillatory ventila-tion: a mathematical model study. PediatrPulmonol 2001; 31:458–463


47. Frantz ID III, Close RH: Alveolar pressureswings during high frequency ventilation inrabbits. Pediatr Res 1985; 19:162–166

48. Allen JL, Frantz ID III, Fredberg JJ: Regionalalveolar pressure during periodic flow. Dualmanifestations of gas inertia. J Clin Invest1985; 76:620–629

49. Pillow JJ, Wilkinson MH, Neil HL, et al: Invitro performance characteristics of high-frequency oscillatory ventilators. Am J RespirCrit Care Med 2001; 164:1019–1024

50. Pillow JJ, Sly PD, Hantos Z: Monitoring of lungvolume recruitment and derecruitment usingoscillatory mechanics during high-frequencyoscillatory ventilation in the preterm lamb. Pe-diatr Crit Care Med 2004; 5:172–180

51. van Genderingen HR, van Vught AJ, DuvalEL, et al: Attenuation of pressure swingsalong the endotracheal tube is indicative ofoptimal distending pressure during high-frequency oscillatory ventilation in a modelof acute lung injury. Pediatr Pulmonol 2002;33:429–436

52. Gattinoni L, Bombino M, Pelosi P, et al: Lungstructure and function in different stages ofsevere adult respiratory distress syndrome.JAMA 1994; 271:1772–1779

53. Piantadosi CA, Schwartz DA: The acute re-spiratory distress syndrome. Ann Intern Med2004; 141:460–470

54. Hall JB: Respiratory system mechanics inadult respiratory distress syndrome. Stretch-ing our understanding. Am J Respir CritCare Med 1998; 158:1–2

55. Kallet RH, Katz JA: Respiratory system me-chanics in acute respiratory distress syndrome.Respir Care Clin North Am 2003; 9:297–319

56. Farre R, Mancini M, Rotger M, et al: Oscilla-tory resistance measured during noninvasiveproportional assist ventilation. Am J RespirCrit Care Med 2001; 164:790–794

57. Tsuzaki K, Hales CA, Strieder DJ, et al: Re-gional lung mechanics and gas transport inlungs with inhomogeneous compliance.J Appl Physiol 1993; 75:206–216

58. Tsuda A, Kamm R, Fredberg JJ: Periodic flowat airway bifurcations. II. Flow partitioning.J Appl Physiol 1990; 69:553–561

59. Tsuda A, Fredberg JJ: Periodic flow at airwaybifurcations. I. Development of steady pressuredifferences. J Appl Physiol 1990; 69:546–552

60. Farre R, Peslin R, Rotger M, et al: Humanlung impedance from spontaneous breathingfrequencies to 32 Hz. J Appl Physiol 1994;76:1176–1183

61. Kalenga M, Battisti O, Francois A, et al:High-frequency oscillatory ventilation inneonatal RDS: initial volume optimizationand respiratory mechanics. J Appl Physiol1998; 84:1174–1177

62. Sipinkova I, Koller EA, Buess C, et al: Me-chanical respiratory system input impedanceduring high-frequency oscillatory ventilationin rabbits. Crit Care Med 1994; 22:S66–70

63. Hantos Z, Daroczy B, Suki B, et al: Inputimpedance and peripheral inhomogeneity ofdog lungs. J Appl Physiol 1992; 72:168–178


High-frequency oscillatory ventilation: Lessons learned frommechanical test lung models

Michael Van de Kieft, MD; David Dorsey, MD; David Morison, MD; Lazaro Bravo, MD;Steven Venticinque, MD; Stephen Derdak, DO

Mechanical ventilation strat-egies for acute respiratorydistress syndrome (ARDS)have been the subject of in-

tense investigation and scrutiny over thepast several years. Numerous experimen-tal studies have demonstrated that venti-lator strategies causing repeated overdis-tension and reopening of alveolar unitslikely contribute to ventilator-inducedlung injury (VILI) (1, 2) and may play arole in the development of multiple organdysfunction syndrome (MODS) (3). Thesefindings have been instrumental in for-mulating lung protective approaches us-ing both conventional ventilation (4) and

high-frequency oscillatory ventilation(HFOV) (5).

HFOV maintains an open lung volumeby application of a constant mean airwaypressure (mPaw). Phasic pressure swingsproduced by an oscillating piston allowthe use of relatively high (e.g., 25–35 cmH2O) mPaw, which, ideally, maintainslung volume and recruits alveolar unitsfor improved gas exchange while mini-mizing lung overdistension and atelec-trauma.

Observational clinical trials of HFOVfor adult ARDS have demonstrated im-proved oxygenation, comparable safety(e.g., no significant differences in baro-trauma or endotracheal tube obstruc-tion), and the absence of significant ad-verse hemodynamic effects (5–7).Although not specifically powered to eval-uate mortality differences, a recent ran-domized trial comparing HFOV withpressure control ventilation suggested atrend toward improved 30-day mortalityin the HFOV group (37% vs. 52% mor-tality, p � .1) (5).

Animal ARDS models have investi-gated inflammatory markers (1, 2) andfactors affecting tidal volume delivery(Vti) during HFOV (8). Similarly, me-chanical test lung studies have assessedVti (9), the effects of endotracheal tube

(ETT) size and ETT cuff leaks (10), andthe effects of various ventilation strate-gies on the transmission of oscillatorypressure and Vti (11). Although mechan-ical test lung models have important lim-itations (e.g., inability to measure bio-logic markers, examine histology, orassess hemodynamics and gas exchange),they can, nevertheless, provide a rapid,reproducible environment in which nu-merous variables may be manipulatedwhile also providing insights about distalairway pressure and volume changes.This chapter reviews HFOV and mechan-ical test lung data as it applies to: 1) Vti

delivery, 2) airway pressure transmission,3) effects of ETT cuff leak placement, and4) simulated clinical situations. Data ob-tained from mechanical test lung modelsmay provide insights into the interactionof HFOV with lung mechanics, and, inconjunction with animal models, maycontribute to developing clinical HFOVstrategies.

Test Lung Models and ExperimentalDesign. Previous tidal volume estimatesusing HFOV have been reported; how-ever, these investigations have generallyused small animal and infant test lungmodels and narrow-diameter cufflessETTs (e.g., 2.5–4.0-mm internal diame-ter [ID]) (8, 9, 11). Understanding the

From Pulmonary/Critical Care Medicine (MVdK, LB,SD), Wilford Hall Medical Center, Lackland AFB, TX;and Pulmonary/Critical Care Medicine (DD, DM, SV)and the Anesthesiology and Surgical Care Unit, BrookeArmy Medical Center, Ft. Sam Houston, TX.

Supported, in part, by SensorMedics Corporation,which provided use of 3100B high-frequency oscilla-tion ventilators for clinical research.

Views expressed in this article are those of theauthors and do not represent the official policy ofthe Department of Defense or other Departmentsof the U.S. Government.


DOI: 10.1097/01.CCM.0000155924.74942.7F

Objective: Review data obtained from high-frequency oscilla-tory ventilation (HFOV) and mechanical test lung models withrespect to delivered tidal volume, distal pressure transmission,endotracheal tube cuff leaks, and simulated clinical conditions.

Design: Review of selected studies from PubMed, publishedabstracts, and institutional mechanical test lung data.

Results: Tidal volume delivery during HFOV is altered by os-cillatory pressure amplitude (�P), frequency (Hz), percent inspira-tory time (IT%), and patient variables. Distal (carinal) oscillatorypressure amplitude transmission is directly correlated with en-dotracheal tube diameter and peripheral airway resistance. En-dotracheal tube cuff leaks promote egress of tracheal gas whileattenuating distal oscillatory pressure amplitude and tidal volume

transmission. Simulated clinical conditions (e.g., increased distalairway resistance, mainstem intubation) may increase observed�P, whereas mean airway pressure is decreased with air leaks.

Conclusion: Mechanical test lung and artificial trachea sim-ulations may provide useful information on the interaction ofHFOV with altered lung mechanics and may contribute to theformulation of HFOV clinical strategies. Important limitations ofthese models include absence of gas exchange, histologic andbiologic markers, or hemodynamic data. (Crit Care Med 2005;33[Suppl.]:S142–S147)

KEY WORDS: high-frequency oscillatory ventilation; test lung;tidal volume; acute respiratory distress syndrome; oscillatorypressure amplitude


method of measuring tidal volume andthe location along the airway where themeasurement was obtained (e.g., at the cir-cuit wye proximal to the ETT or at thecarina, in the presence or absence of anETT cuff leak, in the distal lung down-stream from airway resistors, or at thepleural surface in lung animal models) iscritically important in interpreting andcomparing data between different studies.The measured tidal volume during HFOVvaries depending on the inherent limita-tions and properties of the device used (e.g.,

BICORE CP-100; BICORE Monitoring Sys-tems, Irvine, CA; and Florian; AccutronicMedical Systems AG, Hirzel, Switzerland;heated pneumotachometer; Hans Rudolph,Kansas City, MO; or analog tidal volumescale of the mechanical test lung) andwhether “delivered” tidal volume (Vti) or“exhaled” tidal volume (Vte) is being mea-sured. At Wilford Hall Medical Center(WHMC), our group has used a mechanicaltest lung model comprised of the followingcomponents (Fig. 1): 1) a high-frequencyoscillation ventilator (SensorMedics 3100B;VIASYS Healthcare, Yorba Linda, CA) withadult circuit; 2) an 8.0-mm ID cuffed ETTpositioned within an artificial trachea(0.5-mm plastic tubing, 22 cm in lengthand 10 cm in outer diameter with proximaland distal gas analysis ports) connectedthrough an adaptor to a mechanical testlung (model 5600I; Michigan Instruments,Grand Rapids, MI); and 3) a laptop com-puter (Pneuview software; Michigan Instru-ments, Grand Rapids, MI) interfaced withthe test lung, which provides measure-

ments of internal test lung tidal volumesand pressure amplitudes. Lung complianceand airway resistance may easily be manip-ulated on the test lung, and the circuitallows the placement of pulmonary moni-tors (such as the BICORE CP-100 and Flo-rian) proximal and distal to the ETT cuff,various ETT sizes with or without a cuffleak, and the measurement of gas concen-trations above and below the ETT cuff (Ras-cal II analyzer; Ohmeda, Salt Lake City,UT).

Tidal Volume Delivery During High-Frequency Oscillatory Ventilation. Nu-merous ventilator and patient-relatedvariables influence Vti during HFOV. Ofthe former, oscillatory pressure ampli-tude (�P), frequency (Hz), percent in-spiratory time (IT%), ETT length and di-ameter, and the presence of a cuff leak allaffect Vti and Vte, whereas airway resis-tance and lung compliance comprise thelatter (9–12). A recent study by Sedeek etal. (8) measured delivered tidal volume bythe SensorMedics 3100B HFOV using a

Figure 1. Mechanical test lung model setup usedat Wilford Hall Medical Center. A high-frequencyoscillatory ventilator (SensorMedics 3100B) withadult circuit (not shown) is interfaced with amechanical test lung (Michigan Instrumentsmodel 5600i) through an 8.0-mm cuffed endotra-cheal tube (ETT) positioned inside an artificialtrachea. The trachea is composed of 0.5-mmthick low-compliance plastic tubing (22 cm inlength and 10 cm in outer circumference). Gasanalysis ports are positioned 6.5 and 17 cm fromthe proximal end of the trachea to allow samplingof tracheal gas above (ambient) and below (cari-nal side) the ETT cuff. A Rascal II gas analyzer(Ohmeda) is used to analyze oxygen and nitrogenconcentrations at port sites proximal and distalto the ETT cuff. An adaptor placed at the distalend of the artificial trachea facilitates connectionwith the mechanical test lung. The system en-ables placement of pulmonary monitors (BI-CORE CP-100 or Florian) at locations proximaland distal to the ETT. The mechanical test lungmodel allows manipulation of compliance (0.01–0.10 L/cm H2O) through a steel alloy springstretched between the top plate of the lung andthe test model frame, and airway resistance (5,20, and 50 cm H2O/L/sec fixed orifice airwayresistors) in the tracheal and mainstem bronchuspositions. Additionally, the mechanical test lunghouses an electronic interface module, which al-lows transfer of information from the lung to alaptop computer using Pneuview software (Mich-igan Instruments).

Figure 2. High-frequency oscillatory ventilation (HFOV) tidal volume (Vte in liters) interfaced to amechanical test lung comparing variable power (�P in cm H2O) and fixed �P algorithms. Algorithmsteps, along with corresponding �P and frequency (Hz), are denoted in the accompanying table. Datawere obtained with a SensorMedics 3100B HFOV, Michigan Instruments model 5600i test lung withan 8.0-mm internal diameter endotracheal tube (cuff fully inflated) positioned inside an artificialtrachea (Fig. 1), and Florian neonatal monitor (measures Vte) positioned at the circuit wye. HFOVsettings were as follows: mean airway pressure (mPaw) 30 cm H2O, bias flow (BF) 30 L/min, andinspiratory time (IT) 33%. Test lung compliance was set at 0.08 L/cm H2O with 5-cm H2O/L/secresistors positioned in the trachea and mainstem bronchus positions.


heated pneumotachometer (Hans Ru-dolph, Kansas City, MO) in 30-kg sheepusing a cuffed ETT (internal diameter notspecified). These investigators measureddelivered tidal volumes during HFOVranging between approximately 50 and125 mL (set parameters: mPaw 26 cmH2O, bias flow 30 L/min, �P 60 cm H20,frequency 10 – 4 Hz, and inspiration:expiration [I:E] 1:2). Interestingly, withI:E set at 1:2 (commonly used clinically),the effect of decreasing �P (e.g., from 60to 50 cm H2O) had a more pronouncedeffect on reducing tidal volume deliverythan increasing Hz from 4 to 10. In con-trast, with I:E set at 1:1, changes in Hz(e.g., from 4 to 6) had a more pronouncedeffect on reducing delivered tidal volumethan decreasing �P from 60 to 50 cmH2O. As expected, Vti was found to in-crease with decreasing Hz, increasing �P,and increased IT%. These findings mayhave potential implications for the choiceof �P:Hz combinations for ventilation,depending on which I:E is chosen.

In a preterm lamb-based computermodel (with simulated cuffless ETT IDrange 2.5–4.0 mm), increased trachealtube resistance reduced Vti, with de-creased ETT ID producing a greater de-crease in Vti compared with increasingETT length (11). Increased peripheral air-

way resistance was associated with de-creased Vti; however, larger Vti was notobserved with improved compliance(range of compliances 0.5 mL/cm H2O to2.0 mL/cm H2O).

Using settings of 3 Hz, �P 70 cm H2O,mPaw 40 cm H2O, compliance of 0.05L/cm H2O, and the WHMC test lungmodel described here, Vti and Vte (as mea-sured with the BICORE CP-100 and ver-ified with the analog scale of the testlung) varied significantly depending onIT% (10). At IT% of 50 (I:E 1:1) and 33(I:E 1:2), Vti was 61.5 mL and 36.5 mL,respectively. In contrast, exhaled tidalvolume (Vte) was 40 mL at IT 50% and76.5 mL at IT 33%. Similar to results inthe lamb-based computer model, varyingcompliance (0.05 L/cm H2O vs. 0.02 L/cmH2O) was not associated with significantVti changes.

Test lung models may also be used tostudy the effects of different HFOV venti-lation algorithms (e.g., �P:Hz combina-tions) on Vt. Figure 2 shows Vte usingfixed �P vs. varying �P algorithms usingthe WHMC test lung model and Florianmonitor (where Vte was measured at thecircuit wye using an 8-mm ID cuffedETT). The algorithm using a variable �Pdemonstrated slightly larger Vte com-pared with the fixed �P algorithm. It is

important to realize, however, that al-though Vt may be similar using two dif-ferent �P:Hz combinations, the effect onminute ventilation and gas exchange invivo may be quite different. For example,similar Vt was obtained at �P 55 cmH2O:7 Hz and �P 90 cm H2O:8 Hz, yetminute ventilation (where VE � f x Vt)yielded a difference of approximately 1.5L. The difference in VE between different�P:Hz combinations yielding similartidal volumes would be expected to beeven more pronounced if gas exchange(particularly at higher frequency ranges)behaves as though VE � f x Vt

2 (13).High-Frequency Oscillatory Ventila-

tion and Oscillatory Pressure Amplitude(�P) Transmission. Several studies haveexamined distal airway pressure trans-mission during HFOV. The oscillatorypressure ratio (OPR) is defined as theratio of distal (�Pdistal) and proximal(�Pproximal) ETT pressure swings and isexpressed by the equation:

OPR � �Pdistal/�Pproximal [1]

Using an electrical analog model (with3.0-mm ETT) simulating neonatal idio-pathic respiratory distress (IRDS), theOPR decreased (minimal range 0.37–0.78) as mPaw was increased to 20 cmH2O and subsequently rose as mPaw in-creased to 40 cm H2O (14). This phenom-enon was attributed to changes in lungcompliance; decreasing OPR was ob-served with lung recruitment and im-proved compliance, whereas the risingOPR associated with increased mPaw wasbelieved due to decreased compliancesecondary to alveolar overdistension. As aresult, the OPR was lowest at maximalcompliance, and this value was identifiedas the transition between maximum lungrecruitment and alveolar overdistension(14). In addition, although the minimumOPR was reached at the same mPaw, theactual value was affected by frequency(decreased at higher Hz), �P (decreasedat a maximum of 25 cm H2O), and ETTresistance (lower with increased resis-tance). An acute lung injury animalmodel (Yorkshire pigs using 5.5-mmETT), evaluating OPR and shunt fraction(used to determine maximal lung recruit-ment), showed a similar initial decreasein OPR (minimum value 0.10 � 0.01 atmPaw 31 � 4 cm H2O, with optimalmPaw 32 � 6 cm H2O) followed by anincrease as mPaw was raised further dur-ing lung inflation. Importantly, the min-imal OPR was decreased (0.04 � 0.01) at

Figure 3. Top panel, oscillatory pressure ratio (OPR), defined as �Pcarina/�Pproximal, is increased withan 8.0-mm internal diameter endotracheal tube (ETT) compared with a 3.5-mm ID ETT (no cuff leakwith either ETT) and increases with increasing mean airway pressure (mPaw). High-frequencyoscillatory ventilation (HFOV) settings: �p � 70 cm H2O, Hz 3, IT% 33, bias flow 30 L/min, test lungcompliance � 0.02 L/cm H2O, test lung airway resistance � 20 cm H2O/L/sec. Bottom panel, OPRincreases with increased IT%. HFOV settings: mPaw 30 cm H2O, �p � 70 cm H2O, Hz 3, bias flow 30L/min, test lung compliance 0.02 L/cm H2O, and resistor in the tracheal position 50 cm H2O/L/sec.


mPaw of 18 � 1 cm H2O (with optimalmPaw 14 � 2 cm H2O) during lung de-flation. These data suggest that similaroxygenation (and smaller intrapulmonarypressure swings) after lung recruitmentmaneuvers can be achieved with a lowermPaw set by the deflation limb of thepressure-volume curve rather than by theinflation limb (15).

WHMC mechanical test lung data,comparing OPR response with adult-sized ETT (cuffed 8.0 mm ID) vs. neonatalETT (3.5-mm ID), indicates that factorsother than compliance may influenceOPR (16). A cuffed 8.0-mm ID ETT wasassociated with higher absolute OPR val-ues when mPaw was increased and whenIT% was increased from 33% to 50%(Fig. 3). Additionally, the OPR was notedto increase as airway resistance wasraised, a finding supported by previousmodel data (11), in which higher pres-sure transmission in the trachea was seenin response to increased peripheral air-way resistance. Overall, mechanical lungmodels suggest that, in addition to com-pliance, OPR varies in response to ETTsize, airway resistance, and IT%, and thatthese factors may limit use of the OPR asa clinical tool to optimize mPaw duringHFOV using adult-sized ETTs.

Endotracheal Tube Cuff Leak Effects.Clinical human neonatal and small animalmodel use of HFOV generally use small-diameter cuffless ETTs. The presence of anETT cuff leak may be of importance inpromoting gas exchange (particularly CO2

elimination) and may also potentially at-tenuate the distal transmission of oscilla-tory pressure amplitude (�P) and tidal vol-umes. Interestingly, placement of an ETTcuff leak has been routinely recommendedduring high-frequency percussive ventila-tion/volume diffusive respiration (17, 18).Despite the routine use of cuffless ETTsduring HFOV for neonates, early “rescue”use of HFOV in adults used a deliberateETT cuff leak only as a “last resort” if re-fractory hypercapnia (pH �7.20) persisteddespite maximal �P and lowest Hz settings.ETT cuff leaks are created by withdrawingair from the ETT cuff balloon until mPawdecreases by 5–10 cm H2O, followed by areturn to previous desired mPaw by in-creasing the bias flow or mPaw control. Incases of upper airway edema causing ob-struction (such as with burns or facialtrauma), placement of a small, uncuffedETT or oral/nasal airway into the supraglot-tic hypopharynx may facilitate a path forCO2 egress with cuff leak placement (19).Importantly, failure of the mPaw to de-

crease when deflating the ETT cuff mayindicate that upper airway or glottic edemais preventing an effective cuff leak and sug-gests the need for an additional oral/nasalairway.

The WHMC test lung model (using8.0-mm ID cuffed ETT) has been used tofurther study mechanisms of ETT cuffleaks placed during HFOV (10). A BI-CORE CP-100 monitor (placed at the cir-

Figure 4. Top panel, high-frequency oscillatory ventilation (HFOV) inhaled tidal volume (Vti in mL)delivered to a mechanical test lung model with instilled endotracheal tube (ETT) cuff leaks. IncreasedVti is seen with restoration of mPaw in the presence of a cuff-leak. Vti measurements were obtainedwith the BICORE CP-100 respiratory monitor placed between the HFOV circuit and an 8.0-mm ETT.HFOV baseline settings were: mPaw 40 cm H2O, �P 70 cm H2O, frequency 3 Hz, IT% of 33(inspiration:expiration [I:E] 1:2) and 50 (I:E 1:1), and bias flow 30 L/min. Test lung compliance was0.02 L/cm H2O and airway resistance was 5 and 20 cm H2O/L/sec in the distal tracheal and right andleft mainstem positions, respectively. At a mPaw of 40 cm H2O, a 10 cm H2O ETT cuff leak was placedby withdrawing air from the cuff balloon until a decrease in mPaw of 10 cm H2O was noted. At a mPawof 30 cm H2O (denoted on the x-axis by 30 cm H2O, 10 leak), data were acquired. The mPaw was thenrestored to 40 cm H2O (by increasing the mPaw control, or bias flow if needed) with the cuff leak stillpresent (40 cm H2O, � 10 leak), and data were again acquired. The cuff balloon was then reinflatedand a 15-cm H2O leak was placed (25 cm H2O, 15 cm H2O leak), followed by restoration of the mPawto 40 cm H2O by this method (40 cm H2O � 15 leak) with data acquired at both of these settings. IT%of 50 resulted in greater Vti at all leak settings compared with IT% 33. Middle panel, distal intrapul-monary oscillatory volume (�V, measured in mL) in a mechanical test lung decreases with decreasedmPaw in the presence of a cuff leak, even when mPaw is restored. Volume changes were measured byPneuview software and confirmed by analog scale of the test lung. HFOV baseline settings were asfollows: mPaw 40 cm H2O, �P 70 cm H2O, frequency 3 Hz, IT% 33 (I:E 1:2), and bias flow 30 L/min.Test lung compliance was 0.05 L/cm H2O and airway resistance was 5 and 20 cm H2O/L/sec in thetracheal and right and left mainstem positions, respectively. Data were obtained using the previouslymentioned (top) cuff-leak protocol and an 8.0-mm ETT. Bottom panel, intrapulmonary oscillatorypressure amplitude (�P in cm H2O) in a mechanical test lung decreases with decreased mPaw in thepresence of a cuff leak, even when mPaw is restored. Data were obtained using Pneuview software andusing the same HFOV settings and cuff-leak protocol as outlined in the top panel. IT% of 33 and 50were used with an 8.0-mm ETT, test lung compliance setting of 0.02 L/cm H2O, and airway resistancesof 5 and 20 cm H2O/L/sec in the tracheal and right and left mainstem positions, respectively.


cuit wye) was used to measure Vti and Vte

changes in response to cuff leaks. Addi-tionally, gas flow above and below theETT cuff was evaluated using the proxi-mal and distal artificial trachea ports(Rascal II analyzer). Distal internal testlung oscillatory pressure amplitude andvolume changes were recorded directlyfrom the test lung. Creation of a cuff leakwas demonstrated to acutely increase Vti

and decrease Vte measured at the circuitwye (Fig. 4). Interestingly, despite therestoration of mPaw to the original setlevel by increasing the bias flow, Vti anddistal oscillatory pressure and volume inthe test lung remained attenuated (Fig.4). The slight reduction in distal oscilla-tory pressure amplitude and volumetransmission suggests that the presenceof a cuff leak may allow for retrograde“escape” of phasic oscillatory pressurepulses past the cuff. Gas analysis (usingnitrogen and oxygen sampling at theproximal and distal tracheal ports during100% oxygen administration through theHFOV circuit) demonstrated that cuffleak placement was associated with theimmediate appearance of internal testlung gas (e.g., oxygen) sampled at theproximal port, with no evidence of ambi-ent gas entrainment (e.g., nitrogen) sam-pled at the distal port. Although ventila-tion and resultant CO2 removal duringHFOV likely occurs through several in-teractive mechanisms (as reviewed by Dr.Pillow elsewhere in this supplement),these experiments suggest that cuff-leakplacement may facilitate end-tidal CO2

washout from the trachea, somewhat

similar to the use of a tracheal gas cath-eter (20). As a result of improved CO2

clearance and potential reduction of dis-tal pressure and volume transmission,early ETT cuff-leak placement might al-low the use of potentially more “lungprotective” (lower �P and/or higher Hz)settings. Cuff-leak experiments using theartificial trachea and mechanical testlung, however, cannot address whetheraspiration or ventilator associated pneu-monia might be promoted by the use ofsuch leaks. Clinical HFOV algorithms us-ing “early” vs. “last resort” cuff leaksshould be investigated (including the useof newer ETTs, which provide infraglotticsuction ports above the cuff).

Simulated Clinical Conditions. Unlikeconventional ventilators, in which alarmsand airway pressures warn the clinicianof potential complications, the oscillatoralarm system is essentially limited to ahigh- and low-pressure alarm. Typically,these alarms are bracketed 5–7 cm H2Oabove and below the desired mPaw, andabrupt ventilator stoppage is usually aresult of circuit decompression as a resultof disconnected or loose valve/circuitconnections (e.g., expiratory valve tub-ing).

Data evaluating the effects of commonclinical conditions (e.g., tension pneumo-thorax, ETT occlusion, mainstem intuba-tion, or bronchopleural fistula) on ob-servable HFOV parameters is limited;however, mechanical test lung data sug-gests that changes in �P (assuming aconstant power setting) may be a usefulparameter in evaluating various scenar-

ios. The mechanical test lung (21) hasbeen used to evaluate changes in �P andmPaw (where a significant change wasinterpreted as a variation in �P or mPaw�10 cm H2O) during simulated clinicalsituations (Fig. 5). The following obser-vations were made: 1) increased airwayresistance increased �P without an ap-preciable effect on mPaw, 2) air leak re-sulted in a decreased mPaw while �Premained constant, and 3) changes inlung compliance did not significantly al-ter mPaw or �P.

In practice, these findings highlightthe importance of noting and recordingthe intended and displayed �P in relationto the power setting, as well as the mPaw.Discrepancies should be noted and inves-tigated further rather than arbitrarilymanipulating HFOV settings. Similarly,the HFOV alarms may not alert the clini-cian to potentially catastrophic events(such as a tension pneumothorax), and ahigh level of suspicion must be main-tained in the event of desaturationsand/or hemodynamic instability.

CONCLUSION

Animal and test lung models haveplayed an important role in understand-ing the interaction of HFOV and lungmechanics. Although limited in certainaspects, mechanical test lung modelsmay provide useful insights into HFOVtidal volume delivery, distal airway pres-sure transmission, the effects of cuff-leakplacement, and ventilator parameter al-terations during simulated clinical sce-narios. Lessons learned from these mod-els may contribute to developing HFOVstrategies for clinical use.

REFERENCES

1. Imai Y, Nakagawa S, Ito Y, et al: Comparisonof lung protection strategies using conven-tional and high-frequency oscillatory venti-lation. J Appl Physiol 2001; 91:1836–1844

2. Rotta AT, Gunnarsson B, Fuhrman BP, et al:Comparison of lung protective ventilationstrategies in a rabbit model of acute lunginjury. Crit Care Med 2001; 29:2176–2184

3. Slutsky AS, Tremblay LN: Multiple systemorgan failure. Is mechanical ventilation acontributing factor? Am J Respir Crit CareMed 1998; 157:1721–1735

4. Ventilation with lower tidal volumes as com-pared with traditional tidal volumes for acutelung injury and the acute respiratory distresssyndrome: The Acute Respiratory DistressSyndrome Network. N Engl J Med 2000; 342:1301–1308

5. Derdak S, Mehta S, Stewart T, et al: High-

Figure 5. High-frequency oscillatory ventilation (HFOV) parameter changes in response to simulatedclinical conditions using a mechanical test lung. Data were obtained using a Michigan Instruments2600i mechanical test lung, SensorMedics 3100B HFOV, and Pneuview software. Baseline HFOVsettings were: mPaw 28–30 cm H2O, constant power setting of 8 (correlating to a �P of 66 cm H2O),frequency 5 Hz, IT% 33, and bias flow 30 L/min. Changes were considered clinically significant if avariation of �10 cm H2O was noted in either �P or mPaw. Each1 or2 indicates an increase of �10cm H2O or decrease of �10 cm H2O, respectively (e.g., 11 � an increase of �20 cm H2O, 111� an increase of �30 cm H2O), and7 signifies a change �10 cm H2O. The method of simulating eachcondition is noted in the second column of the figure.


frequency oscillatory ventilation for acute re-spiratory distress syndrome: a randomized,controlled trial. Am J Respir Crit Care Med2002; 166:801–808

6. Fort P, Farmer C, Westerman J, et al: High-frequency oscillatory ventilation for adult re-spiratory distress syndrome: a pilot study.Crit Care Med 1997; 25:937–947

7. Mehta S, Lapinsky S, Hallett D, et al: Pro-spective trial of high-frequency oscillationin adults with acute respiratory distresssyndrome. Crit Care Med 2001; 29:1360 –1369

8. Seedek KA, Takeuchi M, Suchodolski K, et al:Determinants of tidal volume during high-frequency oscillation. Crit Care Med 2003;31:227–231

9. Scalfaro P, Pillow JJ, Sly PD, et al: Reliabletidal volume estimates at an airway openingwith an infant monitor during high-fre-quency oscillatory ventilation. Crit Care Med2001; 29:1925–1930

10. Van de Kieft MW, Dorsey D, Venticinque S, etal: Effects of endotracheal tube cuff-leak ongas flow patterns in a mechanical test lungmodule during high-frequency oscillatory

ventilation. Am J Respir Crit Care 2003; 167:A178


12. Derdak, S: High-frequency oscillatory venti-lation for acute respiratory distress syn-drome in adult patients. Crit Care Med 2003;31:S317–S323

13. Boyton BR, Hammond MD, Fredberg JJ, etal: Gas exchange in healthy rabbits duringhigh-frequency oscillatory ventilation. J ApplPhysiol 1989; 66:1343–1351

14. van Genderingen HR, Versprille A, Leen-hoven T, et al: Reduction of oscillatory pres-sure along the endotracheal tube is indica-tive for maximal respiratory complianceduring high-frequency oscillatory ventila-tion: a mathematical study model. PediatrPulmonol 2001; 31:458–463

15. van Genderingen HR, van Vught AJ, DuvalEL, et al: Attenuation of pressure swingsalong the endotracheal tube is indicative ofoptimal distending pressure during high-frequency oscillatory ventilation in a model

of acute lung injury. Pediatr Pulmonol 2002;33:429–436

16. Dorsey DA, Derdak S, Van de Kieft MW, et al:Effect of endotracheal tube size on oscillatorypressure ratio in a mechanical test lung modelduring high-frequency oscillatory ventilation.Am J Respir Crit Care 2003; 167:A179

17. A Manual on Volumetric Diffusive Respira-tion (VDR) for the Administration of In-trapulmonary Percussive Ventilation (IPV).Sandpoint, ID, Percussionaire Corp, 1993

18. Duotron Users Guide. Sandpoint, ID, Percus-sionaire Corp, 2000

19. Cooper AB, Islur A, Gomez M, et al: Hyper-capnic respiratory failure and partial upperairway obstruction during high-frequencyoscillatory ventilation in an adult burn pa-tient. Can J Anesth 2002; 49:724–728

20. Dolan S, Derdak S, Solomon D, et al: Tra-cheal gas insufflation combined with high-frequency oscillatory ventilation. Crit CareMed 1996; 24:458–465

21. Morison D, Derdak S: High-frequency oscil-latory ventilator parameter changes in re-sponse to simulated clinical conditions usinga mechanical test lung. Am J Respir Crit Care2000; 163:A388


High-frequency oscillatory ventilation: What large-animal studieshave taught us!

Robert M. Kacmarek, PhD, RRT; Atul Malhotra, MD

I nformation on the physiologic ef-fects, the mechanisms of gas ex-change, and the potential clinicalutility of high-frequency oscilla-

tion (HFO) has been acquired in bothlarge- and small-animal studies and fromclinical trials. However, large-animal tri-als, specifically, have provided consider-able information regarding the effects ofventilatory variables on regional gastransport during HFO, the effect of ven-tilatory settings on CO2 elimination, ap-proaches to setting the oscillator to en-sure maximum oxygenation, and thepotential benefits of combining HFO withpartial liquid ventilation (PLV). Large-animal studies have been important inthe development of HFO for clinical usein humans because large animals presentmany of the same concerns and chal-lenges that must be overcome duringventilation of adults. In this article, thespecific contributions from large-animalstudies will be addressed.

Regional Gas Transport

We have learned during the past 30 yrsthat gas transport during conventionalventilation (CV) and HFO (or high-frequency ventilation in general) are sim-ilar in many respects but also differgreatly in others. During CV, gas trans-port along the tracheobronchial tree isprimarily by bulk flow or convection.However, during HFO, the net transportof gas molecules can take place evenwhen fresh gas does not directly reach allregions of the lung or when tidal volumesare less than anatomic dead space vol-ume. In the mid 1980s, Jose Venegas’sgroup (1–5) performed a series of exper-iments in 10- to 30-kg dogs that helped toexplain how ventilation with very smalltidal volumes could provide adequateoverall and regional ventilation. Most ofthese experiments were conducted usingpositron-emitting radioisotope nitro-gen-13 (13NN) imaging. Distribution ofventilation with this process is easilyidentified because of the low solubility of13NN in blood. As a result, 13NN is con-fined to the air space, and its regionaldistribution is not affected by blood flow.

In the first of these experiments Vene-gas et al. (1) estimated alveolar ventila-tion using a high-frequency oscillator ca-pable of generating tidal volumes of 30–120 mL at frequencies of 2–25 Hz withequal and constant inspiratory-to-expira-tory flows and ratios at a mean lung vol-ume equal to the animal’s functional re-

sidual capacity. They found that specificventilation (SPV; ventilation per unit ofcompartment volume) followed the fol-lowing relationship:

SPV � 1.9�VT/VL�2.1 � f [1]

where VT is tidal volume, VL is lung vol-ume, and f is frequency. From this rela-tionship and arterial PaCO2 levels, theywere able to derive an expression for thenormocapnic settings of VT and f, givenmean lung volume (VL) and body weight(W):

�VT � f�n � 0.73 � W � �VT/VL��1.1 [2]

where (VT·f)n represents the normocap-nic product for HFO. These findings weresimilar to those of other groups (6–8)also using similar animal models. Vene-gas et al. (1) went on to point out that theefficiency of ventilation at small tidal vol-umes during HFO is much lower thanwith large tidal volumes, as in CV, andthat to successfully ventilate a subjectwith HFO, a much greater VT·f product isrequired than would be required duringCV (Fig. 1).

In a subsequent study in this series,Yamada et al. (2), using similar-sizedhealthy dogs and 13NN imaging, showedthat changes in inspiratory-to-expiratoryratios (I:E) from 1:1 to 1:4 had no signif-icant effect on PaO2, PaCO2, or alveolarventilation at frequencies of 3, 6, and 9Hz. During these experiments, tidal vol-

From the Departments of Anesthesiology (RMK)and Medicine (AM), Harvard Medical School, Boston,MA; the Department of Respiratory Care, Massachu-setts General Hospital, Boston, MA (RMK); and Brighamand Women’s Hospital and Beth Israel DeaconessMedical Center, Boston, MA (AM).


DOI: 10.1097/01.CCM.0000156786.43935.A0

Background: Much of the information on the physiologic ef-fects, mechanisms of gas exchange, and potential utility of high-frequency oscillation (HFO) has been acquired in animal studies.Specifically, large animal data have been useful in assessingadult application because large animals present many of thesame concerns and challenges as adults.

Objective: To review the literature on HFO testing in largeanimal models, identifying contributions to the understanding ofmechanisms of action and the physiology of HFO.

Results: Large animal studies have clarified the mechanisms

of gas exchange during HFO, identified approaches to settingmean airway pressure based on lung mechanics, and identified apotentially better approach to applying partial liquid ventilation.

Conclusion: The study of HFO in large animal models has beenessential to our understanding of the optimal approach to applyingHFO in human studies. (Crit Care Med 2005; 33[Suppl.]:S148–S154)

KEY WORDS: high-frequency oscillatory ventilation; partial liquidventilation; regional gas transport; conventional ventilation; large-animal models


ume and mean lung volume were heldconstant. These data helped to supportthe assumptions of Permutt et al. (9),that axial dispersion of molecules withinthe lung is independent of flow. Menon etal. (10) also showed that the generalshape of the velocity profiles during in-spiration and expiration in HFO up to 9Hz were insensitive to changes in thebulk flow rate.

The basic effects of mean airway pres-sure on gas transport were documentedby Yamada et al. (3) in 1986. In thisstudy, the authors again used healthydogs. PaO2, PaCO2, and CO2 productionwere measured after 10 mins of HFO atthree levels of mean airway pressure (0, 5,and 10 cm H2O) and three frequencies (3,6, and 9 Hz). At each f, VT was adjusted ata mean airway pressure of 0 cm H2O toobtain eucapnia. In these healthy ani-mals, SpO2 decreased and PaCO2 increasedat each rate setting as mean airway pres-sure was increased to 5 and then to 10 cmH2O (Fig. 2). Alveolar ventilation de-creased about 23% and 40% at mean air-way pressures of 5 and 10 cm H2O, re-spectively. These changes in this healthylarge-animal model were a result of de-creases in alveolar ventilation as meanairway pressure increased, probably a re-sult of a decreased cardiac output (lead-ing to increased physiologic dead space).Because ventilation occurred at an FIO2 of0.21, the decreases in PaO2 observed asPaCO2 increased were most probably a re-sult of the increase in PaCO2, as predictedby the alveolar gas equation.

In a fourth article in this series, Vene-gas et al. (4), again using healthy 16- to

30-kg dogs and 13NN imaging, definedthe regional (basal vs. apical) effect oftidal volume, respiratory frequency, andI:E ratio during HFO. They found that, ata constant VT·f product, increasing VT re-sulted in higher overall lung ventilationper unit of lung volume as a result ofincreased basal ventilation per unit oflung volume but little change in apicalventilation per unit of lung volume (Fig.3). In addition, they determined that in-creasing VT·f at a constant VT increasedoverall ventilation without affecting a dif-ference in the basal-to-apical regionalventilation distribution. As they hadshown previously (3), I:E had no impacton regional distribution of ventilation.They concluded that as VT increases, gastransport changes from inefficient disper-sion to primarily more efficient bulk gasflow (convection), with high regionaltidal volume to dead space ratios. Thesedata are consistent with the findings fromother groups also using HFO in animalmodels (11–13).

In a fifth study in this series, Yamadaet al. (5) studied regional mapping of gastransport during HFO vs. CV. HFO wasprovided at frequencies of 3, 6, and 9 Hzwhere the tidal volume at 6 Hz producedeucapnia. The VT·f obtained at 6 Hz re-mained constant as the rate was de-creased to 3 Hz or increased to 9 Hz.During CV, rate was set at 10 breaths/minand VT set to produce eucapnia. Regionalnonuniformity in gas transport wasgreatest for HFV at 3 and 6 Hz and lowestat 9 Hz and during CV. A central region atthe base of the lungs was preferentiallyventilated during HFO, resulting in a

time-averaged ventilation in this areaequal to that of the main bronchi. Theauthors argued that this finding isstrongly supportive of the fact that con-vection is a primary mechanism for gasexchange during HFO. These data areconsistent with the group’s previous re-ports in large-animal models (2–4) show-ing that overall gas transport in HFO isdependent on VT

2·f and is in agreementwith the findings of others regarding re-gional distribution of ventilation (14–16). Although some debate has previouslyoccurred regarding the magnitude of theexponent on the tidal volume term of thisequation, this controversy has nowlargely resolved. with a value of approxi-mately 2 now being widely accepted.

The phenomenon of “negative fre-quency dependence” is commonly ob-served with the existing high-frequencyoscillators. This phenomenon, whereby areduced respiratory frequency leads to anincrease in CO2 excretion, is likely a prod-uct of the increase in tidal volume thatoccurs with a fixed duty cycle. That is, alower respiratory rate is more than offset

Figure 2. A, PaCO2 vs. mean airway pressure(Paw); B, PaO2 vs. Paw. Each bar represents adifferent frequency (3, 6, 9 Hz). *p � .01 vs. Pawof 0 cm H2O; **p � .001 vs. Paw of 0 cm H2O.Reproduced with permission from Yamada et al(2).

Figure 1. Specific ventilation (SPV) vs. tidal volume � frequency (VT � f) for a constant VT of 80 mL(solid line) and 40 mL (dashed line) from a single dog. Note that a large VT effect is present at constantVT � f, regardless of frequency. VT still has the greatest effect on specific or regional ventilation duringhigh-frequency oscillation. Reproduced with permission from Venegas et al (1).


by the increase in tidal volume becausethe square of the tidal volume is the crit-ical factor governing gas exchange athigh frequencies. Conversely, an increasein respiratory frequency leads to a de-crease in tidal volume, yielding reducedCO2 clearance. This apparent paradox is acommon observation with current tech-nologies for delivering HFO.

Gas Exchange

The ability of HFO to maintain PaO2

and PaCO2 in healthy dogs was nicelydemonstrated by Bohn et al. (17) in 1980.The authors observed a relatively stablePaCO2 of 33.1 � 0.5 mm Hg at a fre-quency of 15 Hz with tidal volumes of 1.9mL/kg using an uncuffed airway. WithFIO2 equal to that of room air, PaO2 couldbe maintained at �95 � 5 mm Hg for 5hrs, and with 100% oxygen, PaO2 was�580 � 9 mm Hg for 5 hrs. This was oneof the first experiments demonstratingthat gas exchange could be maintainedwith tidal volumes less than dead spacevolume.

Slutsky et al. (8) examined the effect ofvarying frequency between 2 and 30 Hzand tidal volume between 1 and 7 mL/kgand lung volume on the efficiency of CO2

elimination in healthy dogs. In all exper-iments CO2 increased with frequency at aconstant VT. However, the most impor-tant variable in determining CO2 was VT·f,but there was considerable variability inthe response of animals to doubling theVT at a constant VT·f. In some animals,CO2 decreased, although in the majorityof animals, CO2 markedly increased. In-creasing lung volume by increasing meanairway pressure up to 25 cm H2O had nosignificant effect on CO2. These data sup-ported the assumptions made by thesame group (18) in a previous article thatgas exchange in HFO was due to a num-ber of augmentative mechanisms in ad-dition to convective flow, Taylor laminar(19) and turbulent dispersion (20), mix-ing due to asymmetrical velocity profilesas proposed by Haselton and Scherer(21), and secondary flows at bifurcations(e.g., pendelluft). In 1982, Thompson etal. (22) compared HFO with CV in 15-kgdogs with oleic acid injury. Cardiac out-put and gas exchange were compared atequivalent mean airway pressures. CVwas performed with a tidal volume of16–21 mL/kg and a frequency of 15–20breaths/min. HFO was delivered at a fre-quency of 15 Hz. FIO2 was maintained at0.5, and mean airway pressure varied over

the range of 7.5–27 cm H2O. With HFO,oxygenation improved as mean airwaypressure increased; however, the im-provement was equal to that during CV atthe same mean airway pressure. Thesedata clearly demonstrated that HFOcould maintain oxygenation in severelung injury at a level at least equivalent tothat achieved during CV.

Setting of Mean AirwayPressure

As with CV and the setting of positiveend-expiratory pressure, there has beenconcern over the methodology used to setmean airway pressure during HFO. Manyhave simply recommended setting meanairway pressure equal to or a few centi-meters of H2O above that used during CV(23–25). Goddon et al. (26), in a 28 �5–kg lavage-injured sheep model investi-gated the use of the pressure–volume

(P-V) curve to set mean airway pressureduring HFO. Inflation and deflation P-Vcurves were measured after lung injuryfollowed by HFO at a frequency of 8 Hz, adelta pressure adjusted to establish PaCO2

with a bias flow of 30 L/min, an FIO2 of1.0, and an I:E ratio of 1:1. The authorsdefined Pflex as the lower inflection pointon the inflation limb of the P-V curve.Gas exchange and hemodynamics wereevaluated at Pflex �2, �6, �10, and �14cm H2O after 1 hr of ventilation at eachlevel. Before the random setting of eachmean airway pressure, the lungs wererecruited with 50 cm H2O continuouspositive airway pressure for 60 secs. Asnoted in Figures 4 and 5, PaO2/FIO2 wasgreatest at Pflex � 6 cm H2O, withoutadversely affecting the cardiac output. In-terestingly, Pflex � 6 cm H2O in thismodel was equal to 26 � 1 cm H2O,which was equivalent to the point of max-

Figure 3. Plots of regional specific ventilation (spVr), normalized by corresponding mean specificventilation (spV), as a function of frequency for four lung regions: right base (RB), left base (LB), rightapex (RA), and left apex (LA). A, tidal volume (VT) � 40 mL; B, VT � 80 mL, mean � SE, n � 6.Reproduced with permission from Venegas et al (4).

Figure 4. PaO2/FIO2 (P/F) ratio at baseline 1 (BL 1) before injury, injury, four settings of mean airwaypressure during high-frequency oscillation (PCL � 2, �6, �10, �14), and at baseline 2 (BL 2). Allvalues are mean � SD. *p � .05 vs. injury. Reproduced with permission from Goddon et al (26).


imum compliance change on the defla-tion limb of the P-V curve, also 26 � 1 cmH2O.

Luecke et al. (27) also evaluated theuse of the P-V to set mean airway pres-sure in lavage-injured large (45.9 � 4.8kg) swine but, in addition, assessed lungvolume at both the base and apex of thelung during these trials using computer-ized tomography. The oscillator was setat 4 Hz, 30 L/min bias flow, delta pressureof 60 cm H2O, FIO2 of 1.0, and I:E of 1:1.Mean airway pressures were based onPflex. Initially, mean airway pressure wasset at Pflex, then 1.5 � Pflex, and then 2 �Pflex, followed again by 1.5 � Pflex, andfinally, again Pflex. At each mean airwaypressure setting, computerized tomo-graphic scans were performed at the baseand apex of the lungs, and blood gas andhemodynamic data were obtained. Gasexchange and hemodynamic data werebest at 1.5 � Pflex on the inflation limb ofthe P-V curve. In fact, at this setting, PaO2

was established at preinjury values. In-creasing the mean airway pressure fur-ther did not further increase oxygenationbut did decrease oxygen delivery by de-creasing cardiac output. In addition, nodifferences in respiratory or hemody-namic variables were observed at equiva-lent mean airway pressure on the ascend-ing and descending limb of the P-V curve.Variation in total slice lung volume wasless than anticipated from the P-V curve.Overdistending lung volume was esti-mated at about 3% and did not varygreatly as mean airway pressure changed.Total slice lung volume was greater dur-ing HFO than predicted by the P-V curveand near the deflation curve volume atcorresponding pressures. That is, the

marked hysteresis observed on the P-Vcurve with this injury model was absentduring HFO.

The findings of Godden et al. (26) andLuecke et al. (27) are remarkably simi-lar—1.5 � Pflex was equal to about 27 cmH2O—almost exactly the value obtainedby Godden et al (26). These data suggestthat the P-V curve is very useful for thesetting of mean airway pressure duringsevere lung injury. Mean airway pressureset at Pflex � 6 cm H2O or at the point ofmaximum compliance change on the de-flation limb of the P-V curve seems ideal.

Tidal Volume during HFO

Sedeek et al. (28), using a 30-kg la-vage-injured sheep model, demonstratedthat tidal volumes during HFO at lowfrequencies and high delta pressure ap-proached those currently recommendedduring CV in acute respiratory distresssyndrome patients (29) (Fig. 6). Animalswere ventilated at a mean airway pressureequal to the maximum compliancechange on the deflation limb of the P-Vcurve (26 � 1.9 cm H2O) through an8-mm internal diameter endotrachealtube with its cuff inflated. Tidal volumewas then measured at all combinations ofrates of 4, 6, 8, and 10 Hz, pressure am-plitudes of 30, 40, 50, and 60 cm H2O,and I:E ratios of 1:1 and 1:2. At both I:Eratios, tidal volume was directly propor-tional to pressure amplitude and in-versely proportional to frequency. A tidalvolume of 4.4 � 1.2 mL/kg was deliveredat an I:E of 1:1, delta pressure of 60 cmH2O, and rate of 4 Hz. These findingswere similar to those observed in othersmaller animal models (30–32). It is in-

teresting to speculate that at the capabil-ity of current adult oscillators with deltapressures of 90 cm H2O and rates of 3 Hz,tidal volumes of �6 mL/kg would be de-livered during HFO in this 30-kg lavageinjury sheep model.

HFO vs. CV

HFO has been compared with CV innumerous small-animal trials, but lim-ited data are available in large-animalmodels for which the approach used dur-ing CV and HFO were similar. Sedeek etal. (33) performed such a study in a 30-kglavage-lung injured sheep model using anopen-lung approach to ventilatory sup-port provided by HFO, pressure control,and intratracheal pressure ventilation.After injury in all three groups, the lungswere recruited and either positive end-expiratory pressure (pressure control, in-tratracheal pressure ventilation) or meanairway pressure was then set using a dec-remental trial. During HFO, rate was 8Hz, delta pressure was set to achieve aPCO2 of 35–45 mm Hg, and I:E was 1:1.Pressure control was provided at a rate of30 breaths/min, an I:E of 1:1, and a tidalvolume to maintain PaCO2 in the targetrange. Intratracheal pressure ventilationwas established by a bias flow via a dou-ble-lumen endotracheal tube in whichthe rate of occlusion was 120/min at anI:E of 1:1. The bias flow (about 18 L/min)was adjusted to ensure the target PaCO2.As noted in Figure 7, PaO2/FIO2 and PaCO2

were maintained at a similar level over a4-hr observation period, regardless of ap-proach. No differences were observed inhemodynamic response, but there was atrend to less lung injury in the groupreceiving HFO and intratracheal pressureventilation. This may be partially ex-plained by a resultant large tidal volume(8.9 � 2.1 mL/kg) and high plateau pres-sure (30.6 � 2.6 cm H2O) during pres-sure control. These data clearly illustratethe fact that HFO can be used with atleast equal efficacy as CV in severe lunginjury.

HFO and PLV

The use of PLV in patients has notresulted in improved outcome (34–36).Some have argued that a primary reasonfor this failure to identify outcome bene-fit is the approach to providing ventila-tory support during PLV (37, 38). All clin-ical trials have been performed using CV.

Figure 5. Cardiac output (CO) at baseline 1 (BL 1) before injury, injury, four settings for mean airwaypressure during high-frequency oscillation (PCL �2, �6, �10, �14), and at baseline 2 (BL 2). Allvalues are mean � SD. *p � .05 vs. injury. Reproduced with permission from Goddon et al (26).


As previously described by Fuhrman etal. (39) and Arnold (37), during PLV,three distinct populations of alveoli existbased on gravitational location. PLV hasbeen referred to as liquid positive end-expiratory pressure (39) because its ef-fects are primarily experienced by themost gravity-dependent lung units. Thispopulation experiences the effect of theperfluorocarbon throughout ventilation

because it remains fluid-filled. The leastgravity-dependent lung units only expe-rience gas ventilation because they areminimally affected by the perfluorocar-bon. It is the population of lung unitsbetween these two extremes that are sub-jected to the actual air–liquid interface.The level of perfluorocarbon in this re-gion varies with inspiration and expira-tion, so during PLV, a substantial part of

the lung still experiences the shear stressof a gas-filled lung or liquid–gas inter-phase (37).

Arnold (37) and Mammel (38) arguethat HFO may be the ideal method ofventilating and oxygenating during PLV.As discussed by Arnold (37), this combi-nation has a number of distinct advan-tages. The perfluorocarbon reverses theatelectasis in the dependent lung and di-rects pulmonary blood flow to the non-dependent lung (40). Because of its lowerVT and peak alveolar pressure, HFO couldventilate the nondependent lung with ad-equate mean airway pressure to avoidoverdistension and atelectasis in this re-gion (41). Arnold (37) proposed that thesmall VT associated with HFO minimizesthe shear stress and subsequent injury inthe regions containing liquid and air. Al-though much of the work addressing theuse of HFO during PLV has been per-formed on small-animal models (42–47),a number of large-animal HFO–PLV stud-ies have been performed by Doctor et al(48–50).

Histopathologic evidence of improvedrecruitment in dependent and nondepen-dent lung during HFO–PLV was demon-strated by Doctor et al. (48) in a swine(29.6 kg) saline lavage–lung injuredmodel. After injury and 1 hr of stabiliza-tion, animals were randomized to HFO orHFO–PLV with 30 mL/kg perfluorocar-bon dosing, then ventilated for 2 hrs.Over the course of the study, there wereno differences between groups in gas ex-change, hemodynamic function, or pul-monary vascular resistance. However, at-electasis scores were reduced greatly inthe HFO–PLV group. The lack of gas ex-change difference may have been a result,at least in part, of a low mean airwaypressure during HFO. Lung mechanicswere not used to set the mean airway; itwas simply set arbitrarily at 12 cm H2Oabove the mean airway pressure duringCMV.

In 2001, Doctor et al. (49) examinedthe perfluorocarbon dose-response dur-ing HFO in a 28.9-kg saline lavage–induced acute lung injury swine model.The gas exchange and hemodynamic ef-fects of 0, 5, 15, and 20 mL/kg of perfluo-rocarbon were evaluated during HFO. Af-ter lung injury, lungs were recruited by astepwise increase in mean airway pres-sure, beginning at 10 cm H2O above themean airway pressure during CV. Afterrecruitment, the animals were stabilizedon HFO at 4 Hz, a delta pressure toachieve a PaCO2 of 45–55 mm Hg, I:E of

Figure 6. A, the combined effect of altering rate and pressure amplitude on actual tidal volume (VT)measured at an inspiratory/expiratory (I:E) ratio of 1:1; B, the combined effect of altering rate andpressure amplitude on actual VT measured at an I:E ratio of 1:2. All values are mean � SD. The numbers30, 40, 50, and 60 at the right of each line represent pressure amplitude in centimeters of H2O. *p �.05 vs. 4 Hz and 60 cm H2O; �p � .05 vs. 4 Hz and 50 cm H2O; #p � .05 vs. 6 Hz and 60 cm H2O;�p � .05 vs.4 Hz and 30 cm H2O; øp � .05 vs. 4 Hz and 40 cm H2O; &p � .05 vs. 8 Hz and 60 cmH2O; ¶p � .05 vs. 6 Hz and 50 cm H2O; fp � .05 vs. same settings at an I:E ratio of 1:1. Reproducedwith permission from Sedeek et al (28).


1:1, FIO2 of 0.6, and mean airway pressureof about 20 cm H2O. After stabilization,animals were randomized to the differentperfluorocarbon dose groups. HFO–PLVwas best tolerated with the 5- and 15-mL/kg dose of perfluorocarbon. PaCO2,pH, cardiac index, and pulmonary vascu-lar resistance did not change significantlyduring any dose or among doses. Thelowest oxygenation index was identifiedduring the 15-mL/kg dosing, and the PO2

tended to be best at the 15-mL/kg dose.These data are in contrast to much of theother dose-response data in animals dur-ing CV in which oxygenation benefit in-creased as dosing increased to 30 mL/kg(51) unless high levels of positive end-expiratory pressure were applied (40).

One of the longest laboratory evalua-tions of PLV was performed by Doctor etal. (50) in 2003. The authors evaluatedthe extended effect of HFO–PLV on gasexchange and injury quantification.Three groups of five lavage-injured, 28.9� 3.1–kg swine were studied. One groupreceived only HFO, the second received10 mL/kg perfluorocarbon in the supinepositions plus HFO, and a third received10 mL/kg perfluorocarbon plus HFO inthe prone position. HFO was applied at arate of 4 Hz to all groups. Mean airwaypressure was adjusted to maintain a SpO2

of 90% � 2% with an FIO2 of 0.6. Deltapressure was set after lung injury toachieve a PaCO2 of 45–55 mm Hg withoutadjustment after perfluorocarbon dosing.Both PaO2 and oxygenation index im-proved rapidly and significantly in theHFO prone–PLV group compared withthe other groups, but this benefit was lostafter about 6 hrs. Tissue myeloperoxidase

activity was reduced globally by HFO–PLV (either group, p � .01), and regionallung injury scores in dependent lungwere improved (p � .05). However, globallung injury scores were improved byHFO–PLV (both groups, p � .05) only inatelectasis, edema, and alveolar disten-sion but not in cumulative score.

This series of studies, along with thedata in small-animal models, does pro-vide some enthusiasm for the continuedexploration of the clinical application ofPLV. The use of CV as opposed to HFO inclinical trials may have contributed totheir negative results, and the use ofHFO–PLV may yield more encouragingresults.

SUMMARY

Large-animal models have contrib-uted greatly to our understanding of thebasic physiology of HFO and methods forclinical application during severe lung in-jury. These studies have clarified themechanisms of gas exchange during HFOand have established the fact that, inadults, HFO is effective because of bulkgas flow. CO2 elimination during HFO isincreased by either decreasing rate or in-creasing delta pressure because bothchanges increase tidal volume. Mean air-way pressure setting during HFO may beset using the static P-V curve or a decre-mental trial following lung recruitment.In either case, the mean airway pressurethat maximizes oxygenation withoutcompromising cardiac output seems tobe about Pflex � 6 cm H2O, 1.5 � Pflex, orthe point of maximum compliancechange on the deflation limb of the P-V

curve. Finally, HFO may provide a gasdelivery mechanism that allows success-ful application of PLV in the managementof acute respiratory distress syndrome.Clearly, large-animal models, becausethey present challenges similar to thoseobserved during human trials, are a crit-ical component of increasing our under-standing of the use of HFO in patients.

REFERENCES

1. Venegas JG, Custer J, Kamm RD, et al: Rela-tionship for gas transport during high-frequency ventilation in dogs. J Appl Physiol1985; 59:1539–1547

2. Yamada Y, Hales CA, Venegas JG: Inspirato-ry-to-expiratory time ratio and alveolar ven-tilation during high-frequency ventilation indogs. J Appl Physiol 1986; 61:1903–1907

3. Yamada Y, Venegas JG, Strieder DJ, et al:Effects of mean airway pressure on gas trans-port during high-frequency ventilation indogs. J Appl Physiol 1986; 61:1896–1902

4. Venegas JG, Yamada Y, Custer J, et al: Effectsof respiratory variables on regional gas trans-port during high-frequency ventilation.J Appl Physiol 1988; 64:2108–2118

5. Yamada Y, Burnham C, Hales CA, et al: Re-gional mapping of gas transport during high-frequency and conventional ventilation.J Appl Physiol 1989; 66:1209–1218

6. Kamm RD, Collins JM, Jushi CH, et al: Axialdispersion of a passive contaminant in oscil-latory flow. Abstr. In: Proceedings of the 35thAnnual Conference on Engineering in Medi-cine and Biology, 1982, p 23

7. Kamm RD, Slutsky AS, Drazen JM: High-frequency ventilation. CRC Crit Rev Bioeng1984; 9:347–379

8. Slutsky AS, Kamm RD, Rossing TH, et al:Effects of frequency, tidal volume, and lungvolume on CO2 elimination in dogs by highfrequency (2–30 Hz), low tidal volume ven-tilation. J Clin Invest 1981; 68:1475–1484

9. Permutt S, Mitzner W, Weinmann G: Modelof gas transport during high-frequency ven-tilation. J Appl Physiol 1985; 58:1956–1970

10. Menon AS, Weber ME, Chang HF: A modelstudy of flow dynamics in human centralairways: III. Oscillatory velocity profiles. Re-spir Physiol 1984; 55:255–275

11. Berdine GB, Lehr JL, McKinley DS, et al:Nonuniformity of canine lung washout byhigh-frequency ventilation. J Appl Physiol1986; 61:1388–1394

12. Moffat SL, Byford L, Forkert L: Interregionalgas mixing and gas transport during high-frequency oscillations. Respir Physiol 1985;62:1–13

13. Schmid ER, Knopp TJ, Rehder K: Intrapul-monary gas transport and perfusion duringhigh-frequency oscillation. J Appl Physiol1981; 51:1507–1514

14. Hoffman EA: Effect of body orientation onregional lung expansion: A computed tomo-

Figure 7. PaO2/FIO2 (P/F) ratio and PaCO2 throughout the course of the study protocol. PaO2/FIO2 amonggroups was similar throughout all phases of the study. No differences in PaCO2 were observedthroughout the study as a result of large SD values. All data are mean values. *p � .05 vs. injury; �p� .05 vs. 30, 60, 90, 120, 180, and 240 mins; BL, baseline; HFO, high-frequency oscillation injury(lavage injury); ITPV, intratracheal pulmonary ventilation; PCV, pressure-control ventilation; RM,immediately after lung recruitment at set positive end-expiratory pressure or mean airway pressure;30, 60, 120, 180, 240 mins, time after lung recruitment. Reproduced with permission from Sedeek etal (33).


graphic approach. J Appl Physiol 1985; 59:468–480

15. Hoffman EA, Ritman EL: Effect of body ori-entation on regional lung expansion in dogand sloth. J Appl Physiol 1985; 59:481–491

16. Hubmayr RD, Walters BJ, Chevalier PA, et al:Topographical distribution of regional lungvolume in anesthetized dogs. J Appl Physicol1983; 54:1048–1056

17. Bohn DJ, Miyasaka K, Marchak BE, et al:Ventilation by high-frequency oscillation.J Appl Physiol 1980; 48:710–716

18. Slutsky AS, Drazen JM, Ingram RH, et al:Effective pulmonary ventilation with smallvolume oscillations at high frequencies. Sci-ence 1980; 209:609–611

19. Taylor GI: Dispersion of soluble matter insolvent flowing slowly through a tube. ProcR Soc London Ser A 1953; 219:186–203

20. Taylor GI: The dispersion of matter in tur-bulent flow through a pipe. Proc R Soc Lon-don Ser A 1954; 223:446–468

21. Haselton FR, Scherer PW: Bronchial bifurca-tions and respiratory mass transport. Science1980; 208:69–71

22. Thompson WK, Marchak BE, Froese AB, etal: High-frequency oscillation compared withstandard ventilation in pulmonary injurymodel. J Appl Physiol 1982; 52:543–548



25. Derdak S, Mehta S, Stewart TE, et al: High-frequency oscillatory ventilation for acute re-spiratory distress syndrome in adults. Am JRespir Crit Care Med 2002; 166:801–808

26. Goddon S, Fujino Y, Hromi JM, et al: Opti-mal mean airway pressure during high-frequency oscillation. Anesthesiology 2001;94:862–869

27. Luecke T, Meinhardt JP, Herrmann P, et al:Setting mean airway pressure during high-frequency oscillatory ventilation accordingto the static pressure-volume curve in sur-factant-deficient lung injury. Anesthesiology2003; 99:1313–1322

28. Sedeek KA, Takeuchi M, Suchodolski K, et al:Determinants of tidal volume during high-frequency oscillation. Crit Care Med 2003;31:227–231

29. Ventilation with lower tidal volumes as com-pared with traditional tidal volumes for acute

lung injury and the acute respiratory distresssyndrome: The Acute Respiratory DistressSyndrome Network. N Engl J Med 2000; 342:1301–1308

30. Dimitriou G, Greenough A, Kavvadia V, et al:Volume delivery during high frequency oscil-lation. Arch Dis Child Fetal Neonatal Ed1998; 78:F148–F150

31. Dimitriou G, Greenough A, Kavvadia V, et al:Comparison of two inspiratory: Expiratoryratios during high frequency oscillation. EurJ Pediatr 1999; 158:796–799

32. Courtney SE, Weber KR, Spohn WA, et al:Measurement of tidal volume using a pneu-motachometer during high-frequency oscil-lation. Crit Care Med 1990; 78:651–653

33. Sedeek KA, Takeuchi M, Suchodolski K, et al:Open-lung protective ventilation with pres-sure control ventilation, high-frequency os-cillation, and intratracheal pulmonary venti-lation results in similar gas exchange,hemodynamics, and lung mechanics. Anes-thesiology 2003; 99:1102–1111

34. Hirschl RB, Pranikoff T, Gauger P, et al.Liquid ventilatory in adults, children, andfull-term neonates. Lancet 1995; 346:1201–1202

35. Hirschl RB, Pranikoff T, Wise C, et al: Initialexperience with partial liquid ventilation inadult patients with the acute respiratory dis-tress syndrome. JAMA 1996; 275:383–389

36. Hirschl RB, Croce M, Gore D, et al: Prospec-tive, randomized controlled pilot study ofpartial liquid ventilation in adult acute respi-ratory distress syndrome. Am J Respir CritCare Med 2002; 165:781–787

37. Arnold JH: High-frequency oscillatory venti-lation and partial liquid ventilation: Liquidbreathing to a different beat (frequency). CritCare Med 2000; 28:2660–2662

38. Mammel M: High-frequency oscillation andpartial liquid ventilation. Crit Care Med2001; 29:1293

39. Fuhrman BP, Herman LJ, Rotta AT: Partialliquid ventilation: A multifaceted approachto acute respiratory distress syndrome. NewHorizons 1999; 7:433–439

40. Kirmse M, Fujino Y, Hess D, et al: Relation-ship between fill-volume during partial liq-uid ventilation (PLV) and the lower inflationpoint on the static pressure-volume curve.Am J Respir Crit Care Med 1998; 158:1550–1556

41. Doctor A, Ibla JC, Grenier EM: Pulmonaryblood flow distribution during partial liquidventilation. J Appl Physiol 1998; 84:1540–1550

42. Gothberg S, Parker TA, Abman SH, et al:

High-frequency oscillatory ventilation andpartial liquid ventilation after acute lung in-jury in premature lambs with respiratory dis-tress syndrome. Crit Car Med 2000; 28:2450–2456

43. Braden HP, Mellema JD, Bratton SL, et al:High-frequency oscillatory ventilation withpartial liquid ventilation in a model of acuterespiratory failure. Crit Care Med 1997; 25:299–302

44. Kinsella JP, Parker TA, Galan H, et al: Inde-pendent and combined effects of inhaled ni-tric oxide, liquid perfluorochemical, andhigh-frequency oscillatory ventilation in pre-mature lambs with respiratory distress syn-drome. Am J Respir Crit Care Med 1999;159:1220–1227

45. Smith KM, Bing DR, Meyers PA, et al: Partialliquid ventilation: A comparison using con-ventional and high-frequency techniques inan animal model of acute respiratory failure.Crit Care Med 1997; 25:1179–1186

46. Smith KM, Mrozek JD, Simonton SC, et al:Prolonged partial liquid ventilation usingconventional and high-frequency ventilatorytechniques: Gas exchange and lung pathol-ogy in an animal model of respiratory dis-tress syndrome. Crit Care Med 1997; 25:1888–1897

47. Sukuman M, Bommaraju M, Fisher JE, et al:High-frequency partial liquid ventilation inrespiratory distress syndrome: Hemodynam-ics and gas exchange. J Appl Physiol 1998;84:327–328

48. Doctor A, Massoni MC, DelBalzo U, et al:High-frequency oscillatory ventilation of theperfluorocarbon-filled lung: Preliminary re-sults in an animal model of acute lung in-jury. Crit Care Med 1999; 27:2500–2507

49. Doctor A, Price B, Bhargava N, et al: High-frequency oscillatory ventilation of the perfluo-rocarbon-filled lung: Dose-response relation-ships in an animal model of acute lung injury.Crit Care Med 2001; 29:847–854

50. Doctor A, Al-Khadra E, Tan P, et al: Extendedhigh-frequency partial liquid ventilation inlung injury: Gas exchange, injury quantifica-tion, and vapor loss. J Appl Physiol 2003;95:1248–1258

51. Tutuncu AS, Faithfull NS, Kachmann B:Intratracheal perfluorocarbon administra-tion combined with mechanical ventilationin experimental respiratory distress syn-drome: Dose-dependent improvement ofgas exchange. Crit Care Med 1993; 21:962–969


Computed tomography scan assessment of lung volume andrecruitment during high-frequency oscillatory ventilation

Thomas Luecke, MD; Peter Herrmann, PhD; Paul Kraincuk, MD; Paolo Pelosi, MD, PhD

Before the introduction of com-puted tomography (CT) scantechnology, imaging of theacute respiratory distress syn-

drome (ARDS) lung was limited to chestradiographs, which showed a widespreadbilateral appearance of pulmonary infil-trates. ARDS at that time was considered tobe a homogeneous disease of the lung pa-renchyma. Computed tomography showedthat in the majority of the patients withARDS, densities are located in the mostdependent part of the lung in supineposition, leaving the nondependent partrelatively well-aerated (1). In addition,Rouby and colleagues (2), performingCT scanning of the whole lung, showed

that the morphologic aspects of thelung parenchyma can be markedly dif-ferent between patients. The patientswere classified as having a “lobar” pat-tern if areas of lung attenuation had alobar or segmental distribution definedby the recognition of anatomic struc-tures such as the major fissures or theinterlobular septa, a “diffuse” pattern iflung attenuation was diffusely distrib-uted throughout the lungs, and a“patchy” pattern if there were lobar orsegmental areas of lungs attenuation insome parts of the lungs but withoutrecognized anatomic limits in others.Furthermore, they showed that patientswith lobar densities on CT were muchless recruitable at high positive end-expiratory pressure (PEEP) levels com-pared with patients with patchy-lobaror diffuse densities. Because compari-sons between CT and plain chest radi-ography have clearly shown the lowaccuracy of chest radiography in assess-ing lung morphology in ARDS (2, 3),CT scan is useful to evaluate regionaldistribution of disease, the nature ofthe infiltrates, and the potential for re-cruitment.

Effect of Edema

Regional analysis of the lung using CTshowed that the excess tissue mass,which likely derives from edema, is notdistributed according to gravity, butevenly distributed throughout the paren-chyma, from ventral to dorsal (4). ARDSis characterized by diffuse, increased per-meability, and edema increases equally ateach lung level, like a sponge filled bywater. The increased lung mass causescompression of the most dependent alve-oli as a result of the increased weight ofthe levels above in a gravitational field(5). When patients with ARDS were stud-ied in a prone position, we observed apartial redistribution of the densities,confirming that the nature of the depen-dent densities in supine position is par-tially reversible (6). Thus, the ARDS lung,in general, can be modeled as composedof normally inflated lung regions (mainlydistributed in the nondependent lung re-gions), consolidated lung regions (wide-spread along the lung parenchyma), andcollapsed–atelectatic lung regions(mainly distributed in the dependent partof the lung). The lung sponge model the-ory has been recently challenged (7, 8),

From the Departments of Anesthesiology and Crit-ical Care Medicine (TL, PH), University Hospital ofMannheim, Faculty of Clinical Medicine Mannheim,University of Heidelberg, Germany; the Department ofAnesthesiology and Critical Care Medicine, Universityof Vienna, Austria; and Dipartimento Ambiente, Salutee Sicurezza (PP), Università degli Studi dell’Insubria,Varese; Servizio di Anestesia e Rianimazione B, Os-pedale di Circolo, Fondazione Macchi, Varese, Italy.


DOI: 10.1097/01.CCM.0000155916.47455.DF

Objective: This review describes how computed tomographyhas increased our understanding of the pathophysiology of acuterespiratory distress syndrome. It summarizes current knowledgeabout lung volume changes and alveolar recruitment during high-frequency oscillatory ventilation (HFOV) assessed by computedtomography (CT), outlines potential problems when comparingHFOV with conventional ventilation (CV) as a result of the differentpressure-time profiles, and describes future research directions.

Data Source: CT allows accurate assessment of total lungvolumes and differentiation between overinflated, normally aer-ated, poorly aerated, and nonaerated lung regions. It allows forclassification of different patterns of consolidation and may bepredictive for the potential for recruitment.

Data summary: Experimental data suggest that HFOV at meanairway pressures (mPaw) set according to a static PV curve leadsto effective lung recruitment but results in overall lung volumesthat are considerably higher than those predicted from the PV

relationship. In saline-lavaged sheep, similar changes in totallung volumes and subvolumes were observed during HFOV andCV. One single study specifically assessed lung volume recruit-ment during HFOV as compared with CV in eight patients withacute respiratory distress syndrome from pneumonia or sepsis.After 48 hrs on HFOV, total ventilated lung volume was signifi-cantly increased, whereas only a minor increase in overinflatedlung volume was observed. These changes correlated with asignificant improvement in gas exchange.

Conclusion: CT is a valuable tool to quantify recruitment andoverinflation during HFOV. Additional studies are needed to bettercharacterize the specific effects of HFOV on lung volume andmorphology. (Crit Care Med 2005; 33[Suppl.]:S155–S162)

KEY WORDS: acute respiratory distress syndrome; high-fre-quency oscillatory ventilation; computed tomography; lung vol-ume; alveolar recruitment


suggesting that in the edematous lung,increasing air-space pressure first causesthe air–fluid interface to penetrate in thealveoli after which the air–fluid interfaceis inside the alveoli and the lung becomescompliant. The basic difference betweenthe two models is that in the spongemodel, edema is believed to be predomi-nantly interstitial (causing alveolar col-lapse by compression), whereas in theair–fluid interface model, edema is pre-dominantly intraalveolar. In both models,the total lung volume is nearly normal(9). The sponge model, however, does notexplain the entire ARDS lung behavior,because this compressive force is not theonly one inducing lung collapse. Puybas-set and coworkers (10) showed the max-imal lung volume loss is caudal, close tothe diaphragm. This suggests that thedecrease of transpulmonary pressure atthe lung base is not just the result of thesuperimposed pressure. It is likely thatthe weight of the heart (11, 12) and anincrease in intraabdominal pressure (12)also contribute to caudal atelectasis.

Effect of Prone Positioning

The prone position increases regionaltranspulmonary pressures and lung vol-umes by reducing the gravitational gra-dient of pleural pressure, which in turncan limit the shear forces and the alveolarvolume excursions in the dorsal regions.Similarly, the prone position causes amore homogeneous pulmonary bloodflow distribution and thus may limit cap-illary stress and reduce lung edema. Re-cent animal data using either healthy orinjured lung models have demonstratedless lung edema and histologic abnormal-ities with prone compared with supinepositioning (13). CT studies in supine pa-tients with ARDS showed that lung col-lapse occurs not only along a verticalgradient, but also along the cephalocau-dal axis. Recruitment maneuvers (RM)have been recommended as a useful toolto reopen collapsed lung regions and im-prove arterial oxygenation. In the supineposition, however, RM may potentiallyoverdistend alveoli already open besidesreopening collapsed dorsal regions. Theprone position, by reducing the pleuralpressure gradient, increases transpulmo-nary pressure in the dorsal regions,which increases lung recruitment of thedorsal regions and improves arterial oxy-genation when RM are applied comparedwith the supine position. This has been

demonstrated experimentally and in pa-tients with acute lung injury/ARDS (14).

PEEP is one of the most importanttools to increase the arterial oxygenationin severe respiratory failure. By opposingthe critical closing pressure (i.e., the su-perimposed pressure), PEEP can keep thelung open at end-expiration. As a result ofthe lower closing pressures and a morefavorable pulmonary blood flow distribu-tion (15) in the prone position, lowerPEEP levels are likely necessary for acomparable increase in arterial oxygen-ation compared with the supine position(14). CT scans have demonstrated thatthe ventral regions remain aerated atlower PEEP levels in the prone position,whereas the dorsal regions, previouslycollapsed in the supine position, becomeaerated. The same dorsal regions needhigher PEEP levels to remain aerated insupine position, which can cause overd-istension of the ventral regions (6).

Pulmonary VersusExtrapulmonary AcuteRespiratory Distress Syndrome

In recent years, a number of studieshave identified differences between pul-monary (ARDSp) and extrapulmonaryARDS (ARDSexp) using CT.

Goodman and colleagues (16) studied33 patients with ARDS (22 ARDSp and 11ARDSexp) by performing three represen-tative scans at the apex, at the hilum, andat the base of the lung. The ventilatorysetting was not standardized duringscans. The lung was scored as follows:“normal lung,” “ground-glass opacifica-tion” (mild increased attenuation withvisible vessels), and “consolidation”(markedly increased attenuation with novisible vessels). They found that in ARD-Sexp, ground-glass opacification wasmore than twice as extensive as consoli-dation. This contrasted markedly withARDSp, in which there was an even bal-ance between ground-glass opacificationand consolidation. When the type ofopacification between the two groups wascompared, the patients with ARDSexphad 40% more ground-glass opacificationthan did those with ARDSp. Conversely,the patients with ARDSp had over 50%more consolidation than did those withARDSexp. The authors also found differ-ences in the regional distribution of thedensities. In ARDSexp, ground-glassopacification was greater in the central(hilar) third of the lung than in the ster-nal or vertebral third. There was no sig-

nificant craniocaudal predominance ofground-glass opacification or consolida-tion, but consolidation showed a prefer-ence for the vertebral position over thesternal and central positions. In ARD-Sexp, ground-glass opacification wasevenly distributed in both the craniocau-dal and sternovertebral directions. Con-solidation tended to favor the middle andbasal levels, but also favored the vertebralposition. CT infiltrates were almostevenly distributed between the left andright lungs in both ARDSp and ARDSexp.However, grossly asymmetric disease wasalways the result of asymmetric consoli-dation. Moreover, the presence of airbronchograms and pneumomediastinumwere prevalent in ARDSp, whereas em-physema-like lesions (bullae) were com-parable in both groups. In contrast,Rouby and coworkers (2) found thatARDSp was more frequent among pa-tients with diffuse and patchy attenua-tion, whereas ARDSexp was more com-mon in patients with lobar attenuation.

These differences in underlying pa-thology and respiratory mechanics mayhave clinical consequences. In fact, thepotential for recruitment is higher in thepresence of alveolar collapse and lower ifalveolar consolidation predominates.Gattinoni and colleagues found that anincrease of PEEP leads to opposite effectson elastance (17). In ARDSp, increasingPEEP caused an increase of the elastanceof the total respiratory system as a resultof an increase in lung elastance with nochange in chest wall elastance. Con-versely, in ARDSexp, the application ofPEEP caused a reduction of the elastanceof to total respiratory system, mainly as aresult of a reduction in lung elastanceand chest wall elastance. Moreover, al-though an increase in PEEP led to anelevation of end-expiratory lung volumein both ARDSp and ARDSexp, it resultedin alveolar recruitment, primarily inARDSexp.

Computed TomographyAssessment of Lung Volumeand Recruitment During High-Frequency Oscillatory Ventilationand Conventional MechanicalVentilation: GeneralConsiderations

Although a large number of CT studieson lung volume and recruitment duringconventional ventilation (CV) have beenpublished over the last 10 years (9, 18),


those studies are still few during HFOV.In addition, comparison of CT findingsobtained during these inherently differ-ent modes will pose methodologic prob-lems. Although CV relies on bulk flow ofgas, associated with considerable pres-sure and volume changes throughout therespiratory cycle, HFOV oscillates arounda constant mean airway pressure (mPaw),yielding only minimal changes in distalpressure and volumes (19–22). CurrentHFOV practice is based on setting initialmPaw equal to, or 3–5 cm H2O higherthan, that observed during conventionalventilation (23–25). Although it would betempting to compare CT findings duringCV and HFOV at similar mean airwaypressures, it is of pivotal importance tokeep in mind that under a “stress andstrain” perspective (26), individual alveoliwill be exposed to highly changing pres-sures and volumes over the respiratorycycle during CV. Using CT of the wholelung, we have previously shown that CVpressure and volume changes are consid-erable, as reflected by the differences indensity distribution patterns from end-inspiration to end-expiration or calcula-tion of cyclic alveolar instability (27, 28).In addition, the amount of lung tissuecontinuously collapsing and reopeningduring the respiratory cycle may be sub-stantial, depending on the level of PEEPapplied (29, 30), and the amount of aer-ated tissue at end-expiration during CVhas been shown to depend on the amountof tissue recruited during previous inspi-rations (29–31). In contrast, distal pres-sure and volume changes during HFOVhave been shown to be small (22). There-fore, it is evident that the mean airwaypressure during CV and HFOV, even if thesame (on average), is not the same ifsingle alveolar units and more impor-tantly their “nature” (e.g., consolidation,atelectasis, or reduced diameter) andtheir location (nondependent or depen-dent, apical, or basal) are considered.During HFOV, the alveoli, which havebeen reopened and subsequently keptopen, are maintained at that mPaw for aprolonged period of time. In other words,the alveoli oscillate around a mean pres-sure volume, but the amplitude of thisoscillation occurs in a substantially lim-ited range of pressures and volumes. Incontrast, the “pressure-time profile” issubstantially different during CV,whereas alveolar excursions occuraround a greater gradient of pressuresand volumes. Another important point isthe expiratory phase, which is rarely con-

sidered. During HFOV, expiration again ismaintained across small pressure changes,whereas during CV, expiration is passive(coming “from high to low” pressures). Insummary, although arithmetic mean air-way pressure is similar, the oscillatory pres-sure amplitudes and volumes during HFOVare smaller compared with CV. We do notknow at present if this is of clinical impor-tance for the lung. Additional factors toconsider include: 1) the instability of thealveolar units (i.e., the presence of intersti-tial or alveolar edema); 2) the amount ofPEEP/mPaw, which is used during HFOVor CV; 3) the compliance of the chest wall,determining the effective transpulmonarypressure; 4) the nature of the densities(consolidation, collapse, or decreased vol-ume); and 5) the position of the patients(supine vs. prone). In fact, it may be possi-ble that, given the level of PEEP is suffi-cient to keep the alveoli open at end-expiration, the constant mPaw associatedwith HFOV may be more dangerous thanCV because alveoli are maintained contin-uously stretched, whereas in CV, there arerelaxation periods during the initial phaseof inspiration and the last phase of expira-tion. Support for this hypothesis is exem-plified by a recent study by Simonson et al.(32), in which lung injury was attenuatedby decreasing inspiratory time, thus pro-longing “pressure-relief time” for the alve-oli. Taking these considerations into thefield of CT analysis, future studies will haveto compare “time-weighted lung volumeand density averages.” For example, whenan inspiratory:expiratory (I:E) ratio of 1:1 isused during CV, the mean lung volume andHounsfield numbers obtained during end-inspiration and end-expiration rather mayhave to be added and divided by two, yield-ing the arithmetic mean. Decreasing theI:E ratio to 1:2 will consequently lower“time-weighted lung volume and density.”For the same reason, interpretation of theoxygenation index (OI � mPaw � FIO2/PaO2 � 100) when comparing CV andHFOV is difficult, because the “time-spent”at mPaw will be different for the twomodes. Therefore, CT lung images obtainedduring CV and HFOV at “comparable meanPaw” can appear substantially different de-pending on when in the respiratory cycle ofCV the image is obtained. In addition,changes in aeration may not necessarilyreflect equal changes in ventilation (33),because aeration in CT does not distinguishbetween open air spaces and those contain-ing trapped gas. Furthermore, dependingon the opening pressure for a given lungunit related to applied airway pressures,

that unit may or may not be effectivelyventilated. Because it is reasonable to as-sume that conditions will be different dur-ing CV and HFOV at comparable meanPAW as a result of the difference of pressurefluctuations over time, comparison be-tween the two modes ideally should includemeasurements of specific ventilation, i.e.,ventilation per given lung volume, usingXenon-enhanced CT (34). In light of thesepotential methodologic challenges, currentinsights derived from CT with respect tolung volume changes, recruitment, andoverdistension during HFOV are reviewed.

Computed Tomography Assessment ofLung Volume and Recruitment DuringHigh-Frequency Oscillatory Ventilationin Experimental Lung Injury. HFOV rep-resents a ventilatory strategy that po-tentially achieves all the goals of lung pro-tective ventilation. This potential, however,appears to be primarily dependent on theappropriate setting of mPaw and, thus, onan optimal lung volume strategy. Accord-ing to Froese (35), optimal lung volumecan be achieved by setting the mPaw dur-ing HFOV above the lower inflection point(Pflex) and the peak alveolar pressure belowthe upper inflection point on the inflationlimb of the static pressure-volume (P-V)curve of the respiratory system. Thus, thetwo “hazard zones” (35), namely the zoneof overdistension and the zone of derecruit-ment and atelectasis, can be avoided andHFOV is thought to take place within the“safe window” of the P-V curve. To test thishypothesis, our group undertook a study inwhich mean airway pressure was set ac-cording to the pressure-volume curve (22).The effect of varying mean airway pressureswas assessed in saline lung-lavaged pigs bycomparing slice lung volumes and aerationpatterns with those obtained during staticP-V recording at the lung apex and lungbase using CT. HFOV was initiated withmPaw set at Pflex, increased to 1.5 and 2times Pflex (termed 1.5*Pflexinc and2*Pflex) and decreased thereafter to 1.5times Pflex and Pflex (termed 1.5*Pflexdec

and Pflexdec). Using this strategy, HFOV atPaw 1.5*Pflexinc reestablished preinjuryPaO2 values. Further increases in mPaw didnot change oxygenation, but decreased ox-ygen delivery as a result of decreased car-diac output. No differences in respiratory orhemodynamic variables were observedwhen comparing HFOV at correspondingPaw during increasing and decreasing Paw.Variation of total slice lung volumes (TLV)was far less than expected from the staticP-V curve. Overdistended lung volume wasconstant and �3% of TLV. TLV during


HFOV at Pflex, 1.5*Pflexinc, and 2*Pflexwere significantly greater than TLV at cor-responding tracheal pressures on the infla-tion limb of the static P-V curve and locatednear the deflation limb. In contrast, TLVduring HFOV at decreasing Paw (i.e.,1.5*Pflexdec and Pflexdec) were not signifi-cantly greater than corresponding TLV onthe deflation limb of the static P-V curves(Fig. 1). These data are consistent with re-sults obtained by Goddon et al. (36), whoused the oxygenation response as a surro-gate marker for recruitment and found thatoptimal mPaw during HFOV was equal toPflex � 6 cm H2O. Of note, this pressurecorrelated with the point of maximum cur-vature on the deflation limb of the pres-sure-volume curve.

Based on these data, it can be con-cluded that during HFOV, mPaw setalong the static P-V curve leads to effec-tive and safe lung recruitment. Lung vol-umes during HFOV, however, are consid-erably higher than those at equivalentpressures on the inflation limb andreadily shifted up to or even beyond thevolumes on the deflation limb of the P-Vcurve. The minimal hysteresis observedover the range of mean airway pressurestested in this study may be the result ofthe fact that Pflex was already close toopen-lung mPaw and the result of theabsence of tidal recruitment duringHFOV (37). Therefore, it is concludedthat the “recruiting effect” of HFOV set atPflex and applied to the lung for a suffi-cient amount of time is largely underes-timated by the static P-V curve. It shouldbe emphasized, however, that these find-ings may not be specific for HFOV, be-cause similar results have been observedduring conventional ventilation (38, 39).

In fact, changes in lung volumes andrecruitment during HFOV may not beinherently different compared with con-ventional ventilation. In a preliminarystudy using fast electron-beam CT(EBCT) of the whole lung, we comparedtotal lung volumes as well as volumes ofoverinflated, normally aerated, poorlyand nonaerated lung at increasing airwaypressures during HFOV and volume-control ventilation in five saline-lavagedsheep. As shown for a representative an-imal in Figure 2, no significant differ-ences could be observed, except from theknown fact that mean airway pressuresduring HFOV tended to be higher forcomparable lung volume changes. Over-distended lung volume was low for allpressures tested. Starting from 118 mL inthe healthy lung at pressure control of 16

cm H2O and PEEP of 5 cm H2O (PC 16/5),maximum overdistension postinjury was150 mL during CV at PC 35/21 and 168mL during HFOV at mPAw of 45 cm H2O(Fig. 2). As mentioned here, however, di-rect comparison between HFOV and CV isdifficult and concepts based on the “timespent at mean airway pressure” and “spe-cific ventilation” need to be developed.

Computed Tomography Assessment ofLung Volume and Recruitment DuringHigh-Frequency Oscillatory Ventilationin Patients with Acute Respiratory Dis-tress Syndrome. To our knowledge, thereis only one study specifically assessinglung volume recruitment during HFOVas compared with conventional ventila-tion in patients with ARDS. Kraincuk etal. (40) studied eight patients with ARDSfrom pneumonia (n � 4) or postoperativesepsis (n � 4) with a mean lung injuryscore of 2.6 � 0.6 and pronounced atel-ectasis in at least two lung quadrants.Initially, all patients were conventionallyventilated in a pressure-limited mode(PCV) with a maximum inspiratory pres-sure of 35 cm H2O and a maximum PEEPof 15 cm H2O (Evita; Dräger, Luebeck,Germany). Respiratory rate (RR) was keptbelow 25 breaths/min and I:E was set at1:1. After adjustments of the airway pres-sures, RR was set according to the PaCO2

determined by arterial blood gas analyses,

allowing for permissive hypercapnia. FIO2

was titrated to keep PaO2 above 75 mmHg and SaO2 above 89%, respectively. Noformal recruitment maneuvers were per-formed. HFOV was initiated after a meanof 4.4 � 1.7 days of conventional venti-lation. HFOV was started at the same FIO2

used during PCV and the following ven-tilator settings: a high-frequency oscilla-tory rate of 5 Hz, an inspiratory timeratio of 33%, and a mPaw 5 cm H2Ohigher than the last mean airway pres-sure noted during PCV (CDPHFOV 28.4 �1.9 vs. mPawCMV 23.8 � 1.1 cm H2O).

Bias flow was set at 25 L/min. Theoscillatory pressure amplitude (�P) wasadjusted to keep the PaCO2 between 35and 69 mm Hg. FIO2 was adjusted as de-scribed for CMV. No recruitment maneu-vers were performed.

Baseline and 48-hr lung volumes onHFOV were determined by volumetry us-ing subsecond multislice spiral CT (Vol-ume Zoom; Siemens, Forchheim, Ger-many). Computed tomography of thewhole lung was performed within 25 secsat end-inspiratory hold during PCV andwithout turning the piston off duringHFOV. Total lung volume significantlyincreased during HFOV (4120 � 290 vs.4822 � 292 mL, p � .05) (Fig. 3). Nor-mally aerated volumes increased from3115 � 241 mL to 3920 � 244 mL (p �

Figure 1. Pressure-slice volume curves (upper panel) and pressure-slice density curves (lower panel)obtained at lung apex and base. Mean slice volumes (sV) and mean densities during high-frequencyoscillatory ventilation (HFOV) are added. Note that for reasons of readability, slice volumes aredisplayed as mean � SEM. *p � .05, sV (mean HU) during HFOV vs. sV (mean HU) at the correspondingtracheal pressure level during static P-V recording. Reproduced (22) with permission.


.05), whereas poorly aerated volumeswere found to be significantly reduced(650 � 54 mL vs. 510 � 52 mL, p � .05).Only a minor increase in hyperinflatedvolumes were observed (256 � 18 mL vs.

302 � 16 mL). No pneumothorax, pneu-mopericardium, or pneumomediastinumoccurred during the study period. Basedon these findings, the investigators con-cluded that HFOV for 48 hrs resulted in

alveolar recruitment in dependent lungareas, and improved gas exchange andarterial oxygenation (40).

Although in this study, no differencesregarding lung volume recruitment orgas exchange were observed between pa-tients with pulmonary ARDS (pneumo-nia) and extrapulmonary ARDS (sepsis),another recent study suggests a differenteffect of HFOV on extrapulmonary (ARD-Sexp) compared with pulmonary ARDS(ARDSp) (41). In this prospective obser-vational study, 30 adults (55 � 19 yrs)with ARDS were treated with HFOV afterfailure of CV (defined as plateau pressure�35 cm H2O, PEEP �10 cm H2O, FIO2

�0.6). HFOV was started at a frequencyof 5 Hz, an I:E ratio of 1:1, and a mPawequal to mPaw during CV � 5 cm H2O.The oscillatory pressure amplitude (�P)was titrated to achieve normocapnia. Sixhours after normocapnic HFOV, therewas no significant increase in FIO2/PaO2

in the ARDSp group (from 129 � 47 to133 � 50 mm Hg) but a significant im-provement in ARDSexp (from 114 � 54to 200 � 65 mm Hg, p � .01). Despitesimilar baseline mPaw on CV betweenARDSp and ARDSexp groups, optimalmPaw for the best FIO2/PaO2 duringHFOV was 20 � 6 cm H2O in ARDSp and28 � 6 cm H2O in ARDSexp (p � .01).The authors concluded that HFOV ismore effective in ARDSexp, using theFIO2/PaO2 response and oxygenation indexas a surrogate parameters for alveolar re-cruitment. The average duration of CVbefore HFOV was 7.7 � 6.4 days in thewhole study group, with 10.7 � 5.9 daysin ARDSp and 4.95 � 5.3 days in ARD-Sexp. As pointed out by the authors, theshorter lead time spent on CVin the ARD-Sexp group may have contributed to themore favorable response observed.

Selection of “Optimal Mean AirwayPressure” During High-Frequency Oscil-latory Ventilation: An Integrated Ap-proach. We propose an integrated ap-proach to optimize mPaw during HFOVby evaluating: 1) lung morphology bychest radiography and CT scan, 2) intra-abdominal pressure, 3) the etiology ofARDS, and 4) the response to PEEP/mPaw.

First of all, a chest radiography shouldbe performed. If lobar characteristics arepresent, poor response to increased meanairway pressure in terms of recruitmentcan be expected. In contrast, patchy ordiffuse injury on chest radiography can-not predict the response to increasedmean airway pressure. In this case, we

Figure 2. Computed tomography (CT) scans obtained at lung base before injury (baseline) at end-inspiration and end-expiration (pressure Control 16, positive end-expiratory pressure [Peep] 5) andpostinjury (pressure control 35, PEEP 0) in saline-lavaged sheep. CT scans during increasing levels ofPEEP (7, 14, 21 cm H2O) are shown on the left below. CT scans during high-frequency oscillatoryventilation (HFOV) at increasing mean airway pressures (25, 35, 45 cm H2O) are depicted on the right,along with histograms showing the density distribution for these scans. Total lung volumes andsubvolumes (overinflated: �1000 to �900 HU; normally aerated: �900 to �500 HU; reduced aerated:�500 to �100 HU; and nonaerated: �100 to �200 HU) for the different settings during conventionalventilation and high-frequency oscillatory ventilation illustrated above.


suggest performing a chest CT scan asearly as possible to evaluate the distribu-tion of the disease. The CT scan should beperformed at two different mean airway

pressures (increasing PEEP) to evaluatethe potential for recruitment. Second, wesuggest measuring intraabdominal pres-sure (IAP) by using the bladder pressure

technique (42). If IAP is lower than 16 cmH2O, we are confident that the airwaypressures likely reflect the real transpul-monary pressures, because the elastanceof the chest wall is likely to be withinnormal limits. If the IAP is �16 cm H2O,the mechanical properties of the chestwall are likely to be altered (17). Thus,the airway pressures selected should betitrated considering that at least 30% to70% of the airway pressures are lost toinflate the chest wall and not the lung(43). Third, the etiology of ARDS shouldbe considered in this setting. Ventilator-associated pneumonia is less likely to re-spond to increased airway pressures thancommunity-acquired pneumonia and ex-trapulmonary ARDS (17). Fourth, a PEEPtrial based on oxygenation, PaCO2, andrespiratory mechanics should be per-formed as proposed by Bohm and Lach-mann (44) and recently investigated inanimal (30, 39, 45) and human studies(29). It consists of a first part to open upthe lung (increasing plateau and PEEPlevels) and a second part to keep the lungopen (progressively decreasing the PEEPlevels). We believe that recruitment atthe bedside can be easily evaluated by theoxygenation response, whereas overdis-tension is likely if PaCO2 increases andcompliance of the respiratory system de-creases. Oxygenation has been found tocorrelate with recruitment in several ex-perimental (27, 30) and clinical studies(46, 47). On the other hand, the increasein PaCO2 and reduction in compliance ofthe respiratory system likely indicateoverstretching of the alveolar units asso-ciated with an increase in dead space(48). The mean airway pressure duringCV � 5 cm H2O is then used as the initialmPaw during HFOV. To account for po-tential rapid derecruitment during thetransition from CV to HFOV, a recruit-ment maneuver such as 40–45 cm H2Ofor 40–60 secs (with the piston turnedoff) can be administered first, as sug-gested by Derdak (24). Subsequently,mPaw should be decreased in steps of 2cm H2O. When reductions in PCO2 areobserved, continue to reduce the mPawdown to the level at which oxygenationstarts to deteriorate. Then, another re-cruitment maneuver is performed andmPaw is adjusted 2 cm H2O higher thanthat level.

CONCLUSION

CT data on the effects of HFOV inpatients with ARDS, although few, sug-

Figure 3. Representative computed tomography scans during conventional ventilation (CV) and 48 hrsafter initiation of high-frequency oscillatory ventilation (HFOV) in a patient with acute respiratorydistress syndrome caused by massive gastric aspiration. Total, normally aerated, poorly aerated, andoverinflated lung volumes during CV and after 48 hrs of HFOV for the individual patients. Time courseof mean FIO2/PaO2 and oxygenation index. OI, numbers indicated divided by 10 during CV (baseline,BL) and during HFOV. (Kraincuk et al., personal communication).


gest effective lung volume recruitmentwithout overt overdistension at pressuresthat may be lower than previously ex-pected, especially if recruitment maneu-vers are incorporated into the HFOVstrategy. Further studies are needed tobetter characterize the CT morphologicalterations induced by HFOV comparedwith CV. Finally, CT may be used to di-rectly validate new approaches to selectoptimum mPaw during HFOV based onthe attenuation of pressure swings alongthe endotracheal tube (19) or oscillatorypressure mechanics (21).

REFERENCES

1. Gattinoni L, Mascheroni D, Torresin A, et al:Morphological response to positive end expi-ratory pressure in acute respiratory failure.Computerized tomography study. IntensiveCare Med 1986; 12:137–142

2. Rouby JJ, Puybasset L, Cluzel P, et al: Re-gional distribution of gas and tissue in acuterespiratory distress syndrome. II. Physiolog-ical correlations and definition of an ARDSSeverity Score. CT Scan ARDS Study Group.Intensive Care Med 2000; 26:1046–1056

3. Puybasset L, Gusman P, Muller JC, et al:Regional distribution of gas and tissue inacute respiratory distress syndrome. III. Con-sequences for the effects of positive end-expiratory pressure. CT Scan ARDS StudyGroup. Adult Respiratory Distress Syndrome.Intensive Care Med 2000; 26:1215–1227

4. Sandiford P, Province MA, Schuster DP: Dis-tribution of regional density and vascularpermeability in the adult respiratory distresssyndrome. Am J Respir Crit Care Med 1995;151:737–742

5. Pelosi P, D’Andrea L, Vitale G, et al: Verticalgradient of regional lung inflation in adultrespiratory distress syndrome. Am J RespirCrit Care Med 1994; 149:8–13

6. Gattinoni L, Pelosi P, Vitale G, et al: Bodyposition changes redistribute lung comput-ed-tomographic density in patients withacute respiratory failure. Anesthesiology1991; 74:15–23

7. Hubmayr RD: Perspective on lung injury andrecruitment: a skeptical look at the openingand collapse story. Am J Respir Crit Care Med2002; 165:1647–1653

8. Wilson TA, Anafi RC, Hubmayr RD: Mechan-ics of edematous lungs. J Appl Physiol 2001;90:2088–2093

9. Gattinoni L, Caironi P, Pelosi P, et al: Whathas computed tomography taught us aboutthe acute respiratory distress syndrome?Am J Respir Crit Care Med 2001; 164:1701–1711

10. Puybasset L, Cluzel P, Chao N, et al: A com-puted tomography scan assessment of re-gional lung volume in acute lung injury. TheCT Scan ARDS Study Group. Am J RespirCrit Care Med 1998; 158:1644–1655

11. Malbouisson LM, Busch CJ, Puybasset L, etal: Role of the heart in the loss of aerationcharacterizing lower lobes in acute respira-tory distress syndrome. CT Scan ARDS StudyGroup. Am J Respir Crit Care Med 2000;161:2005–2012

12. Quintel M, Pelosi P, Caironi P, et al: Anincrease of abdominal pressure increasespulmonary edema in oleic acid-induced lunginjury. Am J Respir Crit Care Med 2004;169:534–541

13. Broccard A: Positive end-expiratory pressureor prone position: is that the question? CritCare Med 2003; 31:2802–2803

14. Gattinoni L, Tognoni G, Pesenti A, et al:Effect of prone positioning on the survival ofpatients with acute respiratory failure.N Engl J Med 2001; 345:568–573

15. Reutershan J, Schmitt A, Dietz K, et al: Non-invasive measurement of pulmonary bloodflow during prone positioning in patientswith early acute respiratory distress syn-drome. Clin Sci (Lond) 2004; 106:3–10

16. Goodman LR, Fumagalli R, Tagliabue P, etal: Adult respiratory distress syndrome dueto pulmonary and extrapulmonary causes:CT, clinical, and functional correlations. Ra-diology 1999; 213:545–552

17. Gattinoni L, Pelosi P, Suter PM, et al: Acuterespiratory distress syndrome caused by pul-monary and extrapulmonary disease. Differ-ent syndromes? Am J Respir Crit Care Med1998; 158:3–11

18. Rouby JJ, Puybasset L, Nieszkowska A, et al:Acute respiratory distress syndrome: lessonsfrom computed tomography of the wholelung. Crit Care Med 2003; 31:S285–295

19. van Genderingen HR, van Vught AJ, DuvalEL, et al: Attenuation of pressure swingsalong the endotracheal tube is indicative ofoptimal distending pressure during high-frequency oscillatory ventilation in a modelof acute lung injury. Pediatr Pulmonol 2002;33:429–436


21. Pillow JJ, Sly PD, Hantos Z: Monitoring oflung volume recruitment and derecruitmentusing oscillatory mechanics during high-frequency oscillatory ventilation in the pre-term lamb. Pediatr Crit Care Med 2004;5:172–180

22. Luecke T, Meinhardt JP, Herrmann P, et al:Setting mean airway pressure during high-frequency oscillatory ventilation accordingto the static pressure-volume curve in sur-factant-deficient lung injury: a computed to-mography study. Anesthesiology 2003; 99:1313–1322

23. Derdak S, Mehta S, Stewart TE, et al: High-frequency oscillatory ventilation for acute re-spiratory distress syndrome in adults: a ran-domized, controlled trial. Am J Respir CritCare Med 2002; 166:801–808

24. Derdak S: High-frequency oscillatory venti-

lation for acute respiratory distress syn-drome in adult patients. Crit Care Med 2003;31:S317–323

25. Mehta S, Granton J, MacDonald RJ, et al:High-frequency oscillatory ventilation inadults: the Toronto experience. Chest 2004;126:518–527

26. Gattinoni L, Carlesso E, Valenza F, et al:Acute respiratory distress syndrome, the crit-ical care paradigm: what we learned andwhat we forgot. Curr Opin Crit Care 2004;10:272–278

27. Luecke T, Roth H, Joachim A, et al: Effects ofend-inspiratory and end-expiratory pressureson alveolar recruitment and derecruitmentin saline-washout-induced lung injury—acomputed tomography study. Acta Anaesthe-siol Scand 2004; 48:82–92

28. Roth H, Luecke T, Deventer B, et al: Pulmo-nary gas distribution during ventilation withdifferent inspiratory flow patterns in experi-mental lung injury—a computed tomogra-phy study. Acta Anaesthesiol Scand 2004;48:851–861

29. Crotti S, Mascheroni D, Caironi P, et al:Recruitment and derecruitment duringacute respiratory failure: a clinical study.Am J Respir Crit Care Med 2001; 164:131–140

30. Pelosi P, Goldner M, McKibben A, et al: Re-cruitment and derecruitment during acuterespiratory failure: an experimental study.Am J Respir Crit Care Med 2001; 164:122–130

31. Richard JC, Brochard L, Vandelet P, et al:Respective effects of end-expiratory and end-inspiratory pressures on alveolar recruit-ment in acute lung injury. Crit Care Med2003; 31:89–92

32. Simonson DA, Adams AB, Wright LA, et al:Effects of ventilatory pattern on experimen-tal lung injury caused by high airway pres-sure. Crit Care Med 2004; 32:781–786

33. Tajik JK, Chon D, Won C, et al: Subsecondmultisection CT of regional pulmonary ven-tilation. Acad Radiol 2002; 9:130–146

34. Marcucci C, Nyhan D, Simon BA: Distribu-tion of pulmonary ventilation using Xe-enhanced computed tomography in proneand supine dogs. J Appl Physiol 2001; 90:421–430

35. Froese AB: High-frequency oscillatory venti-lation for adult respiratory distress syn-drome: let’s get it right this time! Crit CareMed 1997; 25:906–908

36. Goddon S, Fujino Y, Hromi JM, et al: Opti-mal mean airway pressure during high-frequency oscillation: predicted by the pres-sure-volume curve. Anesthesiology 2001; 94:862–869

37. Hickling KG: Best compliance during a dec-remental, but not incremental, positive end-expiratory pressure trial is related to open-lung positive end-expiratory pressure: amathematical model of acute respiratory dis-tress syndrome lungs. Am J Respir Crit CareMed 2001; 163:69–78

38. Downie JM, Nam AJ, Simon BA: Pressure-


volume curve does not predict steady-statevolume in canine lavage lung injury. Am JRespir Crit Care Med 2004; 169:957–962

39. Rimensberger PC, Pristine G, Mullen BM, etal: Lung recruitment during small tidal vol-ume ventilation allows minimal positive end-expiratory pressure without augmenting lunginjury. Crit Care Med 1999; 27:1940–1945

40. Kraincuk P, Ullrich R, Prokop M, et al: Mor-phological and functional effects of high-frequency oscillatory ventilation and super-imposed high-frequency jet ventilation inARDS—a volumetric CT study. In press

41. Pachl J, Roubik K, Fritz M: NormocapnicHFOV differently affects extrapulmonary andpulmonary ARDS. AIC news. Anesthesiologyand Intensive Care 2004; 54:S46–47

42. Iberti TJ, Lieber CE, Benjamin E: Determi-nation of intra-abdominal pressure using atransurethral bladder catheter: clinical vali-dation of the technique. Anesthesiology1989; 70:47–50

43. Pelosi P, Cereda M, Foti G, et al: Alterationsof lung and chest wall mechanics in patientswith acute lung injury: effects of positiveend-expiratory pressure. Am J Respir CritCare Med 1995; 152:531–537

44. Bohm S, Lachmann B: Pressure control ven-tilation: putting a mode into a perspective.International Journal of Intensive Care1996; 4:45–55

45. Rimensberger PC, Cox PN, Frndova H, et al:The open lung during small tidal volumeventilation: concepts of recruitment and ‘op-

timal’ positive end-expiratory pressure. CritCare Med 1999; 27:1946–1952

46. Gattinoni L, Pesenti A, Bombino M, et al:Relationships between lung computed tomo-graphic density, gas exchange, and PEEP inacute respiratory failure. Anesthesiology1988; 69:824–832

47. Maggiore SM, Jonson B, Richard JC, et al:Alveolar derecruitment at decremental posi-tive end-expiratory pressure levels in acutelung injury: comparison with the lower inflec-tion point, oxygenation, and compliance. Am JRespir Crit Care Med 2001; 164:795–801

48. Gattinoni L, Carlesso E, Cadringher P, et al:Physical and biological triggers of ventilator-induced lung injury and its prevention. EurRespir J Suppl 2003; 47:15s–25s


Noninvasive assessment of lung volume: Respiratory inductanceplethysmography and electrical impedance tomography

Gerhard K. Wolf, MD; John H. Arnold, MD

T he use of high-frequency oscil-latory ventilation (HFOV) withan “open-lung” approach toreverse atelectasis requires re-

liable assessment of lung volume and al-veolar recruitment. Currently, lung vol-ume during HFOV is assessed primarilyby arterial oxygenation, inflation of thelung on the patient’s chest radiograph,and subjective assessment of chest wallmovement.

It has been shown that lung injury inacute lung injury (ALI)/acute respiratorydistress syndrome is a heterogenous dis-ease and that regional differences in com-pliance are associated with distinct re-gional differences in opening and closingpressures (1–3). Although computed to-mographic (CT) images provide impor-tant information about alveolar collapseand reversal of atelectasis, the techniqueis not readily available at the bedside. Inaddition, it is associated with significantexposure to radiation and the risk oftransport of a critically ill patient.

Respiratory inductance plethysmogra-phy (RIP) and electrical impedance to-mography (EIT) are two monitoring tech-niques that have been used to nonin-vasively assess lung volume. The promiseof monitoring techniques such as RIPand EIT is that they will guide lung-protective ventilation strategies and allowthe clinician to optimize lung recruit-ment, maintain an open lung, and limitoverdistension. In the following review,we will describe these techniques in de-tail and propose how these new technol-ogies may have a significant future im-pact on the clinical setting.

RIP

RIP is a noninvasive respiratory mon-itoring technique that quantifies changesin the cross-sectional area of the chestwall and the abdominal compartment.The technique uses two elastic bands thatcontain Teflon-coated wires attached tothe bands in a zigzag form. One is typi-cally placed around the patient’s chest, 3cm above the xiphoid process, and thesecond is typically placed around the ab-domen. Each of these bands produces anindependent signal, and the sum of thesetwo signals is calibrated against a knowngas volume. The accuracy of RIP in the

determination of lung volume changesduring the use of HFOV has been inves-tigated in a number of animal studies.

Gothberg et al. (3a) assessed the accu-racy of RIP in measuring lung volumes ina newborn animal model. RIP-derivedlung volumes showed a good correlationwith injected gas volumes (r2 � .98–.99).Pressure–volume curves generated withRIP demonstrated good correlation withthe reference method. The authors fur-ther reported a change in RIP-derivedlung volume after surfactant instillation.This study showed that changes in lungcompliance may be detected with RIPduring a pressure–volume maneuver andconfirmed the accuracy of RIP measure-ments in the quantification of global lungvolumes.

In an animal model of ALI managedwith HFOV, Brazelton et al. (4) quantifiedpressure–volume curves using RIP duringa super-syringe maneuver and demon-strated close correlation with the referencemethod. RIP-derived lung volumes corre-lated with known lung volumes during su-per-syringe (r2 � .78). The authors reportthat RIP was capable of tracking volumechanges and creating a pressure–volumecurve during HFOV. Critical opening pres-sures were identified in three of five ani-mals during a recruitment maneuver with

From the Division of Critical Care Medicine, De-partment of Anesthesia, Children’s Hospital, Boston,MA.


DOI: 10.1097/01.CCM.0000155917.39056.97

Objective: Respiratory inductance plethysmography (RIP) andelectrical impedance tomography (EIT) are two monitoring tech-niques that have been used to assess lung volume noninvasively.

Methods: RIP uses two elastic bands around the chest andabdomen to assess global changes in lung volume. In animalmodels, RIP has been shown to detect changes in lung mechanicsduring high-frequency oscillatory ventilation and has the potentialto quantify lung volumes noninvasively. EIT measures regionalimpedance changes with 16 electrodes around the patient’schest, each of them injecting and receiving small currents. Im-pedance changes have been correlated with volume changes inanimal models and in humans. In a recent animal model, EIT wasshown to be capable of tracking lung volume changes duringhigh-frequency oscillatory ventilation.

Conclusion: The promise of monitoring techniques such as RIPand EIT is that they will guide lung protective ventilation strategiesand allow the clinician to optimize lung recruitment, maintain anopen lung, and limit overdistension. EIT is the only bedside methodthat allows repeated, noninvasive measurements of regional lungvolumes. In the future, it will be important to standardize the defi-nitions of alveolar recruitment and ultimately demonstrate the supe-riority of EIT-guided ventilator management in providing lung pro-tective ventilation. (Crit Care Med 2005; 33[Suppl.]:S163–S169)

KEY WORDS: electrical impedance tomography; respiratory in-ductance plethysmography; high-frequency oscillatory ventila-tion; acute lung injury; acute respiratory distress syndrome


HFOV. At high mean airway pressures, RIPconsistently underestimated lung volumes.Because chest wall expansion is limited, theauthors suggested that isovolumic pressurechanges may not be detected by RIP. Weberet al. (5) used a newborn-animal model todetect lung overdistension during HFOVusing RIP. Lung compliance calculated us-ing lung volumes detected by RIP weredifferent before and after induction of lunginjury by repetitive saline lavage. This find-ing demonstrates the potential for RIP todetect a change in lung mechanics duringHFOV and the potential to quantify lungvolumes noninvasively. Improvements inglobal lung compliance can be monitoredcontinuously, which may prevent inadver-tent lung overdistension produced by de-layed weaning of ventilator settings.

Using the same model of lung injury,the authors subsequently described amathematical prediction tool to optimizemean airway pressure during HFOV us-ing RIP (6). This model is based on theassumption that the time constant variessignificantly, depending on the state oflung inflation; different values for effec-tive lung compliance are derived from thespecific mean airway pressure. Accordingto this model, the optimal mean airwaypressure is indicated by the largest effec-tive compliance. The advantage of thisapproach is that only relative volumechanges are used; therefore, RIP does nothave to be accurately calibrated. In theclinical setting, the approach suggestedin this study is limited due to the inho-mogeneous pattern of alveolar collapse.Average effective compliance includes thecompliance of overdistended and atelec-tatic lung tissue.

A technical modification of RIP is fi-beroptic respiratory plethysmography. Inthis technique, a fiberoptic loop cable,circumferential to the chest, is used todetermine dynamic changes of the tho-racic wall circumference. In animals withinduced lung injury, it was possible todetect the frequency of chest wall oscil-lations generated during HFOV within3% accuracy. However, changes in theamplitude of oscillation could not bequantified (7).

RIP has been used in the clinical set-ting to quantify lung volume changesduring suctioning in mechanically venti-lated patients in two studies (8, 9). Mag-giore et al. (8) compared the loss of lungvolume during in-line vs. open suction-ing using RIP. Open suctioning caused agreater loss of lung volume than in-line

suctioning. Average lung volume loss af-ter disconnection vs. close suctioningwas 1466 mL and 531 mL, respectively (p� .01). Recruitment maneuvers usingthe triggering function of the ventilatorto deliver 40 cm H2O during endotra-cheal suctioning prevented the loss oflung volume during suctioning. In astudy in the pediatric population, Choonget al. (9) came to similar conclusions.These authors further noted that loss oflung volume was related to disconnectionfrom the ventilator rather than to suc-tioning and was most pronounced in pa-tients with low pulmonary compliance.These two studies show the potential ofRIP to contribute to the understanding ofderecruitment in the clinical arena.

Although inductive plethysmographyoffers the potential for noninvasive andcontinuous assessment of lung volume atthe bedside, a number of limitationsmust be acknowledged. As presently con-figured, inductive plethysmography al-lows the quantification of absolute vol-ume change only retrospectively, afterappropriate calibration. Although rapidassessment of lung volume change mayallow early detection of atelectasis andoverdistension, it must be emphasizedthat the peak-to-trough pressure change(amplitude) of HFOV cannot be reliablyquantified.

Finally, it must be acknowledged thateven the accurate assessment of lung vol-ume change with a portable, bedside de-vice has its limitations. Although manyhave suggested that the generation ofpressure–volume curves at the bedsideallows application of lung-protective ven-tilatory strategies and the estimation ofthe optimal mean airway pressure duringHFOV, it is important to emphasize thatthe information derived from pressure–volume curves is limited. CT images inhumans with ALI have convincinglyshown that recruitment occurs along theentire pressure–volume curve, well abovethe lower inflection point (10). Further-more, in a canine-model of acute respi-ratory distress syndrome, reversal ofatelectasis occurred up to airway pres-sures of 54 cm H2O (11), indicating alarge potential for recruitment in dorsallung areas at airway pressures consideredwell beyond the “safe zone.” Clearly, theassessment of global lung mechanicsdoes not provide information aboutchanges in regional lung volume or alter-ations in regional lung behavior.

EIT

The technical aspects of EIT were de-veloped over 20 yrs ago. Recent interestin this technology has been generated bya number of hardware and software im-provements that may allow accurate de-tection of regional lung volume change.The EIT hardware injects small amountsof electrical current sequentially, usingelectrodes applied circumferentially tothe patient’s chest. The receiving elec-trode calculates the voltage differentialand determines the impedance changebetween the transmitting and receivingelectrodes (Fig. 1). This creates a tomo-gram depicting the distribution of tissueelectrical properties in a cross-sectionalimage (Fig. 2). A cross-sectional image ofthe lung, composed of 1024 data points ina 32 � 32 array, is created using a math-ematic algorithm called “back-projec-tion” (12, 13). The thickness of this cross-sectional “slice” of the thorax variesdepending on the circumference of thechest and is typically between 15 and 20cm (10). It is important to emphasize

Figure 1. Generating an electrical impedance to-mographic image: 16 electrodes around the chestinject and receive small currents in a rotatingfashion.

Figure 2. Electrical impedance tomographic im-age of the lung. The orientation is similar to acomputed tomographic image. Both lung fieldsshow equal impedance change during spontane-ous breathing.


that no absolute impedance values aregenerated; rather, impedance changes aregenerated by referencing all measuredvoltages to a baseline measurement.

One of the currently available systemsis the “Goe MF II” system that was devel-oped at the University of Goettingen, Ger-many, and is now distributed by Viasys.Other systems used in the past were theSheffield Mark 1 system and the DAS01-P system (11, 14). A comparison studyof the three systems showed that the GoeMF II system has the most beneficial sig-nal to noise ratio and allows dynamicmeasurements at low lung volumes (15).Recently, a separate group of investiga-tors has used their own EIT system tovalidate regional lung volume changes incomparison with CT (MB Amato, per-sonal communication, 2004).

In current systems, 16 electrodes areapplied around the thorax of the patient,although a prototype device that uses 32electrodes is currently being developed.The scanning rate of the Goe MF II sys-tem is between 13 and 44 scans/sec (Hz).This means that the system can generateup to 44 cross-sectional images per sec-ond. Impedance measurements of up to30 mins in length are possible, whichproduces a data set of approximately 30megabytes. Interference by other electri-cal devices commonly found in the inten-sive care environment has not been sys-tematically investigated. Our ownexperience in the clinical setting suggeststhat a significant amount of interferenceis produced by the 50 mA currents typi-cally used in the respiratory module ofcombined heart rate–apnea monitorswidely used in this country. We have re-cently applied a software modificationthat shifts the EIT current out of therange of the electrocardiograph elec-trodes, and the signal showed less inter-ference.

The advantage of EIT over RIP is thequantification of regional and globalchanges in lung volume. During off-lineanalysis, a functional impedance imageand the time course of impedance changevs. time can be displayed. The so-calledfunctional EIT image is a virtual EITmonitor displaying impedance changeswithin the lung. SD of impedance changeis averaged over 4 secs in this display.This feature provides the investigatorwith an almost continuous real-time im-age of the impedance changes that occurduring the ventilatory cycle and are likelydirectly related to alterations in regionalgas volume.

Local impedance change vs. time im-ages can be generated by selecting a spe-cific region of interest within the tomo-gram. In addition to impedance changes,ventilatory pressures can be recorded si-multaneously using a pressure module.Impedance changes can be correlatedwith airway pressures at a specific timeand provide important information aboutregional changes in lung mechanical be-havior in response to alterations in ven-tilator settings.

The proper interpretation of changesin lung impedance must account for thefact that there are multiple tissues in thethorax and that changes in intrathoracicpressure produce complex interactionsbetween the intrathoracic organs. In gen-eral, an increase in aerated lung volumeresults in a positive impedance changeand a decrease in aerated lung volumeproduces a negative impedance change.

The overall resistivity of lung tissue isabout five times greater than the resistiv-ity of other soft-tissue organs. Typicalvalues of tissue resistivity at a frequencyof 10 kHz are: lungs, 10 � m (changeswith respiratory phase); muscle, 2–4 �m; fat, 20 � m; blood, 1.6 � m; and bone,�40 � m (13). In addition, it is essentialto recognize that intrathoracic blood vol-ume and pulmonary and aortic blood flowinfluence the EIT signal. Typically, anincrease in pulmonary blood volume re-sults in a decrease in relative impedance.

Cardiopulmonary interactions duringmechanical ventilation influence the EITsignal as follows: a higher mean airwaypressure during HFOV will, to some ex-tent, lower pulmonary blood volume be-cause blood is displaced due to increasingalveolar pressure. The displacement ofblood away from the region of interestwill increase measured impedance. Thisincrease in relative impedance changedue to displacement of blood volume oc-curs simultaneously with an increase ofimpedance change secondary to increas-ing lung volume. It may be difficult todifferentiate between the two effects be-cause they occur at the same time.

In addition, the relative impedance isaltered by changes in both right- andleft-heart output (16). The cardiac out-put-related impedance change is signifi-cantly smaller than impedance changescaused by cardiopulmonary interactions.Both physiologic phenomena—alterationof intrathoracic blood volume and bloodflow— can be altered by either directchanges in cardiac function or, second-

arily, by cardiopulmonary interactionduring mechanical ventilation.

In an animal model of ALI managedwith HFOV, van Genderingen et al. (17)assessed regional lung mechanics withEIT (18). Regional and local impedancechange vs. time plots were described dur-ing an inflation–deflation maneuver onHFOV. EIT measurements were com-pared with strain-gauge plethysmographyand helium dilution. During a pressure–volume maneuver using a super-syringemethod, the authors reported a regionalinhomogeneity of lung disease, mani-fested by different shapes of the pressure–volume curves in dependent (collapsed)and nondependent (recruited) areas. Thisfinding has previously been described byother authors and is related to collapse ofalveolar tissue along the gravitationalaxis (19–21). During the same pressure–volume maneuver using HFOV, the pres-sure-impedance curves in dependent andnondependent lung areas were more ho-mogeneous. The authors suggested thatHFOV has a “homogenizing effect” andthat alveolar recruitment is achieved bythe opening of “sticky airways” duringHFOV. This finding was not verified byother imaging techniques, such as rever-sal of atelectasis documented by chestCT. This is the first study demonstratingthat EIT is capable of tracking lung vol-ume changes during HFOV in an animalmodel. It also shows the significance ofEIT as a functional imaging technique toobtain more insight regarding recruit-ment phenomena during HFOV.

Frerichs et al. (21) compared regionalatelectasis using EIT and CT in an animalmodel of ALI. The animals were venti-lated with five different tidal volumes atthree different PEEP levels. Local air con-tent was compared with CT and EIT inventral, middle, and dorsal lung areas on

Figure 3. Comparison of a computed tomo-graphic (CT) scan and an electrical impedancetomographic (EIT) image during mechanicalventilation of a patient with acute lung injury.The gray and white areas in the EIT image reflectareas with impedance change (volume changes)and correlate to well-inflated areas in the CTscan. There is no impedance change on the EITscan in the dorsal areas, where atelectasis ispresent on the CT scan (23).


Figure 4. Three-dimensional array of recruitment after suctioning on high-frequency oscillatory ventilation. The SD of impedance change after reconnectionto the ventilator is displayed.

Figure 5. Recruitment on high-frequency oscillatory ventilation after suctioning in four different quadrants of the lung.


the left and on the right side of the lung.The correlation was strongest in the de-pendent areas (r2 � .86) and acceptablein nondependent areas (r2 � .66). Theauthors attributed the worsening corre-lation in nondependent areas to move-ment artifacts during tidal ventilationthat increased variability in the lung den-sity measured by CT. These artifacts donot occur during EIT. Time resolution ofthe EIT scans (13 scans/sec) was higherthan time resolution of CT scans (3.3scans/sec). The significance of this studyis that it confirmed the correlation ofimpedance changes and volume changesusing CT as a reference method in ananimal model.

Kunst et al. (20) identified regionalpressure–volume curves in an animalmodel with ALI. Pressure volume curvesobtained with the super-syringe methodwere correlated to pressure-impedancecurves determined by EIT. The authorswere able to show distinct differences inregional pressure-impedance curves bydividing the lung into four different lay-ers. The lower inflection point was 8 cmhigher in the most dependent lung com-pared with the nondependent part of the

lung. The authors attributed these differ-ences to increased collapse of atelectasis-prone lung in dorsal areas after the in-duction of ALI. This study demonstratesthe potential for EIT to detect regionalalveolar collapse during mechanical ven-tilation.

A dynamic approach to assess regionalrecruitment was described in a similaranimal model by the same authors. Dur-ing conventional ventilation, regionalimpedance changes over time were com-pared in dependent and nondependent ar-eas (19). During ALI, impedance changeswere decreased in dorsal lung areas, sug-gesting regional collapse. After a recruit-ment maneuver, the authors reportedimproving regional impedance changesin previously collapsed areas. In theory, aratio of equal impedance changes in de-pendent and nondependent parts of thelung suggests homogeneous distributionof ventilation to all lung regions. How-ever, this concept has not yet been asso-ciated with improved physiologic orpathologic outcomes. Therefore, it re-mains speculative whether the equaliza-tion of impedance changes in all lung

areas detected by EIT scans produces op-timal lung protection.

The correlation of regional impedancechanges in patients with ALI with lungdensity measurements by CT has beenrecently verified in a study by Victorino etal (22). In this study, a slow inflationmaneuver was recorded with EIT in tenintubated and mechanically ventilatedadult patients. Subsequently, the sameinflation maneuver was performed in theCT scanner. Air content seen on the CTscan was compared with regional imped-ance changes in the EIT slice (Fig. 3).Both techniques detected imbalances inventilation of dependent and nondepen-dent lung areas (upper/lower ratio �82%/18% and 75%/25% for EIT and CT,respectively). Regional impedancechanges on the EIT image showed goodcorrelation (r2 � .92) with changes in aircontent detected by CT. Furthermore,EIT scans showed good reproducibility(SD, 4.9 %) between repeated measure-ments on the same patient. It should benoted that CT and EIT images were notobtained simultaneously due to electro-magnetic interference of the EIT equip-ment in the CT scanner. Nevertheless,

Figure 6. Oscillations during high-frequency oscillatory ventilation. This 9-yr-old patient with acute respiratory distress syndrome was ventilated with theSensorMedics 3100A (Viasys, Yorba Linda, CA) at a rate of 6 Hz. The electrical impedance tomographic scanning rate was 44 Hz. Impedance changes ofsix cycles per second are displayed after filtering the heart rate.


the authors attempted to minimize thismethodologic problem by averaging EITscans obtained before and after CT and bythe use of standardized recruitment ma-neuvers before each inflation sequence.This study confirmed that regional im-pedance changes are closely correlatedwith regional volume changes identifiedby CT. This is relevant to the future use ofEIT at the bedside, where ventilation andrecruitment strategies could potentiallytarget regional atelectasis identified bythe EIT image.

Frerichs et al. (23, 24) used EIT todescribe the effect of gravity on regionalalveolar collapse. Under microgravity in-duced by a parabolic flight in seven spon-taneously breathing volunteers, depen-dent areas showed larger impedancechanges (better ventilation) than undernormal circumstances. This finding sug-gests that gravity causes regional collapsein spontaneously breathing patients.

Our group has utilized EIT to detectlung volume changes during a standard-ized suctioning maneuver in childrenventilated with HFOV (SensorMedics3100A, Viasys, Yorba Linda, CA). We wereable to demonstrate that EIT can be usedto detect and quantify regional lung vol-ume changes in the pediatric populationduring disconnection from the ventilatorand suctioning. We have demonstratedconsiderable regional heterogeneity involume changes during a derecruitmentmaneuver. The SD of impedance changebetween suctioning and reinstitution ofHFOV in a sample patient is displayed ina three-dimensional array in Figure 4.Local impedance vs. time plots of fourquadrants demonstrate a significant dif-ference in impedance change during re-recruitment, with larger recruitmentoccurring in nondependent lung areas(Fig. 5).

The experimental data regarding cur-rent EIT technology suggest a reliablecorrelation between global lung volumechanges and global impedance changes inmultiple animal models. Furthermore,regional impedance changes have dem-onstrated acceptable correlation with re-gional changes in lung volume seen onCT during mechanical ventilation. Alveo-lar recruitment has been demonstratedusing EIT by generating regional pres-sure-impedance curves for different lungareas. In the clinical setting, the mostnoncompliant area with the highestlower inflection point can be carefullymonitored to apply adequate airway pres-sure for recruitment. Furthermore, by

calculation of upper-to-lower lung im-pedance ratios, the clinician may be ableto optimize the distribution of ventilationand, in particular, prevent the overdisten-sion of nondependent lung during ag-gressive lung recruitment maneuvers.

It must be noted that the use of cur-rent EIT technology to assess alveolarrecruitment during HFOV has a numberof technical limitations. The tidal vol-umes to be detected during HFOV aremuch smaller than during conventionalventilation; in small infants, deliveredtidal volumes are between 1.5 and 2mL/kg and may range up to 6 mL/kg inlarger patients managed with lower oscil-latory frequencies (25). The resolutionrequired to detect impedance changes ofthis magnitude are at the limit of thecurrent hardware/software specifications.We have recently modified the softwarefiltering function and increased the sam-pling rate to detect impedance changesproduced by the peak-to-trough oscilla-tory pressure waveform generated duringHFOV (Fig. 6).

Conclusion

Neither RIP nor EIT are technicallyperfect at this point. EIT is the only bed-side method that allows the potential forrepeated, noninvasive measurements ofregional lung volume changes. It offersthe potential to detect regional atelectasisat the bedside and to alter the ventilationstrategy in real time to reverse it. Anoscillatory pressure waveform deliveredat 5–15 Hz during HFOV presents tech-nical challenges that limit the accuracy ofcurrent devices. As EIT technologyevolves to meet these challenges, it willbe important for investigators to stan-dardize the definitions of alveolar recruit-ment, explore the optimal relationshipsbetween regional impedance change ra-tios, and ultimately, to demonstrate thesuperiority of EIT-guided ventilator man-agement in providing lung-protectiveventilation.

ACKNOWLEDGMENT

We thank Bartlomiej Grychtol for as-sistance in preparing the figures for thisarticle.

REFERENCES

1. Gattinoni L, Presenti A, Torresin A, et al:Adult respiratory distress syndrome profilesby computed tomography. J Thorac Imag-ing 1986; 1:25–30

2. Gattinoni L, D’Andrea L, Pelosi P, et al:Regional effects and mechanism of positiveend-expiratory pressure in early adult respi-ratory distress syndrome. JAMA 1993; 269:2122–2127

3. Gattinoni L, Caironi P, Pelosi P, et al: Whathas computed tomography taught us aboutthe acute respiratory distress syndrome?Am J Respir Crit Care Med 2001; 164:1701–1711

3a.Gothberg S, et al: Lung volume recruitmentin lambs during high-frequency oscillatoryventilation using respiratory inductiveplethysmography. Pediatr Res 2001; 49:38–44

4. Brazelton TB III, Watson KF, Murphy M, etal: Identification of optimal lung volumeduring high-frequency oscillatory ventila-tion using respiratory inductive plethys-mography. Crit Care Med 2001; 29:2349–2359

5. Weber K, Courtney SE, Pyon KH, et al:Detecting lung overdistention in newbornstreated with high-frequency oscillatory ven-tilation. J Appl Physiol 2000; 89:364–372

6. Habib RH, Pyon KH, Courtney SE: Optimalhigh-frequency oscillatory ventilation set-tings by nonlinear lung mechanics analysis.Am J Respir Crit Care Med 2002; 166:950–953

7. Davis C, Mazzolini A, Mills J, et al: A newsensor for monitoring chest wall motionduring high-frequency oscillatory ventila-tion. Med Eng Phys 1999; 21:619–623

8. Maggiore SM, Lellouche F, Pigeot J, et al:Prevention of endotracheal suctioning-induced alveolar derecruitment in acutelung injury. Am J Respir Crit Care Med2003; 167:1215–1224

9. Choong K, Chatrkaw P, Frndova H, et al:Comparison of loss in lung volume withopen versus in-line catheter endotrachealsuctioning. Pediatr Crit Care Med 2003;4:69 –73

10. Blue RS, Isaacson D, Newell JC: Real-timethree-dimensional electrical impedance im-aging. Physiol Meas 2000; 21:15–26

11. Frerichs I: Electrical impedance tomogra-phy (EIT) in applications related to lungand ventilation: A review of experimentaland clinical activities. Physiol Meas 2000;21:R1–R21

12. Barber DC: A review of image reconstruc-tion techniques for electrical impedance to-mography. Med Phys 1989; 16:162–169

13. Brown BH: Electrical impedance tomogra-phy (EIT): A review. J Med Eng Technol2003; 27:97–108

14. Hahn G, Beer M, Frerichs I, et al: A simplemethod to check the dynamic performanceof electrical impedance tomography sys-tems. Physiol Meas 2000; 21:53–60

15. Hahn G, Thiel F, Dudykevych T, et al: Quan-titative evaluation of the performance ofdifferent electrical tomography devices.Biomed Tech (Berl) 2001; 46:91–95


16. Smit HJ, Vonk Noordegraaf A, Marcus JT, etal: Determinants of pulmonary perfusionmeasured by electrical impedance tomogra-phy. Eur J Appl Physiol 2004; 92:45–49

17. van Genderingen HR, van Vught AJ, JansenJR: Regional lung volume during high-frequency oscillatory ventilation by electri-cal impedance tomography. Crit Care Med2004; 32:787–794

18. Arnold JH: Electrical impedance tomogra-phy: On the path to the Holy Grail. CritCare Med 2004; 32:894–895

19. Kunst PW, Vazquez de Anda G, Bohm SH,et al: Monitoring of recruitment and dere-cruitment by electrical impedance tomog-

raphy in a model of acute lung injury. CritCare Med 2000; 28:3891–3895

20. Kunst PW, de Vries PM, Postmus PE, et al:Evaluation of electrical impedance tomog-raphy in the measurement of PEEP-inducedchanges in lung volume. Chest 1999; 115:1102–1106

21. Frerichs I, Hinz J, Herrmann P, et al: De-tection of local lung air content by electri-cal impedance tomography compared withelectron beam CT. J Appl Physiol 2002; 93:660–666

22. Victorino JA, Borges JB, Okamoto VN, et al:Imbalances in regional lung ventilation: Avalidation study on electrical impedance to-

mography. Am J Respir Crit Care Med 2004;169:791–800

23. Frerichs I, Dudykevych T, Hinz J, et al:Gravity effects on regional lung ventila-tion determined by functional EIT duringparabolic flights. J Appl Physiol 2001; 91:39 –50

24. Frerichs I, Hahn G, Hellige G: Gravity-dependent phenomena in lung ventilationdetermined by functional EIT. Physiol Meas1996; 17(Suppl 4A):A149–A157

25. Sedeek KA, Takeuchi M, Suchodolski K, etal: Determinants of tidal volume duringhigh-frequency oscillation. Crit Care Med2003; 31:227–231


Clinical use of high-frequency oscillatory ventilation in adultpatients with acute respiratory distress syndrome

Kenneth P. W. Chan, MBBS, MMed, FCCP; Thomas E. Stewart, MD, FRCPC

W ith the publication of arandomized, controlledtrial documenting the ef-ficacy of a low–tidal vol-

ume (V̇T) ventilatory strategy in reducingmortality compared with a higher V̇T

strategy, there has been renewed appre-ciation that mechanical ventilation in apatient with acute respiratory distresssyndrome (ARDS), although necessary, isoften injurious and may contribute tofurther lung damage (1, 2). Positive-pressure ventilation may injure the lungvia several different mechanisms, includ-

ing alveolar overdistension (volutrauma),repeated closing and opening of collapsedalveolar units (atelectrauma), and oxygentoxicity (3). In addition, these mecha-nisms seem to lead to the release ofproinflammatory cytokines, which fur-ther perpetuate the ongoing lung damageand may have a role in multiple organdysfunction syndrome, which is often en-countered in these patients (4, 5).

An understanding of these processeshas led to the search for ventilatory strat-egies that are “lung-protective” (i.e., fur-ther lung injury is minimized while stillmaintaining adequate oxygenation andsupport to off-load the respiratory mus-cles). This involves the use of limited V̇T

to avoid alveolar overdistension, adequateend-expiratory lung volumes utilizingpositive end-expiratory pressure (PEEP),and higher mean airway pressures(maws) to minimize atelectrauma and

improve oxygenation. In a randomized,controlled trial conducted by the Na-tional Institutes of Health’s ARDS Net-work, a ventilatory strategy using con-ventional ventilation (CV), low tidalvolumes (V̇T � 6 mL/kg predicted bodyweight) and limited inspiratory plateaupressures (�30 cm H2O) was associatedwith an absolute reduction in mortality of9% when compared with a larger V̇T strat-egy (target, 12 mL/kg) (2). Although thisis exciting, there are still patients withARDS who are unable to achieve adequategas exchange while receiving CV, and themortality rate of these patients remainshigh (6).

High-frequency oscillatory ventilation(HFOV) is a unique mode of mechanicalventilation first used successfully in theneonatal population (7–9). It is charac-terized by rapid oscillations of a recipro-cating diaphragm, leading to high–

From the Inter-Departmental Division of CriticalCare, Department of Medicine and Anaesthesiology,University of Toronto, Mount Sinai Hospital and Uni-versity Health Network, Toronto, Ontario, Canada.


DOI: 10.1097/01.CCM.0000155915.97462.80

Objective: High-frequency oscillatory ventilation (HFOV) is anemerging ventilatory strategy for adults that has been used suc-cessfully in the neonatal and pediatric population. This modalityutilizes high mean airway pressures to maintain an open lung andlow tidal volumes at a high frequency that allow for adequateventilation while at the same time preventing alveolar overdis-tension. With the current understanding that excessive lungstretch and inadequate end-expiratory ventilatory volume may beinjurious to the lungs, HFOV seems to be the ideal lung-protectiveventilatory mode. During the past 8 yrs, there have been increas-ing numbers of studies describing its use in adult patients withacute respiratory distress syndrome. This article aims to reviewthe published studies of HFOV in adults with acute respiratorydistress syndrome with regard to its safety and efficacy.

Data Source: To assist us with our review, we did a search ofMEDLINE (from 1966 to present) and EMBASE (1980 to present)databases to identify adult, clinical, English-language, researcharticles related to HFOV use. In addition, we reviewed relevantanimal and mechanical ventilation studies. We did not perform aformal systematic review.

Data Synthesis: The application of HFOV was mainly reportedas a rescue ventilatory mode in adult patients with acute respi-ratory distress syndrome who were thought to have failed con-ventional ventilation. In these patients, HFOV has consistently

been shown to improve oxygenation without obvious increases incomplications measured. There was only one randomized, con-trolled trial comparing HFOV with conventional ventilation. Thisstudy showed that there was a nonsignificant trend toward alower mortality rate in the HFOV group. In addition, HFOV was aseffective and safe as conventional ventilation. Although there arelimitations, multiple studies have shown that earlier initiation ofHFOV in patients with severe acute respiratory distress syndromemay also be associated with a lower mortality.

Conclusions: HFOV seems to be safe and effective for adults withsevere acute respiratory distress syndrome who have failed conven-tional ventilation. Further research is needed to determine the idealpatients, timing, and optimal technique with which to provide HFOV.When considering HFOV as an early, lung-protective mode of venti-lation, there is still a need to perform an adequately powered,randomized, controlled trial comparing it with the best available formof conventional ventilation. However, we believe that such a trialshould wait until we have a better understanding of HFOV in adults.(Crit Care Med 2005; 33[Suppl.]:S170–S174)

KEY WORDS: acute respiratory distress syndrome; high-fre-quency ventilation; high-frequency oscillatory ventilation; me-chanical ventilation; respiratory failure; ventilator-induced lunginjury; mortality; recruitment maneuver


respiratory cycle frequencies, usuallybetween 3 and 9 Hz in adults, and verylow V̇T. Although no data in adult patientswith ARDS exist as yet, measurements ina sheep model of ARDS suggests that V̇T

delivered during HFOV is about 1–4 mL/kg, depending on the frequency (10).HFOV is conceptually very attractive, as itachieves many of the goals of lung-protective ventilation. The application ofa constant mPaw maintains an “openlung” and optimizes lung recruitment.Ventilation in HFOV is primarily achievedby oscillations of air around the setmPaw, usually at much lower V̇Ts thanthose achieved with CV, thus theoreti-cally avoiding alveolar overdistension.The mechanism of ventilation duringhigh-frequency ventilation is signifi-cantly different from CV and is describedelsewhere (11). Importantly, unlike otherforms of high-frequency ventilation, ex-piration is active during HFOV. This islikely an essential mechanism in the pre-vention of gas trapping and the optimi-zation of ventilation.

Comparative studies in animals indeedsuggest that HFOV reduces ventilator-induced lung injury (12, 13). Imai et al.compared pathophysiologic and bio-chemical markers of acute lung injury ina saline-lavaged rabbit model of ARDSwith different ventilatory strategies: acontrol group utilizing CV with moderateV̇T (10–12 mL/kg) and three differentlung-protective strategies: 1) low V̇T (5–6mL/kg) with PEEP 2–3 cm H2O higherthan the lower inflection point utilizingCV, 2) low V̇T with PEEP of 8–10 cm H2O,and 3) HFOV (12). The group utilizingHFOV had lower neutrophil infiltration,lower levels of tumor necrosis factor, anddecreased pathologic changes in the alve-olar spaces when compared with theother lung-protective groups.

In addition, as higher mPaws areachievable with HFOV when comparedwith CV, HFOV could theoretically lead toimprovements in oxygenation and, inturn, a reduction in FIO2. Certainly, in theneonatal literature, the use of HFOV hasbeen associated with improvements inoxygenation, a decrease in surfactant use,and possibly, a decrease in lung injury asassessed by a reduction in the need forsupplemental oxygen and a lower roent-genographic score at 30 days (7, 9).

In view of the success of HFOV in theneonatal and pediatric population and ofits theoretical physiologic advantages inregard to lung-protective ventilation,there has been accumulating interest in

the application of this ventilatory modein adult patients with ARDS. This articleaims to review the published experiencewith HFOV in adult patients with ARDS.We will describe the safety and efficacy ofthis relatively new mode of ventilation interms of its complications and clinicallyrelevant outcomes.

Clinical Studies of HFOV inAdult Patients

The first study we reviewed describedHFOV use in 17 adult patients withARDS, mainly due to sepsis and pneumo-nia, who failed CV (14). Failure of me-chanical ventilation was defined by anyone of the following criteria: an FIO2 of �0.7 with a PaO2 of � 65 mm Hg, a peakinspiratory pressure of �65 cm H2O, or aPEEP of � 15 cm H2O. The patients hada mean PaO2/FIO2 ratio of 68.6 mm Hg, amean oxygenation index (OI � FIO2 �mPaw � 100/PaO2) of 48.56, and a meanAcute Physiology and Chronic HealthEvaluation (APACHE) II score of 23.3.Thirteen of the patients had an improve-ment in the PaO2/FIO2 ratio and the OI. At30 days, nine patients did not survive(53% mortality), with four patients dyingof respiratory failure.

Subsequent to this study, a number ofother authors published their experiencewith HFOV (Table 1) (15–19). All of thesestudies have several similar characteris-tics. First, the number of patients wasgenerally small, ranging between 5 and42. Second, most of the patients hadARDS secondary to pneumonia or sepsis,although two of the studies looked spe-cifically at patients with multiple traumaand burns, respectively (15, 17). Third,HFOV was primarily employed as a rescuemode of ventilation for patients who hadfailed CV. As such, the severity of lunginjury was higher than that reported inother studies of ventilatory modes inARDS, with PaO2/FIO2 ratios ranging from68.6 to 114 (2, 4, 20). Fourth, the strat-egies used to implement HFOV were sim-ilar; the mPaw was set at 5 cm H2O abovethe last mPaw measured while receivingCV, the mPaw was then titrated upwardat 2–3 cm H2O increments until an FIO2

of �0.6 or SpO2 of �92% was achievedwith a reasonable mPaw, and initial fre-quency was at 4–5 Hz and bias flow atbetween 30–40 L/min.

All of these studies showed that theutilization of HFOV led to an improve-ment in oxygenation. This was not en-tirely surprising, as one of the main ad-

vantages of HFOV over CV is the ability toemploy higher mPaws, leading to higherend-expiratory lung volumes and betterrecruitment. Despite the ability to im-prove gas exchange, mortality rates wererelatively high, ranging from 20% to83.3%. Given that these studies were un-controlled, it is difficult to make conclu-sions from these rates. Certainly, a highmortality rate is not surprising becausefailure to respond to CV reflects a patientpopulation with a greater severity of ill-ness. For example, the mean APACHE IIscores in the studies by Mehta et al. (16)and David et al. (19) were 21.5 and 28,respectively. In addition, two of the stud-ies had patients with severe burns or wererecipients of bone-marrow transplants,factors associated with an increased mor-tality (16, 17). Indeed, combining thesetwo studies, patients with severe burnshad a 90.9% mortality rate. Consistentwith other studies of outcomes in ARDS,only a minority of patients receivingHFOV died because of respiratory failure(21). Nevertheless, this observation maybe impressive when one considers thefact that most of these patients had failedCV.

The largest case series so far was pub-lished recently by Mehta et al (22). Theyretrospectively reviewed the experiencewith HFOV in 156 adult patients withsevere ARDS at three academic, teachinghospitals. The patients had a mean age of48 yrs, were severely ill (mean APACHEII, 23.8), and had severe ARDS (meanPaO2/FIO2 ratio of 91 mm Hg and OI of31). Again, HFOV was generally employedas a rescue therapy. They found signifi-cant improvements in the PaO2/FIO2 ratioand the OI, but they had a 30-day mor-tality rate of 61.7%. Independent predic-tors of mortality on multivariate analysiswere older age, higher APACHE II score,lower pH, and a greater number of daysreceiving CV before HFOV. Twenty-sixpercent of patients had HFOV discontin-ued because of difficulties with oxygen-ation, ventilation, or hemodynamics.Importantly, we believe this study dem-onstrates that similar results can be ob-tained in multiple centers.

Thus far, there has only been one pro-spective, randomized trial comparing thesafety and efficiency of HFOV with CV inadults with ARDS. Derdak et al. (23) ran-domized 148 adults, in 13 university-affiliated medical centers, with ARDS toHFOV or CV utilizing pressure-controlledventilation. Patients had a mean age of49.5 yrs, APACHE II score of 22, and


similar PaO2/FIO2 ratios (114 � 37 in theHFOV group and 111 � 42 in the CVgroup). The authors found an earlier im-provement in the PaO2/FIO2 ratio (�16hrs) in the HFOV group; however, thisdifference did not persist beyond 24 hrs.There were a similar number of adverseevents in both groups (intractable hypo-tension, 0% in the HFOV group and 3%in the CV group; air leak, 9% and 12%;mucus-plugged endotracheal tube, 5%and 4%, respectively). It was interesting

that there was a nonsignificant trend to-ward a lower 30-day mortality in theHFOV group (37% vs. 52%, p � .102). Itwould be attractive to postulate that themortality difference could be due to thereduction in ventilator-induced lung in-jury among the patients receiving HFOV.However, as this study was primarily asafety and efficacy study, it was not ade-quately powered to detect a mortality dif-ference. The authors have also been crit-icized for using relatively large V̇Ts (of up

to 10 mL/kg) that led to high Paws (38 �9 cm H2O at 48 hrs) in the control group,in comparison with V̇Ts of 6 mL/kg, nowconsidered a standard of care (3). It isimportant to bear in mind that this studywas conducted before the publication ofthe ARDS Network trial, which demon-strated a survival benefit among patientsventilated with a lower V̇T strategy (2).

Adjunctive Therapies to HFOV. IfHFOV fails to improve oxygenation in theadult patient with ARDS or the patient

Table 1. Comparison of studies evaluating the use of high-frequency oscillatory ventilation in adult patients with acute respiratory distress syndrome

Author(Reference No.) Study Design Patients Mortality

Death Due to RespiratoryFailure, % Selected Complications

Fort et al. (14) Prospective,observational

17 patientsMean age, 38 yrsPaO2/FIO2 ratio, 68.6OI, 48.6APACHE II score,

23.3

30-day mortalityrate, 53%

33.0 3 (17.6%) patients withdrawnfrom HFOV because ofhypotension

Claridge et al. (15) Prospective,observational

5 patients (alltrauma)

Mean age, 36.6 yrsPaO2/FIO2 ratio, 52.2APACHE II score,

28.5

Mortality rate, 20% 0 None reported

Mehta et al. (16) Prospective,observational

24 patientsMean age, 48.5 yrsPaO2/FIO2 ratio, 98.8OI, 32.5APACHE II score,

21.5


6.25 2 patients (8.3%) hadpneumothoraces

Cartotto et al. (17) Retrospective 6 patients (allburns)

Mean age, 34 yrsPaO2/FIO2 ratio, 92OI, 32APACHE II score, 16

Mortality rate,83.3%

0 Not reported

Derdak et al. (23) Randomized, controlledtrial

148 patientsMean age, 49.5 yrsPaO2/FIO2 ratio,

112.5OI, 25.2APACHE II score, 22

30-day mortality was37% in the HFOVgroup and 52% inthe CV group

16.0 in both arms Similar in both groups

Andersen et al. (18) Retrospective 16 patientsMean age, 38.2 yrsPaO2/FIO2 ratio, 92OI, 28.1SAPS II, 40.3

Mortality rate,31.2% at 3 mos

Not reported 1 patient (6.3%) had apneumothorax

David et al. (19) Prospective,observational

42 patientsMedian age, 49 yrsPaO2/FIO2 ratio, 94OI, 23APACHE II score, 28


33.3 1 patient (2.4%) had apneumothorax

Mehta et al. (22) Retrospective 156 patientsMedian age, 47.8 yrsPaO2/FIO2 ratio, 91OI, 31.2APACHE II score,

23.8

30-day mortalityrate, 61.7%

Not reported 34 patients (21.8%) had apneumothorax

OI, oxygenation index (FIO2 � mean airway pressure � 100/PaO2); APACHE, Acute Physiology and Chronic Health Evaluation score; HFOV,high-frequency oscillatory ventilation; SAPS, Simplified Acute Physiology Score, CV, conventional ventilation.


subsequently deteriorates, other adjunc-tive therapies, including prone position-ing and inhaled nitric oxide, have beenreported to be used successfully to im-prove gas exchange (24, 25). In a study of23 adults with ARDS by Mehta et al. (24),titrated inhaled nitric oxide at doses be-tween 5 and 20 ppm during HFOV in-creased PaO2/FIO2 by 38% at 30 mins in 19of those patients (83%). The intensivecare unit survival was only 39%, an ob-servation that may not be surprising con-sidering that these patients had beendeemed to fail CV and were also deterio-rating while receiving HFOV.

Recently, recruitment maneuvers incombination with HFOV have beenshown to be safe and result in rapid andsustained improvements in oxygenationwhen compared with CV (26). Further-more, because of the earlier improve-ment in oxygenation when comparedwith other studies of HFOV, it is possiblethat the effects of recruitment maneuversare, at the very least, additive.

Overall, the effect of these adjunctivetherapies on further ventilator-inducedlung injury, duration of mechanical ven-tilation, and mortality remains to be de-fined. Further research also needs to bedone on patient selection and ideal tim-ing in applying these adjuncts to HFOV.

Predictors of Mortality Among Pa-tients Receiving HFOV. Many of the stud-ies showed that a delay in initiatingHFOV was a predictor of death (14, 16,19, 22). For example, in the study by Fortet al. (14), the mean duration of CV in thesurvivors was 2.5 days, as compared with7.2 days in the nonsurvivors (p � .09). Asimilar finding was later reported byMehta et al. (16) (1.6 days vs. 7.8 days,respectively). In addition, David et al. (19)found a greater 30-day mortality in pa-tients ventilated with CV for �3 daysbefore commencing HFOV (64% vs.20%). A subsequent study involving 156patients also found that a higher numberof days receiving CV before HFOV inde-pendently predicted mortality in a multi-variable analysis (22). It is possible thatthese observations may have been con-founded by both a greater severity of ill-ness and lung injury in the nonsurvivorgroup. For instance, in the study by Fortet al. (14), nonsurvivors had higher meanAPACHE II scores (27.17 vs. 20.0 in thesurvivor group), OI (60.0 vs. 34.0), andLung Injury Scores (3.92 vs. 3.69) atbaseline. However, Mehta et al. (16)found similar baseline APACHE II andLung Injury Scores in both groups.

Multiple animal studies have alsoshown that exposure to excessive V̇Ts andinsufficient PEEP to prevent repeated al-veolar collapse could further exacerbatethe ongoing lung injury (27–29). In ad-dition, exposure to excessive airway pres-sures has been shown to increase thelevels of proinflammatory cytokines andhas been implicated in the developmentof multiple organ failure (4, 5). Thus, it isattractive to postulate that a longer du-ration of exposure to CV could have led toworsening lung injury, perhaps multipleorgan failure, and hence, a higher mor-tality. As mentioned before, in the ran-domized trial by Derdak et al. (23) com-paring the safety and efficacy, there was atrend toward a lower mortality in theHFOV group. Nevertheless, the readerneeds to bear in mind that patients withARDS who have severe illness early intheir intensive care unit course (and thusneed HFOV earlier) may have an entirelydifferent prognosis than those who havesevere illness later. Clearly, the hypothe-sis that patients with severe ARDS mayderive benefit from earlier interventionwith HFOV is worthwhile in testing in aprospective, randomized fashion.

Some of the studies have also identi-fied the OI as a predictor of mortality (14,22, 23). Fort et al. (14) found that a base-line OI � 47 was predictive of death, with100% sensitivity and 100% specificity. Inaddition, both Derdak et al. (23) andMehta et al. (22) found that the posttreat-ment OI was also important. For exam-ple, in a multivariable analysis, the OI at24 hrs after commencing HFOV wasfound to be the most significant predictorof mortality (22). This is not entirely un-expected, as the OI has also been found tobe an independent predictor of mortalityin the pediatric literature studying theutility of HFOV and in adult patients re-ceiving CV (7, 30, 31). The OI may bemore powerful than traditional indiceslike the PaO2/FIO2 ratio, as it assesses the“pressure cost” of oxygenation.

Other variables that have been shownto correlate independently with survivalinclude the age of the patient, pH, andthe APACHE II score (22, 23). Unfortu-nately these predictors of poor outcomehave not been substantiated to the pointthat they can be incorporated into deci-sions around ongoing care.

Complications of HFOV. The compli-cations reported with the use of HFOVare generally low (Table 1). These areusually related to barotrauma or hemo-dynamic compromise. For example, in

the series by Mehta et al. (16), two pa-tients (8.3%) had pneumothoraces. Inone of these patients, the pneumothoraxcould possibly be related to a right mainbronchus intubation. However, in a laterretrospective study involving a largergroup of patients, again by Mehta et al.(22), they found a much higher preva-lence of pneumothoraces (21.8%). Thisrate is higher than that reported in otherstudies evaluating HFOV and also in stud-ies evaluating conventional ventilatorystrategies (2, 14–19, 23, 32). The reasonfor the higher rate remains unclear.There were no obvious differences in theway HFOV was employed compared withthe other studies. Differences in ventila-tory strategies and patient populationsacross the different centers may have ledto greater barotrauma.

The higher mPaws utilized duringHFOV could conceivably impede venousreturn and lead to hypotension. However,this complication seemed to be uncom-mon, and only one study documentedsignificant hypotension after the initia-tion of HFOV (14). In this study, HFOVwas withdrawn from three patients(17.6%) because of hypotension. In one ofthese patients, the hypotension was at-tributed to postoperative bleeding. In pa-tients with a pulmonary artery catheterin situ, the utilization of HFOV has beenassociated with an increase in central ve-nous pressure and pulmonary artery oc-clusion pressure and a small decrease incardiac output (14, 16, 22, 23). As anexample, in the randomized study by Der-dak et al. (23), the central venous pres-sure increased from 14 mm Hg at base-line to 16 mm Hg at 2 hrs. Thecorresponding values for the pulmonaryartery occlusion pressure were 16 mm Hgand 18 mm Hg. For cardiac output, theywere 7.4 L/min and 7.0 L/min. Most ofthese changes were transient, as only thepulmonary artery occlusion pressure re-mained significantly increased after 3days. The increase in central venous pres-sure and pulmonary artery occlusionpressure likely reflected the increase inmPaw. The relevance of the small de-crease in cardiac output is unclear, par-ticularly given that the cardiac outputremained within normal or supernormalrange, and it did not seem to be associ-ated with a decrease in mean arterialpressure or increase in heart rate (22,23). However, in our experience, clini-cians do need to be aware that as themPaw increases, techniques to improvethe cardiac output (e.g., intravascular


volume loading) may need to be em-ployed.

Mucus inspissation is a potential prob-lem, which could cause endotrachealtube obstruction and refractory hyper-capnia (6). It should be suspected in anotherwise stable patient who has had asudden increase in PCO2. Fortunately, theprevalence of this complication is rela-tively low (4–5%) and not reported by allthe studies (14, 16). The extent that fre-quent airway suctioning, chest physio-therapy, or fluid instillation into the air-way ameliorates this is uncertain. In therandomized study by Derdak et al. (23),the frequency of endotracheal tube occlu-sion requiring a tube change was similarin both the HFOV and the CV groups, atabout 5% and 4%, respectively.

Conclusions

HFOV has been shown to be a safe andeffective mode of ventilation in adult pa-tients with severe ARDS who have failedCV. Earlier initiation of HFOV in adultpatients with ARDS seems to be at least assafe as CV, and preliminary data suggestthat there may be a survival advantage.Future research should be geared towardidentifying patients who may benefitfrom early initiation of HFOV and thebest technique with which to utilize it.Finally, although we are encouraged bythe data we have reviewed, HFOV foradults with ARDS is still in its infancy,and we look forward to its continued eval-uation.

REFERENCES

1. International consensus conferences in in-tensive care medicine: Ventilator-associatedLung Injury in ARDS. This official confer-ence report was cosponsored by the Ameri-can Thoracic Society, The European Societyof Intensive Care Medicine, and The Societede Reanimation de Langue Francaise, andwas approved by the ATS Board of Directors,July 1999. Am J Respir Crit Care Med 1999;160:2118–2124


3. Singh JM, Stewart TE: High-frequency oscil-latory ventilation in adults with acute respi-ratory distress syndrome. Curr Opin CritCare 2003; 9:28–32

4. Ranieri VM, Suter PM, Tortorella C, et al:Effect of mechanical ventilation on inflam-matory mediators in patients with acute re-

spiratory distress syndrome: A randomizedcontrolled trial. JAMA 1999; 282:54–61

5. Slutsky AS, Tremblay LN: Multiple systemorgan failure: Is mechanical ventilation acontributing factor? Am J Respir Crit CareMed 1998; 157(6 Pt 1):1721–1725

6. Derdak S: High-frequency oscillatory venti-lation for acute respiratory distress syn-drome in adult patients. Crit Care Med 2003;31(4 Suppl):S317–S323


8. Gerstmann DR, Minton SD, Stoddard RA, etal: The Provo multicenter early high-frequency oscillatory ventilation trial: Im-proved pulmonary and clinical outcome inrespiratory distress syndrome. Pediatrics1996; 98(6 Pt 1):1044–1057

9. Plavka R, Kopecky P, Sebron V, et al: A pro-spective randomized comparison of conven-tional mechanical ventilation and very earlyhigh frequency oscillatory ventilation in ex-tremely premature newborns with respira-tory distress syndrome. Intensive Care Med1999; 25:68–75

10. Sedeek KA, Takeuchi M, Suchodolski K, et al:Determinants of tidal volume during high-frequency oscillation. Crit Care Med 2003;31:227–231


12. Imai Y, Nakagawa S, Ito Y, et al: Comparisonof lung protection strategies using conven-tional and high-frequency oscillatory venti-lation. J Appl Physiol 2001; 91:1836–1844

13. Rotta AT, Gunnarsson B, Fuhrman BP, et al:Comparison of lung protective ventilationstrategies in a rabbit model of acute lunginjury. Crit Care Med 2001; 29:2176–2184


15. Claridge JA, Hostetter RG, Lowson SM, et al:High-frequency oscillatory ventilation can beeffective as rescue therapy for refractoryacute lung dysfunction. Am Surg 1999; 65:1092–1096


17. Cartotto R, Cooper AB, Esmond JR, et al:Early clinical experience with high-fre-quency oscillatory ventilation for ARDS inadult burn patients. J Burn Care Rehabil2001; 22:325–333

18. Andersen FA, Guttormsen AB, Flaatten HK:High frequency oscillatory ventilation inadult patients with acute respiratory distresssyndrome: A retrospective study. Acta Anaes-thesiol Scand 2002; 46:1082–1088

19. David M, Weiler N, Heinrichs W, et al: High-frequency oscillatory ventilation in adultacute respiratory distress syndrome. Inten-sive Care Med 2003; 29:1656–1665

20. Esteban A, Alia I, Gordo F, et al: Prospectiverandomized trial comparing pressure-con-trolled ventilation and volume-controlledventilation in ARDS: For the Spanish LungFailure Collaborative Group. Chest 2000;117:1690–1696

21. Frutos-Vivar F, Nin N, Esteban A: Epidemi-ology of acute lung injury and acute respira-tory distress syndrome. Curr Opin Crit Care2004; 10:1–6



24. Mehta S, MacDonald R, Hallett DC, et al:Acute oxygenation response to inhaled nitricoxide when combined with high-frequencyoscillatory ventilation in adults with acuterespiratory distress syndrome. Crit Care Med2003; 31:383–389

25. Varkul MD, Stewart TE, Lapinsky SE, et al:Successful use of combined high-frequencyoscillatory ventilation, inhaled nitric oxide,and prone positioning in the acute respira-tory distress syndrome. Anesthesiology 2001;95:797–799

26. Ferguson ND, Chiche JD, Kacmarek RM, etal: High frequency oscillatory ventilation andrecruitment maneuvers are safe and improveoxygenation in early severe ARDS. Am J Re-spir Crit Care Med 2003; 167(Suppl):A177

27. Muscedere J, Mullen J, Gan K, et al: Tidalventilation at low airway pressures can aug-ment lung injury. Am J Respir Crit Care Med1994; 149:1327–1334

28. Parker JC, Hernandez LA, Peevy KJ: Mecha-nisms of ventilator-induced lung injury. CritCare Med 1993; 21:131–143

29. Tsuno K, Miura K, Takeya M, et al: His-topathologic pulmonary changes from me-chanical ventilation at high peak airway pres-sures. Am Rev Respir Dis 1991; 143(5 Pt1):1115–1120

30. Monchi M, Bellenfant F, Cariou A, et al: Earlypredictive factors of survival in the acuterespiratory distress syndrome: A multivariateanalysis. Am J Respir Crit Care Med 1998;158:1076–1081

31. Arnold JH, Anas NG, Luckett P, et al: High-frequency oscillatory ventilation in pediatricrespiratory failure: A multicenter experience.Crit Care Med 2000; 28:3913–3919

32. Stewart TE, Meade MO, Cook DJ: Evaluationof a ventilatory strategy to prevent baro-trauma in patients at high risk for acuterespiratory distress syndrome. N Engl J Med1998; 338:355–361


Use of high-frequency oscillatory ventilation in burn patients

Robert Cartotto, MD, FRCS(C); Sandi Ellis, RRT; Terry Smith, MD, FRCP(C)

Patients with major burn inju-ries are prone to the develop-ment of acute lung injury andthe acute respiratory distress

syndrome (ARDS) (1). These conditionsmay arise from direct lung injury aftersmoke inhalation or pneumonia, from in-direct lung injury due to systemic inflam-mation from the burn wound, or fromsepsis arising from any number ofsources such as the lungs, the burnwound, the gastrointestinal and urinarytracts, and intravascular catheters. Theprevalence of ARDS among mechanicallyventilated adults with major burns hasbeen estimated to be as high as 54% (2),using the American–European Consen-sus definition (3).

Management of acute lung injury andARDS in burn patients involves two keyprinciples. The first is to strike a delicatebalance between provision of respiratorysupport to a hypoxemic patient with stiff,noncompliant lungs, while utilizing alung-protective ventilation strategy thatminimizes ventilator-induced lung injury

(4–6). The second principle is early exci-sion and closure of the burn wound.Timely removal of burned tissue has beenidentified as an important determinant ofsurvival after a major burn injury (7–9)and is immensely important in removingthe very force that may be contributing tothe lung injury. Hence, unlike many pa-tients in the medical-surgical intensivecare unit, patients with significant ther-mal injury typically require a series ofmajor, staged operations under generalanesthesia. In a patient with ARDS andsevere oxygenation failure, this poses aformidable challenge, and surgical inter-vention should not be delayed or deferredin the scenario in which a patient is fail-ing conventional mechanical ventilation(CMV). In this situation, a ventilatorystrategy that effectively reverses oxygen-ation failure and that can be continued inthe operating room would be distinctlyadvantageous.

In response to these unique require-ments, we have utilized high-frequencyoscillatory ventilation (HFOV) in ouradult regional burn center since 1999.Initially, HFOV was employed as a rescueventilation strategy for patients with ex-treme oxygenation failure from ARDS,despite maximum CMV. The marked andrapid improvements in oxygenation thatoccurred with HFOV allowed us to surgi-

cally excise the burn wounds of patientswho would otherwise have been too un-stable to tolerate a major operation undergeneral anesthesia while continuing onmaximal CMV. With recognition of thepotential lung-protective benefits ofHFOV (10 –14), we have progressivelyused HFOV more frequently and earlierin the course of caring for a burn patientwith acute lung injury or ARDS. Thissection of the Supplement will review ourexperience with HFOV in burn patients,our specific approach to the initiationand termination of HFOV, the specialconsiderations for use of HFOV in burnpatients with smoke inhalation injury,and our approach to use of HFOV duringsurgery.

Clinical Experience with High-Frequency Ventilation in BurnPatients

Published experience describing theuse of HFOV in burn patients is fairlylimited. Early case reports (14) and smallcase series (15, 16), including our ownpreliminary experience (17), generallyfound that HFOV was highly effective as a“rescue therapy” when maximal or nearmaximal CMV had been unsuccessful inreversing extreme oxygenation failure as-sociated with ARDS. To date, we have

From the Ross Tilley Burn Centre, Sunnybrook andWomen’s College Health Sciences Centre, Toronto,Canada.


DOI: 10.1097/01.CCM.0000157232.31910.4B

Background: Patients with major burn injuries frequently de-velop acute respiratory distress syndrome (ARDS). High-fre-quency oscillatory ventilation (HFOV) has been used successfullyin our regional burn center since 1999 for the management ofoxygenation failure secondary to ARDS and as a method ofintraoperative ventilation to allow surgical burn wound excision toproceed, despite the presence of severe ARDS.

Objective: The objective of this article is to review the use ofHFOV in burn patients, with an emphasis on the indications andselection criteria for the initiation of HFOV, special considerationsfor patients with smoke inhalation injury, and our approach tointra-operative HFOV.

Setting: Adult regional burn center in a university-affiliatedtertiary care hospital.

Results: We have now used HFOV in 36 severely burned pa-

tients, 33% of whom had an associated smoke inhalation injury.HFOV produced significant improvements in oxygenation amongburn patients with oxygenation failure secondary to ARDS. HFOVproduced a slower and less robust reversal of oxygenation failurein those with smoke inhalation compared with patients with burnsalone. HFOV has been used intraoperatively for 33 procedures in18 patients without complications.

Conclusion: HFOV has been an indispensable ventilation mo-dality in our burn center, and has played an important role inreversing oxygenation failure in patients with ARDS and in facil-itating early excision and closure of the burn wound in thesepatients. (Crit Care Med 2005; 33[Suppl.]:S175–S181)

KEY WORDS: high-frequency oscillatory ventilation; burns;smoke inhalation; acute respiratory distress syndrome; acute lunginjury


used HFOV in 36 severely burned patients(mean � SD age, 42 � 15 yrs; percentageof total body surface area burn, 42% �17%, with 33% rate of smoke inhalation),and we have analyzed data on our expe-rience with the first 25 cases (18). In thatstudy, 24 of 25 patients met American–European Consensus criteria for ARDS(3) and had a mean � SD PaO2/FIO2 ratioof 98 � 26 and a mean oxygenation index(OI � mean airway pressure � FIO2 �100/PaO2) of 27 � 10 just before startingHFOV. Conventional ventilation just be-fore HFOV had consisted of pressure-controlled ventilation using an FIO2 of 0.8� 0.2, positive end-expiratory pressure of13.7 � 3.5 cm H2O, inspiratory-to-expiratory ratio of 1:1.6, and peak inspira-tory pressure of 28 � 5 cm H2O, with sixpatients receiving 16 � 6 ppm of inhalednitric oxide (iNO). There was a significantand sustained increase in the PaO2/FIO2

ratio within 1 hr and a significant and

sustained reduction in the OI within 48hrs of starting HFOV (Fig. 1). Among thesix patients who had been receiving iNO,all were weaned off iNO within 24 hrs ofstarting HFOV. Aside from three episodesof severe hypercapnia (PaCO2 range, 92–136 mm Hg), there were no other com-plications related to HFOV such as grossbarotrauma, hemodynamic instability, orinspissation of mucous secretions. HFOVwas maintained for 6.1 � 5.8 days (range,2 hrs to 26 days). Seven patients (28%)died receiving HFOV after a duration ofHFOV treatment of 5.9 � 5.8 days (range,2 hrs to 15 days). In the patients whodied, the PaO2/FIO2 ratio and OI immedi-ately before death were 170 � 116 and 22� 11, respectively. Refractory oxygen-ation failure was judged to be contribu-tory in only one case (PaO2/FIO2 ratio, 74;OI, 39). The underlying cause of death inall cases was sepsis with multiple organdysfunction. The 18 survivors were con-

verted back to CMV after 6.1 � 5.9 days(range, 5 hrs to 26 days) at an FIO2 of 0.4� 0.1 and a mean airway pressure(mPOaw) of 24.4 � 4.0 cm H2O. At thistime, their PaO2/FIO2 ratio was 238 � 109and their OI was 13 � 9, both of whichwere significantly better than present onCMV immediately before starting HFOV(p � .001, each).

To summarize, HFOV has producedpronounced and sustained improvementin oxygenation in several of our burnpatients with severe oxygenation failuresecondary to ARDS. A progressive declinein the OI was also observed, indicatingthat the improvements in oxygenationoccurred at a progressively lower “airwaypressure cost” (19). Our findings in thisgroup of burn patients are very similar toobservations made in the first two adultstudies of HFOV in critically ill medical-surgical intensive care unit patients (19,20).

High-Frequency PercussiveVentilation/Volume-DiffusiveRespirator

High-frequency percussive ventilation(HFPV) has been more widely used inthermally injured patients and has re-ceived considerably more attention in theburn literature than has HFOV. HFPV isadministered using a volumetric-diffusiverespirator that incorporates a slidingVenturi called the Phasitron (21). Thevolumetric-diffusive respirator deliverssub–dead space breaths at frequenciesbetween 0.6 and 15 Hz. The sliding Ven-turi stacks the rapid pulsatile breaths to aselected peak airway pressure that is theninterrupted at regular intervals (usuallyevery 2 secs) by a phase of passive exha-lation that allows the waveform and air-way pressure to return to baseline con-tinuous positive airway pressure, which isusually set between 5 and 10 cm H2O(21). Thus, the high-frequency percussivewaveform is superimposed on a sinusoi-dal cyclic waveform of increasing and de-creasing airway pressures (the phasicrate). There are several fundamental dif-ferences between HFOV and HFPV. First,the mPOaw is relatively constant and sus-tained during HFOV, whereas it fluctu-ates and is not sustained during HFPV.Second, during HFOV, the mPOaw is di-rectly set, which immediately influencesalveolar recruitment and oxygenation. InHFPV, the mPOaw can only be indirectlyaltered. Third, exhalation during HFOV isactive, whereas exhalation is passive dur-

Figure 1. Top, mean � SD PaO2/FIO2 ratios on conventional mechanical ventilation (CMV) immediatelybefore high-frequency oscillatory ventilation (HFOV) and at different time points after initiation ofHFOV. N, the number of measurements; *p � .04; **p � .004 compared with CMV using repeated-measures analysis of variance. Bottom, mean � SD oxygenation index (OI) on CMV immediately beforeHFOV and at different time points after initiation of HFOV. N, the number of measurements; *p � .03;**p � .01 compared with CMV using repeated-measures analysis of variance.


ing HFPV. Fourth, because the Phasitronis pneumatically driven, the frequencyand pressure amplitude of the high-frequency sub– dead space breaths are“coupled” (e.g., an increase in frequencyresults in a decrease in pressure ampli-tude and vice versa); in contrast, duringHFOV, the frequency and proximal oscil-latory pressure swings (�P) are adjust-able independently. Finally, the percus-sive nature of the pulsatile breath patternin HFPV promotes mucokinesis andclearance of secretions, which is aided bythe fact that the endotracheal tube (ETT)cuff is kept deflated during HFPV.

The reported use of HFPV among burnpatients has largely been restricted tothose with an associated smoke inhala-tion injury. Two small case series suggestthat HFPV is highly effective as a salvageventilation strategy for patients with se-vere oxygenation failure after burns andsmoke inhalation (22, 23). HFPV has alsobeen applied as a prophylactic ventilationstrategy in burn patients with inhalationinjury. A retrospective study by Cioffi etal. (24), in which 54 patients were placedon HFPV within 1 hr of admission to theburn unit, found a significant reductionin the rate of pneumonia and the mortal-ity rate, compared with predicted valuesfrom a historical cohort. More recently, aprospective study involving 35 adultswith burns and smoke inhalation ran-domized patients to conventional vol-ume-controlled ventilation or to HFPV(25). The PaO2/FIO2 ratios in the HFPVgroup were significantly higher than inthe conventional group, but there wereno differences between groups in peakairway pressures, rate of pneumonia, ormortality. To our knowledge, there havebeen no prospective, randomized clinicaltrials comparing HFPV with HFOV.

Specific Clinical Issues in UsingHFOV in Burn Patients

Patient Selection. In a burn patientwith ARDS, we generally consider usingHFOV when there is moderate to severeoxygenation failure (usually with a PaO2/FIO2 ratio of �150), despite relatively ag-gressive or escalating conventional me-chanical ventilatory support (typicallycharacterized by either an FIO2 of �0.6with positive end-expiratory pressure of�12.5 cm H2O or the need for inverseratio ventilation or use of iNO to supportoxygenation). The OI is also an importantconsideration. An OI of �30 is generallyaccepted to represent failure of CMV (19),

but we have generally instituted HFOVwhen the OI is �25. The presence orabsence of an associated smoke inhala-tion injury does not affect our decision touse HFOV. However, as will be discussed,we have found HFOV to be less effectivein burn patients with smoke inhalation.Finally, the need to take a patient to sur-gery for burn wound excision also influ-ences our decision to start HFOV. If burnexcision is required and the patient is ator approaching the gas exchange andventilatory variables described above, weare more inclined to switch over to HFOVas soon as possible before surgery. Ourrationale for this approach is that in ourhands, HFOV typically produces markedimprovements in oxygenation over a rel-atively short period and thus provides agreater margin of safety and flexibilitywith respect to intraoperative ventilatorymanagement than if the patient had beenbrought to the operating room remainingon CMV with poor oxygenation and highairway pressures.

Timing. Over the past 5 yrs, we haveprogressively initiated HFOV earlier inthe patient’s course. In 1999, we startedHFOV after a mean of 7.3 days postburnin patients with a mean OI of 30, whereasin 2003, HFOV was started after a meanof only 2.5 days postburn in patients witha mean OI of 30. The effect of timing ofinitiation of HFOV on the course of ARDSremains unknown, but recent adultHFOV trials suggest that earlier institu-tion of HFOV may improve outcome (19,20, 26).

Sedation Strategy. Early in our expe-rience, we paralyzed all patients usingvecuronium infusions. However, becausemost of our mechanically ventilated burnpatients are heavily sedated with contin-uous infusions of morphine and midazo-lam, we have found that neuromuscularblockade is not absolutely required dur-ing HFOV, as long as the patient is wellsedated. If the patient is inadequately se-dated and making any respiratory efforts,one will observe swinging fluctuations inthe set mPOaw. This suggests the need forincreased sedation or paralysis. This con-trasts sharply with mechanically venti-lated patients in the medical-surgical in-tensive care unit, who are usually not asdeeply sedated and who will likely requireparalysis during HFOV.

Oxygenation Strategy. We set the ini-tial FIO2 at 1.0 and arbitrarily start with amPOaw of 5 cm H2O above the mPOaw thatwas present while the patient was receiv-ing CMV. This typically produces a tran-

sient increase in the OI, which should notbe interpreted as a sign of HFOV failure.The mean � SD starting mPOaw in ourburn patients has been 33 � 5 cm H2O(18). The percentage of inspiration timeis set at 33% and the bias flow is set at 30L/min to start. We now also routinely useone or more lung recruitment maneuvers(LRMs) just after initiation of HFOV. Thisconsists of temporarily increasing themPOaw to 40 cm H2O for a period of20–40 secs. There is substantial evidencefrom animal models that LRMs are essen-tial in recruiting alveoli for optimal ap-plication of HFOV (27). As long as theSpO2 is �92%, the FIO2 can then be ti-trated down from 1.0 to the pre-HFOVlevel. If the pre-HFOV FIO2 cannot beachieved, then, guided by oximetry andarterial blood gasses, gradual increases of1 or 2 cm H2O are made in the mPOaw, andthese may be combined with LRMs. Themost common error in our experience isto use an inadequate starting mPOaw or toprematurely reduce the mPOaw. We havenot had to use a mPOaw in excess of 40 cmH2O. Careful attention must be paid tothe patient’s intravascular volume status.High mPOaw, especially during a LRMs,may elevate intrathoracic pressure andimpair preload, causing hypotension.Similarly, overinflation with excessivelyhigh mPOaw may impair ventilation/perfusion matching and cause transientdesaturation, especially during a LRM.The important clinical message is thatcareful titration and patience are oftenneeded in the initial selection of appro-priate FIO2 and mPOaw combinations.

Once HFOV has been initiated and theinitial FIO2 and mPOaw settings are stabi-lized, we will attempt to titrate down theFIO2 every 8 to 12 hrs as long as SpO2 is�92% and the PaO2 is �65 mm Hg. As ageneral rule, FIO2 is always titrated downbefore reducing the mPOaw. The goal is toachieve an FIO2 of �0.4 for 12 consecu-tive hours. Then, and only then, will webegin to reduce the mPOaw. This is done inincrements of 1–2 cm H2O every 8–12hrs. If acceptable oxygenation on an FIO2

of �0.4 with a mPOaw of �25 cm H2O canbe achieved for 12 consecutive hours, wethen consider conversion back to CMV.

Ventilation Strategy. During HFOV,alveolar ventilation is improved by in-creasing the power to increase the ampli-tude of oscillation or by reducing thefrequency to increase the stroke volume.The latter maneuver is generally the mosteffective but is the least desirable becausethe larger tidal volumes at low frequen-


cies may be sufficient to overstretch al-veoli and cause injury (with an adult-sized ETT and no cuff leak, using high-power settings and frequencies of 3–6Hz, tidal volumes of 150–260 mL areproduced, which may be sufficient to pro-duce stretch injury when the lung is in-flated with a mPOaw of 30 cm H2O) (28,29). Most adult studies of HFOV (19, 20,26, 30), including our own (18), haveinitiated HFOV at approximately 5 Hz anddecreased this toward 3 Hz if increasedCO2 removal was necessary. There is noscientific basis for limiting frequencies to5 Hz, and in neonates, HFOV frequenciesof 10–15 Hz are routinely used (23). Op-timal frequencies in adults are unknownbut are probably higher than 3–5 Hz (23).Our current approach is to start withmaximum power (9, 10) and a frequencyaround 6–8 Hz. Chest and body vibrationare then used to roughly guide the initialpower and frequency settings. If vibrationis excessive (e.g., chest vibrations aretransmitted distal to the mid-thigh), thenthe power is decreased by increments of 1unit every 5 mins until the target of mid-

thigh “wiggle” is obtained. If chest wallvibration is too little (e.g., not transmit-ted to the mid-thigh level), then the fre-quency is reduced by 1 Hz every 5 minsuntil mid-thigh wiggle is obtained (usu-ally this involves a reduction to around 5Hz but should not go lower than 3 Hz).Subsequent adjustments are made basedon arterial blood gasses. A permissive hy-percapnia approach is used, allowing aPaCO2 of �65 mm Hg and pH � 7.25.Hypercapnia with acidemia beyond theselimits is first treated by increasing powerto maximum if it is not already at 10 and,second, by reducing frequency by 1-Hzincrements to a minimum of 3 Hz. If thisis unsuccessful, we occasionally resort toan intentional endotracheal cuff leak, asdescribed by Derdak (29). An intentionalcuff leak is produced by withdrawing airfrom the cuff until the mPOaw decreasesby 5 cm H2O and then restoring themPOaw to the preleak level with the mPOawcontrol dial. If this is unsuccessful inrestoring the preleak mPOaw, the bias flowcan be increased. We have used cuff leaksin eight cases to date. These were insti-

tuted for PaCO2 levels ranging from 61 to136 mm Hg, while administering fre-quencies between 2.8 and 5 Hz and powersettings of 9 or 10. In two cases, the cuffleak alone was successful in correctingthe PaCO2 and pH to acceptable levels. Infour cases, the cuff leak produced modestbut insufficient corrections in the PaCO2,and ultimately, further reductions in fre-quency were required. In one case, aswitch to CMV was required to correcthypercapnia. In the final case, severe hy-percapnia (PaCO2 of 136 mm Hg and pHof 7.10) persisted, despite a frequency of 3Hz, amplitude of 109 cm H2O, and anintentional cuff leak. It was recognizedthat the intentional cuff leak from theoral ETT was probably ineffective due tosevere head and neck edema that wasblocking passive CO2 egress in this par-ticular patient. A 6.0-mm nasal ETT wasinserted into the supraglottic hypophar-ynx to act as a “vent” for the cuff leak,which resulted in a PaCO2 of 85 and a pHof 7.3 within 4 hrs and further reductionin the PaCO2 to 60 mm Hg by 6 hrs whilecontinuing HFOV at the same frequencyand power settings (31). The improve-ment in alveolar ventilation persisted for48 hrs until the patient expired frommultiple organ failure. Notwithstandingthe aforementioned cases, hypercapniahas not been a significant problem in thevast majority of our HFOV cases, inwhich the mean � SD PaCO2 was 51 � 12mm Hg on a frequency of 4.6 � 1.1 Hzover the first 72 hrs of HFOV (18) .

Smoke Inhalation Injury and HFOV.Approximately 30% of our burn patientswho receive HFOV have an associatedsmoke inhalation injury. In our institu-tion, smoke inhalation injury is diag-nosed at admission by direct visualizationof the tracheobronchial mucosa using fi-beroptic bronchoscopy. The presence ofinfraglottic soot, mucosal hyperemia,edema, slough, or ulceration are all con-sidered positive signs of smoke inhalationinjury. Serial examinations by the sameexperienced observer prove to be themost reliable. Unfortunately, there is anindistinct relationship between the sever-ity of findings at bronchoscopy and theeventual extent of lung dysfunction fromsmoke inhalation injury. The response toHFOV in our patients who have had asmoke inhalation injury has been less im-pressive than among patients with a burnalone. Significant improvement in thePaO2/FIO2 ratio from that on CMV oc-curred only after 72 hrs of HFOV insmoke inhalation cases, whereas in pa-

Figure 2. Top, comparison of the mean � SD PaO2/FIO2 ratios of patients with inhalation injury (solidcircles) and patients without inhalation injury (open circles) receiving conventional mechanicalventilation (CMV) and at different time points after initiation of high-frequency oscillatory ventilation(HFOV). N, the number of measurements; *p � .03 for patients without smoke inhalation; **p � .02for patients with smoke inhalation compared with CMV using repeated-measures analysis of variance.Bottom, comparison of the mean � SD oxygenation index (OI) of patients with inhalation injury (solidcircles) and patients without inhalation injury (open circles) receiving CMV and at different timepoints after initiation of HFOV. N, the number of measurements; *p � .05 for patients without smokeinhalation injury compared with CMV using repeated-measures analysis of variance.


tients without inhalation injury, signifi-cant improvements were seen after only12 hrs. Similarly, although smoke inha-lation patients showed a trend toward alower OI, this never became significantlybetter compared with that on CMV. Inthose patients who had a burn alone, theOI was significantly better than on CMVwithin 24 hrs (Fig. 2). There are no hu-man studies of HFOV in smoke inhala-tion, but a study in primates has com-pared the prophylactic use of HFOVagainst HFPV after an experimentally in-duced smoke inhalation injury (32). Inthat study, subjects receiving HFOV (n �3) were less successfully oxygenated andseemed to have more histologic evidenceof ventilator-induced pulmonary injurythan subjects receiving HFPV (n � 5).Although this study was somewhat lim-ited in extent, the findings mirror ourclinical observations.

A possible explanation may be theunique pathologic changes in the smallairways that occur with smoke inhalationinjury. Obstruction of the small airwaysby edema and by sloughing, necrotic ep-ithelial mucosa combined with broncho-spasm may prevent adequate alveolar re-cruitment, whereas in other areas of thelung, this same pathology may contributeto segmental gas trapping and overdis-tention with the application of sustainedand elevated mPOaw, thus promoting alve-olar stretch injury. Because lung volumeis allowed to periodically return to base-

line during HFPV, gas trapping and seg-mental overdistention may be betteravoided relative to HFOV. Also, the mu-cokinetic effects of HFPV may promoteenhanced clearing of the airways, thuspreventing atelectasis and improving al-veolar recruitment. In summary, al-though we do not consider smoke inha-lation injury to be a contraindication toHFOV, we do recognize that the responseto HFOV will likely be slower and lessrobust than in cases without an inhala-tion injury. Furthermore, in selected in-stances, especially when bronchorrhea isa prominent feature, the mucokinetic ef-fects of HFPV offer a distinct advantageover HFOV for the burn patient with asmoke inhalation injury.

Use of HFOV in the OperatingRoom

Early surgical excision and closure ofthe burn wound is an integral part of themanagement of a patient with a majorburn injury. However, subjecting a pa-tient with acute lung injury or ARDS andmoderate to severe oxygenation failure toa major operation under general anesthe-sia poses a substantial challenge, partic-ularly with respect to perioperative me-chanical ventilation. We realized early inour experience that the ability of HFOV toreverse severe oxygenation failure in arelatively short period of time created animportant window of opportunity for the

excision and closure of the patients’ burnwounds. However, this presented thechallenge of whether HFOV could be usedin the operating room as opposed to tem-porarily converting back to CMV for thesurgical procedure. We believe that inter-ruption of HFOV for these operationswould not only have posed a risk of alve-olar derecruitment with the interruptionin mPOaw but would also have diminishedthe consistency of the lung-protective ef-fects of HFOV. Therefore, in an effort totake advantage of the improvements inoxygenation that occurred with HFOV,and wishing to maintain a consistentlung-protection strategy, we have devel-oped an approach for continuing HFOVduring surgery in the operating room.

The main obstacle is that HFOV can-not be continued during transport to theoperating room. The electrical power re-quirements of the oscillator preclude theuse of a battery, and two independent gassources must be maintained (pressurizedroom air and oxygen). Hence, the patientmust be temporarily detached from theoscillator, which poses a risk of alveolarderecruitment when the applied mPOaw isinterrupted. To compensate for this, wehave developed the following protocol. Inthe burn center, the patient is preoxygen-ated with 100% oxygen. The anesthesiol-ogist transiently clamps the ETT with aKelly clamp just distal to the teeth (Fig.3). Next, the HFOV circuit is detachedand is exchanged for a Laerdal bag, withthe positive end-expiratory pressure valveset at 20 cm H2O, which is then reat-tached to the still clamped ETT. Whenthe Laerdal bag is attached, the ETT isunclamped. We believe that these stepsminimize alveolar derecruitment due tothe loss of mPOaw when HFOV is inter-rupted. Next, the burn unit respiratorytherapist transfers the oscillator to theoperating room ahead of the patientwhile the anesthesiologist manually ven-tilates the patient with 100% oxygen us-ing short rapid breaths while followingbehind. In the operating room, the samesequence of ETT clamping and unclamp-ing is used to switch the patient from theLaerdal bag to the HFOV circuit. HFOVthen resumes using the same mPOaw andan FIO2 of 1.0. At this point, it is oftennecessary to perform one or more LRMs(mPOaw increased to 40 cm H2O for 20–40secs) if there has been any decrease inoxygen saturation by pulse oximetry. Ifoxygenation has remained stable, patientpositioning and surgery then proceed asusual after induction of general anesthe-

Figure 3. Transient clamping of the endotracheal tube before detachment of the high-frequencyoscillatory ventilation circuit (bottom left) and reattachment of the Laerdal bag (right).


sia using a total intravenous anesthetic(typically infusions of fentanyl and propo-fol are added to the existing infusions ofmorphine and benzodiazepines). If pronepositioning is needed, the aforemen-tioned steps of clamping and unclampingthe ETT are taken to disconnect the pa-tient during the positioning. During theprocedure, the FIO2 and mPOaw are ad-justed by the anesthesiologist to maintainan SpO2 of �92%, the goal being to re-sume the preoperative FIO2 level as soonas possible. Frequency and power are ma-nipulated to maintain the PaCO2 between35 and 60 mm Hg and the pH at �7.25.At the conclusion of the procedure, thesteps of clamping and unclamping theETT during changeover between HFOVand the transport Laerdal bag ventilation,as described above, are repeated.

To date, we have used HFOV for 33surgical procedures in 18 patients. ThePaO2/FIO2 ratio and OI of these patients,just before being converted from CMV toHFOV, were 88 � 17 and 28 � 13, re-spectively. These indices reflect severeoxygenation failure on near maximalCMV and highlight the precarious state ofthese patients who required major sur-gery under general anesthesia. However,with initiation of HFOV and after a meanof 84 � 93 hrs (range, 0–264 hrs) oftreatment, the preoperative PaO2/FIO2 ra-tio and OI had improved significantly to216 � 106 and 19 � 10, respectively.These are levels of oxygenation at whichthere is clearly considerably more flexi-bility and safety. These improvements inoxygenation illustrate the concept of us-ing HFOV to create a window of oppor-tunity to allow surgery to proceed. Onecould argue that at these levels, continu-ation of HFOV was unnecessary and thatresumption of CMV for surgery wouldhave been feasible. However, as describedabove, we believe this would have carriedthe risk of allowing renewed alveolar de-recruitment, along with loss of the poten-tial lung-protective benefits of HFOV.During the operative procedures, a mean� SD of 20% � 11% of total body surfacearea burn was excised and closed withautograft, allograft, or skin substitutesduring a mean operating room durationof 302 � 107 mins. Prone positioningoccurred in 10 of 33 of the procedures(30%), and iNO was continued intraoper-atively for two patients at concentrationsof 7 and 10 ppm, with one patient receiv-ing iNO and HFOV while prone. Themean � SD postoperative PaO2/FIO2 ratioand OI were 242 � 78 and 17 � 6, re-

spectively, which did not differ signifi-cantly from preoperative values. Theminimum FIO2 achieved intraoperativelywas 0.6 � 0.2. There have been no com-plications associated with intraoperativeHFOV use.

Conclusion

HFOV has been particularly useful inthe care of our burn patients with ARDS.Initially, HFOV was used as a rescue ven-tilation strategy for severe oxygenationfailure. However, it has been used pro-gressively earlier in the course of burnpatients care and has facilitated transferof patients to the operating room for sur-gical excision and closure of the burnwound. Our enthusiasm for HFOV in thesmoke inhalation patient is somewhatguarded at this time, and larger, prospec-tive, randomized clinical trials are neededto assess the usefulness of HFOV aftersmoke inhalation.

REFERENCES

1. Ware LB, Matthay MA: The acute respiratorydistress syndrome. N Engl J Med 2000; 342:1334–1348

2. Dancey DR, Hayes J, Gomez M, et al: ARDS inpatients with thermal injury. Intensive CareMed 1999; 25:1231–1236

3. Bernard GR, Artigas A, Carlet J, et al: TheAmerican–European Consensus Conferenceon ARDS- definitions, mechanisms, relevantoutcomes and clinical trial co-ordination.Am J Respir Crit Care Med 1994; 149:818–824

4. Amato MB, Barbas CS, Medeiros DM, et al:Effect of a protective ventilation strategyon mortality in the acute respiratory dis-tress syndrome. N Engl J Med 1998; 338:347–354


6. Stewart TE: Lung protection during me-chanical ventilation. Ontario Thorac Rev1996; 9:1–4

7. Tompkins RG, Burke JF, Schoenfeld DA, etal: Prompt eschar excision: A treatment sys-tem contributing to reduced burn mortality.Ann Surg 1986; 204:272–281

8. Thompson P, Herndon DN, Abston S, et al:Effects of early excision on patients with ma-jor thermal injury. J Trauma 1987; 27:205–207

9. Munster AM, Smith-Meek M, Sharkey P: Theeffect of early surgical intervention on mor-tality and cost-effectiveness in burn care,1978–91. Burns 1994; 20:61–64

10. Froese AB: High-frequency oscillatory venti-

lation for the acute respiratory distress syn-drome: Let’s get it right this time! Crit CareMed 1997; 25:906–908

11. Simma B, Luz G, Trawager R, et al: Compar-ison of different modes of high frequencyventilation in surfactant deficient rabbits.Pediatr Pulmonol 1996; 22:263–270

12. Shimaoku M, Fujino Y, Taenaka N, et al:High frequency oscillatory ventilation atten-uates the activation of alveolar macrophagesand neutrophils in lung injury. Crit Care1998; 2:35–39

13. Imai Y, Nakagawa S, Ito Y, et al: Compar-ison of lung protection strategies usingconventional and high frequency oscilla-tory ventilation. J Appl Physiol 2001; 91:1836 –1844

14. Merz U, Schfels J, Hendricks H, et al: Com-bination of high frequency oscillatory venti-lation, NO inhalation, and surfactant replace-ment in a child with acute respiratorydistress syndrome [in German]. Klin Pediatr1999; 211:83–85

15. Andersen FA, GuttormsenAB, Flaatten HK:High frequency oscillatory ventilation inadult patients with acute respiratory distresssyndrome: A retrospective study. Acta An-aesth Scand 2002; 46:1082–1088

16. Whitford KM, Hibbert MJ, Caruso DM, et al:A new strategy for burn patients that failconventional ventilation: High-frequency os-cillation. Abstr. J Burn Care Rehabil 2003;24:S122

17. Cartotto R, Cooper AB, Esmond JR, et al:Early clinical experience with high-fre-quency oscillatory ventilation for ARDS inadult burn patients. J Burn Care Rehabil2001; 22:325–333

18. Cartotto R, Ellis S, Gomez M, et al: Highfrequency oscillatory ventilation in burn pa-tients with the acute respiratory distress syn-drome. Burns 2004; 30:453–463

19. Fort P, Farmer C, Westerman J, et al: HFOVfor adult acute respiratory distress syn-drome: A pilot study. Crit Care Med 1997;25:937–947

20. Mehta S, Lapinsky SE, Hallett DC, et al:Prospective trial of HFOV in adults withacute respiratory distress syndrome. CritCare Med 2001; 29:1360–1369

21. Carmen B, Cahill T, Warden G, et al: Aprospective randomized comparison of thevolumetric diffusive respirator vs conven-tional ventilation for ventilation of burnedchildren. J Burn Care Rehabil 2002; 23:444 – 447

22. Cioffi WG, Graves TA, McManus WF, et al:High frequency percussive ventilation in pa-tients with inhalation injury. J Trauma 1989;29:350–354

23. Reper P, Dankaert R, van Hille F, et al: Theusefulness of combined high frequency per-cussive ventilation during acute respiratoryfailure after smoke inhalation. Burns 1998;24:34–38


24. Cioffi WG, Loring WR, Graves TA, et al: Pro-phylactic use of high frequency percussiveventilation in patients with inhalation injury.Ann Surg 1991; 213:575–580

25. Reper P, Wibaux O, van Laeke P, et al: Highfrequency percussive ventilation and conven-tional ventilation after smoke inhalation: Arandomized study. Burns 2002; 28:503–508

26. Derdak S, Mehta S, Stewart TE, et al: High-frequency oscillatory ventilation for acute re-spiratory distress syndrome in adults: A ran-domized, controlled trial. MulticenterOscillatory Ventilation for Acute RespiratoryDistress Syndrome Trial (MOAT) Study In-

vestigators. Am J Respir Crit Care Med 2002;166:801–808

27. Froese AB: The incremental application oflung protective high-frequency oscillatoryventilation. Am J Respir Crit Care Med 2002;166:786–787

28. 3100B High-Frequency Oscillatory Ventila-tor Operator’s Manuel. Yorba Linda, CA,SensorMedics, 2002

29. Derdak S: High-frequency oscillatory venti-lation for acute respiratory distress syn-drome in adult patients. Crit Care Med 2003;31:S317–S323

30. Mehta S, Granton J, MacDonald RJ, et al:

High-frequency oscillatory ventilation inadults: The Toronto experience. Chest 2004;126:518–527

31. Cooper AB, Islur A, Gomez M, et al: Hyper-capnic respiratory failure and partial upperairway obstruction during high-frequencyoscillatory ventilation in an adult burnpatient. Can J Anesth 2002; 49:724 –728

32. Cioffi WG, deLemos RA, Coalson JL, et al:Decreased pulmonary damage in primateswith inhalation injury treated with high-frequency ventilation. Ann Surg 1993; 218:328–337


High-frequency oscillatory ventilation and adjunctive therapies:Inhaled nitric oxide and prone positioning

Eddy Fan, MD; Sangeeta Mehta, MD, FRCPC

Hypoxemic respiratory failurefrom acute lung injury (ALI)and acute respiratory dis-tress syndrome (ARDS) con-

tinues to be a leading cause of admissionto the intensive care unit (ICU) (1). On-going basic science and clinical researchhas increased our knowledge of the un-derlying pathophysiology of ARDS buthas not translated into any significanttherapeutic advances. Supportive careand mechanical ventilation (MV) are themainstays of treatment. However, tradi-tional approaches to MV in ARDS withliberal tidal volumes and pressures havebeen demonstrated to be injurious, with avolume- and pressure-limited lung-protective strategy resulting in a signifi-cant survival advantage (2). The survivalbenefit has been postulated to be the re-sult of decreased ventilator-associatedlung injury (VALI) resulting from highinflation pressures (barotrauma), overd-

istension (volutrauma), and repetitive re-cruitment– derecruitment (atelec-trauma). Despite increasing acceptanceof this new standard (3), patients maydevelop progressive refractory hypox-emia. As a result, there is much interestin exploring other ventilation strategiesthat may further improve ARDS out-comes.

High-frequency oscillatory ventilation(HFOV) is an alternative mode of ventila-tion that theoretically fulfills the goals oflung-protective ventilation (4). HFOV hasbeen most extensively studied in criticallyill neonates with respiratory distress syn-drome (RDS) (5, 6). The ventilation strat-egy uses both active inspiration and ex-piration at frequencies between 3 and 10Hz with the delivery of subdead spacetidal volumes. Studies of HFOV in adultswith ARDS (4, 7–10) have been less com-pelling, and its place in the therapeuticarmamentarium for ARDS remains un-clear.

Various adjunctive therapies to con-ventional mechanical ventilation (CMV)have been evaluated in patients withARDS, including inhaled nitric oxide(iNO) and prone positioning (PP). Al-though iNO is beneficial in some criti-cally ill neonates, both iNO and PP have

not shown significant benefit in adultclinical studies when used as adjunctivetherapy to CMV. However, both tech-niques continue to be used alone or incombination in refractory ARDS as res-cue therapy. In this review, we examinethe evidence for these therapies alone(with CMV), and in combination (HFOVwith iNO and/or PP), in adults with re-fractory ALI/ARDS.

Physiological Rationale for InhaledNitric Oxide in Acute Lung Injury/AcuteRespiratory Distress Syndrome. Inhalednitric oxide is a selective pulmonary va-sodilator that acts on the alveolar endo-thelium to produce regional vasodilationin well-ventilated lung units where it isdistributed. Various studies have demon-strated that iNO can improve ventilation–perfusion mismatch, hypoxemia, and pul-monary hypertension that are allcharacteristic of ARDS (11). This is instark contrast to the use of systemic va-sodilators, which often aggravate ventila-tion–perfusion mismatch by causing dif-fuse pulmonary vasodilation, includingareas of nonventilated lung. Inhaled NOalso has bronchodilatory, antiinflamma-tory, and antiproliferative effects in thelung (12), which have been exploited withvariable success in reducing ischemia–

From the Interdepartmental Division of CriticalCare, University of Toronto, Toronto, Canada; and theDepartment of Medicine, Mount Sinai Hospital, To-ronto, Canada.


DOI: 10.1097/01.CCM.0000155927.54034.34

Objective: To review the use of high-frequency oscillatoryventilation (HFOV) with adjunctive therapies (inhaled nitric oxide[iNO] and prone positioning [PP]) in adult patients with acuterespiratory distress syndrome (ARDS).

Data Sources: Published studies evaluating the use of iNO, PP,and HFOV in adult patients with ARDS.

Data Summary: Despite ongoing preclinical and clinical re-search, the therapeutic armamentarium for ARDS remains limited.Although a pressure- and volume-limited strategy aimed at mit-igating ventilator-associated lung injury has demonstrated mor-tality benefit, patients with severe ARDS may still develop life-threatening hypoxemia. As a result, various salvage therapiesaimed at improving oxygenation, including HFOV, iNO, and PPalone or in combination, have been evaluated in patients withrefractory ARDS. Although the few preclinical and clinical trials of

combination therapy to date have shown promising improvementsin oxygenation and other physiological variables, with few ad-verse clinical events, the impact on survival awaits the perfor-mance of large randomized trials.

Conclusions: There is limited clinical data to recommend thewidespread use of combination therapy in patients with ARDS. In thesubset of patients with life-threatening hypoxemia from refractoryARDS, combination therapy is safe and may be considered for sal-vage therapy. More rigorous randomized, controlled trials are neededto help delineate the therapeutic role of combination therapy inadults with ARDS. (Crit Care Med 2005; 33[Suppl.]:S182–S187)

KEY WORDS: acute lung injury; acute respiratory distress syn-drome; high-frequency oscillatory ventilation; inhaled nitric oxide;intensive care; mechanical ventilation; oxygenation; prone posi-tion; ventilator-associated lung injury


reperfusion lung injury after lung trans-plantation (13, 14). After inhalation, iNOdiffuses into the bloodstream, where it israpidly metabolized and its inactive by-products are excreted in the urine (12).Thus, iNO has predominantly local effectswith virtually no systemic vasodilation.

Clinical Trials of Inhaled Nitric Oxidefor Acute Lung Injury/Acute RespiratoryDistress Syndrome. Given its theoreti-cally favorable effects, iNO has been eval-uated in a number of clinical trials inboth adults and children with acute hy-poxemic respiratory failure (including pa-tients with ALI, ARDS, and other diag-noses) (15–21). Five of the studies wereanalyzed in a recent systematic review(22) representing results from a hetero-geneous group of over 500 patients. Thereview concluded that although the useof iNO leads to a transient improvementin oxygenation for up to 72 hrs, this didnot translate into any significant survivaladvantage (pooled relative risk [RR] usingfixed-effects model 0.98; 95% confidenceinterval [CI] 0.66–1.44). There was nodifference in ventilator-free days betweenthe iNO treatment and control groups. Adose-dependent relationship between iNOand the measured outcomes was not ob-served.

Although many of the trials enrolledmodest numbers and lacked significantmethodologic rigor, these preliminary re-sults suggest that iNO may be used asrescue therapy in refractory hypoxemia toimprove oxygenation. The lack of mortal-ity benefit is not surprising, given thatmost of these patients die of sepsis andmultiorgan failure (MOF) rather than hy-poxemia (23, 24). Thus, a therapy de-signed to treat hypoxemia alone may notbe expected to have a significant impacton mortality. This point remains contro-versial, however, because recent data sug-gest that hypoxemia in patients withacute respiratory failure (ARF) may berelated to mortality in a dose-dependentfashion (25).

Although iNO may transiently im-prove oxygenation, there is little evidencethat this leads to a subsequent decreasein the intensity of mechanical ventila-tion, commonly measured by the oxygen-ation index (OI; mean airway pressuredivided by PaO2/FIO2 � 100). A reductionin OI could be theoretically advantageousas VALI has been implicated in potentiat-ing and perpetuating the systemic in-flammatory response (biotrauma) andMOF seen in ALI/ARDS (26). Two trials(15, 16) reported a decrease in OI with

iNO administration. This effect, however,is related to an improvement in PaO2/FIO2

ratio rather than a change in mean air-way pressure (mPaw) (27). The inabilityto reduce mPaw may partly explain whyiNO administration did not lead to a mor-tality benefit in these trials. Of note, noneof the trials used a lung protective venti-lation strategy with low tidal volumes,although most endeavored to limit pla-teau pressures. This distinction is impor-tant because there is no safe level of pla-teau pressure below which limitation oftidal volumes is not advantageous (27,28).

More recently, Taylor et al. (20) con-ducted a randomized, placebo-controlledtrial of iNO at five parts per million (ppm)in 385 patients with moderately severeALI/ARDS. Patients with any nonpulmo-nary organ dysfunction and/or sepsiswere excluded, because benefits to pul-monary optimization were unlikely to af-fect mortality in these patients. This spe-cific dose was used based on a post hocanalysis of a previous trial (16) that sug-gested improved ventilator-free survivalat 28 days with iNO at 5 ppm when com-pared with doses ranging from 1.25 to 80ppm. Of note, no patients received HFOVas a cointervention. As seen in previoustrials, oxygenation improved up to 24 hrswith iNO but there was no difference inmortality (23% iNO vs. 20% placebo; p �.54) or ventilator-free survival (mean,10.7 days iNO vs. 10.6 days placebo; p �.97) between groups. Overall, the numberof adverse events was similar in bothgroups, although there were more noso-comial infections in the iNO group. Theauthors conclude, in keeping with previ-ous studies, that routine use of iNO inmoderately severe ALI/ARDS without sep-sis or MOF is not warranted, although itmay be considered as salvage therapy inpatients with life-threatening hypoxemiadespite optimal CMV support (22).

Dose Response of Inhaled Nitric Oxidein Acute Lung Injury/Acute RespiratoryDistress Syndrome. Most clinical trials ofiNO in ALI/ARDS have not demonstratedimproved outcomes with continuousfixed-dose iNO (usually �5 ppm) (15–20).However, it is clear that the effect of iNOis dose-dependent and is heterogeneousamong individual patients (29–32). Infact, lower iNO concentrations (�1 ppm)may be as effective as higher doses atimproving oxygenation in ARDS (33).Furthermore, patients exhibit time-dependent variation in the dose-responseto iNO with respect to oxygenation and

pulmonary vascular resistance (PVR), re-sulting from increased sensitization ofthe pulmonary vasculature to lower iNOconcentrations (32). Thus, the beneficialeffects of iNO on oxygenation and pulmo-nary hypertension may be attenuated bycontinuous fixed-dose regimens leadingto unintentional “overdosing.” Althoughthis hypothesis has yet to be proven, itmay be an important contributor to thenegative results in recent studies of iNOin ALI/ARDS.

Physiological Rationale for Prone Po-sitioning in Acute Lung Injury/Acute Re-spiratory Distress Syndrome. Althoughthe exact physiological mechanism bywhich PP augments oxygenation in ALI/ARDS is not known, it is thought thatimproved respiratory mechanics and de-creased regional ventilation–perfusion(V/Q) mismatch contribute to this effect(34). In the prone position, there is effec-tive reduction in the pleural pressure ondorsal (nondependent) lung regions,leading to recruitment of these previ-ously atelectatic areas (35). In fact, as aresult of the improved fit of the lungsinto the thorax with PP, the degree ofhyperinflation and atelectrauma is re-duced in patients ventilated in the proneposition, irrespective of the tidal volume,positive end-expiratory pressure (PEEP),or recruitment strategy used (36). Secre-tion drainage may be more efficient inthe prone position (37). Experimental an-imal models have also suggested im-proved and/or sustained responses to re-cruitment maneuvers (38), as well as areduction and redistribution of the inju-rious mechanical forces associated withVALI (39, 40) in the prone position com-pared with the supine position.

Clinical Trials of Prone Positioningfor Acute Lung Injury/Acute RespiratoryDistress Syndrome. Improvement in ox-ygenation during PP was first reported ina retrospective study of five patients withARDS in 1976 (41), although its theoret-ical benefits were espoused a few yearsearlier (42). Many small prospective stud-ies of PP in ALI/ARDS have been per-formed in the past 20 yrs, encompassingover 450 patients (34). However, only onelarge, multicentered, randomized, con-trolled trial of PP in ALI/ARDS has beenreported in the literature (43).

In this study, 304 patients with ALI/ARDS were randomized to supine venti-lation or proning for at least 6 hrs per day(mean, 7.0 � 1.8 hrs per day) over a10-day period. The PaO2/FIO2 ratio im-proved significantly in the prone group;


however, there was no difference in in-tensive care unit discharge (50.7% vs.48.0%; relative risk [RR] 1.05; 95% con-fidence interval [CI] 0.84–1.32), 10-daymortality (21.1% vs. 25.0%; RR 0.84; 95%CI 0.56–1.27), or 6-mo mortality (62.5%vs. 58.6%; RR 1.06; 95% CI 0.88–1.28)between prone and supine groups. Therewas no difference between groups in dayswithout nonpulmonary organ dysfunc-tion. A post hoc analysis of 10-day mor-tality in the most severely affected pa-tients (by quartiles), based on PaO2/FIO2

ratio (mean ratio, �88), Simplified AcutePhysiology Score (SAPS) II (�49), orhighest tidal volume (�12 mL/kg pre-dicted body weight), revealed a signifi-cant survival advantage to PP (20.5% vs.40.0% for at least one characteristic; RR0.54; 95% CI 0.32–0.90) in these patientsubgroups. However, the difference inmortality did not persist beyond intensivecare unit discharge.

Complications related to PP includeda greater number of pressure sores perprone patient, especially in those areassubject to more pressure in the proneposition (i.e., thorax, cheekbone, iliaccrest, breast, and knee). Interestingly,there was no difference in the percentageof patients with new or worsening pres-sure sores, or with displacement of endo-tracheal tubes, vascular catheters, or tho-racotomy tubes between groups. Themost frequent adverse events resultingfrom PP were the need for increased se-dation, immediate airway suctioning onturning, and facial edema.

In conclusion, PP in ALI/ARDS is rea-sonably well tolerated, with few seriouscomplications, and can lead to improvedoxygenation. This effect is partially pre-

served after the patient is returned to thesupine position. Like iNO therapy, thisphysiological improvement does nottranslate into a clear survival advantage,although a single post hoc analysis sug-gests benefit in a more severely ill sub-population of patients with ALI/ARDS(43). Pending the results of further clin-ical trials, the routine use of PP in ALI/ARDS is not warranted and should bereserved for critically ill ALI/ARDS pa-tients with refractory hypoxemia.

Combination Therapy: Preclinicaland Clinical Studies in Adult Acute LungInjury/Acute Respiratory Distress Syn-drome. Although both iNO and PP havenot demonstrated clear mortality benefitin ALI/ARDS individually, there is muchspeculation and interest in the possibilityof additive or synergistic effects whenused in combination with either CMV orHFOV. Several small clinical studiescombining PP and/or iNO with CMV havebeen conducted in adult patients withARDS (44–47) (Table 1). None of thesestudies were designed or powered forhard clinical outcomes such as mortality,but were performed to elucidate possiblesynergistic or additive effects of these dif-ferent therapeutic modalities. Thesestudies demonstrate both an independenteffect of iNO and PP on oxygenation, aswell as significant additive effects of iNOcombined with PP on oxygenation. Nosignificant interaction between the twotreatments was observed (47).

The physiological rationale for com-bining iNO with HFOV is sound. HFOVrecruits alveolar volume, thereby increas-ing the surface area for iNO to act, po-tentially improving V/Q matching to agreater degree than either of these ther-

apies alone. This hypothesis is supportedby a few studies in which iNO nonre-sponders were converted to iNO respond-ers after positive end-expiratory pressure-induced alveolar recruitment (48 –50).Early animal studies have demonstratedthat the use of HFOV in ALI/ARDS resultsin an early and transient improvement inoxygenation, more homogeneous lunginflation, and decreased histopathologicevidence of VALI (51–53). In a subse-quent study by Kinsella et al. (54) inpremature lambs, the use of lung recruit-ment strategies (HFOV and/or partial liq-uid ventilation) augmented the oxygen-ation effect of low-dose iNO (5 ppm).Lung neutrophil accumulation was alsoreduced by this strategy (54). Similar re-sults have been obtained in clinical trialsevaluating HFOV and iNO use in persis-tent pulmonary hypertension of the new-born (55) and neonatal/pediatric ARDS(56).

There is a paucity of clinical data re-garding the use of adjunctive therapieswith HFOV (Table 2). Published evidenceis limited to case reports (57) or subsetsof patients in other HFOV trials (9, 10).To date, only one prospective clinical trialof HFOV and iNO has been reported inthe literature (58). No clinical studies us-ing all three modalities simultaneouslyhave been conducted, although the suc-cessful use of combined HFOV, iNO, andPP in a patient with ARDS has been re-ported (57).

In the study designed to evaluate theequivalence of HFOV and CMV, Derdak etal. (9) randomized 148 patients withARDS to HFOV or CMV. Goals of ventila-tion were similar in both groups, al-though the CMV protocol was not de-

Table 1. Combination therapy with conventional mechanical ventilation (CMV)

StudyARDS

Patients Strategy Outcomes Adverse events

Papazian et al. (44)(1998)

14 Volume control CMV combinedwith iNO and/or PP

Additive, nonsynergistic effects ofiNO � PP on increasing PaO2/FIO2

ratio and decreasing shuntfraction

No clinically relevantdeleterious respiratory orhemodynamic effects noted

Martinez et al. (45)(1999)

14 Volume assist/control CMVcombined with iNO and/orPP

Additive effects of iNO � PP onincreasing PaO2/FIO2 ratio anddecreasing venous admixture

No clinically relevant adverseevents from iNO or PP

Dupont et al. (46)(2000)

27 Volume control CMV combinedwith iNO followed by PP

PP improved hypoxemia significantlybetter than iNO; no interactionbetween therapies seen

Mild cutaneous and mucosaldamage in seven patients(26%) from PP

Borelli et al. (47)(2000)

14 Volume control CMV combinedwith iNO and/or PP (2 � 2factorial design)

Additive effects of iNO � PP onincreasing PaO2; no interactionbetween therapies seen

No clinically relevant adverseevents from iNO or PP

ARDS, acute respiratory distress syndrome; iNO, inhaled nitric oxide; PP, prone positioning.


signed to limit inspiratory plateau orpeak pressures. The HFOV group demon-strated significant early improvement inoxygenation as compared with the CMVgroup; however, this effect did not persistbeyond 24 hrs. There was a trend towarddecreased mortality at 30 days and at 6mos in the HFOV group, but this did notreach statistical significance (37% vs.52%, p � .102 at 30 days; 47% vs. 59%, p� .143 at 6 mos). Of the 75 patientsrandomized to HFOV, seven (9%) re-ceived rescue therapies, including fourpatients who received iNO and two whowere prone. Of the 73 patients random-ized to CMV, 12 (16%) received rescuetherapies, including eight patients whoreceived iNO and three who were prone.In this small subset of patients who re-ceived adjunctive therapies, there was nosignificant difference in 30-day mortalitybetween the two groups (71% HFOV vs.50% CMV, p � not significant).

In a retrospective study by Mehta et al.(10), 156 patients with severe ARDStreated with HFOV were examined. Acuteimprovement in oxygenation, as mea-sured by PaO2/FIO2 ratio and OI, was ob-served and persisted for the duration ofthe 72-hr study period. Patients receivedan average of 5 days of HFOV. Mortality at30 days was 61.7% in this cohort. Two-thirds of patients received some form ofadjunctive therapy, including iNO in 68patients (43.6%) and PP in ten patients(6.4%). There were no significant differ-ences in the number of patients receiving

adjunctive therapy among survivors andnonsurvivors.

In the only prospective study of HFOVcombination therapy reported, Mehta etal. (58) treated 23 adult patients withARDS with oxygenation failure despiteongoing HFOV, with various doses of iNO(5–20 ppm) according to a predefined al-gorithm. The majority of patients (83%)responded with improved oxygenation(�20% increase in PaO2/FIO2 ratio) afteriNO administration (mean improvementin PaO2/FIO2 ratio 38% at 30 mins), allow-ing significant reductions in FIO2 within8–12 hrs of initiation. Although therewas also a reduction in OI, this was com-pletely attributable to improvements inPaO2/FIO2 ratio, because mean airwaypressure (Paw) on HFOV remained con-stant during iNO therapy. There was noclear iNO dose (5, 10, or 20 ppm) associ-ated with peak improvement in PaO2/FIO2

ratio among all responders. Intensivecare unit and hospital survival rates were39% and 30%, respectively. All deathswere the result of MOF or withdrawal oflife support. Nonsurvivors tended to beolder, have higher baseline OI andAPACHE II scores, and were ventilatedlonger with HFOV before starting iNO.The latter may be an important factorbecause a recent study of iNO in childrenwith ARDS (59) demonstrated that earlytreatment (median 1.5 hrs) with iNO ledto acute and sustained improvements inoxygenation, with earlier reduction of

ventilation intensity, and subsequent im-provement in survival.

Finally, there is only a single case re-port in the literature of the successful useof HFOV, iNO, and PP combination ther-apy in an adult patient with ARDS (57).The report described a 56-yr-old manwho developed ARDS after a multidrugoverdose with subsequent aspiration ofgastric contents. Despite heavy sedationand neuromuscular blockade, the patientdeveloped progressive hypoxemia onmaximal CMV support. Therapy with iNOwas instituted, which led to an early im-provement in oxygenation and allowedtitration of the FIO2 from 1.0 to 0.55.However, this response was not sus-tained, and he continued to deterioratewith both ventilation and oxygenationfailure despite high peak inspiratory pres-sures (40 cm H2O). He was placed onHFOV with a Paw of 32 cm H2O, which ledto an improvement in ventilation. How-ever, despite increasing Paw, he had wors-ening hypoxemia, and he was subse-quently placed prone. This led to animmediate improvement in his oxygen-ation and allowed for reductions in FIO2

and Paw. After 4 days of combined HFOV,iNO, and PP, he was returned to CMV inthe supine position. He was graduallyweaned from iNO after 9 days of therapy.After a prolonged intensive care unit andhospital admission, the patient was dis-charged from the hospital. Obviously, noclear conclusions regarding combinationtherapy can be elucidated from the single

Table 2. Combination therapy with high-frequency oscillatory ventilation (HFOV)

StudyARDS

Patients Strategy Outcomes Adverse events

Derdak et al. (9)(2002)

148 HFOV vs. CMV;cointerventions includediNO (4 in HFOV group, 8in CMV group) and PP (2in HFOV, 3 in CMV)

No significant difference in mortalitybetween HFOV and CMV groupsreceiving adjunctive therapies, oroverall 30-day mortality (37%HFOV vs. 52% CMV, p � .102)

No significant difference inhemodynamic variables,oxygenation failure, ventilationfailure, barotrauma, or mucusplugging between groups

Mehta et al. (10)(2004)

156 Retrospective review ofpatients treated with HFOV;43.6% treated with iNO;6.4% treated with PP

No significant difference in use ofadjunctive therapies betweensurvivors vs. nonsurvivors; overall30-day mortality 61.7%

26% did not tolerate HFOV and werereturned to CMV; 21.8% had apneumothorax

Varkul et al. (57)(2001)

1 Case report of single patienttreated with HFOV, iNO,and PP

iNO days 4–12 (from intensive careunit admission); HFOV days 5–9;PP days 6–9

No significant clinically adverseoutcomes seen

Mehta et al. (58)(2003)

23 HFOV combined with iNO(5–20 ppm)

Significantly increased PaO2/FIO2

ratio after iNO administration;trend toward decreased OI

Five patients developedpneumothorax; 1 patient hadcardiac arrest; 1 each developedintracranial and gastrointestinalbleeding

ARDS, acute respiratory distress syndrome; CMV, conventional mechanical ventilation; iNO, inhaled nitric oxide; PP, prone positioning; OI, oxygenationindex.


case report, and more clinical trials areneeded.

CONCLUSIONS

Currently, the only ventilation strat-egy that improves survival in patientswith ARDS incorporates both plateaupressure and tidal volume limitation, po-tentially leading to attenuation of VALI(2, 26). Although many other therapies,including novel ventilation protocols andpharmacologic agents, have been pro-posed and tested in ARDS, none havedemonstrated any mortality benefit.Studies of HFOV in adult patients withARDS have shown it to be a safe andeffective mode of ventilation, especiallywhen initiated early, in patients failingCMV (4, 60).

Most patients with ARDS succumb toMOF. However, some patients with ARDSwill develop refractory life-threateninghypoxemia. In these patients, treatmentstrategies that include HFOV with iNOand/or PP may lead to acute improve-ments in oxygenation, providing an im-portant temporary bridge to possible re-covery. The use of these therapies, aloneor in combination, may indirectly lead toimproved outcomes by mitigating the ex-posure to, or need for, intensified CMVsupport that may result in progressiveARDS and MOF resulting from VALI.

Many questions remain unansweredregarding the use of HFOV, iNO, and PPin adult patients with ARDS. More rigor-ous randomized clinical trials comparingstandard lung protective ventilation andHFOV, with or without adjunctive thera-pies, are needed to ascertain their thera-peutic niche. Until then, the routine andsystematic use of these treatments can-not be recommended, and their use

should be restricted to salvage therapy forpatients with severe, refractory, life-threatening hypoxemia failing conven-tional supportive measures.

REFERENCES

1. Rubenfeld GD: Epidemiology of acute lunginjury. Crit Care Med 2003; 31:S276–S284

2. The Acute Respiratory Distress Network In-vestigators: Ventilation with lower tidal vol-umes as compared with traditional tidal vol-umes for acute lung injury and the acuterespiratory distress syndrome. N Engl J Med2000; 342:1301–1308

3. Young MP, Manning HL, Wilson DL, et al:Ventilation of patients with acute lung injuryand acute respiratory distress syndrome: hasnew evidence changed clinical practice? CritCare Med 2004; 32:1260–1265

4. Ritacca FV, Stewart TE: Clinical review:High-frequency oscillatory ventilation inadults—a review of the literature and prac-tical applications. Crit Care 2003; 7:385–390

5. Courtney SE, Durand DJ, Asselin JM, et al:Pro/con clinical debate: high-frequency os-cillatory ventilation is better than conven-tional ventilation for premature infants. CritCare 2003; 7:423–426

6. Mehta S: High-frequency ventilation. CritCare 2000; 6:38–45

7. Fort P, Farmer C, Westerman J, et al: High-frequency oscillatory ventilation for adult re-spiratory distress syndrome: a pilot study.Crit Care Med 1997; 25:937–947

8. Mehta S, Lapinsky SE, Hallet DC, et al: Aprospective trial of high-frequency oscilla-tion in adults with acute respiratory distresssyndrome. Crit Care Med 2001; 29:1360–1369

9. Derdak S, Mehta S, Stewart TE, et al: High-frequency oscillatory ventilation for acute re-spiratory distress syndrome in adults. Am JRespir Crit Care Med 2002; 166:801–808

10. Mehta S, Granton J, MacDonald RJ, et al:High-frequency oscillatory ventilation inadults. The Toronto experience. Chest 2004;126:518–527

11. Rossaint R, Falke KJ, Lopez F, et al: Inhalednitric oxide for the adult respiratory distresssyndrome. N Engl J Med 1993; 328:399–405

12. Ichinose F, Roberts JD, Zapol WM: Inhalednitric oxide. A selective pulmonary vasodila-tor. Current uses and therapeutic potential.Circulation 2004; 109:3106–3111

13. Ardehali A, Laks H, Levine M, et al: A pro-spective trial of inhaled nitric oxide in clini-cal lung transplantation. Transplantation2001; 72:112–115

14. Meade MO, Granton JT, Matte-Martyn A, etal: A randomized trial of inhaled nitric oxideto prevent ischemia-reperfusion injury afterlung transplantation. Am J Respir Crit CareMed 2003; 167:1483–1489

15. Dobyns EL, Cornfield DN, Anas NG, et al:Multicenter randomized controlled trial ofthe effects of inhaled nitric oxide therapy on

gas exchange in children with acute hypox-emic respiratory failure. J Pediatr 1999; 134:406–412

16. Dellinger RP, Zimmerman JL, Taylor RW, etal: Effects of inhaled nitric oxide in patientswith acute respiratory distress syndrome: re-sults of a randomized phase II trial. Crit CareMed 1998; 26:15–23

17. Troncy E, Collet JP, Shapiro S, et al: Inhalednitric oxide in acute respiratory distress syn-drome: a pilot randomized controlled study.Am J Respir Crit Care Med 1998; 157:1483–1488

18. Lundin S, Mang H, Smithies M, et al: Inha-lation of nitric oxide in acute lung injury:results of a European multicentre study. In-tensive Care Med 1999; 25:911–919

19. Michael JR, Barton RG, Saffle JR, et al: In-haled nitric oxide versus conventional ther-apy: effect on oxygenation in ARDS. Am JRespir Crit Care Med 1998; 157:1372–1380

20. Taylor RW, Zimmerman JL, Dellinger RP, etal: Low-dose inhaled nitric oxide in patientswith acute lung injury. JAMA 2004; 291:1603–1609

21. Mehta S, Simms HH, Levy MM, et al: Ex-tended therapy with inhaled nitric oxide failsto improve oxygenation in patients withacute respiratory distress syndrome: a ran-domized controlled trial. J Appl Res 2001;1:73–84

22. Sokol J, Jacobs SE, Bohn D: Inhaled nitricoxide for acute hypoxic respiratory failure inchildren and adults: a meta-analysis. AnesthAnalg 2003; 97:989–998

23. Montgomery AB, Stager MA, Carrico CJ, etal: Causes of mortality in patients with theadult respiratory distress syndrome. Am RevRespir Dis 1985; 132:485–489

24. Jiminez P, Torres A, Roca J, et al: Arterialoxygenation does not predict the outcome ofpatients with acute respiratory failure need-ing mechanical ventilation. Eur Respir J1994; 7:730–735

25. Esteban A, Anzueto A, Frutos F, et al: Char-acteristics and outcomes in adult patientsreceiving mechanical ventilation. A 28-dayinternational study. JAMA 2002; 287:345–355

26. Crimi E, Slutsky AS: Inflammation and theacute respiratory distress syndrome. BestPrac Res Clin Anaesth 2004; 18:477–492

27. Adhikari N, Granton JT: Inhaled nitric oxidefor acute lung injury. No place for NO? JAMA2004; 291:1629–1631

28. Brower RG, Matthay M, Schoenfeld D: Meta-analysis of acute lung injury and acute respi-ratory distress syndrome trials. Am J RespirCrit Care Med 2002; 166:1515–1517

29. Gerlach H, Rossaint R, Pappert D, et al:Time-course and dose-response of nitric ox-ide inhalation for systemic oxygenation andpulmonary hypertension in patients withadult respiratory distress syndrome. EurJ Clin Invest 1993; 23:499–502

30. Puybasset L, Rouby JJ, Mourgeon E, et al:Inhaled nitric oxide in acute respiratory fail-

I n the subset of pa-

tients with life-threat-

ening hypoxemia

from refractory acute respi-

ratory distress syndrome,

combination therapy is safe

and may be considered for

salvage therapy.


ure: dose-response curves. Intensive CareMed 1994; 20:319–327

31. Lowson SM, Rich GF, McArdle PA, et al: Theresponse to varying concentrations of in-haled nitric oxide in patients with acute re-spiratory distress syndrome. Anesth Analg1996; 82:574–581

32. Gerlach H, Keh D, Semmerow A, et al: Dose-response characteristics during long-terminhalation of nitric oxide in patients withsevere acute respiratory distress syndrome.Am J Respir Crit Care Med 2003; 167:1008–1015

33. Gerlach H, Pappert D, Lewandowski K, et al:Long-term inhalation with evaluated lowdoses of nitric oxide for selective improve-ment of oxygenation in patients with adultrespiratory distress syndrome. IntensiveCare Med 1993; 19:443–439

34. Pelosi P, Brazzi L, Gattinoni L: Prone posi-tion in acute respiratory distress syndrome.Eur Respir J 2002; 20:1017–1028

35. Mehta S: Lung volume recruitment. CurrOpin Crit Care 1998; 4:6–15

36. Albert RK: Turnabout may be more than fairplay. Crit Care Med 2000; 28:571–572

37. Marini JJ, Rubenfeld G: Pro/con clinical de-bate: the use of prone positioning in themanagement of patients with acute respira-tory distress syndrome. Crit Care 2002;6:15–17

38. Cakar N, Van der Kloot T, Youngblood M, etal: Oxygenation response to a recruitmentmaneuver during supine and prone positionsin an oleic acid-induced lung injury model.Am J Respir Crit Care Med 2000; 161:1949–1956

39. Broccard AF, Shapiro RS, Schmitz LL, et al:Influence of prone position on the extent anddistribution of lung injury in a high tidalvolume oleic acid model of acute respiratorydistress syndrome. Crit Care Med 1997; 25:16–27

40. Broccard AF, Shapiro RS, Schmitz LL, et al:Prone positioning attenuates and redistrib-utes ventilator-induced lung injury in dogs.Crit Care Med 2000; 28:295–303

41. Piehl MA, Brown RS: Use of extreme position

changes in acute respiratory failure. CritCare Med 1976; 4:13–14

42. Bryan AC: Comments of a devil’s advocate.Am Rev Respir Dis 1974; 110:143–144

43. Gattinoni L, Tognoni G, Pesenti A, et al:Effect of prone positioning on the survival ofpatients with acute respiratory failure.N Engl J Med 2001; 345:568–573

44. Papazian L, Bregeon F, Gaillat F, et al: Re-spective and combined effects of prone posi-tion and inhaled nitric oxide in patients withacute respiratory distress syndrome. Am JRespir Crit Care Med 1999; 157:580–585

45. Martinez M, Diaz E, Joseph D, et al: Improve-ment in oxygenation by prone position andnitric oxide in patients with acute respiratorydistress syndrome. Intensive Care Med 1999;25:29–36

46. Dupont H, Mentec H, Cheval C, et al: Short-term effect of inhaled nitric oxide and pronepositioning on gas exchange in patients withsevere acute respiratory distress syndrome.Crit Care Med 2000; 28:304–308

47. Borelli M, Lampati L, Vascotto E, et al: He-modynamic and gas exchange response toinhaled nitric oxide and prone positioning inacute respiratory distress syndrome. CritCare Med 2000; 28:2707–2712

48. Puybasset L, Rouby JJ, Mourgeon E, et al:Factors influencing cardiopulmonary effectsof inhaled nitric oxide in acute respiratoryfailure. Am J Respir Crit Care Med 1995;152:318–328

49. Johannigman JA, Davis K, Campbell RS, etal: Positive end-expiratory pressure and re-sponse to inhaled nitric oxide: changing non-responders into responders. Surgery 2000;127:390–394

50. Okamoto K, Kukita I, Hamaguchi M, et al:Combined effects of inhaled nitric oxide andpositive end-expiratory pressure during me-chanical ventilation in acute respiratory dis-tress syndrome. Artif Organs 2000; 24:390–395

51. Hamilton PP, Onayemi A, Smyth JA, et al:Comparison of high-frequency ventilation:oxygenation and lung pathology. J ApplPhysiol 1983; 55:131–138

52. McCulloch PR, Forkert PG, Froese AB: Lungvolume maintenance prevents lung injuryduring high frequency oscillatory ventilationin surfactant deficient rabbits. Am Rev RespirDis 1998; 137:1185–1192

53. Coalson JJ, deLemos RA: Pathologic featuresof various ventilatory strategies. Acta Anaes-thesiol Scand Suppl 1989; 90:108–116

54. Kinsella JP, Parker TA, Galan H, et al: Inde-pendent and combined effects of inhaled ni-tric oxide, liquid perfluorochemical, andhigh-frequency oscillatory ventilation in pre-mature lambs with respiratory distress syn-drome. Am J Respir Crit Care Med 1999;159:1220–1227

55. Kinsella JP, Truog WE, Walsh WF, et al:Randomized multicenter trial of inhaled ni-tric oxide and high-frequency oscillatoryventilation in severe, persistent pulmonaryhypertension of the newborn. J Pediatr 1997;131:55–62

56. Dobyns EL, Anas NG, Fortenberry JD, et al:Interactive effects of high-frequency oscilla-tory ventilation and inhaled nitric oxide inacute hypoxemic respiratory failure in pedi-atrics. Crit Care Med 2000; 30:2425–2429

57. Varkul MD, Stewart TE, Lapinsky SE, et al:Successful use of combined high-frequencyoscillatory ventilation, inhaled nitric oxide,and prone positioning in the acute respira-tory distress syndrome. Anesthesiology 2001;95:797–799

58. Mehta S, MacDonald R, Hallett DC, et al:Acute oxygenation response to inhaled nitricoxide when combined with high-frequencyoscillatory ventilation in adults with acuterespiratory distress syndrome. Crit Care Med2003; 31:383–389

59. Fioretto JR, de Moraes MA, Bonatto RC, et al:Acute and sustained effects of early adminis-tration of inhaled nitric oxide to childrenwith acute respiratory distress syndrome. Pe-diatr Crit Care Med 2004; 5:469–474

60. Derdak S: High-frequency oscillatory venti-lation for acute respiratory distress syn-drome in adult patients. Crit Care Med 2003;31(suppl):S317–323


Inhaled alternatives to nitric oxide

Stuart M. Lowson, MBBs, FRCA, MRCP

As reviewed by Dr. Mehta in thissupplement, there have beenseveral articles describing theuse of inhaled nitric oxide

(iNO) in conjunction with high-fre-quency oscillatory ventilation (HFOV).iNO was one of the first agents shown toproduce selective vasodilation of the pul-monary vasculature without affectingsystemic arterial pressures (SAP) in pa-tients with pulmonary hypertension(PH). There are now a number of alter-natives to iNO described in the literature.Many of these alternative agents havebeen shown to have comparable efficacywith iNO in both animal models and inclinical disease states, potentially have abetter safety profile, and are certainlymore competitive in terms of cost. Thesealternatives have become an establishedpart of the treatment plan in a number ofdisease states, but in comparison withiNO, there have been very few studiesdescribing their use with HFOV. All of thealternative agents are formulated as liq-uids, as opposed to being a gas, and re-quire aerosolization to be delivered to thelung. This presents a problem becausethere is a paucity of studies describing

the performance of nebulizers withHFOV.

Aerosolization of Drugs andHFOV

Previous studies have identified vari-ous ventilator, equipment, and patientfactors that influence the efficacy of theadministration of aerosolized drugs tothe lung. In particular, it has been shownthat tidal volumes in adults must be ofsufficient size (�500 mL) to ensure ade-quate aerosol delivery to the lung (1).This is obviously a potential concern withHFOV, with which delivered tidal vol-umes are frequently equivalent to onlythe dead-space volume of the lung.

Studies have demonstrated that me-tered-dose inhalers do not function wellduring HFOV. Garner et al. (2) found that�1% of a dose from a metered-dose in-haler was deposited in a pediatric lungmodel ventilated with HFOV at varyingfrequencies of 5 to 15 Hz, a mean airwaypressure of 28 cm H2O, pressure ampli-tude of 55 cm H2O, and varying inspira-tory times (30–50%). Changing the ven-tilatory frequency or inspiratory time wasfound to have no effect on drug delivery.

In contrast, nebulization of surfactantwas shown to be feasible in a rabbit modelof acute lung injury using a Miniheartlow flow volume nebulizer (Westmed,Tucson, AZ) at a gas flow of 1.5 L/min (3).This nebulizer produces a particle sizethat should permit peripheral distribu-tion of surfactant within the lung (mass

median aerodynamic diameter of 3 �m,with 70% of the particles being between 1and 5 �m). The animals were ventilatedwith high-frequency ventilation at a fre-quency of 8 Hz and a mean airway pres-sure of 12 cm H2O. Although high-frequency ventilation differs from HFOVin that expiration is a passive process asopposed to being active, it is not knownwhat effect this has on nebulizer perfor-mance and the delivery of drug to thelung. With this caveat in mind, theseauthors found that 9.8% of the initialdose of surfactant was deposited in thelung. In an identical animal model butusing conventional ventilation, the au-thors found that 8.4% of a nebulized dosewas delivered to the lung, demonstratingthat the nebulizer performed equally wellwith both conventional and high-fre-quency ventilation. There was a nonuni-form distribution of surfactant within thelung, with the majority of the dose beingpreferentially deposited in the right up-per and lower lobes of the lung. The au-thors noted that under test conditions,jet nebulizers that operated at high gasflows increased mean airway pressure,whereas the Miniheart low-flow jet neb-ulizer did not. They also noted that acti-vation of the nebulizer produced an ini-tial increase in PaCO2 on high-frequencyventilation that returned to normal whenthe nebulizer was turned off. The cause ofthe change in PaCO2 was not ascertained.

Fink et al. (4) compared the perfor-mance of different nebulizers in a pediat-

From the Department of Anesthesiology, Universityof Virginia Health Services Foundation, Charlottesville,VA.


DOI: 10.1097/01.CCM.0000156792.40298.5A

Objective: Inhaled nitric oxide has gained an establishedplace in the management of pulmonary hypertension. However,cost, potential toxicity, and the lack of positive outcome datawith inhaled nitric oxide therapy has generated interest inalternative inhaled, selective pulmonary vasodilators. This ar-ticle describes those alternatives that have been studied todate.

Design: Literature review of inhaled, selective pulmonary va-sodilators other than nitric oxide.

Methods: A review of the molecular mechanisms, potentialside effects, and the studies to date in both animal models and

clinical studies describing the physiologic effects of alternativeagents to inhaled nitric oxide.

Conclusion: There are a number of available agents that have com-parable physiologic effects as inhaled nitric oxide. The best studied ofthese are the inhaled prostanoids (prostacyclin and iloprost), and thereis growing interest in novel therapies such as phosphodiesterase inhib-itors and neuropeptides. (Crit Care Med 2005; 33[Suppl.]:S188–S195)

KEY WORDS: high-frequency oscillatory ventilation; nitric oxide;prostacyclin; iloprost; phosphodiesterase inhibitors; ad-renomedullin; vasoactive intestinal peptide; pulmonary hyperten-sion; acute respiratory distress syndrome


ric test lung ventilated with HFOV at afrequency of 8 Hz, mean airway pressureof 20 cm H2O, pressure amplitude of 25cm H2O, and an inspiratory time of 33%.Conventional jet nebulizers delivered ap-proximately 8% of the initial drug dose tothe lung and altered mean airway pres-sure during HFOV. The Aeroneb nebu-lizer (Aeroneb Pro, Aerogen, Sunnyvale,CA), which uses a vibrational element toproduce droplets without the require-ment for a driving gas flow, delivered23% of the initial drug dose and did notalter ventilation variables.

The limited studies that have beenperformed to date suggest that delivery ofaerosolized drug is possible with HFOVand that nebulizer performance is com-parable with that achieved during con-ventional ventilation. Some nebulizersseem to be better suited to HFOV thanothers. Nebulizers requiring low drivingflows will alter ventilation variables suchas mean airway pressure less than thoserequiring high flows. Furthermore, de-vices using novel methods to generateaerosolized particles may find a particularniche during HFOV (5). Further studiesare very much needed to examine theissue of the delivery of aerosolized drugsand HFOV.

Alternatives to iNO

iNO has found a place in the manage-ment of a number of disease states, in-cluding persistent PH and hypoxemia innewborns, acute respiratory distress syn-drome (ARDS), acute PH (particularly af-ter cardiac surgery), and sickle cell dis-ease. Although iNO produces selectivepulmonary vasodilation in these condi-tions, outcome studies have shown thatiNO does not decrease mortality in ARDSand produces only a temporary improve-ment in ventilator variables. These disap-pointing findings may simply reflect thefact that any inhaled pulmonary vasodi-lator does not influence or ameliorate theunderlying disease process in ARDS. Al-ternatively, the potential beneficial ef-fects of iNO (decreased pulmonary arterypressures [PAPs] and lung edema, im-proved oxygenation, effects on plateletand leukocyte adhesiveness in the pulmo-nary capillaries) are offset by its knowntoxic effects (free-radical formation, pro-duction of nitrogen dioxide). Many of theproposed alternatives to iNO share itsbeneficial properties but lack its toxicside effects. These agents therefore have

the untested potential to improve out-come in ARDS.

Inhaled Prostaglandins. The prosta-glandins, in particular prostacyclin(PGI2) and its longer acting analog, ilo-prost, have received the most interest inthe literature as possible substitutes foriNO. Ironically, the clinical effect of in-haled PGI2 was investigated decades be-fore the publication by Rossaint et al. (6)in 1993 describing the effect of iNO inpatients with ARDS. The flood of interestand publications pertaining to iNO thatfollowed this publication, effectively sti-fled further investigations into inhaledPGI2 or any other inhaled agents. Interestin alternatives to iNO has, however, beenrenewed as a result of both the failure todemonstrate that iNO improves outcomein ARDS and, particularly in the UnitedStates, the cost of iNO that has beenlevied by its manufactures (INO Thera-peutics, Clinton, NJ).

PGI2 is derived from arachidonic acidand is synthesized mainly in endothelialcells, particularly those of the pulmonarycirculation. PGI2 has a half-life of only3–6 mins and undergoes spontaneous hy-drolysis to its inactive metabolite, 6-keto-prostaglandin-F1�. It causes vascularsmooth muscle relaxation and is the mostpotent known inhibitor of platelet aggre-gation. These properties suggest thatPGI2 has a role in maintaining blood flowand preventing clot formation in bloodvessels (7). Prostaglandins (PG) exerttheir biological effects via activation ofprostanoid receptors. The PG receptorshave been classified into IP, DP, EP, FP,and TP receptors according to their en-dogenous ligands, respectively, PGI2,PGD2, PGE2, PGF2�, and TxA2. The acti-vation of some receptors (TP, EP1, EP3,FP) induces vasoconstriction, whereasothers (IP) induce vasodilation. Activa-tion of the IP receptor by PGI2 producesactivation of the enzyme adenylate cy-clase, increased intracellular cyclic aden-osine monophosphate levels, and openingof calcium-activated potassium channels(8). Increased potassium conductanceproduces hyperpolarization of the cellmembrane, blockade of L-type calciumchannels, and decreased cytosolic cal-cium. The net result of this process in thevascular smooth muscle cell is relaxationwith consequent vasodilation. Stimula-tion of the IP receptor also results inactivation of protein kinase A. Proteinkinase A is linked to the cellular growthinhibitory effects of both PGI2 and ilo-prost, which includes down-regulation of

cell cycle regulatory proteins, inhibitionof extracellular matrix protein produc-tion by both fibroblasts and endothelialcells (9), and inhibition of mitogen-induced proliferation of vascular smoothmuscle cells (10). These growth inhibi-tory effects are thought to play an impor-tant role in modulating disease processessuch as primary PH that is associatedwith cellular hyperplasia within the vas-cular wall, causing mechanical obstruc-tion of the pulmonary vasculature. Al-though these cellular processes areparticularly apparent in chronic forms ofPH, they may well be operant in acuteforms of PH such as that associated withARDS. Inhibition of platelet aggregation,a recognized property of PGI2, may alsoplay a role in maintaining the patency ofthe pulmonary capillaries. Both PGI2 andiloprost possess anti-inflammatory prop-erties. Iloprost-induced increased cellularcyclic adenosine monophosphate levelsinhibit neutrophil adherence to endothe-lial monolayers and prevented lung dam-age in an animal model of neutrophil-induced lung injury (11). Iloprost hasalso been shown to down-regulate in-flammatory mediator release from neu-trophils (12).

There have been a number of studiesinvestigating the use of PGI2 in ARDS.PGI2 may be beneficial in ARDS fromseveral standpoints. PGI2-induced im-provements in intrapulmonary shuntingand, thereby, oxygenation may poten-tially permit a reduction in inspired oxy-gen levels or airway pressures to “safe”levels. Acute PH is a common associationwith ARDS and may be of sufficient se-verity to induce right ventricular (RV)failure. There is a correlation between theseverity of PH and the associated RV fail-ure and both the severity of lung injuryand mortality (13). PH and RV failuremay cause inadequate left ventricle fill-ing, decreased cardiac output, and sys-temic hypotension. Decreased SAPs maycompromise RV coronary perfusion at atime when RV end-diastolic pressures andRV myocardial oxygen consumption areincreased secondary to increased RV walltension. Decreased oxygen delivery to theRV will exacerbate the RV failure, causinga further decrease in cardiac output andSAP. A vicious cycle may ensue unlessPAPs can be decreased, permitting in-creased RV ejection (Fig. 1). All systemicvasodilators will decrease both systemicand pulmonary arterial pressures. PH isameliorated but at the expense of wors-ened systemic hypotension. Inhaled se-


lective pulmonary vasodilators such asiNO and PGI2 decrease PAPs without af-fecting SAPs and have been a major ad-vance in the management of hemody-namically significant acute PH.

Several studies demonstrate that in-haled PGI2 is effective in the managementof acute PH and produces comparable de-creases in PAP as iNO. Mikhail et al. (14)compared increasing doses of inhaled PGI2

(15–50 ng·kg�1·min�1) with iNO (10–100ppm) in 12 patients with chronic PH. Bothagents produced selective decreases in PAP,with inhaled PGI2 producing a greater de-crease in pulmonary vascular resistance(PVR) than iNO (38% vs. 12%). Haraldssonet al. (15) compared the short-term re-sponse to iNO (40 ppm) and inhaled PGI2

(20–30 �g bolus dose) in ten patients withPH awaiting heart transplantation. iNO andinhaled PGI2 decreased the PVR to a similarextent without effecting SAP. Interestingly,an 11% increase in cardiac output was ob-served with inhaled PGI2 therapy but notiNO. Both agents increased the pulmonaryartery occlusion pressure that, in somecases, was associated with overt pulmonaryedema. A similar increase in pulmonaryartery occlusion pressure has been notedpreviously when iNO was administered topatients with severely decreased left ventri-cle ejection fraction (16) (see below). In-haled PGI2 has also been used for the man-agement of the PH and hypoxia associatedwith acute pulmonary embolism. Webb etal. (17) administered a 200-ng·kg�1·min�1

dose of inhaled PGI2 to a cyanotic, shockedpatient with acute on chronic pulmonaryembolism. The mean PAP decreased from59 to 53 mm Hg with no effect on SAP,

even at this high a dose, and increased thePaO2/FIO2 ratio from 66 to 225 mm Hg.Inhaled PGI2 has also been used success-fully in the management of portopulmo-nary hypertension (18) and in patients withPH after lung transplantation (19). In thelatter study (19), 10 ng·kg�1·min�1 inhaledPGI2 produced an 11% decrease in meanPAP and a 25% decrease in intrapulmonaryshunt.

Studies investigating the effects of in-travenous PGI2 to treat the PH associatedwith ARDS demonstrated that the bene-ficial effect of decreased PAPs was offsetby the propensity of the intravenous for-mulation to cause systemic hypotension(above) and increase pulmonary venousadmixture. In contrast, inhaled PGI2 hasbeen repeatedly shown to produce selec-tive vasodilation of the pulmonary vascu-lature and potentially improve oxygen-ation. In 1993, Walmrath et al. (20)administered inhaled PGI2 (17–50ng·kg�1·min�1) to three patients withARDS. Inhaled PGI2 decreased meanPAPs (from 40 to 32 mm Hg), decreasedPVR by 30%, and increased the PaO2/FIO2

ratio from 120 to 173 mm Hg. Van Heer-den et al. (21) studied the response to a10- to 50-ng·kg�1·min�1 dose of inhaledPGI2 in nine patients with ARDS. Oxygen-ation significantly improved in responseto the 10-ng·kg�1·min�1 dose. InhaledPGI2 had no effect on PAPs in this study;however, none of the patients had ele-vated PAPs. In common with iNO, in-haled PGI2 has no apparent pulmonaryvasodilating effect in the absence of PH(22). Inhaled PGI2 has been comparedwith iNO in patients with ARDS.

Walmrath et al. (23) compared the lowestdose of either iNO or inhaled PGI2 thatproduced the maximum increase in PaO2.This corresponded to a dose of 2–40 ppmof iNO and 1.5–34 ng·kg�1·min�1 inhaledPGI2. Although both agents produced a7% decrease in intrapulmonary shuntingand a comparable increase in PaO2, in-haled PGI2 produced a greater decrease inPVR than iNO. Some of the patients wereadministered either drug for 48 hrs orlonger, demonstrating that the beneficialeffects were sustained. Zwissler et al. (24)compared three doses of iNO (1, 4, and 8ppm) with three doses of inhaled PGI2 (1,10, and 25 ng·kg�1·min�1) in eight pa-tients with ARDS. There was a significantdecrease in PVR in response to both 10-and 25-ng·kg�1·min�1 doses of inhaledPGI2 but not to any of the doses of iNO(Fig. 2). The PaO2 increased in responseto the 10- and 25-ng·kg�1·min�1 doses ofinhaled PGI2 and all three doses of iNO(Fig. 3). A dose of 8 ppm of iNO produceda greater increase in PaO2 (�45%) thanthe 25-ng·kg�1·min�1 dose of inhaledPGI2 (�25%). This study suggests thatinhaled PGI2 may be more effective thaniNO at decreasing PVR, whereas iNO maybe more effective at improving oxygen-ation. This is supported by comparativestudies of the effects of the PGI2 analogiloprost vs. iNO in patients with primaryPH (see below).

Some studies have suggested that theeffect of inhaled PGI2 in ARDS may differaccording to the cause of the lung injury(25). Overall, 53% of patients respondedto inhaled PGI2 with a �10% increase inthe PaO2/FIO2 ratio. However, the major-ity of “responders” were patients in whomthe ARDS was caused by extrapulmonarydisease (sepsis, major trauma). Therewere few responders in patients whoseARDS was caused by a primary pulmo-nary injury (pneumonia, aspiration). Asignificant inhaled PGI2-induced de-crease in PAP was also only observed inthe former group of patients. In contrast,Walmrath et al. (26) demonstrated thatpatients with pneumonia had a good re-sponse to inhaled PGI2, provided thatlung fibrosis was absent. Interestingly,the overall response to inhaled PGI2 inthe study by Domenighetti et al. (25) wasremarkably similar to the percentage ofpatients with ARDS who respond to iNO(27).

A number of agents have been shownto enhance the effect of inhaled PGI2.Inhaled PGI2 and iNO exert their cellulareffects via adenylate cyclase, producing

Figure 1. Vicious cycle of right ventricle (RV) failure, decreased cardiac output (CO), systemichypotension, decreased coronary perfusion, and worsening right ventricular failure. PAP, pulmonaryartery pressure; RVEDP, right ventricle end-diastolic pressure; RVEDV, right ventricle end-diastolicvolume; LVEDV, left ventricle end-diastolic volume; CO, cardiac output.


cyclic adenosine-3,'5' monophosphate,and guanylate cyclase, producing cyclicguanosine monophosphate, respectively(Fig. 4). The net effect of activation ofeither of these pathways in vascularsmooth muscle is relaxation, raising thepossibility of an additive effect if the twoagents are given together. This has beendemonstrated in both a rodent model ofchronic PH (28) and in a clinical study ofpatients after lung transplantation (29).Cyclic adenosine monophosphate is me-tabolized by the enzyme phosphodiester-ase (PDE), predominantly the type-3form. PDE inhibitors (PDEIs) can there-fore increase the intracellular concentra-tions of cyclic adenosine monophosphateand augment the biological effects ofPGI2. The pulmonary vasodilating re-

sponse to inhaled PGI2 was augmentedand significantly prolonged in an animalmodel of PH by various types of PDEIstested but, particularly, the combinedtype 3 and 4 inhibitors. This finding wasrecently confirmed in a clinical study.Haraldsson et al. (30) demonstrated thatinhaled milrinone, a type-3 PDEI, aug-mented the pulmonary vasodilating ef-fects of inhaled PGI2.

The optimum dose of inhaled PGI2 invarious clinical settings has yet to be estab-lished. In patients with ARDS, Van Heerdenet al. (21) found a fairly flat dose-responsecurve for the observed increase in PaO2/FIO2

ratio between 10 and 50 ng·kg�1·min�1.Although there was a trend for oxygenationto further improve over the range of dosestested, this was not significant. Similarly,

Mikhail et al. (14) found a significantdecrease in PAP (above) in response to a15-ng·kg�1·min�1 dose of inhaled PGI2

with no further decrease in PAP up to a50-ng·kg�1·min�1 dose. At our institution,we start therapy with 50 ng·kg�1·min�1

inhaled PGI2 and titrate the dose down ac-cording to the patient’s response. We haveseen a further decrease in PAP in re-sponse to a 100-ng·kg�1·min�1 dose inthe rare patient but usually do not ex-ceed 50 ng·kg�1·min�1.

There are a number of possible sideeffects associated with the use of inhaledPGI2. Systemic hypotension is a theoret-ical possibility, particularly as systemicabsorption of PGI2 from the lungs canoccur, as shown by the detection of itsmetabolites in the systemic circulation insome (21) but not all (31) studies. How-ever, inhaled PGI2 therapy is not associ-ated with the systemic effects normallyassociated with systemic use of PGI2,namely, facial flushing, headache, jawpain, and diarrhea. Furthermore, sys-temic hypotension has not been noted inany of the clinical trials of inhaled PGI2

described above. One study (32) did re-port a decrease in diastolic arterial pres-sures in women (but not men), after abolus dose of 250–500 �g; however, thisdose is far greater than that given in anyother published clinical study. Inhibitionof platelet aggregation is a recognizedproperty (side effect) of PGI2. Haraldssonet al. (31) studied platelet aggregationand clinical bleeding in patients after car-diac surgery. Inhaled PGI2 (either 30 or62 ng·kg�1·min�1) was given for 6 hrs inthe postoperative period. Impaired invitro platelet aggregation, beyond thatexpected after cardiopulmonary bypass,was noted after 2 hrs of inhaled PGI2,with a trend for further impairment at 6hrs. Despite in vitro evidence of plateletdysfunction, there was no difference withrespect to bleeding times, chest tubedrainage, or transfusion requirementsbetween those patients who received in-haled PGI2 and controls. There are con-flicting reports concerning the effect ofPGI2 on bronchial tone, with some stud-ies showing bronchodilation and othersbronchoconstriction (7). Bronchospasmhas not been reported in any of the nu-merous clinical trials cited above. How-ever, we have experienced one possiblecase of bronchospasm in a child withcongenital heart disease that necessitatedstopping inhaled PGI2 therapy. The effectof long-term PGI2 on pulmonary functionis not known. PGI2 is supplied as a pow-

Figure 3. Effect of three doses of inhaled prostacyclin (PGI2) and inhaled nitric oxide (NO) on arterialoxygen tensions (PaO2; closed circles) and intrapulmonary shunt (QsQt; open circles). All data aremean � SEM. *p � .05 vs. control value before administration of inhaled PGI2; #p � .05 vs. controlvalue before administration of inhaled NO. Reprinted with permission from Zwissler et al (24).

Figure 2. Effects of three doses of inhaled prostacyclin (PGI2) and inhaled nitric oxide (NO) on meanpulmonary artery pressure (PAP; closed circles) and pulmonary vascular resistance (PVR; open circles).All data are mean � SEM. *p � .05 vs. control data before administration of inhaled PGI2; #p � .05 vs.control data before administration of inhaled NO. Reprinted with permission from Zwissler et al (24).


der that must be dissolved in a glycinebuffer (supplied by the manufacturers) toproduce a final solution with a pH of 10.5.There is one report of PGI2 causing amild tracheitis when administered intra-tracheally to piglets (33); however, ani-mals were given nine times the normaldose of diluent that would be adminis-tered to an adult human. A separate studywas unable to detect any evidence of pul-monary toxicity after 8 hrs of inhaledPGI2 in healthy lambs (34). We have ad-ministered inhaled PGI2 for 22 days toone patient after a lung transplantationwithout any overt evidence of pulmonarytoxicity. As with iNO, abrupt withdrawalof inhaled PGI2 may be associated withrebound PH (35). Although this phenom-enon is rare (we have not observed it in 3yrs of using inhaled PGI2 therapy in both

adult and pediatric patients), it cautionsthat therapy should be weaned gradually.Unfortunately, there are no definitiveguidelines as to how gradual this shouldbe. Overall, PGI2 and its metabolites areremarkably nontoxic, and our experienceand that of others (36) have found few ifany adverse effects.

This solution must be protected fromlight and is stable for 12 hrs at roomtemperature and for 48 hrs if refriger-ated. At the University of Virginia, PGI2 isprepared at the designated dilution in thehospital pharmacy and delivered in60-mL syringes. A syringe pump delivers8 mL/hr of solution (to replace the vol-ume nebulized) to a low-flow jet nebu-lizer positioned toward the end of theinspiratory circuit. Both the syringepump and the nebulizer are wrapped in

silver foil to protect the solution fromlight. A spacer is placed between the neb-ulizer and the Y-piece. More sophisti-cated solutions have also been described(13). In contrast to iNO, the lack of tox-icity of PGI2 or its metabolites permitsinhaled PGI2 to be administered via asimple, readily available and inexpensivedelivery system, without the need for themeasurements of inhaled drug levels orthe exhaled levels of its metabolites. Thissystem works well in clinical practice us-ing conventional ventilators but, as de-scribed above, has not definitively beenshown to be feasible with HFOV. There isa major difference in cost between iNOand inhaled PGI2. Whereas the cost ofiNO is $3,000 per day (in the UnitedStates), the daily cost of PGI2 in an adultpatient is $150. In contrast to iNO, PGI2

is not currently approved by the U.S.Food and Drug Administration for use bythe inhaled route in the United States.

Although improved oxygenation anddecreased PH might be expected to ben-efit patients with ARDS, it should be re-membered that iNO therapy also pro-duces these effects but has not beenshown to improve outcomes. As sug-gested above, this may be due to the ben-eficial effects of iNO being offset by itsknown toxicity (37). PGI2 and its metab-olites seem to be remarkably free fromtoxic side effects. Furthermore, as de-scribed above, PGI2 and its analogs havepotent effects at the cellular level to in-hibit coagulation, cellular proliferation,and inflammation. This does at least raisethe possibility that PGI2 might influencethe pathology underlying ARDS in a pos-itive manner.

Inhaled Iloprost. Iloprost is the stablecarbacyclin analog of PGI2. In contrast toPGI2, iloprost is stable at room tempera-ture, physiologic pH, and normal lightconditions. Whereas PGI2 has a half-lifeof only 3 mins, iloprost has a half-life of20–30 mins and exerts its pulmonary va-sodilating effects for 30–90 mins. Clinicalstudies have shown that inhaled iloprosthas comparable pulmonary hemody-namic effects as iNO and inhaled PGI2. Ithas been shown to be a selective pulmo-nary vasodilator in patients with primaryPH, producing a 30% decrease in PVR,which was greater than that produced by40 ppm of iNO (38). Leuchte et al. (39)also demonstrated that inhaled iloprost(15–20 �g) was significantly more effec-tive at lowering PVR than 40 ppm of iNOor 50–100 mg of oral sildenafil (see be-low). Although most clinical studies to

Figure 4. Physiologic effect of nitric oxide (NO) and prostacyclin (PGI2) production by the endothelialcell (EC) on the vascular smooth muscle cell (SMC). Cyclic guanosine monophosphate (cGMP)activates protein kinase G (PKG) and, to a lesser extent, protein kinase A (PKA) and is metabolizedpredominantly by phosphodiesterase (PDE) type-5. Cyclic adenosine monophosphate (cAMP) activatesPKA and, to a lesser extent, PKG and is metabolized predominantly by PDE type-3. Both protein kinaseA and G act via the sarcolemmal potassium (K�) and calcium (Ca2�) pumps to decrease cytosoliccalcium and thereby induce smooth muscle relaxation. R, receptor; NOS, nitric oxide synthase; PCS,prostacyclin synthase; SNP, sodium nitroprusside; NTG, nitroglycerin; GC, guanylate cyclase; AC,adenylate cyclase; GTP, guanosine triphosphate; ATP, adenosine triphosphate.


date have focused on patients withchronic PH (9), there is a growing num-ber of reports of its use in acute-onsetPH. Inhaled iloprost has been shown toproduce selective pulmonary vasodilationin both adult (40, 41) and pediatric (42)patients with PH undergoing cardiac sur-gery. Whereas the dose of inhaled iloprostadministered to patients with chronic PHis typically 100 �g, the dose of iloprostgiven acutely to adults was 12 �g (41),and the dose in the pediatric case reportwas 2.5 �g administered over 20 mins(42). Although intermittent dosing is fea-sible in the patient with chronic PH,acute fluctuations in pulmonary hemody-namics may not be desirable in patientswith severe acute PH. Therefore, iloprostmay not have any major advantages overPGI2 in the management of acute PH.Iloprost is presently not available in theUnited States but has been approved inboth New Zealand and Germany.

PDEIs. As described above, PGI2 is me-tabolized predominantly by type-3 PDE,whereas nitric oxide is metabolized pre-dominantly by type-5 PDE (Fig. 2).Type-5 PDE is expressed in relatively highamounts in the pulmonary vasculature.Animal models suggest that type-5 PDEIssuch as sildenafil (Viagra, Pfizer) aremore effective pulmonary vasodilatorsthat the type-3 PDEIs such as milrinone(43). Inhaled type-5 PDEIs (sildenafil, za-prinast) not only augmented the hemo-dynamic effect of iNO but produced se-lective pulmonary vasodilation whengiven alone in animal models of PH (44,45). Haraldsson et al. (30) recently dem-onstrated that inhaled milrinone, atype-3 PDEI, also produced selective pul-monary vasodilation in cardiac surgicalpatients with postoperative PH. Whereasinhalation of aerosolized PDEIs has beenshown to be a practical option, recentinterest has focused on the effects of theoral formulation. Oral sildenafil enhancesthe pulmonary vasodilator effects of iNO(46) and, interestingly, also enhances theeffect of inhaled iloprost (47). Intrave-nous PDEI has been found to augmentand prolong the effect of inhaled PGI2,without decreasing SAP, in an animalmodel of PH (48). Sildenafil has beenused effectively in patients with chronicPH who were deteriorating despite treat-ment with inhaled iloprost (49). The lat-ter phenomenon can best be explained bythe fact the various PDE types are notcompletely specific for a particular li-gand. Oral sildenafil has also been shownto ameliorate the rebound PH associated

with iNO withdrawal (50). Mickelakis etal. (51) demonstrated that oral sildenafilproduces selective pulmonary vasodila-tion in patients with primary PH withcomparable efficacy to iNO. Not all stud-ies, however, have confirmed this pulmo-nary selectivity. Intravenous sildenafil(0.35 mg/kg) significantly decreased PVRand enhanced the effects of iNO in pedi-atric patients after cardiac surgery butalso decreased SAP and caused a worsen-ing of oxygenation from 138 to 108 mmHg (52). A decrease in both PAP and SAPin response to systemically administeredsildenafil has also been reported in ani-mal models of PH and in patients withchronic PH (53). The dose of oral silde-nafil used in most studies ranges from 25to 100 mg. The finding that effective andselective pulmonary vasodilation may beachieved by an orally administered agentis very promising for long-term therapyin chronic forms of PH and deserves fur-ther study in acute forms of PH. PDEIsmay also be used in the acute setting toaugment the pulmonary vasodilation pro-duced by either iNO or inhaled PGI2 andiloprost.

Nitric Oxide Donors. Both sodium ni-troprusside (SNP) and nitroglycerin pro-duce vasodilation via enzymatic release ofnitric oxide. When given intravenously,these agents decrease both pulmonaryand systemic pressures and have the po-tential to increase intrapulmonary shuntby inhibition of hypoxic pulmonary vaso-constriction. However, when aerosolizedand inhaled, these agents can both pro-duce selective pulmonary vasodilationand improvements in oxygenation. In anacute lung injury model in newborn pig-lets, inhaled SNP produced a significantdecrease in PAP (from 32 to 17 mm Hg)and an increase in arterial oxygen tensionfrom 71 to 130 mm Hg, with no evidenceof systemic hypotension (54). In animalmodels, the pulmonary selectivity of SNPseems to be dose dependent, with loss ofselectivity at higher doses (54, 55). Thereis also one study that suggests that theresponse to inhaled SNP may differ ac-cording to the cause of the PH. InhaledSNP selectively decreased PAP in a por-cine model of hypoxia-induced PH buthad much less of an effect when the PHwas induced by group B streptococci (56).

There is one clinical report of the ad-ministration of inhaled SNP in ten new-borns with severe hypoxia (57). InhaledSNP, at a concentration of 0.25 mg/mL indistilled water, produced a significant in-crease in the mean PaO2/FIO2 ratio from

32 to 94 mm Hg. The effect was sustainedfor the treatment duration, which rangedfrom 97 to 157 hrs.

Inhaled nitroglycerin has been shownto produce selective pulmonary vasodila-tion in a canine model of thromboxane-induced PH (58); however, other studiessuggest that it may be less potent in thisregard than other agents such as SNP(59). Inhaled nitroglycerin has recentlybeen shown to produce selective pulmo-nary vasodilation in patients with PH af-ter cardiac surgery (60). The authors ad-ministered a dose of 2.5 �g·kg�1·min�1

inhaled nitroglycerin that produced a sig-nificant decrease in PAP, PVR, and in-trapulmonary shunting without effectingSAPs.

Lung Neuropeptides. There are anumber of neuropeptides found in thelung that influence vascular smoothmuscle tone, RV function, endothelialfunction, collagen and elastin deposition,and fluid balance (61). Pulmonary vascu-lar tone is regulated by the balance ofpressor peptides (endothelin-1, angioten-sin-II, substance P) vs. dilating peptides(calcitonin gene-related peptide, ad-renomedullin, atrial natriuretic peptide,vasoactive intestinal peptide, endothelin-3), and this balance seems to be disturbedin disease states associated with PH.Drugs that antagonize the biological ef-fects of pressor peptides, such as endo-thelin-1, have been extensively investi-gated in both animal models and clinicalstudies of chronic PH. Clinical data sug-gest that bosentan, an endothelin antag-onist, produces a functional improve-ment in patients with chronic PH (42).

Recent studies suggest that there is adeficiency of vasoactive intestinal peptidein patients with primary PH, and inhala-tion of vasoactive intestinal peptide pro-duces a significant decrease in PAP (62).In contrast, plasma adrenomedullin lev-els seem to increase in proportion to theseverity of PH. Inhalation of ad-renomedullin produced selective pulmo-nary vasodilation in 11 patients with pri-mary PH (63).

Combination Therapy. This reviewhas described several agents that can de-crease PAP in patients with PH, many ofwhich act through different cellularmechanisms. This raises the potential ofpossible additive effects when theseagents are used in combination. Suchadditive effects have already been demon-strated for combinations of iNO and in-haled PGI2, combinations of PDEIs andeither iNO or inhaled prostanoids


(above), and combinations of bosentanand inhaled iloprost (64). Combinationtherapy is being actively investigated inthe management of chronic PH (65), andthere are clearly a number of combina-tions that deserve further study for bothchronic and acute forms of PH.

Conclusion

As we understand more about the lungand how the pulmonary circulation isregulated, there seems to be an ever in-creasing number of agents that have thepotential for clinical utility in the man-agement of both PH and ARDS. It is prob-able that no one agent used alone will betotally effective for the treatment of alltypes of PH, and severe cases will requirecombination therapy, as suggested above.Although the ideal would be to find anoral agent to reliably produce selectivepulmonary vasodilation, this ideal has yetto be achieved, and inhalational therapyseems to be the next best option. Inhala-tional therapy remains a largely unknownissue in patients with ARDS who are me-chanically ventilated with HFOV, as therehas been very little research investigatingthe practicality of this approach. It ishoped that the next few years will pro-duce carefully designed clinical trials in-vestigating the potential of these agentsto improve outcomes from ARDS.

REFERENCES

1. Duarte A: Inhaled bronchodilator adminis-tration during mechanical ventilation. Re-spir Care 2004; 49:623–634

2. Garner S, Wiest D, Bradley W: Albuterol de-livery by metered-dose inhaler in a pediatrichigh-frequency oscillatory ventilation model.Crit Care Med 2000; 28:2086–2089

3. Dijk P, Heikamp A, Oetomo S: Surfactantnebulization versus instillation during highfrequency ventilation in surfactant-deficientrabbits. Ped Res 1998; 44:699–704

4. Fink J, Barraza P, Bisgaard J, et al: Aerosoldelivery during mechanical ventilation withhigh frequency oscillation: An in vitro eval-uation. Abstr. Chest 2001;120:277S

5. Dhand R: New frontiers in aerosol deliveryduring mechanical ventilation. Respir Care2004; 49:666–677

6. Rossaint R, Falke K, Lopez F, et al: Inhalednitric oxide for the adult respiratory distresssyndrome. N Engl J Med 1993; 328:399–405

7. Lowson S: Inhaled alternatives to nitric ox-ide. Anesthesiology 2002; 96:1504–1513

8. Schubert R, Serebryakov V, Engel H, et al:Iloprost activates KCa channels of vascularcell muscle cells: Role of cAMP-dependentprotein kinase. Am J Physiol 1996; 271:C1203–C1211

9. Olschewski H, Rose F, Schermuly R, et al:Prostacyclin and its analogues in the treat-ment of pulmonary hypertension. Pharma-col Ther 2004; 102:139–153

10. Phillips P, Long L, Wilkins M, et al: cAMPphosphodiesterase inhibitors potentiate ef-fects of prostacyclin analogs in hypoxic pul-monary vascular remodeling. Am J PhysiolLung Cell Mol Physiol 2005; 288:L103–L115

11. Riva C, Morgonroth M, Ljungman A, et al:Inhaled iloprost inhibits neutrophil-inducedlung injury and neutrophil adherence to en-dothelial monolayers. Am J Respir Cell MolBiol 1990; 3:301–309

12. Rose F, Hattar K, Gakish S, et al: Increasedneutrophil mediator release in patients withpulmonary hypertension: Suppression by in-haled iloprost. Thromb Haemost 2003; 90:1141–1149

13. Siobal M: Aerosolized prostacyclins. RespirCare 2004; 49:640–652

14. Mikhail G, Gibbs S, Richardson G, et al: Anevaluation of nebulized prostacyclin in pa-tients with primary and secondary pulmo-nary hypertension. Eur Heart J 1997; 18:1499–1504

15. Haraldsson A, Kieler-Jensen N, Nathorst-Westfelt U, et al: Comparison of inhaled ni-tric oxide and inhaled aerosolized prostacy-clin in the evaluation of heart transplantcandidates with elevated pulmonary vascularresistance. Chest 1998; 114:780–786

16. Semigram M, Cockrill B, Kacmarek R, et al:Hemodynamic effects of inhaled nitric oxidein heart failure. J Am Coll Cardiol 1994;24:982–988

17. Webb S, Stott S, Van Heerden P: The use ofinhaled aerosolized prostacyclin (IAP) in thetreatment of pulmonary hypertension sec-ondary to pulmonary embolism. IntensiveCare Med 1996; 22:353–355

18. Schroeder R, Rafii A, Plotkin J, et al: Use ofaerosolized inhaled epoprostenol in thetreatment of portopulmonary hypertension.Transplantation 2000; 70:548–550

19. Della Rocca G, Coccia C, Costa M, et al:Inhaled aerosolized prostacyclin and pulmo-nary hypertension during anesthesia for lungtransplantation. Transplant Proc 2001; 33:1634–1636

20. Walmrath D, Schneider T, Pilch J, et al: Aero-solized prostacyclin in adult respiratory dis-tress syndrome. Lancet 1993; 342:961–962

21. Van Heerden P, Barden A, Michalopoulos N,et al: Dose response to inhaled aerosolizedprostacyclin for hypoxemia due to ARDS.Chest 2000; 117:819–827

22. Brett S, Chambers J, Bush A, et al: Pulmo-nary response of normal human subjects toinhaled vasodilator substances. Clin Sci1998; 95:621–627

23. Walmrath D, Schneider T, Pilch J, et al: Di-rect comparison of inhaled nitric oxide andaerosolized prostacyclin in acute respiratorydistress syndrome. Am J Respir Crit CareMed 1996; 153:991–996

24. Zwissler B, Kemming G, Habler O, et al:Inhaled prostacyclin versus inhaled nitric ox-

ide in adult respiratory distress syndrome.Am J Respir Crit Care Med 1996; 154:1671–1677

25. Domenighetti G, Stricker H, Waldispuehl B:Nebulized prostacyclin in acute respiratorydistress syndrome: Impact of primary (pul-monary injury) and secondary (extrapulmo-nary injury) disease on gas exchange re-sponse. Crit Care Med 2001; 29:57–62

26. Walmrath D, Schneider T, Pilch J, et al: Ef-fects of aerosolized prostacyclin in severepneumonia: Impact of fibrosis. Am J RespirCrit Care Med 1995; 151:724–730

27. Dellinger R, Zimmerman J, Taylor W, et al:Effects of inhaled nitric oxide in patientswith acute respiratory distress syndrome: Re-sults of a randomized phase II trial. Crit CareMed 1998; 26:15–23

28. Hill L, Pearl R: Combined inhaled nitric ox-ide and inhaled prostacyclin during experi-mental chronic pulmonary hypertension.J Appl Physiol 1999; 86:1160–1164

29. Della Rocca G, Coccia C, Pompei L, et al:Hemodynamic and oxygenation changes ofcombined therapy with inhaled nitric oxideand inhaled aerosolized prostacyclin. J Car-diothorac Vasc Anesth 2001; 15:224–227

30. Haraldsson A, Kieler-Jensen N, Ricksten S:The additive pulmonary vasodilating effectsof inhaled prostacyclin and inhaled milri-none in postcardiac surgical patients withpulmonary hypertension. Anesth Analg 2001;93:1439–1445

31. Haraldsson A, Kieler-Jensen N, Wadenvik H,et al: Inhaled prostacyclin and platelet func-tion after cardiac surgery and cardiopulmo-nary bypass. Intensive Care Med 2000; 26:188–194

32. Parsargiklian M, Bianco S: Ventilatory andcardiovascular effects of prostacyclin and6-oxo-PGF1� by inhalation. Adv Prostaglan-din Thromboxane Res 1980; 7:943–951

33. Van Heerden P, Caterina P, Filion P, et al:Pulmonary toxicity of inhaled aerosolizedprostacyclin therapy: An observational study.Anaesth Intensive Care 2000; 28:161–166

34. Habler O, Kleen M, Zwissler B, et al: Inhala-tion of prostacyclin for 8 hours does notproduce signs of acute pulmonary toxicity inhealthy lambs. Intensive Care Med 1996; 22:426–433

35. Augoustides J, Culp K, Smith S: Reboundpulmonary hypertension and cardiogenicshock after withdrawal of inhaled prostacy-clin. Anesthesiology 2004; 100:1023–1025

36. De Wet C, Affleck D, Jacobsohn E, et al:Inhaled prostacyclin is safe, effective, andaffordable in patients with pulmonary hyper-tension, right heart dysfunction, and refrac-tory hypoxemia after cardiothoracic surgery.J Thoracic Cardiovasc Surg 2004; 127:1058–1067

37. Weinberger B, Laskin D, Heck D, et al: Thetoxicology of inhaled nitric oxide. Toxicol Sci2001; 59:5–16

38. Hoeper M, Olschewski H, Ghofrani H, et al: Acomparison of the acute hemodynamic ef-fects of inhaled nitric oxide and aerosolized


iloprost in primary pulmonary hypertension.J Am Coll Cardiol 2000; 35:176–182

39. Leuchte H, Schwaiblmair M, Baumgartner R,et al: Hemodynamic response to sildenafil,nitric oxide and iloprost in pulmonary arte-rial hypertension. Chest 2004; 125:580–586

40. Langer F, Wilhelm W, Tscholl D, et al: Intra-operative inhalation of the long-acting pros-tacyclin analog iloprost for pulmonary hy-pertension. J Thorac Cardiovasc Surgery2003; 126:874–875

41. Rex S, Busch T, Vettelschoss M, et al: Intra-operative management of severe pulmonaryhypertension during cardiac surgery with in-haled iloprost. Anesthesiology 2003; 99:745–747

42. Muller M, Scholz S, Kwapisz M, et al: Use ofinhaled iloprost in a case of pulmonary hy-pertension during pediatric congenital heartsurgery. Anesthesiology 2003; 99:743–744

43. Lowson S: Alternatives to nitric oxide. BrMed Bulletin 2004;70:119–131

44. Ichinose F, Adrie C, Hurford W, et al: Selec-tive pulmonary vasodilation induced by aero-solized zaprinast. Anesthesiology 1998; 88:410–416

45. Ichinose F, Erana-Garcia J, Hromi J, et al:Nebulized sildenafil is a selective pulmonaryvasodilator in lambs with acute pulmonaryhypertension. Crit Care Med 2001; 29:1000–1005

46. Atz A, Lefler A, Fairbrother D, et al: Sildenafilaugments the effect of inhaled nitric oxidefor postoperative pulmonary hypertensivecrisis. J Thorac Cardiovasc Surg 2002; 124:628–629

47. Ghofrani H, Wiedermann R, Rose F, et al:Combination therapy with oral sildenafil andinhaled iloprost for severe pulmonary hyper-tension. Ann Intern Med 2002; 136:515–522

48. Schermuly R, Ghofrani H, Weissman N, et al:Low-dose systemic phosphodiesterase inhib-itors amplify the pulmonary vasodilator re-sponse to inhaled prostacyclin in experimen-tal pulmonary hypertension. Am J RespirCrit Care Med 1999; 160:1500–1506

49. Ghofrani A, Rose F, Schermuly R, et al: Oralsildenafil as long term adjuvant to inhalediloprost in severe pulmonary hypertension.J Am Coll Cardiol 2003; 42:158–164

50. Atz A, Wessel D: Sildenafil ameliorates theeffect of nitric oxide withdrawal. Anesthesi-ology 1999; 91:307–310

51. Mickelakis E, Tymchak W, Lien D, et al: Oralsildenafil is an effective and specific pulmo-nary vasodilator in patients with pulmonaryarterial hypertension. Circulation 2002; 105:2398–2403

52. Stocker C, Penny D, Brizzard C, et al: Intra-venous sildenafil and inhaled nitric oxide: Arandomized trial in infants after cardiac sur-gery. Intensive Care Med 2003; 29:1996–2003

53. Ghofrani H, Wiedemann R, Rose F, et al:Sildenafil for treatment of lung fibrosis andpulmonary hypertension: A randomized con-trolled trial. Lancet 2002; 360:895–900

54. Yu X, Saugstad D: Nebulization of sodiumnitroprusside in lung-lavaged newborn pig-lets. Pediatr Res 1999; 45:255–259

55. Adrie C, Ichinose F, Holzmann A, et al: Pul-monary vasodilation by nitric oxide gas andprodrug aerosols in acute pulmonary hyper-tension. J Appl Physiol 1998; 84:435–441

56. Meadow W, Rudinsky B, Bell A, et al: Effectsof nebulized nitroprusside on pulmonary andsystemic hemodynamics during pulmonaryhypertension in piglets. Pediatr Res 1998;44:181–186

57. Palhares D, Figueiredo C, Moura A: Endotra-

cheal inhalatory sodium nitroprusside in se-verely hypoxic newborns. J Perinat Med1998; 26:219–224

58. Gong F, Shiraishi H, Kikuchi Y, et al: Inha-lation of nebulized nitroglycerin in dogs withexperimental pulmonary hypertension in-duced by U46619. Pediatr Int 2000; 42:255–258

59. Schutte H, Grimminger F, Otterbein J, et al:Efficiency of aerosolized nitric oxide donordrugs to achieve sustained pulmonary vaso-dilation. J Pharmacol Exp Ther 1997; 282:985–994

60. Yurtseven N, Karaca P, Kaplan M, et al: Effectof nitroglycerin inhalation on patients withpulmonary hypertension undergoing mitralvalve replacement surgery. Anesthesiology2003; 99:855–858

61. Keith I: The role of endogenous lung neu-ropeptides in regulation of the pulmonarycirculation. Physiol Res 2000; 49:519–537

62. Petkov V, Mosgoeller W, Ziesche R, et al:Vasoactive intestinal peptide as a new drugfor treatment of primary pulmonary hyper-tension. J Clin Invest 2003; 111:1339–1346

63. Nagaya N, Kyotani S, Uematsu M, et al: Ef-fects of adrenomedullin inhalation on hemo-dynamics and exercise capacity in patientswith idiopathic pulmonary arterial hyperten-sion. Circulation 2004; 109:351–356

64. Hoeper M, Taha N, Bekjarova A, et al: Bosen-tan treatment in patients with primary pul-monary hypertension receiving non-paren-teral prostanoids. Eur Respir J 2003; 22:330–334

65. Hoeper M, Dinh-Xuan A: Combination ther-apy for pulmonary arterial hypertension:Still more questions than answers. Eur Re-spir J 2004; 24:339–340


High-frequency oscillatory ventilation in adults: Respiratorytherapy issues

Jason Higgins, BS, RRT; Bob Estetter, RRT; Dean Holland, RRT; Brian Smith, RRT; Stephen Derdak, DO

High-frequency oscillatoryventilation (HFOV) is an al-ternative mode of ventila-tion that may be considered

for acute respiratory distress syndrome(ARDS) in adult patients who are failingconventional ventilation (CV). Observa-tional studies have demonstrated thatHFOV may improve oxygenation whenused as a rescue modality in adult pa-tients with severe ARDS failing CV (1, 2).A recent multicentered, randomized,controlled trial of HFOV (3), comparedwith pressure control ventilation, dem-onstrated that HFOV was safe and effec-tive for adult ARDS. Of particular impor-tance to the respiratory therapist (RT),HFOV does not offer traditional monitor-ing capabilities (e.g., tidal volume, flow-time graphs, flow-volume loops, and soon) used to identify and optimize conven-tional ventilator strategies for changes inpulmonary mechanics. A multidisci-plinary approach is required to optimizethe management of patients on HFOV.RTs caring for adult patients with ARDSshould develop increased knowledge ofmechanical properties intrinsic to HFOV,

an understanding of the underlyingpathophysiology, and advanced patientassessment skills unique to this mode ofventilation.

Clinical expertise coupled with ad-vanced patient assessment skills placeRTs in a key position in the managementof patients on HFOV. The purpose of thischapter is to summarize clinical informa-tion pertinent to RTs caring for adultpatients on HFOV. Active involvement ofa critical care respiratory therapy team isessential to successful implementation ofan adult HFOV program.

IDENTIFYING PATIENTS FORHIGH-FREQUENCYOSCILLATORY VENTILATION

Observational rescue trials suggestthat early initiation of HFOV in patientswith severe ARDS may be important tosuccessful outcomes. Patients transi-tioned to HFOV within 72 hrs may have abetter chance of survival than those pa-tients on CV for �7 days (1, 2). RTs, as anintegral member of the critical care team,are in a frontline position to activelyidentify patients with ARDS who may bepotential candidates for a trial of HFOV.Although the exact severity threshold atwhich to initiate a trial of HFOV remainsunclear, an emerging approach in centersexperienced in treating adults with HFOVmay serve as guidelines (4, 5). HFOV maybe considered for patients with ARDS

when they meet the following severitycriteria:

● FIO2 �.60 and/or SpO2 �88% on CV withpositive end-expiratory pressure �15 cmH2O, or

● Plateau pressures (Pplat) �30 cm H2O,or

● Mean airway pressure (MAP) �24 cmH2O, or

● Airway pressure release ventilationPhigh �35 cm H2O.

Once severity criteria are met, and adecision is made to initiate HFOV, em-phasis should be placed on transitioningto HFOV as soon as feasible (within 12–24hrs).

HIGH-FREQUENCYOSCILLATORY VENTILATION‘TEAM APPROACH’

At Parkland Hospital, the HFOV man-agement team consists of the attendingintensivist, respiratory care team leader,respiratory care area manager, intensivecare unit respiratory therapist, criticalcare nurse, and an identified consultteam member who is on-call 24 hr a day/7days a wk for troubleshooting (Table 1).Once a potential candidate is identified,the HFOV team will discuss possible rea-sons the patient is failing CV, reviewother available options (e.g., prone posi-tioning), timing of transition from CV toHFOV (e.g., will bronchoscopy be per-

From Parkland Health and Hospital System (JH,BE, DH, BS), Respiratory Care Department, Dallas, TX;and Pulmonary/Critical Care Medicine (SD), Wilford HallMedical Center, Lackland AFB, TX.


DOI: 10.1097/01.CCM.0000155922.78943.2D

Objective: To summarize clinical information and assessmenttechniques relevant to respiratory therapists caring for adultpatients on high-frequency oscillatory ventilation (HFOV).

Data Source: Review of observational studies, controlled trials,case reports, institutional experience, and hospital HFOV guide-lines for adult patients.

Data Summary: Respiratory therapists require unique physicalassessment skills and knowledge in managing patients on HFOV.Respiratory therapy procedures relevant to HFOV include settingendotracheal tube cuff leaks, performing lung recruiting maneu-

vers, endotracheal suctioning, and monitoring ventilator param-eters. Respiratory therapists serve as essential team members inthe creation and implementation of written HFOV guidelines (e.g.,algorithms) to optimize patient care.

Conclusion: Respiratory therapy assessment and proceduralskills are essential in providing optimal care to adult patients onHFOV. (Crit Care Med 2005; 33[Suppl.]:S196–S203)

KEY WORDS: high-frequency oscillatory ventilation; respiratorytherapy; acute respiratory distress syndrome; lung recruitmentmaneuvers; endotracheal cuff leak


formed before initiation?), initial HFOVsettings, baseline monitoring measure-ments, and HFOV strategies for a failure-to-respond scenario. An important goal ofthe HFOV team approach is to facilitatecommunication between disciplines andto ensure that parameter adjustments areappropriate. The intensive care unit RT isresponsible for identification andhands-on care of the patient after initia-tion. He or she is the liaison to the HFOVteam and has been trained and demon-strated to be competent in caring for pa-tients on HFOV.

We believe the “team approach” pro-vides our patients with the best care byoptimizing HFOV management and pro-viding ongoing education for those indi-viduals not as familiar with the manage-ment of HFOV.

Patient Assessment andMonitoring Techniques

RTs must have a broad understandingof ARDS pathophysiology, mechanicalproperties of the mode of ventilation be-ing used, and the specific ventilator strat-egy and goals chosen for the patient. Athorough physical assessment is, un-doubtedly, the most important skillneeded to provide quality care to patientson HFOV. RTs receive extensive trainingin auscultation, inspection, palpation,and percussion (6). In addition, they re-ceive thorough training on how to inter-

pret the monitoring indices associatedwith CV, as well as how to troubleshoot aventilator (7). At present, literature re-garding physical assessment and specificmonitoring parameters for adult patientson HFOV is limited. Patients on HFOVmay experience acute changes requiringrapid recognition to provide optimalmanagement.

Identifying Diaphragm Position. Lunginflation during HFOV may be estimatedby monitoring diaphragm position on su-pine portable chest radiographs (CXR).Hyperinflation may be suspected on CXRif the apical to diaphragm distance ex-ceeds 24–25 cm and/or the anterior sixthrib is visible above the diaphragm (8).Because changes in lung inflation may beaccompanied by positional changes of thediaphragm, it is beneficial to identifybaseline diaphragm position at the initi-ation of HFOV. The traditional techniqueof percussion to locate diaphragm posi-tion is difficult as a result of the noisegenerated by the ventilator. To accuratelypercuss the diaphragm, ventilation mayneed to be temporarily interrupted bystopping piston oscillation.

At Parkland Hospital, we have used aDoppler technique to quickly identify theposition of the diaphragm at the bedside.After initiation of HFOV, and simulta-neously with the initial CXR, a pencilDoppler (9.3 mHz, model 915BL; ParksMedical Electronics, Aloha, OR) is used tolocate the diaphragm. Diaphragm posi-

tion is then marked directly on the pa-tient with a surgical marker pen. Thismark is used to correlate the position ofthe diaphragm with its location on theCXR. Once the location of the diaphragmhas been confirmed, we then monitorlung displacement as part of routine pa-tient assessment with every ventilatorcheck. If trended consistently, this tech-nique assists with early recognition ofchanges in lung inflation and mayprompt ordering of a CXR if a significantchange is suspected. Other techniquesthat may be used at the bedside to locatethe position of the diaphragm are dia-phragmatic auscultation or conventionalultrasound visualization.

Auscultation. Breath sounds shouldbe auscultated with every ventilatorcheck and during patient compromise.Recommendations for frequency of HFOVventilator checks by the RT are every 30mins during the first hour after initia-tion, then every hour for 2 hrs, and thenevery 2 hrs (Table 2). Although bilateralauscultation of the chest may not revealadventitious breath sounds, it can assistin the identification of lung inflation, di-aphragm position, and may be helpfulwith detection of other complicationssuch as pneumothorax, atelectasis, endo-tracheal tube (ETT) obstruction, or mu-cus plugging. Unlike the assessment ofbreath sounds during CV, the clinicianmay not be able to identify wheezes,rhonchi, or crackles during HFOV be-cause of the small tidal volumes deliveredand the noise of the oscillator piston. Forthis reason, breath sounds should be aus-cultated whenever HFOV is interruptedfor manual ventilation. During HFOV,the clinician should listen closely to thequality of the percussions delivered bythe ventilator. Breath sounds should beauscultated over all accessible regions ofthe chest and compared with the oppositeside for symmetry. Unilateral decreasedbreath sounds may be detected withpneumothorax, mucous plugging, atelec-tasis, mainstem intubation, and pleuraleffusions. Bilaterally decreased breathsounds may be observed with accidentalextubation, ETT occlusion (partial orcomplete), alveolar collapse, alveolaroverdistension, fluid overload/pulmonaryedema, and obesity.

Visual and Tactile Inspection. It isvery important to visualize and palpatethe chest with each HFOV assessment.Visual inspection of the chest for “chestwiggle” and movement of the abdomenmay assist with identifying changes in

Table 1. High-frequency oscillatory ventilation (HFOV) consultation team

Responsibility Action/circumstance

Intensive care unit respiratorytherapist

Identify patients that meet the indications forimplementation of HFOV

Contact RC team leader when patient is identifiedRecord current settings on conventional ventilation before

implementation of HFOVContact RC team leader when ventilator changes are

indicatedConsult with TL, consultation team members, and physician

staff regarding all setting changesRespiratory care team leader Assure that HFOV is available and properly calibrated

Play an active role in the management of the HFOVContact identified consultation team member when

ventilator changes are indicatedConsult with intensive care unit therapists, consultation

team members, and physician staff regarding all settingchanges

Consultation team members Play an active role in the management of the HFOVConsult with team leaders, therapists, other team members,

and physician staff regarding all setting changesRespond promptly when on-call for consultationProvide pertinent articles, troubleshooting guides, and other

material to interested therapists and physicians

RC, respiratory care; TL, team leader.


lung compliance, lung resistance, and/orairway resistance. If the chest is notbouncing as much as it was 2 hrs previ-ously and there appears to be slightchanges in ventilator parameters, theremay be significant changes occurring inlung mechanics. Serial evaluations of ribspaces may also assist in determininglung inflation. Rib spaces that have in-creased over time may indicate the lunghas been recruited to the point of overd-istension. Obesity or chest wall edemamay make these monitoring techniquesdifficult.

ENDOTRACHEAL TUBES

Proper ETT management is of extremeimportance in all critically ill patients,especially those receiving HFOV. ETTsize, position, and patency have directeffects on gas exchange independent ofalterations in the patients’ underlyinglung pathology. Smaller-diameter ETTs(e.g., �7 mm internal diameter) attenu-ate delivered tidal volume and make ef-fective ventilation of large adults moredifficult.

Endotracheal Tube Position

ETT position should be checked regu-larly and maintained. As a result of highlevels of mean airway pressure (mPaw)during HFOV, migration of the ETT prox-imally in the trachea may occur. The po-sition of the ETT relative to a fixed ana-tomic site (e.g., upper front teeth orgum) should be recorded and monitoredfrequently. Migration of the ETT as littleas 2 to 3 cm can adversely affect theability to ventilate the patient.

Tracheal Suctioning

Gross pulmonary edema, hemorrhage,or foaming into the airway, ETT, and/oroscillator circuit will impede the abilityto oxygenate and ventilate during HFOV.Obvious filling of the ETT tube withedema, blood, or foam must be cleared bytracheal suction (usually combined withvigorous manual ventilation with an at-tached positive end-expiratory pressurevalve) before initiation of HFOV. Simi-larly, excessive secretions or mucus plug-ging in the distal airways or ETT can

adversely affect adequate gas exchangeduring HFOV. Because the mechanism ofinjury associated with ARDS predisposesto alveolar collapse, the RT must beaware that tracheal suctioning (TS) canalso be detrimental by creating negativecarinal pressure, which promotes addi-tional alveolar derecruitment. For thisreason, TS should be performed onlywhen clinically indicated (e.g., visible se-cretions), especially in patients with mar-ginal oxygenation requiring high meanairway pressures (mPaw). To ensure ETTpatency, the inline TS catheter can bepassed (without turning on suction) ev-ery 2 to 4 hrs along with instillation of asmall volume of sterile saline (2–3 mL).We perform tracheal suction on CV justbefore initiation of HFOV and then brieflyclamp the ETT during transition to min-imize alveolar derecruitment. It shouldbe noted that this clamping technique isused in any instances the patient requiresa ventilator disconnect (e.g., bronchos-copy, TS, transporting). If possible, TS isavoided during the first 12 hrs on HFOVto allow alveolar recruitment. After this

Table 2. High-frequency oscillatory ventilator (HFOV) check sheet HFOV checks to be done every 30 mins � 2, every 1 hr � 2, then every 2 hrs or asordered

Date and time 3 (30 Mins) (60 Mins) (2 Hrs) (3 Hrs) (5 Hrs) (7 Hrs) (9 Hrs)

FIO2

mPawPower�PHz% I-timeSystem temperatureBias flowCuff leakSpO2

Heart rateBlood pressureOIDiaphragm monitoredBilateral BSSpon resp rate�s in mPaw�s in �P1 mPaw alarm2 mPaw alarmPressure limit

Consult Team Notification Parameters

● SpO2 drop of �3–5% without recovery● mPaw drift � �5 cm H2O (with spontaneous breathing)

� May need increase in paralytic or sedation● �P drift �5 cm H2O with significant diaphragm change or �P drift �10 cm H2O without significant diaphragm change● ABG parameters

� pH �7.45 or �7.20 or a � of �10� PaCO2 �60 mm Hg or �30 mm Hg or a � of �10 mm Hg� PaO2 �90 mm Hg or �55 mm Hg or a � of �10 mm Hg

Changes made 3


initial timeframe, in-line, closed TS isperformed only when necessary and noton a “routine” scheduled interval. Addi-tional indications for TS include an abruptincrease in proximal oscillatory amplitude(�P) coupled with decreased “chest wig-gle,” unexplained hypercapnia or increas-ing oxygen requirements, and followingprone positioning. If the patient’s respira-tory status continues to deteriorate in thepresence of excessive pulmonary secretionsafter TS, bronchoscopy may be considered.Patients who desaturate (SpO2 drops �5%)after TS should be considered for perfor-mance of a lung recruiting maneuver(LRM; see subsequently).

Endotracheal Tube Obstruction

ETT narrowing caused by inspissatedmucus or blood clot accumulation willresult in increased airway resistance sec-ondary to a decreased lumen size. Thisincrease in airway resistance can result inincreased proximal oscillatory pressureamplitude (�Pproximal) displayed on theventilator and decreased carinal oscilla-tory pressure amplitude (�Pcarinal). Par-tial occlusion of the ETT should be sus-pected if a previously stable PaCO2 is nowincreasing. Identification may also be re-flected by an acute or gradual increase in�Pproximal (�5 cm H2O) coupled with adecrease in chest wiggle, a decrease inbreath sounds bilaterally, and an increas-ing oxygen requirement. Acute occlusionof the ETT from mucous plugging or akinked tube presents with a sudden in-crease in �Pproximal (�10 cm H2O), de-creased chest wiggle, decreased breathsounds bilaterally, and rapid oxygen de-saturation with hypercapnia. It is imper-ative the RT understands that decreasingthe power setting in an attempt to main-tain an ordered �Pproximal will only maskthe underlying problem. Moreover, de-creasing the power may result in furtherreduction of �Pcarinal, which may com-promise the patient more, even thoughthe proximal amplitude appears to be un-changed.

The possibility of complete ETT ob-struction should initially be assessed bypassage of a suction catheter and can bedefinitively diagnosed with emergent fi-beroptic bronchoscopy (FOB). If a suc-tion catheter cannot be passed and man-ual ventilation produces no airmovement, emergent reintubation is re-quired. An attempt may be made to passan ETT exchange catheter or stylet in aneffort to open the occlusion while equip-

ment is readied to reintubate. “Quick-look” bronchoscopy should be consideredbefore initiation of HFOV to ensure ETTand airway, especially if on CV �3 days.Patients who desaturate significantly af-ter bronchoscopy may benefit from LRM(see subsequently) before resumption ofthe desired mPaw setting.

Cuff Leaks

Cuff leaks during HFOV may promotePaCO2 clearance by several mechanismsand may allow for the use of lower �P andhigher Hz (which are conceptually morelung protective) (9–11). A small cuff leak,approximately 5–7 cm H2O, may be triedwhen refractory hypercapnia (pH �7.20)occurs despite maximal �P and lowestHz. Failure of a cuff leak to lower PaCO2

may indicate upper airway edema aroundthe ETT and may respond to placement ofan additional oropharyngeal airway to al-low gas egress (12). Some centers useETT cuff leaks at the initiation of HFOVin all patients. Whether to reserve use ofa deliberate cuff leak for refractory hyper-capnia or to use a leak in all patients tofacilitate use of lower �P and higher Hzstrategies should be investigated. Beforecreating a cuff leak, the mouth and pos-terior pharynx should be suctioned. Thelow mPaw and high mPaw alarm shouldbe reset to avoid triggering by a drop orrise in mPaw with initial setting of thecuff leak. At Parkland Hospital, our ap-proach is to initiate a cuff leak by increasebias flow by 5 L/min, then slowly removeair from the ETT cuff pilot balloon whilemonitoring for a 5- to 7-cm H2O drop inmPaw on the ventilator. Once the appro-priate leak has been applied, the mPawcontrol is readjusted to return the mPawto the original setting. It should be notedthat increasing bias flow after institutionof a cuff leak to achieve a set mPaw mayresult in an elevated mPaw as a result ofa decreased cuff leak. This is of clinicalimportance because the magnitude of thecuff leak may change as a result of tra-cheal edema, secretion accumulation,and body positioning. The mPaw alarmsand mPaw pressure limit should be setappropriately (e.g., bracket desired mPawby 5–7 cm H2O) to protect the patient inthe event of a decreasing cuff leak. Thetechnique used to create a cuff leak atWilford Hall Medical Center is detailed inAppendix 1.

HUMDIFICATION

Humidification is often overlooked asan important aspect during any form ofmechanical ventilation. Because endotra-cheal intubation bypasses the upper air-way, it becomes necessary for inspiredgases to be heated and humidified artifi-cially to mimic normal respiratory phys-iology (13). Complications that may oc-cur as a result of ineffective heat andhumidification are, but not limited to,hypothermia, inspissation of airway se-cretions, destruction of airway epithe-lium, and atelectasis (14). There are cur-rently two forms of delivering heat andhumidification to patients requiring me-chanical ventilation: external active hu-midifiers and passive heat and moistureexchangers (HME). HMEs have not beenadequately studied with HFOV andshould not be used.

During HFOV, the bias flow circuit isconnected directly to an external activehumidifier to provide humidified gas en-tering the inspiratory limb of the circuit.Temperature settings should resemblethose normally used during conventionalmechanical ventilation and should be setto establish the desired gas temperatureat the patient airway temperature port(15). We suggest maintaining tempera-ture settings at 37°C to 39°C. Water lev-els in the chamber should always bemaintained at the appropriate levels toprevent the chamber from becoming dry.This will result in the patient receivingonly heated gas without proper humidi-fication and may result in complications.The HFOV circuit has two temperatureports on the inspiratory limb, one nearthe patient’s airway and one near thepressure limit valve. Temperature shouldalways be monitored as close to the pa-tient’s airway opening as possible. Be-cause ambient air temperatures can affectthe temperature and relative humidity inthe circuit, caution should be exercised ifambient temperatures exceed 84°F (e.g.,burn intensive care units). Also, to pre-vent excessive rainout in the circuit, aheated wire circuit should be used. At ourinstitution, active external humidifiersare checked with each ventilator checkand appropriate documentation is per-formed.

LUNG RECRUITMENTMANEUVERS

Lung recruitment maneuvers are usedto improve oxygenation after derecruit-


ing events (e.g., suction, bronchoscopy,circuit disconnects) or for patients whocontinue to have marginal oxygenationduring HFOV. LRMs should be consid-ered for an acute drop in SpO2 �5% afterinitial HFOV transition, TS, bronchos-copy, or circuit disconnect. When usingan in-line closed suction catheter, perfor-mance of a recruitment maneuver duringTS has been shown to prevent alveolarderecruitment during CV (16). Lung re-cruiting maneuvers are typically per-formed by briefly (40–60 secs) raisingmPaw approximately 10 cm H2O abovethe original set mPaw. Before performingthe maneuver, the mPaw high alarmmust be reset (e.g., to 50 cm H2O) andany ETT–cuff leak removed. The oscilla-tor piston is turned off during the ma-neuver to minimize additional distaltransmission of the �P while the mPaw iselevated. Suggested LRM guidelines forHFOV are detailed in Appendix 1.

PRONE VENTILATION

Prone positioning is becoming a morewidely used therapeutic modality in pa-tients with refractory hypoxemia. Studiesshow an improvement in oxygenation inapproximately 60% of patients with ARDS(17, 18). Case reports have observed im-provements in oxygenation and ventila-tion with the combination of HFOV andprone ventilation (19). When patients re-ceiving HFOV are placed in the proneposition, patient and ventilator assess-ment become extremely important. Care-ful adherence to a detailed proning algo-rithm is essential (20). Like with allpatients receiving mechanical ventila-tion, maintaining a patent airway is ofextreme importance. The actual turningof the patient usually requires brief dis-connect from the ventilator circuit andmanual ventilation with a positive end-expiratory pressure valve. ETT placementmust be confirmed and the circuit shouldbe resecured after prone positioning.HFOV parameters need to be verified anddocumented immediately before and im-mediately after prone positioning. Lungmechanics and ETT leaks can rapidlychange with patient turning, and RTsmust be prepared to recognize thesechanges and address them rapidly. LRMsmay be performed in the prone position,particularly if a circuit disconnect wasrequired and desaturation persists.

TENSION PNEUMOTHORAX

Like with all forms of positive pressureventilation, tension pneumothorax maydevelop as a manifestation of “vo-lutrauma” or secondary to the cystic na-ture of the underlying lung disease. Ad-ditionally, vigilance must be maintainedfor pneumothorax in patients undergoingcentral line placement (e.g., subclavianor internal jugular) and thoracentesis.Studies have demonstrated pneumotho-rax occurrence rates during HFOV simi-lar to those observed with conventionalventilation. However, in contrast to CV,the presence of a pneumothorax can beparticularly difficult to detect in patientson HFOV because no alarms on the ven-tilator will reliably signal that tension isdeveloping (21). Quick assessment of ETTplacement, hemodynamic parameters,tracheal position, visualization of thechest for unilateral hyperinflation, de-creased chest movement, auscultation forbreath sounds, diaphragmatic position,and palpation to identify the presence ofsubcutaneous emphysema assist the RTin detection of pneumothorax. If timepermits and the patient is relatively sta-ble, the diagnosis can be confirmed by a“stat” portable CXR. After placement of achest tube, the RT should anticipate thatadjustments in mPaw and �P:Hz will berequired. The degree of leak (e.g., from abronchopleural fistula) should be quanti-tated on the chest tube suction chamberdevice, and changes in the leak should berecorded in response to changing HFOVsettings and as part of the routine sched-uled ventilator checks. Air leak through abronchopleural fistula can be minimizedby using the highest Hz, the lowest �P,the lowest mPaw, and the shortest in-spiratory time (IT%) allowable to achieveacceptable oxygenation and ventilation(22).

HEMODYNAMICS, ARTERIALPRESSURE TRACINGS, ANDPULSE OXIMETRY

Central vein and pulmonary arterypressure monitoring, peripheral arterialcatheters, and pulse oximetry are formsof real-time monitoring that assess pa-tient hemodynamic stability duringHFOV. Hemodynamic status is extremelyimportant before and after initiation ofHFOV. HFOV maintains alveolar recruit-ment by sustaining an essentially con-stant mPaw (in contrast to the cyclicpressure excursions of conventional vol-

ume-cycled ventilation). This increasesmean intrathoracic pressure, reducescentral venous return (e.g., right heartpreload), and may cause potential adverseeffects on the cardiovascular system.Therefore, patients requiring HFOVshould be hemodynamically stabilized be-fore initiation and a fluid bolus and/orvasopressors should be readily available ifhypotension occurs. We commonly usecentral venous pressure monitoring (e.g.,internal jugular or subclavian centralvein) or pulmonary artery flotation cath-eter monitoring to ensure optimal hemo-dynamics in unstable patients. In the ab-sence of a pulmonary artery flotationcatheter, heart rate and blood pressurecan be useful. Dampened arterial andpulse oximetry waveforms may indicatecompromised cardiac output and requirespecial attention by the RT to trend (7).In hypovolemic patients, heart rate read-ings on pulse oximetry may be observedthat reflect the oscillatory frequency be-ing delivered. This phenomenon maysuggest that cardiac output is being com-promised as a result of high intrathoracicpressures. In addition, waveform tracingsof central venous pressure or pulmonarycapillary wedge pressure (PCWP) maysometimes show low-amplitude superim-posed waves that correspond to the HFOVfrequency (Hz). Brief interruption of theoscillator piston (while maintainingmPaw) may be considered when trying toobtain an accurate assessment of centralvenous pressure or PCWP waveforms.Changes in mPaw may produce similarchanges in central venous pressure orPCWP, suggesting that at least some ofthe mPaw is being transmitted to theintravascular pressures. The response ofblood pressure, central venous pressure,or PCWP trends in response to fluid chal-lenges may be a better indicator of thepatient’s intravascular volume statusthan any absolute value, particularly inhypotensive patients requiring highmPaw.

OPTIMIZING HIGH-FREQUENCYOSCILLATORY VENTILATIONSETTINGS

To optimally manage critically ill pa-tients on HFOV, it is important to under-stand the machine’s capabilities and lim-itations. HFOV has been considered a“decoupling device.” By definition, a de-coupling device uses individual controlsthat affect only certain parameters andnothing else. For example, mPaw, FIO2,


and inspiratory time % (IT%) primarilyaffect oxygenation, whereas increasingoscillatory pressure amplitude (�P), de-creasing frequency (Hz), and creatingETT cuff leaks primarily increase ventila-tion. Clinical observations, however, sug-gest that HFOV is not an absolute decou-pling device. To maximize gas exchangewhile minimizing ventilator-associatedlung injury, it is imperative to maintainan optimal mPaw while avoiding excesstidal volume delivery. Alveolar overdis-tension or underdistension from an im-properly set mPaw or �P:Hz combinationcan adversely affect both oxygenation andventilation respectively and potentiatefurther lung injury.

Optimizing Mean AirwayPressure

Determining an appropriate mPawsetting can prove to be particularly chal-lenging, especially during recruitmentphases. Optimal mPaw can be describedas that mPaw which causes sufficientlung inflation to maximize gas exchangewhile protecting the lung from alveolaroverdistension, underdistension, or im-peding hemodynamics. In hemodynami-cally stable adults, mPaw is typicallystarted at 2–5 cm H2O above the mPawobserved during CV. Subsequent in-creases in mPaw by 1–2 cm H2O every30–60 mins are used (up to a maximumof 40–45 cm H2O) to achieve a targetSpO2 �88% with an FIO2 �60%. Whetherto wean mPaw before reducing FIO2 inpatients who require high mPaw (e.g.,�35 cm H2O) for optimal lung protectionremains unclear. One approach to usingcombinations of mPaw, FIO2, and lung

recruiting maneuvers for oxygenationand weaning is outlined in Appendix 1.

Application of technologies such as re-spiratory inductance plethysmographyand electrical impedance tomographyduring HFOV may offer more precise bed-side tools for assessing lung inflation andare reviewed elsewhere in this supple-ment.

OSCILLATORY PRESSUREAMPLITUDE (�P)—RANDOMDRIFT OR PATIENTFEEDBACK?

During HFOV, “real-time” feedbackand trending of airway and lung mechan-ics may be available through monitoringchanges in the displayed �P (21, 23).Understanding that a number of clinicalvariables may cause changes in �P canprovide the bedside clinician with a use-ful monitoring tool.

Increases in �P (assuming constantpower setting) may occur with increasesin ETT resistance, bronchospasm, ormainstem intubation (21). Thoracic com-pliance changes may have variable effectson �P and may not be distinguishablefrom airway resistance effects. Similarly,variable ETT cuff leaks or spontaneousbreathing may cause fluctuations in �P.As a result of the many contributing fac-tors associated with changes in �P, pa-tient care decisions should not be madesolely on this measurement. Rather, �Pshould be included as an additional mon-itoring parameter in conjunction withother patient assessment techniques, he-modynamic parameters, and arterialblood gas analysis. We routinely recordthe observed �P during ventilator checks

and monitor trends as an early indicatorof possible changes in pulmonary me-chanics. In situations of acute decompen-sation, an understanding of the variablesaffecting �P may give the clinician anadditional tool to determine whether thecause of decompensation is pulmonary ornonpulmonary. Table 3 depicts variousclinical situations and their possible ef-fects on �P.

HIGH-FREQUENCYOSCILLATORY VENTILATIONAND AEROSOL MEDICATIONDELIVERY

Very few studies have examined meth-ods to optimize aerosolized medicationdelivery during HFOV. Metered dose in-halers (MDI) appear relatively ineffectivein delivering optimal drug amounts. MDIdelivery during HFOV has been shown todeliver approximately 1% to 2% of theaerosolized drug in a neonatal lungmodel and 2.5% to 6.3% in a pediatriclung model (24). In this study, the lowdeposition was attributed to turbulentflow in conjunction with the high biasflow in the system and the small diameterof the ETT. No difference was noted be-tween differing settings on the HFOV.

Use of flow-driven nebulizers in-lineduring HFOV has not been thoroughlystudied. This form of aerosol generatorprovides an additional amount of flow tothe circuit that will result in alterationsof the mPaw and the �P. During deliveryof aerosolized medication, the RT mustreduce oscillator bias flow to maintainconstant mPaw (25).

New aerosol generators that use a vi-brational element to generate a low-

Table 3. Affects of clinical situations on proximal amplitude (�Pproximal) and possible treatments

Condition Compliance Vt �Pcarinal �Pproximal Treatment

Alveolar overdistension Decreased Decreased Increased Decreased Decrease mPaw incrementallyTension pneumothorax Decreased Decreased Increased Decreased Decrease mPaw for adverse reactions

Chest tube placementMucous plugging Decreased or same Decreased Increased Decreased Bag, saline lavage and suction

BronchoscopyBronchoconstriction Decreased Decreased Increased Decreased Bronchodilator

SteroidsFulminating pulmonary edema

(endotracheal tube frothing)Decreased Decreased Decreased Increaseda Increase mPaw

Increasing endotracheal tuberesistance (partial occlusion)

No Change Decreased Decreased Increased Bag, saline lavage, and suction

BronchoscopyEndotracheal tube occlusion No Change Decreased Decreased Increaseda Bag, saline lavage, and suction

ReintubationAlveolar recruitment Increased Increased Decreased Increased Monitor for over-distension

aSituation may result in �Pproximal changes � 10 cm H2O; �Pproximal is measured in the oscillator circuit and displayed on the ventilator.


velocity, small droplet medication haveshowed promising results in optimizingdrug delivery. This form of aerosol gen-erator delivers up to the three times theamount of drug without effecting venti-lator parameters when compared withtraditional flow-driven aerosol generators(26). Aerosol drug delivery during HFOVshould be investigated further to deter-mine optimal techniques. The use ofaerosolized selective pulmonary vasodila-tor medications is reviewed elsewhere inthis supplement.

CONCLUSION

Integration of respiratory therapy ex-pertise into the management of adult pa-tients on HFOV is an essential compo-nent to successful outcomes with thisnovel mode of ventilation. Respiratorytherapists serve as team leaders in thedevelopment and implementation ofHFOV treatment algorithms. In addition,the RT plays a vital role in monitoringpatients on HFOV, in early recognition ofchanging clinical conditions (e.g., ob-structing ETTs, pneumothorax, hyperin-flation), and in HFOV-related procedures(e.g., creating ETT cuff leaks, lung re-cruitment maneuvers). Optimal tech-niques to deliver aerosol medicationsduring HFOV remain unclear and requirefurther study.

REFERENCES

1. Fort P, Farmer C, Westerman J, et al: High-frequency oscillatory ventilation for adult re-spiratory distress syndrome—a pilot study.Crit Care Med 1197; 25:937–947

2. Mehta S, Lapinsky SE, Hallet DC, et al: Aprospective trial of high-frequency oscilla-tion in adults with acute respiratory distresssyndrome. Crit Care Med 2001; 29:1360–1369

3. Derdak S, Mehta S, Stewart TE, et al: Highfrequency oscillatory ventilation for acute re-spiratory distress syndrome: A randomized,

controlled trial. Am J Respir Crit Care Med2002; 166:801–808


5. Mehta S, Granton J, MacDonald RJ, et al:High-frequency oscillatory ventilation inadults. The Toronto experience. Chest 2004;126:518–527

6. Wilkins RL, Krider SJ, Sheldon RL: ClinicalAssessment in Respiratory Care. Third Edi-tion, St. Louis, Mosby, 1995

7. Scanlan CL, Spearman CB, Sheldon RL:Egan’s Fundamentals of Respiratory Care.Sixth Edition. St. Louis, Mosby, 1995

8. Johnson MM, Ely W, Chiles C, et al: Radio-graphic assessment of hyperinflation. Corre-lation with objective chest radiographic mea-surements and mechanical ventilatorparameters. Chest 1998; 113:1698–1704

9. Dolan S, Derdak S, Solomon D, et al: Tra-cheal gas insufflation combined with high-frequency oscillatory ventilation. Crit CareMed 1996; 24:458–465

10. Van de Kieft M, Dorsey D, Venticinque S, etal: effects of endotracheal tube (ETT) cuffleak on gas flow patterns in a mechanicallung model during high-frequency oscilla-tory ventilation (HFOV). Am J Respir CritCare Med 2003; 167:A178

11. Cartotto R, Ellis S, Gomez M, et al: Highfrequency oscillatory ventilation in burn pa-tients with the acute respiratory distress syn-drome. Burns 2004; 30:453–463

12. Cooper AB, Islur A, Gomez M, et al: Hyper-capneic respiratory failure and partial upperairway obstruction during high-frequencyoscillatory ventilation in an adult burn pa-tient. Can J Anesth 2002; 49:724–728

13. Schiffmann H, et al: Determination of airwayhumidification in high-frequency oscillatoryventilation using an artificial neonatal lungmodel: comparison of a heated humidifierand a heat and moisture exchanger. Inten-sive Care Med 1999; 25:997–1002

14. American Association of Respiratory CareClinical Practice Guidelines. Humidificationduring mechanical ventilation. Respir Care1992; 37:887–890

15. 3100B Operator’s Manual. Yorba Linda, CA,Sensormedics Corp, 2002

16. Maggiore SM, Lellouche F, Pigeot J, et al:

Prevention of endotracheal suctioning-induced alveolar derecruitment in acute lunginjury. Am J Respir Crit Care Med 2003;167:1215–1224

17. Lee D, Chiang H, Lin S, et al: Prone-positionventilation induces sustained improvementin oxygenation in patients with acute respi-ratory distress syndrome who have a largeshunt. Crit Care Med 2002; 30:1446–1452

18. Gattinoni L, Tognoni G, Presenti A, et al:Effect of prone positioning on the survival ofpatients with acute respiratory failure.N Engl J Med 2001; 345:568–573

19. Varkul M, Stewart T, Lapinsky S, et al: Suc-cessful use of combined high-frequency os-cillatory ventilation, inhaled nitric oxide, andprone positioning in the acute respiratorydistress syndrome. Anesthesiology 2001; 95:797–799

20. Messerole E, Peine P, Wittkopp S, et al: Thepragmatics of prone positioning. Am J RespirCrit Care Med 2002; 165:1359–1363

21. Morison D, Derdak S: High frequency oscil-latory ventilator parameter changes in re-sponse to simulated clinical conditions usinga mechanical test lung. Am J Respir Crit Care2000; 163:A388

22. Ellsbury DL, Klein JM, Segar JL: Optimiza-tion of high-frequency oscillatory ventilationfor the treatment of experimental pneumo-thorax. Crit Care Med 2002; 30:1131–1135

23. Holland D, O’Keefe G: Monitoring proximaland carinal amplitude changes to identifychanges in airway resistance and lung com-pliance during high frequency oscillatoryventilation (HFOV). Journal of RespiratoryCare 1999; 44:1243A

24. Garner SS, West DB, Bradley JW: Albuteroldelivery by metered dose inhaler in a pediat-ric high-frequency oscillatory ventilationmodel. Crit Care Med 2000; 28:2086–2089

25. Higgins J, Diebold A, Mellor S, et al: Theevaluation of aerosolized albuterol deposi-tion in-line via conventional ventilation ver-sus high frequency oscillatory ventilation(HFOV). Journal of Respiratory Care 2001;46:1080A

26. Fink JB, Barraza P, Bisgaard J: Aerosol deliv-ery during mechanical ventilation with high-frequency oscillation: an in vitro evaluation.Chest 2001; 120:277S

The Appendix is on the next page.


APPENDIX


Nursing and infection-control issues during high-frequencyoscillatory ventilation

Anne-Marie Sweeney, RN, BSc, BScN, CNN(N), CNN(C), PHCNP, MN; Joseph Lyle, BSc, MBA, RRCP;Niall D. Ferguson, MD, FRCPC, MSc

Clinicians are becoming in-creasingly conscious of thepotentially harmful effects ofmechanical ventilation in

adults with hypoxemic respiratory failure(1–3). More attention is now being fo-cused on methods that may limit venti-lator-induced lung injury, including theuse of high-frequency oscillatory ventila-tion (HFOV) (4, 5). These factors, alongwith the advent of a commercially avail-able ventilator capable of oscillatingadults, have led to a significant increasein the use of HFOV in adult intensive careunits (ICUs) in the past 10 yrs. Introduc-ing a new HFOV program in an adult ICU,however, involves a great deal more thansimply obtaining a new ventilator andidentifying the appropriate patient popu-lation. We must be cognizant of the neo-natal HFOV experience, in which therapid expansion of HFOV to centers with

insufficient experience and training mayhave led to adverse patient outcomes (6,7). As detailed in other articles in thisissue of Critical Care Medicine, specificknowledge regarding optimal methodsfor setting and managing HFOV is neededfrom both physicians and respiratorytherapists. There are, however, severalnursing issues that must be addressedbefore the successful implementation ofan HFOV program in an adult ICU, whichwe will explore in this article. In addition,our experience with severe acute respira-tory syndrome (SARS) (8–11) has high-lighted the importance of infection-control issues with HFOV, which we willoutline below.

HFOV Nursing Issues

Early in our experience with this mo-dality, nurses often saw HFOV as a ther-apy of last resort for patients who werefailing conventional ventilation and whowere already close to death. It is not sur-prising, therefore, that HFOV was oftenperceived as futile. More recently, how-ever, this perception has been changingwith the introduction of HFOV earlier inthe course of management. In our unit,HFOV has become one of the primarymodes of early rescue therapy for patients

with severe acute respiratory distress syn-drome. These extremely ill patients usu-ally require a nurse-to-patient ratio of1:1, although we have occasionallyneeded to have two nurses for one pa-tient.

Patient Assessment. Assessment of thepatient receiving HFOV is challenging fora number of reasons. One reason whyassessment may be difficult is noise. Thepiston pump of the oscillator produces anoise that can be perceived as loud andrepetitive. When we first started usingHFOV, patients were given earplugs fornoise protection because HFOV was sig-nificantly louder than conventional ven-tilation (12). Newer versions of adult os-cillatory ventilators generally containimproved noise muffling systems, andnoise pollution from the ventilator is nowless of a concern. Nonetheless, the respi-ratory breath sounds are different duringHFOV compared with conventional ven-tilation. The very high frequencies used(typically 240–360 breaths/min in adults)and the very small tidal volumes deliv-ered make it difficult to impossible tohear usual sounds. In institutions wherenursing protocols have been developedregarding HFOV, no mention is made ofroutine auscultation (13). When one isauscultating the chest, the interest

From the Toronto Western Hospital, UniversityHealth Network (AMS, JL, NDF); the Michener Instituteof Applied Health Sciences (JL); and the Interdepart-mental Division of Critical Care Medicine and the De-partment of Medicine, Division of Respirology, Univer-sity of Toronto (NDF), Toronto, Ontario, Canada.


DOI: 10.1097/01.CCM.0000155918.29268.84

Objectives: To review the specific nursing and infection-con-trol issues that arise during the care of patients receiving high-frequency oscillatory ventilation (HFOV).

Data Source: Published articles, governmental guidelines, andhospital procedures and practices.

Data Summary: Nurses, respiratory therapists, and other cli-nicians caring for patients receiving HFOV need to be aware ofspecific differences in patient assessment, including close obser-vation for symmetric chest-wall vibrations. In addition, manage-ment of sedation with or without neuromuscular blockade andeffective communication with the patients are essential nursingskills needed with the use of HFOV. From an infection-controlstandpoint, HFOV is considered a high-risk respiratory procedure

because of the inability to effectively filter all respiratory secre-tions. Appropriate infection-control precautions, including patientlocation and use of personal protective equipment, need to beconsidered when implementing HFOV in the intensive care unit.

Conclusions: Important infection-control and nursing issuesexist that are specific to the use of HFOV. These issues shouldbe addressed with appropriate staff education before the im-plementation of HFOV in an intensive care unit. (Crit Care Med2005; 33[Suppl.]:S204 –S208)

KEY WORDS: high-frequency ventilation; nursing assessment;infection control; respiratory therapy; family communication; se-vere acute respiratory syndrome; acute respiratory distress syn-drome


should not be in the absolute quality ofbreath sounds, but rather in a changefrom what was heard at baseline (14).Nurses caring for patients receivingHFOV want to ensure that equal breathsounds are heard bilaterally.

Inspection and palpation of the respi-ratory system is also different while ad-ministering HFOV, as the patients arecontinually moving. The way HFOV isdelivered—by small volumes at high fre-quencies—causes patients to vibrate; thepatients seem to be wiggling. Despite thisdifference, inspection to determinewhether chest inflation and vibration aresymmetrical is a very important compo-nent of the assessment by nurses (andother clinicians) of the patient receivingHFOV. Vibration that is not symmetricalis a finding of concern, as this could in-dicate the presence of a pneumothorax,main-stem bronchus intubation, mucousplugging, or lung collapse (12, 14).

Nurses caring for the patient receivingHFOV should document the presence andextent of vibration or wiggle. A wigglethat extends to the mid-thigh region hasbeen reported to be an indicator of ade-quate oscillation and has been used as anindex against which to titrate initialpower settings (12). Empirical evidencesupporting this convention, however, islacking. It is therefore important that ar-terial blood gas measurements are per-formed shortly after making the transi-tion to HFOV so that power andfrequency can be adjusted to ventilationas needed. As with asymmetric chest vi-bration, sudden changes in the extent ofwiggle or vibration are important to noteand investigate.

Careful assessment and ongoing closeobservation of the HFOV patient is im-portant because the usual ventilatoralarms that might alert clinicians toproblems, such as low-volume or high-pressure alarms, do not function in thesame way with HFOV. Patients may havea large pneumothorax, and indeed, as anextreme example, one can clamp the en-dotracheal tube without any significantchange in mean airway pressure or anyalarm sounding. It is therefore importantto assess the patient’s respiratory statususing a combination of bedside observa-tion, continuous blood pressure and ar-terial oxygen saturation monitoring, andperiodic arterial blood gas measure-ments. Chest radiographs should usuallybe performed at least daily to monitor thedegree of lung inflation and observe forgross barotrauma. A key monitoring mes-

sage is that the absence of a ventilatoralarm does not rule out a change in or aproblem with the patient’s respiratorysystem. With HFOV, and indeed with anyventilatory mode, we must assess the pa-tient as a whole and not just look at theventilator.

Patient Comfort and Monitoring ofSedation and Paralysis. Ensuring patientcomfort has always been a nursing prior-ity and challenge. Unlike neonates, whoare able to breath spontaneously whilereceiving HFOV, adults who requireHFOV usually have their spontaneous re-spiratory efforts suppressed. This sup-pression is needed because inspiratoryflow rates generated by these patients arelarge enough outstrip the constant flowof gas in the ventilator circuit (15). This,in turn, leads to both a drop in the meanairway pressure of the circuit and a cutoffof inspiratory flow to the patient. Alladults switching to HFOV therefore needto be deeply sedated, and in many cases,chemical paralysis must be induced usingneuromuscular blocking agents (NMBs).This is one major reason why HFOV iscurrently reserved for patients with moresevere forms of acute lung injury, as wewould not want to deeply sedate and par-alyze patients with mild disease.

Neuromuscular blockers are beingused less frequently in recent years asclinicians have become more concernedwith their associations with long-termadverse outcomes (16). When HFOV wasinitially introduced to larger children andadults, all patients were treated with con-tinuous infusions of NMBs (17). More re-cently, however, we are discovering thatsome patients can be managed on HFOVwith deep sedation but without requiringparalytic agents. NMBs do not have anysedative or analgesic properties, but theydo make it difficult to assess a patient’slevel of sedation and analgesia. All pa-tients who are receiving NMBs shouldalso be receiving continuous infusions ofsedatives and analgesics. It is importantto remember that unexplained tachycar-dia or hypertension in a paralyzed patientmay be an indicator of inadequate seda-tion or analgesia. When switching toHFOV, whenever possible, our practice isto first titrate infusions of sedatives andanalgesics to a level at which the patientis unresponsive to stimuli. We thenswitch to HFOV and observe the patientclosely—if the patient is making respira-tory efforts that are large enough to causethe mean airway pressure to drop bymore than 2–3 cm H2O, or if their oxygen

saturation is low, we will usually give abolus dose of neuromuscular blockers.The patient is then observed again, and ifthese issues recur after the initial NMBdose has worn off, only then do we begina continuous infusion of NMBs.

Patients who are receiving NMBsshould be assessed regularly to gauge thedegree of neuromuscular blockade usinga peripheral nerve stimulator and train-of-four monitoring (TO4). The nervestimulator is attached (using standardelectrocardiographic monitoring elec-trodes) to deliver a mild electrical currentto (for example) the median nerve. Ade-quate neuromuscular blockade is indi-cated by the presence of one to twothumb twitches on TO4 testing, indicat-ing a 75–90% blockade (18–20). This as-sessment of neuromuscular blockadeshould be done every hour to ensure thegoal of one to two twitches is maintained.Excessive blockade or prolonged use ofNMBs have been associated with persis-tent paralysis, muscle weakness, and dif-ficulty weaning from mechanical ventila-tion (18–20). For a more comprehensivediscussion of ongoing neuromuscularblockade, sedation, and analgesia in thecritical care setting, the reader is referredto two recent clinical practice guidelines(21, 22).

TO4 monitoring is performed largelyto guard against excessive neuromuscu-lar blockade. When the TO4 indicates thepresence of minimal or no blockade (TO4� 3–4), the patient’s clinical context ofoxygenation and ventilation should beevaluated. A TO4 of �2 accompanied bydesaturations or worsening gas exchangeis usually indicative of the need to in-crease the dose of NMBs. In contrast, aTO4 of 3 or 4 with no desaturations,small ventilatory efforts, mean airwaypressure fluctuations, and favorable he-modynamics should be discussed withthe healthcare team to determine thecourse of treatment. This assessmentmay indicate that a trial of discontinua-tion of NMBs would be appropriate. Un-less the patient has persistent severe hy-poxemia with high FIO2 requirementswile receiving HFOV, we generally at-tempt to discontinue NMBs on a dailybasis.

Family Communication and Teach-ing. For families of critically ill patients,the ICU is very often an unknown, over-whelming, and frightening place. Likethose of other patients, the family of thepatient receiving HFOV must have ongo-ing two-way communication with the


healthcare team (23). In addition, we be-lieve that it is important to discuss HFOVin particular and to educate families re-garding: 1) the purpose of HFOV, 2) howHFOV is expected to benefit the patient,3) how the patient’s care while receivingHFOV will differ from previous care pro-visions, and 4) what to expect when theysee the patient receiving HFOV. Theseexplanations should be given either be-fore or immediately on the family’s ar-rival at the bedside and, obviously, mustbe given in terms the family can under-stand and comprehend.

In general, when we are discussingthis issue with families, we explain thatHFOV is a new type of ventilator thatoften is able to provide additional helpgetting oxygen into the blood in cases ofsevere respiratory failure. Depending onthe situation, we may also touch on basicconcepts of ventilator-induced lung in-jury and lung-protective ventilation. Weexplain that HFOV delivers breaths thatare very small in size but that are deliv-ered very rapidly. It is important to makeclear to the family that the patient willlook different while receiving HFOV andthat this is to be expected. They should betold that the patient will be deeply se-dated (with or without muscle relaxants,as appropriate to the situation). We usu-ally say that the patient may not seem tobe breathing in a normal fashion, butrather, they will seem to be shaking orvibrating. Despite these appearances,however, gas exchange with HFOV is ad-equate and is being monitored closely.The amount of equipment surroundingthe bed may also be increased, with in-travenous pumps for medications, moni-tors, and ventilators. The family shouldbe reassured that the patient is being keptcomfortable and pain free; they maytouch and talk to the patient but shouldnot expect a response. We find that thesefamily discussions are most effectivewhen they are conducted in tandem withboth nursing and physician input andwith reassurance being provided to thefamily. In addition, because patients re-ceiving HFOV are among the most se-verely ill in the ICU, other family sup-ports within the critical care setting suchas chaplaincy or social work may be ap-propriate.

Infection-Control Issues

Usual Precautions. The SARS crisis of2003 has forced us and other centers toreconsider our approach to many “rou-

tine” ICU procedures. Because of its po-tential to generate droplets from patients’respiratory secretions, HFOV is groupedwith other procedures, such as noninva-sive ventilation, intubation, and nebu-lized therapies, as a high-risk respiratoryprocedure (24, 25). A full discussion re-garding preparedness and procedures fora respiratory outbreak such as SARS isbeyond the scope of this article, but thereader is referred to online resourcessuch as those of the Centers for DiseaseControl (www.cdc.gov) or the OntarioMinistry of Health (www.health.gov.on.ca). In general, the risk of SARS trans-mission to healthcare workers caring forSARS patients has been shown to be re-lated to the type (e.g., intubation), theproximity, and the duration of exposure(10, 11). These principles need to be con-sidered when dealing with any patientwho has or may have SARS.

The issue with HFOV that distin-guishes it from conventional ventilatorsis the constant venting of gas out of themean airway pressure control diaphragm.Using the standard circuit for the Sensor-Medics 3100B oscillator (Viasys, YorbaLinda, CA), these gases are unfiltered andaerosolized into the room, posing an oc-cupational hazard for clinicians. Becauseof this, in the post-SARS era, appropriateprecautions must be taken when usingthis kind of ventilator.

The degree of precautions that shouldbe used when applying HFOV is depen-dent on a number of factors, includingthe presence of a febrile respiratory ill-ness, SARS risk factors, contact history,and whether an outbreak is ongoing (24).For example, in a nonoutbreak setting, apatient with an undifferentiated febrilerespiratory illness who required HFOVwould be treated in a single room withnegative-pressure ventilation. In this sit-uation, staff treating this patient woulduse personal protective equipment in-cluding goggles, N95 mask, gown, andgloves. In a patient with either a provenbacterial infection or a known noninfec-tious cause of acute respiratory distresssyndrome, we employ droplet precau-tions during HFOV, consisting of the useof (at least) a surgical mask, goggles, andgloves for staff providing patient care. Asingle patient room is still preferred dur-ing droplet precautions, but if unavail-able, these patients can be treated in amultiple-patient room provided that aseparation of �1 m be maintained fromother patients. In outbreak settings,HFOV has not been proven to be strongly

associated with SARS transmission, incontrast to endotracheal intubation (11).Nevertheless, because of concerns out-lined above, during the SARS outbreak inToronto, HFOV use was avoided, and thismodality was used only when deemedmedically essential.

Additional Precautions. As a means offurther protection for our staff at theUniversity Health Network, we have nowbegun using the SensorMedics Scavenger(Viasys) as a protective device over themean airway pressure control diaphragm(Fig. 1 and close-up in Fig. 2). We shouldpoint out that this scavenger is not ap-proved by the U.S. Food and Drug Admin-istration for this purpose. The coveringcap is attached to a short length of cor-rugated tubing with a water trap and ahigh-efficiency bacterial/viral filter placedat the end. We use the DAR SteriventMini filter (Mallinckrodt DAR, Mirandola,Italy), which is changed every 12 hrs orwhen visibly wet. The scavenger also hasa connection for suction, which we attachto wall suction with a collection device,setting the suction level so that minimalflow is felt through the filter. This appa-ratus greatly reduces the spray of waterinto the area, thus further protectinghealthcare professionals when dealingwith these patients. We wish to stress,

Figure 1. High-frequency oscillator with resis-tance-valve scavenger. A SensorMedics 3100B os-cillator (Viasys, Yorba Linda, CA) is shown with ascavenger cap (white arrow) attached over themean airway pressure control diaphragm.


however, that this is used as adjunctiveprotection and not as a substitute forpatient isolation or personal protectiveprocedures.

Humidification and Secretion Man-agement. Secretion and humidificationmanagement deserve mention when con-sidering infection-control issues withHFOV because they may influence thetype and duration of exposures for health-care workers. Adequate humidification ofthe bias flow is essential during HFOV toprevent the development of thick tena-cious secretions, mucous plugging, andtracheobronchitis. Because of high–biasflow rates, we find that the humidifierchamber in our vaporizers can often usewater at a very high rate. Special dili-gence must be paid, therefore, to ensurethat an acceptable water level is main-tained in the humidifier chamber. Grav-ity-feed water filling systems have, in ourexperience, not been able to cope withthe high airway pressures employed dur-

ing HFOV, especially as the water bagnears empty. To deal with this problem,we use a nondisposable chamber and aclamping water transfer set, manually en-suring that the humidifier chamber re-mains adequately filled. The main disad-vantage of this system is that it requiresfrequent monitoring and intervention bythe respiratory therapists. This is an issuewhen patients are in respiratory isolationbecause monitoring becomes more diffi-cult, and the manual intervention leadsto longer duration of staff exposure toisolated patients. This is another reasonwhy HFOV was used infrequently duringour SARS outbreak. For suctioning dur-ing HFOV, we believe that in-line suctioncatheters should be used instead of con-ventional suctioning. Their use protectsclinicians from being exposed to patientsecretions and spray from the ventilator,as there is no stand-by application on theSensorMedics 3100B. In addition, the useof in-line suctioning minimizes discon-nection times and probably results in lesssuction-related derecruitment than con-ventional suctioning.

Conclusions

Both nursing considerations and in-fection-control precautions are impor-tant to consider when implementingHFOV. From a nursing standpoint, keyfactors in caring for the patient receivingHFOV include awareness of the differ-ences in patient assessment, the propermanagement of sedation with or withoutneuromuscular blockade, and effectivecommunication with the patient’s family.Infection-control issues arise with HFOVbecause of our current inability to ade-quately filter exhaled respiratory secre-tions. HFOV is therefore a high-risk re-spiratory procedure, and appropriateprecautions should be employed when itis used. The precise nature of these pre-cautions should be dictated both by spe-cific patient factors and by whether theyoccur in an outbreak setting.

REFERENCES

1. Amato MB, Barbas CS, Medeiros DM, et al:Effect of a protective-ventilation strategy onmortality in the acute respiratory distresssyndrome. N Engl J Med 1998; 338:347–354


3. Ranieri VM, Suter PM, Tortorella C, et al:Effect of mechanical ventilation on inflam-matory mediators in patients with acute re-spiratory distress syndrome: A randomizedcontrolled trial. JAMA 1999; 282:54–61

4. Froese AB: High-frequency oscillatory venti-lation for adult respiratory distress syn-drome: Let’s get it right this time. Crit CareMed 1997; 25:906–908

5. Ferguson ND, Stewart TE: New therapies foradults with acute lung injury: High-fre-quency oscillatory ventilation. Crit Care Clin2002; 18:91–106

6. High-frequency oscillatory ventilation com-pared with conventional mechanical ventila-tion in the treatment of respiratory failure inpreterm infants: The HIFI Study Group.N Engl J Med 1989; 320:88–93

7. Bryan AC, Froese AB: Reflections on the HIFItrial. Pediatrics 1991; 87:565–567

8. Fowler RA, Lapinsky SE, Hallett D, et al:Critically ill patients with severe acute respi-ratory syndrome. JAMA 2003; 290:367–373

9. Booth CM, Matukas LM, Tomlinson GA, et al:Clinical features and short-term outcomes of144 patients with SARS in the greater To-ronto area. JAMA 2003; 289:2801–2809

10. Scales DC, Green K, Chan AK, et al: Illness inintensive care staff after brief exposure tosevere acute respiratory syndrome. EmergInfect Dis 2003; 9:1205–1210

11. Fowler RA, Guest CB, Lapinsky SE, et al:Transmission of severe acute respiratory syn-drome during intubation and mechanicalventilation. Am J Respir Crit Care Med 2004;169:1198–1202

12. Hynes-Gay P, MacDonald R: Using high-frequency oscillatory ventilation to treatadults with acute respiratory distress syn-drome. Crit Care Nurse 2001; 21:38–47

13. Waisman D, Weintraub Z, Rotschild A, et al:High-frequency oscillatory ventilation:“Please do not forget about me,” said thestethoscope. Pediatrics 2001; 108:819

14. Ritacca FV, Stewart TE: Clinical review:High-frequency oscillatory ventilation inadults. A review of the literature and practi-cal applications. Crit Care 2003; 7:385–390

15. Ferguson ND, Stewart TE, Slutsky AS: High-frequency oscillatory ventilation: A tool todecrease ventilator-induced lung injury? In:2000 Yearbook of Intensive Care and Emer-gency Medicine. Vincent JL (Ed). Berlin,Springer, 2000, pp 400–410

16. Herridge MS, Cheung AM, Tansey CM, et al:One-year outcomes in survivors of the acuterespiratory distress syndrome. N Engl J Med2003; 348:683–693


18. Arbour R: Mastering neuromuscular block-ade: Part 1. Nursing 2000; 30:CC1–CC5

19. Arbour R: Mastering neuromuscular block-ade: Part 2. Nursing 2000; 30:CC1–CC6

Figure 2. Close-up view of scavenger device. ASensorMedics 3100B oscillator (Viasys, YorbaLinda, CA) is shown with a scavenger cap (largewhite arrow) attached over the mean airway pres-sure control diaphragm. The scavenger is, inturn, attached to wall suction and to a high-efficiency filter (arrowhead) and a water trap(right arrow).


20. Luer JM: Sedation and neuromuscular block-ade in patients with acute respiratory failure.Crit Care Nurse 2002; 22:70–75

21. Murray MJ, Cowen J, DeBlock H, et al: Clin-ical practice guidelines for sustained neuro-muscular blockade in the adult critically illpatient. Crit Care Med 2002; 30:142–156

22. Jacobi J, Fraser GL, Coursin DB, et al: Clin-ical practice guidelines for the sustained useof sedatives and analgesics in the critically illadult. Crit Care Med 2002; 30:119–141

23. McDonagh JR, Elliott TB, Engelberg RA, et

al: Family satisfaction with family confer-ences about end-of-life care in the intensivecare unit: Increased proportion of familyspeech is associated with increased satisfac-tion. Crit Care Med 2004; 32:1484–1488

24. Government of Ontario, Ministry of Healthand Long-Term Care: Directive to All OntarioAcute Care Facilities for High-Risk Respira-tory Procedures (includes both non-outbreakand outbreak conditions). Available onlineat: http://www.health.gov.on.ca/english/providers/program/emu/sars/sars_obc/ppe/

dir_2102203_high_risk_respiratory.pdf. Ac-cessed November 8, 2004

25. Centers for Disease Control and Prevention:Public Health Guidance for Community-Level Preparedness and Response to SevereAcute Respiratory Syndrome (SARS) Supple-ment I: Infection Control in Healthcare,Home, and Community Settings. III. Infec-tion Control in Healthcare Facilities. Avail-able online at: http://www.cdc.gov/ncidod/sars/guidance/I/healthcare.htm. AccessedNovember 8, 2004


Sedation, analgesia, and neuromuscular blockade forhigh-frequency oscillatory ventilation

Curtis N. Sessler, MD, FCCM, FCCP

A constant principle in the man-agement of critically ill pa-tients is the provision of safetyand comfort (1). Achieving

these goals requires clinician awarenessof the numerous conditions that can con-tribute to patient distress or interferewith effective patient care (2). These con-ditions include underlying medical con-ditions, acute medical or surgical illness,mechanical ventilation, invasive medicaland nursing interventions, medicationsthat contribute to delirium, hospital-acquired illness, and intensive care unit(ICU) environmental influences (2).Comprehensive management includesmodification of these underlying causesof distress and pharmacologic and non-pharmacologic measures to optimizecomfort and enhance tolerance of the ICUenvironment (2, 3). Conditions such asacute respiratory distress syndrome(ARDS) present particular challenges insedation management and may require

use of neuromuscular blocking agents(NMBAs) for effective mechanical ventila-tion and oxygenation (4), particularlyduring use of strategies such as high-frequency oscillatory ventilation (HFOV).

FREQUENCY OF SEDATIVE,ANALGESIC, AND NMBA USEIN MECHANICAL VENTILATION,ARDS, AND HFOV

Sedation and Analgesia

Facilitation of mechanical ventilationis among the most commonly cited indi-cations for sedative and analgesic medi-cations, and surveys indicate their use inthe majority of mechanically ventilatedpatients (5–10). In an international, pro-spective, observational study (11), 68% of5,183 mechanically ventilated patientsreceived sedative medications duringventilatory support (12). However, only4.5% of patients in this study had ARDS,a condition that often requires high-intensity ventilatory support, listed as thecause of acute respiratory failure (11). Asubsequent analysis of data from theARDS Network trial of low– vs. high–tidalvolume ventilation for ARDS (13) re-vealed that sedative medications were uti-lized in over 90% of patients during thefirst several days of mechanical ventila-

tion (14). Approximately 80% and 75% ofpatients continued to require sedativemedications on day 7 and day 14, respec-tively, with no significant difference insedative administration between the twotreatment groups based on size of tidalvolume (14). Virtually all patients treatedwith HFOV reported in recent clinicaltrials and case series received sedativemedications by continuous infusion tofacilitate mechanical ventilation (15–17).

Neuromuscular Blockade Use

The most commonly cited indicationfor administering NMBAs is to facilitatemechanical ventilation (18). In contrastto the widespread use of sedative medica-tions, administration of NMBAs is infre-quent for mechanically ventilated pa-tients, even including patients withARDS. Only 9% of 5,183 consecutive me-chanically ventilated adults from the pre-viously mentioned international study(11) received NMBAs during mechanicalventilation (19). Twenty-five percent of902 ARDS patients were receiving NMBAsat the time of enrollment into the ARDSNetwork low– vs. high–tidal volume trial(13), and 10–15% of patients continuedto receive NMBAs as far out as 14 days ofmechanical ventilation (14).

The published experience from clini-cal trials and recent case series suggests

From the Division of Pulmonary and Critical CareMedicine, Department of Internal Medicine, VirginiaCommonwealth University, Richmond, VA; and CriticalCare and Medical Respiratory Intensive Care Unit,Medical College of Virginia Hospitals, Richmond, VA.


DOI: 10.1097/01.CCM.0000156794.96880.DF

Objective: To provide a comprehensive review of the issuerelated to the administration of sedative, analgesic, and neuro-muscular blocking agents (NMBA) to patients who are receivingventilatory support for acute respiratory distress syndrome(ARDS) with high-frequency oscillatory ventilation.

Results: Sedative, analgesic, and NMBA are used with greatfrequency in patients with severe ARDS who are undergoing high-frequency oscillatory ventilation. In particular, the use of NMBA hasbeen higher than for other ARDS populations. Important consider-ations for effective treatment include careful patient evaluation,patient-based medication selection, identification of treatment goalswith periodic re-assessment, titration of medications to objectiveparameters such as sedation scales and peripheral nerve stimula-

tion, use of intermittent therapy when feasible, implementation ofdrug interruption strategies, and discontinuation of medications atthe earliest possible time. It is important to recognize that patientsevolve from severe ARDS through phases of recovery to the resolu-tion of respiratory failure and that ventilatory management, as wellas sedative and related medication requirements, will vary markedlyover the course of this process.

Conclusions: A multidisciplinary, structured approach that isbased on the considerations described should help achieve opti-mal results in this challenging patient population. (Crit Care Med2005; 33[Suppl.]:S209–S216)

KEY WORDS: high-frequency oscillatory ventilation; sedation;analgesia; neuromuscular blocking agents; protocols; scales


that NMBAs have been routinely utilizedfor the vast majority of patients treatedwith HFOV (15–17). Specifically, in a se-ries of 156 adult patients treated withHFOV at three ICUs since 1998, 90% ofpatients received NMBAs by continuousinfusion (17). In the recent multiple-center, controlled clinical trial, NMBAswere administered to all patients ran-domized to HFOV (15). The higher rate ofNMBA utilization with HFOV may be re-lated in part to the marked severity ofARDS, when compared with other ARDSpopulations. However, it is more likelythat a ventilatory strategy in which spon-taneous ventilation can be disruptive togas exchange, such as HFOV, will requiredeep sedation or chemical paralysis. Thisis in contradistinction to ventilatorymodes, like airway pressure–release ven-tilation, in which spontaneous ventila-tion is encouraged and deep sedation andparalysis may be avoided (20). Recent ex-pert recommendations now emphasizethat not all patients require NMBAs dur-ing HFOV and that attempts should bemade to avoid NMBAs altogether or limittreatment to intermittent boluses (21).

SEDATION, ANALGESIA, ANDNEUROMUSCULAR BLOCKADEMANAGEMENT STRATEGIES

The use of sedative agents and NMBAsis associated with longer duration of me-chanical ventilation and longer ICUlength of stay (12, 19, 22). This is notunexpected and may be related primarilyto higher severity of illness. However,there is evidence that strategies for effec-tive and targeted sedation managementcan achieve important outcomes such asreduced duration of mechanical ventila-tion, lower prevalence of tracheostomy,shorter ICU length of stay, reduced drugutilization, and reduced costs (23–26).Guidelines for the management of seda-tion, analgesia, and neuromuscularblockade in mechanical ventilation havebeen published (2, 3, 18, 27, 28), andcommon themes include multidisci-plinary structured management, drug ti-tration to objectively measured endpoints, individualized drug selection,strategies to use the lowest effective doseand avoid drug accumulation, and avoid-ance of adverse effects.

Challenges of ARDS and HFOV

Sedation and analgesia managementfor patients with ARDS is particularly

challenging because these patients gen-erally have a long duration of mechanicalventilation, often have concomitant non-pulmonary organ dysfunction that mayinfluence drug metabolism, require deepsedation to achieve patient-ventilatorysynchrony, and have serious conse-quences of inadequate synchrony such asprolonged oxygen desaturation or baro-trauma. During HFOV, the consequencesof uncontrolled spontaneous respirationmay be even more significant because astrong inspiratory effort by some patientscan cause a reduction in airway pressurebelow the preset lower limit, which isinterpreted by the ventilator as a circuitdisconnection, and oscillation is termi-nated (17). In addition, although sponta-neous breathing during HFOV may becomfortable for the patient, such breathsare not augmented— contrasting withmost conventional ventilatory modes(20).

Mechanical ventilatory needs, andtherefore sedative and NMBA needs, varymarkedly over the course of ARDS, frominitial stabilization of severe gas ex-change impairment to recovery. Accord-ingly, there are a number of transitionpoints of management, such as from neu-romuscular blockade plus sedation to se-dation alone, as the patient progressesfrom phase to phase (Fig. 1). Thesephases and management transitions eachrequire structured approaches, adding tothe complexity of management. Further,mechanical ventilation and sedationmanagement are often intertwined. Forexample, mental status is an importantdeterminant of successful weaning frommechanical ventilation (29), and direct

measures of mental status (29–31) or ofsedation therapy (32) are incorporatedinto many weaning protocols.

Key Components of SedationManagement Strategies

Common themes for management ofsedative and analgesic medications areoutlined in Table 1. Development ofguidelines or management protocolsmust be multidisciplinary, including phy-sicians, nurses, and respiratory therapistswho will play key roles in managing thesepatients. Successful implementation willrequire establishing “buy-in” from indi-viduals involved in patient care and frommanagers of involved units. Sustainabil-ity requires continual attention to theprocess.

Patient Evaluation. Many patientshave conditions that contribute to thedevelopment of pain, anxiety, and deliri-um—which are key components of pa-tient distress (2). Patient evaluationshould focus on detecting and then elim-inating or controlling these factors whenpossible. In addition, detection and quan-tification of pain should be specificallysought using a pain scale (3, 33). Use ofdeep sedation or neuromuscular block-ade, or both, makes conventional assess-ment difficult, and reliance on observa-tion of pain-related behaviors such asfacial grimacing or unexplained tachycar-dia or hypertension becomes necessary,recognizing the lack of specificity of theseobservations (3, 34). Use of intermittenttherapy (24) or scheduled interruption oftherapy (23) allows additional opportuni-ties for detection of pain.

Figure 1. Schematic representation of the progression of the stage of management, stage of mechan-ical ventilation, and stage of sedative and neuromuscular blockade therapy over the course of time insevere acute respiratory distress syndrome. Note the transitions, for example, from conventionalventilation to high-frequency oscillatory ventilation (HFOV) or from deep sedation to moderatesedation.


Goals of Sedation Therapy and Seda-tion Evaluation. The goals of treatmentwith sedative and analgesic medicationsshould be explicitly established for eachpatient, communicated within the ICUteam, and reexamined at least daily. Pro-vision of comfort and relief of distress isalways one of the goals; however, inARDS, and particularly with HFOV man-agement, patient-ventilator synchrony orapnea may be the target. In goal-directedsedation and analgesia, the use of a vali-dated sedation scale is useful (2, 3). De-sirable features of a sedation scale in-clude: a) multidisciplinary development,b) ease of use, recall and interpretation,c) well-defined discrete criteria for eachlevel, d) sufficient sedation levels for ti-tration of sedative and analgesic medica-tions, e) an assessment of agitation, and f)rigorous testing of reliability and validityin relevant patient populations (35).There are many scales (36, 37); however,sedation scales that have been tested forvalidity and reliability in adult patientsinclude the Ramsay sedation scale (38),

the sedation agitation scale (39), the mo-tor activity assessment scale (40), theVancouver interaction and calmness scale(41), the Richmond agitation-sedationscale (42), the adaptation to the intensivecare environment instrument (43), andthe Minnesota sedation assessment tool(44). The arousal components of recentlydeveloped scales, such as the Richmondagitation-sedation scale, the adaptationto the intensive care environment instru-ment, the Minnesota sedation assessmenttool, and the Ramsay sedation scale, relyon direct observation of simple responses(i.e., eye opening, movement) to progres-sively intense stimuli (the tester’s voice,then physical stimulation), yielding mul-tiple discrete levels of arousal/sedationthat are suitable for titrating sedativemedications (38, 42–44). The Richmondagitation-sedation scale also incorporateslevels of agitation and a measure of cog-nition within a single scale, which can allbe assessed in 30 secs (42). The adapta-tion to the intensive care environment isunique among the sedation assessment

tools in that patient tolerance of the ICUenvironment is also addressed, althoughapproximately 20 steps are needed to as-sess all five subscales (43). Use of a seda-tion scale can help reduce the frequencyof oversedation (45) and has been incor-porated into successful protocols (23, 24,26, 30).

Assessment of the level of sedation ismore complicated when an NMBA isadded because the patient responses thatare examined in sedation scales may bemasked by muscle paralysis. Unexplainedtearing, tachycardia, or hypertension mayraise suspicions that sedation and analge-sia is insufficient; however, these arenonspecific findings (34). Most helpful isthe use of intermittent bolus NMBA ther-apy or scheduled NMBA withdrawal thatprovides paralysis-free periods for seda-tion scale testing and assessment of pain.The employment of a device, such as theBispectral index, that converts electroen-cephalographic signals into a digital scalefrom 100 (fully alert) to 0 (isoelectricelectroencephalogram) with continuous

Table 1. Important components of treatment guidelines for sedation, analgesia, and neuromuscular blockade in patients with acute respiratory distresssyndrome

1. Perform patient assessment: evaluate for treatable factors contributing to pain, anxiety, delirium, and patient-ventilator dyssynchrony; optimizepatient comfort

2. Select sedative, analgesic, or neuromuscular blocking agent (NMBA) drugs based on clinical characteristics, risk of adverse effects, and costa. Sedatives

i. If rapid onset of action desired, select midazolamii. If rapid emergence desired, select propofol

iii. If renal insufficiency is present, avoid midazolamiv. If hemodynamic instability is present, avoid propofolv. If duration of �48 hrs, avoid propofol

b. Analgesicsi. If rapid onset of action desired, select fentanyl

ii. If renal insufficiency is present, avoid morphineiii. If hemodynamic instability is present, avoid morphineiv. For intermittent therapy, select morphine or hydromorphone

c. NMBAsi. If renal insufficiency or liver dysfunction is present, select atracurium or cisatracurium

ii. If tachycardia or hypertension are unacceptable, avoid pancuroniumiii. If intermittent dosing acceptable, select pancuronium (or vecuronium)

3. Establish treatment goals for sedative, analgesic, and NMBA medicationsa. Comfortable and free of distressb. Satisfactory patient-ventilator synchronyc. Apnead. Complete muscle relaxation

4. Utilize objective measures of sedation, analgesia, or neuromuscular blockadea. Sedation and analgesia

i. Sedation scale: document at least every 4 hrsii. Consider Bispectral index monitoring if neuromuscular blockade is in use

iii. Pain scaleb. Neuromuscular blockade

i. Clinical evaluationii. Train-of-four peripheral nerve stimulator monitoring—every 4 hrs (especially for vecuronium infusion or deep paralysis)

5. Titrate medications to achieve effectiveness based on treatment goals and safety using objective measures to avoid overdosage. Attempt to use thelowest effective dose, utilizing intermittent therapy or scheduled drug interruption, as tolerateda. Establish target (sedation scale, train-of-four, etc) and frequently adjust dosage to achieve target, using lowest effective doseb. Use intermittent sedative, analgesic, and NMBA therapy if feasiblec. If continuous infusion therapy, use daily interruption of medications (sedatives/analgesics, NMBAs)

6. Take steps to avoid adverse effects of treatment


display has been recommended as usefulfor assessing sedation and analgesianeeds for paralyzed patients (46, 47). Thecorrelation between the Bispectral indexand various sedation measures has beenonly moderately good for ICU patients ingeneral (48), dampening enthusiasm forwidespread implementation (3, 28). De-spite the observation that the addition ofan NMBA reduces muscle artifact thatfalsely elevates Bispectral index values(49), some investigators have noted thatBispectral index values vary widely, evenin paralyzed patients (50). However, ad-vances in methodology with newerBispectral index devices (51) seem to im-prove the accuracy and reduce artifact.Other monitoring devices, such as pa-tient state index, entropy monitoring, au-ditory-evoked potentials, and actigraphy,may eventually prove to be helpful formonitoring the level of arousal in criti-cally ill patients (52–55).

Sedative and Analgesic Medications. Avariety of sedative and analgesic medica-tions are used for long-term intravenousadministration to mechanically venti-lated ARDS patients, including propofol,midazolam, lorazepam, fentanyl, mor-phine, and hydromorphone (2, 3, 28).Dexmedetomidine, a selective alpha-2 ag-onist, has been approved only for short-term ICU administration at this time(56). No agent is considered clearly supe-rior than others (3, 28), and the choice ofdrugs should be influenced by patientcharacteristics (2). Propofol has a veryshort half-life, with numerous studiesdemonstrating more rapid recovery thanwith other agents; thus, it is preferredwhen rapid emergence is important (57).However, propofol has a higher preva-lence of hypotension when administeredand has been associated with lactic acido-sis, hypertriglyceridemia, and pancreati-tis (57). We generally avoid using propo-fol in hemodynamically unstable patients(2). From the standpoint of acquisitioncost, propofol is considerably more ex-pensive than generic midazolam or loraz-epam (2), but the difference in cost maybe offset by faster extubation (58, 59).Experts generally recommend thatpropofol infusion be limited to 24 to 48hrs. In a randomized trial, a strategy ofswitching to propofol when the antici-pated need for sedation was �24 hrs wasassociated with shorter weaning time andless agitation compared with continuingmidazolam (60).

The benzodiazepine midazolam has arapid onset of action but a more variable

awakening time. Midazolam is metabo-lized in the liver by oxidation, yielding anactive metabolite, alpha-hydroxymidazo-lam, which is excreted by the kidney.Patients with renal insufficiency mayhave markedly prolonged sedation (61),and we avoid using midazolam in thesepatients (2). Some experts recommendagainst using midazolam beyond 24 hrsof sedation (3) because of concerns aboutprolonged and variable recovery com-pared with lorazepam (62). However,other studies demonstrate 2- to 3-foldshorter recovery times with midazolamcompared with lorazepam (63). Loraz-epam has a more gradual onset of actionand is metabolized in the liver by glucu-ronidation to inactive metabolites withan intermediate duration half-life. Veryhigh-dose and prolonged infusions havebeen associated with a syndrome of hy-perosmolality, lactic acidosis, and revers-ible renal insufficiency (64). The imple-mentation of a protocol that focused onincreased utilization of lorazepam was as-sociated with lower costs without pro-longing weaning (65). Although loraz-epam is recommended for long-termsedation (3, 66), a 1998 survey indicatedthat midazolam and propofol are fre-quently used for �24 hrs (10). Interest-ingly, in a survey of 647 European ICUphysicians, midazolam was used often oralways by 63% of respondents, and propo-fol was used by 35% (67).

Fentanyl, morphine, and hydromor-phone are the most widely used analge-sics for ICU patients by bolus or contin-uous intravenous delivery, and there isno routinely preferred agent for long-term analgesia (3, 28). Fentanyl has arapid onset of action and the shortesthalf-life, but repeated dosing may causeaccumulation (3). Morphine has sloweronset and longer duration of action. It ismetabolized by glucuronidation to anumber of compounds, including mor-phine-6-glucuronide, which is more po-tent than the parent compound and ac-cumulates in renal insufficiency (68).Morphine is associated with histamine-related hypotension. We avoid adminis-tering morphine to patients with hemo-dynamic instability or renal insufficiency(2). Hydromorphone has an intermediateduration of action, is metabolized by glu-curonidation, and has no active metabo-lites. It is not associated with histaminerelease. Surveys indicate that morphineand fentanyl are used most widely (10,67).

Sedation and Analgesia Protocols.Strategies to optimally utilize sedativeand analgesic medications for mechani-cally ventilated patients include properdrug selection based on patient charac-teristics, titration to specific end pointsusing a sedation scale and goals of ther-apy, and utilizing protocols to avoid ac-cumulation of drugs through use of in-termittent therapy or scheduled cessationof medications. Several groups of investi-gators have introduced sedation proto-cols with varying levels of success. Afterimplementing a sedation scoring systemand sedation protocol, Brattebo et al. (26)were able to reduce duration of mechan-ical ventilation and ICU length of stay.Mascia et al. (25) developed and imple-mented a protocol for sedation, analgesia,and neuromuscular blockade for me-chanically ventilated ICU patients. Com-pared with the baseline period, protocolpatients had lower drug costs, ventilatortime, and ICU length of stay. Impres-sively, the use of NMBAs declined from30% of patients to 5%. MacLaren et al.(69) implemented an evidence-based se-dation and analgesia protocol, with ahigh protocol adherence rate, and notedmixed results when compared with thepre-protocol period. Patients experiencedless pain and discomfort and had lowerhourly cost of sedation, probably due toincreased lorazepam use. However, dura-tion of sedation use and ventilator wean-ing were each about 24 hrs longer withthe new protocol, offsetting the drug-related cost savings. Kollef et al. (22) ob-served that continuous intravenous seda-tion and analgesia was associated withlonger duration of mechanical ventilationafter adjusting for numerous co-vari-ables. This observation led to a prospec-tive trial at the same institution thatcompared nurse-led protocol-based man-agement with traditional management(24). This strategy emphasized conver-sion from continuous to intermittenttherapy using primarily lorazepam andfentanyl. The protocol patients had sig-nificantly shorter duration of mechanicalventilation, shorter ICU and hospitallength of stays, and one half the trache-ostomy rate of nonprotocol patients. It isnoteworthy that the protocol is complexand that the frequent dosing of intermit-tent therapy (as frequent as every 2 hrsbefore continuous infusion is initiated)might not be embraced by all nurses.

Kress et al. (23) tested scheduled dailyinterruption of sedative and analgesiccontinuous intravenous infusions. In this


approach, the medications are discontin-ued, and the patient is observed closelyuntil he or she is able to complete threeof the following four tasks: 1) open theeyes in response to voice, 2) use the eyesto follow the investigator on request, 3)squeeze a hand on request, or 4) stick outthe tongue on request, at which time theinfusions are restarted at one half thedose. The daily interruption protocol wasassociated with a significant reduction induration of mechanical ventilation,shorter ICU length of stay, and fewer di-agnostic studies to assess altered mentalstatus. There was no increase in seriousagitation such as self-removal of impor-tant tubes and catheters. Subsequentpost hoc analysis of the study populationshowed the daily interruption strategywas associated with fewer ICU-relatedcomplications (70). Concerns that theshock of daily withdrawal of all sedativeand analgesic medications might predis-pose the patient to posttraumatic stressdisorder (71) were quelled when the sameinvestigators demonstrated strong trendsfor fewer posttraumatic stress disordersymptoms and less psychological stressthan with conventional management(72). Patients in the original study (23)were also randomized to midazolam orpropofol (both groups also received mor-phine). Interestingly, daily interruptionof midazolam and morphine resulted inlower total doses compared with conven-tional care, yet there was no difference inthe total propofol or morphine with dailyinterruption in the propofol-infused pa-tients. Thus, the shorter duration of me-chanical ventilation is likely related toless drug accumulation among the mida-zolam-treated patients yet cannot explainthe benefit observed in the propofol-infusion patients. It may be that the dailyawakening event provides an opportunityfor ventilator weaning to be contem-plated at an earlier time than those with-out daily sedation interruption (73).

Extrapolation of these findings toHFOV-managed ARDS patients should bewith caution because the consequences ofsedation withdrawal with resulting pa-tient-ventilator dyssynchrony may bemore serious (21). Approximately 25% ofpatients in the daily interruption clinicaltrial (23) had ARDS and 12% were receiv-ing NMBAs, although none were receiv-ing HFOV. Application of a daily interrup-tion strategy or intermittent therapyprotocol during the conventional ventila-tion phase of patient management seemslogical. Clinical judgment must be exer-

cised during HFOV as to the safety ofapproaches that promote an awake or ag-itated state. It may be that a modifiedapproach whereby partial awakening isachieved, perhaps by targeting a lightersedation level, will safely reduce drug ac-cumulation and present an opportunityfor reducing ventilatory support.

Neuromuscular Blockade

Similar to sedation and analgesia,management of neuromuscular blockaderequires patient-based drug selection,identification of goals of therapy andmonitoring for safety and effectiveness,and titration of drugs to the lowest effec-tive dose for the shortest possible dura-tion using intermittent therapy or dailydrug interruption (18, 27, 74).

NMBAs. Among the various NMBAs,pancuronium, vecuronium, atracurium,and cis-atracurium are used most oftenfor ICU management of severe respiratoryfailure. Pancuronium is an amino-steroidagent with intermediate duration of ac-tion such that intermittent dosing is fea-sible. Primary limiting characteristics in-clude vasolytic and sympathomimeticproperties that can result in tachycardiaand hypertension. In addition, like vecu-ronium, it is metabolized to active com-pounds that accumulate in the setting ofrenal insufficiency. Vecuronium, also anamino-steroid, has a shorter duration ofaction and thus is used primarily via con-tinuous infusion. Atracurium is an inter-mediate acting benzylisoquinoliniumthat is eliminated through Hofmann deg-radation, which is independent of organfunction. Atracurium promotes hista-mine release that can lead to bronchos-pasm or hypotension. Cisatracurium isan isomer of atracurium with reducedhistamine releasing properties. Atra-curium and cisatracurium have fasterand more predictable recovery profilesthan vecuronium (75–77). Accordingly, ifrenal insufficiency or liver dysfunction ispresent, atracurium or cisatracuriumshould be selected. Otherwise, if inter-mittent therapy is desired, pancuroniumis used unless tachycardia or hyperten-sion would be poorly tolerated, in whichcase, vecuronium, atracurium, or cisatra-curium would be preferred (18, 27, 74).

Goals and Monitoring of Neuromus-cular Blockade. The clinical goal of neu-romuscular blockade should be estab-lished at onset of therapy and periodicallyreevaluated. In some cases, spontaneousrespiration with patient-ventilator syn-

chrony is sufficient; however, eliminationof spontaneous breaths may be necessaryto achieve adequate gas exchange. Moni-toring of the depth of neuromuscularblockade includes an assessment of effec-tiveness (i.e., are goals being met?). Sec-ond, some measure of the depth of block-ade is important to examine to reduce thelikelihood of drug or active metaboliteaccumulation and prolonged drug effect.The components of such monitoring in-clude a periodic clinical evaluation andperipheral nerve stimulation using“train-of-four” testing. Clinical evalua-tion may include observation of skeletalmuscle movement and respiratory effortand testing of deep tendon reflexes (78).Train-of-four testing involves applicationof an electrical current in four briefbursts 0.5 sec apart to leads positionedover the ulnar or facial nerve, with obser-vation of innervated muscle contraction(adductor pollicis or orbicularis oculimuscle, respectively) with each stimuli.In a 1998 survey, 91.8% of critical carephysicians reported utilizing train-of-four testing to evaluate depth of blockade(10). Most clinicians titrate NMBAs toachieve one to three twitches; however, alighter degree of paralysis (i.e., four offour twitches) is acceptable if clinicalgoals are being met. Train-of-four testingis principally a safety measure—the pres-ence of at least one twitch reduces thelikelihood that the patient is receiving anexcessive dose that might result in pro-longed neuromuscular blockade. Pro-longed neuromuscular blockade is a well-established complication of vecuroniuminfusion (79, 80), which is due in part toaccumulation of an active metabolite,3-desacetyl-vecuronium, in patients withimpaired renal function (81). Vecuro-nium clearance is poorly predictable withprolonged infusion (82), and train-of-fourtesting is particularly helpful in reducingthe prevalence of prolonged neuromus-cular blockade with vecuronium (83). Incontrast, drug clearance is more reliablewith atracurium and cisatracurium (76),and several prospective trials have failedto demonstrate benefit of train-of-fourtesting when compared with clinical as-sessment alone for patients treated withatracurium or cisatracurium (84, 85). Allpatients who receive NMBAs should re-ceive periodic clinical assessment andtrain-of-four testing every 4 hrs (18). Anargument can be made that patients whoreceive vecuronium or who undergo deepneuromuscular blockade will derive the


greatest benefit from scheduled train-of-four testing (74).

ICU-Acquired Myopathy. Althoughprolonged neuromuscular blockade isimportant, ICU-acquired myopathy is amore serious NMBA-related cause ofweakness in the ICU (86) because of thedelayed recovery that can require weeksof rehabilitation (87). ICU-acquired my-opathy, also known as acute quadriplegicmyopathy syndrome, among other terms,is most commonly associated with con-comitant prolonged corticosteroid ad-ministration and is likely due to a com-bination of myonecrosis (88) and reducedmyosin production (89). Unfortunately,ICU-acquired myopathy occurs despiteconscientious avoidance of drug overdoseby train-of-four monitoring. Originallydescribed with amino-steroid agents, ithas subsequently been described with allNMBAs, although the relative frequencyamong agents is not known (83, 90, 91).

Selected patients with severe ARDSand HFOV management may require con-tinuous neuromuscular blockade; how-ever, considerable effort should be madeto limit the duration of NMBA adminis-tration, particularly if corticosteroids aregiven concomitantly. In addition, al-though not well studied, the concept of adaily interruption of therapy or the use ofintermittent therapy a) helps avoid un-necessary drug accumulation, b) allowsperiodic evaluation of neurologic status,c) permits assessment of sedation andanalgesic needs, and d) promotes periodicreassessment of the need for further neu-romuscular blockade (74). An importantfactor, individualized for each patient andchanging over time, is the tolerance oftemporary cessation of paralysis.

CONCLUSIONS

A principle goal of implementation ofprotocols and guidelines is to consis-tently apply well-reasoned strategies toachieve more effective and streamlinedpatient care across multiple care provid-ers. Simpler protocols often work better.However, ventilatory and sedative man-agement of respiratory failure in severeARDS is complex, in part because of thephases of illness and recovery that pa-tients will proceed through. The conceptsdiscussed in this and other reviewsshould aid in implementing successfulstrategies.

REFERENCES

1. Sessler CN: Comfort and distress in the ICU:Scope of the problem. Semin Respir CritCare Med 2001; 22:111–113

2. Sessler CN, Grap MJ, Brophy GM: Multidis-ciplinary management of sedation and anal-gesia in critical care. Semin Respir Crit CareMed 2001; 22:211–225

3. Jacobi J, Fraser GL, Coursin DB, et al: Clin-ical practice guidelines for the sustained useof sedatives and analgesics in the critically illadult. Crit Care Med 2002; 30:119–141

4. Gainnier M, Roch A, Forel JM, et al: Effect ofneuromuscular blocking agents on gas ex-change in patients presenting with acute re-spiratory distress syndrome. Crit Care Med2004; 32:113–119

5. Hansen-Flaschen JH, Brazinsky S, Basile C,et al: Use of sedating drugs and neuromus-cular blocking agents in patients requiringmechanical ventilation for respiratory fail-ure: A national survey. JAMA 1991; 266:2870–2875

6. Christensen BV, Thunedborg LP: Use of sed-atives, analgesics and neuromuscular block-ing agents in Danish ICUs 1996/97: A na-tional survey. Intensive Care Med 1999; 25:186–191

7. Magarey JM: Sedation of adult critically illventilated patients in intensive care units: Anational survey. Aust Crit Care 1997; 10:90–93

8. Watling SM, Dasta JF, Seidl EC: Sedatives,analgesics, and paralytics in the ICU. AnnPharmacother 1997; 31:148–153

9. Dasta JF, Fuhrman TM, McCandles C: Pat-terns of prescribing and administering drugsfor agitation and pain in patients in a surgi-cal intensive care unit. Crit Care Med 1994;22:974–980

10. Rhoney DH, Murry KR: National survey ofthe use of sedating drugs, neuromuscularblocking agents, and reversal agents in theintensive care unit. J Intensive Care Med2003; 18:139–145

11. Esteban A, Anzueto A, Frutos F, et al: Char-acteristics and outcomes in adult patientsreceiving mechanical ventilation: A 28-dayinternational study. JAMA 2002; 287:345–355

12. Alfa I, Arroliga AC, Frutos-Vivar F, et al: Useof sedatives in mechanically ventilated pa-tients. Am J Respir Crit Care Med 2003; 167:A969


14. Gottlieb JE, Park P, Girod A, et al: Compar-ison of neuromuscular blocker and sedativeuse among patients with ARDS treated withlow versus high tidal volume. Am J RespirCrit Care Med 2000; 161:A506

15. Derdak S, Mehta S, Stewart TE, et al: High-frequency oscillatory ventilation for acute re-

spiratory distress syndrome in adults: A ran-domized, controlled trial. Am J Respir CritCare Med 2002; 166:801–808



18. Murray MJ, Cowen J, DeBlock H, et al: Clin-ical practice guidelines for sustained neuro-muscular blockade in the adult critically illpatient. Crit Care Med 2002; 30:142–156

19. Arroliga AC, Frutos-Vivar F, Alfa I, et al: Useof neuromuscular blockers (NMBs) in me-chanically ventilated patients. Am J RespirCrit Care Med 2003; 167:A970

20. Kaplan LJ, Bailey H, Formosa V: Airway pres-sure release ventilation increases cardiacperformance in patients with acute lung in-jury/adult respiratory distress syndrome. CritCare 2001; 5:221–226


22. Kollef MH, Levy NT, Ahrens TS, et al: The useof continuous i.v. sedation is associated withprolongation of mechanical ventilation.Chest 1998; 114:541–548

23. Kress JP, Pohlman AS, O’Connor MF, et al:Daily interruption of sedative infusions incritically ill patients undergoing mechanicalventilation. N Engl J Med 2000; 342:1471–1477

24. Brook AD, Ahrens TS, Schaiff R, et al: Effectof a nursing-implemented sedation protocolon the duration of mechanical ventilation.Crit Care Med 1999; 27:2609–2615

25. Mascia MF, Koch M, Medicis JJ: Pharmaco-economic impact of rational use guidelineson the provision of analgesia, sedation, andneuromuscular blockade in critical care. CritCare Med 2000; 28:2300–2306

26. Brattebo G, Hofoss D, Flaatten H, et al: Effectof a scoring system and protocol for sedationon duration of patients’ need for ventilatorsupport in a surgical intensive care unit. BMJ2002; 324:1386–1389

27. Gehr LC, Sessler CN: Neuromuscular block-ade in the intensive care unit. Semin RespirCrit Care Med 2001; 22:175–188

28. Vender JS, Szokol JW, Murphy GS, et al:Sedation, analgesia, and neuromuscularblockade in sepsis: An evidence-based review.Crit Care Med 2004; 32:S554–S561

29. Salam A, Tilluckdharry L, Amoateng-Adjepong Y, et al: Neurologic status, cough,secretions and extubation outcomes. Inten-sive Care Med 2004; 30:1334–1339

30. Grap MJ, Strickland D, Tormey L, et al: Col-laborative practice: Development, implemen-tation, and evaluation of a weaning protocolfor patients receiving mechanical ventila-tion. Am J Crit Care 2003; 12:454–460


31. Krishnan JA, Moore D, Robeson C, et al: Aprospective, controlled trial of a protocol-based strategy to discontinue mechanicalventilation. Am J Respir Crit Care Med 2004;169:673–678

32. Ely EW, Baker AM, Dunagan DP, et al: Effecton the duration of mechanical ventilation ofidentifying patients capable of breathingspontaneously. N Engl J Med 1996; 335:1864–1869

33. Sanders KD, McArdle P, Lang JD: Pain in theICU: Recognition, measurement, manage-ment. Semin Respir Crit Care Med 2001;22:127–135

34. Frazier SK, Moser DK, Riegel B, et al: Criticalcare nurses’ assessment of patients’ anxiety:Reliance on physiological and behavioral pa-rameters. Am J Crit Care 2002; 11:57–64

35. Sessler CN: Sedation scales in the ICU. Chest2004; 126:1727–1730

36. De Jonghe B, Cook D, Appere-De-Vecchi C, etal: Using and understanding sedation scoringsystems: A systematic review. Intensive CareMed 2000; 26:275–285

37. Riker RR, Fraser GL: Monitoring sedation,agitation, analgesia, neuromuscular block-ade, and delirium in adult ICU patients. Se-min Respir Crit Care Med 2001; 22:189–198

38. Ramsay MA, Savege TM, Simpson BR, et al:Controlled sedation with alphaxalone-alphadolone. BMJ 1974; 2:656–659

39. Riker RR, Picard JT, Fraser GL: Prospectiveevaluation of the Sedation-Agitation Scalefor adult critically ill patients. Crit Care Med1999; 27:1325–1329

40. Devlin JW, Boleski G, Mlynarek M, et al:Motor Activity Assessment Scale: A valid andreliable sedation scale for use with mechan-ically ventilated patients in an adult surgicalintensive care unit. Crit Care Med 1999; 27:1271–1275

41. de Lemos J, Tweeddale M, Chittock D: Mea-suring quality of sedation in adult mechani-cally ventilated critically ill patients: TheVancouver Interaction and Calmness Scale.J Clin Epidemiol 2000; 53:908–919

42. Sessler CN, Gosnell MS, Grap MJ, et al: TheRichmond Agitation-Sedation Scale: Validityand reliability in adult intensive care unitpatients. Am J Respir Crit Care Med 2002;166:1338–1344

43. De Jonghe B, Cook D, Griffith L, et al: Adap-tation to the Intensive Care Environment(ATICE): Development and validation of anew sedation assessment instrument. CritCare Med 2003; 31:2344–2354

44. Weinert C, McFarland L: The state of intu-bated ICU patients: Development of a two-dimensional sedation rating scale for criti-cally ill adults. Chest 2004; 126:1883–1890

45. Costa J, Cabre L, Molina R, et al: Cost of ICUsedation: Comparison of empirical and con-trolled sedation methods. Clin IntensiveCare 1994; 5:17–21

46. Riker RR, Fraser GL: Sedation in the inten-sive care unit: Refining the models and de-fining the questions. Crit Care Med 2002;30:1661–1663

47. Arbour R: Using bispectral index monitoringto detect potential breakthrough awarenessand limit duration of neuromuscular block-ade. Am J Crit Care 2004; 13:66–73

48. Nasraway SS Jr, Wu EC, Kelleher RM, et al:How reliable is the Bispectral Index in criti-cally ill patients? A prospective, comparative,single-blinded observer study. Crit Care Med2002; 30:1483–1487

49. Vivien B, Di Maria S, Ouattara A, et al: Over-estimation of Bispectral Index in sedated in-tensive care unit patients revealed by admin-istration of muscle relaxant. Anesthesiology2003; 99:9–17

50. Dasta JF, Kane SL, Gerlach AT, et al: Bispec-tral Index in the intensive care setting. CritCare Med 2003; 31:998

51. Deogaonkar A, Gupta R, Degeorgia M, et al:Bispectral Index monitoring correlates withsedation scales in brain-injured patients. CritCare Med 2004; 32:2403–2406

52. Schneider G, Heglmeier S, Schneider J, et al:Patient State Index (PSI) measures depth ofsedation in intensive care patients. IntensiveCare Med 2004; 30:213–216

53. Watson BD, Kane-Gill SL: Sedation assess-ment in critically ill adults: 2001–2004 up-date. Ann Pharmacother 2004; 38:1898–1906

54. Schmidt GN, Bischoff P, Standl T, et al: Com-parative evaluation of the Datex-Ohmeda S/5Entropy Module and the Bispectral Indexmonitor during propofol-remifentanil anes-thesia. Anesthesiology 2004; 101:1283–1290

55. Grap MJ, Borchers CT, Munro CL, et al: Ac-tigraphy in the critically ill: Correlation withactivity, agitation, and sedation. Am J CritCare 2005; 14:52–60

56. Coursin DB, Maccioli GA: Dexmedetomidine.Curr Opin Crit Care 2001; 7:221–226

57. McKeage K, Perry CM: Propofol: A review ofits use in intensive care sedation of adults.CNS Drugs 2003; 17:235–272

58. Anis AH, Wang XH, Leon H, et al: Economicevaluation of propofol for sedation of patientsadmitted to intensive care units. Anesthesi-ology 2002; 96:196–201

59. Carrasco G, Molina R, Costa J, et al: Propofolvs midazolam in short-, medium-, and long-term sedation of critically ill patients: A cost-benefit analysis. Chest 1993; 103:557–564

60. Saito M, Terao Y, Fukusaki M, et al: Sequen-tial use of midazolam and propofol for long-term sedation in postoperative mechanicallyventilated patients. Anesth Analg 2003; 96:834–838

61. Bauer TM, Ritz R, Haberthur C, et al: Pro-longed sedation due to accumulation of con-jugated metabolites of midazolam. Lancet1995; 346:145–147

62. Pohlman AS, Simpson KP, Hall JB: Contin-uous intravenous infusions of lorazepam ver-sus midazolam for sedation during mechan-ical ventilatory support: A prospective,randomized study. Crit Care Med 1994; 22:1241–1247

63. Barr J, Zomorodi K, Bertaccini EJ, et al: Adouble-blind, randomized comparison of i.v.

lorazepam versus midazolam for sedation ofICU patients via a pharmacologic model. An-esthesiology 2001; 95:286–298

64. Yaucher NE, Fish JT, Smith HW, et al: Pro-pylene glycol-associated renal toxicity fromlorazepam infusion. Pharmacotherapy 2003;23:1094–1099

65. Devlin JW, Holbrook AM, Fuller HD: Theeffect of ICU sedation guidelines and phar-macist interventions on clinical outcomesand drug cost. Ann Pharmacother 1997; 31:689–695

66. Shapiro BA, Warren J, Egol AB, et al: Practiceparameters for intravenous analgesia and se-dation for adult patients in the intensive careunit: An executive summary. Society of Crit-ical Care Medicine. Crit Care Med 1995; 23:1596–1600

67. Soliman HM, Melot C, Vincent JL: Sedativeand analgesic practice in the intensive careunit: The results of a European survey. Br JAnaesth 2001; 87:186–192

68. Bion JF, Logan BK, Newman PM, et al: Se-dation in intensive care: Morphine and renalfunction. Intensive Care Med 1986; 12:359–365

69. MacLaren R, Plamondon JM, Ramsay KB, etal: A prospective evaluation of empiric versusprotocol-based sedation and analgesia. Phar-macotherapy 2000; 20:662–672

70. Schweickert WD, Gehlbach BK, Pohlman AS,et al: Daily interruption of sedative infusionsand complications of critical illness in me-chanically ventilated patients. Crit Care Med2004; 32:1272–1276

71. Heffner JE: A wake-up call in the intensivecare unit. N Engl J Med 2000; 342:1520–1522

72. Kress JP, Gehlbach B, Lacy M, et al: Thelong-term psychological effects of daily sed-ative interruption on critically ill patients.Am J Respir Crit Care Med 2003; 168:1457–1461

73. Sessler CN: Wake up and breathe. Crit CareMed 2004; 32:1413–1414

74. Sessler CN: Train-of-four to monitor neuro-muscular blockade? Chest 2004; 126:1018–1022

75. Prielipp RC, Jackson MJ, Coursin DB: Com-parison of the neuromuscular recovery afterparalysis with atracurium versus vecuro-nium in an ICU patient with renal insuffi-ciency. Anesth Analg 1994; 78:775–778

76. Pearson AJ, Harper NJ, Pollard BJ: The infu-sion requirements and recovery characteris-tics of cisatracurium or atracurium in inten-sive care patients. Intensive Care Med 1996;22:694–698

77. Newman PJ, Quinn AC, Grounds RM, et al: Acomparison of cisatracurium (51W89) andatracurium by infusion in critically ill pa-tients. Crit Care Med 1997; 25:1139–1142

78. Tavernier B, Rannou JJ, Vallet B: Peripheralnerve stimulation and clinical assessment fordosing of neuromuscular blocking agents incritically ill patients. Crit Care Med 1998;26:804–805

79. Segredo V, Caldwell JE, Matthay MA, et al:


Persistent paralysis in critically ill patientsafter long-term administration of vecuro-nium. N Engl J Med 1992; 327:524–528

80. Watling SM, Dasta JF: Prolonged paralysis inintensive care unit patients after the use ofneuromuscular blocking agents: A review ofthe literature. Crit Care Med 1994; 22:884–893

81. Caldwell JE, Szenohradszky J, Segredo V, etal: The pharmacodynamics and pharmacoki-netics of the metabolite 3- desacetylvecuro-nium (ORG 7268) and its parent compound,vecuronium, in human volunteers. J Phar-macol Exp Ther 1994; 270:1216–1222

82. Segredo V, Caldwell JE, Wright PM, et al: Dothe pharmacokinetics of vecuronium changeduring prolonged administration in criticallyill patients? Br J Anaesth 1998; 80:715–719

83. Rudis MI, Sikora CA, Angus E, et al: A pro-spective, randomized, controlled evaluation

of peripheral nerve stimulation versus stan-dard clinical dosing of neuromuscular block-ing agents in critically ill patients. Crit CareMed 1997; 25:575–583

84. Strange C, Vaughan L, Franklin C, et al: Com-parison of train-of-four and best clinical assess-ment during continuous paralysis. Am J RespirCrit Care Med 1997; 156:1556–1561

85. Baumann MH, McAlpin BW, Brown K, et al:A prospective randomized comparison oftrain-of-four monitoring and clinical assess-ment during continuous ICU cisatracuriumparalysis. Chest 2004; 126:1267–1273

86. Deem S, Lee CM, Curtis JR: Acquired neuro-muscular disorders in the intensive careunit. Am J Respir Crit Care Med 2003; 168:735–739

87. Rudis MI, Guslits BJ, Peterson EL, et al:Economic impact of prolonged motor weak-ness complicating neuromuscular blockade

in the intensive care unit. Crit Care Med1996; 24:1749–1756

88. Fischer JR, Baer RK: Acute myopathy asso-ciated with combined use of corticosteroidsand neuromuscular blocking agents. AnnPharmacother 1996; 30:1437–1445

89. Larsson L, Li X, Edstrom L, et al: Acutequadriplegia and loss of muscle myosin inpatients treated with nondepolarizing neuro-muscular blocking agents and corticoste-roids: Mechanisms at the cellular and molec-ular levels. Crit Care Med 2000; 28:34–45

90. Prielipp RC, Coursin DB, Scuderi PE, et al:Comparison of the infusion requirementsand recovery profiles of vecuronium and ci-satracurium 51W89 in intensive care unitpatients. Anesth Analg 1995; 81:3–12

91. Marik PE: Doxacurium-corticosteroid acutemyopathy: Another piece to the puzzle. CritCare Med 1996; 24:1266–1267


Prognostic determinants of acute respiratory distress syndrome inadults: Impact on clinical trial design

Lorraine B. Ware, MD

Acute lung injury (ALI) andacute respiratory distress syn-drome (ARDS) are majorcauses of acute respiratory fail-

ure that are associated with high morbid-ity and mortality (1). There is a pressingneed for the development and clinicaltesting of new therapies such as high-frequency oscillatory ventilation thatmight improve clinical outcomes in ALIand ARDS. However, the design of clini-cal trials for this complex and heteroge-neous syndrome is not straightforward.Selection of patients for enrollment inclinical trials at higher risk for adverseclinical outcomes using clinical or bio-logic predictors is one strategy to select amore severely ill group of patients whomay be more likely to benefit from new

clinical interventions. This article brieflydiscusses some of the clinical features ofALI/ARDS that are relevant to the designof clinical trials for ALI/ARDS. Then theclinical characteristics and biologicmarkers that can be used to identify pa-tients at higher risk of adverse clinicaloutcomes are discussed, including strat-egies to identify patients later in thecourse of ALI/ARDS who are failing toimprove with conventional therapy.

Acute Lung Injury: Magnitude ofthe Clinical Problem

The exact incidence of ARDS is diffi-cult to measure, in part because of thelack of a clinical diagnostic test and inpart because ARDS is largely undiagnosed(2). A recent study that used the enroll-ment logs from the National Heart Lungand Blood Institute’s (NHLBI) ARDS Net-work (3) estimated a U.S. incidence ashigh as 64 cases per 100,000 population,or 150,000 per year. Overall, approxi-mately 7% of patients admitted to theintensive care unit will develop ALI orARDS (4), and among mechanically ven-

tilated patients with acute respiratoryfailure, the incidence of ALI and ARDS is11% to 23% (4). Thus, ALI/ARDS is amajor public health problem encoun-tered frequently by all physicians whocare for critically ill patients. Currently,the only therapy that has been proven tobe effective at reducing mortality in ALI/ARDS in a large randomized, multicentercontrolled trial (5) is a protective venti-latory strategy.

In and of itself, the diagnosis of ALI orARDS portends a poor prognosis. Themajority of recent studies report mortal-ity in the 35% to 60% range when allpatients who meet the American Euro-pean Consensus Conference definitions(6) are included (4). For example, in therecently published ALIVE study (7), a co-hort study of 6,522 patients admitted to78 intensive care units in ten Europeancountries over a 2-month period in 1999,the hospital mortality was 32.7% in pa-tients with ALI and 57.9% in patientswith ARDS. Despite these discouragingnumbers, there is some evidence thatwhen single centers are considered, mor-tality from ALI/ARDS has declined over

From Allergy, Pulmonary and Critical Care Medi-cine, Department of Medicine, Vanderbilt University,Nashville, TN.

Supported by National Institutes of Health grantHL70521.


DOI: 10.1097/01.CCM.0000155788.39101.7E

Objective: The objective of this study was to review knownclinical predictors and biologic markers of adverse clinical out-comes in acute lung injury (ALI) and acute respiratory distresssyndrome (ARDS) that might be used as selection criteria inclinical trials of novel therapies for ALI/ARDS.

Data Source: Published studies on clinical predictors andbiologic markers of adverse outcomes in ALI/ARDS.

Main Results: In large epidemiologic studies, a number ofclinical factors have been identified consistently as independentpredictors of mortality in ALI/ARDS. These include age, comor-bidities, including chronic liver disease and immunosuppression,severity of illness scores, and the degree of multisystem organfailure. Several biologic markers of mortality have also beenidentified in large studies, including von Willebrand factor anti-gen, surfactant protein D, protein C, plasminogen activator inhib-itor-1, interleukins 6 and 8, and the TNF receptors. The PaO2/FIO2

ratio at the onset of ALI/ARDS does not predict clinical outcomebut may be more useful after the first day of ALI/ARDS. A persis-

tently low PaO2/FIO2 ratio is associated with worse outcomes andmay be a marker of failure to respond to conventional therapy.Changes in IL-6, IL-8, TNF receptors, and SP-D over the first 3days of ALI/ARDS are also associated with adverse clinical out-comes. The use of a combination of clinical factors and biologicmarkers is a promising strategy that needs to be prospectivelyvalidated.

Conclusions: The design of clinical trials for new therapies for ALIand ARDS is a complex problem that ultimately will have a majorimpact on both trial outcome and generalizability. A number ofclinical factors and biologic markers can be used to differentiategroups of patients at highest risk for adverse clinical outcomes.Whether enriching study populations with these sicker patients willincrease or decrease the likelihood of a treatment effect for a giventherapy is unknown. (Crit Care Med 2005; 33[Suppl.]:S217–S222)

KEY WORDS: acute lung injury; acute respiratory distress syn-drome; clinical trial; mortality; sepsis; trauma; mechanical venti-lation; high-frequency oscillatory ventilation; epidemiology


time (8, 9). For example, at the Universityof California, San Francisco, mortality ina cohort of patients with ALI/ARDS stud-ied in 1991–1992 was 58% (10) comparedwith 42% in a similarly selected cohort ofpatients with ALI/ARDS studied at thesame center from 1998–2000 (11). Bothcohorts were identified using the Ameri-can European Consensus Conference def-initions (6). In addition to the high mor-tality, survivors of ALI/ARDS havesignificant morbidity and long-term dis-ability (12–15). The high incidence com-bined with the substantial morbidity andmortality emphasizes the pressing needfor development and testing of new treat-ments.

Clinical Trial Design in AcuteLung Injury/Acute RespiratoryDistress Syndrome

The design of clinical trials for newtherapies for ALI/ARDS has a major im-pact on both trial outcome and general-izability (16). Unfortunately, the optimalclinical trial design for new therapies ofARDS is not straightforward (17). Al-though it is beyond the scope of thisreview to discuss all of the importantissues that impact clinical trial design inARDS (see reference (16) for a review),several are relevant to a discussion ofprognostic indicators. For a therapy to begeneralizable, it is preferable to includethe majority of patients with ALI/ARDSregardless of prognostic indicators. Thistype of study design has the advantage ofrelative ease of enrollment of large num-bers of patients compared with more re-strictive enrollment criteria. However,

there may be some advantage to limitingenrollment to a subset of patients. Forexample, as the predicted mortality in thecontrol group declines, a larger samplesize will be required to demonstrate aclinically and statistically significant ef-fect of a treatment on mortality. To re-duce the study size needed to demon-strate a mortality benefit, the studypopulation could be limited to patients athigher risk of dying. This approach hasseveral pitfalls, however. First, althoughwe can identify patients at higher risk ofdying based on a variety of clinical factors(see subsequently), there is no guaranteethat the therapy that is being tested willbe effective in this more severely ill sub-group. In fact, this group might not beresponsive to treatment in the same waythat a larger unselected group might be(16). Second, as we restrict the patientgroup to patients more likely to die, themortality that is actually attributable toALI/ARDS may fall while the attributablemortality from other causes rises.

A second reason for attempting to se-lect a subset of patients for clinical trialsis to identify patients who are more likelyto be treatment-responsive. This ap-proach may be useful for novel therapiessuch as high-frequency oscillatory venti-lation that are cumbersome and unfamil-iar to the majority of clinicians becauseenthusiasm for enrolling lower-risk pa-tients into clinical trials of these thera-pies may be low. However, the factorsthat predict treatment responsiveness arelargely unknown and are not necessarilyintuitive. For example, an improvementin PaO2/FIO2 ratio after institution of a

therapy might intuitively be expected tobe an indicator of treatment responsive-ness that would be associated with betterclinical outcomes. This has not been thecase, however, in several large clinicaltrials. For example, both prone position-ing and inhaled nitric oxide can producerapid improvements in oxygenation insome patients with ALI/ARDS, yet neithertreatment has a demonstrable impact onclinical outcome (18, 19). Low tidal vol-ume ventilation actually worsens oxygen-ation during the first few days of therapycompared with a higher tidal volume, yetis associated with better clinical out-comes (5). The remainder of this articleconsiders the known clinical and biologicmarkers that predict outcome both earlyand later in the course of ALI/ARDS thatmight be used to select patients for en-rollment in trials of novel therapies.

Clinical Factors That PredictOutcome at the Onset of AcuteLung Injury/Acute RespiratoryDistress Syndrome

Although the clinical factors that pre-dict outcome at the onset of ALI/ARDShave varied substantially from study tostudy (7, 10, 11, 20–25), there are severalfactors that are consistently predictive ofmortality in a number of large epidemi-ologic studies (Table 1). The largest studyto date (7) was done over a 2-month in-ception period in 78 intensive care unitsin ten European countries. In that studyof 463 patients with ALI or ARDS, byAmerican European Consensus Confer-ence definitions, age, immunosuppres-sion, barotrauma, organ dysfunction,

Table 1. Summary of independent predictors of mortality in epidemiologic studies of acute lung injury and acute respiratory distress syndrome

Author Year Country No. Patients Independent Predictors of Mortality

Doyle (10) 1995 USA 123 AECC ARDS Nonpulmonary organ failure, chronic liver disease,sepsis

Zilberberg (20) 1998 USA 107 AECC ARDS Age �65 yrs, organ transplant, HIV, cirrhosis,malignancy, sepsis

Monchi (21) 1998 France 259 AECC ARDS SAPS II, McCabe score, cirrhosis, length of mechanicalventilation before ARDS, oxygenation index

Luhr (22) 1999 Scandinavia 287 AECC ARDS Age, chronic liver disease, P/F ratio less than 100 mm HgRocco (23) 2001 USA 111 AECC ARDS and LIS �2.50 Age, MODS �8, Lung Injury Score �2.76Nuckton (11) 2002 USA 179 AECC ARDS SAPS II, respiratory system compliance, dead-space

fractionEstenssoro (24) 2002 Argentina 217 ARDS alive at 24 hrs Severe comorbidities, PaO2/FIO2 on day 3, SOFA score

day 3Venet (25) 2003 France 125 ARDS alive at 7 days SAPS II, McCabe score, use of prone positioningBrun-Buisson (7) 2004 Europe 463 AECC ALI or ARDS Age, immunosuppression, barotraumas, organ

dysfunction, SAPS II, pH �7.30

AECC, American European Consensus Conference definition; HIV, human immunodeficiency virus infection; SAPS II, Simplified Acute PhysiologyScore II; LIS, Lung Injury Score; MODS, Multi-Organ Dysfunction Score; SOFA, Sequential Organ Failure Assessment.


Simplified Acute Physiology Score II(SAPS II), and a pH �7.30 were all asso-ciated with higher mortality. Of these,age, organ dysfunction, immunosuppres-sion, and SAPS II have also been identi-fied in other, smaller studies. Sepsis,chronic liver disease, and other comorbidillnesses also were independent predic-tors of mortality in several studies. Al-though it is difficult to generalize fromthese diverse studies that were done ondifferent continents with different meth-odologies, the repeated identification ofage, organ dysfunction, comorbid dis-eases, including chronic liver disease, andseverity of illness as measured by scoringsystems such as SAPS II or APACHE iden-tifies these factors as important predic-tors of mortality across disparate popula-tions of patients with ALI/ARDS. Whetherthese factors also consistently predictother clinically relevant outcomes suchas ventilator-free days is less clear. Al-though trauma patients were not univer-sally included in these epidemiologicstudies, the majority of data suggests thatALI/ARDS that results from multipletrauma has a better outcome that ALI/ARDS from other etiologies (8, 26).

It is important to note that oxygen-ation at the onset of ALI/ARDS as mea-sured by the PaO2/FIO2 ratio has not beenan independent predictor of mortality instudies of adults with ALI/ARDS (7, 10,11, 20–22, 24, 25, 27–31). Table 2 sum-marizes a number of epidemiologic stud-ies that have examined the PaO2/FIO2 ratioand outcomes of ALI and ARDS. Althoughthe PaO2/FIO2 has been predictive of clin-ical outcomes in some studies in univar-

iate analysis, it has not been a predictorin multivariate analysis. The PaO2/FIO2 ra-tio also was not predictive in a meta-analysis of 101 prior studies of ALI andARDS that was published in 1996 (32).Likewise, the distinction between ALI andARDS has not consistently distinguishedbetween groups with lower and highermortality. One possible explanation forthe lack of discriminatory power of thePaO2/FIO2 ratio is that many of the mul-tivariate analyses included organ dys-function scores such as SAPS II that in-clude a measurement of PaO2/FIO2 ratio. Asecond possibility is that the PaO2/FIO2 ishighly variable depending on the ventila-tory strategy chosen. This explanation issupported by the fact that the initial PaO2/FIO2 ratio was predictive of outcome inseveral large clinical trials that mandatedor recommended a protocolized ventila-tory strategy (5, 33) (and see subse-quently). In addition, several investiga-tors have shown in small studies that thePaO2/FIO2 in patients with ARDS can bemodified greatly be the application ofeven small levels of positive end-expira-tory pressure (34, 35). Another explana-tion is that the degree of respiratory fail-ure is not the primary determinant ofoutcome in patients with ALI/ARDS,rather the degree of nonpulmonary organfailure is the primary determinant. Thus,strategies for selecting the most severelyill patients with ALI/ARDS should not fo-cus solely on oxygenation as a primarydeterminant of severity of illness. Like-wise, new therapies for ARDS may needto target multiorgan dysfunction as well

as pulmonary dysfunction to be effective(17).

Although less well studied than adultALI/ARDS, predictors of outcome in pe-diatric ALI/ARDS are probably somewhatdifferent from in adults in part becausepediatric ALI/ARDS has different epide-miologic characteristics than adult ALI/ARDS. In a recently completed prospec-tive observational study of 328 childrenwho met the American European Con-sensus Conference Definitions of ALI/ARDS, the mortality was only 22%. Themost common cause of pediatric ALI/ARDS is pneumonia. Perhaps for this rea-son, mortality in children is governedprimarily by the severity of lung injury asmeasured by the oxygenation deficitrather than comorbidities. Thus, in con-trast to adults, the PaO2/FIO2 ratio is astrong independent predictor of mortalityin children (36). Other factors that wereassociated with mortality and prolongedmechanical ventilation from ARDS in aprospective cohort of 328 pediatric pa-tients were the presence of nonpulmo-nary and noncentral nervous system(central nervous system) organ dysfunc-tion and the presence of central nervoussystem dysfunction (36).

In addition to examining prognosticindicators from large epidemiologic stud-ies, it is important to examine clinicalprognosticators in patients who actuallywere enrolled in clinical trials. To carryout a risk-adjusted analysis on patientsenrolled in the recently completed ALVE-OLI study of two different levels of posi-tive end-expiratory pressure in ARDS(37), the NHLBI ARDS network identified

Table 2. Association of oxygenation at study enrollment with mortality in clinical studies in acute lung injury and acute respiratory distress syndrome

Author Year

Association of Initial PaO2/FIO2

Ratio with Death

Patients ReferenceUnivariate Multivariate

Bonea 1989 No Not done ARDS with PaO2/FIO2 �150 or �200 with PEEP 27Sloanea 1992 No No ALI with PaO2/FIO2 ratio �250 28Knausa 1994 No No ARDS by PaO2/FIO2 ratio and ICD-9 diagnosis 29Doyle 1995 No No ARDS 10Zilberberg 1998 No No ALI and ARDS 20Monchi 1998 No No ARDS 21Luhr 1999 No No ALI and ARDS 22Luhr 2000 No No ARDS 30Estenssoro 2002 Yes No ARDS 24Nuckton 2002 Yes No ARDS 11Bersten 2002 Yes Not done ALI and ARDS 31Venet 2003 No No ARDS 25Brun-Buisson 2004 Yes No ALI and ARDS 7

ARDS, acute respiratory distress syndrome; PEEP, positive end-expiratory pressure; ALI, acute lung injury; ICD-9, International Classification ofDiseases, 9th Revision.

aDid not use American European Consensus Conference definitions. All others defined ALI and/or ARDS using these definitions.


the clinical variables associated withmortality in their prior 861 patient trialof a protective ventilatory strategy (5).The baseline variables that were associ-ated with a significantly higher risk ofdeath were similar to those found in thelarge epidemiologic studies and includedage, sepsis, number of organ failures, el-evated APACHE III score, the alveolar–arterial oxygen difference, and the pla-teau airway pressure (38). This trialexcluded patients with other importantrisk factors for mortality, including im-munosuppression and chronic liver dis-ease. In another large dataset, 448 pa-tients in a North American, European,and South African study of surfactant re-placement therapy in ARDS had indepen-dent predictors of mortality of age,APACHE II score, and baseline PaO2/FIO2

ratio (33).

Biologic Factors That PredictOutcome at the Onset of AcuteLung Injury/Acute RespiratoryDistress Syndrome

For many major lung diseases such asasthma and chronic obstructive pulmo-nary disease, there is not yet a singlebiomarker that is of potential clinicalvalue. By contrast, there is a large body ofresearch on biomarkers in ARDS reflect-ing many facets of the complex patho-physiology (1, 39–41). For assessing theseverity and outcome of ARDS, a numberof biomarkers may have some clinicaluse. For example, elevated plasma levelsof the endothelial marker von Willebrandfactor antigen had an independent asso-ciation with hospital mortality whenmeasured early in ARDS in a single cen-ter study (42). Alterations in circulatingcoagulation and fibrinolytic proteins arealso associated with adverse clinical out-

comes in ARDS (43, 44). Others haveshown the prognostic value of interleu-kin-8 (45), hepatocyte growth factor (46),transforming growth factor-� (47, 48)and many others (49–54).

Until recently, however, none of thesebiologic markers have been studied inlarge multicenter populations of patientswith ALI and ARDS. In collaboration withthe National Institutes of Health (NIH)ARDS Network, we have recently com-pleted studies of a large number of bio-logic markers in the patients enrolled inthe NIH ARDS Network trial of a protec-tive ventilatory strategy (55–60). As sum-marized in Table 3, there were a variety ofbiologic markers that were associatedwith mortality, including levels of cyto-kines, markers of endothelial and alveolarepithelial injury, and abnormal coagula-tion and fibrinolysis. Use of one or moreof these biologic markers to select agroup of patients at higher risk of adverseclinical outcome could be used to restrictor stratify enrollment in future clinicaltrials. However, implementation of thisstrategy will require further prospectivevalidation of these markers and the de-velopment of readily available rapid clin-ical tests for these markers. In addition,development of a prognostic index thatcombines clinical factors and biologicmarkers may be useful.

Identifying Patients Who AreFailing Conventional Therapy

The precise role for novel ventilatortherapies such as high-frequency oscilla-tory ventilation remains to be delineated.It may be that the primary role will be asa rescue therapy for patients who arefailing to improve with conventional lungprotective strategies. Both clinical factorsand biologic markers may be useful to

identify patients who are failing conven-tional therapy. Although the initial PaO2/FIO2 ratio is not a reliable predictor ofclinical outcome, changes in PaO2/FIO2

ratio over the first week may be quiteuseful in segregating patients at high riskfor adverse outcomes (27, 28, 61, 62).Similarly, the evolution of organ failuresover the first few days may also be a goodpredictor of adverse clinical outcomes.Pulmonary dead space fraction is easy tomeasure noninvasively and is indepen-dently associated with adverse clinicaloutcomes at the onset of ALI/ARDS (11).Recently, pulmonary dead space fractionhas been measured sequentially over thefirst 6 days of ARDS in 59 patients (63).Persistence of a dead space fraction of�0.60 was highly associated with mortal-ity and may be an excellent marker ofpersistent pulmonary organ dysfunction.

Changes in plasma biomarker levelsmay also be indicative of failure to re-spond to a protective ventilatory strategy.In collaboration with the NIH ARDS Net-work, we have measured a number ofbiomarkers on day 0 and day 3 after ran-domization to 6 mL/kg vs 12 mL/kg pre-dicted body weight tidal volume. Severalbiomarkers were clearly modulated by aprotective ventilator strategy, includingSP-D (60), IL-6, IL-8 (57), IL-10, sTNFRI(58), and protein C (55). For example, thecytokines IL-6, IL-8, and IL-10 all werereduced in the plasma by day 3 to agreater extent in the 6-mL/kg group thanin the 12-mL/kg group, a finding that hasalso been observed by others (64). Unlikethe cytokines, SP-D rose in all patientsfrom day 0 to day 3. However, the rise inplasma SP-D levels was significantly lessin the 6-mL/kg group than in the 12-mL/kg group (60). These findings suggestthat there are multiple biomarkers thatcan be used to measure the lung and

Table 3. Plasma biologic markers associated with death in patients enrolled in the Acute Respiratory Distress Syndrome Network trial of 6 vs 12 mL/kgtidal volume

Study Design Plasma MarkersNo. of

Patients Results Reference

Acute inflammation IL-6, IL-8 781 IL-6 and IL-8 predictive of death 57Endothelial injury VWF 559 VWF predictive of death 59Epithelial type II cell molecules SP-D 565 SP-D predictive of death 60Adhesion molecule ICAM-1 565 ICAM-1 predictive of death Ware et al.

unpublished dataNeutrophil-endothelial interaction sTNFRI/II 562 sTNFRI/II predictive of death 58Procoagulant activity Protein C 778 Low protein C predictive of death 55Fibrinolytic activity PAI-1 778 PAI-1 predictive of death 56

IL, interleukin; VWF, von Willebrand factor antigen; SP-D, surfactant protein-D; ICAM-1, intercellular adhesion molecule-1; sTNFR1/II, soluble tumornecrosis factor receptors I and II; PAI-1, plasminogen activator inhibitor-1.


systemic response to a protective ventila-tory strategy and potentially to identifypatients who are inadequately treated andmight be candidates for rescue therapies.

CONCLUSIONS

Acute lung injury and ARDS are com-mon, life-threatening disorders. Despitethe success of a low tidal volume strategyin reducing mortality, new therapies arestill needed. The design of clinical trialsfor new therapies for ALI and ARDS is acomplex problem that ultimately willhave a major impact on both trial out-come and generalizability. A number ofclinical factors such as age, comorbid ill-nesses, severity of illness scores, and in-dices of multiorgan system failure can beused to differentiate groups of patients athighest risk for adverse clinical out-comes. In addition, biologic markerssuch as the endothelial marker von Wil-lebrand factor antigen, the alveolar epi-thelial marker surfactant protein D, co-agulation and fibrinolysis proteins, andthe cytokines IL-6 and IL-8 and solubleTNF receptors are also independentmarkers of adverse outcomes. Ultimately,some combination of clinical factors andbiologic marker measurements may beuseful to select more homogeneousgroups of patients with ALI/ARDS for fu-ture clinical trials. Prospective evaluationof the potential value of combining clin-ical and biologic risk factors in a prog-nostic index is needed. However, al-though we have now identified a largenumber of factors associated with adverseprognosis in ALI/ARDS, it remains un-clear whether enriching study popula-tions with these sicker patients will in-crease or decrease the likelihood of atreatment effect for a given therapy. Fornovel ventilatory strategies such as high-frequency oscillatory ventilation, theidentification of patients who are failingconventional therapy using clinical indi-ces, biologic markers, and perhaps pul-monary dead space may be an alternativeway of enriching study populations withpatients more likely to benefit from anovel ventilatory strategy.

REFERENCES

1. Ware LB, Matthay MA: Medical progress: theacute respiratory distress syndrome. N EnglJ Med 2000; 342:1334–1349

2. Mahoney AM, Caldwell E, Hudson LD, et al:Barriers to physician recognition of acutelung injury. Am J Respir Crit Care Med 2003;167:A738

3. Goss CH, Brower RG, Hudson LD, et al: In-cidence of acute lung injury in the UnitedStates. Crit Care Med 2003; 31:1607–1611

4. Vincent JL, Sakr Y, Ranieri VM: Epidemiol-ogy and outcome of acute respiratory failurein intensive care unit patients. Crit Care Med2003; 31:S296–S299


6. Bernard GR, Artigas A, Brigham KL, et al:The American–European Consensus Confer-ence on ARDS. Definitions, mechanisms, rel-evant outcomes, and clinical trial coordina-tion. Am J Respir Crit Care Med 1994; 149:818–824

7. Brun-Buisson C, Minelli C, Bertolini G, et al:Epidemiology and outcome of acute lung in-jury in European intensive care units. Inten-sive Care Med 2004; 30:51–61

8. Milberg JA, Davis DR, Steinberg KP, et al:Improved survival of patients with acute re-spiratory distress syndrome (ARDS):1983–1993. JAMA 1995; 273:306–309

9. Abel SJC, Finney SJ, Brett SJ, et al: Reducedmortality in association with the acute respi-ratory distress syndrome. Thorax 1998; 53:292–294

10. Doyle RL, Szaflarski N, Modin GW, et al:Identification of patients with acute lung in-jury. Predictors of mortality. Am J RespirCrit Care Med 1995; 152:1818–1824

11. Nuckton TJ, Alonso JA, Kallet RH, et al: Pul-monary dead-space fraction as a risk factorfor death in the acute respiratory distresssyndrome. N Engl J Med 2002; 346:1281–1286

12. Schelling G, Stoll C, Haller M, et al: Health-related quality of life and posttraumaticstress disorder in survivors of the acute re-spiratory distress syndrome. Crit Care Med1998; 26:651–659

13. Elliott CG, Morris AH, Cengiz M: Pulmonaryfunction and exercise gas exchange in survi-vors of adult respiratory distress syndrome.Am Rev Respir Dis 1981; 123:492–495

14. Davidson TA, Caldwell ES, Curtis JR, et al:Reduced quality of life in survivors of acuterespiratory distress syndrome compared withcritically ill control patients. JAMA 1999;281:354–360

15. Herridge MS, Cheung AM, Tansey CM, et al:One-year outcomes in survivors of the acuterespiratory distress syndrome. N Engl J Med2003; 348:683–693

16. Wood KA, Huang D, Angus DC: Improvingclinical trial design in acute lung injury. CritCare Med 2003; 31:S305–S311

17. Fuhrman BP, Abraham E, Dellinger RP: Fu-tility of randomized, controlled ARDS tri-als—a new approach is needed. Crit CareMed 1999; 27:431–433

18. Gattinoni L, Tognoni G, Pesenti A, et al:Effect of prone positioning on the survival of

patients with acute respiratory failure.N Engl J Med 2001; 345:568–573

19. Taylor RW, Zimmerman JL, Dellinger RP, etal: Low-dose inhaled nitric oxide in patientswith acute lung injury: a randomized con-trolled trial. JAMA 2004; 291:1603–1609

20. Zilberberg MD, Epstein SK: Acute lung in-jury in the medical ICU. Comorbid condi-tions, age, etiology and hospital outcome.Am J Respir Crit Care Med 1998; 157:1159–1164

21. Monchi M, Bellenfant F, Cariou A, et al: Earlypredictive factors of survival in the acuterespiratory distress syndrome. A multivariateanalysis. Am J Respir Crit Care Med 1998;158:1076–1081

22. Luhr OR, Antonsen K, Karlsson M, et al:Incidence and mortality after acute respira-tory failure and acute respiratory distresssyndrome in Sweden, Denmark and Iceland.Am J Respir Crit Care Med 1999; 159:1849–1861

23. Rocco TR Jr, Reinert SE, Cioffi W, et al: A9-year, single-institution, retrospective re-view of death rate and prognostic factors inadult respiratory distress syndrome. AnnSurg 2001; 233:414–422

24. Estenssoro E, Dubin A, Laffaire E, et al: In-cidence, clinical course, and outcome in 217patients with acute respiratory distress syn-drome. Crit Care Med 2002; 30:2450–2456

25. Venet C, Guyomarc’h S, Pingat J, et al: Prog-nostic factors in acute respiratory distresssyndrome: a retrospective multivariate anal-ysis including prone positioning in manage-ment strategy. Intensive Care Med 2003; 29:1435–1441

26. Eberhard LW, Morabito DJ, Matthay MA, etal: Initial severity of metabolic acidosis pre-dicts the development of acute lung injury inseverely traumatized patients. Crit Care Med2000; 28:125–131

27. Bone RC, Maunder R, Slotman G, et al: Anearly test of survival in patients with theadult respiratory distress syndrome. Chest1989; 96:849–851

28. Sloane PJ, Gee MH, Gottlieb JE, et al: Amulticenter registry of patients with acuterespiratory distress syndrome. Am Rev Re-spir Dis 1992; 146:419–426

29. Knaus WA, Sun X, Hakim RB, et al: Evalua-tion of definitions for adult respiratory dis-tress syndrome. Am J Respir Crit Care Med1994; 150:311–317

30. Luhr OR, Karlsson M, Thorsteinsson A, et al:The impact of respiratory variables on mor-tality in non-ARDS and ARDS patients re-quiring mechanical ventilation. IntensiveCare Med 2000; 26:508–517

31. Bersten AD, Edibam C, Hunt T, et al: Inci-dence and mortality of acute lung injury andthe acute respiratory distress syndrome inthree Australian states. Am J Respir Crit CareMed 2002; 165:443–448

32. Krafft P, Fridrich P, Pernerstorfer T, et al:The acute respiratory distress syndrome: def-initions, severity and clinical outcome—an


analysis of 101 clinical investigations. Inten-sive Care Med 1996; 22:519–529

33. Spragg RG, Lewis JF, Walmarath H-D, et al:Effect of recombinant surfactant protein C-based surfactant on the acute respiratory dis-tress syndrome. N Engl J Med 2004; 351:884–892

34. Ferguson ND, Kacmarek RM, Chiche J-D, etal: Screening of ARDS patients using stan-dardized ventilator settings: influence on en-rollment in a clinical trial. Intensive CareMed. 2004; 30:1111–1116

35. Villar J, Perez-Mendez L, Kacmarek RM: Cur-rent definitions of acute lung injury and theacute respiratory distress syndrome do notreflect their true severity and outcome. In-tensive Care Med 1999; 25:930–935

36. Flori HR, Glidden DV, Rutherford GW, et al:Pediatric acute lung injury: prospective eval-uation of risk factors associated with mortal-ity. Am J Respir Crit Care Med 2005; in press

37. The Acute Respiratory Distress SyndromeNetwork: A trial of mechanical ventilationwith higher versus lower positive end-expiratory pressures in patients with acutelung injury and acute respiratory distresssyndrome. N Engl J Med 2004; 351:327–336

38. The ARDS Network: A trial of mechanicalventilation with higher versus lower positiveend-expiratory pressures in patients withacute lung injury and acute respiratory dis-tress syndrome. N Engl J Med 2004; 351:327–336

39. Parsons PE, Moss M: Early detection andmarkers of sepsis. Clin Chest Med 1996; 17:199–212

40. Pittet JF, MacKersie RC, Martin TR, et al:Biological markers of acute lung injury:prognostic and pathogenetic significance.Am J Respir Crit Care Med 1997; 155:1187–1205

41. Goodman R, Pugin J, Lee JS, et al: Cytokinemediated inflammation in acute lung injury.Cytokine Growth Factor Rev 2003; 14:523–535

42. Ware LB, Conner ER, Matthay MA: von Wil-lebrand factor antigen is an independentmarker of poor outcome in patients withearly acute lung injury. Crit Care Med 2001;29:2325–2331

43. Prabhakaran P, Ware L, White K, et al: Ele-vated levels of plasminogen activator inhibi-tor-1 in pulmonary edema fluid are associ-

ated with mortality in acute lung injury.Am J Physiol Lung Cell Mol Physiol 2003;285:L20–L28

44. Ware LB, Fang X, Matthay MA: Protein C andthrombomodulin in human acute lung in-jury. Am J Physiol Lung Cell Mol Physiol2003; 285:L514–L521

45. Miller EJ, Cohen AB, Nagao S, et al: Elevatedlevels of NAP-1/interleukin-8 are present inthe airspaces of patients with adult respira-tory distress syndrome and are associatedwith increased mortality. Am Rev Respir Dis1992; 146:427–432

46. Verghese GM, McCormick-Shannon K, Ma-son RJ, et al: Hepatocyte growth factor andkeratinocyte growth factor in the pulmonaryedema fluid of patients with acute lung in-jury. Am J Respir Crit Care Med 1998; 158:386–394

47. Chesnutt AN, Kheradmand F, Folkesson HG,et al: Soluble transforming growth factor-�is present in the pulmonary edema fluid ofpatients with acute lung injury. Chest 1997;111:652–656

48. Madtes DK, Rubenfeld G, Klima LD, et al:Elevated transforming growth factor-alphalevels in bronchoalveolar lavage fluid of pa-tients with acute respiratory distress syn-drome. Am J Respir Crit Care Med 1998;158:424–433

49. Chesnutt AN, Matthay MA, Tibayan FA, et al:Early detection of type III procollagen pep-tide in acute lung injury. Am J Respir CritCare Med 1997; 156:840–845

50. Clark JG, Milberg JA, Steinberg KP, et al:Type III procollagen peptide in the adult re-spiratory distress syndrome. Ann Intern Med1995; 122:17–23

51. Pugin J, Verghese G, Widmer M-C, et al: Thealveolar space is the site of intense inflam-matory and profibrotic reactions in the earlyphase of ARDS. Crit Care Med 1999; 27:304–312

52. Greene KE, Wright JR, Steinberg KP, et al:Serial changes in surfactant-associated pro-teins in lung and serum before and afteronset of ARDS. Am J Respir Crit Care Med1999; 160:1843–1850

53. Moss MD, Gillespie MK, Ackerson L, et al:Endothelial cell activity varies in patients atrisk for the adult respiratory distress syn-drome. Crit Care Med 1996; 24:1782–1786

54. Meduri GU, Kohler G, Headley S, et al: In-

flammatory cytokines in the BAL of patientswith ARDS. Persistent elevation over timepredicts poor outcome. Chest 1995; 108:1303–1314

55. Ware LB, Ancukiewicz M, Parsons PI, et al:Prognostic value of plasma protein C in pa-tients with acute lung injury. Am J RespirCrit Care Med 2003; 167:A661

56. Ware LB, Eisner MD, Wickersham N, et al:Plasminogen activator inhibitor-1 (PAI-1) amarker of impaired fibrinolysis, is associatedwith higher mortality in patients with ALI/ARDS. Am J Respir Crit Care Med 2004;169:A115

57. Parsons PE, Eisner MD, Thompson BT, et al:Lower tidal volume ventilation and plasmacytokine markers of inflammation in patientswith acute lung injury. Crit Care Med 2005;33:1–6

58. Parsons PE, Matthay MA, Ware LB, et al:Elevated plasma levels of soluble TNF recep-tors are associated with morbidity and mor-tality in patients with acute lung injury. Am JPhysiol Lung Cell Mol Physiol 2005; in press

59. Ware LB, Eisner MD, Thompson BT, et al:Significance of von Willebrand factor in sep-tic and non-septic patients with acute lunginjury. Am J Respir Crit Care Med 2004;170:766–772

60. Eisner MD, Parsons P, Matthay MA, et al:Serum surfactant protein levels and clinicaloutcomes in patients with acute lung injury.Thorax 2003; 58:983–988

61. Heffner JE, Brown LK, Barbieri CA, et al:Prospective validation of an acute respiratorydistress syndrome predictive score. Am J Re-spir Crit Care Med 1995; 152:1518–1526

62. Offner PJ, Moore EE: Lung injury severityscoring in the era of lung protective mechani-cal ventilation: the PaO2/FIO2 ratio. J Trauma2003; 55:285–289

63. Kallet RH, Alonso JA, Pittet J-F, et al: Prog-nostic value of the pulmonary dead-spacefraction during the first 6 days of acute re-spiratory distress syndrome. Respir Care2004; 49:1008–1014

64. Ranieri VM, Suter PM, Tortorella C, et al:Effect of mechanical ventilation on inflam-matory mediators in patients with acute re-spiratory distress syndrome. JAMA 1999; 282:54–61


Protocols for lung protective ventilation

Henry E. Fessler, MD; Roy G. Brower, MD

A spectrum of decision supporttools is available to assist in themanagement of complex pa-tients. Protocols, which lie

near one end of the spectrum, are sets ofexplicit, algorithmic rules, which directclinical management or research. Theymust be distinguished from guidelines, aset of principles, goals, or suggestionsthat may also direct clinical or researchdecisions. In general, guidelines are moregeneral, flexible, and tolerant of latitudeamong clinicians. They lie closer to theother pole of the decision support toolsspectrum. Protocols, although not neces-sarily inviolate, are more specific in theirinstructions. The distinction is somewhatarbitrary, but a useful working definitionis that a protocol is a set of rules that willlead varied practitioners, faced with theidentical clinical situation, to reach theidentical decision.

Protocols have long been recognizedas essential to the performance of pro-spective clinical research and are beingused increasingly for clinical care. Webelieve they are particularly valuable toguide the use of unfamiliar or complexinterventions such as high-frequency os-cillatory ventilation (HFOV). However,we will first distinguish between their usefor clinical research and clinical practice,because these distinctions have implica-

tions for the design, implementation, andeffects of protocols.

Research Protocols

In clinical research, the most apparentneed for a protocol is to direct the appli-cation of the intervention under study,for example, in a randomized trial. Anal-ogous to specifying the quantity, fre-quency, and administration route of aninvestigational drug, the application of anew intervention must be well-defined toassure that the study group receives anappropriate “dose.” For example, in theARDS Network (1) trial comparing smallwith traditional tidal volumes in acutelung injury/acute respiratory distresssyndrome (ALI/ARDS), there were ex-plicit instructions to determine the sizeof the tidal volumes in the lower tidalvolume arm.

A somewhat more controversial appli-cation of protocols in clinical research istheir use to direct the management of thecontrol or comparison arm in regard tothe intervention under study (2). One ap-proach is to allow the comparison arm toreceive “usual care,” care delivered by thejudgment (or whim) of the treating phy-sician. The alternative is to provide anexplicit protocol, which limits the clini-cians’ range of choices while embodying aprocess of care that many practitionerswould have chosen were they providingusual care to similar patients. Each trialdesign may be appropriate for differentresearch questions. However, “usualcare” in critical care is often highly vari-able, and the determinants of clinicians’varied choices are obscure. In unblindedtrials, patient management decisions may

also be altered by the very existence of thestudy (the Hawthorne effect). Therefore,use of usual care as a control arm willtend to reduce a treatment effect, perhapsrendering it immeasurable. If effects aredetected, the specific determinants of theoutcome are more difficult to define. Arecent randomized trial of prone vs. su-pine positioning for hypoxemic respira-tory failure exemplifies this problem (3).This study tested the effect on mortalityof an intervention, prone positioning,that may protect the lung from ventila-tor-induced injury. However, there wasno systematic control of the strategy ofmechanical ventilation in either arm, andprone positioning was allowed in the con-trol arm if clinicians felt it was indicated(despite the absence of data on its effi-cacy). This resulted in a 21% crossoverrate, and there was no difference in mor-tality between the study groups.

A final role for protocols in clinicalresearch is to reduce confounding byother aspects of care that may affect out-comes but are not themselves the inter-vention under study. Given the complex-ity of care for the critically ill, this mayappear unobtainable. Some trials havegone to great lengths to control aspectsof clinical care (4). However, the mostimportant issues to control by protocolwould be those likely to both affect out-comes and to vary between study arms.Examples in studies of mechanical venti-lation are the approach to weaning (1) orthe use of sedatives and paralytics (5). Inthe ARDS Network tidal volume study,ventilator-free days was an importantoutcome variable. Because it was impos-sible to blind the clinicians to study

From the Division of Pulmonary and Critical CareMedicine, Johns Hopkins Medical Institutions, Balti-more, MD.


DOI: 10.1097/01.CCM.0000155919.53727.D5

Protocols have a well-established role in clinical research and areincreasingly being used to direct routine clinical care. In this article,we review the differing goals of research and clinical protocols andoutline the similar process for their development. We use the me-chanical ventilation protocol of the ARDS Network trial comparingsmall with traditional tidal volumes as an example. As a startingpoint for debate, we also suggest guiding principles and specific

components of a protocol for high-frequency oscillatory ventilation.(Crit Care Med 2005; 33[Suppl.]:S223–S227)

KEY WORDS: process of care; protocols; decision support; clin-ical research; clinical trials; mechanical ventilation; high-fre-quency oscillatory ventilation; acute respiratory distress syn-drome; ventilator-induced lung injury; lung-protective ventilation


group assignment, there was a potentialfor systematic differences in the approachto weaning between study groups. There-fore, weaning in both arms was con-trolled by the same protocol. The com-plex environment of the intensive careunit creates a very “noisy” field in whichto distinguish a treatment effect. The useof protocols to manage ancillary aspectsof care helps to decrease the noise attrib-utable to unnecessary practice variationand potentially reveal the signal.

Clinical Care Protocols

The use of protocols to guide routineclinical care outside of the research set-ting is a more recent development incritical care with a distinct set of goalsand requirements. At least four goals canbe served. First, the protocols reduce thedegree of variability inherent in usualcare. To the extent that evidence or ex-pert opinion supports one managementstyle over others, patient outcomes mayimprove by increasing compliance withthe evidence.

A second, but related, goal is the rapidintegration of new information into clin-ical practice. Several studies have shownthat clinicians are slow to change theirpractice after the publication of new find-ings (6, 7). Several explanations for thisinertia have been suggested (8). However,protocols can speed this process by mak-ing the preferred approach the defaultapproach and requiring more effort orthought to deviate to other strategies.

A third goal is to redistribute work-load. A well-designed clinical protocolcan clarify decision-making, allowingpoorly understood judgment by physi-cians to be replaced by more determinis-tic decisions that can be made by nursesor therapists. An example of this is thevarious protocols for ventilator weaningby nursing and respiratory care staff,which have been shown to decrease timeon mechanical ventilation in several set-tings (9–11). To the extent that physiciantime is more expensive, this redistribu-tion of workload can result in cost sav-ings even if it has no effect on clinicaloutcomes. However, involving additionalskilled personnel in a process of care tra-ditionally reserved for physicians can of-ten both reduce costs and improve out-comes. Relatively simple protocols forscheduled weaning by nurses and thera-pists have often hastened extubationcompared with the complex physiologicalmeasures and judgments of physicians.

This suggests that the regularity of stepsenforced by a protocol as executed bynurses or therapists trumps the rarifiedindividual decisions made sporadically bybusy physicians.

Finally, protocols can facilitate medi-cal education. It may seem counterintui-tive that a tool that reduces the need tothink carefully and individually abouteach patient can be a teaching tool. How-ever, the process of protocol development(discussed subsequently) includes a thor-ough review of relevant data and expertopinion, consideration of current localpractice and patient needs. It clarifies, aswell as codifies, the state-of-the-art on aspecific patient management question.The protocol provides a template for ex-plaining why it represents the preferredmode of therapy, much like the periodictable of the elements can be used as ei-ther a simple reference chart or a richdistillate of chemistry and physics.

Clinical protocols have also been crit-icized. First, there is the potential loss ofthe individualization of care, or subver-sion of the art of medicine (12). Thiscriticism, in part, assumes that variationsin usual care result from thoughtful ti-tration rather than individual prejudice,routine, style, or inattention. The criti-cism is also more validly leveled at apoorly designed protocol that does notanticipate the range of circumstances inwhich it will be applied or allow for indi-vidual titration within a goal range. Sec-ond, protocols may be criticized if theyare applied in populations beyond whichthey have been tested. This, too, is a fail-ure of protocol design rather than of pro-tocols per se.

Despite these potential shortcomings,clinical protocols in critical care havebeen shown to improve medication use(13), weaning (9), glucose control andassociated mortality (14), and nutritionsupport (15). Their use has been encour-aged by published practice guidelines(16). As results from studies using re-search protocols continue to inform cli-nicians about the best management deci-sions, clinical protocols will remainpractical tools to implement those deci-sions at the bedside.

How to Construct a Protocol

Research protocols are often imple-mented by focused, motivated researchpersonnel or by clinicians closely super-vised by researchers. This allows researchprotocols to be detailed and relatively

complex while still assuring compliance.In contrast, clinical protocols are appliedby busy doctors, nurses, and therapistswho are juggling numerous tasks. Im-proved patient outcomes are a high pri-ority, but the benefits of protocol adher-ence for individual patients are notdirectly visible. Protocols may also beperceived as constraints to clinicianchoice or judgment. Therefore, cliniciansmay not be as committed as researchersto adherence with a protocol. Clinicalprotocols will be followed in proportionto their simplicity, the consistency oftheir decision instructions with reasonedclinical judgment, the thoroughness andenthusiasm with which staff are trained,and the payback they provide in reducedworkload or obvious improved outcomes.Despite differences in their final struc-ture, the processes of protocol develop-ment for research or clinical care aresimilar. In either case, the effort neededto design and implement a protocol suc-cessfully is also frequently underesti-mated.

Protocol development, like criticalcare, is a team process. Protocols im-posed by an individual are doomed tofailure. The team must include broadmedical knowledge and representationfrom the disciplines that will be imple-menting the protocol. Broad representa-tion will quickly uncover flaws that wouldimpede implementation and will build asense of ownership and commitmentfrom participants. The process should be-gin by identifying the goals of the proto-col and the population in which it will beused. This will be assisted by a review ofavailable data, including published simi-lar protocols, local practices and opin-ions, and any available guidelines.

The review will assemble a variety ofpractices, preferences, and protocols sup-ported by evidence of variable quality.The next step is winnowing these optionsdown to a consensus. Formal methodol-ogies for consensus-building have beendeveloped such as the Delphi techniqueor the Nominal Group Process. These areseldom used in protocol development,which typically relies on informal “brain-storming.” Informal methods, however,risk domination by those with the moststrongly held or passionately expressedopinions. For that reason, it is importantthat candidate protocols be reviewed bypractitioners outside of the developmentcommittee early in their genesis. Thiswill ensure that they do not stray so far


from accepted practice that they are un-likely to be followed.

After an approach is agreed on, thegeneral strategy must be distilled into aset of rules or a decision tree. This is acritical step, which requires unambigu-ous decision rules that anticipate a broadrange of possible contingencies. Decisionrules will be applied most consistentlywhen based on objective and quantifiabledata such as a respiratory rate or PaO2

rather than subjective factors such aswhether a patient appears dyspneic. Com-promises must be made between thespecificity and complexity of the protocol;as protocols attempt to control more de-tails, they become exponentially morecomplicated and adherence suffers.

Once a protocol has been committedto paper, it should be submitted to aniterative process of pilot testing and revi-sion. This will reveal aspects that are toorestrictive, too lax, too complicated, orsimply illogical. When a practical proto-col has been built, a more extended pilotshould be undertaken. This will uncoverflaws by exposure to a wider range ofpatients and staff. For example, we per-formed a trial of small vs. traditional tidalvolumes in ARDS in four hospitals inBaltimore (17). The experience from thattrial assisted protocol development forthe much larger multicenter trial under-taken by the ARDS Network (1). The pilotis also the opportunity to develop and testthe educational program that will beneeded for wide implementation and torefine data reporting forms that will beused to track adherence or efficacy.

Before implementation, staff must beeducated. The details of this process willvary with the specific aspects of the pro-tocol. It will involve personnel from var-ious disciplines with different skills andlevels of insight and motivation whomust all be motivated to alter their prac-tice. When the protocol is implemented,plans must be in place to monitor theircompliance and provide feedback and en-couragement. Finally, because compli-ance with the protocol is not the goal initself, it is useful to collect data docu-menting the impact of the protocol onpatient outcomes, costs, or other goals.This will serve to motivate the cliniciansand justify the investment in protocoldevelopment.

This process is laborious and not in-variably successful. Ely et al. (18) docu-mented the steps they used to extend asimple weaning protocol throughouttheir hospital after demonstrating its ef-

ficacy in the medical intensive care unit.The process took a full year. Even in theirexperienced hands, compliance was�65% and quickly fell when ongoing ed-ucation and reinforcement was stopped.A weaning protocol for pediatric patients,which was the subject of a recent multi-center trial (19), took an estimated 250man-hours to develop (20).

Protocols for High-FrequencyOscillatory Ventilation

HFOV for adult ALI/ARDS is both acomplex and unfamiliar technology forwhich research protocols are essential.Guidelines for the use of HFOV in adultsare widely available and discussed else-where in this supplement. However, out-come benefits of HFOV compared withthe best lung-protective mechanical ven-tilation strategy have yet to be demon-strated. Given the high acquisition costsof HFO ventilators, suitable for only aminority of ventilated patients, outcomeadvantages must be demonstrated to jus-tify the widespread use of this technol-ogy. This will require rigorous compari-son between well-defined conventionaland HFOV strategies in large clinical tri-als. These, in turn, will require that bothstrategies be directed by protocols. Untilthe results of such trials are available,clinical protocols will also be useful todirect the use of this unfamiliar technol-ogy in a way that maximizes its safety andefficacy.

The �2 decades of experience withHFOV in neonatology provides some in-structive lessons for adult practitioners.The earliest large randomized trial ofHFOV in neonates failed to recognize theimportance of lung recruitment to mini-mize ventilator-associated lung injury.HFOV was applied using relatively lowmean airway pressures (mPaw) of only8–10 cm H2O. In this study, HFOV didnot reduce the incidence of chronic lungdisease or mortality, but a higher inci-dence of some complications occurred inthe HFOV group (21). This experiencereminds us that protocols not based onsound pathophysiological insight mayharm patients and doom a promisingtechnology.

More recent trials comparing HFOVwith conventional ventilation in neonateshave used protocols that favored highermPaw. Nevertheless, two recent large,multicenter trials enrolling similar pa-tients came to differing conclusions. Inone, HFOV reduced the incidence of

chronic lung disease at full gestationalage (22). The other found no benefit ofHFOV (23). These discordant conclusionssuggest that the treatment effect is mod-est, and that small variations betweenprotocols for HFOV or conventional ven-tilation may be sufficient to alter the out-come of the trial.

It is difficult to trace the ontogeny ofthe ventilator-setting guidelines forHFOV in adults. We believe they haveevolved from settings used in pediatricsand trial and error during early experi-ence in adults. Moreover, they were builtlargely on the immediate feedback fromphysiological outcomes such as blood gasimprovement (positive feedback) or he-modynamic compromise (negative feed-back). This resulted in an approach thatyields acceptable gas exchange and hemo-dynamics in most patients. There is ex-tensive evidence, however, that gas ex-change is not an accurate surrogate forsurvival in ALI/ARDS. Interventions suchas inhaled nitric oxide (24) or higher lev-els of PEEP (25) improved oxygenationwithout improving survival. The use oftidal volumes of 6 mL/kg predicted bodyweight had deleterious effects on gas ex-change but improved survival (1). Thus,with survival as the goal, we have littleevidence that current guidelines forHFOV in adults are optimal.

We propose that the protocols forHFOV in adults be guided by the follow-ing principles, each of which is eitherdirectly supported by animal and humandata or extrapolated from such data:

1. Smaller tidal volumes are less injuri-ous than larger tidal volumes.

2. For any given steady-state level of ar-terial CO2, the benefits of smaller tidalvolumes are not negated by associatedhigher respiratory rates.

3. Relatively high airway pressures areless injurious if the lungs are dis-tended at a nearly static volume thanif exposed to the same pressure andvolume only at the peak of each con-ventional tidal breath. These threepoints are the foundation for the useof HFOV as a form of lung protectiveventilation.

4. High pleural pressures accompanyingeffective lung recruitment will com-promise venous return, and hightranspulmonary pressure may in-crease right ventricular afterload. Theresulting need for volume resuscita-tion may negate beneficial effects ofrecruitment on lung injury.


5. Neuromuscular blockade or deep se-dation may prolong the duration ofmechanical ventilation. Because thesemedications are often required duringHFOV, the duration of HFOV useshould be minimized.

6. Lung protective ventilation will bemost beneficial when applied early inthe course of ALI/ARDS, before venti-lator-induced injury is widespread.

From these principles, we derive thefollowing more specific recommenda-tions for protocol design (each of whichwe offer as a starting point for debate):

Patient Selection. HFOV has beenshown to improve gas exchange, but itseffect on patient outcomes relative tolung-protective conventional ventilationis unknown. Presumptive beneficial ef-fects on lung injury are accompanied bythe potentially deleterious effects on he-modynamics, sedation requirements, andimpaired ability to perform a physical ex-amination. For these reasons, we believethe clinical application of HFOV shouldbe limited to patients who are failing con-ventional ventilation. Such failure shouldbe documented by inability to maintainthresholds of oxygenation or ventilationon lung-protective conventional ventila-tion. For a research protocol, on theother hand, lung-protective benefits ofHFOV will be most demonstrable if it isbegun early in the course, within 24–48hrs of meeting criteria for ARDS. Limit-ing such a study to patients who havefailed conventional ventilation will makeit ethically difficult to justify randomiza-tion, because physicians may insist on itsuse as rescue therapy. Moreover, such atrial may select a population of patientstoo late to benefit from HFOV.

pH Goal. Modest degrees of hypercar-bic acidosis (pH �7.25) should be accept-able, because a goal of normal pH willrequire either larger or more frequentbreaths.

Frequency. Given the influence oftidal volume on survival in ARDS, itwould be very useful to know the tidalvolume during HFOV to titrate settingsto individual patient mechanics. Lackingthis information, we should choose set-tings that minimize tidal volume. Ratherthan begin HFOV at 5 Hz and only reducefrequency if necessary to correct hyper-carbia, we recommend that HFOV beused at the highest tolerated frequencytogether with maximal oscillation ampli-tude (delta P). Our rationale for this isthat, for a given PaCO2, tidal volume will

be smaller at higher frequency even withhigher delta P. Applying this approach toour last nine patients, we found that sixcould maintain pH �7.25 at frequenciesbetween 7 and 12 Hz. The other threerequired frequencies of 3–4 Hz as a resultof refractory acidosis.

Mean Airway Pressure. The mPawshould be maintained in the 25- to 35-cmH2O range to promote lung recruitment,but FIO2 should be increased instead ofmPaw to maintain oxygenation whenblood pressure is compromised.

Recruitment Maneuvers. Attempts to

recruit lung by briefly (�1 min) raisingmPaw to 40 –50 cm H2O may be at-tempted when initiating HFOV or aftersuctioning or patient disconnections. Weknow of no clinical data to support theuse of recruitment maneuvers at regularintervals or repeated attempts in patientswho fail to demonstrate a brisk and sus-tained improvement in oxygenation aftera maneuver.

Transition to Conventional Ventila-tion. To minimize the duration of HFOV,patients in whom HFOV is used as rescuetherapy for gas exchange should have a

Figure 1. The HFOV Quick Guide is a chart summarizing the protocol for clinical application ofhigh-frequency oscillatory ventilation (HFOV) used at Johns Hopkins Hospital. It is posted at thebedside of all patients on HFOV as a reference for housestaff, nurses, and respiratory therapists.Greater detail is available in the protocol manuals for nursing and respiratory therapy.


trial of conventional ventilation dailywhen gas exchange improves to require amPaw �30 cm H2O and �50% oxygen.Transition criteria will differ for researchpatients in whom HFOV is applied earlyin their course to study its effect on ven-tilator-induced lung injury. In that situ-ation, a minimum duration of HFOV maybe required. In addition, lung mechanicsand gas exchange should improve to thepoint where the risk of substantial venti-lator-induced injury on conventionalventilation will be low. This might beindicated by a plateau pressure �25 cmH2O, need for �50% FIO2, and a minuteventilation �12 L during a trial on con-ventional, lung-protective ventilation.

Sedation. To minimize the total dura-tion of mechanical ventilation, sedation,and neuromuscular blockade on bothHFOV and conventional ventilationshould be controlled by a separate proto-col. See chapter 16 of this supplement.

The introduction of an endotrachealtube cuff leak can also be used to improveCO2 clearance during HFOV. This willallow the use of lower tidal volumes, po-tentially improving the degree of lungprotection. However, deflating the endo-tracheal tube cuff may also promote ven-tilator-associated pneumonia by allowingoral secretions to enter the lower airways.We have no data on which to base arecommendation as to when a cuff leakshould be added in a management algo-rithm. However, rules for a cuff leakshould be designed into a research pro-tocol so that its use will be consistent andreproducible.

Currently, we use HFOV for patientsfailing conventional lung-protective ven-tilation. Our clinical protocol for HFOV issummarized in Figure 1.

CONCLUSIONS

Well-designed management protocolsare essential for interpretable clinical tri-als and have been shown to improve nu-merous aspects of care for the criticallyill. Protocol development can be arduousbut can yield the returns of more efficientand effective patient care. Protocol devel-opment for HFOV is in its formativestage, hampered by limited human phys-iological data and the inability to mea-sure tidal volume. These protocols willmature with improved instrumentation

(see chapters 10 and 11), focused pilotstudies, and consensus-building amongexperienced users. We will then be pre-pared to undertake a clinical efficacytrial, the results of which can be defini-tive and robust.

REFERENCES

1. ARDS Network Investigators: Ventilationwith lower tidal volumes as compared withtraditional tidal volumes for acute lung in-jury and the acute respiratory distress syn-drome. N Engl J Med 2000; 342:1301–1308

2. Silverman HJ, Miller FG: Control group se-lection in critical care randomized controlledtrials evaluating interventional strategies: anethical assessment. Crit Care Med 2004; 32:852–857

3. Guerin C, Gaillard S, Lemasson S, et al: Ef-fects of systematic prone positioning in hy-poxemic acute respiratory failure: a random-ized controlled trial. JAMA 2004; 292:2379–2387

4. Morris AH, Wallace CJ, Menlove RL, et al:Randomized clinical trial of pressure-controlled inverse ratio ventilation and ex-tracorporeal CO2 removal for adult respira-tory distress syndrome. Am J Respir CritCare Med 1994; 149:295–305

5. Kress JP, Pohlman AS, O’Connor MF, et al:Daily interruption of sedative infusions incritically ill patients undergoing mechanicalventilation. N Engl J Med 2000; 342:1471–1477

6. Pronovost PJ, Rinke ML, Emery K, et al:Interventions to reduce mortality among pa-tients treated in intensive care units. J CritCare 2004; 19:158–164

7. Weinert CR, Gross CR, Marinelli WA: Impactof randomized trial results on acute lunginjury ventilator therapy in teaching hospi-tals. Am J Respir Crit Care Med 2003; 167:1304–1309

8. Berwick DM: Disseminating innovations inhealth care. JAMA 2003; 289:1969–1975

9. Ely EW, Baker AM, Dunagan DP, et al: Effecton the duration of mechanical ventilation ofidentifying patients capable of breathingspontaneously. N Engl J Med 1996; 335:1864–1869

10. Kollef MH, Shapiro SD, Silver P, et al: Arandomized, controlled trial of protocol-directed versus physician-directed weaningfrom mechanical ventilation. Crit Care Med1997; 25:567–574

11. Marelich GP, Murin S, Battistella F, et al:Protocol weaning of mechanical ventilationin medical and surgical patients by respira-tory care practitioners and nurses: effect onweaning time and incidence of ventilator-associated pneumonia. Chest 2000; 118:459–467

12. Tobin MJ: Of principles and protocols andweaning. Am J Respir Crit Care Med 2004;169:661–662

13. Aldea GS, O’Gara P, Shapira OM, et al: Effectof anticoagulation protocol on outcome inpatients undergoing CABG with heparin-bonded cardiopulmonary bypass circuits.Ann Thorac Surg 1998; 65:425–433

14. van den Berghe G, Wouters P, Weekers F, etal: Intensive insulin therapy in the criticallyill patients. N Engl J Med 2001; 345:1359–1367

15. Hedberg AM, Lairson DR, Aday LA, et al:Economic implications of an early postoper-ative enteral feeding protocol. J Am Diet As-soc 1999; 99:802–807

16. Ely EW, Meade MO, Haponik EF, et al: Me-chanical ventilator weaning protocols drivenby nonphysician health-care professionals:evidence-based clinical practice guidelines.Chest 2001; 120(suppl):454S–463S

17. Brower RG, Shanholtz CB, Fessler HE, et al:Prospective, randomized, controlled clinicaltrial comparing traditional versus reducedtidal volume ventilation in acute respiratorydistress syndrome patients. Crit Care Med1999; 27:1492–1498

18. Ely EW, Bennett PA, Bowton DL, et al: Largescale implementation of a respiratory thera-pist-driven protocol for ventilator weaning.Am J Respir Crit Care Med 1999; 159:439–446

19. Randolph A, Wypij D, Venkataraman S, et al:Effect of mechanical ventilator weaning pro-tocols on respiratory outcomes in infants andchildren. JAMA 2002; 288:2561–2568

20. Randolph A: A practical approach to evi-dence-based medicine. Crit Care Clin 2004;19:515–527

21. The HIFI Study Group: High-frequency os-cillatory ventilation compared with conven-tional mechanical ventilation in the treat-ment of respiratory failure in preterminfants. N Engl J Med 1989; 320:88–93

22. Courtney SE, Durand DJ, Asselin JM, et al:High-frequency oscillatory ventilation versusconventional mechanical ventilation forvery-low-birth-weight infants. N Engl J Med2002; 347:643–652

23. Johnson AH, Peacock JL, Greenough A, et al:High-frequency oscillatory ventilation forthe prevention of chronic lung disease ofprematurity. N Engl J Med 2002; 347:633–642

24. Taylor RW, Zimmerman JL, Dellinger RP, etal: Low-dose inhaled nitric oxide in patientswith acute lung injury: a randomized con-trolled trial. JAMA 2004; 291:1603–1609

25. Brower RG, Lanken PN, MacIntyre N, et al:Higher versus lower positive end-expiratorypressures in patients with the acute respira-tory distress syndrome. N Engl J Med 2004;351:327–336


Other approaches to open-lung ventilation: Airway pressure releaseventilation

Nader M. Habashi, MD, FACP, FCCP

Airway pressure release ventila-tion (APRV) was initially de-scribed by Stock and Downs (1,2) as continuous positive air-

way pressure (CPAP) with an intermittentpressure release phase. Conceptually,APRV applies a continuous airway pres-sure (Phigh) identical to CPAP to maintainadequate lung volume and promote alve-olar recruitment. However, APRV adds atime-cycled release phase to a lower setpressure (Plow). In addition, spontaneousbreathing can be integrated and is inde-pendent of the ventilator cycle (Fig. 1).CPAP breathing mimics the gas distribu-tion of spontaneous breaths as opposed tomechanically controlled, assisted, or sup-ported breaths, which produce less phys-iological distribution (3–6). Mechanicalbreaths shift ventilation to nondependentlung regions as the passive respiratorysystem accommodates the displacementof gas in to the lungs. However, sponta-neous breathing during APRV results in amore dependent gas distribution whenthe active respiratory system draws gasinto the lung as pressure changes and

flow follow a similar time course (7–9).As a result, by allowing patients to spon-taneously breathe during APRV, depen-dent lung regions may be preferentiallyrecruited without the need to raise ap-plied airway pressure.

APRV has been used in neonatal, pe-diatric, and adult forms of respiratoryfailure (1–4, 6, 10–22). Clinical studiesusing APRV are summarized in Table 1(1–4, 12, 18, 23–28).

In patients with decreased functionalresidual capacity (FRC), elastic work ofbreathing (WOB) is effectively reducedwith the application of CPAP. As FRC isrestored, inspiration begins from a morefavorable pressure/volume relationship,facilitating spontaneous ventilation andimproving oxygenation (29).

However, in acute lung injury/acuterespiratory distress syndrome (ALI/ARDS), the surface area available for gasexchange is significantly reduced. Despiteoptimal lung volume, CPAP mandatesthat unaided spontaneous breathingmanage the entire metabolic load or CO2

production. However, CPAP alone may beinadequate to accomplish necessary CO2

removal without producing excessiveWOB. In contrast to CPAP, APRV inter-rupts airway pressure briefly to supple-ment spontaneous minute ventilation.During APRV, ventilation is augmented

by releasing airway pressure to a lowerCPAP level termed Plow. The intermittentrelease in airway pressure during APRVprovides CO2 removal and partially un-loads the metabolic burden of pure CPAPbreathing.

By using a release phase for ventila-tion, APRV uncouples the traditional re-quirement of elevating airway pressure,lung volume, and distension during tidalventilation. Rather than generating atidal volume by raising airway pressureabove the set positive end-expiratorypressure (PEEP) (like in conventionalventilation), release volumes in APRV aregenerated by briefly releasing airwaypressure from Phigh to Plow. Because ven-tilation with APRV results as airway pres-sure and lung volume decrease (releasevolume), the risk of overdistension maybe reduced. In contrast, conventionalventilation raises airway pressure, elevat-ing lung volumes, potentially increasingthe threat of overdistension (Fig. 2).

Ventilation generated by the releasephase of APRV may have additional ad-vantages in ALI/ARDS. Increased elasticrecoil is common to restrictive lung dis-eases such as ALI/ARDS. With APRV, asairway pressure is briefly interrupted, therelease volume is driven by gas compres-sion and lung recoil (potential energy)stored during the Phigh time period or

From the Multi-trauma ICU, R Adams CowleyShock Trauma Center, Baltimore, MD.


DOI: 10.1097/01.CCM.0000155920.11893.37

Objective: To review the use of airway pressure release ven-tilation (APRV) in the treatment of acute lung injury/acute respi-ratory distress syndrome.

Data Source: Published animal studies, human studies, andreview articles of APRV.

Data Summary: APRV has been successfully used in neonatal,pediatric, and adult forms of respiratory failure. Experimental andclinical use of APRV has been shown to facilitate spontaneousbreathing and is associated with decreased peak airway pres-sures and improved oxygenation/ventilation when compared withconventional ventilation. Additionally, improvements in hemody-namic parameters, splanchnic perfusion, and reduced sedation/neuromuscular blocker requirements have been reported.

Conclusion: APRV may offer potential clinical advantages forventilator management of acute lung injury/acute respiratorydistress syndrome and may be considered as an alternative“open lung approach” to mechanical ventilation. WhetherAPRV reduces mortality or increases ventilator-free days com-pared with a conventional volume-cycled “lung protective”strategy will require future randomized, controlled trials. (CritCare Med 2005; 33[Suppl.]:S228 –S240)

KEY WORDS: airway pressure release ventilation; airway pres-sure release ventilation; spontaneous breathing; lung protectivestrategies; acute lung injury; acute respiratory distress syndrome


Thigh. During conventional ventilation,inspiratory tidal volumes must overcomeairway impedance and elastic forces ofthe restricted lung from a lower baselineresting volume, increasing the energy orpressure required to distend the lung andchest wall. Furthermore, as thoraciccompliance decreases, the inspiratorylimb of the volume/pressure curve shiftsto the right, i.e., more pressure is re-quired to deliver a set tidal volume. How-ever, the expiratory limb remains unaf-fected by the prevailing volume/pressurerelationship and extends throughout allphases of injury (30). APRV uses the morefavorable volume/pressure relationship ofthe expiratory limb for ventilation by ap-plying a near-sustained inflation or re-cruitment state (31).

Alveolar recruitment is a pan-inspira-tory phenomenon. Successful recruitingpressure depends on the yield or thresh-old opening pressure (TOP) of lung units.ALI/ARDS may have a multitude of TOPdistributed throughout the lung (32–35).In addition to TOP, the time-dependentnature of recruitment should also be con-sidered. Although the exact mechanismsare not known, the lung is interdepen-dent and recruitment of air spaces resultsin radial traction of neighboring alveoli,producing a time-dependent ripple effectof recruitment (36–38).

As lung units recruit, the additionaltime (Thigh) at Phigh provides stability asan “avalanche” of lung units pop open(37, 38). Conceptually, superimposed

spontaneous breaths at a high lung vol-ume rather than brief and frequent tidalventilation between PEEP and end-inspiratory pressure may be more suc-cessful in achieving progressive and sus-tained alveolar recruitment.

Airway opening is dynamic as the lungcreeps to the recruited lung volume.Compliance and resistance (time con-stants) of recently recruited lung unitsdecrease the inflating or sustaining pres-sure requirements. Therefore, progres-sive extensions of Thigh may be critical forsustaining recruitment as time constantsevolve (38). Furthermore, the sustainedThigh period may encourage spontaneousbreathing at an upper and open lung vol-ume, improving efficiency of ventilation.

Although recruitment maneuvers maybe effective in improving gas exchangeand compliance, these effects appear tobe nonsustained, requiring repeated ma-neuvers (39, 40). Alternatively, APRV maybe viewed as a nearly continuous recruit-ment maneuver with the Phigh providing80% to 95% of the cycle time creating astabilized “open lung” while facilitatingspontaneous breathing. Fundamentally,assisted mechanical breaths cannot pro-vide the same gas distribution as sponta-neous breaths. Therefore, during a re-cruitment maneuver in a passiverespiratory system, the nondependentlung regions distend first until appliedairway pressure reaches and exceeds thehigh TOP of the dependent lung units,increasing the threat of overdistention.

Conversely, spontaneous breathing favorsdependent lung recruitment through theapplication of pleural pressure. Sponta-neous breaths at the CPAP level (Phigh)improve dependent ventilation throughpleural pressure changes rather than theapplication of additional applied airwaypressure (5, 6, 26). The recruited lungrequires less pressure than the recruitinglung. Therefore, maintaining lung vol-ume and allowing spontaneous breathingfrom the time of intubation by usingAPRV (CPAP with release) may reduce theneed for recurrent high CPAP recruit-ment maneuvers (41). If a recruitmentmaneuver is desired during APRV, thePhigh and Thigh can be adjusted to simu-late a conventional CPAP-type recruit-ment maneuver (e.g., Phigh 40–50 cmH2O and Thigh 30–60 secs).

Conventional volume ventilation lim-its recruitment to brief cyclic intervals atend-inspiration or plateau pressure. Lungregions that are recruited only duringbrief end-inspiratory pressure cycles pro-duce inadequate mean alveolar volume.Because alveolar volume is not main-tained, compliance does not improve, re-quiring the same inflation pressure onsubsequent breaths. Reapplication of thesame distending pressure without ade-quate lung recruitment is likely to pro-duce recurrent shear forces and does notattenuate potential lung injury (42). Con-versely, sustained recruitment is associ-ated with increased compliance allowingsuccessful, sequential airway pressure re-duction and improving gas exchange byincreasing alveolar surface area (4, 42–44). Increased alveolar surface area mayimprove stress distribution in the lung.

Gallagher and coworkers (45) demon-strated a direct correlation among meanairway pressure, lung volume, and oxy-genation. The use of APRV to optimizemean airway pressure/lung volume pro-vides a greater surface area for gas ex-change. Allowing sustained duration(Thigh) of Phigh and limiting duration andfrequency of the release phase (Tlow) ofPlow permits only partial emptying, lim-iting lung volume loss during ventilation.As lung recruitment is sustained, gas re-distribution and diffusion along concen-tration gradients have time to occur. Themixture of alveolar and inspired gaswithin the anatomic dead space results ina greater equilibration of gas concentra-tions in all lung regions, improved oxy-genation, and reduced dead-space venti-lation (26, 46) (Fig. 3).

Figure 1. Airway pressure release ventilation is a form of continuous positive airway pressure (CPAP).The Phigh is equivalent to a CPAP level; Thigh is the duration of Phigh. The CPAP phase (Phigh) isintermittently released to a Plow for a brief duration (Tlow) reestablishing the CPAP level on thesubsequent breath. Spontaneous breathing may be superimposed at both pressure levels and isindependent of time-cycling. Reprinted from ICON educational supplement 2004 with permission.


In addition to FIO2 and slope, clini-cian-controlled APRV parameters are:Phigh, Thigh, Plow, and Tlow. Time parame-ters in APRV are independent rather thanan inspiratory:expiratory (I:E) ratio al-lowing precise adjustment.

The Phigh and Thigh regulate end-inspiratory lung volume and provide asignificant contribution to the mean air-way pressure. Mean airway pressure cor-relates to mean alveolar volume and iscritical for maintaining an increased sur-

face area of open air spaces for diffusivegas movement. As a result, these param-eters control oxygenation and alveolarventilation. Counterintuitive to conven-tional concepts of ventilation, the exten-sion of Thigh can be associated with a

Table 1. Clinical studies using airway pressure-release ventilation (APRV)

Author(yr Published) Study Measurements Findings Study design

Stock (1987) APRV vs. IPPV; dogs with ALI(n � 10)

Blood gases, hemodynamics,lung volume, airwaypressure, f, VT, VE

Hemodynamics were notdifferent at equivalent VE; withAPRV, PIP and physiologicaldead space were lower, meanairway pressure was higher,and oxygenation was better

Animal study, small n, andshort-term observations

Garner (1988) APRV vs. conventional ventilation,patients after cardiac surgery(n � 14)


Similar oxygenation andventilation at lower peakairway pressure

Observational, crossovertrial

Rasanen (1988) APRV vs. conventional ventilationvs. CPAP; anesthetized dogs(n � 10)


APRV had similar effects onblood gases but withsignificantly fewer adversehemodynamic effects

Animal studies, small n,and short-termobservations

Martin (1991) APRV vs. CPAP vs. conventionalventilation vs. spontaneousbreathing; neonatal sheep witholeic-acid lung injury (n � 7)


APRV increased VE more thanCPAP; APRV provided similargas exchange to conventionalventilation, but with feweradverse hemodynamic effects


Davis (1993) APRV vs. SIMV; surgery patientswith ALI(n � 15)


APRV provided similar gasexchange with lower PIP, butno hemodynamic advantagewas identified

Prospective, crossover trialwith short-termobservations

Putensen (1994) APRV (with and withoutspontaneous breathing) vs.PSV; anesthetized dogs(n � 10)

Blood gases, hemodynamics,lung volume, airwaypressure, f, VT, VE,ventilation/perfusiondetermined by multipleinert-gas-eliminationtechnique

PSV had highest VE; APRV hadhigher cardiac output, PaO2,and oxygen delivery; APRV hadbetter V/Q and less dead space


Sydow (1994) APRV vs. volume-controlledinverse-ratio ventilation;patients with ALI; 24-hrobservation periods (n � 18)


APRV provided 30% lower PIP,less venous admixture (14 vs.21%), and better oxygenation;no difference inhemodynamics

Prospective, randomized,crossover trial

Calzia (1994) BiPAP vs. CPAP; patients afterbypass surgery (n � 19)

WOB and PTP No difference Prospective, crossover trial

Rathgeber (1997) BiPAP vs. conventional ventilationvs. SIMV; patients after cardiacsurgery (n � 596)

Duration of intubation,sedation requirement,analgesia requirement

APRV had shorter duration ofintubation (10 hrs) than SIMV(15 hrs) or conventionalventilation (13 hrs);conventional ventilation wasassociated with greater dosesof midazolam; APRV wasassociated with less need foranalgesia

Prospective, randomized,controlled, open trialover 18 months, unevenrandomization

Kazmaier (2000) BiPAP vs. SIMV vs. PSV; pPatientsafter coronary artery bypass(n � 24)


No differences in blood gases orhemodynamics

Prospective, crossover trialwith short-termobservations

Putensen (2001) APRV vs. pressure controlledconventional ventilation;patients with ALI after trauma(n � 30)

Gas exchange,hemodynamics, sedationrequirement,hemodynamic support,duration of ventilation,ICU stay

APRV was associated with fewerICU days, fewer ventilatordays, better gas exchange,better hemodynamicperformance, better lungcompliance, and less need forsedation and vasopressors

Randomized controlled,prospective trial, smalln; the conventionalventilation groupreceived paralysis for thefirst 3 days, potentiallyconfounding results

Varpula (2003) Combined effects of proning andSIMV PC/PS vs. APRV; patientswith ALI (n � 45)

Blood gases, oxygenation(PaO2/FIO2 ratio),hemodynamic, sedationrequirement

Oxygenation was significantlybetter in APRV group beforeand after proning; sedation useand hemodynamics weresimilar

Prospective, randomizedintervention study

IPPV, intermittent positive-pressure ventilation; ALI, acute lung injury; f, respiratory frequency; VE, minute volume; VT, tidal volume; PIP, peakinspiratory pressure; CPAP, continuous positive airway pressure; PSV, pressure support ventilation; V/Q, ventilation/perfusion ratio; BiPAP, bilevel positiveairway pressure; SIMV, synchronized intermittent mandatory ventilation; WOB, work of breathing; PTP, pressure-time product; ICU, intensive care unit.

Reprinted with permission from Branson RD, Johannigman JA: What is the evidence base for the newer ventilation modes? Respir Care 2004;49:742–760.


decrease in PaCO2 as machine frequencydecreases. This has been previously de-scribed and is similar to improved CO2

clearance with increasing I:E ratios (47–51). Despite the intermittent nature ofventilation, CO2 delivery to the lung iscontinuous as cardiac output transfersCO2 into the alveolar space, provided air-ways remain open (52). During the briefTlow, released gas is exchanged with freshgas to regenerate the gradient for CO2

diffusion. In addition, cardiogenic mixingresults in CO2 movement toward centralairways during the Thigh or breathholdperiod (46, 53–55), improving the efficacyof the release for ventilation. The addi-tion of spontaneous breaths during theThigh period at Phigh (higher lung volume)further enhances recruitment and venti-lation efficiency (Fig. 3).

The risk of using APRV as a cyclicmode and attempting to increase the ma-chine frequency and minute ventilationby reducing Thigh may sacrifice alveolarventilation and oxygenation. Reducing

Thigh will lead to a reduction in meanairway pressure, potentially resulting inairway closure, decreasing alveolar sur-face area for gas exchange.

In addition to spontaneous breathing,ventilation is augmented during APRV asa result of the release phase. The releasephase is determined by the driving pres-sure differential (Phigh - Plow), inspiratorylung volume (Phigh), the potential energy(recoil or compliance of the thorax andthe amount of energy stored duringThigh), and downstream resistance (artifi-cial airway). Plow and Tlow regulate end-expiratory lung volume and should beoptimized to reduce airway closure/derecruitment and not as a primary ven-tilation adjustment. Generally, to main-tain maximal recruitment, the majorityof the time or Thigh (80–95% of the totalcycle time) occurs at the Phigh or CPAPlevel. To minimize derecruitment, thetime (Tlow) at Plow is brief (usually be-tween 0.2 and 0.8 secs in adults).

Because patients can maintain theirnative respiratory drive during APRV,spontaneous inspiratory and expiratorytime intervals are independent of theThigh, Tlow cycle (56). Thus, the releasephase does not reflect the only expiratorytime during APRV when patients arebreathing spontaneously. Therefore,spontaneous expirations will occur at theupper pressure or Phigh phase. Active ex-halation during the Phigh phase may re-sult in additional recruitment and vol-ume redistribution analogous togrunting respiration in neonates, therebyimproving ventilation/perfusion (V/Q)matching (22, 57–61).

The release time (Tlow) may be titratedto maintain end-expiratory lung volume(EELV)/(end-release lung volume [ERLV]).The end-release lung volume can be ad-justed and continually assessed by usingthe expiratory flow pattern (Fig. 4). Theexpiratory gas flow is a result of the inspira-tory lung volume, the recoil or drive pres-sure of the lung, and downstream resis-tance (artificial airway, circuit, and PEEPvalve) (Fig. 5). Experimental data in a por-cine ALI model using dynamic computedtomography scanning shows that airwayclosure occurs rapidly (within 0.6 secs) (38,62). However, the rapid airway closure inpig models of ALI may be related to poorcollateral ventilation as opposed to humanlungs. Collateral ventilation may play a sig-nificant role in recruitment/derecruitmentin ALI (63).

Using a Plow of zero allows end-expiratory/release lung volume to be con-trolled by one parameter (time). The in-herent resistance of the artificial airwaybehaves as a flow resistor/limiter and, ifcoupled with a brief release time, caneffectively trap gas volume to maintainend-release or expiratory pressure(PEEP) (64, 65). During passive expira-tion or release in patients with ARDS,expiratory time constants are signifi-cantly modified (increased threefold) bythe flow-dependent resistance of the arti-ficial airway (66, 67).

Because the artificial airway produces anonlinear, flow-dependent resistive loadand the release results from a high lungvolume, flow resistance will be highest atthe initial portion of the release phase (67–69). The Tlow or release phase is terminatedT-PEFR rapidly before the flow-dependentexpiratory load is dissipated, resulting inend-expiratory volume and pressure.

The residual pressure and volume inthe lung during the brief release phasetypically yields end-release or end-

Figure 2. Ventilation during airway pressure release ventilation is augmented by release volumes andis associated with decreasing airway pressure and lung distension. Conversely, tidal volumes duringconventional ventilation are generated by increasing airway pressure and lung distension. Reprintedfrom ICON educational supplement 2004 with permission.

Figure 3. Gas exchange during airway pressure release ventilation. A, mean airway pressure (lungvolume) provides sustained mean alveolar volume for gas diffusion. B, alveolar gas volume combinedwith cardiac output provides continuous diffusive gas exchange between alveolar and blood compart-ments despite the cyclic nature of ventilation. C, CO2-enriched gas is released to accommodateoxygen-enriched gas delivered with the subsequent inspiratory cycle. New inspiratory volume isintroduced, regenerating diffusion gradients. Reprinted from ICON educational supplement 2004 withpermission.


expiratory pressure greater than the Plow

arbitrarily set at the machine’s peep valve(approximately 8 ft away). In fact, com-mercially available ventilators with tubecompensation algorithms for resistance

of artificial airways provide inadequateexpiratory compensation when PEEP isreduced to atmospheric pressure. The ad-dition of a negative pressure source tobriefly lower the end-expiratory pressure

to subatmospheric is required to fullycompensate the expiratory resistance im-posed by the artificial airway (70).

By using a Plow �0 cm H2O, peak expi-ratory flow rate (PEFR) is delayed, whereasa Plow of 0 cm H2O accelerates PEFR con-cluding the release phase earlier and en-abling the Phigh phase to be resumed earlierin the cycle. A greater percent of the cycletime at Thigh increases the potential forrecruitment, maintains lung volume, limitsderecruitment, and induces spontaneousbreathing. See Table 2 for goals, set up,oxygenation, ventilation, weaning, and pre-cautions during utilization of APRV.

Spontaneous Breathing DuringAirway Pressure ReleaseVentilation

During APRV, patients can control thefrequency and duration of spontaneousinspiration and expiration. Patients arenot confined to a preset I:E ratio, andspontaneous tidal volumes maintain a si-nusoidal flow pattern similar to normalspontaneous breaths. The ability of criti-cally ill patients to effectively augment

Figure 5. Patient interface to mechanical ventilator circuit and inherent resistance to expiratory flowfrom artificial airway (R1) and positive end-expiratory pressure (PEEP) valve (R2). Because the releaseoccurs from a high lung volume during airway pressure release ventilation, flow resistance developsat the distal end of R1 and R2. The proximal end of R1 decompresses more rapidly than the distal end.Despite zero end-expiratory pressure (ZEEP), flow resistance at R2 (typically measured approximately8 ft away) contributes to tracheal pressure elevation above end-expiratory pressure. Flow resistance ishighest at the onset of the release (�0.2 L/sec) and decreases as expiratory flow rate declines. Releasetime is terminated after a brief duration before flow resistance dissipates to maintain end-expiratorylung volume (67–69). Reprinted from ICON educational supplement 2004 with permission.

Figure 4. End-expiratory lung volume. Expiratory flow pattern during the release phase of airway pressure release ventilation. Initial portion of expiratoryflow limb is the peak expiratory flow rate (PEFR) as a result of Phigh to Plow pressure reduction. Deceleration of gas flow occurs as driving pressure dissipatesproducing the decelerating limb. PEFR and rate of deceleration are affected by inspiratory lung volume, thoracic recoil, and downstream resistance(artificial airway resistance with Plow of zero). A, expiratory flow pattern demonstrating normal deceleration at 45° as airways empty sequentially (67).Release time is adjusted to regulate T-PEFR to 60% of PEFR. The flow/time beyond the T-PEFR correlates with the end-release lung volume (ERLV)(end-expiratory lung volume [EELV]). The angle of deceleration (ADEC) can suggest alterations in lung mechanics resulting in altered expiratory gas flow([a] Normal (45°), [b] restrictive, e.g., decrease thoracic compliance (�45°) [c] obstructive, e.g., OLD, small or obstructed artificial airway (�45°)]. B,expiratory flow pattern with lung changes indicating derecruitment. As lung compliance worsens (less end-inspiratory lung volume and increasing thoracicrecoil, e.g., increased lung water or decreased thoracic/abdominal compliance), the expiratory flow pattern will become more restricted (the deceleratinglimb will become steeper, i.e., ADEC �45°) and the set release time will result in a lower T-PEFR and less end-expiratory lung volume (or ERLV); lowerERLV with resulting airway closure/derecruitment. C, the release time can be adjusted to limit airway closure and prevent derecruitment. D, conversely,as respiratory mechanics improve (lung, thoracic, abdominal compliance), recruitment is reflected as the decelerating limb returns to a 45° angle and theset release time increases the T-PEFR regulating ERLV (EELV). E, the release time can be readjusted to maintain T-PEFR at 50% to 75%. Reprinted fromICON educational supplement 2004 with permission.


Table 2. Airway pressure release ventilation (APRV) clinical guide

GoalsAcute (recruitable) restrictive lung disease (RLD)

Increase (recruit) and maintain lung volume (Phigh and Thigh)Decrease elastic WOB with CPAP (Phigh and Thigh)ATC set at 100% to maximally compensate for artificial airway resistance and decrease resistive WOB imposed by the artificial airwaya

Minimize number of releases to supplement ventilation from spontaneous breathing (71)Limit derecruitment; Tlow set to ensure T-PEFR is �50 and �75%Allow spontaneous breathing within 24 hrs of APRV application

Acute obstructive lung disease (OLD)Decrease lung volumeMaintain Phigh at or 1–2 cm H2O above PEEPiMinimize number of releases to supplement ventilation from spontaneous breathing (71)Stint airways; Tlow set for 25–50% T-PEFR

Allow spontaneous breathing within 24 hrs of APRV application. May require a brief course of NMBA (�24 hrs) to control high spontaneousbreathing frequency and artificial airway contribution to dynamic hyperinflation.

Set-up—adultsNewly intubated

Phigh—set at desired plateau pressure (typically 20–35 cm H2O)Note: Phigh �35 cm H2O may be necessary in patients with decreased thoracic/abdominal compliance or morbid obesity (73, 74). With a Phigh

�25 cm H2O, use of noncompliant ventilator circuit is recommended to minimize circuit volume compression (75, 76).Plow—0 cm H2OThigh—4–6 secsTlow—0.2–0.8 secs (RLD)0.8–1.5 secs (OLD)

Transition from conventional ventilationPhigh—plateau pressure in volume-cycled mode or peak airway pressure in pressure-cycled modePlow—0 cm H2OThigh—4–6 secsTlow—0.2–0.8 secs (RLD)0.8–1.5 secs (OLD)

Transition from HFOVb—use noncompliant ventilator circuitPhigh—mPaw on HFOV plus 2–4 cm H2OPlow—0 cm H2OThigh—4–6 secsTlow—0.2–0.8 secs (RLD)0.8–1.5 secs (OLD)

Set-up—PediatricsNewly intubated

Phigh—set at desired plateau pressure (typically 20–30 cm H2O)Note: Phigh �30 cm H2O may be necessary in patients with decreased thoracic/abdominal compliance or morbid obesity (73, 74). With a Phigh

�25 cm H2O, use of noncompliant ventilator circuit is recommended to minimize circuit volume compression (75, 76).Plow—0 cm H2OThigh—3–5 secsTlow—0.2–0.8

Transition from conventional ventilationPhigh—plateau pressure in volume-cycled mode or peak airway pressure in pressure-cycled modePlow—0 cm H2OThigh—3–5 secsTlow—0.2–0.8

Transition from HFOVb

Phigh—mPaw on HFOV plus 2–4 cm H2OPlow—0 cm H2OThigh—3–5 secsTlow—0.2–0.8

Set-up—NeonatesNewly intubated

Phigh—set at desired plateau pressure (typically 10–25 cm H2O)Note: Phigh �25 cm H2O may be necessary in patients with decreased thoracic/abdominal compliance (73, 74). With a Phigh�25 cm H2O,

use of noncompliant ventilator circuit is recommended to minimize circuit volume compression (75, 76).Plow—0 cm H2OThigh—2–3 secsTlow—0.2–0.4

Transition from conventional ventilationPhigh—plateau pressure in volume-cycled mode or peak airway pressure in pressure-cycled modePlow—0 cm H2OThigh—2–3 secsTlow—0.2–0.4

Transition from HFOVb

Phigh—mPaw on HFOV plus 0–2 cm H2OPlow—0 cm H2OThigh—2–3 secsTlow—0.2–0.4

Continues


spontaneous ventilation in response tochanging metabolic needs may promotesynchrony during mechanical ventilationand improve V/Q matching (3–6, 10, 11).In contrast, patients transitioned fromspontaneous breathing to mechanicalventilation through the induction of an-esthesia exhibit worsening gas exchangeand dependent atelectasis on computedtomography scan within minutes (7–9,76–83). These studies suggest rapid al-teration of ventilation distribution whenthe respiratory system becomes passive.

Most mechanical ventilators monitorairway pressures; however, transpulmo-nary pressures ultimately determine lungvolume change. Although difficult tomonitor clinically, the effects of pleuralpressure on transpulmonary pressuresshould not be excluded from manage-ment principles. For example, patientswith reduced thoracic and abdominalcompliance demonstrate higher airwaypressure yet may have lower transpulmo-nary pressure (73, 74).

Spontaneous breathing, diaphrag-matic tone, and prone positioning modify

pleural pressure, improving transalveolarpressure gradients in dependent lung re-gions (7, 84–87). Increased dependentlung ventilation during spontaneousbreathing recruits alveoli improving V/Qmatching without raising applied airwaypressure (3–6, 10, 11).

APRV and prone positioning mayhave an additive effect on recruitmentand gas exchange. Varpula demon-strated greater improvement in gas ex-change when prone positioning wascombined with APRV rather than syn-chronized intermittent mandatory ven-tilation (10).

Hemodynamic Effects of AirwayPressure Release Ventilation

The descent of the diaphragm into theabdomen during a spontaneous breathingeffort simultaneously decreases pleuralpressures and increases abdominal pres-sure. This effectively lowers the right atrial(RA) pressure while compressing abdomi-nal viscera propelling blood (preload) intothe inferior vena cava (IVC). Increasing the

mean systemic pressure (MSP)/RA gradientcouples the thoracic and cardiac pumps,increasing venous return, improving car-diac output, and decreasing dead space ven-tilation (88, 89). Conversely, when sponta-neous breathing is limited or thediaphragm is paralyzed, the passive decentof the diaphragm is no longer linked withlower pleural/right atrial pressure, mini-mizing the IVC–right atrial pressure gradi-ent (MSP-RA) and limiting venous return/cardiac output.

Restoration of cardiopulmonary inter-action with spontaneous breathing dur-ing APRV produces improvements in sys-temic perfusion. Animal and humanstudies document improved splanchnicand renal perfusion during APRV withspontaneous breathing (13, 90)

Use of Pressure SupportVentilation with Airway PressureRelease Ventilation

Currently, some ventilator manufac-turers incorporate pressure support ven-tilation (PSV) above Phigh. The addition of

Table 2. Continued

OxygenationOptimize end-expiratory or release lung volume

Reassess release volume to ensure T-PEFR is �50 and �75%If oxygenation poor and T-PEFR �50%, decrease release time until T-PEFR 75%Optimize gas exchange surface area by adjusting mPaw

Increase Phigh or Phigh and Thigh simultaneouslyAdjustment of Phigh to recruit by achieving TOPAdjustment of Thigh increases gas mixing and recruits lung units with high resistance time constants

Assess hemodynamicsVentilation

Assess for oversedation; consider using sedation scale (77)Optimize end-expiratory or release lung volume; reassess release volume to ensure at 50–75% T-PEFRIf T-PEFR �75% and oxygenation is acceptable, consider increasing Tlow by 0.05–0.1 increments to achieve 50% T-PEFRIf T-PEFR �50%, decrease Tlow to achieve minimum T-PEFR of 50%Increase alveolar ventilation (preferred method)—increase Phigh or Phigh and Thigh simultaneouslyIncrease minute ventilation—decrease Thigh and increase Phigh simultaneously (see precautions below)

WeaningSimultaneously reduce Phigh and increase Thigh for a gradual reduction of mPaw and to increase the contribution of spontaneous to total minute

ventilation.Progress to CPAP with automatic tube compensation when Phigh �16 and Thigh �12–15 sec (APRV � 90% CPAP)Wean CPAP (with automatic tube compensation) and consider extubation when CPAP 5–10 cm H2O

PrecautionsAdjustment of Tlow differs with lung disease, lung volume and artificial airway size. Tlow values provided are typical but not absolute; see goals forOLD and RLDIf minute ventilation is increased by decreasing Thigh in an attempt to improve CO2 clearance, mPaw and gas exchanging surface area will be

reduced; more so if Phigh is not simultaneously increased as CO2 may paradoxically increase (see text for details). May need to decrease Tlow asThigh reduction may produce less mean alveolar volume (lung volume) and will result in shorter emptying time.

Tlow should not be extended solely to lower CO2 as this may lead to airway closure (derecruitment) (38, 56, 71). Additionally, Tlow should not beviewed as an expiratory time as the patient may exhale throughout the respiratory cycle if permitted (58).

ATC, automatic tube compensation; PEEPi, intrinsic PEEP; mPaw, mean airway pressure; PEFR, peak expiratory flow rate; T-PEFR, peak expiratory flowrate termination; HFOV, high-frequency oscillatory ventilation; WOB, work of breathing; APRV, airway pressure release ventilation; TOP, threshold openingpressure.

aIn vitro resistance may be greater than in vivo resistance calculations and measurements (commercial tube compensation algorithms) due todeformation, kinks and secretion in the artificial airway (64, 65, 72); bmPaw during HFOV is equal to CPAP; mPaw during APRV is typically 2–4 cm H2Oless than CPAP as a result of the release phase (airway pressure interruption) and proximal vs. distal mPaw measurement. Reprinted from ICON EducationalSupplement—2004 with permission.


PSV to APRV contradicts limiting airwaypressure and lung distension during ven-tilation by not restricting lung inflationto the Phigh level.

PSV above Phigh may lead to signifi-cant elevation in transpulmonary pres-sure (Fig. 6). When PSV is triggeredduring the Phigh phase, the higher base-line lung volume distends further as thesum of Phigh, PSV, and pleural pressureraises transpulmonary pressure. Theadditional lung distension above Phigh

and the transpulmonary pressure eleva-tion will not be completely reflected inthe airway pressure because the pleuralpressure remains unknown (91). Fur-thermore, the imposition of PSV toAPRV reduces the benefits of spontane-ous breathing by altering sinusoidalspontaneous breaths to decelerating as-sisted mechanical breaths as flow andpressure development are uncoupledfrom patient effort (Fig. 7). Ultimately,PSV during APRV defeats improvementsin distribution of ventilation and V/Qmatching associated with unassistedspontaneous breathing (4, 6, 26, 92–94). During weaning, even low levels ofPSV used to overcome tube resistancemay overcompensate and convert pa-tient-triggered efforts to assisted rather

than spontaneous breaths, especially ifadequate PEEP levels are used (95, 96).If patient efforts are more vigorous, thePSV will undercompensate artificial air-way resistance.

Artificial Airway CompensationAlgorithms and Airway PressureRelease Ventilation

Computerized ventilator algorithms,which attempt to match inspiratory flow tocalculated resistance of the artificial airwayduring APRV may reduce spontaneousWOB (14). Unlike PSV, tube compensationalgorithms apply inspiratory flow in pro-portion to the pressure drop across theartificial airway resulting from patienteffort or flow demands (95, 96). As aresult, the dynamic pressure applied tothe artificial airway is determinedwithin the breath cycle, limiting over-or undercompensation of artificial air-way resistance. Furthermore, as tubecompensation is coupled to patient ef-fort, flow and resulting applied airwaypressure do not exceed inspiratory pres-sure generated by the respiratory mus-cles (Pmus) (patient’s effort), preservingthe sinusoidal flow pattern of spontane-ous breathing (97) (Fig. 6). Conversely,

applying a fixed airway pressure likePSV may result in both over- and un-dercompensation of artificial airway re-sistance as patient effort (flow) varies.

Commercially available ventilators offerforms of tube compensation but vary in theefficiency of the algorithms applied. Al-though many ventilators may compensateinspiratory resistance effectively, expiratorycompensation by lowering PEEP levels toatmosphere or ZEEP (at the initial phase ofexpiration) may not unload expiratory re-sistance imposed by the artificial airway.The application of negative airway pressureduring the initial expiration phase may benecessary to negate the pressure dropacross the artificial airway (70).

Airway Pressure ReleaseVentilation and Use of Sedationand Neuromuscular-BlockingAgents

Sedation is essential when caring forcritically ill and injured patients re-quiring mechanical ventilation. In se-vere cases of patient–ventilator dys-synchrony, neuromuscular-blockingagents (NMBAs) are frequently used.Excessive sedation has been associatedwith increased duration of mechanicalventilation in patients with acute respi-ratory failure (98 –103). Reducing the

Figure 7. Gas flow pattern comparing pressuresupport ventilation (PSV) and spontaneousbreathing. PSV produces a decelerating gas flowpattern because the gas flow and patient effort (PMus) do not follow a similar time course. Typi-cally, patient effort is outpaced by applied flowand pressure development. Gas distribution dur-ing PSV is similar to an assisted breath ratherthan a spontaneous breath. Spontaneous or un-assisted continuous positive airway pressure(CPAP) breath producing a sinusoidal gas flowpattern as gas flow and patient effort (P Mus) arecoupled and follow a similar time course. Sponta-neous ventilation (unassisted) is associated withimproved ventilation/perfusion distribution, unlikePSV (4, 6, 26, 92–94). Reprinted from ICON educa-tional supplement 2004 with permission.

Figure 6. A, pressure tracing represents a pressure support ventilation (PSV) breath at the Phigh level.During passive efforts, once the PSV trigger threshold is reached, airway pressure elevates above thePhigh level. Alternatively, if the patient generates vigorous inspiratory efforts, the transpulmonary pressuredifferential (sum of PSV, Phigh, Pmus) can be significant and may result in overdistension. B, flow tracingdemonstrates a triggered PSV breath with resultant decelerating gas flow as opposed to sinusoidal gas flowtypical of spontaneous breathing (see D). C, pressure tracing represents airway pressure release ventilation(APRV) with automatic tube compensation; note minimal airway pressure elevation above Phigh. D,spontaneous breaths during APRV with automatic tube compensation, preserving a sinusoidal flow pattern.Reprinted from ICON educational supplement 2004 with permission.


duration of mechanical ventilation de-creases patient exposure to artificialairways, sedation, and NMBAs and thelikelihood of ventilator-associatedpneumonia (VAP) (101, 104 –106).

The negative impact of sedation andNMBAs on the duration of mechanicalventilation and the risk of VAP is likely tobe in part related to depression of thecough reflex, increasing the risk of aspi-ration of pharyngeal secretions (107).Watando suggested that improved coughreflex may limit aspiration pneumonia inhigh-risk groups (108). Furthermore, aneffective cough may be a predictor of hos-pital mortality and of successful extuba-tion (109).

Because APRV uses an open breathingsystem and requires less sedation, pa-tients can exhale or cough throughoutthe respiratory cycle. As a result, coughand secretion clearance can be facilitatedwithout significant intrathoracic pres-sure elevation or airway pressure-limit-ing as would occur with a closed expira-tory valve system.

APRV has been associated with a 70%reduction in NMBA requirements and a30% to 40% reduction in sedation re-quirements when compared with conven-tional ventilation (3, 4, 11, 12, 23, 25,110). In addition, some studies suggest adecrease in ventilator days and intensivecare unit and hospital length of stay as aresult of using APRV (4).

The ARDS Network (ARDSNet) re-ported a significant reduction of mortal-ity from 39.8% to 31.0% with a low tidalvolume strategy (6 mL/kg of ideal bodyweight) and a limited inspiratory plateaupressure (30 cm H2O) using a volume-cycled mode (111). However, patientsventilated using low tidal volumes mayexperience more dyssynchrony and re-quire additional sedation (Kallet RH, per-sonal communication) (112–118).

Airway Pressure ReleaseVentilation for Trauma-Associated Acute RespiratoryDistress Syndrome: ClinicalExperience

APRV has been used at R Adams Cow-ley Shock Trauma Center (STC) in Balti-more, MD, since 1994 and has become astandard of care. In the early 1990s, STCestablished a regional advanced respira-tory failure service, including the devel-opment of ventilation protocols aimed toreduce airway pressure with APRV, pronepositioning, and an extracorporeal lung

assist technique (119). The STC haslogged over 50,000 patient-hours annu-ally on APRV since 1994, developing sig-nificant clinical experience with APRV. AtSTC, the 2-yr period post-APRV imple-mentation for the management of ad-vanced respiratory failure was studiedand documented a reduction in ARDSmortality and multisystem organ failure.The mortality rates after the implemen-tation of APRV in patients meeting crite-ria for ARDS were lower than reported inthe ARDSNet trial, 21.4% vs. 31% (120).In addition, sedation requirements werereduced and NMBA use essentially elimi-nated from routine practice at STC.

Weaning from Airway PressureRelease Ventilation

Patients with improved oxygenationon APRV (e.g., FIO2 �40% with SpO2

�95%) can be progressively weaned bylowering the Phigh and extending theThigh. By decreasing the number of re-leases, the minute ventilation output ofthe ventilator is reduced while simulta-neously (if permitted) the patient’s spon-taneous minute ventilation increases,enabling a progressive spontaneousbreathing trial (99) (Fig. 8). The total andspontaneous minute ventilation shouldbe carefully monitored during weaning toanticipate changes in PaCO2. Eventually,the ventilator’s minute ventilation outputis significantly reduced or eliminated andthe patient has gradually transitioned topure CPAP. CPAP, when combined withtube compensation, can be used to effec-tively overcome artificial airway resis-tance during the final phase of weaning.When used in the final weaning phase,tube compensation may be a useful pre-dictor of successful extubation, particu-larly in the difficult-to-wean patient whofails PSV and T-piece weaning methods(96). This author believes that progres-sive extension of Thigh during APRV wean-ing increases spontaneous breathingthrough a gradual transition to pureCPAP (with tube compensation). There-fore, transitioning the weaning APRV pa-tient to PSV (a form of assisted breathing)may be counterproductive and unneces-sary (121).

Airway Pressure ReleaseVentilation and High-FrequencyOscillatory Ventilation

Fundamentally APRV and high-fre-quency oscillatory ventilation (HFOV)

have similar goals. Both techniques focuson maintaining lung volume while limit-ing the peak ventilating pressure. Main-taining lung volume optimizes V/Qmatching, improves gas exchange, andimproves stress distribution, minimizingshear forces. During HFOV, the continu-ous, high-flow gas pattern facilitates aconstant airway pressure profile mini-mizing derecruitment. In contrast toHFOV, APRV actively promotes spontane-ous breathing.

In addition, APRV does not require asingle-purpose ventilator, effectively usesconventional humidification systems,and is associated with reduced sedationand NMBA use. Furthermore, becauseALI can develop in 24% of patients receiv-ing mechanical ventilation who did nothave ALI at the onset (122), APRV as alung protective strategy, may be used ear-lier rather than at advanced stages ofrespiratory failure. APRV can be appliedas the initial ventilator mode for respira-tory failure or typically before HFOV cri-teria are reached. For patients showingimprovement on HFOV, APRV may rep-resent an ideal transition/weaning modal-ity because Phigh on APRV can be matchedto the mean airway pressure (mPaw) dur-ing HFOV, permitting continued gradualreduction of lung volume (123).

Noninvasive Ventilation withAirway Pressure ReleaseVentilation

APRV may also be applied noninva-sively pre- or postintubation (124). Non-invasive APRV has the advantage of anadjustable degree of mandatory ventila-tion without the need for a trigger (onlyforms of APRV that do not use pressuresupport), decreasing the likelihood of au-tocycling from leaks common to nonin-vasive ventilation.

Future clinical research with APRVshould test different algorithms for oxy-genation, ventilation, and weaning. In ad-dition, hemodynamic and systemic perfu-sion during APRV should be assessed.Animal studies should be performed withAPRV (coupled with spontaneous breath-ing) to compare histologic and biomarkerevidence of VILI/VALI compared withother mechanical ventilation approaches.The delivery of aerosol medications to thelung is poorly understood during APRVand should be studied. Ultimately, clini-cal studies should be conducted using avalidated, protocolized approach to APRV


compared with a protective conventionalventilation strategy.

CONCLUSION

Clinical and experimental studieswith APRV demonstrate improvementsin physiological end points such as gasexchange, cardiac output, and systemicblood flow (3, 4, 6, 11, 13, 26, 90). APRVfacilitates spontaneous breathing andimproves patient tolerance to mechan-ical ventilation by decreasing patient–ventilator dyssynchrony. Additionalstudies document reduction in sedationand NMBAs with APRV, and some, butnot all (10), studies suggest less venti-lator days and shorter length of inten-sive care unit stay (3, 4, 11, 12, 23, 25,111). An adequately designed and pow-ered study to demonstrate a reductionin mortality or ventilator days withAPRV compared with optimal lung pro-tective conventional ventilation has notyet been performed. APRV (combinedwith tube compensation software) re-mains unique among potential “open

lung” approaches to lung protectivemechanical ventilation with the abilityto facilitate spontaneous breathing.

ACKNOWLEDGMENTS

We thank John Downs for review ofthe manuscript and Penny Andrews forreview and assistance with the manu-script; Ulf Borg for review of the manu-script; and Alastair A. Hutchison for ob-servation and contribution on neonatalbreathing pattern during airway pressurerelease ventilation.

REFERENCES

1. Downs JB, Stock MC: Airway pressure re-lease ventilation: a new concept in ventila-tory support. Crit Care Med 1987; 15:459–461

2. Stock CM, Downs JB: Airway pressure re-lease ventilation. Crit Care Med 1987; 15:462–466

3. Sydow M, Burchardi H, Ephraim E, et al:Long-term effects of two different ventila-tory modes on oxygenation in acute lunginjury. Comparison of airway pressure re-

lease ventilation and volume-controlled in-verse ratio ventilation. Am J Respir CritCare Med 1994; 149:1550–1556

4. Putensen C, Zech S, Wrigge H, et al: Long-term effects of spontaneous breathing dur-ing ventilatory support in patients withacute lung injury. Am J Respir Crit CareMed 2001; 164:43–49

5. Wrigge H, Zinserling J, Neumann P, et al:Spontaneous breathing improves lung aer-ation in oleic acid-induced lung injury. An-esthesiology 2003; 99:376–384

6. Putensen C, Mutz N, Putensen-Himmer G,et al: Spontaneous breathing during venti-latory support improves ventilation-perfu-sion distributions in patients with acute re-spiratory distress syndrome. Am J RespirCrit Care Med 1999; 159:1241–1248

7. Froese A, Bryan C: Effects of anesthesia andparalysis on diaphragmatic mechanics inman. Anesthesiology 1974; 41:242–255

8. Rehder K, Sessler AD: Regional intrapulmo-nary gas distribution in awake and anesthe-tized-paralyzed man. J Appl Physiol 1977;42:391–402

9. Tokics L, Hedenstierna G, Svensson L, et al:V/Q distribution and correlation to atelec-tasis in anesthetized paralyzed humans.J Appl Physiol 1996; 81:1822–1833

Figure 8. Airway pressure release ventilation weaning and gradual transition to pure continuous positive airway pressure (CPAP). Tube compensation maybe used with the onset of spontaneous breathing and is not limited to the final weaning phase. A, sequentially decreasing Phigh and increasing Thigh

simultaneously results in a gradual reduction of mean airway pressure. Mean airway pressure is reshaped to produce a lower and extended pressure profile.B, patient’s increased spontaneous minute ventilation increases as fewer releases in Phigh contribute less to the total minute ventilation. Eventually, therelease phase provides minimal contribution to the total minute ventilation, and the patient has transitioned to CPAP (tube compensate remains active).Reprinted from ICON educational supplement 2004 with permission.


10. Varpula T, Jousela I, Niemi R, et al: Com-bined effects of prone positioning and air-way pressure release ventilation on gas ex-change inpatients with acute lung injury.Acta Anasthesiol Scand 2003; 47:516–524

11. Kaplan LF, Bailey H, Formosa V: Airwaypressure release ventilation increases car-diac performance in patients with acutelung injury/adult respiratory distress syn-drome. Crit Care 2001; 5:221–226

12. Rathgeber J, Schorn B, Falk V, et al: Theinfluence of controlled mandatory ventila-tion (CMV), intermittent mandatory venti-lation (IMV) and biphasic intermittent pos-itive airway pressure (BIPAP) on duration ofintubation and consumption of analgesicsand sedatives: a prospective analysis in 596patients following adult cardiac surgery.Eur J Anaesthesiol 1997; 14:576–582

13. Hering R, Peters D, Zinserling J, et al: Ef-fects of spontaneous breathing during air-way pressure release ventilation on renalperfusion and function in patients withlung injury. Intensive Care Med 2002; 28:1426–1433

14. Wrigge H, Zinserling J, Hering R, et al:Cardiorespiratory effects of automatic tubecompensation during airway pressure re-lease ventilation in patients with acute lunginjury. Anesthesiology 2001; 95:382–389

15. Hales R, Colborn S, Tyler L, et al: Retro-spective review in pediatrics: response timeto airway pressure release ventilation(APRV) [Abstract]. Respiratory Care 2004;49:1441

16. De Carvalho WB, Kopelman BI, GurgueiraGL, et al: Airway pressure release in post-operative cardiac surgery in pediatric pa-tients. Rev Assoc Med Bras 2000; 46:166–173

17. Foland JA, Martin J, Novotny T, et al: Airwaypressure release ventilation with a shortrelease time in a child with acute respira-tory distress syndrome. Respiratory Care2001; 46:1019–1023

18. Martin LD, Wetzel RC, Bilenki AL: Airwaypressure release ventilation in a neonatallamb model of acute lung injury. Crit CareMed 1991; 19:373–378

19. Schultz TR, Costarino AT, Durning SM, etal: Airway pressure release ventilation inpediatrics. Pediatr Crit Care Med 2001;2:243–246

20. Crooke C, Blake B, Aucott S: Case studycomparing airway release ventilation(APRV) to synchronized intermittent man-datory ventilation (SIMV) in the critically illneonate [Abstract]. Respir Care 2004; 49:1376

21. Jones R, Roberts T, Christensen: Airwaypressure release ventilation in pediatricpopulation [Abstract]. Respir Care 2004; 49:1414

22. Hutchison AA, Leaderstorf MR, HabashiNM: Airway pressure release ventilation in apreterm infant with severe bronchopulmo-nary dysplasia. 10th Annual REaSoN meet-

ing; June 27–28, 2004; University of War-wick, Coventry, UK; Abstract

23. Garner W, Downs JB, Stock MC, et al: Air-way pressure release ventilation (APRV). Ahuman trail. Chest 1988; 94:779–781

24. Rasanen J, Down JB, Stock MC: Cardiovas-cular effects of conventional positive pres-sure ventilation and airway pressure releaseventilation. Chest 1988; 93:911–915

25. Davis K, Johnson DJ, Branson RD, et al:Airway pressure release ventilation. ArchSurg 1993; 128:1348–1352

26. Putensen C, Rasanen J, Lopez FA, et al:Effect of interfacing between spontaneousbreathing and mechanical cycles on theventilation–perfusion distribution in caninelung injury. Anesthesiology 1994; 81:921–930

27. Calzia E, Lindner KH, Witt S, et al: Pres-sure-time product and work of breathingduring biphasic continuous positive airwaypressure and assisted spontaneous breath-ing. Am J Respir Crit Care Med 1994; 150:904–910

28. Kazmaier S, Rathgeber J, Buhre W, et al:Comparison of ventilatory and haemody-namic effects of BIPAP and S-IMV/PSV forpostoperative short-term ventilation in pa-tients after coronary artery bypass grafting.Eur J Anaesthesiol 2000; 17:601–610

29. Katz JA, Marks JD: Inspiratory work withand without continuous positive airwaypressure in patients with acute respiratoryfailure. Anesthesiology 1985; 63:598–607

30. Matamis D, Lemaire F, Harf A, et al: Totalrespiratory pressure-volume curves in theadult respiratory distress syndrome. Chest1984; 86:58–66

31. Rimensberger P, Cox P, Frndova H, et al:The open lung during small tidal volumeventilation: concepts of recruitment and‘optimal’ positive end-expiratory pressure.Crit Care Med 1999; 27:1946–1952

32. Hickling KG: The pressure-volume curve isgreatly modified by recruitment a mathe-matical model of ARDS lungs. Am J RespirCrit Care Med 1998; 158:194–202

33. Gattinoni L, Caironi P, Pelosi P, et al: Whathas computerized tomography taught usabout acute respiratory distress syndrome?Am J Respir Crit Care Med 2001; 164:1701–1711

34. Pelosi P, Goldner M, McKibben A, et al:Recruitment and derecruitment duringacute respiratory failure: an experimentalstudy. Am J Respir Crit Care Med 2001;164:122–130

35. Crotti S, Mascheroni D, Caironi P, et al:Recruitment and derecruitment duringacute respiratory failure: a clinical study.Am J Respir Crit Care Med 2001; 164:131–140

36. Greaves IA, Hildebrandt J, Hoppin FG: Mi-cromechanics of the lung. In: Handbook ofPhysiology, Third Edition. Macklem PT,Ead J (Eds). Bethesda, MD: American Phys-iological Society, 1986

37. Yap DYK, Liebkemann WD, Solway J, et al:

Influences of parenchymal tethering on thereopening of closed pulmonary airways.J Appl Physiol 1994; 75:2095–2105

38. Markstaller K, Eberle B, Kauczor H, et al:Temporal dynamics of lung aeration deter-mined by dynamic CT in a porcine model ofARDS. Br J Anaesth 2001; 87:459–468

39. Foti, Cereda M, Sparacino ME, et al: Effectsof periodic lung recruitment maneuvers ongas exchange and respiratory mechanics inmechanically ventilated acute respiratorydistress syndrome (ARDS) patients. Inten-sive Care Med, 2000; 26:501–507

40. Oczenski W, Hormann C, Keller C, et al:Recruitment maneuvers after a positiveend-expiratory pressure trial do not inducesustained effects in early adult respiratorydistress syndrome. Anesthesiology 2004;101:620–625

41. Medoff BD, Harris RS, Kesselman H, et al:Use of recruitment maneuvers and high-positive end-expiratory pressure in a patientwith acute respiratory distress syndrome.Crit Care Med 2000; 28:1210–1216

42. Van Kaam A, Haitsma J, DeJaegere A, et al:Open lung ventilation improves gas ex-change and attenuates secondary lung in-jury in a piglet model of meconium aspira-tion. Crit Care Med 2004; 32:443–449

43. Amato M, Barbas C, Medeiros D, et al: Effectof a protective-ventilation strategy on mor-tality in the acute respiratory distress syn-drome. N Engl J Med 1998; 338:347–354

44. Schreiter D, Reske A, Stichert B, et al: Al-veolar recruitment in combination withsufficient positive end-expiratory pressureincreases oxygenation and lung aeration inpatients with severe chest trauma. Crit CareMed 2004; 32:968–975

45. Gallagher TJ, Banner MJ: Mean airway pres-sure as a determinant of oxygenation. CritCare Med 1980; 8:244

46. Engel L, Menkes L, Wood L, et al: Gasmixing during breath holding studied byintrapulmonary gas sampling. J ApplPhysiol 1973; 35:9–17

47. Mercat A, Diehl J, Michard F, et al: Extend-ing inspiratory time in acute respiratorydistress syndrome. Crit Care Med 2001; 29:40–44

48. Knelson JH, Howatt WF, De Muth GR: Ef-fect of respiratory pattern on alveolar gasexchange. J Appl Physiol 1970; 29:328–331

49. Fuleihan SF, Wilson RS, Pontoppidan H:Effect of mechanical ventilation with end-inspiratory pause on blood-gas exchange.Anesth Analg 1976; 55:122–130

50. Pesenti A, Marcolin R, Prato P, et al: Meanairway pressure vs positive end-expiratorypressure during mechanical ventilation.Crit Care Med 1985; 13:34–37

51. Cole AGH, Weller SF, Sykes MK: Inverseratio ventilation compared with PEEP inadult respiratory failure. Intensive CareMed 1984; 10:227–232

52. Rose G, Cassidy S, Johnson R: Diffusingcapacity at different lung volumes during


breath holding and rebreathing. J ApplPhysiol 1979; 49:32–37

53. Fredberg JJ: Augmented diffusion in theairways can support pulmonary gas ex-change. J Appl Physiol 1980; 49:232–238

54. Fukuchi Y, Roussos CS, Macklem PT, et al:Convection, diffusion and cardiogenic mix-ing of inspired gas in the lung; an experi-mental approach. Respir Physiol 1976; 26:77–90

55. Haycroft JB, Edie R: The cardiopneumaticmovements. J Physiol 1891; 12:426–437

56. Neumann P, Golisch W, Strohmeyer A, etal: Influence of different release times onspontaneous breathing pattern during air-way pressure release ventilation. IntensiveCare Med 2002; 28:1742–1749

57. Kosch PC, Stark AR: Dynamic maintenanceof endexpiratory lung volume in full terminfants. J Appl Physiol 1984; 57:1126–1133

58. Chernick V: Continuous distending pres-sure in hyaline membrane disease. Pediat-rics 1973; 52:114–115

59. Harrison VC, Heese HB, Klein M: The sig-nificance of grunting in hyaline membranedisease. Pediatrics 1968; 41:549

60. Martin RJ, Okken A, Katona PG, et al: Effectof lung volume on expiratory time in thenewborn infant. J Appl Physiol 1978; 45:18–23

61. Gregory GA, Klitterman JA, Phibbs RH:Treatment of idiopathic respiratory syn-drome with continuous positive airwaypressure. N Engl J Med 1971; 284:1333–1339

62. Neumann P, Berglund JE, Mondéjar EF, etal: Effect of different pressure levels on thedynamics of lung collapse and recruitmentin oleic-acid-induced lung injury. Am J Re-spir Crit Care Med 1998; 158:1636–1643

63. Inners CF, Terry PB, Traystman RJ, et al:Collateral ventilation and the middle lobesyndrome. Am Rev Respir Dis 1978; 118:305–310

64. Wright PE, Marini JJ, Bernard GR: In vitroversus in vivo comparison of endotrachealtube airflow resistance. Am Rev Respir Dis1989; 140:10–16

65. Bersten AD, Rutten AJ, Vedig AE, et al:Additional work of breathing imposed byendotracheal tubes, breathing circuits, andintensive care ventilators. Crit Care Med1989; 17:671–677

66. Marini JJ, Culver BH, Kirk W: Flow resis-tance of exhalation valves and positive end-expiratory pressure devices used in me-chanical ventilation. Am Rev Respir Dis1985; 131:850–854

67. Guttmann J, Eberhard L, Fabry B, et al:Time constant/volume relationship of pas-sive expiration in mechanically ventilatedARDS patients. Eur Respir 1995; 8:114–120

68. Guttmann J, Eberhard L, Fabry B, et al:Continuous calculation of intratrachealpressure in tracheally intubated patients.Anesthesiology 1993; 79:503–513

69. Koutsoukou A, Armaganidis A, Stavrakaki-Kallergi C, et al: Expiratory flow limitation

and intrinsic positive end-expiratory pres-sure at zero positive end-expiratory pres-sure in patients with adult respiratory dis-tress syndrome. Am J Respir Crit Care Med2000; 161:1590–1596

70. Elsasser S, Guttmann J, Stocker R, et al:Accuracy of automatic tube compensationin new-generation mechanical ventilators.Crit Care Med 2003; 31:2619–2626

71. Neumann P, Hedenstierna G: Ventilatorsupport by continuous positive airway pres-sure breathing improves gas exchange ascompared with partial ventilatory supportwith airway pressure release ventilation.Anesth Analg 2001; 92:950–958

72. Shah C, Kollef M: Endotracheal tube in-traluminal volume loss among mechani-cally ventilated patients. Crit Care Med2004; 32:120–125

73. Malbrain ML: Abdominal pressure in thecritically ill: measurement and clinical rel-evance. Intensive Care Med 1999; 25:1453–1458

74. Ranieri VM, Brienza N, Santostasi S, et al:Impairment of lung and chest wall mechan-ics in patients with acute respiratory dis-tress syndrome: role of abdominal disten-sion. Am J Respir Crit Care Med 1997; 156:1082–1091

75. Verheecke G, Gilbertson A: Compliance oftubing compartment of lung ventilators. In-tensive Care Med 1981; 7:309–310

76. Bartel LP, Bazik JR, Powner DJ: Compres-sion volume during mechanical ventilation:comparison of ventilators and tubing cir-cuits. Crit Care Med 1985; 13:851–854

77. Ely E, Truman B, Shintani A, et al: Moni-toring sedation status over time in ICU pa-tients. Reliability and validity of the Rich-mond Agitation-Sedation Scale (RASS).JAMA 2003; 289:2983–2291

78. Rehder K, Hatch DJ, Sessler AD, et al: Thefunction of each lung of anesthetized andparalyzed man during mechanical ventila-tion. Anesthesiology 1972; 37:16–26

79. Reber A, Nylund U, Hedenstierna G: Posi-tion and shape of the diaphragm: implica-tions for atelectasis formation. Anaesthesia1998; 53:1054–1061

80. Krayer S, Rehder K, Vettermann J, et al:Position and motion of the human dia-phragm during anesthesia-paralysis. Anes-thesiology 1989; 70:891–898

81. Lundquist H, Hedenstierna G, StrandbergA, et al: CT-assessment of dependent lungdensities in man during general anaesthe-sia. Acta Radiol Scand 1995; 36:626–632

82. Brismar B, Hedenstierna G, Lundquist H, etal: Pulmonary densities during anesthesiawith muscular relaxation: a proposal of at-electasis. Anesthesiology 1985; 62:422–428

83. Tokics L, Hedenstierna G, Strandberg A, etal: Lung collapse and gas exchange duringgeneral anesthesia: effects of spontaneousbreathing, muscle paralysis, and positiveend-expiratory pressure. Anesthesiology1987; 66:157–167

84. Gattinoni L, D’Andrea L, Pelosi P, et al:

Regional effects and mechanism of positiveend-expiratory pressure in early adult respi-ratory distress syndrome. JAMA 1993; 269:2122–2127

85. Mutoh T, Guest RJ, Lamm WJE, et al: Proneposition alters the effect of volume overloadon regional pleural pressures and improveshypoxemia in pigs in vivo. Am Rev RespirDis 1992; 46:300–306

86. Lamm WJE, Graham MM, Albert RK: Mech-anism by which the prone position im-proves oxygenation in acute lung injury.Am J Respir Crit Care Med 1994; 150:184–193

87. Pelosi P, D’Andrea L, Vitale G, et al: Verticalgradient of regional inflation in adult respi-ratory distress syndrome. Am J Respir CritCare Med 1994; 149:8–13

88. Willeput R, Rondeux C, De Troyer A:Breathing affects venous return from legsin humans. J Appl Physiol 1984; 57:971–976

89. Takata M, Wise R, Robotham J: Effects ofabdominal pressure on venous return: ab-dominal vascular zone conditions. J ApplPhysiol 1990; 69:1961–1972

90. Hering R, Viehöfer A, Zinserling J, et al:Effects of spontaneous breathing duringairway pressure release ventilation on intes-tinal blood flow in experimental lung in-jury. Anesthesiology 2003; 99:1137–1144

91. Beck J, Gottfired SB, Navalesi P, et al: Elec-trical activity of the diaphragm during pres-sure support ventilation in acute respira-tory failure. Am J Respir Crit Care Med2001; 164:419–424

92. Putensen C, Hering R, Wrigge H: Con-trolled versus assisted mechanical ventila-tion. Curr Opin Crit Care 2002; 8:51–57

93. Dembinski R, Max M, Bensberg R, et al:Pressure support compared with controlledmechanical ventilation in experimentallung injury. Anesth Analg 2002; 94:1570–1576

94. Santak B, Radermacher P, Sandmann W, etal: Influence of SIMV plus inspiratory pres-sure support on VA/Q distributions duringpostoperative weaning. Intensive Care Med1991; 17:136–140

95. Guttmann J, Bernhard H, Mols G, et al:Respiratory comfort of automatic tube com-pensation and inspiratory pressure supportin conscious humans. Intensive Care Med1997; 23:1119–1124

96. Haberthür C, Elsasser S, Eberhard L, et al:Total versus tube-related additional work ofbreathing in ventilator dependent patients.Acta Anaesthesiol Scand 2000; 44:749–757

97. Haberthür C, Fabry B, Zappe D, et al: Ef-fects of mechanical unloading and mechan-ical loading on respiratory loop gain andperiodic breathing in man. Respir Physiol1998; 112:23–36

98. Kress JP, Pohlman AS, O’Connor MF, et al:Daily interruption of sedative infusions incritically ill patients undergoing mechani-cal ventilation. N Engl J Med 2000; 342:1471–1477


99. Kollef MH, Levy NT, Ahrens TS, et al: Theuse of continuous IV sedation is associatedwith prolongation of mechanical ventila-tion. Chest 1998; 114:541–548

100. Ely W, Baker A, Dunagan D, et al: Effects onthe duration of mechanical ventilation ofidentifying patients capable of breathingspontaneously. N Engl J Med 1996; 335:1864–1869

101. Coplin W, Pierson D, Cooley K, et al: Impli-cations of extubation delay in brain-injuredpatients meeting standard weaning criteria.Am J Respir Crit Care Med 2000; 161:1530–1536

102. Ibrahim EH, Tracy L, Hill C, et al: Theoccurrence of ventilator-associated pneu-monia in a community hospital: risk factorsand clinical outcomes. Chest 2001; 120:555–561

103. Cook DJ, Walter SD, Cook RJ, et al: Inci-dence of and risk factors for ventilator as-sociated pneumonia in critically ill patients.Ann Intern Med 1998; 129:433–440

104. Brook AD, Ahrens TS, Schaiff R, et al: Effectof a nursing-implemented sedation protocolon the duration of mechanical ventilation.Crit Care Med 1999; 27:2609–2615

105. Ely EW, Meade MO, Haponik EF, et al: Me-chanical ventilator weaning protocolsdriven by nonphysician health-care profes-sionals: evidence-based clinical practiceguidelines. Chest 2001; 120:454S–463S

106. Chastre J, Fagon JY: Ventilator-associatedpneumonia. State of the art. Am J RespirCrit Care Med 2002; 165:867–903

107. Craven DE, Steger KA: Epidemiology ofnosocomial pneumonia: new perspectiveson an old disease. Chest 1995; 108(suppl):1S–16S

108. Watando A, Ebihara S, Ebihara T, et al:Daily oral care and cough reflex sensitivity

in elderly nursing home patients. Chest2004; 126:1066–1070

109. Smina M, Salam A, Khamiees M, et al:Cough peak flows and extubation outcomes.Chest 2003; 124:262–268

110. Hormann C, Baum M, Putensen C, et al:Biphasic positive airway pressure(BIPAP)—a new mode of ventilatory sup-port. Eur J Anaesthesiol 1994; 11:37– 42


112. Kallet R, Corral W, Silverman H, et al: Im-plementation of a low tidal volume ventila-tion protocol for patients with acute lunginjury and the acute respiratory distresssyndrome. Respir Care 2001; 46:1024–1037

113. Kallet RH, Luce JM: Detection of patient-ventilator asynchrony during low tidal vol-ume ventilation, using ventilator waveformgraphics. Respir Care 2002; 47:183–185

114. Kallet RH, Alonso JA, Martin L, et al, for theNIH NHLBI ARDS Network: Incidence ofrespiratory distress or patients ventilatordyssynchrony during low tidal volume ven-tilation in patients with acute lung injury.Respir Care 2001; 46:1122

115. Kallet R, Eisner M, Luce JM, for the NIHNHLBI ARDS Network: Sedation and neu-romuscular blocking agent requirementsduring initiation of low tidal volume venti-lation in patients with acute lung injury(ALI). Respir Care 2001; 46:1122

116. Kallet RH, Alonso JA, Martin L, et al, for theNIH NHLBI ARDS Network: Characteristicsof acute lung injury patients exhibiting re-spiratory distress or dyssynchrony during

low tidal volume ventilation. Respir Care2001; 46:1122

117. Kallet RH, Corral W, Hayden D, et al, for theNIH NHLBI ARDS Network: Does high levelminute ventilation demand limit imple-mentation of the Acute Respiratory DistressSyndrome (ARDS) Network low tidal vol-ume ventilation protocol? Respir Care2001; 46:1122

118. Eichacker PQ, Gerstenberger EP, BanksSM, et al: Meta-analysis of acute lung injuryand acute respiratory distress syndrome tri-als testing low tidal volumes. Am J RespirCrit Care Med 2002; 166:1510–1514

119. Gebremichael M, Borg U, Habashi N, et al:Interhospital transport of the extremely illpatient: The mobile intensive care unit. CritCare Med 2000; 28:79–85

120. Navarrete-Navarra P, Rodriguez A, ReynoldsH, et al: Acute respiratory distress syn-drome among trauma patients: trends inICU mortality, risk factors, complicationsand resource utilization. Intensive CareMed 2001; 27:1133–1140

121. Haberthür C, Mols G, Elsasser S, et al: Ex-tubation after spontaneous breathing auto-matic tube compensation, t-tube, or pres-sure support ventilation. Acta AnaesthesiolScand 2002; 46:973–979

122. Gajic O, Dara S, Mendez J, et al: Ventilator-associated lung injury in patients withoutacute lung injury at the onset of mechanicalventilation. Crit Care Med 2004; 32:1817–1824

123. Derdak S: High-frequency oscillatory venti-lation for acute respiratory distress syn-drome in adult patients. Crit Care Med2003; 31(suppl):S317–S323

124. Jousela IT, Nikkio P, Tahvanainen J: Airwaypressure release ventilation by mask. CritCare Med 1988; 16:1250–12


High-frequency percussive ventilation

Ali Salim, MD, FACS; Matthew Martin, MD

High-frequency ventilation(HFV) was introduced in thelate 1960s as a new type ofventilation designed to re-

duce the complications observed withconventional ventilation (CV). The majorcharacteristics of HFV include a ventila-tory frequency of �60 breaths/min, in-creased functional reserve capacity, tidalvolumes of less than dead space, lowerpeak airway pressures, lower transpulmo-nary pressure, and more efficient gas ex-change than CV (1). These supposed ad-vantages have been successfullytranslated into clinical practice and havebeen reported to cause less circulatoryinterference than CV, diminish the rate ofbarotrauma, reduce air leaks in broncho-pleural fistulas, and create similar or im-proved gas exchange at lower pressures(2). High-frequency percussive ventila-tion (HFPV), a newer mode introducedduring the last 20 yrs, combines the ben-eficial effects of HFV with conventionalventilatory support (3). HFPV has beendescribed as an exceptionally versatileform of HFV that delivers subphysiologic

tidal volumes at rapid rates (up to 500breaths/min) using the volume-diffusiverespirator (1). It has been shown to pro-vide the same or improved oxygenationand ventilation at lower peak, mean, andend-expiratory pressures when comparedwith CV (3). There is a small, but growingbody of literature that has detailed thephysiologic and clinical responses toHFPV in a variety of patient populationsand disease states (Table 1).

Mechanism of Action

The high-frequency percussive venti-lator (Percussionaire, Bird Technologies,Sandpoint, ID) is a pneumatically pow-ered, time-cycled, and pressure-limitedventilator with inspiratory and expiratoryoscillation. The tidal volume delivery is aproduct of the peak inspiratory pressure(PIP) setting and subtidal volumes pro-duced by the oscillatory function. Uniqueto the HFPV is the presence of a pha-sitron or piston mechanism situated atthe end of the endotracheal tube, whichis a sliding Venturi that acts as both aninspiratory and expiratory valve. The pha-sitron is driven by a high-pressure gassupply at a high-frequency rate of 200–900 beats/min superimposed on a con-ventional inspiratory/expiratory pressure-controlled cycle that is set at a rate of10–15 breaths/min. During inspiration,lung volumes are progressively increasedin a controlled, stepwise fashion by repet-

itively diminishing subtidal volume deliv-eries until an oscillatory plateau is en-tered and maintained (4). At the end ofinspiration, the lung is allowed to emptypassively until the preset expiratory base-line is reached.

Gas exchange has been noted to be asgood, if not better, than CV at lower air-way pressures. As described by Krishnanand Brower (5), there are six mechanismsthat may contribute to gas exchange dur-ing all forms of HFV: 1) direct bulk flow,the flow of inspired air in proximal alveolileading to gas exchange by traditionalmethods (as with CV); 2) longitudinal(Taylor) dispersion, which is secondary tomixing from turbulent and swirling flowpatterns; 3) pendelluft, the variable flowdirectly between adjacent lung regionswith differences in compliance and resis-tance; 4) asymmetric velocity profiles, thelaminar flow pattern in which gas in thecenter of the airway advances inward andgas outside the center flows in a retro-grade fashion; 5) cardiogenic mixing, themechanical agitation from the beatingheart, especially in peripheral lung unitsadjacent to the heart; and 6) moleculardiffusion, the mixing of air in the small-est lung units near the alveolocapillarymembrane.

There are typically seven control vari-ables that need to be addressed in HFPV:1) PIP, 2) positive end-expiratory pres-sure, 3) continuous positive airway pres-sure, 4) inspiratory time, 5) expiratory

From the Department of Surgery, Division ofTrauma and Critical Care, University of Southern Cal-ifornia Keck School of Medicine; and the Los AngelesCounty–University of Southern California Medical Cen-ter, Los Angeles, CA.


DOI: 10.1097/01.CCM.0000155921.32083.CE

Objective: To review the technique and clinical application ofhigh-frequency percussive ventilation in critically ill patients.

Design: Literature search and descriptive review.Results: High-frequency percussive ventilation is a time-cycled,

pressure-limited mode of ventilation that delivers subphysiologictidal volumes at rates that can exceed 500 breaths/min. It offers thepotential advantage over conventional ventilation of providing equalor improved oxygenation and ventilation at lower peak and end-expiratory pressures. This modality has been used to manage severelung disease in the neonatal and pediatric population, treat inhalationinjury in pediatric and adult patients, and as salvage therapy in adultpatients with acute respiratory distress syndrome.

Conclusions: High-frequency percussive ventilation has beenshown to provide favorable gas exchange in several well-definedpatient populations. It reliably improves oxygenation and providesadequate ventilation at lower peak pressures than conventionalventilation. Adequately powered, randomized, prospective studiesdemonstrating significant mortality benefit have not yet beenperformed. (Crit Care Med 2005; 33[Suppl.]:S241–S245)

KEY WORDS: high-frequency percussive ventilation; volume-dif-fusive respirator; acute respiratory distress syndrome; inhalationinjury; subtidal volume; conventional ventilation; intracranial hy-pertension


Table 1. Physiologic and clinical responses to high-frequency percussive ventilation (HFPV) in reviewed studies

Study Patients HFPV Indication n Design Results with HFPV

Biarent et al., 1985 (11) Neonates ARF, failing CV 5 Case series 25% reduction in PEEP and FIO2;no change in gas exchange

Pfenninger et al., 1987(12)

Neonates HMD with ARF 8 Case series Oxygenation index doubled, PaCO2

decreased; no change inhemodynamics

Pfenninger et al., 1988(13)

Neonates HMD with ARF 8 Prospective, interventionaltrial (CV vs. HFPV)

Improved static compliance, lowerMAP (17 vs. 20) with equal lungvolumes

Cortiella et al., 1999 (15) Pediatric burn Inhalation injury 13 Case-control (CV vs.HFPV)

Higher P/F (463 vs. 380), lowerPIP (27 vs. 42), lowerpneumonia with HFPV

Mlcak et al., 2002 (16) Pediatric burn Inhalation injury 43 PRCT (CV vs. HFPV) Lower PIP (30 vs. 43); pneumonia9% with HFPV vs. 14% with CV;mortality 14% with HFPV vs.24% with CV

Carman et al., 2002 (17) Pediatric burn Burn with ARF 64 PRCT (CV vs. HFPV) Lower PIP (30 vs. 39), higher P/F(563 vs. 507) with HFPV; nooutcome difference

Reper et al., 1998 (18) Adult burn Inhalation injury 11 Case series Improved P/F (85 to 303), PIPdecreased (50 to 30), PaCO2

decreased (53 to 34)Cioffi et al., 1991 (19) Adult burn Inhalation injury 54 Case series Pneumonia and mortality less than

predicted by historic controlsRue et al., 1993 (20) Adult burn Inhalation injury 61 Case control (CV vs.

HFPV)Pneumonia 29% with HFPV vs.

52% with CV; mortality 16%with HFPV vs. 43% with CV

Cioffi et al., 1989 (1) Adult burn Inhalation injury 5 Case series Decreased FIO2 in 5/5, decreasedPIP in 3/5, increased P/F in 5/5,decreased PaCO2 in 4/5

Reper et al., 2003 (21) Adult burn Normal lungs 8 Prospective, interventionaltrial (CV vs. HFPV)

Increased PaO2 and P/F ratio,decreased PaCO2, decreased PIP

Reper et al., 2002 (22) Adult burn Inhalation injury 35 PRCT (CV vs. HFPV) Increased P/F first 3 days,decreased FIO2; no difference ininfections or mortality

Barrette et al., 1987 (33) Adult trauma TBI 66 Prospective, interventionaltrial (CV vs. HFPV vs.HFJV)

ICP decreased from 22 to 14 andPIP from 46 to 29 withimproved PaO2

Paulsen et al., 2002 (25) Adult surgical/trauma ARDS, failing CV 10 Case series Improved P/F in 8/10 (�300%increase in 7/10); no change inPIP

Velmahos et al., 1999 (26) Adult medical/surgical ARDS, failing CV 32 Case series Improved P/F and SaO2 withdecreased PIP; BP increasedwith HFPV

Hurst et al., 1987 (27) Adult trauma ARDS, failing CV 54 Case series Increased PaO2, decreased shuntfraction and PIP in hypoxicgroup; no barotrauma

Hurst et al., 1990 (2) Adult surgical ARF 100 PRCT (CV vs. HFPV) ARDS group had lower PIP andMAP on HFPV; no outcomedifference

Salim et al., 2004 (28) Adult TBI ARDS, failing CV 10 Case series Improved P/F (92 to 270),decreased PaCO2 (38 to 32); ICPdecreased from 31 to 17

Gallagher et al., 1989 (30) Adult medical/surgical ARF 7 Prospective, interventionaltrial (CV vs. HFPV)

Increased PaO2 (105 to 259),decreased PaCO2 (48 to 40); nochange in cardiac output

Hurst et al., 1988 (34) Adult TBI ARF, elevatedICP

38 Case series Lower PIP (34 vs. 62) and MAP (15vs. 22), lower ICP (15 vs. 27),same PCO2

Nates et al., 1999 (29) Adult neurosurgical ARDS, failing CV 18 Prospective, interventionaltrial (CV vs. HFPV)

Decreased PaCO2, increased PaO2

(130 vs. 84), decreased ICP (12vs. 17)

ARF, acute respiratory failure; CV, conventional ventilation; PEEP, positive end-expiratory pressure; HMD, hyaline membrane disease; MAP, meanairway pressure; P/F, PaO2/FIO2 ratio; PIP, peak inspiratory pressure; PRCT, prospective, randomized, controlled trial; TBI, traumatic brain injury; HFJV,high-frequency jet ventilation; ICP, intracranial pressure; BP, blood pressure.


time, 6) percussive frequency, and 7)rate. The endotracheal tube cuff is leftpartially deflated to allow for a continu-ous air leak through the trachea. Thispartially deflated cuff combined with thecontinuous pneumatic compressions al-lows for a dramatic mobilization of secre-tions and clearance of pulmonary infil-trates and for improved ventilation.Oxygenation is controlled by FIO2, theratio of positive end-expiratory pressureand continuous positive airway pressure,PIP, inspiratory time, and frequency.Ventilation is governed by the relation-ship between inspiratory time and expi-ratory time, PIP, and frequency.

Typical starting settings at our insti-tution are as follows: rate of 15, frequencyof 500, inspiratory time/expiratory timeof 1:1, PIP of approximately two thirdsthe pressure utilized in CV, and a positiveend-expiratory pressure/continuous posi-tive airway pressure ratio of 2:8 cm H2O.Table 2 summarizes the algorithm weutilize to maximize gas exchange. Ourcurrent main indication for the use ofHFPV is for salvage of patients with acutelung injury or acute respiratory distresssyndrome (ARDS) who demonstrate on-going hypoxemia or severe hypercarbia,despite maximal conventional support.Other less common indications includerefractory bronchopleural fistulas, severeunilateral lung injury (independent-lungventilation), or repeated lobar collapsedue to airway secretions. HFPV is alsoused extensively by our burn service as aprimary ventilatory modality in patientswith significant inhalation injury or air-way debris. Neuromuscular blockade isnot required with this mode of ventila-tion and is frequently able to be discon-tinued within hours of initiation ofHFPV.

Clinical Experience with HFPV

Animal Models. Although other formsof HFV (high-frequency jet or oscillatoryventilation) have been studied in a varietyof animal lung injury models (6 – 8),there are few relevant animal studies withHFPV. Smolarz et al. (9) performed aprospective, randomized trial comparingHFPV with a lung-protective ventilationstrategy of low-pressure (PIP � 30 cmH2O) CV combined with arteriovenousCO2 removal in a sheep model of ARDS.Both methods provided adequate andequivalent oxygenation at low peak air-way pressures (mean, 26 cm H2O) and nosignificant difference in CO2 removal(PCO2, 51 mm Hg with HFPV vs. 46 mmHg with CV).

Cioffi et al. (10) performed a prospec-tive, randomized trial of CV comparedwith both HFPV and high-frequency os-cillatory ventilation in a primate burnand inhalation injury model. The HFPVgroup demonstrated significantly loweroxygen requirements (FIO2) and PIPscompared with high-frequency oscilla-tory ventilation and a lower respiratoryrate and barotrauma index when com-pared with CV. Histologic examination oflungs from all three groups demonstratedsignificantly less pathologic injury in theanimals ventilated with HFPV, whichthey attributed to decreased barotrauma.

Pediatric and Burn Patients. As withother modes of HFV, some of the earliestapplications of HFPV were for the man-agement of severe lung disease in theneonatal and pediatric population. Biar-ent et al. (11) described a series of fiveneonates with acute respiratory distresswho were changed to HFPV after failingwith conventional ventilatory support.They observed immediate improvementsin oxygenation, with a 25% reduction in

FIO2 and positive end-expiratory pressurerequirements within 12 hrs. There wasno significant difference in gas exchangeor hemodynamic variables observed withHFPV.

Pfenninger and Gerber (12) and Pfen-ninger and Minder (13) described theirexperience with HFPV in two studies ofneonates with hyaline membrane diseaseand acute respiratory failure (ARF). Inthe first series of eight neonates, therewas a marked improvement in oxygen-ation, with a doubling of the oxygenationindex, after switching to HFPV. Ventila-tion was also improved, with a significantdecrease in PCO2 and no change in centralhemodynamics (12). They then per-formed a prospective interventional trialin eight neonates, investigating the effectof HFPV on lung compliance and gasexchange. HFPV demonstrated a markedimprovement in static compliance andmaintained equivalent lung volumes atlower mean airway pressures (17 vs. 20cm H2O) compared with CV. In addition,it seemed that HFPV maintained themean airway pressures within the idealrange on the pressure–volume curves,whereas the mean airway pressures on CVwere consistently above the upper inflex-ion point (13).

An additional proposed benefit ofHFPV is the generation of intrapulmo-nary mechanical percussive waves, whichmay aid in the lysis and clearance of air-way mucus and secretions (14). This hasgenerated much interest from the burncommunity, particularly for the treat-ment of children and adults with inhala-tion injury (1, 15–22). Two prospective,randomized trials of HFPV have been re-ported in the pediatric burn population.Mlcak et al. (16) studied 43 children withinhalation injury randomized to CV orHFPV at the time of hospital admission.The HFPV group was adequately venti-lated at significantly lower PIPs (30 vs.43) compared with the CV group. Al-though pneumonia and mortality rateswere lower in the HFPV group, the dif-ferences failed to reach statistical signif-icance. Carman et al. (17) randomized 64burned children with ARF to HFPV or CV,and this represents one of the only stud-ies using a lung-protective strategy withtidal volumes of 6–8 mL/kg in the CVgroup. They again demonstrated im-proved oxygenation with HFPV (PaO2/FIO2

ratio, 563 vs. 507) at lower peak airwaypressures (30 vs. 39 cm H2O) comparedwith CV. However, there was no signifi-cant difference noted in any outcome

Table 2. Algorithm for basic management of volume-diffusive respirator

To increase PaO2 onlyA. 1PIPB. 1FIO2

C. 1PEEP/CPAP level in increments of 2 cm H2OTo decrease PaCO2 only

A. 1PIP (PIPs at the carina are approximately one third the level set on the HFPV)B. If PIP is �80 cm H2O: 2 high-frequency rate from 500 to 350 cycles/minC. If PaO2 is acceptable and high levels of CPAP are present (i.e., 20 cm H2O), 2level of CPAP

To increase PaO2 and decrease PaCO2

A. 1PIPs in increments of 5 cm H2OTo increase PaCO2 in presence of low PIPs

A. 1CPAP level by 4 cm H2O

PIP, peak inspiratory pressure; PEEP, positive end-expiratory pressure; CPAP, continuous positiveairway pressure; HFPV, high-frequency percussive ventilation.


measures such as sepsis, pneumonia,ARDS, or survival.

Multiple studies have examined therole of HFPV in the adult burn and inha-lation injury population. Cioffi et al. (19)described a case series of 54 burn admis-sions with inhalation injury placed on“prophylactic” HFPV within 1 hr of hos-pital admission. They compared results inthis population with historic controls andused a logistic regression–derived equa-tion to predict mortality. The HFPVgroup had a significantly lower pneumo-nia rate (26% vs. 46%) than historic con-trols and a mortality rate of 18% com-pared with a predicted mortality of 43%.The same group reported a subsequentcase-control series of inhalation injuriesover a 6-yr period managed with HFPV (n� 61) or CV (n � 199). They again dem-onstrated lower pneumonia rates withHFPV (29% vs. 52%) and improved mor-tality (16% vs. 43%) compared with CV(20). However, the validity of compari-sons with these retrospective and un-matched control populations pose signif-icant limitations to the interpretation oftheir results.

In the only prospective, randomizedtrial in adults with inhalation injury,Reper et al. (22) randomized 35 patientsto HFPV or CV (volume control, 10mL/kg tidal volume) at hospital admis-sion. The HFPV group demonstrated asignificant improvement in the PaO2/FIO2

ratio (325 vs. 175) over the first 72 hrscompared with the CV group, with nochange in hemodynamics and no clini-cally evident barotrauma. Although therewas no significant difference betweenstudy groups in ventilation, airway pres-sures, pneumonia, or survival, these re-sults are severely limited by the smallsample size.

ARF/ARDS. With the accumulatingbody of evidence relating volutrauma andbarotrauma to the development of acutelung injury and ARDS (5, 23), high-frequency ventilatory techniques seem tooffer a unique advantage of adequate ox-ygenation and ventilation with lower tidalvolumes and airway pressures than CV.Multiple techniques of HFV have beenstudied in animal and human models ofacute lung injury and ARDS, with vari-able results (6, 7, 23, 24). HFPV hasmainly been studied as a salvage therapyfor patients with ARF/ARDS who are fail-ing CV (25–30). Velmahos et al. (26) re-ported a series of 32 adult medical andsurgical intensive care unit patients withARDS who were failing conventional me-

chanical ventilation. The mean PaO2/FIO2

ratio on CV was 111, which was improvedto 163 by 1 hr after conversion to HFPVand 193 at 48 hrs. PIPs decreased from42.4 cm H2O on CV to 33.2 cm H2O after1 hr of HFPV and 32.5 at 48 hrs, butmean airway pressures increased from 21cm H2O on CV to 27 cm H2O on HFPV.There was no change in hemodynamicvariables. Paulsen et al. (25) reported aseries of ten adult surgical and traumapatients with ARDS who were failing CVand were switched to HFPV. There was asignificant improvement in oxygenationwith HFPV in the majority of patients(80%), with 70% demonstrating a�300% increase in the PaO2/FIO2 ratio.Ventilation was also improved on HFPV,with the mean PaCO2 decreasing from 52to 44 mm Hg.

Hurst et al. (27) demonstrated similarresults in a series of 54 adult traumapatients with postinjury ARDS. Hypoxicpatients switched to HFPV demonstratedimproved oxygenation with decreasedshunt fraction and PIPs and no clinicallysignificant barotrauma. The same groupalso reported the only prospective, ran-domized trial of HFPV for adults withARF (2). A total of 100 patients with ARFwere randomized to CV or HFPV andtreated to predefined respiratory endpoints (PaO2/FIO2 � 225 or shunt �20%).There was no difference between groupsin the time to reach the study end points,but among patients with ARDS, the HFPVgroup maintained significantly lower air-way pressures throughout the study pe-riod. There was no difference observed inlength of stay or mortality.

Intracranial Hypertension. Preven-tion and treatment of elevated intracra-nial pressures (ICP) is one of the primarygoals in the management of the patientwith brain injury. The management ofintracranial hypertension (ICP � 20) re-fractory to standard therapy is an area ofactive investigation, with some suggest-ing drastic maneuvers such as craniec-tomy or decompressive laparotomy tolower ICP (31, 32). Several interestingstudies have examined the salutary effectof HFPV on lowering ICP (28, 29, 33, 34).Barrette et al. (33) performed a prospec-tive, interventional trial on adult patientswith traumatic brain injury that com-pared CV with HFPV and high-frequencyjet ventilation. Both modes of HFV sig-nificantly lowered ICP, with improved ox-ygenation in the HFPV and worsening ofoxygenation with jet ventilation. Hurst etal. (34) used HFPV as a salvage therapy in

38 patients with head injury and ARF anddemonstrated a decrease in ICP from amean of 27 to 15 mm Hg with lower peakand mean airway pressures. More re-cently, Salim et al. (28), using HFPV inhead-injured patients with ARDS, dem-onstrated a significant decrease in ICP(31 to 17) and improved oxygenation(PaO2/FIO2, 92 to 270) and ventilation(PaCO2, 38 to 32).

Conclusion

HFPV has been demonstrated to pro-vide effective oxygenation and ventilationacross a broad spectrum of patient popu-lations and disease processes. In the gen-eral intensive care unit population, HFPVis most commonly used as a salvage ther-apy in patients with worsening respira-tory failure or ARDS who are failing con-ventional ventilatory support. In thissetting, HFPV reliably improves oxygen-ation with adequate or improved ventila-tion at lower PIPs and with minimal ef-fect on central hemodynamics. Additionaldemonstrated benefits of HFPV includeimproved clearance of airway debris andsecretions with less barotrauma in pa-tients with inhalation injury and signifi-cant decreases of ICP in head-injured pa-tients.

Further animal studies examining thelung mechanics, gas exchange, and phys-iologic effects with HFPV in a variety ofnormal and pathologic lung conditionsare needed to better understand this ven-tilatory modality. Although there hasbeen no demonstrated outcome benefit interms of mortality or morbidity withHFPV, larger and more appropriatelypowered prospective studies will beneeded to definitively address this ques-tion. In addition, the long-term effectsand the optimal timing for initiation ofHFPV still need further exploration.

REFERENCES

1. Cioffi WG, Graves TA, McManus WF, et al:High-frequency percussive ventilation in pa-tients with inhalation injury. J Trauma 1989;29:350–354

2. Hurst JM, Branson RD, Davis K, et al: Com-parison of conventional mechanical ventila-tion and high-frequency ventilation. AnnSurg 1990; 211:486–491

3. Davis K, Hurst JM, Branson RD: High fre-quency percussive ventilation. Probl RespirCare 1989; 2:39–47

4. Mlcak R, Cortiella J, Desai M, et al: Lungcompliance, airway resistance, and work of


breathing in children after inhalation injury.J Burn Care Rehabil 1997; 18:531–534

5. Krishnan JA, Brower RG: High-frequencyventilation for acute lung injury and ARDS.Chest 2000; 118:795–807


7. Hamm CR, Millan JC, Curtet N, et al: Highfrequency jet ventilation preceded by lungvolume recruitment decreases hyaline mem-brane formation in surfactant deficientlungs. Pediatr Res 1990; 27(Suppl):305A

8. Froese AB, McCulloch PR, Sugiura M, et al:Optimizing alveolar expansion prolongs theeffectiveness of exogenous surfactant therapyin the adult rabbit. Am Rev Respir Dis 1993;148:569–577

9. Smolarz JR, Wang D, Alpard SK, et al: High-frequency percussive ventilation vs. arterio-venous carbon dioxide removal in a LD100model of ARDS. Annu Natl Student Res Fo-rum Publ 2002:21–39

10. Cioffi WG, deLemos RA, Coalson JJ, et al:Decreased pulmonary damage in primateswith inhalation injury treated with high-frequency ventilation. Ann Surg 1993; 218:328–337

11. Biarent D, Steppe M, Muller F, et al: High-frequency jet percussive ventilation in new-borns and infants with damaged lungs. ActaAnaesthiol Belgica 1985; 3:127–128

12. Pfenninger J, Gerber AC: High-frequencyventilation (HFV) in hyaline membrane dis-ease: A preliminary report. Intensive CareMed 1987; 13:71–75

13. Pfenninger J, Minder C: Pressure-volumecurves, static compliances and gas exchangein hyaline membrane disease during conven-tional mechanical and high-frequency venti-lation. Intensive Care Med 1988; 14:364–372

14. Freitag L, Long WM, Kim CS, et al: Removalof excessive bronchial secretions by asym-

metric high-frequency oscillations. J ApplPhysiol 1989; 67:614–619

15. Cortiella J, Mlcak R, Herndon D: High fre-quency percussive ventilation in pediatric pa-tients with inhalation injury. J Burn CareRehabil 1999; 20:232–235

16. Mlcak RP, Suman OE, Sanford AP, et al:Comparison of high frequency percussiveventilation and conventional. J Burns SurgWound Care [serial online] 2002; 1:19

17. Carman B, Cahill T, Warden G, et al: A pro-spective, randomized comparison of the vol-ume diffusive respirator vs conventional ven-tilation for ventilation of burned children:2001 ABA paper. J Burn Care Rehabil 2002;23:444–448

18. Reper P, Dankaert R, van Hille F, et al: Theusefulness of combined high-frequency per-cussive ventilation during acute respiratoryfailure after smoke inhalation. Burns 1998;24:34–38

19. Cioffi WG, Rue LW, Graves TA, et al: Prophy-lactic use of high-frequency percussive ven-tilation in patients with inhalation injury.Ann Surg 1991; 213:575–582

20. Rue LW, Cioffi WG, Mason AD, et al: Im-proved survival of burned patients with inha-lation injury. Arch Surg 1993; 128:772–778

21. Reper P, Van Bos R, Van Loey K, et al: Highfrequency percussive ventilation in burn pa-tients: Hemodynamics and gas exchange.Burns 2003; 29:603–608

22. Reper P, Wibaux O, Van Laeke D, et al: Highfrequency percussive ventilation and conven-tional ventilation after smoke inhalation: Arandomised study. Burns 2002; 28:503–508

23. Derdak S: High-frequency oscillatory venti-lation for acute respiratory distress syn-drome in adult patients. Crit Care Med 2003;31(Suppl):S317–S322

24. Delemos RA, Coalson JJ, Gerstmann DR, etal: Ventilatory management of infant ba-boons with hyaline membrane disease: Theuse of high frequency ventilation. PediatrRes 1987; 21:594–602

25. Paulsen SM, Killyon GW, Barillo DJ: High-frequency percussive ventilation as a salvagemodality in adult respiratory distress syn-drome: A preliminary study. Am Surg 2002;10:854–856

26. Velmahos GC, Chan LS, Tatevossian R, et al:High-frequency percussive ventilation im-proves oxygenation in patients with ARDS.Chest 1999; 116:440–446

27. Hurst JM, Branson RD, DeHaven CB: Therole of high-frequency ventilation in post-traumatic respiratory insufficiency.J Trauma 1987; 27:236–242

28. Salim A, Miller K, Dangleben D, et al: High-frequency percussive ventilation: An alterna-tive mode of ventilation for head injured pa-tients with adult respiratory distresssyndrome. J Trauma 2004; 57:542–546

29. Nates JL, Cravens J, Hudgens C, et al: Effectsof volumetric diffusive respiration with nor-mal or inverse I:E ratio on intracranial pres-sure. Crit Care Med 1999; 27:A73

30. Gallagher TJ, Boysen PG, Davidson DD:High-frequency percussive ventilation com-pared with conventional mechanical ventila-tion. Crit Care Med 1989; 17:364–366

31. Albanese J: Decompressive craniectomy forsevere traumatic brain injury: Evaluation ofthe effects at one year. Crit Care Med 2003;31:2535–2538

32. Joseph DK, Dutton RP, Aarabi B, et al: De-compressive laparotomy to treat intractableintracranial hypertension after traumaticbrain injury. J Trauma 2004; 57:687–695

33. Barrette RR, Hurst JM, Branson RD, et al: Acomparison of conventional mechanical hy-perventilation with two forms of high-frequency ventilation for the control of in-tracranial pressure in closed head injury.Respir Care 1987; 32:733–740

34. Hurst JM, Branson RD, Davis K: High-frequency percussive ventilation in the man-agement of elevated intracranial pressure.J Trauma 1988; 28:1363–1367


High Frequency Oscillatory Ventilation - A Decade of Progress

Documents

Transcript of High Frequency Oscillatory Ventilation - A Decade of Progress