Marc MossMichele YangMadison MachtPeter SottileLaura GrayMonica McNultyDianna Quan

Screening for critical illnesspolyneuromyopathy with single nerveconduction studies

Received: 11 November 2013Accepted: 18 February 2014Published online: 13 March 2014� Springer-Verlag Berlin Heidelberg andESICM 2014

Take-home message: Neuromuscularweakness is common in patients recoveringfrom critical illness. This weakness is oftendue to the development of critical illnesspolyneuromyopathy (CIPNM); however, theability to diagnose CIPNM is hampered byimpaired patient sensorium, technicallimitations, and the time-intensive nature ofperforming complete electrophysiologicaltesting. In this study, we demonstrated thatunilateral peroneal and sural nerveconduction studies can accurately screen forCIPNM in critically ill patients; theseresults identify a more streamlined methodto identify those at highest risk for CIPNMand may facilitate routine assessment andmonitoring of weakness in critically illpatients.

M. Moss ()) � M. Macht � P. Sottile �L. GrayDivision of Pulmonary Sciences andCritical Care Medicine, University ofColorado Denver School of Medicine,AMC, RC2, C-272, 12700 E 19th Ave,Aurora, CO 80045, USAe-mail: marc.moss@ucdenver.eduTel.: ?1-303-7246079Fax: ?1-303-7246042

M. YangSection of Child Neurology, Department ofPediatrics, University of Colorado DenverSchool of Medicine, Aurora, CO, USA

M. McNultyColorado Health Outcomes Group,Department of Medicine, University ofColorado Denver School of Medicine,Aurora, CO, USA

D. QuanDepartment of Neurology, University ofColorado Denver School of Medicine,Aurora, CO, USA

Abstract Purpose: The ability todiagnose patients with critical illnesspolyneuromyopathy (CIPNM) is ham-pered by impaired patient sensorium,technical limitations, and the time-intensive nature of performing electro-physiological testing. Therefore, wesought to determine whether singlenerve conduction studies (NCS) couldaccurately screen for CIPNM. Meth-ods: Critically ill patients atincreased risk for developing CIPNMwere identified. Bilateral NCS of sixnerves, and concentric needle electro-myography were performed within24 h of meeting inclusion criteria, andsubsequently on a weekly basis untilCIPNM was diagnosed or the patientwas discharged from the intensive careunit (ICU). Results: A total of 75patients were enrolled into the study.Patients who developed CIPNM had ahigher hospital mortality (50 vs. 13 %,p = 0.002), and fewer ICU-free days(0 vs. 11, p = 0.04). There were nodifferences between the right and leftamplitudes (p = 0.59, 0.91, and 0.21)for nerves that could be

simultaneously tested bilaterally(sural, peroneal, and tibial). Theamplitudes for each of the six indi-vidual nerves were significantlydiminished in patients with CIPNMwhen compared to patients withoutCIPNM. The nerves with the bestdiagnostic accuracy were the peronealnerve [AUC = 0.8856; sensitiv-ity = 94 % (95 % CI = 88–100 %);specificity = 74 % (95 %CI = 63–85 %)], and the sural nerve[AUC = 0.8611; sensitivity = 94 %(95 % CI = 88–100 %); specific-ity = 70 % (95 % CI = 59–81 %)].The combined diagnostic accuracy forthe amplitudes of the peroneal andsural nerves increased significantly[AUC = 0.9336; sensitivity = 100 %(95 % CI = 100–100 %) and speci-ficity = 81 % (95 % CI =71–91 %)]. Conclusions: Unilateralperoneal and sural NCS can accuratelyscreen for CIPNM in ICU patients anddetect a limited number of patients thatwould need concentric needle elec-tromyography to confirm a diagnosisof CIPNM. These results identify amore streamlined method to diagnoseCIPNM that may facilitate routinediagnostic testing and monitoring ofweakness in critically ill patients.

Keywords Critical illnesspolyneuropathy � Critical illnessmyopathy � Severe sepsis �Sensitivity � Specificity

Intensive Care Med (2014) 40:683–690DOI 10.1007/s00134-014-3251-6 ORIGINAL

Introduction

Critical illness polyneuropathy (CIP) and critical illnessmyopathy (CIM) are neuromuscular disorders thatdevelop after admission to an intensive care unit (ICU),and impair short- and long-term outcomes [1–5]. BecauseCIP and CIM can occur concurrently in patients, they aresometimes combined into a variety of terms includingICU-acquired paresis, critical illness myopathy and/orneuropathy, or critical illness polyneuromyopathy (CIP-NM) [6–15].

Nerve conduction studies (NCS) in conjunction withneedle electromyography (EMG) are necessary toproperly diagnose CIPNM [16]. These electrophysio-logical tests, particularly the EMG assessment of motorunit potential morphology and recruitment, are ham-pered by decreased patient effort resulting fromdelirium, encephalopathy, and the use of sedative agents[9, 17]. Furthermore electrical devices in the ICU cancause interference and impede adequate NCS and EMGsignals. Therefore, the performance of these neuro-physiological tests can be particularly difficult incritically ill patients [15, 18]. In addition, comprehensiveelectrophysiological testing of multiple nerves is time-consuming, requiring as long as 90 min to complete incritically ill patients [15]. Comprehensive electrophysi-ological testing of multiple nerves can also causeprolonged discomfort to some patients [15, 18]. Finally,the performance of bilateral NCS may be unnecessarybecause CIPNM usually symmetrically affects multiplenerves.

The ability to diagnose CIPNM could benefit from thedevelopment of more streamlined testing that could helpefficiently screen critically ill patients. Reductions in theperoneal compound muscle action potential have previ-ously been reported to diagnose CIPNM [15]. However, alimitation of this study was that complete electrophysio-logical studies were performed only if the sural sensorynerve action potential (SNAP) or peroneal compoundmuscle action potential (CMAP) became abnormal [15].If a focused diagnostic testing algorithm for CIPNMcould be developed, neurophysiological testing might beutilized more frequently and effectively in critically illpatients. As a result, healthcare providers could moreeasily identify patients who would benefit from potentialtherapies for CIPNM such as early mobilization, and alsoprovide patients and families with better informationconcerning the likelihood of prolonged hospitalizationand residual physical limitations [19, 20]. Therefore, weconducted a prospective observational cohort study ofcritically ill patients to answer the following question:do specific motor or sensory nerves more accuratelyscreen for CIPNM? If so, focused NCS testing of theseparticular nerves may effectively identify likely cases ofCIPNM.

Methods

Patients were prospectively enrolled from ICUs at oneuniversity and one community hospital. The institutionalreview board for both hospitals approved this study, andeach patient or authorized representative gave informedwritten consent. Some of these data were previouslypresented as an abstract at the International AmericanThoracic Society Conference [21].

Inclusion and exclusion criteria

We studied two groups of patients at increased risk for thedevelopment of CIPNM: patients with severe sepsis, andnon-septic patients with acute respiratory failure and non-pulmonary organ dysfunction. Patients with severe sepsismet the following two criteria: severe sepsis by consensusdefinition; and acute respiratory failure requiringmechanical ventilation [22]. Patients with respiratoryfailure and non-pulmonary organ dysfunction met thefollowing criteria: acute respiratory failure defined asrequiring invasive or non-invasive ventilation with a p/f ratio of 250 or less; and one non-pulmonary organdysfunction including cardiovascular, renal, hematologi-cal, or acidosis [22, 23].

Exclusion criteria were age less than 18 years, diag-nosis of pre-existing peripheral nervous system disorderaffecting motor function, a pre-existing central nervoussystem disorder, fewer than two limbs in which musclestrength could be tested, inability to obtain informedconsent, or patient/surrogate or physician refusal to par-ticipate in the study.

Study procedures

All aspects of patient care, including nutritional support,ventilator management, glycemic control, and generalICU supportive care, were left to the discretion of theprimary physician.

Baseline neurological examination, NCS, and con-centric needle EMG examinations were performed within24 h of meeting inclusion criteria. Electrophysiologicstudies were performed in the ICU according to Kimura’sstandard techniques using a Viking IV or Viking Selectsystem (Madison, WI) [24]. The serial studies for eachsubject were performed by a single examiner on the samepiece of equipment to minimize the inter-examiner vari-ability and improve reliability [25, 26]. The neurologicalexamination and NCS were repeated on a weekly basisuntil CIPNM was diagnosed or the patient was dischargedfrom the ICU [4].

When possible, NCS were performed bilaterally forsix different nerves (three sensory nerves: sural, radial,

684

and median; and three motor nerves: peroneal, tibial, andmedian). The compound muscle action potential (CMAP)was elicited by stimulating standard distal and proximalsites to calculate a conduction velocity and to exclude thepresence of conduction block or temporal dispersion. Fwaves were also recorded from the tibial and mediannerves when the distal CMAP amplitudes were greaterthan 1 mV. Finally, on the initial study of each patient,3-Hz repetitive stimulation of the median motor nervewas performed, over 2-s recording at the abductor pollicisbrevis muscle. If the patient was able to cooperate,additional trains of six consecutive stimuli were givenafter brief (15 s) exercise and 2 s after prolonged (1 min)exercise to screen further for defects of neuromusculartransmission. After reviewing the studies to excludepatients with electrophysiologic features of acquireddemyelination or neuromuscular transmission defects, theamplitudes of the sensory and motor responses wereanalyzed for abnormalities.

Electromyographic (EMG) studies were performedusing standard precautions. Insertional activity, sponta-neous activity, motor unit potential (MUP) morphologyand recruitment pattern were recorded from the tibialisanterior, vastus medialis, deltoid, abductor pollicis brevis,and extensor digitorum communis muscles. However, themuscles studied varied according to the patient’s level ofconsciousness and ability to activate the muscles eithervoluntarily or during spontaneous limb movement. Theinsertional and spontaneous activity was analyzed using asensitivity of 50 mV/division. The MUP morphology andrecruitment patterns were analyzed using a sensitivity of200 mV/division and sweep speed of 10 ms/division.

Diagnostic criteria for CIP and CIM

Patients were diagnosed as having CIP or CIM if they metthe criteria in Table 1 [4, 27]. To determine the MRCscore, six muscle groups were tested bilaterally: shoulderabduction, elbow flexion, wrist extension, hip flexion,knee extension, and foot dorsiflexion [28]. If the perfor-mance of a formal MRC assessment was limited by

sedation or encephalopathy, the presence of any evidenceof clinical weakness on examination was necessary tomeet the diagnosis of CIP or CIM. Clinical weakness onexamination was defined as more than two muscle groupswith an MRC score of four or less. Patients were definedas having CIPNM if they met either of the criteria for CIPor CIM. In cases where CIP and CIM appeared to coexist,the CIM criterion for SNAP amplitudes greater than 80 %of the lower limit of normal was waived. For patients whowere diagnosed as having CIP or CIM, the electrophysi-ological testing at the time they met the diagnostic criteriawere used in all analyses. For the remaining patients whodid not meet the diagnostic criteria for CIPNM, data fromtheir last electrophysiological tests were used. ICU-freedays are defined as the number of days out of 28 that thepatient was alive and out of the ICU [29]. Hospital-freedays were defined as the number of days out of 90 that thepatient was alive and out of the ICU. Patients withoutelectrophysiological evidence of CIPNM were furtherclassified as being deconditioned (MRC 48 or less) orhaving relatively normal neuromuscular function (MRCgreater than 48).

Statistical analysis

All univariate comparisons of demographic informationwere evaluated with v2 or non-parametric Wilcoxonanalyses. Non-parametric testing was performed becauseall of the data were not normally distributed, and data arereported as median values with 25–75 % quartiles. Apaired non-parametric analysis was performed to deter-mine difference in the distribution of the scores betweenright and left nerve amplitudes. We also determined acoefficient of determination (R2) comparing the left andright nerve amplitudes. For each of the six nerves, weconstructed individual ROC curves for the diagnosis ofCIPNM, and then calculated a c-statistic that is commonlyused to quantify the capacity of a diagnostic test to dis-criminate between outcomes. Scores range from 0.5 to 1.0with higher values indicating a better predictive model. Adifference of 0.05 in the c-statistic is generally considered

Table 1 Diagnostic criteria for critical illness polyneuropathy and myopathy

Critical illness polyneuropathy Critical illness myopathy

MRC score of at most 48 or clinical weakness on examination MRC score of at most 48 or clinical weakness on examinationSNAP amplitudes less than 80 % of the lower limit

of normal in two or more nervesSNAP amplitudes greater than 80 % of the lower limit of

normal in two or more nervesReduced recruitment of motor unit potentials on needle

EMG examinationCMAP amplitudes less than 80 % of the lower limit of

normal in two or more nerves without conduction blockAbsence of a decremental response on repetitive nerve

stimulationNeedle EMG with short-duration, low amplitude motor unit potentials

with early recruitment or evidence of muscle inexcitabilityRetained muscle excitability Absence of a decremental response on repetitive nerve stimulation

MRC Medical Research Council, SNAP sensory nerve action potential, CMAP compound muscle action potentials

685

to be clinically significant [30, 31]. For each nerve, thebest cutoff value for the specific amplitude was deter-mined by identifying the value that maximized theoptimal sensitivity and specificity. Assuming a sensitivityof 90 % and a specificity of 90 %, a sample size of 64patients would be able to obtain 95 % CI 83–97 % forboth sensitivity and specificity. All data were analyzedusing JMP software. Reported p values are two sided, andan a value of 0.05 was used in all analyses.

Results

From July 2009 through April 2012, a total of 208patients met inclusion criteria for the study. A total of 133patients were excluded for the following reasons: age lessthan 18 years (n = 2), pre-existing peripheral motornervous system disorder (n = 15), a pre-existing centralnervous system disorder (n = 24), fewer than two limbsin which muscle strength could be tested (n = 2),inability to obtain informed consent (n = 42), or refusalto participate in the study (patient/family refusal, n = 40;primary physician refusal, n = 8). Therefore, 75 patientswere enrolled; 66 met the criteria for severe sepsis, and anadditional nine met the criteria for respiratory failure withadditional organ dysfunction. Eleven patients did not haveany electrophysiological testing performed for the fol-lowing reasons: patient died before testing could beperformed (n = 5), patient’s goals of care were changedto comfort measures only before testing could be per-formed (n = 3), and other reasons (withdrawn from studyby surrogate, n = 1; development of new cerebrovascularaccident (CVA), n = 1; experimental medication startedwith known neurological effects, n = 1). Therefore thefinal cohort included 64 patients. A total of 69 % ofpatients (n = 44) were responsive to verbal stimuli at thetime of the electrophysiological testing.

The demographics of these patients are included inTable 2. A total of 24 % (n = 18) of the patients devel-oped CIPNM. All of these patients met the diagnosticcriteria for both CIP and CIM. The median number of

NCS performed was not different between patients whodeveloped CIPNM [n = 1, (25 % quartiles: 1, 2)] andthose who did not develop CIPNM [n = 1, (25 % quar-tiles: 1, 2)], p = 0.9. There was no difference in thediagnosis of CIPNM between patients who could andcould not respond to verbal stimuli at the time the NCSwas performed, p = 0.2. Patients who developed CIPNMhad a higher hospital mortality (50 vs. 13 %, p = 0.002),and fewer ICU-free days (0 vs. 11 p = 0.04) whencompared to patients who did not develop CIPNM. Onlyhalf (n = 25) of the 50 surviving patients were dischargedto home. The rest of the patients were discharged torehabilitation or long-term acute care hospitals (n = 16),hospice (n = 4), nursing facilities (n = 3), or other(n = 2). There was no difference in discharge locationbetween the patients with and without CIPNM.

Of the 46 patients without electrophysiological evi-dence of CIPNM, 22 of them met criteria fordeconditioning, and 24 had relatively normal neuromus-cular function. Hospital mortality was increased in thepatients with CIPNM when compared to those with de-conditioning or relatively normal neuromuscular function(50, 14, and 13 % respectively, p = 0.01). Hospital-freedays (at day 90) were also significantly decreased in theCIPNM patients when compared to those with decondi-tioning or normal function [4.5 (0–70.3) vs. 59.5(36.5–76.5) vs. 72 (54.5–73), p = 0.022].

We were unable to routinely test all of the six nervesbilaterally. The most frequent nerves that could not be testedbilaterally were the median and radial nerves. The mostcommon reason for the inability to perform NCS was limitedphysical access to the nerve due to the presence of intrave-nous or arterial catheters (Table 3). Environmental electricalinterference occurred infrequently. Bilateral examinationswere routinely performed for the sural, peroneal, and tibialnerves, and there were no statistical differences in the dis-tribution of the scores for the right and left SNAP and CMAPamplitudes (Table 4: p = 0.59, 0.91, and 0.21, respec-tively). The correlation between the right and leftsamplitudes for the sural, peroneal, and tibial nerves was alsohighly significant (R2 values of 0.87, 0.54, and 0.70,respectively, p \ 0.0001 for all analyses).

Table 2 Patient demographics and outcomes

Total cohort(n = 64)

Patients with CIPNM(n = 18)

Patients without CIPNM(n = 46)

p value

Gender (% male) 42 (65 %) 11/18 (61 %) 31/46 (67 %) 0.63Median age 50 (37–63) 58 (39–66) 49 (36–62) 0.30Median APACHE II score 20.5 (16–24) 22 (19–24) 20 (16–23) 0.51Race (% Caucasian) 43 (67 %) 11/18 (61 %) 32/46 (70 %) 0.52History of diabetes 21 (33 %) 9/18 (50 %) 12/46 (26 %) 0.07History of HIV 3 (5 %) 1/18 (6 %) 2/46 (4 %) 0.84History of alcohol use disorder 19 (31 %) 4/18 (22 %) 15/44 (34 %) 0.35Median ICU-free days 9 (0–20) 0 (0–19) 11 (0–20) 0.04Median hospital LOS 18 (13–33) 21 (14–33) 17 (13–33) 0.77Mortality 23 % (15/64) 50 % (9/18) 13 % (6/46) 0.002

686

Because there were no significant differences inbilateral nerve amplitudes, we averaged the right and leftnerve amplitudes when they were both performed.Though the diagnostic criteria for CIPNM requireabnormal NCS in only two nerves, the median SNAPand CMAP amplitudes were significantly lower inpatients with CIPNM when compared to those patientswithout CIPNM in all six of the tested nerves (Table 4).To further determine the diagnostic accuracy of the NCSfor each single nerve, we constructed individual ROCcurves (for each of the six nerves). The best amplitudecutoff value and the resulting sensitivities and specific-ities for all of the six nerves are displayed in Table 5.Overall, 32 patients had peroneal NCS within the normalrange for our laboratory (amplitude of greater than 0.8).Only one of these patients met the diagnostic criteria forCIPNM yielding a sensitivity of 97 % (95 %CI = 91–100 %). The lesser sensitivity of the tibialnerve may be explained by the broad amplitude rangeand typically high amplitude of the tibial CMAP. Usinga difference in the c-statistic of 0.05 as being clinicallysignificant, the motor nerve with the best diagnosticaccuracy was the peroneal nerve. The sensory nerve withthe best diagnostic accuracy was the sural nerve. Whenthe diagnostic accuracy of the peroneal and sural nerveamplitudes were examined together, the c-statistic sig-nificantly increased to 0.9336 with a sensitivity of 100 %

(95 % CI = 100–100 %) and specificity of 81 % (95 %CI = 71–91 %).

Discussion

In this prospective study, we confirmed that the devel-opment of CIPNM is common in critically ill patients, andis associated with both an increased mortality and adecrease in the number of ICU-free days [32, 33]. Inaddition, we identified that the performance of NCS ismore likely to be technically limited in the upper ascompared to the lower extremities. We also determinedthat the mean SNAP and CMAP amplitudes are sym-metric between the right and left sides for the same nerve.Considering that CIPNM is a diffuse disorder, symmetricchanges in the SNAP and CMAP amplitudes are expec-ted. Most importantly, we identified that the peronealnerve is the most accurate motor nerve and the sural nerveis the most accurate sensory nerve to identify the presenceof CIPNM. When the NCS of the peroneal and suralnerves are combined, their diagnostic accuracy for CIP-NM is extremely high.

These results identify a more streamlined electro-physiological screening strategy to identify CIPNM. Thetime required to perform a single NCS has been estimated

Table 3 Ability to perform NCS for each one of the 12 specific nerves

NCS able tobe performed

Inability to perform NCSdue to presence ofarterial or venous catheters

Inability to perform NCSdue to electrical interference

Inability to perform NCSdue to patient refusal

R sural 84 % (n = 54) 9 % (n = 6) 2 % (n = 1) 5 % (n = 3)L sural 86 % (n = 55) 8 % (n = 5) 3 % (n = 2) 3 % (n = 2)R radial 52 % (n = 33) 45 % (n = 29) 0 % (n = 0) 3 % (n = 2)L radial 43 % (n = 28) 53 % (n = 34) 2 % (n = 1) 2 % (n = 1)R sensory median 52 % (n = 33) 45 % (n = 29) 0 % (n = 0) 3 % (n = 2)L sensory median 40 % (n = 26) 55 % (n = 35) 2 % (n = 1) 3 % (n = 2)R peroneal 84 % (n = 54) 5 % (n = 3) 5 % (n = 3) 6 % (n = 4)L peroneal 82 % (n = 53) 10 % (n = 6) 3 % (n = 2) 5 % (n = 3)R tibial 87 % (n = 56) 5 % (n = 3) 2 % (n = 1) 6 % (n = 4)L tibial 86 % (n = 55) 6 % (n = 4) 3 % (n = 2) 5 % (n = 3)R motor median 55 % (n = 35) 40 % (n = 26) 0 % (n = 0) 5 % (n = 3)L motor median 42 % (n = 27) 50 % (n = 32) 3 % (n = 2) 5 % (n = 3)

Table 4 Nerve amplitudes and right to left differences at the time of the last NCS

Nerve (recording site) Amplitude in patientswith CIPNM

Amplitude in patientswithout CIPNM

p value Median difference betweenR and L nerve amplitudes

p value

Sural antidromic (n = 61) 0 (0–2.12) 5.5 (2.6–13.1) \0.0001 0 (-1, 1.23) 0.59Radial antidromic (n = 61) 14.0 (7.6–20.70) 21 (13.7–28.4) 0.02 – –Sensory median orthodromic (n = 59) 3.6 (0–5.1) 7.0 (4.0–11.5) 0.005 – –Peroneal (EDB) (n = 60) 0 (0–0.5) 1.4 (0.7–2.5) \0.0001 0.1 (-0.4, 0.7) 0.91Tibial (AHB) (n = 63) 2.6 (0.8–4.4) 7.6 (5.1–8.5) \0.0001 -0.3 (-1.7, 1.1) 0.21Motor median (APB) (n = 60) 3.2 (1.4–5.2) 5.5 (3.8–7.2) 0.0093 – –

EDB extensor digitorum brevis, AHB abductor hallucis brevis, APB abductor pollicis brevis

687

to be 5–10 min, instead of up to 90 min for a compre-hensive electrophysiological examination [15]. Since theNCS studies of the peroneal and sural nerves are highlysensitive, patients with normal studies are very unlikely tohave CIPNM and needle EMG examination is unlikely toalter this conclusion. If NCS of the peroneal and suralnerves are abnormal, the diagnosis can then be exploredfurther with more extensive testing such as EMG studies.If electrophysiological testing can be simplified, thediagnosis of CIPNM may be made more rapidly with lessdiscomfort to the patient. Therefore, electrophysiologicaltesting may appropriately become more available forcritically ill patients and serve as a routine method fordiagnosing and monitoring weakness in these patients.

Our results extend the findings of Weber-Carstens whoexamined the predictive ability of two sensory and twomotor nerves for the development of ICU-acquiredweakness in critically ill patients [34, 35]. In this study,ICU-acquired weakness was defined as reduced musclestrength as assessed by the Medical Research Council(MRC) scale. The study identified that reductions in thecompound muscle action potentials after direct musclestimulation of less than 3 mV were predictive of theeventual development of ICU-acquired weakness. How-ever, these investigators did not compare the predictiveability of the four different nerves that were tested. Ourresults are also consistent with the findings of Latronicoand colleagues. These investigators studied ICU patientsto determine the ability of unilateral NCS examinations ofthe median, peroneal, and sural nerves to diagnose CIP-NM. In their study, the diagnosis of CIPNM was made onthe basis of significant reductions in the NCS of twonerves. Concentric needle examination was not routinelyperformed. They identified that reduced amplitudes of theperoneal nerve resulted in the best diagnostic accuracy forCIPNM [15].

Our study has several important limitations. BecauseNCS values are included as one of the diagnostic criteriafor CIPNM, our results are prone to incorporation bias[36, 37]. However, because we are comparing severalnerves that are all prone to incorporation bias, the com-parisons between each nerve should not be affected.Secondly, we did not validate our findings in a separatecohort of patients. Further studies are planned to validate

our findings and determine their generalizability in criti-cally ill patients. Thirdly, it is also possible thatabnormalities in a single nerve may be related to acutesingle nerve palsies in response to trauma or positionalcompression. A visual assessment of the limbs was per-formed to help assess these possibilities, and no patientswere excluded from the analysis on the basis of a suspi-cion of focal nerve injury. Finally, it is difficult todifferentiate whether our electrophysiological findings areindicative of the development of an acute CIPNM or thediscovery of a chronic pre-existing polyneuromyopathy. Itis possible that the reduced sural SNAP and peronealCMAP may have reflected pre-existing polyneuropathy insome patients, as reductions in these parameters are wellknown to be associated with various chronic polyneur-opathies not just CIP [27]. However, patients with knownpre-existing polyneuromyopathy were excluded from thisstudy. Therefore, the utility of screening for CIPNM withsingle nerve conduction studies will most likely not besuitable for patients with pre-existing chronic neuropa-thies or for patients with chronic conditions such asdiabetes that are associated with chronic neuropathies.

Due to delirium, sedation, and/or encephalopathy,many critically ill patients are unable to give a full effortduring bedside strength testing and the use of the MRCscore to define weakness can be imperfect. However,some investigators have recommended that the diagnosisof CIPNM be made on the MRC score alone withoutelectrophysiological testing [18]. As recognized by theseinvestigators, the timing of the clinical examination incritically ill patients may be difficult because patientsmust be cooperative to perform the different componentsof the examination. In one study of patients requiringmechanical ventilation for more than 3 days, manualmuscle testing during critical illness was only possible in25 % of patients [28]. Before the patient regains theability to participate in formal manual muscle testing,electrophysiological testing can differentiate a weakpatient who is deconditioned from those with CIPNM.Our data demonstrated different outcomes in patients withCIPNM when compared to those who were deconditionedor had relatively normal neurological function. Accuratedifferentiation of deconditioning from CIPNM may havepotentially important therapeutic implications specifically

Table 5 Accuracy of each nerve amplitude for the diagnosis of CIPNM

Nerve (recording site) c-StatisticAUC

Best cutoffamplitude value

Normal nerveamplitude

Sensitivity(with 95 % CI)

Specificity(with 95 % CI)

Sural antidromic (n = 61) 0.8611 4.0 lV [10 lV 94 % (88–100 %) 70 % (59–81 %)Radial antidromic (n = 61) 0.6903 16.2 lV [20 lV 75 % (64–86 %) 71 % (60–82 %)Sensory median orthodromic (n = 61) 0.7369 5.2 lV [10 lV 81 % (71–91 %) 67 % (55–79 %)Peroneal (EDB) (n = 60) 0.8856 0.65 mV [0.8 mV 94 % (88–100 %) 74 % (63–85 %)Tibial (AHB) (n = 63) 0.8315 5.8 mV [1 mV 94 % (88–100 %) 69 % (58–80 %)Motor median (APB) (n = 60) 0.7209 3.8 mV [5 mV 63 % (51–75 %) 77 % (65–89 %)

EDB extensor digitorum brevis, AHB abductor hallucis brevis, APB abductor pollicis brevis

688

in the utilization of early physical therapy andmobilization.

In conclusion, amplitude reductions in the actionpotentials of the sural and peroneal nerves provide thebest accuracy for the diagnosis of CIPNM. In light of thedifficulties in examining neuromuscular function in crit-ically ill patents, these results may have importantimplications in the utilization of electrophysiologicaltesting for critically ill patients. Our findings will require

future validation in a separate group of critically illpatients before they should be implemented in diagnosticalgorithms for CIPNM.

Acknowledgments Financial support: NIH R01 NR011051 (MM).

Conflicts of interest Drs. Moss, Yang, Macht, Sottile, and Quan,and Ms. Gray and McNulty report no disclosures.

References

1. Ali NA, O’Brien JM Jr, Hoffmann SPet al (2008) Acquired weakness,handgrip strength, and mortality incritically ill patients. Am J Respir CritCare Med 178:261–268

2. Batt J, dos Santos CC, Cameron JI,Herridge MS (2013) Intensive care unit-acquired weakness: clinical phenotypesand molecular mechanisms. Am JRespir Crit Care Med 187:238–246

3. Latronico N, Shehu I, Seghelini E(2005) Neuromuscular sequelae ofcritical illness. Curr Opin Crit Care11:381–390

4. Latronico N, Bolton CF (2011) Criticalillness polyneuropathy and myopathy: amajor cause of muscle weakness andparalysis. Lancet Neurol 10:931–941

5. Needham DM (2008) Mobilizingpatients in the intensive care unit:improving neuromuscular weakness andphysical function. JAMA300:1685–1690

6. Bolton CF, Gilbert JJ, Hahn AF,Sibbald WJ (1984) Polyneuropathy incritically ill patients. J NeurolNeurosurg Psychiatry 47:1223–1231

7. Bolton CF, Laverty DA, Brown JD,Witt NJ, Hahn AF, Sibbald WJ (1986)Critically ill polyneuropathy:electrophysiological studies anddifferentiation from Guillain-Barresyndrome. J Neurol NeurosurgPsychiatry 49:563–573

8. Bolton CF (2005) Neuromuscularmanifestations of critical illness.Muscle Nerve 32:140–163

9. De Jonghe B, Sharshar T, LefaucheurJP et al (2002) Paresis acquired in theintensive care unit: a prospectivemulticenter study. JAMA288:2859–2867

10. Khan J, Harrison TB, Rich MM, MossM (2006) Early development of criticalillness myopathy and neuropathy inpatients with severe sepsis. Neurology67:1421–1425

11. Lacomis D, Giuliani MJ, Van CA,Kramer DJ (1996) Acute myopathy ofintensive care: clinical,electromyographic, and pathologicalaspects. Ann Neurol 40:645–654

12. Lacomis D, Petrella JT, Giuliani MJ(1998) Causes of neuromuscularweakness in the intensive care unit: astudy of ninety-two patients. MuscleNerve 21:610–617

13. Lacomis D, Zochodne DW, Bird SJ(2003) Critical illness myopathy.Muscle Nerve 23:1785–1788

14. Latronico N, Fenzi F, Recupero D et al(1996) Critical illness myopathy andneuropathy. Lancet 347:1579–1582

15. Latronico N, Bertolini G, Guarneri Bet al (2007) Simplifiedelectrophysiological evaluation ofperipheral nerves in critically illpatients: the Italian multi-centreCRIMYNE study. Crit Care 11:R11

16. Stevens RD, Marshall SA, CornblathDR et al (2009) A framework fordiagnosing and classifying intensivecare unit-acquired weakness. Crit CareMed 37:S299–S308

17. Leijten FS, Harinck-de Weerd JE,Poortvliet DC, de Weerd AW (1995)The role of polyneuropathy in motorconvalescence after prolongedmechanical ventilation. JAMA274:1221–1225

18. De JB, Finfer S (2007) Critical illnessneuromyopathy: from risk factors toprevention. Am J Respir Crit Care Med175:424–425

19. Carson SS (2006) Outcomes ofprolonged mechanical ventilation. CurrOpin Crit Care 12:405–411

20. Schweickert WD, Pohlman MC,Pohlman AS et al (2009) Early physicaland occupational therapy inmechanically ventilated, critically illpatients: a randomised controlled trial.Lancet 373:1874–1882

21. Macht M, Quan Q, Clark BJ, BensonAB, Burnham EL, Moss M (2011)Development of an efficient screeningtest for critical illness polyneuropathy.Am J Respir Crit Care Med 183:A2649

22. Bernard GR, Vincent JL, Laterre PFet al (2001) Efficacy and safety ofrecombinant human activated protein Cfor severe sepsis. N Engl J Med344:699–709

23. Martin GS, Mannino DM, Eaton S, MossM (2003) The epidemiology of sepsis inthe United States from 1979 through2000. N Engl J Med 348:1546–1554

24. Kimura J (ed) (2001) Electrodiagnosisin disease of nerve and muscle:principles and practice, 3rd edn. OxfordUniversity Press, New York

25. Chaudhry V, Cornblath DR, Mellits EDet al (1991) Inter- and intra-examinerreliability of nerve conductionmeasurements in normal subjects. AnnNeurol 30:841–843

26. Chaudhry V, Corse AM, Freimer MLet al (1994) Inter- and intraexaminerreliability of nerve conductionmeasurements in patients with diabeticneuropathy. Neurology 44:1459–1462

27. England JD, Gronseth GS, Franklin Get al (2005) Distal symmetricalpolyneuropathy: definition for clinicalresearch. Muscle Nerve 31:113–123

28. Hough CL, Lieu BK, Caldwell ES (2011)Manual muscle strength testing ofcritically ill patients: feasibility andinterobserver agreement. Crit Care 15:R43

29. Wiedemann HP, Wheeler AP, BernardGR et al (2006) Comparison of two fluid-management strategies in acute lunginjury. N Engl J Med 354:2564–2575

30. Pencina MJ, D’Agostino RB Sr,D’Agostino RB Jr, Vasan RS (2008)Evaluating the added predictive abilityof a new marker: from area under theROC curve to reclassification andbeyond. Stat Med 27:157–172

31. Ho JE, Wang TJ (2012) Growthdifferentiation factor 15: a canary in acoal mine? Clin Chem 58:3–5

32. Garnacho-Montero J, Madrazo-Osuna J,Garcia-Garmendia JL et al (2001)Critical illness polyneuropathy: riskfactors and clinical consequences. Acohort study in septic patients. IntensiveCare Med 27:1288–1296

33. Garnacho-Montero J, Amaya-Villar R,Garcia-Garmendia JL, Madrazo-OsunaJ, Ortiz-Leyba C (2005) Effect ofcritical illness polyneuropathy on thewithdrawal from mechanical ventilationand the length of stay in septic patients.Crit Care Med 33:349–354

689

34. Weber-Carstens S, Koch S, Spuler Set al (2009) Nonexcitable musclemembrane predicts intensive care unit-acquired paresis in mechanicallyventilated, sedated patients. Crit CareMed 37:2632–2637

35. Weber-Carstens S, Deja M, Koch S et al(2010) Risk factors in critical illnessmyopathy during the early course ofcritical illness: a prospectiveobservational study. Crit Care 14:R119

36. Lijmer JG, Mol BW, Heisterkamp Set al (1999) Empirical evidence ofdesign-related bias in studies ofdiagnostic tests. JAMA 282:1061–1066

37. Rutjes AW, Reitsma JB, Di NM, SmidtN, van Rijn JC, Bossuyt PM (2006)Evidence of bias and variation indiagnostic accuracy studies. CMAJ174:469–476

690