Efficacy of Therapy with Recombinant Human Activated Protein C of Critically Ill Surgical Patients with Infection Complicated by Septic Shock and Multiple Organ Dysfunction Syndrome

Abstract

Background:

Septic shock causing or complicating critical surgical illness results in high mortality. Drotrecogin alfa (activated), known also as recombinant human activated protein C (rhAPC) has become controversial as therapy, owing to persisting questions of efficacy and safety. We hypothesized rhAPC to be effective therapy for critically ill surgical patients with septic shock.

Methods:

Open-label therapy with rhAPC (by predefined criteria) of 108 critically ill surgical patients. Treated patients were matched individually in prospect for age, gender, Acute Physiology and Chronic Health Evaluation (APACHE)-II and –III scores, site of infection, and organism (0–2 points each, maximum 12 points) with 108 patients from our 15,000-patient surgical intensive care unit database who did not receive rhAPC. No match was accepted if <6 points. Multiple organ dysfunction (MOD) scores and data regarding cortisol concentrations, bleeding complications, and transfusion requirements were collected. The primary endpoint was 28-day mortality, with mortality for hospitalization and resolution of organ dysfunction as secondary endpoints. Statistical analyses included ANOVA, c statistic, binary logistic regression, and Kaplan-Meier time-to-event and Cox proportional hazards analyses; α=0.05.

Results:

The mean match score was 9.2±0.1 points (range, 6–12 points). Patients were well matched by all criteria, including baseline MOD score (9.5±0.7 vs. 9.8±0.3 points, p=0.66). Mean age was 68.1±1.1 years (p=0.49), Mean APACHE-III score was 99.6±1.5 points (p=0.87). Mean time to rhAPC administration was 25±3 h. Survival at 28 days after rhAPC was 71.3% vs. 49.1% (p=0.001); hospital survival was 57.4% vs. 40.7% (p=0.02). By logistic regression, rhAPC therapy resulted in improved 28-day survival (OR 2.57, 95% CI 1.46–4.52, p=0.001) (model χ2 11.244, p=0.001); and hospital survival (OR 1.96, 95% CI 1.14–3.36, p=0.015) (model χ2 6.03, p=0.014). The MOD score decreased significantly (p=0.012) during rhAPC therapy.

Conclusion:

Therapy with rhAPC appeared to improve survival in surgical ICU patients with life-threatening infection characterized by septic shock and organ dysfunction.

Surgical patients comprise a substantial proportion (25% to 40%) of patients with severe sepsis and septic shock [1,2]. Although necrotizing soft tissue infections are associated commonly with critical illness [3], surgical cases of intra-abdominal infection that result in critical illness are relatively unusual [4], but pose an increased risk of failed therapy and consequent development or exacerbation of multiple organ dysfunction syndrome (MODS), which may be lethal [5]. Severe sepsis may also be a consequence of a secondary nosocomial infection developing after an elective or emergency surgical procedure [6]; nosocomial infections lead to worse outcomes [6-8]. Factors that may increase the risk of clinical failure of intra-abdominal infection include increased severity of illness [9], initial inappropriate antibiotic therapy [10,11], peritonitis rather than abscess formation (failure of intra-peritoneal host defenses) [9], and inadequate surgical control of the primary source of infection (“source control”) [12,14]. Clinical failure increases the risks of re-operation, wound failure, fistula formation, MODS, and death [13].

The mortality of intra-abdominal infection complicated by critical illness is approximately 22–30% [14]. Although contemporary critical care and improved source control may be reducing mortality [14], there is always the potential for improvement. Whether improved antibiotic therapy can yield incremental benefit is unknown, because critically ill patients are usually excluded from clinical trials. Thus, the optimal pharmacotherapy of serious infections of surgical patients is still being defined.

Drotrecogin alfa (activated) (Xigris®, Eli Lilly and Co., Inc., Indianapolis, IN), or recombinant human activated protein C (rhAPC), was approved in 2001 for the treatment of severe sepsis in patients with a high risk of death (i.e., Acute Physiology and Chronic Health Evaluation (APACHE)-II score ≥25 points, dysfunction of ≥2 organs). In a phase III trial (PROWESS), the risk of 30-day, all-cause mortality with severe sepsis was reduced by 6.1% [1], but apparently only among patients with an enrollment Acute Physiology and Chronic Health Evaluation (APACHE)-II score ≥25 points. Moreover, among the 532 “surgical” patients enrolled in PROWESS, mortality decreased only by 3.1% (p=NS), raising the question of efficacy for surgical sepsis, a concern compounded by questions of safety because of a perceived higher risk of bleeding after surgery.

A blinded post-hoc adjudication of the PROWESS database assessed the adequacy of surgical source control among 474 confirmed surgical patients [15]. The 28-day, all-cause reduction of mortality in high-risk (APACHE-II ≥25 points) intra-abdominal surgery patients treated with rhAPC (absolute risk reduction 18.2%; relative risk of death [RR] 0.60, 95% confidence interval [CI] 0.36-1.00) was consistent with the overall PROWESS study sample. Increased risk of bleeding in surgical patients was not detected.

Hypothesizing that rhAPC is efficacious for therapy of critically ill patients with intra-abdominal infection, a pilot prospective, matched-control, inception-cohort study was conducted (35 patients with intra-abdominal infection and septic shock, 35 control patients selected from our ICU database who were matched prospectively based on six criteria) [16]. Treated patients were extremely ill (mean APACHE III score, 102±2 points) and had substantial MODS. Mortality at 30 days was 30% for patients treated with rhAPC compared with 70% for the control patients (odds ratio (OR) for survival 5.29, 95% CI 1.85–15.10, p<0.0001). Patients who survived rhAPC therapy showed objective resolution of organ dysfunction before the 96-h infusion period was completed.

This study was undertaken to test further the hypothesis that treatment with rhAPC reduces mortality among critically ill surgical patients with infection, septic shock, and MODS. Although patients with non-abdominal infections were eligible for inclusion, the enrollment and case-matching criteria were essentially similar to the pilot study.

Patients and Methods

This study was a prospective, open-label, single-blind, comparative trial of rhAPC administration to critically ill surgical patients with infection and septic shock, as defined by consensus [17]. Informed consent was waived by the Committee on Human Rights and Research of Weill Cornell Medical College, because all patients received approved, indicated therapy.

Hospital and study inclusion and exclusion criteria were established to mimic as closely as possible the inclusion and exclusion criteria from PROWESS [1] (Table 1), excepting the requirement of septic shock. Consecutive patients were enrolled if they required surgical critical care for a surgically treated or post-operative infection, had MODS ≥ 2 organs, and required vasopressor therapy. Absolute contraindications included a high risk of bleeding after surgery, active hemorrhage, and expected survival of less than 24 h (Table 1).

Table 1.

Institutional Contraindications for Therapy with Recombinant Human Activated Protein C (rhAPC)

Absolute

Active hemorrhage

Acute pancreatitis

Congential uncorrectable bleeding diathesis

Epidural catheter

Hypersensitivity to rhAPC

Onset of organ dysfunction >48 h prior to onset of therapy

Prior intracranial arteriovenous malformation, aneurysm, neoplasm, or cerebral herniation

Recent (3 months) hemorrhagic or ischemic stroke

Recent (2 months) neurosurgery or major traumatic brain injury

Therapeutic unfractionated heparin (within 8 h) or low molecular weight heparin (within 12 h)

Thrombocytopenia (<30,000/mm3) unresponsive to transfusion of platelets

Warfarin therapy with International Normalized Ratio (INR) >3.0

Severe chronic liver disease (Child-Pugh Class C)

Refusal of transfusion of blood/blood products

Refusal to receive life-sustaining therapy

Relative

Antiplatelet agent, aspirin (>650 mg/day), or thrombolytic therapy within prior 3 days

Antithrombin therapy (>10,000 U) within prior 12 h

Chronic renal failure

Conditions where bleeding constitutes a hazard or would be particularly difficult to manage

Gastrointestinal hemorrhage (that required therapy) within 6 weeks

Glycoprotein Iib/IIIa receptor antagonist therapy within prior 7 days

Major surgical procedure (i.e., body cavity) within 12 h

Minor surgical procedure (i.e., percutaneous catheter placement) within 2 h

Pregnancy or lactation

Protein C concentrate therapy within prior 24 h

Recent (within 3 months) venous thromboembolic disease

Thrombocytopenia (<30,000/mm3) responsive to transfusion of platelets

Warfarin therapy with International Normalized Ratio (INR) <3.0

Treated patients were compared with a cohort of patients who were treated previously in the surgical ICU, matched individually from a prospective database of more than 15,000 patients dating from January 1991, configured in prospect to de-identify patients and outcomes. After a patient began treatment with rhAPC, he or she was matched within 12 h by one of us (LJH), who had no clinical responsibilities and did not enter the ICU. Allocation and outcome of all matched patients were concealed until data lock.

Matching was performed with validated methodology [18], according to six criteria defined prospectively (Table 2): Age, gender, APACHE II and -III scores, infection-related diagnosis, and microbiology. Use of both APACHE II and -III scores ensured the closest possible match for severity of illness [19]. The closeness of the match for each parameter was awarded 0–2 points, thus a perfect match received 12 points. A patient was considered unmatched (and therefore excluded) if a minimum six-point match could not be achieved. All matching cases were selected from a time period as close to contemporaneous management as possible. When multiple candidate matches were identified for a treated case, the matching case treated most recently was chosen.

Table 2.

Matching Schema for Identification of Case-Matched Control Patients

Points awarded
Parameter	2 points	1 point	0 points
Age (years)	+2	+3-5	>5
APACHE II (points)	+1	+2-3	>3
APACHE III (points)	+2	+3-5	>5
Site of infection	Identical		No Match
Gender	Identical		No Match
Organism	Identical	Same gram stain	No Match

Patients were matched on six criteria; a perfect score was 12 points; whereas patients were considered unmatchable if a match of >6 points could not be found. Patients were matched by both APACHE II and APACHE III scores, so as to describe severity of illness with utmost precision (see text). “Same gram stain” refers to both patients having “gram-positive cocci,” “gram-negative bacilli,” or the like.

Demographic data collected included age, gender, admission diagnosis, site of perforation, APACHE II and –III scores, MOD scores (criteria of Marshall et al.; six organs, 0–4 point scale, maximum score 24 points) [5] on days 1–7, and the cumulative MOD score. In the cumulative score paradigm, points are awarded for worsening organ dysfunction, but not rescinded if organ function improved subsequently. Serial activated partial thromboplastin time (aPTT), serum glucose concentrations, and basal and stimulated (60 min after cosyntropin 1 mcg) cortisol concentrations were monitored. Transfusion of blood and blood products was recorded. Mortality was noted at 28 days and for the hospitalization.

The surgical ICU is “semi-closed,” with care coordinated with the surgical team but provided by surgical intensivists [20]. Sepsis care was provided according to Surviving Sepsis Campaign guidelines [21]. Empiric antibiotic therapy of complicated intra-abdominal infection and nosocomial pneumonia was administered according to guidelines of the Surgical Infection Society [22] and the American Thoracic Society [23], respectively, but administered by a “rotation” schedule (monthly rotation of cefepime, levofloxacin, imipenem-cilastatin, piperacillin-tazobactam; quarterly rotation of vancomycin and linezolid) [24]. Double-coverage (e.g., an aminoglycoside) of gram-negative sepsis was unusual. Empiric antifungal therapy was not administered unless the patient had a hospital-acquired perforated viscus (e.g., colon anastomotic dehiscence). Antibiotic therapy was de-escalated as appropriate based on the antibiotic susceptibility of isolated organisms.

Statistical analysis

Statistical analysis was performed using commercial software (SPSS 11.0 for Macintosh, SPSS, Inc., Chicago, IL). The study was designed to have 80% power to detect a 13% absolute decrease in mortality from therapy with drotrecogin alfa (activated) with 95% confidence. The primary endpoint was 28-day mortality. Secondary endpoints were mortality for the hospitalization and the resolution of organ dysfunction. Univariate analysis of coordinate variables was performed by chi-square contingency table analysis with the Fisher exact test, and univariate analysis of continuous variables was performed using the Mann-Whitney U-test. The Kolmogorov-Smirnov non-parameteric test was used to test differences in data that were not distributed normally. Changes in continuous variables with respect to time were analyzed by repeated measures analysis of variance (ANOVA) with the Tukey post-hoc test, and analyzed by the c statistic. Differences in mortality to 28 days were analyzed by Kaplan-Meier time-to-event analysis and compared by Cox proportional hazard ratio. Statistical significance was defined at α=0.05. Binary logistic regression was used to assess the independence of a single independent variable (therapy with rhAPC) upon mortality, considering that several “independent” variables of relevance to outcome were controlled for by the case-matching schema. The OR and 95% CI were determined, and the sensitivity, specificity, and goodness of fit (model χ2, Hosmer-Lemeshow test) were determined for the logistic regression model. Data are expressed as mean values±standard error.

Results

One hundred eight treated patients were compared with 108 individually matched control patients. Comparative demographics for the matched patients are listed in Table 3. The mean age of the 216 patients was 68.1±1.1 years, and the mean APACHE II and APACHE III scores were 28.6±0.4 and 99.6±1.5 points, respectively. The mean MOD score was 9.6±1.5 points at study entry. None of these values differed between groups. The results of the matching process are shown in Table 4 and Figure 1. All treated patients were matched without exception. The mean matching score was 9.2±0.1 points, with comparable degrees of matching for all six components. Ninety-five percent of matched patients received at least an eight-point match.

FIG. 1.

Frequency distribution of the matching scores.

Table 3.

Patient Demographics

Parameter	Matched control patients	rhAPC	p
Age, years	68.9±1.6	67.3±1.7	0.49
APACHE II	28.6±0.6	28.5±0.6	0.94
APACHE III	99.8±2.2	99.4±2.1	0.87
MOD Score (Day 1)	9.5±0.7	9.8±0.4	0.66

APACHE=Acute Physiology and Chronic Health Evaluation; MOD=multiple organ dysfunction.

Table 4.

Results of the Matching Schema

Parameter	Mean score	SEM	% 2-Point
Total Points	9.19	0.14	-----
Age	1.57	0.07	69.4
APACHE II	1.63	0.06	68.5
APACHE III	1.64	0.06	73.1
Site of infection	1.83	0.05	91.6
Gender	1.63	0.08	81.3
Organism	0.91	0.09	36.1

APACHE=Acute Physiology and Chronic Health Evaluation; MOD=multiple organ dysfunction; SEM=standard error of the mean.

Among patients operated on for source control of intra-abdominal infection (72/108, 66.6%), rhAPC was initiated 25±3 h after surgery (29/72, 40.2% within 24 h). The infusion was stopped for 10/108 patients (9.3%) to manage bleeding (n=4) or anemia (n=6) with transfusions of blood and blood products (Table 5). The infusion was resumed and completed for all 10 patients who were transfused, for 100% completion of the intended 96-h infusion. Patients treated with rhAPC were likely to develop an increased aPTT. The aPTT was prolonged throughout the infusion period compared with matched controls (Fig. 2). Mean aPTT increased from 38.7±1.2 seconds to 64.6±3.1 seconds by 12.8±1.3 h after the initiation of therapy, and increased above the upper limit of normal for the assay in 64.8% of patients. Fresh frozen plasma was administered if the aPTT increased above 60 seconds, regardless of the status of hemostasis. The incidence of FFP transfusion was 38.5% among rhAPC-treated patients, totaling 1.6±0.3 units. Mean basal and 60-min stimulated serum cortisol concentrations did not change in either the rhAPC–treated patients or matched controls (34.9±8.1 vs. 29.7±2.3 mcg/dL, p=0.494, zero vs. 60 min, matched cohort; 33.0±6.5 vs. 35.8±2.8 mcg/dL, p=0.776, zero vs. 60 min, rhAPC).

FIG. 2.

Mean activated partial thromboplastin times (in seconds), recombinant human activated protein C (rhAPC)-treated patients and matched control patients, during infusion.

Table 5.

Characteristics of Patients Who Required Transfusion during rhAPC Therapy

Overt Bleeding (n=4, 3.6%)-all transfused

GI hemorrhage

1 lower GI

1 upper GI

Endotracheal tube-exact site not determined

Subdural hematoma^*-craniotomy

Non-specific decrease of hematocrit to transfusion trigger (n=6, 4.8%)-all transfused

rhAPC therapy was contraindicated in this trauma patient with traumatic brain injury

rhAPC=recombinant human activated protein C.

Patients treated with rhAPC were significantly more likely to survive to the 28-day primary endpoint and the hospital discharge secondary endpoint. Survival at 28 days was 71.3% vs. 49.1% (p<0.001) for an absolute reduction of the risk of death of 22.2% and a relative risk reduction of 44.0%. Survival to hospital discharge was 57.7% vs. 40.7% (p=0.02), for an absolute reduction of the risk of death of 17.0% and a relative risk reduction of 29.0%. The Kaplan-Meier analysis of mortality is shown in Figure 3; the survival benefit of rhAPC therapy was significant out to more than 90 days. (p<0.01). By binary logistic regression, rhAPC-treated patients were more than twice as likely to survive to 28 days (OR 2.58, 95% CI 1.47-4.52), and nearly twice as likely to survive to hospital discharge (OR 1.96, 95% CI 1.14–3.36, with good model calibration and discrimination for both (Table 6). Among the 72 rhAPC-treated patients with complicated intra-abdominal infections, 71.6% were alive at 28 days vs. 48.5% of patients in the matched cohort (p=0.004). Among the 36 rhAPC-treated patients with all other types of infections (ventilator-associated pneumonia, most commonly), 70.6% were alive at 28 days vs. 50% of patients in the matched cohort (p=0.059). Binary logistic regression of 28-day mortality in the infection type subsets confirmed these observations, including the trend toward efficacy for “other” infections.

FIG. 3.

Therapy with recombinant human activated protein C (Xigris) increased survival significantly (p<0.01) at both 28 days and at hospital discharge, compared with untreated matched controls. The survival curves are also statistically distinct at 100 days post-therapy (p<0.05).

Table 6.

Binary Logistic Regression Analyses

Parameter	p	OR	95% CI
Dependent Variable: Alive at 28 days
rhAPC	0.001	2.58	1.47-4.52
Model chi-square 11.244, p<0.001
Goodness of fit p>0.05
ROC area 0.616
61.1% classified correctly
Dependent Variable: Alive at hospital discharge
rhAPC	0.002	1.96	1.14-3.36
Model chi-square 6.03, p<0.015
Goodness of fit p>0.05
ROC area 0.583
58.5% classified correctly
Infection Subsets
Dependent Variable: Alive at hospital discharge
Intra-abdominal infection
rhAPC	0.005	2.68	1.34-5.36
Model chi-square 7.98, p<0.005
Other infection
rhAPC	0.075	2.40	0.92-6.29
Model chi-square 3.28, p=0.07

OR=odds ratio; CI=confidence interval.

Although the initial magnitude of organ dysfunction was comparable between treated and matched patients (9.8±0.4 vs. 9.5±0.7 points, respectively, p=0.66), statistically significant improvement of organ function (p<0.05) was observed during rhAPC therapy before the 96-hour infusion was completed, compared with matched controls (Fig. 4). The improvement in MOD scores was a significant predictor of outcome as analyzed by the c statistic and displayed by receiver-operator characteristic curves (Fig. 5). Length of stay in the ICU (ULOS) and hospital (HLOS) were longer after rhAPC therapy despite comparable severity of illness. For ULOS, there was a trend toward longer duration (28±3 vs. 18±2 days, p=0.068), whereas HLOS was prolonged unequivocally (47±6 vs. 34±4 days, p=0.006), likely a direct reflection of increased survival.

FIG. 4.

Daily changes of multiple organ dysfunction score among recombinant human activated protein C (rhAPC)-treated patients stratified by hospital survival. The differences at days 2–4 and 4–7 are statistically significant.

FIG. 5.

Influence of change in daily multiple organ dysfunction (MOD) scores during recombinant human activated protein C (rhAPC) infusion on mortality from septic shock in surgical patients. The areas under the receiver-operator characteristic curves are 0.759 (day 1 to 2), 0.696 (day 2 to 4), and 0.737 (day 4 to 7) (all p<0.05).

Discussion

Compared with matched controls, critically ill surgical patients treated with rhAPC for septic shock had significantly improved survival with an associated resolution of organ dysfunction. The observed reduction of risk was greater than that observed by PROWESS [1], from analysis of the sickest 50% of patients from PROWESS (APACHE II ≥25 points), and from analysis of the cohort of surgical patients from PROWESS [17]. These results may have been due to the focus on high severity of illness, or to the emphasis on administration of drug within 24 h, which has been associated with improved outcomes [2]. Therapy of intra-abdominal infection was particularly effective.

Our cohort-matching strategy creating equivalent groups for comparison. Potential assignment bias was minimized by assigning the process to an individual who had no clinical responsibilities, before the outcomes of the rhAPC-treated patients were known. Using the 12-point scale, comparators were selected based on the closest match; a “perfect” (12-point) match was achieved for 13 patients, and a majority had a match of eight points or better. Examples of temporal bias include the reduced mortality that may have resulted from faster resolution of organ dysfunction, which may also be attributable to improved glycemic control; a protocol for glycemic control was followed during the rhAPC therapy period [25], but not for some of the matched controls. It is also possible that the observed effects are attributable to some imperceptible effect other than therapy with rhAPC, such as improved surgical source control, or changes in the way antimicrobial agents were administered. The latter is unlikely, given that antibiotics are administered by protocol and that initial antimicrobial therapy was appropriate in more than 90% of cases [24].

The open-label cohort-matching design was chosen because, at the time the study was designed, a randomized, placebo-controlled trial of approved therapy raised ethical questions regarding the withholding of therapy for the purpose of the study. However, controversy developed subsequently because of the publication of non-confirmatory trials [26], and equipoise supported the conduct of a randomized, double-blind, placebo-controlled trial of rhAPC for therapy of septic shock (PROWESS-SHOCK) [27 –30]. The patient population studied herein fell clearly within the population indicated in the product literature, but given that media reports indicate that PROWESS-SHOCK was a negative study and that rhAPC was withdrawn from the market worldwide on October 25, 2011 [31,32], questions of efficacy and safety have been rendered moot, and now will center on why PROWESS-SHOCK failed.

The limitations of the study are several. Foremost, this was not a randomized controlled trial, and some other aspects of critical care therapeutics did change during the study period, but the rigor of the cohort matching process should have mitigated that effect. An alternative to matching individual patients by specific criteria would have been to calculate a propensity score for therapy with rhAPC based on measured variables, and match based on the propensity score [33]. Propensity scores have been validated for comparisons of groups treated and not treated, but rhAPC was not available clinically during some of the time period when the matching patients were treated, hence there was no propensity whatsoever for their treatment with rhAPC, and therefore no treatment selection bias. Propensity score methods do reduce bias by 2–13% in the estimation of marginal ORs compared with logistic regression [34,35], but not to the degree that this study would have produced different results.

Footnotes

Author Disclosure Statement

Doctor Barie is a consultant to and has received honoraria from Eli Lilly and Co., Inc., and is a member of the Academic Steering Committee for the PROWESS-SHOCK trial. None of the other authors have a conflicting financial interest.

References

Bernard

, Vincent

, Laterre

et al. Efficacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med, 2001; 344:699–709.

Bernard

, Margolis

, Shanies

et al. Extended evaluation of recombinant human activated protein C United States Trial (ENHANCE US): A single-arm, phase 3B, multicenter study of drotrecogin alfa (activated) in severe sepsis. Chest, 2004; 125:2206–2216.

Barie

. Serious intra-abdominal infections. Curr Opin Crit Care, 2001; 7:263–267.

Barie

, Vogel

, Dellinger

et al. for the Cefepime Intra-abdominal Infection Study Group. A randomized double-blind clinical trial comparing cefepime plus metronidazole with imipenem-cilastatin in the treatment of complicated intraabdominal infection. Arch Surg, 1997; 132:1294–1302.

Barie

, Hydo

. Epidemiology of multiple organ dysfunction syndrome in critical surgical illness. Surg Infect, 2000; 1:173–186.

Merlino

, Yowler

, Malangoni

. Nosocomial infections adversely affect the outcomes of patients with serious intraabdominal infections. Surg Infect, 2004; 5:21–27.

Barie

. Importance, morbidity, and mortality of pneumonia in the surgical intensive care unit. Am J Surg, 2000; 179,2A Suppl:2S–7S.

Mulier

, Penninckx

, Verwaest

et al. Factors affecting mortality in generalized postoperative peritonitis: Multivariate analysis in 96 patients. World J Surg, 2003; 27:379–384.

Anaya

, Nathens

. Risk factors for severe sepsis in secondary peritonitis. Surg Infect, 2003; 4:355–362.

10.

Mosdell

, Morris

, Voltura

et al. Antibiotic treatment for surgical peritonitis. Ann Surg, 1991; 214:543–549.

11.

Montravers

, Gauzit

, Muller

et al. Emergence of antibiotic-resistant bacteria in cases of peritonitis after intraabdominal surgery affects the efficacy of empirical antimicrobial therapy. Clin Infect Dis, 1996; 23:486–494.

12.

Marshall

, Schein

. Source control for surgical infections. World J Surg, 2004; 28:638–645.

13.

Barie

, Eachempati

. Consequences of failed source control. Schein

, Marshall

. Source Control. Berlin: Springer-Verlag, 2003; 71–81.

14.

Barie

, Hydo

, Eachempati

. Longitudinal outcomes of intra-abdominal infection complicated by critical illness. Surg Infect, 2004; 5:365–373.

15.

Barie

, Lowry

, Williams

et al. Benefit/risk profile of therapy with drotrecogin alfa (activated) in surgical patients with severe sepsis. Am J Surg, 2004; 188:212–220.

16.

Barie

, Hydo

, Shou

, Eachempati

. Efficacy and safety of drotrecogin alfa (activated) for the therapy of surgical patients with severe sepsis. Surg Infect, 2006; 7,Suppl 2:S77–S80.

17.

Members of the American College of Chest Physicians and the Society of Critical Care Medicine Consensus Conference Committee. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Crit Care Med, 1992; 20:864–871.

18.

Pittet

, Tarara

, Wenzel

. Nosocomial bloodstream infection in critically ill patients: Excess length of stay, extra costs, and attributable mortality. JAMA, 1994; 271:1598–1601.

19.

Barie

, Hydo

, Fischer

. Comparison of APACHE II and III scoring systems for mortality prediction in critical surgical illness. Arch Surg, 1995; 130:77–82.

20.

Barie

, Bacchetta

, Eachempati

. The contemporary surgical intensive care unit: structure, staffing, and issues. Surg Clin North Am, 2000; 80:791–807.

21.

Dellinger

, Levy

, Carlet

et al. International Surviving Sepsis Campaign Guidelines Committee; American Association of Critical-Care Nurses; American College of Chest Physicians; American College of Emergency Physicians; Canadian Critical Care Society; European Society of Clinical Microbiology and Infectious Diseases; European Society of Intensive Care Medicine; European Respiratory Society; International Sepsis Forum; Japanese Association for Acute Medicine; Japanese Society of Intensive Care Medicine; Society of Critical Care Medicine; Society of Hospital Medicine; Surgical Infection Society; World Federation of Societies of Intensive and Critical Care Medicine. Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock: 2008. Crit Care Med, 2008; 36:296–327Erratum in: Crit Care Med 2008;36:1394–1396.

22.

Solomkin

, Mazuski

, Bradley

et al. Diagnosis and management of complicated intra-abdominal infection in adults and children: guidelines by the Surgical Infection Society and the Infectious Diseases Society of America. Surg Infect (Larchmt), 2010; 11:79–109.

23.

American Thoracic Society; Infectious Diseases Society of America. Guidelines for the management of adults with hospital-acquired, ventilator-associated, and healthcare-associated pneumonia. Am J Respir Crit Care Med, 2005; 15,171:388–416.

24.

Barie

, Hydo

, Shou

et al. Influence of antibiotic therapy on mortality of critical surgical illness caused or complicated by infection. Surg Infect, 2005; 6:41–54.

25.

Eachempati

, Hydo

, Shou

, Barie

. Implementation of tight glucose control for critically ill surgical patients: a process improvement analysis. Surg Infect (Larchmt), 2009; 10:523–531.

26.

Barie

. Current role of activated protein C therapy for severe sepsis and septic shock. Curr Infect Dis Rep, 2008; 10:368–376.

27.

Finfer

, Ranieri

, Thompson

et al. Design, conduct, analysis and reporting of a multi-national placebo-controlled trial of activated protein C for persistent septic shock. Intensive Care Med, 2008; 34:1935–194728. Barie PS. “All in” for a huge pot: The PROWESS-SHOCK trial for refractory septic shock. Surg Infect 2007;8:491–494.

28.

Thompson

, Ranieri

, Finfer

et al. Statistical analysis plan of PROWESS SHOCK study. PROWESS SHOCK Steering Committee. Intensive Care Med., 2010; 36:1972–1973.

29.

Ranieri

, Thompson

, Finfer

et al. Unblinding plan of PROWESS-SHOCK trial. PROWESS-SHOCK Academic Steering Committee. Intensive Care Med, 2011; 37:1384–1385.

30.

Barie

. The last Xigris^®* survivor. Surg Infect, 2011; 12:423–425.

31.

Lilly announces withdrawal of Xigris® following recent clinical trial results. https://investor.lilly.com/releasedetail2.cfm?ReleaseID=617602. 2011 November 17.

32.

Austin

. The performance of different propensity-score methods for estimating relative risks. J Clin Epidemiol, 2008; 61:537–545.

33.

Martins

, Pestman

, de Boer

et al. Systematic differences in treatment effect estimates between propensity score methods and logistic regression. Int J Epidemiol, 2008; 37:1142–1147.

34.

Austin

. The performance of different propensity score methods for estimating marginal odds ratios. Statist Med, 2007; 26:3078–3094.