Gilles Clermont | University of Pittsburgh (original) (raw)
Papers by Gilles Clermont
Critical Care Nephrology, 1998
Proceedings of the 2nd ACM SIGHIT symposium on International health informatics - IHI '12, 2012
We present a novel method for obtaining a concise and mathematically grounded description of mult... more We present a novel method for obtaining a concise and mathematically grounded description of multivariate differences between a pair of clinical datasets. Often data collected under similar circumstances reflect fundamentally different patterns. For example, information about patients undergoing similar treatments in different intensive care units (ICUs), or within the same ICU during different periods, may show systematically different outcomes. In
Classification, Clustering, and Data Mining Applications, 2004
I Mathematical and Statistical Modeling of Acute Inflammation Gilles Clermont1, Carson C. Chow1, ... more I Mathematical and Statistical Modeling of Acute Inflammation Gilles Clermont1, Carson C. Chow1, Gregory M. Constantine1, Yoram Vodovotz1, and John Bartels2 1 University of Pittsburgh, USA clermontg0ccm. upmc. edu, ccchowQpitt. edu, gmcfleuler. math. pitt. edu, ...
Intensive care medicine, 2013
Intensive care medicine, 2013
Mathematical biosciences, 2015
Biological systems present particular challengers to model for the purposes of formulating predic... more Biological systems present particular challengers to model for the purposes of formulating predictions of generating biological insight. These systems are typically multi-scale, complex, and empirical observations are often sparse and subject to variability and uncertainty. This manuscript will review some of these specific challenges and introduce current methods used by modelers to construct meaningful solutions, in the context of preserving biological relevance. Opportunities to expand these methods are also discussed.
Intensive care medicine, 2013
AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium, 2010
We develop and evaluate a data-driven approach for detecting unusual (anomalous) patient-manageme... more We develop and evaluate a data-driven approach for detecting unusual (anomalous) patient-management actions using past patient cases stored in an electronic health record (EHR) system. Our hypothesis is that patient-management actions that are unusual with respect to past patients may be due to a potential error and that it is worthwhile to raise an alert if such a condition is encountered. We evaluate this hypothesis using data obtained from the electronic health records of 4,486 post-cardiac surgical patients. We base the evaluation on the opinions of a panel of experts. The results support that anomaly-based alerting can have reasonably low false alert rates and that stronger anomalies are correlated with higher alert rates.
AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium, 2010
Developing methods to detect deviations from usual medical care may be useful in the development ... more Developing methods to detect deviations from usual medical care may be useful in the development of automated clinical alerting systems to alert clinicians to treatment choices that warrant additional consideration. We developed a method for identifying deviations in medication administration in the intensive care unit that is based on learning logistic regression models from past patient data that when applied to current patient data identifies statistically unusual treatment decisions. The models predicted a total of 53 deviations for 6 medications on a set of 3000 patient cases. A set of 12 predicted deviations and 12 non-deviations was evaluated by a group of intensive care physicians. Overall, the predicted deviations were assessed to often warrant an alert and to be clinically useful, and furthermore, the frequency with which such alerts would be raised is not likely to be disruptive in a clinical setting.
AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium, 2007
Detecting deviations from usual medical care is valuable in identifying potentially concerning pa... more Detecting deviations from usual medical care is valuable in identifying potentially concerning patient management events, both in real time and retrospectively. We describe a statistical method for identification of deviations in medication administration. The preliminary results reported here characterize the statistical properties of the identified deviations. Future research will investigate which deviations are clinically useful.
Critical care medicine, 2004
To determine the feasibility and potential usefulness of mathematical models in evaluating immuno... more To determine the feasibility and potential usefulness of mathematical models in evaluating immunomodulatory strategies in clinical trials of severe sepsis. Mathematical modeling of immunomodulation in simulated patients. Computer laboratory. We introduce and evaluate the concept of conducting a randomized clinical trial in silico based on simulated patients generated from a mechanistic mathematical model of bacterial infection, the acute inflammatory response, global tissue dysfunction, and a therapeutic intervention. Trial populations are constructed to reflect heterogeneity in bacterial load and virulence as well as propensity to mount and modulate an inflammatory response. We constructed a cohort of 1,000 trial patients submitted to therapy with one of three different doses of a neutralizing antibody directed against tumor necrosis factor (anti-TNF) for 6, 24, or 48 hrs. We present cytokine profiles over time and expected outcome for each cohort. We identify subgroups with high p...
Critical care medicine, 2003
Clinical trials of therapies for sepsis have been mostly unsuccessful in impacting mortality. Thi... more Clinical trials of therapies for sepsis have been mostly unsuccessful in impacting mortality. This may be partly due to the use of insensitive mortality end points. We explored whether modeling survival was more sensitive than traditional end points in detecting mortality differences in cohorts of patients with sepsis. Patients were stratified into seven a priori defined paired subgroups that reflected high and low mortality risk according to known clinical risk factors. We fitted an exponential survival model to the high- and low-risk cohort of each subgroup, providing estimates of the rate of dying, long-term survival, and excess day 1 mortality. Mortality in the high- and low-risk cohorts in each subgroup was compared using model parameters, fixed-point mortality, and Kaplan-Meier survival analysis. Eight intensive care units within a university teaching institution. One hundred thirty patients with severe sepsis or suspected Gram-negative bacteremia. None. Overall mortality of t...
American journal of respiratory and critical care medicine, 2002
Despite careful evaluation of changes in hospital care for community-acquired pneumonia (CAP), li... more Despite careful evaluation of changes in hospital care for community-acquired pneumonia (CAP), little is known about intensive care unit (ICU) use in the treatment of this disease. There are criteria that define CAP as "severe," but evaluation of their predictive value is limited. We compared characteristics, course, and outcome of inpatients who did (n = 170) and did not (n = 1,169) receive ICU care in the Pneumonia Patient Outcomes Research Team prospective cohort. We also assessed the predictive characteristics of four prediction rules (the original and revised American Thoracic Society criteria, the British Thoracic Society criteria, and the Pneumonia Severity Index [PSI]) for ICU admission, mechanical ventilation, medical complications, and death (as proxies for severe CAP). ICU patients were more likely to be admitted from home and had more comorbid conditions. Reasons for ICU admission included respiratory failure (57%), hemodynamic monitoring (32%), and shock (16%)...
Critical care (London, England), 2006
The lack of a standard definition for acute kidney injury has resulted in a large variation in th... more The lack of a standard definition for acute kidney injury has resulted in a large variation in the reported incidence and associated mortality. RIFLE, a newly developed international consensus classification for acute kidney injury, defines three grades of severity--risk (class R), injury (class I) and failure (class F)--but has not yet been evaluated in a clinical series. We performed a retrospective cohort study, in seven intensive care units in a single tertiary care academic center, on 5,383 patients admitted during a one year period (1 July 2000-30 June 2001). Acute kidney injury occurred in 67% of intensive care unit admissions, with maximum RIFLE class R, class I and class F in 12%, 27% and 28%, respectively. Of the 1,510 patients (28%) that reached a level of risk, 840 (56%) progressed. Patients with maximum RIFLE class R, class I and class F had hospital mortality rates of 8.8%, 11.4% and 26.3%, respectively, compared with 5.5% for patients without acute kidney injury. Addi...
Molecular medicine (Cambridge, Mass.)
The inflammatory phenotype of genetically modified mice is complex, and the role of Gram-negative... more The inflammatory phenotype of genetically modified mice is complex, and the role of Gram-negative lipopolysaccharide (LPS) in acute inflammation induced by surgical cannulation trauma, alone or in combination with hemorrhage and resuscitation ("hemorrhagic shock"), is both complex and controversial. We sought to determine if a mathematical model of acute inflammation could elucidate both the phenotype of CD14-deficient (CD14(-/-)) mice--following LPS, cannulation, or hemorrhagic shock--and the role of LPS in trauma/hemorrhage-induced inflammation. A mathematical model of inflammation initially calibrated in wild-type (C57Bl/6) mice subjected to LPS, cannulation, and hemorrhagic shock was recalibrated in CD14(-/-) mice subjected to the same insults, yielding an ensemble of models that suggested specific differences at the cellular and molecular levels (for example, 43-fold lower activation of leukocytes by LPS). The CD14(-/-)-specific model ensemble could account for comple...
Current opinion in critical care, 2004
Trauma and infection elicit an acute inflammatory response. In certain circumstances the degree o... more Trauma and infection elicit an acute inflammatory response. In certain circumstances the degree of the acute inflammatory response may result in pathologic manifestations, namely, sepsis and multiple organ failure. Despite an extensive series of clinical trials designed to modulate inflammation in sepsis, only one compound, activated protein C, has emerged from more than 250 failed trials. There is a growing recognition that the complexity of the acute inflammatory response precludes the efficient development of therapies for sepsis and multiple organ failure until systems approaches are brought to bear on this problem. Work carried out by the authors' groups suggests that mathematical modeling can provide a means by which in vitro and in vivo data can be synthesized into system-level analytic models of the acute inflammatory response. The authors have focused on agent-based modeling and modeling with ordinary differential equations. Some of the advantages and disadvantages of t...
Archives of internal medicine, Jan 10, 2003
Hospital mortality of patients admitted with community-acquired pneumonia (CAP) has been well des... more Hospital mortality of patients admitted with community-acquired pneumonia (CAP) has been well described. However, the long-term survival of those discharged alive is less clear. We sought to determine long-term survival of patients hospitalized with CAP and compare the outcome with controls hospitalized for reasons other than CAP. We performed a matched case-control analysis using the Medicare hospital discharge database from the first quarter of 1997. We compared all Medicare recipients 65 years or older hospitalized with CAP and controls matched for age, sex, and race hospitalized for reasons other than CAP. We measured 1-year mortality determined from the Medicare Beneficiary Entitlement file and the Social Security Administration. We identified 158 960 CAP patients and 794 333 hospitalized controls. Hospital mortality rates for the CAP cohort and hospitalized controls were 11.0% and 5.5%, respectively (P<.001). One-year mortality rates for the CAP cohort and hospitalized cont...
Critical Care, 2014
Prior work suggests that leukocyte trafficking is determined by local chemokine gradients between... more Prior work suggests that leukocyte trafficking is determined by local chemokine gradients between the nidus of infection and the plasma. We recently demonstrated that therapeutic apheresis can alter immune mediator concentrations in the plasma, protect against organ injury, and improve survival. Here we aimed to determine whether the removal of chemokines from the plasma by apheresis in experimental peritonitis changes chemokine gradients and subsequently enhances leukocyte localization into the infected compartment, and away from healthy tissues. In total, 76 male adult Sprague-Dawley rats weighing 400 g to 600 g were included in this study. Eighteen hours after inducing sepsis by cecal ligation and puncture, we randomized these rats to apheresis or sham treatment for 4 hours. Cytokines, chemokines, and leukocyte counts from blood, peritoneal cavity, and lung were measured. In a separate experiment, we labeled neutrophils from septic donor animals and injected them into either apheresis or sham-treated animals. All numeric data with normal distributions were compared with one-way analysis of variance, and numeric data not normally distributed were compared with the Mann-Whitney U test. Apheresis significantly removed plasma cytokines and chemokines, increased peritoneal fluid-to-blood chemokine (C-X-C motif ligand 1, ligand 2, and C-C motif ligand 2) ratios, and decreased bronchoalveolar lavage fluid-to-blood chemokine ratios, resulting in enhanced leukocyte recruitment into the peritoneal cavity and improved bacterial clearance, but decreased recruitment into the lung. Apheresis also reduced myeloperoxidase activity and histologic injury in the lung, liver, and kidney. These Labeled donor neutrophils exhibited decreased localization in the lung when infused into apheresis-treated animals. Our results support the concept of chemokine gradient control of leukocyte trafficking and demonstrate the efficacy of apheresis to target this mechanism and reduce leukocyte infiltration into the lung.
Critical care medicine, 2004
To compare differences in hospital mortality and resource use in adult severe sepsis subjects ran... more To compare differences in hospital mortality and resource use in adult severe sepsis subjects randomized to receive drotrecogin alfa (activated) (DrotAA) or placebo in the PROWESS trial. Retrospective, cross-sectional, blinded follow-up of subjects enrolled in a previous randomized, controlled trial. One hundred sixty-four tertiary care institutions in 11 countries. The 1,690 subjects with severe sepsis enrolled and treated with study drug in PROWESS, of whom 1,220 were alive at 28 days (the end of the original PROWESS follow-up). DrotAA (n = 850), 24 microg/kg/hr for 96 hrs, or placebo (n = 840). New follow-up data through hospital discharge were merged with existing 28-day follow-up data. Hospital mortality was calculated for designated subgroups. Intensive care unit and hospital length of stay and Simplified Therapeutic Intervention Scoring System-28 (TISS-28) scores were calculated overall and in designated subgroups. Hospital discharge location was recorded. The 95% confidence ...
Critical Care Nephrology, 1998
Proceedings of the 2nd ACM SIGHIT symposium on International health informatics - IHI '12, 2012
We present a novel method for obtaining a concise and mathematically grounded description of mult... more We present a novel method for obtaining a concise and mathematically grounded description of multivariate differences between a pair of clinical datasets. Often data collected under similar circumstances reflect fundamentally different patterns. For example, information about patients undergoing similar treatments in different intensive care units (ICUs), or within the same ICU during different periods, may show systematically different outcomes. In
Classification, Clustering, and Data Mining Applications, 2004
I Mathematical and Statistical Modeling of Acute Inflammation Gilles Clermont1, Carson C. Chow1, ... more I Mathematical and Statistical Modeling of Acute Inflammation Gilles Clermont1, Carson C. Chow1, Gregory M. Constantine1, Yoram Vodovotz1, and John Bartels2 1 University of Pittsburgh, USA clermontg0ccm. upmc. edu, ccchowQpitt. edu, gmcfleuler. math. pitt. edu, ...
Intensive care medicine, 2013
Intensive care medicine, 2013
Mathematical biosciences, 2015
Biological systems present particular challengers to model for the purposes of formulating predic... more Biological systems present particular challengers to model for the purposes of formulating predictions of generating biological insight. These systems are typically multi-scale, complex, and empirical observations are often sparse and subject to variability and uncertainty. This manuscript will review some of these specific challenges and introduce current methods used by modelers to construct meaningful solutions, in the context of preserving biological relevance. Opportunities to expand these methods are also discussed.
Intensive care medicine, 2013
AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium, 2010
We develop and evaluate a data-driven approach for detecting unusual (anomalous) patient-manageme... more We develop and evaluate a data-driven approach for detecting unusual (anomalous) patient-management actions using past patient cases stored in an electronic health record (EHR) system. Our hypothesis is that patient-management actions that are unusual with respect to past patients may be due to a potential error and that it is worthwhile to raise an alert if such a condition is encountered. We evaluate this hypothesis using data obtained from the electronic health records of 4,486 post-cardiac surgical patients. We base the evaluation on the opinions of a panel of experts. The results support that anomaly-based alerting can have reasonably low false alert rates and that stronger anomalies are correlated with higher alert rates.
AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium, 2010
Developing methods to detect deviations from usual medical care may be useful in the development ... more Developing methods to detect deviations from usual medical care may be useful in the development of automated clinical alerting systems to alert clinicians to treatment choices that warrant additional consideration. We developed a method for identifying deviations in medication administration in the intensive care unit that is based on learning logistic regression models from past patient data that when applied to current patient data identifies statistically unusual treatment decisions. The models predicted a total of 53 deviations for 6 medications on a set of 3000 patient cases. A set of 12 predicted deviations and 12 non-deviations was evaluated by a group of intensive care physicians. Overall, the predicted deviations were assessed to often warrant an alert and to be clinically useful, and furthermore, the frequency with which such alerts would be raised is not likely to be disruptive in a clinical setting.
AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium, 2007
Detecting deviations from usual medical care is valuable in identifying potentially concerning pa... more Detecting deviations from usual medical care is valuable in identifying potentially concerning patient management events, both in real time and retrospectively. We describe a statistical method for identification of deviations in medication administration. The preliminary results reported here characterize the statistical properties of the identified deviations. Future research will investigate which deviations are clinically useful.
Critical care medicine, 2004
To determine the feasibility and potential usefulness of mathematical models in evaluating immuno... more To determine the feasibility and potential usefulness of mathematical models in evaluating immunomodulatory strategies in clinical trials of severe sepsis. Mathematical modeling of immunomodulation in simulated patients. Computer laboratory. We introduce and evaluate the concept of conducting a randomized clinical trial in silico based on simulated patients generated from a mechanistic mathematical model of bacterial infection, the acute inflammatory response, global tissue dysfunction, and a therapeutic intervention. Trial populations are constructed to reflect heterogeneity in bacterial load and virulence as well as propensity to mount and modulate an inflammatory response. We constructed a cohort of 1,000 trial patients submitted to therapy with one of three different doses of a neutralizing antibody directed against tumor necrosis factor (anti-TNF) for 6, 24, or 48 hrs. We present cytokine profiles over time and expected outcome for each cohort. We identify subgroups with high p...
Critical care medicine, 2003
Clinical trials of therapies for sepsis have been mostly unsuccessful in impacting mortality. Thi... more Clinical trials of therapies for sepsis have been mostly unsuccessful in impacting mortality. This may be partly due to the use of insensitive mortality end points. We explored whether modeling survival was more sensitive than traditional end points in detecting mortality differences in cohorts of patients with sepsis. Patients were stratified into seven a priori defined paired subgroups that reflected high and low mortality risk according to known clinical risk factors. We fitted an exponential survival model to the high- and low-risk cohort of each subgroup, providing estimates of the rate of dying, long-term survival, and excess day 1 mortality. Mortality in the high- and low-risk cohorts in each subgroup was compared using model parameters, fixed-point mortality, and Kaplan-Meier survival analysis. Eight intensive care units within a university teaching institution. One hundred thirty patients with severe sepsis or suspected Gram-negative bacteremia. None. Overall mortality of t...
American journal of respiratory and critical care medicine, 2002
Despite careful evaluation of changes in hospital care for community-acquired pneumonia (CAP), li... more Despite careful evaluation of changes in hospital care for community-acquired pneumonia (CAP), little is known about intensive care unit (ICU) use in the treatment of this disease. There are criteria that define CAP as "severe," but evaluation of their predictive value is limited. We compared characteristics, course, and outcome of inpatients who did (n = 170) and did not (n = 1,169) receive ICU care in the Pneumonia Patient Outcomes Research Team prospective cohort. We also assessed the predictive characteristics of four prediction rules (the original and revised American Thoracic Society criteria, the British Thoracic Society criteria, and the Pneumonia Severity Index [PSI]) for ICU admission, mechanical ventilation, medical complications, and death (as proxies for severe CAP). ICU patients were more likely to be admitted from home and had more comorbid conditions. Reasons for ICU admission included respiratory failure (57%), hemodynamic monitoring (32%), and shock (16%)...
Critical care (London, England), 2006
The lack of a standard definition for acute kidney injury has resulted in a large variation in th... more The lack of a standard definition for acute kidney injury has resulted in a large variation in the reported incidence and associated mortality. RIFLE, a newly developed international consensus classification for acute kidney injury, defines three grades of severity--risk (class R), injury (class I) and failure (class F)--but has not yet been evaluated in a clinical series. We performed a retrospective cohort study, in seven intensive care units in a single tertiary care academic center, on 5,383 patients admitted during a one year period (1 July 2000-30 June 2001). Acute kidney injury occurred in 67% of intensive care unit admissions, with maximum RIFLE class R, class I and class F in 12%, 27% and 28%, respectively. Of the 1,510 patients (28%) that reached a level of risk, 840 (56%) progressed. Patients with maximum RIFLE class R, class I and class F had hospital mortality rates of 8.8%, 11.4% and 26.3%, respectively, compared with 5.5% for patients without acute kidney injury. Addi...
Molecular medicine (Cambridge, Mass.)
The inflammatory phenotype of genetically modified mice is complex, and the role of Gram-negative... more The inflammatory phenotype of genetically modified mice is complex, and the role of Gram-negative lipopolysaccharide (LPS) in acute inflammation induced by surgical cannulation trauma, alone or in combination with hemorrhage and resuscitation ("hemorrhagic shock"), is both complex and controversial. We sought to determine if a mathematical model of acute inflammation could elucidate both the phenotype of CD14-deficient (CD14(-/-)) mice--following LPS, cannulation, or hemorrhagic shock--and the role of LPS in trauma/hemorrhage-induced inflammation. A mathematical model of inflammation initially calibrated in wild-type (C57Bl/6) mice subjected to LPS, cannulation, and hemorrhagic shock was recalibrated in CD14(-/-) mice subjected to the same insults, yielding an ensemble of models that suggested specific differences at the cellular and molecular levels (for example, 43-fold lower activation of leukocytes by LPS). The CD14(-/-)-specific model ensemble could account for comple...
Current opinion in critical care, 2004
Trauma and infection elicit an acute inflammatory response. In certain circumstances the degree o... more Trauma and infection elicit an acute inflammatory response. In certain circumstances the degree of the acute inflammatory response may result in pathologic manifestations, namely, sepsis and multiple organ failure. Despite an extensive series of clinical trials designed to modulate inflammation in sepsis, only one compound, activated protein C, has emerged from more than 250 failed trials. There is a growing recognition that the complexity of the acute inflammatory response precludes the efficient development of therapies for sepsis and multiple organ failure until systems approaches are brought to bear on this problem. Work carried out by the authors' groups suggests that mathematical modeling can provide a means by which in vitro and in vivo data can be synthesized into system-level analytic models of the acute inflammatory response. The authors have focused on agent-based modeling and modeling with ordinary differential equations. Some of the advantages and disadvantages of t...
Archives of internal medicine, Jan 10, 2003
Hospital mortality of patients admitted with community-acquired pneumonia (CAP) has been well des... more Hospital mortality of patients admitted with community-acquired pneumonia (CAP) has been well described. However, the long-term survival of those discharged alive is less clear. We sought to determine long-term survival of patients hospitalized with CAP and compare the outcome with controls hospitalized for reasons other than CAP. We performed a matched case-control analysis using the Medicare hospital discharge database from the first quarter of 1997. We compared all Medicare recipients 65 years or older hospitalized with CAP and controls matched for age, sex, and race hospitalized for reasons other than CAP. We measured 1-year mortality determined from the Medicare Beneficiary Entitlement file and the Social Security Administration. We identified 158 960 CAP patients and 794 333 hospitalized controls. Hospital mortality rates for the CAP cohort and hospitalized controls were 11.0% and 5.5%, respectively (P<.001). One-year mortality rates for the CAP cohort and hospitalized cont...
Critical Care, 2014
Prior work suggests that leukocyte trafficking is determined by local chemokine gradients between... more Prior work suggests that leukocyte trafficking is determined by local chemokine gradients between the nidus of infection and the plasma. We recently demonstrated that therapeutic apheresis can alter immune mediator concentrations in the plasma, protect against organ injury, and improve survival. Here we aimed to determine whether the removal of chemokines from the plasma by apheresis in experimental peritonitis changes chemokine gradients and subsequently enhances leukocyte localization into the infected compartment, and away from healthy tissues. In total, 76 male adult Sprague-Dawley rats weighing 400 g to 600 g were included in this study. Eighteen hours after inducing sepsis by cecal ligation and puncture, we randomized these rats to apheresis or sham treatment for 4 hours. Cytokines, chemokines, and leukocyte counts from blood, peritoneal cavity, and lung were measured. In a separate experiment, we labeled neutrophils from septic donor animals and injected them into either apheresis or sham-treated animals. All numeric data with normal distributions were compared with one-way analysis of variance, and numeric data not normally distributed were compared with the Mann-Whitney U test. Apheresis significantly removed plasma cytokines and chemokines, increased peritoneal fluid-to-blood chemokine (C-X-C motif ligand 1, ligand 2, and C-C motif ligand 2) ratios, and decreased bronchoalveolar lavage fluid-to-blood chemokine ratios, resulting in enhanced leukocyte recruitment into the peritoneal cavity and improved bacterial clearance, but decreased recruitment into the lung. Apheresis also reduced myeloperoxidase activity and histologic injury in the lung, liver, and kidney. These Labeled donor neutrophils exhibited decreased localization in the lung when infused into apheresis-treated animals. Our results support the concept of chemokine gradient control of leukocyte trafficking and demonstrate the efficacy of apheresis to target this mechanism and reduce leukocyte infiltration into the lung.
Critical care medicine, 2004
To compare differences in hospital mortality and resource use in adult severe sepsis subjects ran... more To compare differences in hospital mortality and resource use in adult severe sepsis subjects randomized to receive drotrecogin alfa (activated) (DrotAA) or placebo in the PROWESS trial. Retrospective, cross-sectional, blinded follow-up of subjects enrolled in a previous randomized, controlled trial. One hundred sixty-four tertiary care institutions in 11 countries. The 1,690 subjects with severe sepsis enrolled and treated with study drug in PROWESS, of whom 1,220 were alive at 28 days (the end of the original PROWESS follow-up). DrotAA (n = 850), 24 microg/kg/hr for 96 hrs, or placebo (n = 840). New follow-up data through hospital discharge were merged with existing 28-day follow-up data. Hospital mortality was calculated for designated subgroups. Intensive care unit and hospital length of stay and Simplified Therapeutic Intervention Scoring System-28 (TISS-28) scores were calculated overall and in designated subgroups. Hospital discharge location was recorded. The 95% confidence ...