Before–after study of a standardized hospital order set for … · ment and intensive care units...

7
Feature Articles Before–after study of a standardized hospital order set for the management of septic shock* Scott T. Micek, PharmD; Nareg Roubinian, MD; Tim Heuring, PharmD; Meghan Bode, PharmD; Jennifer Williams, APRN; Courtney Harrison, RN; Theresa Murphy, PharmD; Donna Prentice, MSN; Brent E. Ruoff, MD; Marin H. Kollef, MD S evere sepsis is an infection- induced syndrome resulting in a systemic inflammatory re- sponse that is complicated by dysfunction of at least one organ system (1). In the United States, approximately 750,000 cases of sepsis occur each year (2, 3). The mortality associated with se- vere sepsis ranges from 30% to 50%, with mortality increasing with advancing age (3, 4). Although complex, the patho- physiology of sepsis involves a series of interacting pathways involving immune stimulation, immune suppression, hy- percoagulation, and hypofibrinolysis (5, 6). Cardiovascular management plays an important role in the treatment of septic shock. Hypotension occurs due to failure of vasoconstriction by vascular smooth muscle, resulting in peripheral vasodila- tion (7, 8). Goal-directed cardiovascular resuscitation has been demonstrated to be an important determinant of survival in patients with septic shock (9). In addi- tion to cardiovascular management, ap- propriate initial antimicrobial treatment of patients with severe sepsis also seems to be an important determinant of patient outcome (10 –12). Barnes–Jewish Hospital is an urban tertiary referral center where patients who develop septic shock in the commu- nity setting are first evaluated in the emergency department. These patients typically are treated for 6 –24 hrs in the emergency department while awaiting a bed in the intensive care unit. The goal of this study was to determine whether a standardized physician order set focus- ing on intravenous fluid administration and the appropriateness of initial antimi- crobial therapy could improve patient management as recommended by the re- cent Surviving Sepsis Campaign guide- lines (13). MATERIALS AND METHODS Study Location and Patients. The study was conducted within the emergency depart- ment and intensive care units (medical, sur- gical–trauma) of an academic medical center, Barnes–Jewish Hospital/Washington Univer- sity Medical Center (1,200 beds) in St. Louis, MO, from December 2004 to November 2005. The emergency department is a closed unit where patient care is delivered by emergency department resident physicians under the su- *See also p. 2842. From the Departments of Pharmacy (STM, TH, MB, TM) and Heart Services (DP), Barnes–Jewish Hospital, St. Louis, MO; and the Pulmonary and Critical Care Division (NR, MHK) and Department of Emergency Medicine (JW, CH, BER), Washington University School of Medicine, St. Louis, MO. The authors have not disclosed any potential con- flicts of interest. Copyright © 2006 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/01.CCM.0000241151.25426.D7 Objective: To evaluate a standardized hospital order set for the management of septic shock in the emergency department. Design: Before–after study design with prospective consecu- tive data collection. Setting: Emergency department of a 1,200-bed academic med- ical center. Patients: A total of 120 patients with septic shock. Interventions: Implementation of a standardized hospital order set for the management of septic shock. Measurements and Main Results: A total of 120 consecutive patients with septic shock were identified. Sixty patients (50.0%) were managed before the implementation of the standardized order set, constituting the before group, and 60 (50.0%) were evaluated after the implementation of the standardized order set, making up the after group. Demographic variables and severity of illness measured by the Acute Physiology and Chronic Health Evaluation II were similar for both groups. Patients in the after group received statistically more intravenous fluids while in the emergency department (2825 1624 mL vs. 3789 1730 mL, p .002), were more likely to receive intravenous fluids of >20 mL/kg body weight before vasopressor administration (58.3% vs. 88.3%, p < .001), and were more likely to be treated with an appropriate initial antimicrobial regimen (71.7% vs. 86.7%, p .043) compared with patients in the before group. Patients in the after group were less likely to require vasopressor administration at the time of transfer to the intensive care unit (100.0% vs. 71.7%, p < .001), had a shorter hospital length of stay (12.1 9.2 days vs. 8.9 7.2 days, p .038), and a lower risk for 28-day mortality (48.3% vs. 30.0%, p .040). Conclusions: Our study found that the implementation of a standardized order set for the management of septic shock in the emergency department was associated with statistically more rigorous fluid resuscitation of patients, greater administration of appropriate initial antibiotic treatment, and a lower 28-day mor- tality. These data suggest that the use of standardized order sets for the management of septic shock should be routinely em- ployed. (Crit Care Med 2006; 34:2707–2713) KEY WORDS: septic shock; outcomes; vasopressors; intensive care; hospital mortality 2707 Crit Care Med 2006 Vol. 34, No. 11

Transcript of Before–after study of a standardized hospital order set for … · ment and intensive care units...

Feature Articles

Before–after study of a standardized hospital order set for themanagement of septic shock*

Scott T. Micek, PharmD; Nareg Roubinian, MD; Tim Heuring, PharmD; Meghan Bode, PharmD;Jennifer Williams, APRN; Courtney Harrison, RN; Theresa Murphy, PharmD; Donna Prentice, MSN;Brent E. Ruoff, MD; Marin H. Kollef, MD

Severe sepsis is an infection-induced syndrome resulting ina systemic inflammatory re-sponse that is complicated by

dysfunction of at least one organ system(1). In the United States, approximately750,000 cases of sepsis occur each year(2, 3). The mortality associated with se-vere sepsis ranges from 30% to 50%, withmortality increasing with advancing age(3, 4). Although complex, the patho-

physiology of sepsis involves a series ofinteracting pathways involving immunestimulation, immune suppression, hy-percoagulation, and hypofibrinolysis(5, 6).

Cardiovascular management plays animportant role in the treatment of septicshock. Hypotension occurs due to failureof vasoconstriction by vascular smoothmuscle, resulting in peripheral vasodila-tion (7, 8). Goal-directed cardiovascularresuscitation has been demonstrated tobe an important determinant of survivalin patients with septic shock (9). In addi-tion to cardiovascular management, ap-propriate initial antimicrobial treatmentof patients with severe sepsis also seemsto be an important determinant of patientoutcome (10–12).

Barnes–Jewish Hospital is an urbantertiary referral center where patientswho develop septic shock in the commu-nity setting are first evaluated in the

emergency department. These patientstypically are treated for 6–24 hrs in theemergency department while awaiting abed in the intensive care unit. The goal ofthis study was to determine whether astandardized physician order set focus-ing on intravenous fluid administrationand the appropriateness of initial antimi-crobial therapy could improve patientmanagement as recommended by the re-cent Surviving Sepsis Campaign guide-lines (13).

MATERIALS AND METHODS

Study Location and Patients. The studywas conducted within the emergency depart-ment and intensive care units (medical, sur-gical–trauma) of an academic medical center,Barnes–Jewish Hospital/Washington Univer-sity Medical Center (1,200 beds) in St. Louis,MO, from December 2004 to November 2005.The emergency department is a closed unitwhere patient care is delivered by emergencydepartment resident physicians under the su-

*See also p. 2842.From the Departments of Pharmacy (STM, TH, MB,

TM) and Heart Services (DP), Barnes–Jewish Hospital,St. Louis, MO; and the Pulmonary and Critical CareDivision (NR, MHK) and Department of EmergencyMedicine (JW, CH, BER), Washington University Schoolof Medicine, St. Louis, MO.

The authors have not disclosed any potential con-flicts of interest.

Copyright © 2006 by the Society of Critical CareMedicine and Lippincott Williams & Wilkins

DOI: 10.1097/01.CCM.0000241151.25426.D7

Objective: To evaluate a standardized hospital order set for themanagement of septic shock in the emergency department.

Design: Before–after study design with prospective consecu-tive data collection.

Setting: Emergency department of a 1,200-bed academic med-ical center.

Patients: A total of 120 patients with septic shock.Interventions: Implementation of a standardized hospital order

set for the management of septic shock.Measurements and Main Results: A total of 120 consecutive

patients with septic shock were identified. Sixty patients (50.0%)were managed before the implementation of the standardizedorder set, constituting the before group, and 60 (50.0%) wereevaluated after the implementation of the standardized order set,making up the after group. Demographic variables and severity ofillness measured by the Acute Physiology and Chronic HealthEvaluation II were similar for both groups. Patients in the aftergroup received statistically more intravenous fluids while in theemergency department (2825 � 1624 mL vs. 3789 � 1730 mL,p � .002), were more likely to receive intravenous fluids of >20

mL/kg body weight before vasopressor administration (58.3% vs.88.3%, p < .001), and were more likely to be treated with anappropriate initial antimicrobial regimen (71.7% vs. 86.7%, p �.043) compared with patients in the before group. Patients in theafter group were less likely to require vasopressor administrationat the time of transfer to the intensive care unit (100.0% vs.71.7%, p < .001), had a shorter hospital length of stay (12.1 � 9.2days vs. 8.9 � 7.2 days, p � .038), and a lower risk for 28-daymortality (48.3% vs. 30.0%, p � .040).

Conclusions: Our study found that the implementation of astandardized order set for the management of septic shock in theemergency department was associated with statistically morerigorous fluid resuscitation of patients, greater administration ofappropriate initial antibiotic treatment, and a lower 28-day mor-tality. These data suggest that the use of standardized order setsfor the management of septic shock should be routinely em-ployed. (Crit Care Med 2006; 34:2707–2713)

KEY WORDS: septic shock; outcomes; vasopressors; intensivecare; hospital mortality

2707Crit Care Med 2006 Vol. 34, No. 11

pervision of attending physicians board certi-fied in emergency medicine. The intensivecare units are closed units employing multi-disciplinary rounds directed by a physicianboard certified in critical care. The study wasapproved by the Washington University Schoolof Medicine Human Studies Committee. In-

formed consent was obtained for collection ofpatient data.

Hospital Order Set. The hospital order setfor the management of severe sepsis and septicshock focusing on hemodynamic resuscitationand antimicrobial treatment was developed bya committee comprising emergency depart-

ment physicians, pharmacists, and criticalcare physicians chaired by one of the authors(S. T. Micek) (Figs. 1 and 2). An additionalorder set focusing on secondary managementissues, including the administration of corti-costeroids and drotrecogin alfa (activated),was employed but is not shown. The order set

Figure 1. Hospital order set for the administration of antibiotic therapy for severe sepsis and septic shock in the emergency department. IVPB, intravenouspiggyback; Q, every; PCN, penicillin; max, maximum.

2708 Crit Care Med 2006 Vol. 34, No. 11

was based on recommendations derived fromthe Surviving Sepsis Campaign (13). Formaleducation and clinical training of all physi-cians, nurses, and patient care technicians inthe emergency department pertaining to theprocesses and procedures related to the orderset forms was completed in a 4-wk period by

two of the investigators before the after phase(J. Williams and C. Harrison). These educa-tional endeavors included training in sepsispathophysiology, monitoring of central ve-nous pressures, assessment of central venousblood oxygen saturation, and the pharmaco-therapy of sepsis (1, 13).

Study Entry Criteria. To be enrolled intothe study, patients had to have a clinical diag-nosis of septic shock that included: 1) thepresence of two or more signs of the systemicinflammatory response syndrome, 2) a pre-sumed site of infection documented by radio-graphic findings consistent with infection or a

Figure 2. Hospital order set for hemodynamic resuscitation of patients with severe sepsis and septic shock in the emergency department. NS, normal saline;IV, intravenous.

2709Crit Care Med 2006 Vol. 34, No. 11

clinical syndrome associated with a high prob-ability of infection, and 3) vasodilatory shockrequiring fluid resuscitation plus vasopressoradministration in the emergency department.To more fully assess the effect of the order seton hemodynamic resuscitation, only patientswith septic shock were enrolled. The prescrip-tion of vasopressors (e.g., norepinephrine ordopamine) was not based on predeterminedhemodynamic cutoffs but was prescribedsolely at the discretion of the treating physi-cians. Patients with refractory septic shockdying in the emergency department despitevasopressor administration were excluded tomore accurately assess the effect of the orderset on hospital outcomes. Patients with analternative diagnosis accounting for the shockstate (e.g., myocardial infarction, pulmonaryembolism) and those with a preexisting do-not-resuscitate order were also excluded. Ta-pering of vasopressors occurred by reducingthe norepinephrine or dopamine infusions byincrements of 1.0 �g/min and 0.05–1.0�g·kg�1·min�1, respectively.

Study Design and Data Collection. A pro-spective, before–after study design was used,enrolling consecutive patients meeting thestudy entry criteria. The primary outcomesevaluated were the total quantity of intrave-nous fluids administered and the prescriptionof appropriate initial antimicrobial treatmentin the emergency department. Secondary out-comes included hospital mortality and hospi-tal length of stay.

For all study patients, the following patientcharacteristics were prospectively recorded:age, sex, race, body weight, and severity ofillness as assessed by the Acute Physiology andChronic Health Evaluation (APACHE) II scorebased on the worst values obtained while thepatient was in the emergency department.Process-of-care variables evaluated in theemergency department included administra-tion of antibiotics within 3 hrs of arrival, ob-taining blood cultures before the start of an-tibiotics, administration of �20 mL/kgintravenous fluids before vasopressors, totalintravenous fluid volume administered, unitsof packed red blood cells transfused in theemergency department, assessment of a se-rum lactate measurement, use of vasopressorsat the time of transfer out of the emergencydepartment, employing a central vein pressuretransducer, documenting the achievement ofa central venous pressure of �8 mm Hg, eval-uation of a central venous blood oxygen satu-ration, and the use of corticosteroids and dro-trecogin alfa (activated).

One of the investigators made daily roundson all study patients, recording relevant datafrom the medical records, bedside flowsheets,and the hospital’s mainframe computer forreports of microbiological studies (Gramstains and cultures of blood, urine, sputum,lower respiratory tract specimens, tissue, andwounds). All pharmacotherapies administeredin the emergency department and the inten-sive care units were evaluated using patients’

medical records and the hospital’s computer-ized bedside workstations (EMTEK HealthCare Systems, Tempe, AZ; and Clinical Desk-top, BJC Healthcare, St. Louis, MO).

Definitions. All definitions were selected apriori as part of the original study design.Vasopressors were targeted to maintain amean arterial pressure of �65 mm Hg. Refrac-tory septic shock was defined as the inabilityto maintain a mean arterial pressure of �65mm Hg with the administration of vasopres-sors. The fluid resuscitation of patients wasevaluated during their emergency departmentstay before transfer to an intensive care unit.The types of crystalloid fluids administered inthe emergency department included 0.9% so-dium chloride and lactated Ringer solution.For the purposes of this investigation, appro-priate initial antimicrobial treatment was de-fined as the microbiological documentation ofan infection (i.e., a positive culture result) thatwas being effectively treated based on in vitrosusceptibility results at the time of its identi-fication (14). Corticosteroid therapy was com-posed of 200–300 mg/day hydrocortisone orits equivalent dose. Patients were not requiredto have a diagnosis of adrenal insufficiency basedon random cortisol levels or an adrenocortico-tropic hormone (ACTH) stimulation test.

APACHE II scores were calculated on thebasis of the worst clinical data available forpatients while in the emergency department(15). The definition for systemic inflammatoryresponse syndrome was that proposed by theAmerican College of Chest Physicians/Societyof Critical Care Medicine Consensus Confer-ence (16). The use of drotrecogin alfa (acti-vated) was based on the criteria in the originalstudy design (17).

Statistical Analysis. Continuous data werereported as mean � SD, and the Student’st-test was employed when comparisons weremade for parametric data. Nonparametric datawere analyzed with the Mann-Whitney U test.Categorical variables were reported as fre-

quency distributions, and chi-square or Fish-er’s exact tests were used to test whether dif-ferences existed between groups. After theseunivariate analyses, multivariate logistic re-gression was undertaken to determine inde-pendent risk factors for hospital mortality.Risk factors significant at the .2 level in theunivariate analysis were entered into themodel. Adjusted odds ratios and their corre-sponding 95% confidence intervals are re-ported. Kaplan-Meier curves representing the28-day mortality stratified according to groupassignment were compared with the use of alog-rank test. All tests were two-tailed, and a pvalue of �.05 was predetermined to representstatistical significance. Analyses were done us-ing the SPSS 10.1 software package (SPSS,Chicago, IL). We calculated post hoc that wewould need a sample size of 128 patients toidentify an absolute difference in the amountof fluid administration to patients before start-ing vasopressors of 5 mL/kg (estimated SD, 10mL/kg) with a power of 0.8 (two-tailed) at asignificance level of .05.

RESULTS

Patients. A total of 125 consecutivepatients with septic shock were prospec-tively evaluated. Five patients with refrac-tory septic shock dying in the emergencydepartment were excluded (four in thebefore group and one in the after group)(Tables 1 and 2). The mean age of thepatients was 64.7 � 18.2 yrs (range,19–97 yrs); 53 patients (44.2%) were menand 67 (55.8%) were women. The meanAPACHE II score of the entire cohort was22.5 � 8.3 (range, 5–46). Patients in thebefore and after groups were similar interms of baseline demographics, primaryinfection site, and APACHE II scores (Ta-ble 1).

Table 1. Patient demographics and severity of illness

VariableBefore Group

(n � 60)After Group

(n � 60) p Value

Age, yrs 68.0 � 16.1 61.4 � 20.0 .066Sex, n (%) .581

Male 25 (41.7) 28 (46.7)Female 35 (58.3) 32 (53.3)

Race, n (%) .091White 24 (40.0) 34 (56.7)African-American 34 (56.7) 26 (43.3)Other 2 (3.3) 0 (0.0)

Body weight, kg 73.7 � 24.9 71.3 � 13.9 .806Primary infection site, n (%) .969

Lung 17 (28.3) 17 (28.3)Intraabdominal 17 (28.3) 15 (25.0)Urinary tract 15 (25.0) 17 (28.3)Other 11 (18.3) 11 (18.3)

APACHE II score 21.7 � 6.8 23.3 � 9.6 .305

APACHE II, Acute Physiology and Chronic Health Evaluation II.Unless indicated otherwise, data provided as mean � SD.

2710 Crit Care Med 2006 Vol. 34, No. 11

Process-of-Care Variables. All patientsin the before and after groups had centralvenous catheters placed in the emergencydepartment. Patients in the after groupwere statistically more likely to have hadintravenous antibiotics administeredwithin 3 hrs of arrival at the emergencydepartment and to have received an ini-tial antibiotic regimen that was appropri-ate for their infection compared with pa-

tients in the before group (Table 2).Patients in the after group also receivedsignificantly greater total volume of in-travenous fluids in the emergency depart-ment (Fig. 3) and total volume of intra-venous fluids before the start ofvasopressors, and they were statisticallymore likely to receive �20 mL/kg of in-travenous fluids administered before va-sopressors.

Serum lactate and central venous ox-ygen saturation measurements were sta-tistically more often performed amongpatients in the after group. Patients inthe after group were more likely to

achieve a documented central venouspressure of �8 cm H2O, less likely torequire vasopressors at the time of trans-fer to an intensive care unit, less likely toreceive corticosteroids and drotrecoginalfa (activated), and had a longer emer-gency department length of stay com-pared with patients in the before group.Of patients having central venous oxygensaturation assessments performed in theemergency department during the afterperiod, 27 of 29 (93.1%) achieved a valueof �70%. The mean central venous oxy-gen saturation achieved among patientsin the after group before transfer out ofthe emergency department was 79.5% �10.9%.

Outcomes. Patients in the after grouphad a statistically lower risk of 28-daymortality, (48.3% vs. 30.0%, p � .040)compared with patients in the beforegroup. Kaplan-Meier plots of the proba-bility of remaining alive are shown inFigure 4. Patients in the after group hada statistically higher probability of 28-daysurvival compared with patients in thebefore group (p � .001 by log-rank test).Hospital mortality was similar, with48.3% of patients in the before group and35.0% of the patients in the after groupdying by the end of their hospitalization(p � .139). The hospital length of staywas significantly lower for patients in theafter group (12.1 � 9.2 days vs. 8.9 � 7.2days, p � .038). The intensive care unitlength of stay did not differ between groupsstatistically (6.6 � 7.0 days vs. 5.1 � 5.2days, p � .228). Logistic regression analysisdemonstrated that not achieving �20mL/kg intravenous fluid administration be-fore vasopressors and increasing patientage were independently associated withhospital mortality (Table 3).

DISCUSSION

This study demonstrated that use of astandardized physician order set em-ployed in the emergency department wasstatistically more likely to result in theadministration of appropriate initial anti-microbial treatment and targeted goal-directed intravenous fluids among pa-tients with septic shock. We also foundthat the 28-day mortality rate and hospi-tal length of stay favored patients man-aged in the after group with the standard-ized order set. Finally, patients in theafter group were less likely to requiretreatment with vasopressors at the timeof transfer to the intensive care unit com-pared with patients in the before group.

Table 2. Processes of medical care

VariableBefore Group

(n � 60)After Group

(n � 60) p Value

Antibiotic administration within 3 hrs of EDarrival, n (%)

36 (60.0) 52 (86.7) .001

Appropriate initial antibiotic treatment,n (%)

43 (71.7) 52 (86.7) .043

Blood cultures obtained before antibioticadministration, n (%)

47 (78.3) 51 (85.0) .345

Intravenous fluids administered in ED, mL 2825 � 1624 3789 � 1730 .002Intravenous fluids administered before

vasopressors, mL1740 � 1267 2771 � 1242 �.001

Achieved 20 mL/kg intravenous fluids beforevasopressors, n (%)

35 (58.3) 53 (88.3) �.001

Transfused RBC units, n (%) 4 (6.7) 12 (20.0) .032ED length of stay, hrs 5.8 � 3.6 7.3 � 4.0 .015Serum lactate measurement, n (%) 10 (16.7) 47 (78.3) �.001Documented central venous pressure of �8

mm Hg in ED, n (%)3 (5.0) 29 (48.3) �.001

Central venous oxygen saturationassessment in the ED, n (%)

1 (1.7) 29 (48.3) �.001

Vasopressor administrationa, n (%) 60 (100.00) 43 (71.7) �.001Corticosteroid administration, n (%) 30 (50.0) 13 (21.7) .001Drotrecogin alfa (activated) administration,

n (%)7 (11.7) 2 (3.3) .083

ED, emergency department; RBC, red blood cell. Unless indicated otherwise, data provided asmean � SD.

aAt the time of transfer to the intensive care unit.

Figure 3. Box plots of the total intravenous fluidadministered to patients in the emergency de-partment for septic shock before transfer to theintensive care unit. Boxes represent the 25th and75th percentiles with 50th percentile valuewithin the boxes (solid line). The mean value isrepresented as a dotted line within the boxes. The10th and 90th percentiles are shown as cappedbars, and filled circles mark the 5th and 95thpercentiles.

Figure 4. Kaplan-Meier curves depicting theprobability of survival in patients with septicshock managed during the before group (filledcircles) and patients managed during the aftergroup (open circles) according to the duration ofsurvival (p � .001 by log-rank test).

2711Crit Care Med 2006 Vol. 34, No. 11

Physicians practicing in the emer-gency department setting, similar to theintensive care unit setting, are frequentlyfaced with the challenge of caring formultiple patients with potentially life-threatening illnesses. Septic shock is adisease state requiring timely and di-rected interventions. Rivers et al. (9)demonstrated a 16% decrease in absolute28-day mortality by implementing anearly goal-directed therapy protocol intheir emergency department. This proto-col focused on intensive care unit–levelcare in the emergency department withspecific targeted end points for the ad-ministration of intravenous fluids and va-sopressors. The major differences intreatment between the intervention andcontrol groups were in the volume ofintravenous fluids received, the percent-age of patients transfused packed redblood cells in the first 6 hrs, the use ofdobutamine, and the presence of a dedi-cated study team (9).

As it has become clear that severe sep-sis and septic shock are common causesof death (2, 3), other groups have alsofocused on the management of this dis-order to improve patient outcomes (18).Gao et al. (19) evaluated the outcomes ofpatients with sepsis in terms of whethercompliance with sepsis care bundles oc-curred during their management. Thesepsis care bundles provided recommen-dations for the management of intrave-nous fluids, blood transfusions, antibiot-ics, and vasopressors. There was anoverall compliance of 52% with the sepsisbundles in the study by Gao et al (19).Despite being comparable in terms ofbaseline demographics and severity of ill-ness, the compliant patients had a statis-tically lower risk of hospital mortality

compared with the noncompliant pa-tients (29% vs. 55%, p � .045). Our in-vestigation differed from the study of Gaoet al. (19) in that we focused on theemergency department population, at-tempting to target the management ofintravenous fluid administration andappropriate initial antibiotic treatment.

Previous investigations have shownthat antimicrobial regimens lacking ac-tivity against identified microorganismscausing serious infections (e.g., hospital-acquired pneumonia, bloodstream infec-tions) are associated with greater hospitalmortality (14, 20, 21). More recently, thesame finding has been demonstrated forpatients with severe sepsis (10–12, 22).Inappropriate antimicrobial treatmenthas been shown to be an important inde-pendent risk factor for mortality amonghospitalized patients with bloodstreaminfections (20). Unfortunately, changingantimicrobial therapy to an appropriateregimen after susceptibility data becomeavailable has not been demonstrated toimprove clinical outcomes (23, 24). Pre-vious studies have also suggested that thetiming of appropriate antibiotic treat-ment for sterile-site infections is an im-portant determinant of outcome. Ireguiet al. (25) showed that patients treated for�12 hrs after clinical suspicion of venti-lator-associated pneumonia had a statis-tically greater risk of hospital mortality.Similarly, Morrell et al. (26) demon-strated that delayed appropriate treat-ment for Candida bloodstream infectionis associated with increased hospital mor-tality. These investigations suggest thatearly accurate microbiological diagnosisto guide specific treatment or appropriateempirical treatment before culture re-

sults becoming available is necessary toimprove outcomes of serious infections.

Standardization of evidence-basedpractices in the care of critically ill pa-tients has become accepted as the opti-mal method for their management. Lib-eration from mechanical ventilation,avoiding excess sedation administration,and antibiotic treatment of serious infec-tions are several examples in which prac-tice protocols have improved clinical out-comes (27–29). By standardizing care,clinicians ensure that necessary proce-dures and therapies are carried out in atimely manner. They also allow new prac-tice changes to be more accurately mon-itored in terms of their effect on patientoutcomes. The management of septicshock easily lends itself to standardiza-tion because of the importance of achiev-ing early goal-directed resuscitation andadministration of appropriate antimicro-bial treatment (9–12).

Our study has several important limi-tations. First, it was performed within asingle emergency department and maynot be generalizable to other treatmentsettings. However, the results are consis-tent with those demonstrated by otherinvestigations performed in emergencydepartments (9, 19). Second, our studydesign and relatively small sample sizelimit our ability to determine a causalrelationship between the use of the stan-dardized order sets and the outcomes weevaluated. Nonetheless, the absolute re-duction in 28-day mortality of 15% thatwe achieved is similar to that found byRivers et al. (9) and is consistent with theother findings of our investigation. Third,we primarily focused on two elements inthe management of septic shock. Thestandardized order sets focused on fluidadministration and appropriate initial an-tibiotic therapy, although recommenda-tions on vasopressor use and packed redblood cell transfusion targets were alsoprovided. It is possible that our order setcould be improved by incorporating otherelements such as glycemic control, lung-protective mechanical ventilation, andnutritional therapy, especially if em-ployed in the intensive care unit setting(13, 30). In addition, despite all 60 pa-tients having a central venous catheterplaced in the after group, only 48% had adocumented central venous pressuremeasurement in the emergency depart-ment, suggesting the need for furtherimprovement.

Another important limitation of thisinvestigation is that it was an observa-

Table 3. Multivariate analyses of independent risk factors for 28-day mortalitya

Variables Adjusted Odds Ratio 95% CI p Value

Model 1Not achieving 20 mL/kg intravenous fluid

administration before vasopressors2.66 1.67–4.24 .036

APACHE II score (1-point increments) 1.10 1.07–1.13 .001Model 2

Before group assignment 2.48 1.65–3.89 .031APACHE II score (1-point increments) 1.11 1.08–1.14 .001

APACHE, Acute Physiology and Chronic Health Evaluation; CI, confidence interval.aBecause of colinearity between patient group assignment and not achieving at least 20 mL/kg

intravenous fluid administration before vasopressors, two separate models are displayed, with only oneof these two variables entered into each model, respectively.

Other covariates not presented in the table had a p value of �.05, including appropriate initialantibiotic administration, requiring vasopressors at the time of transfer out of the emergencydepartment, and increasing age in 1-point increments (Hosmer-Lemeshow goodness-of-fit test,p � .45).

2712 Crit Care Med 2006 Vol. 34, No. 11

tional, nonrandomized study. Therefore,unforeseen biases on the part of practi-tioners caring for these patients, includ-ing uncontrolled changes in practices,could have occurred, contributing to theresults. However, the routine use of prac-tice protocols in our intensive care unitsshould have minimized this effect (28,29). Finally, our study design does notallow us to determine which specific in-terventions accounted for the observedbenefit in outcomes. Only 48% of ourpatients had a documented central ve-nous pressure measurement in the emer-gency department. Therefore, it is possi-ble that elements of the order set otherthan aggressive fluid resuscitation mayhave contributed to the observed out-come benefit. Despite these limitations,our study supports the importance ofachieving targeted end points for themanagement of septic shock to improvepatient outcomes.

The findings of this study confirm thepremise that patients with septic shockare often under-resuscitated in the emer-gency department setting and will benefitfrom a standardized order set includingaggressive early hemodynamic resuscita-tion that meets defined measurable endpoints and a prescribed antibiotic path-way. Unfortunately, it seems that mostinstitutions have been slow to adopt theprotocol of early goal-directed fluid re-suscitation, despite the evidence in sup-port of its effect on patient outcomes (9,19, 31). There is a requirement for extraresources, time, and equipment in imple-menting this strategy. However, the po-tential outcomes benefit seems to easilyjustify these resources, especially in pa-tients with septic shock, given the likeli-hood for reductions in morbidity andmortality.

In summary, the initial managementof patients with septic shock seems to becritical in terms of determining outcome(9). Most patients developing septic shockin the community setting will initially betreated in a hospital emergency depart-ment. Our findings suggest that emer-gency departments should apply stan-dardized physician order sets, or someother systematic approach, for the man-agement of patients with septic shock.Given that physician order sets exposepatients to no additional risks and areassociated with little to no acquisitioncosts, their implementation should be-come the standard of care for the man-agement of septic shock. We have cur-rently broadened the scope of use of these

order sets at Barnes–Jewish Hospital sothat they are now applied hospital-wide tomanage patients with septic shock.

REFERENCES

1. Micek ST, Shah RA, Kollef MH: Managementof severe sepsis: Integration of multiple phar-macologic interventions. Pharmacotherapy2003; 23:1486–1496

2. Angus DC, Linde-Zwirble WT, Lidicker J, etal: Epidemiology of severe sepsis in theUnited States: Analysis of incidence outcomeand associated costs of care. Crit Care Med2001; 29:1303–1310

3. Martin GS, Mannino DM, Eaton S, et al: Theepidemiology of sepsis in the United Statesfrom 1979 through 2000. N Engl J Med 2003;348:1546–1554

4. Angus DC, Wax RS: Epidemiology of sepsis:An update. Crit Care Med 2001; 29(Suppl):S109–S116

5. Hotchkiss RS, Karl IE: The pathophysiologyand treatment of sepsis. N Engl J Med 2003;348:138–150

6. Wheeler AP, Bernard GR: Treating patientswith severe sepsis. N Engl J Med 1999; 340:207–214

7. Dellinger RP: Cardiovascular management ofseptic shock. Crit Care Med 2003; 31:946–955

8. Landry DW, Oliver JA: The pathogenesis ofvasodilatory shock. N Engl J Med 2001; 345:588–595

9. Rivers E, Nguyen B, Havstad S, et al: Earlygoal-directed therapy in the treatment of se-vere sepsis and septic shock. N Engl J Med2001; 345:1368–1377

10. Harbarth S, Garbino J, Pugin J, et al: In-appropriate initial antimicrobial therapyand its effect on survival in a clinical trialof immunomodulating therapy for severesepsis. Am J Med 2003; 115:529 –535

11. Garnacho-Montero J, Garcia-Garmendia JL,Barrero-Almodovar AF, et al: Impact of ade-quate empiric antibiotic therapy on the out-come of patients admitted to the intensivecare unit with sepsis. Crit Care Med 2003;31:2742–2751

12. Micek ST, Isakow W, Shannon W, et al: Predic-tors of hospital mortality for patients with se-vere sepsis treated with Drotrecogin alfa (acti-vated). Pharmacotherapy 2005; 25:26–34

13. Dellinger RP, Carlet JM, Masur H, et al: Sur-viving Sepsis Campaign guidelines for man-agement of severe sepsis and septic shock.Crit Care Med 2004; 32:858–873

14. Kollef MH: Inadequate antimicrobial treat-ment: An important determinant of outcomefor hospitalized patients. Clin Infect Dis2000; 31:S131–S138

15. Knaus WA, Draper EA, Wagner DP, et al:APACHE II: A severity of disease classificationsystem. Crit Care Med 1985; 13:818–829

16. Bone RC, Balk RA, Cerra FB, et al: ACCP/SCCM Consensus Conference: Definitions forsepsis and organ failure and guidelines forthe use of innovative therapies in sepsis.Chest 1992; 101:1644–1655

17. Bernard GR, Vincent JL, Laterre PF, et al:Efficacy and safety of recombinant humanactivated protein C for severe sepsis. N EnglJ Med 2001; 344:699–709

18. Ruokonen E, Pettila V: Surviving sepsis cam-paign: Outcome of severe sepsis can be im-proved by revising procedural standards.Acta Anaesthesiol Scand 2005; 49:597–598

19. Gao F, Melody T, Daniels D, et al: The impactof compliance with 6-hour and 24-hour sep-sis bundles on hospital mortality in patientswith severe sepsis: A prospective observa-tional study. Crit Care 2005; 9:R764–R770

20. Ibrahim EH, Sherman G, Ward S, et al: Theinfluence of inadequate antimicrobial treat-ment of bloodstream infections on patientoutcomes in the ICU setting. Chest 2000;118:146–155

21. Kollef MH, Sherman G, Ward S, et al: In-adequate antimicrobial treatment of infec-tions: A risk factor for hospital mortalityamong critically ill patients. Chest 1999;115:462– 474

22. Dhainaut JF, Laterre PF, LaRosa SP, et al:The clinical evaluation committee in a largemulticenter phase 3 trial of drotrecogin alfa(activated) in patients with severe sepsis(PROWESS): Role, methodology, and results.Crit Care Med 2003; 31:2291–2301

23. Kollef MH, Ward S: The influence of mini-BAL cultures on patient outcomes: Implica-tions for the antibiotic management of ven-tilator-associated pneumonia. Chest 1998;113:412–420

24. Rello J, Gallego M, Mariscal D, et al: Thevalue of routine microbial investigation inventilator-associated pneumonia. Am J Re-spir Crit Care Med 1997; 156:196–200

25. Iregui M, Ward S, Sherman G, et al: Clinicalimportance of delays in the initiation of appro-priate antibiotic treatment for ventilator-associated pneumonia. Chest 2002; 122:262–268

26. Morrell M, Fraser VJ, Kollef MH: Delayingthe empiric treatment of candida blood-stream infection until positive blood cultureresults are obtained: A potential risk factorfor hospital mortality. Antimicrob AgentsChemother 2005; 49:3640–3645

27. Ely EW, Baker AM, Dunagan DP, et al: Effecton the duration of mechanical ventilation ofidentifying patients capable of breathing spon-taneously. N Engl J Med 1996; 335:1864–1869

28. 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

29. Ibrahim EH, Ward S, Sherman G, et al Ex-perience with a clinical guideline for thetreatment of ventilator-associated pneumo-nia. Crit Care Med 2001; 29:1109–1115

30. Rhodes A, Bennett D: Early-goal directedtherapy: An evidence-based review. Crit CareMed 2004; 32:S448–S450

31. Shapiro N, Howell M, Talmor D: A blueprintfor a sepsis protocol. Acad Emerg Med 2005;12:352–359

2713Crit Care Med 2006 Vol. 34, No. 11