Influence of Gender on the Outcome of Severe Sepsis *

DOI 10.1378/chest.07-0420 2007;132;1786-1793; Prepublished online September 21, 2007;Chest

 Goldgran-Toledano, Maïté Garrouste-Orgeas and Jean François TimsitDarques, Carole Schwebel, Didier Nakache, Samir Jamali, Dany Christophe Adrie, Elie Azoulay, Adrien Francais, Christophe Clec'h, Loic 

: A Reappraisal*Severe SepsisInfluence of Gender on the Outcome of

  http://chestjournal.chestpubs.org/content/132/6/1786.full.html

services can be found online on the World Wide Web at: The online version of this article, along with updated information and 

ISSN:0012-3692)http://chestjournal.chestpubs.org/site/misc/reprints.xhtml(

written permission of the copyright holder.this article or PDF may be reproduced or distributed without the priorDundee Road, Northbrook, IL 60062. All rights reserved. No part of Copyright2007by the American College of Chest Physicians, 3300Physicians. It has been published monthly since 1935.

is the official journal of the American College of ChestChest

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from

Influence of Gender on the Outcome ofSevere Sepsis*A Reappraisal

Christophe Adrie, MD, PhD; Elie Azoulay, MD, PhD; Adrien Francais, PhD;Christophe Clec’h, MD; Loic Darques, MD; Carole Schwebel, MD;Didier Nakache, PhD; Samir Jamali, MD; Dany Goldgran-Toledano, MD;Maıte Garrouste-Orgeas, MD; and Jean Francois Timsit, MD, PhD; for theOutcomeRea Study Group†

Background: The influence of gender on survival of patients with severe sepsis is unclear. Earlierstudies suggested better survival in women, possibly related to the sex-steroid profile.Methods: To investigate whether mortality from severe sepsis was higher in men than in womenand whether the difference varied with menopausal status, we studied 1,692 patients with severesepsis included in the OutcomeRea database over an 8-year period. We conducted a nestedcase-control study, accurately matching men and women on three criteria: a death propensityscore, age, and center. Subgroup analyses were performed on individuals < 50 years old (men vspremenopausal women) and > 50 years old (men vs postmenopausal women).Results: We matched 1,000 men to 608 women with severe sepsis before and after adjustment forconfounding factors (ie, chronic respiratory failure; metastatic cancer; immunocompromisedstatus; emergency surgery, acute respiratory failure, and shock at admission; urinary tractinfection; and type of microorganism). Overall hospital mortality was significantly lower in women(adjusted odds ratio [OR], 0.75; 95% confidence interval [CI], 0.57 to 0.97; p � 0.02). In the group> 50 years old (481 women, 778 men), hospital mortality was significantly lower in women (OR,0.69; 95% CI, 0.52 to 0.93; p � 0.014). Hospital mortality was not significantly different betweenmen and women in the younger group (127 women, 222 men) [OR, 1.01; 95% CI, 0.52 to 1.97;p � 0.98]. Level of care, as assessed using the nine equivalents of nursing manpower use score,was identical in men and women.Conclusions: Among individuals > 50 years old with severe sepsis, women have a lower risk ofhospital mortality than men. (CHEST 2007; 132:1786–1793)

Key words: critical care; gender; outcome; sepsis; severe sepsis; therapeutic interventions

Abbreviations: CI � confidence interval; DNR � do not resuscitate; LOD � logistic organ dysfunction; NEMS � nineequivalents of nursing manpower use score; OR � odds ratio; SAPS � simplified acute physiology score

S evere sepsis remains a leading cause of death inindustrialized countries, and the number of deaths

caused by sepsis is increasing despite improved survivalrates.1,2 Mortality ranges from 20 to 50% in patientswith severe sepsis. The incidence of sepsis is loweramong women in the US general population for allinfection sources except the genitourinary tract.1,2

Men were more likely to have sepsis than women ineach year of the 22-year period from 1979 through2000, with a mean annual relative risk of 1.28.1 Theinfluence of gender on mortality in patients with

established sepsis is less clear. The greater immunesystem activity in women than in men is consistentwith better survival in women with severe sepsis. Sexhormones3 or sex-related gene polymorphisms4,5

For editorial comment see page 1725

may protect women against sepsis and death fromsepsis. Differences in hormone profiles have beenwidely suggested as the cause of gender-based dif-ferences in the incidence and outcome of sepsis. In

CHEST Original ResearchCRITICAL CARE MEDICINE

1786 Original Research

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from

mice, proestrus females tolerated polymicrobial sep-sis better than males,6 and survival improved inmales after testosterone receptor blockade.7 How-ever, epidemiologic studies3,8–14 produced conflict-ing results, perhaps reflecting effects of age, case-mix differences, nature of the injury preceding sepsisdevelopment (eg, trauma or burns), infection source,comorbidities, and menopausal status. Another pos-sible source of gender-based differences may be thereported lower use of invasive procedures in criti-cally ill women compared to men, despite greaterseverity of illness in women and even after adjust-ment on age.15

The objective of our study was to clarify the influ-ence of gender on survival of patients admitted to theICU for severe community-acquired sepsis. We stud-ied patients in a vast prospective database, and we useda propensity score to control for potential confounders.We then conducted a nested case-control study toinvestigate the hypothesis that men are at greater risk ofdeath than premenopausal women.

Materials and Methods

Study Design and Data Source

We conducted a prospective observational study in a multi-center database (OutcomeRea; Rosny-sous-Bois, France) fromJanuary 1997 to September 2005. The database, fed by 12 FrenchICUs, contains data on daily disease severity, iatrogenic events,and nosocomial infections. A random sample of at least 50patients � 16 years old and having ICU stays � 24 h was enteredinto the database each year. Each participating ICU chose toperform random sampling by taking either consecutive admis-sions in selected ICU beds all year long or consecutive admissionsin all ICU beds in a given month.

Data Collection

Data were collected daily by senior physicians in the partici-pating ICUs. For each patient, the investigators entered the datainto a computer case-report form using data-capture software(VIGIREA; OutcomeRea) and imported all records into theOutcomeRea database. All codes and definitions were establishedprior to study initiation. The following information was recorded:age and sex, admission category (medical, scheduled surgery, orunscheduled surgery), origin (home, ward, or emergency depart-ment), and McCabe score.16 Severity of illness was evaluated onthe first ICU day using the simplified acute physiology score(SAPS) II,17 logistic organ dysfunction (LOD) score,18 and acutephysiologic and chronic health evaluation II score.19 Knaus scaledefinitions were used to record preexisting chronic organ failuresincluding respiratory, cardiac, hepatic, renal, and immune systemfailure.19 The nine equivalents of nursing manpower use score(NEMS) was determined to measure the nursing workload foreach patient, which was taken as an indicator of treatmentintensity.20

Quality of the Database

For most of the study variables, the data-capture softwareimmediately ran an automatic check for internal consistency,generating queries that were sent to the ICUs before incorpora-tion of the new data into the database. In each participating ICU,data quality was checked by having a senior physician fromanother participating ICU review a 2% random sample of thestudy data. All the variables introduced in the analyses had a �coefficient � 0.6 for qualitative variables and an interrater coef-ficient of 0.67 to 1, indicating good to excellent reproducibility.

Study Population

The presence or absence of infections was documentedaccording to the standard definitions developed by the Cen-ters for Disease Control and Prevention21; in addition, aquantitative protected plugged catheter culture showing � 103

cfu/mL was used to diagnose pneumonia.22 Community-acquiredinfection was defined as infection manifesting before or within48 h after hospital admission. Hospital-acquired infection wasinfection manifesting at least 48 h after hospital admission butbefore ICU admission. Infection sites were categorized asfollows: pneumonia, peritonitis, urinary tract infection, exac-erbation of COPD, primary bacteremia (excluding untreatedStaphylococcus epidermidis), miscellaneous sites (mediastini-tis, prostatitis, osteomyelitis, and others), and multiple sites.Lengths of ICU and hospital stays were computed startingfrom ICU admission.

Severe sepsis was defined as infection with two or more criteriafor systemic inflammatory response syndrome and at least onecriterion for organ dysfunction. Criteria for systemic inflamma-tory response syndrome included core temperature � 38°C or� 36°C, heart rate � 90 beats/min, respiratory rate � 20 breaths/min, Pco2 � 32 mm Hg or use of mechanical ventilation, andperipheral leukocyte count � 12,000/�L or � 4,000/�L. Organdysfunction was defined as follows: (1) cardiovascular systemfailure with a need for vasopressors and/or inotropic drugs,and/or a systolic BP � 90 mm Hg, and/or a drop in systolic BP� 40 mm Hg from baseline; (2) renal dysfunction with urinaryoutput � 700 mL/d in a patient not previously receiving hemo-dialysis for chronic renal failure; (3) respiratory dysfunction witha Pao2 � 70 mm Hg or mechanical ventilation or a Pao2/fractionof inspired oxygen ratio � 250 (or � 200 in patients withpneumonia); (4) bone marrow failure with a platelet count� 80,000/�L; and (5) metabolic acidosis with a plasma lactatelevel � 3 mmol/L.

*From the Medical-Surgical ICU (Drs. Adrie and Darques),Delafontaine Hospital, Saint Denis; Medical ICU (Dr. Azoulay),Saint Louis Teaching Hospital, Paris; INSERM U823 (Drs.Francais and Timsit), Epidemiology of Cancer and Severe Ill-nesses, Albert Bonniot Institute, Grenoble; Medical-SurgicalICU (Dr. Clec’h), Avicenne Teaching Hospital, Bobigny; MedicalICU (Dr. Schwebel), Albert Michallon Teaching Hospital,Grenoble; Laboratory of Computer Sciences (Dr. Nakache),Centre National des Arts et Metiers, Paris; Medical-Surgical ICU(Dr. Jamali), Dourdan Hospital, Dourdan; Medical-Surgical ICU(Dr. Goldgran-Toledano), Gonesse Hospital, Gonesse; and Medical-Surgical ICU (Dr. Garrouste-Orgeas), Saint Joseph Hospital, Paris,France.†A list of participants is given in the Appendix.OutcomeRea is supported by nonexclusive educational grants fromAventis Pharma, France, and Wyeth, as well as by public funds fromthe Centre National de la Recherche Scientifique.The authors have no conflicts of interest to disclose.Manuscript received February 14, 2007; revision accepted July10, 2007.Reproduction of this article is prohibited without written permissionfrom the American College of Chest Physicians (www.chestjournal.org/misc/reprints.shtml).Correspondence to: Christophe Adrie, MD, PhD, Service de Reani-mation Polyvalente, Hopital Delafontaine, 2, rue du Dr Delafontaine,93205 Sant Denis, France; e-mail: [email protected]: 10.1378/chest.07-0420

www.chestjournal.org CHEST / 132 / 6 / DECEMBER, 2007 1787

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from

End Points

The primary end points were all-cause ICU mortality andpost-ICU mortality. Secondary end points were workloads withinthe first 2 days in the ICU.

Several analyses were planned. First, we assessed the influenceof gender on hospital mortality in the overall population ofpatients with severe sepsis. We then looked at the influence ofage by carefully matching male and female patients � 50 yearsold and those � 50 years old. The 50-year cut-off was chosen toseparate premenopausal and postmenopausal women.

Statistical Analysis

Results are expressed as numerical values and percentages forcategorical variables, and as medians and first and third quartilesfor continuous variables. For categorical data, gender compari-sons in the overall cohort were based on �2 tests or Fisher exacttests depending on sample size; for continuous data, Kruskal-Wallis or Wilcoxon tests were used. Since characteristics ofsevere sepsis may markedly influence the risk of death indepen-dently from gender, we developed a predictive model of death inthe overall population, which was used as a matching criterionwhen selecting men and women.

Logistic regression was performed subsequently to identifyindependent risk factors for hospital mortality in men and womenwith severe sepsis.23 Because of colinearity, severity and organfailure scores were not introduced simultaneously in the modelsbut, instead, tested consecutively and chosen according to theAkaike criterion. Calibration and discrimination of the finalmodel were assessed using the Hosmer-Lemeshow �2 test and Cstatistic, respectively.

We then designed a nested case-control study to comparefemale and male patients. This score was based on the results ofthe above-described multivariate logistic regression analysis. Us-ing an algorithm (available at http://www.outcomerea.org/ehtm/matchmacro.pdf), we matched female and male patients based onthree criteria: center, death propensity score24 (� 10%), and10-year age group. Wald �2 tests were used to determine thesignificance of each variable. Adjusted odds ratios (ORs) and 95%confidence intervals (CIs) were calculated for each parameterestimate, using conditional logistic regression. We then did asimilar analysis comparing premenopausal women (� 50 years)to men matched on the propensity score. Finally, we adjusted theconditional logistic regression on variables not balanced betweenmale and female patients and previously reported to be associ-ated with mortality; p values � 0.05 were considered significant.Analyses were computed using statistical software (SAS 8.2; SASInstitute; Cary, NC).

Results

Study Population

Of the 4,860 patients included in the database (allwith ICU stays � 48 h), 1,692 met our criteria forsevere sepsis; 63% were male and 37% were female(Fig 1). The women were older and had a higher rateof emergency surgery, lower organ dysfunctionscores (LOD and sequential organ failure assess-ment), and similar SAPS II scores at hospital admis-sion (Table 1). Chronic pulmonary dysfunction asassessed by Knaus definitions was more common inmale patients, but all other comorbidities were

evenly distributed between genders. Cardiovasculardysfunction was the leading organ dysfunction, withsimilar rates in men and women (67.3% vs 69.6%)and in patients who did and did not require vaso-pressor support (indicating septic shock, present in53% of male and 54% of female patients). Respira-tory failure was the only organ dysfunction that wassignificantly more common in men. Women hadlower rates of pneumonia and of multiple sources ofinfection but a higher rate of urinary tract infection.

Severe Sepsis and Gender

We evaluated the effect of gender based on thevariables that were independently associated withdeath in the multivariate logistic regression analysis(Table 2). We matched 1,000 men to 608 women inthe overall population of patients with severe sepsis.After matching on risk factors for death and adjust-ing for confounding factors (ie, chronic respiratoryfailure; metastatic cancer; immunocompromised sta-tus; emergency surgery; acute respiratory failure andshock at hospital admission; urinary tract infection asthe site of infection and E coli, S pneumoniae, andEnterobacter species as the causative microorgan-ism), the risk of hospital death was lower in women(OR, 0.75; 95% CI, 0.57 to 0.97; p � 0.02) [Table 3].

When we separated the patients based on age� 50 years or � 50 years, we found that mortalitywas significantly lower in women � 50 years old(postmenopausal, n � 481) than in men � 50 yearsold (n � 778) [OR, 0.69; 95% CI, 0.52 to 0.93] afteradjusting for confounding variables (p � 0.014).

1692 patients with severe sepsis

1061 males (63%)

Age 50 yearsN=261

Mortality: 54 (21%)

Age > 50 yearsN=800

Mortality: 271 (34%)

631 females(37%)

Age 50 yearsN= 137

Mortality: 31 (23%)

Age >50 yearsN= 494

Mortality: 138 (28%)

1970 patients with SIRS

2045 patients with infection

Outcomerea Database:4860 patients

<_ <_

Figure 1. Flow diagram of the 1,692 patients with severe sepsiswho formed the basis for the study and who were taken from the4,860 patients included in the OutcomeRea database. Data areexpressed as No. (%).

1788 Original Research

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from

Table 1—Baseline Characteristics of the 1,692 Patients With Severe Sepsis According to Gender*

VariablesAll

(n � 1,692)Men

(n � 1,061) �63%Women

(n � 631) �37% p Value

Age, yr 66 (51–76) 65 (51–75) 69 (53–77) 104

SAPS II 45 (35–59) 45 (34–60) 44 (33–58) 0.23LOD score 5 (2–7) 5 (2–7) 4 (2–6) 103

Sequential organ failure assessment score 6 (4–9) 6 (4–9) 6 (4–8) 0.04APACHE II score 19 (14–24) 19 (14–24) 19 (14–24) 0.95Body weight, kg 67 (57–80) 70 (60–80) 61 (52–74) 104

Height, cm 168 (160–175) 172 (168–177) 160 (155–165) 104

Body mass index† 23 (20–27) 23 (21–27) 24 (20–29) 102

Admission categoriesMedical 1,294 (76.7) 830 (78.4) 464 (74) 0.03Scheduled surgery 77 (4.6) 54 (5.1) 23 (3.7) 0.20Emergency surgery 316 (18.7) 174 (16.4) 142 (22.3) 103

Transferred from another ward 817 (48.3) 503 (47.4) 286 (45.8) 0.55McCabe score (n � 17 missing)

1 824 (48.7) 500 (47.5) 324 (52) 0.072 670 (39.6) 433 (41.2) 237 (38)3 181 (10.1) 119 (11.3) 62 (10)

Comorbidities (Knaus definitions)At least one comorbidity 871 (51) 564 (53) 307 (49) 0.08Chronic pulmonary failure 389 (22.3) 266 (25) 123 (19.5) 0.01Immunocompromised status 326 (19.3) 207 (19.5) 119 (18.9) 0.80Chronic heart failure 194 (11.5) 123 (11.6) 71 (11.2) 0.90Chronic hepatic failure 94 (5.5) 66 (6.2) 28 (4.4) 0.15Chronic renal failure 48 (2.8) 29 (2.7) 19 (3) 0.85

Main symptom at hospital admission‡Acute respiratory failure 622 (36.7) 411 (38.7) 211 (32.9) 0.03Shock 499 (29.5) 294 (27.8) 205 (32.5) 0.02COPD exacerbation 130 (8) 84 (7.9) 46 (7.2) 0.03Coma 185 (11.9) 115 (10.8) 70 (10.9) 0.20Monitoring and scheduled surgery 17 (1) 12 (1.1) 5 (0.8) 0.67Multiple organ failure 80 (4.7) 54 (5.1) 26 (4.1) 0.43Acute renal failure 45 (2.6) 26 (2.4) 19 (2.9) 0.59Trauma 17 (1) 5 (0.5) 12 (1.9) 0.93

Sites of infectionPneumonia 726 (42.9) 500 (47.1) 226 (35.8) � 104

COPD exacerbation or bronchitis 242 (14.3) 149 (14) 93 (14.7) 0.75Primary bacteremia 198 (11.7) 112 (10.5) 86 (13.6) 0.07Peritonitis 152 (8.9) 83 (7.8) 69 (10.9) 0.15Urinary tract infection 119 (7) 51 (4.8) 68 (10.8) � 104

Multiple 301 (18) 123 (21) 178 (16) 0.004Organ dysfunction§

Cardiovascular dysfunction 1,163 (68.7) 738 (69.6) 425 (67.3) 0.34Renal failure 845 (50) 520 (49) 325 (51.5) 0.32Respiratory failure 871 (51) 571 (53.8) 300 (47.5) 0.01Hematologic dysfunction 345 (20.4) 221 (20.8) 124 (19.7) 0.56Unexplained metabolic acidosis 434 (25.6) 268 (25.3) 166 (26.3) 0.63

Type of microorganism�Streptococcus pneumoniae 88 (5.2) 66 (6.2) 22 (3.5) 0.01Methicillin-sensitive Staphylococcus aureus 11 (0.7) 9 (0.8) 2 (0.3) 0.19Methicillin-resistant S aureus 60 (3.5) 41 (3.9) 19 (3.0) 0.36Coagulase-negative staphylococci 13 (0.8) 11 (1) 2 (0.3) 0.10Escherichia coli 121 (7.2) 57 (5.4) 64 (10.1) 104

Other Enterobacteriaceae 49 (2.9) 38 (3.6) 11 (1.7) 0.03Acinetobacter baumanii 2 (0.1) 1 (0.1) 1 (0.2) 0.71Pseudomonas aeruginosa 39 (2.3) 28 (2.6) 11 (1.7) 0.24Yeasts 9 (0.5) 4 (0.4) 5 (0.8) 0.26Multimicrobial 61 (3.6) 44 (4.1) 17 (2.7) 0.12

*Data are presented as median (first and third quartiles) or No. (%). The total number of sites of infection is greater than the number of patientsbecause some patients had infection at more than one site. APACHE � acute physiology and chronic health evaluation.

†As there were 124 missing values in the height and/or weight measurements, body mass index was not taken into account in the logistic regressionmodel of factors associated with death. A subgroup analysis of clusters without missing body mass index values produced a similar result.

‡Referred to the main symptom that led to ICU admission.§Organ dysfunction from severe sepsis (see text for definition).�Type of microorganism causing sepsis for all sites of infection. We specified only main types. Some patients had multiple microorganisms.

www.chestjournal.org CHEST / 132 / 6 / DECEMBER, 2007 1789

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from

Mortality was not significantly different between the127 women and the 222 men � 50 years of age(Table 3). These results remained unchanged whenpatients with early do-not-resuscitate (DNR) orderswere excluded.

In the matched population, women had a lowerrate of central venous line use, shorter mechanicalventilation times, and shorter ICU stay lengths;however, these differences were not significant afteradjustment for confounding factors. Furthermore,workload as assessed by the NEMS score on day 1,day 2, and mean NEMS for day 1 and day 2 was notsignificantly different between men and women,indicating a similar level of care (Table 4).

Discussion

Mortality from severe sepsis was higher in menthan in women, after adjustment for confoundingfactors. This difference was due to higher mortalityin men � 50 years old compared to same-age (post-menopausal) women; mortality was not significantlydifferent between younger men and women. Thelevel of care and rate of invasive procedures weresimilar in women and men.

Our findings agree with previous data showing ahigher incidence of sepsis in men1,2 compared towomen. Numerous studies1,3,8–14 have evaluated theinfluence of gender on survival in patients with

established sepsis, with conflicting results. For instance,in surgical units, survival was better in women,3 betterin men,14 or similar in men and women.8 Althoughdifferences in case-mix, as stated earlier, and samplesize contributed to these discrepancies, the mainfactor was probably imperfect matching of male andfemale patients. An important strength of thepresent study is the use of a propensity score in alarge cohort of patients (n � 1,692, predominantlymedical patients), which allowed us to obtain twogroups that were very accurately matched on con-founding factors. We found that mortality was higherin men in the overall cohort of patients with severesepsis.

Differences in level of care may lead to differencesin survival between men and women. Several stud-ies25–27 showed that women were less likely thanmen to undergo intensive evaluation and invasive

Table 2—Variables Independently Associated WithDeath in the Logistic Regression Analysis of Data

From Patients With Severe Sepsis*

ParametersDegree ofFreedom Estimate

p Value,�2 OR

Intercept 1 3.9959 � 0.0001 0.018DNR order within the

first 2 d1 1.4801 � 0.0001 4.393

Transfer from anotherward

1 0.4153 0.0005 1.515

Multiorgan failure 1 1.1478 � 0.0001 3.151At least one chronic

disease†1 0.7024 � 0.0001 2.019

Swan-Ganz catheterwithin the first 2 d

1 0.7939 0.0001 2.212

Pneumonia as a sourceof severe sepsis

1 0.3423 0.0042 1.408

SAPS at admission 1 0.0481 � 0.0001 1.049

*The variables not found significant in the multivariate regressionwere age; shock; acute respiratory failure; exacerbation of COPD;McCabe score; peritonitis, primary bacteremia, or multiple sites ofinfection; chronic hepatic failure; hematologic malignancy; need forarterial and central venous lines; hemodialysis within the first 2 days,and type of microorganism (E coli, S pneumoniae, and Enterobacterspecies). Final model: Hosmer-Lemeshow of 11.3 (p � 0.18) indi-cated a good fit (C statistic � 0.80).

†According to Knaus definitions.

Table 3—Influence of Gender on Mortality in PatientsWith Severe Sepsis*

Variables OR‡ 95% CI p Value

Overall matched cohort (608women, 1,000 men)

After matching on risk factors fordeath

ICU mortality 0.80 0.62–1.03 0.09Hospital mortality 0.78 0.61–1.01 0.06

After adjusting for confoundingvariables†

ICU mortality 0.75 0.58–0.98 0.03Hospital mortality 0.75 0.57–0.97 0.02

Patients � 50 yr old (481 women,778 men)

After matching on risk factors fordeath

ICU mortality 0.73 0.55–0.97 0.03Hospital mortality 0.71 0.54–0.94 0.02

After adjusting for confoundingvariables†

ICU mortality 0.70 0.52–0.94 0.018Hospital mortality 0.69 0.52–0.93 0.014

Patient � 50 yr old (127 women,222 men)

After matching on risk factors fordeath

ICU mortality 1.33 0.72–2.46 0.36Hospital mortality 1.34 0.73–2.44 0.35

After adjusting for confoundingvariables†

ICU mortality 1.01 0.51–1.99 0.98Hospital mortality 1.01 0.52–1.97 0.98

*These results were obtained using conditional logistic regressionwith matching on age, death propensity score, and center.

†Chronic respiratory failure; metastatic cancer; immunocompro-mised status; emergency surgery; acute respiratory failure and shockat hospital admission; urinary tract infection as a cause of sepsis; andtype of microorganism (E coli, S pneumoniae, and Enterobacterspecies).

‡OR of death according to conditional logistic regression.

1790 Original Research

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from

treatment for cardiovascular disease. Data from theUnited States indicate that women are more likelythan men to receive recommended preventive andchronic care but less likely to receive recommendedacute care.28 However, greater utilization of acute-care resources in men may be ascribable to a fewwidely used procedures, such as invasive proceduresfor cardiovascular disease, and may mask lowerutilization of resources for specific acute disorders.In Austria, Valentin et al15 documented a higherlevel of care with greater use of invasive proceduresin men compared to women admitted to ICUs forany reason. This gender difference was found in allage groups, including the oldest patients. Althoughdisease severity was greater in women, survival wasnot significantly different, suggesting either an inap-propriately high level of care in men or a betterpotential for survival in women masked by an inap-propriately low level of care. Resource use accordingto gender may vary across health-care systems. Inaddition, Valentin et al15 studied the overall popula-tion of ICU patients, as opposed to patients withsevere sepsis. In our study, the NEMS values sug-gested similar levels of care in men and women. Thissimilar level of care may explain the higher survivalrate in women in our study, in contradiction to theresults reported by Valentin et al.15

As stated in the introduction, hypothesized mech-anisms of gender-based differences in the responseto sepsis would predict better survival in premeno-pausal women than in men. However, we foundbetter survival in women � 50 years old (ie, post-menopausal) than in same-age men, with no signifi-cant survival difference between younger womenand men. First, we cannot rule out that the absenceof a significant gender-based mortality difference inour younger population was due to the small numberof patients and lower fatality rate. Severe sepsis is farmore common in older individuals than in youngerage groups. Second, estrogens produced outside theovaries may confer protection to postmenopausalwomen; the main source is probably the adrenalcortex, although T cells and macrophages or fat29

may also contribute to the high sex-steroid levelsobserved in women. The metabolism of the adrenalhormone dehydroepiandrosterone is a major deter-minant of sex-steroid status in postmenopausal women.Dehydroepiandrosterone is a very weak androgen butcan be converted to either more potent androgens orestrogens by peripheral tissue enzymes (5�-reductasefor conversion to dihydrotestosterone and aromatasefor conversion to 17�-estradiol).29 Both advancingage and higher adipocyte mass are known to increasearomatase activity. The higher body mass indexobserved in women than men may have led to betterprotection as a result of greater aromatase activity infat tissue via estrogen production.29 Third, hormonereplacement therapy used by some postmenopausalwomen may improve responses to infection, al-though this hypothesis needs evaluation. Fourth, inwomen, high levels of estrogen for years may even-tually lead to health benefits becoming apparent onlylater in life, compared to men. Fifth, gender-baseddifferences in cytokine secretion by peripheral bloodmononuclear cells may lead to poorer outcomes inmale patients.30,31 Conceivably, these differences maybe more marked in postmenopausal than premeno-pausal women, or their effects may be masked bycounterbalancing factors in premenopausal women.Cytokine secretion differences between premeno-pausal and postmenopausal women with sepsis de-serve to be investigated. Finally, differences inhealth-related behaviors between men and womenover the life span may eventually lead to differencesin outcomes late in life.

In a recent study,32 survival in elderly patients withsevere infection was similar in men and women butvaried with the sex-steroid profile. In this study, theabsence of a gender difference may be ascribable tothe smaller sample size and to the inclusion ofpatients with sepsis, as opposed to severe sepsis, inour study. Furthermore, confounding factors werenot well taken into account.32 We did not assay sex

Table 4—Level of Care and Use of InvasiveProcedures at Hospital Admission in the Cross-

Matched Population of Patients With Severe Sepsis*

VariablesMen

(n � 1,000)Women

(n � 608) p Value

Mechanical ventilation 347 (35) 188 (30) 0.12Duration, d 9 (4–18) 7 (4–16) 0.01

Central venous catheter 485 (49) 324 (52) 0.05Duration, d 10 (5–17) 9 (5–15) 0.23

Arterial line 267 (27) 137 (22) 0.14Duration, d 4 (3–8) 5 (3–8) 0.22

Renal support 45 (5) 32 (5) 0.49Duration, d 3 (2–7) 3 (2–5) 0.11

Antibiotics, % 913 (91) 539 (86) 0.08Duration, d 8 (5–16) 8 (5–13) 0.07

Corticosteroid treatment 266 (27) 183 (29) 0.06Duration, d 7 (4–12) 6 (4–10) 0.29

Vasopressor support, % 511 (51) 303 (50) 0.62Duration, d 5 (3–10) 4 (3–8) 0.1

NEMS on day 1 27 (21–39) 27 (21–39) 0.12NEMS on day 2 33 (26–39) 33 (21–39) 0.73Mean NEMS days 1–2 33 (21–39) 33 (21–39) 0.10DNR order, % 204 (20) 132 (21) 0.57Time from hospital admission

to DNR order, d26 (14–48) 25 (14–47) 0.92

ICU length of stay, d 10 (5–19) 9 (5–17) 0.03Hospital length of stay, d 26 (14–48) 25 (14–47) 0.92ICU mortality 289 (29) 159 (26) 0.23Hospital mortality 336 (34) 188 (31) 0.26

*Data are presented as No. (%) or median (first-third quartiles).

www.chestjournal.org CHEST / 132 / 6 / DECEMBER, 2007 1791

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from

hormones in our study. However sex hormone pro-files during severe sepsis may fail to reflect baselinehormone production, since severe sepsis is oftenpreceded by several days of systemic inflammation, aprocess known to decrease testosterone levels33,34

and to increase 17� estradiol synthesis via an in-crease in aromatase activity.35,36

In conclusion, women with severe sepsis had alower risk of hospital mortality than did men care-fully matched on confounding variables. This differ-ence was present only in the group of individuals� 50 years old and was not ascribable to differencesin level of care. Further studies are required toevaluate whether or not there is a need for specifictreatment depending on the gender.

Appendix

Members of the OutcomeRea Study Group

Scientific Committee: Jean-Francois Timsit (Hopital AlbertMichallon and INSERM U823, Grenoble, France); Pierre Moine(Surgical ICU, Denver, CO); Arnaud De Lassence (ICU, HopitalLouis Mourier, Combes, France); Elie Azoulay (Medical ICU,Hopital Saint Louis, Paris, France); Yves Cohen (ICU, HopitalAvicenne, Bobigny, France); Maıte Garrouste-Orgeas (ICU Ho-pital Saint-Joseph, Paris, France); Lilia Soufir (ICU, HopitalSaint-Joseph, Paris, France); Jean-Ralph Zahar (MicrobiologyDepartment, Hopital Necker, Paris, France); Christophe Adrie(ICU, Hopital Delafontaine, Saint Denis, France); Adel Benali(Microbiology and Infectious Diseases, Hopital Saint-Joseph,Paris France); Christophe Clec’h (ICU, Hopital Avicenne, Bo-bigny, France); and Jean Carlet (ICU, Hopital Saint-Joseph,Paris, France).

Biostatistical and Informatics Expertise: Jean-Francois Timsit(Group of Epidemiology, INSERM U823, Grenoble, France);Sylvie Chevret (Medical Computer Sciences and BiostatisticsDepartment, Hopital Saint-Louis, Paris, France); Corinne Al-berti (Medical Computer Sciences and Biostatistics Department,Robert Debre, Paris, France); Aurelien Vesin (Group of Epide-miology, INSERM U823, Grenoble, France); Adrien Francais(Group of Epidemiology, INSERM U823, Grenoble, France);Muriel Tafflet (Outcomerea, France); Frederik Lecorre (Su-pelec, France); and Didier Nakache (Conservatoire National desArts et Metiers, Paris, France).

Investigators of the OutcomeRea Database: Christophe Adrie(ICU, Hopital Delafontaine, Saint Denis, France); Bernard Al-laouchiche (surgical ICU, Hopital Edouard Herriot, Lyon); CarolineBornstain (ICU, Hopital de Montfermeil, France); Alexandre Boyer(ICU, Hopital Pellegrin, Bordeaux, France); Antoine Caubel (ICU,Hopital Saint-Joseph, Paris, France); Christine Cheval (SICU, Ho-pital Saint-Joseph, Paris, France); Marie-Alliette Costa de Beaure-gard (Nephrology, Hopital Tenon, Paris, France); Jean-Pierre Colin(ICU, Hopital de Dourdan, Dourdan, France); Anne-SylvieDumenil (Hopital Antoine Beclere, Clamart France); AdrienDescorps-Declere (Hopital Antoine Beclere, Clamart France);Jean-Philippe Fosse (ICU, Hopital Avicenne, Bobigny, France);Samir Jamali (ICU, Hopital de Dourdan, Dourdan, France); Chris-tian Laplace (ICU, Hopital Kremlin-Bicetre, Bicetre, France);Thierry Lazard (ICU, Hopital de la Croix Saint-Simon, Paris,France); Eric Le Miere (ICU, Hopital Louis Mourier,Combes,France); Laurent Montesino (ICU, Hopital Bichat, Paris,

France); Bruno Mourvillier (ICU, Hopital Bichat, France);Benoît Misset (ICU, Hopital Saint-Joseph, Paris, France); Del-phine Moreau (ICU, Hopital Saint-Louis, Paris, France); EtiennePigne (ICU, Hopital Louis Mourier, Combes, France); CaroleSchwebel (CHU A Michallon, Grenoble, France); Gilles Troche(Hopital Antoine, Beclere, Clamart France); Marie Thuong(ICU, Hopital Delafontaine, Saint Denis, France); GuillaumeThierry (ICU, Hopital Saint-Louis, Paris, France); Dany Tole-dano (CH Gonesse, France); Eric Vantalon (SICU, HopitalSaint-Joseph, Paris, France); and Francois Vincent (ICU, HopitalAvicenne, Bobigny, France).

ACKNOWLEDGMENT: We are indebted to A. Wolfe, MD, forhelping with this manuscript.

References1 Martin GS, Mannino DM, Eaton S, et al. The epidemiology

of sepsis in the United States from 1979 through 2000.N Engl J Med 2003; 348:1546–1554

2 Adrie C, Alberti C, Chaix-Couturier C, et al. Epidemiologyand economic evaluation of severe sepsis in France: age,severity, infection site, and place of acquisition (community,hospital, or intensive care unit) as determinants of workloadand cost. J Crit Care 2005; 20:46–58

3 Schroder J, Kahlke V, Staubach KH, et al. Gender differencesin human sepsis. Arch Surg 1998; 133:1200–1205

4 Hubacek JA, Stuber F, Frohlich D, et al. Gene variants of thebactericidal/permeability increasing protein and lipopolysaccha-ride binding protein in sepsis patients: gender-specific geneticpredisposition to sepsis. Crit Care Med 2001; 29:557–561

5 Schroder J, Kahlke V, Book M, et al. Gender differences insepsis: genetically determined? Shock 2000; 14:307–310;discussion 310–303

6 Zellweger R, Wichmann MW, Ayala A, et al. Females inproestrus state maintain splenic immune functions and toler-ate sepsis better than males. Crit Care Med 1997; 25:106–110

7 Angele MK, Wichmann MW, Ayala A, et al. Testosteronereceptor blockade after hemorrhage in males: restoration ofthe depressed immune functions and improved survival fol-lowing subsequent sepsis. Arch Surg 1997; 132:1207–1214

8 Wichmann MW, Inthorn D, Andress HJ, et al. Incidence andmortality of severe sepsis in surgical intensive care patients:the influence of patient gender on disease process andoutcome. Intensive Care Med 2000; 26:167–172

9 Offner PJ, Moore EE, Biffl WL. Male gender is a risk factorfor major infections after surgery. Arch Surg 1999; 134:935–938; discussion 938–940

10 Crabtree TD, Pelletier SJ, Gleason TG, et al. Gender-dependent differences in outcome after the treatment ofinfection in hospitalized patients. JAMA 1999; 282:2143–2148

11 Bindl L, Buderus S, Dahlem P, et al. Gender-based differ-ences in children with sepsis and ARDS: the ESPNIC ARDSDatabase Group. Intensive Care Med 2003; 29:1770–1773

12 Mostafa G, Huynh T, Sing RF, et al. Gender-related out-comes in trauma. J Trauma 2002; 53:430–434; discussion434–435

13 George RL, McGwin G Jr, Schwacha MG, et al. Theassociation between sex and mortality among burn patients asmodified by age. J Burn Care Rehabil 2005; 26:416–421

14 Eachempati SR, Hydo L, Barie PS. Gender-based differencesin outcome in patients with sepsis. Arch Surg 1999; 134:1342–1347

15 Valentin A, Jordan B, Lang T, et al. Gender-related differ-ences in intensive care: a multiple-center cohort study oftherapeutic interventions and outcome in critically ill pa-tients. Crit Care Med 2003; 31:1901–1907

1792 Original Research

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from

16 McCabe WR, Jackson GG. Gram-negative bacteremia: I.Etiology and ecology. Arch Intern Med 1962; 110:847–853

17 Le Gall JR, Lemeshow S, Saulnier F. A new simplified acutephysiology score (SAPS II) based on a European/NorthAmerican multicenter study. JAMA 1993; 270:2957–2963

18 Le Gall JR, Klar J, Lemeshow S, et al. The logistic organdysfunction system: a new way to assess organ dysfunction inthe intensive care unit; ICU Scoring Group. JAMA 1996;276:802–810

19 Knaus WA, Draper EA, Wagner DP, et al. APACHE II: aseverity of disease classification system. Crit Care Med 1985;13:818–829

20 Reis Miranda D, Moreno R, Iapichino G. Nine equivalents ofnursing manpower use score (NEMS). Intensive Care Med1997; 23:760–765

21 Garner JS, Jarvis WR, Emori TG, et al. CDC definitions fornosocomial infections, 1988. Am J Infect Control 1988;16:128–140

22 Pham LH, Brun-Buisson C, Legrand P, et al. Diagnosis ofnosocomial pneumonia in mechanically ventilated patients:comparison of a plugged telescoping catheter with the pro-tected specimen brush. Am Rev Respir Dis 1991; 143:1055–1061

23 Hosmer D, Lemeshow S. Applied logistic regression, 1989ed. New York, NY: John Wiley & Sons, 1989

24 Joffe MM, Rosenbaum PR. Invited commentary: propensityscores. Am J Epidemiol 1999; 150:327–333

25 Ayanian JZ, Epstein AM. Differences in the use of proce-dures between women and men hospitalized for coronaryheart disease. N Engl J Med 1991; 325:221–225

26 Jaglal SB, Goel V, Naylor CD. Sex differences in the use ofinvasive coronary procedures in Ontario. Can J Cardiol 1994;10:239–244

27 Vaccarino V, Rathore SS, Wenger NK, et al. Sex and racialdifferences in the management of acute myocardial infarc-tion, 1994 through 2002. N Engl J Med 2005; 353:671–682

28 Asch SM, Kerr EA, Keesey J, et al. Who is at greatest risk forreceiving poor-quality health care? N Engl J Med 2006;354:1147–1156

29 Federman DD. The biology of human sex differences.N Engl J Med 2006; 354:1507–1514

30 van Eijk LT, Dorresteijn MJ, Smits P, et al. Gender differ-ences in the innate immune response and vascular reactivityfollowing the administration of endotoxin to human volun-teers. Crit Care Med 2007; 35:1464–1469

31 Asai K, Hiki N, Mimura Y, et al. Gender differences incytokine secretion by human peripheral blood mononuclearcells: role of estrogen in modulating LPS-induced cytokinesecretion in an ex vivo septic model. Shock 2001; 16:340–343

32 Angstwurm MW, Gaertner R, Schopohl J. Outcome in elderlypatients with severe infection is influenced by sex hormonesbut not gender. Crit Care Med 2005; 33:2786–2793

33 Sam AD II, Sharma AC, Lee LY, et al. Sepsis producesdepression of testosterone and steroidogenic acute regulatory(StAR) protein. Shock 1999; 11:298–301

34 Mechanick JI, Nierman DM. Gonadal steroids in criticalillness. Crit Care Clin 2006; 22:87–103

35 Zhao Y, Nichols JE, Valdez R, et al. Tumor necrosis factor-�stimulates aromatase gene expression in human adiposestromal cells through use of an activating protein-1 bindingsite upstream of promoter 1.4. Mol Endocrinol 1996; 10:1350–1357

36 Macdiarmid F, Wang D, Duncan LJ, et al. Stimulation ofaromatase activity in breast fibroblasts by tumor necrosisfactor �. Mol Cell Endocrinol 1994; 106:17–21

www.chestjournal.org CHEST / 132 / 6 / DECEMBER, 2007 1793

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from

DOI 10.1378/chest.07-0420; Prepublished online September 21, 2007; 2007;132; 1786-1793Chest

Goldgran-Toledano, Maïté Garrouste-Orgeas and Jean François TimsitDarques, Carole Schwebel, Didier Nakache, Samir Jamali, Dany

Christophe Adrie, Elie Azoulay, Adrien Francais, Christophe Clec'h, Loic : A Reappraisal*Influence of Gender on the Outcome of Severe Sepsis

 May 17, 2011This information is current as of

http://chestjournal.chestpubs.org/content/132/6/1786.full.htmlUpdated Information and services can be found at:

Updated Information & Services

http://chestjournal.chestpubs.org/content/132/6/1786.full.html#ref-list-1This article cites 35 articles, 10 of which can be accessed free at:

References

http://chestjournal.chestpubs.org/content/132/6/1786.full.html#related-urlsThis article has been cited by 6 HighWire-hosted articles:

Cited Bys

http://www.chestpubs.org/site/misc/reprints.xhtmlfound online at: Information about reproducing this article in parts (figures, tables) or in its entirety can bePermissions & Licensing

http://www.chestpubs.org/site/misc/reprints.xhtmlInformation about ordering reprints can be found online:

Reprints

"Services" link to the right of the online article.Receive free e-mail alerts when new articles cite this article. To sign up, select the

Citation Alerts

PowerPoint slide format. See any online figure for directions. articles can be downloaded for teaching purposes inCHESTFigures that appear in Images in PowerPoint format

 © 2007 American College of Chest Physicians by guest on May 17, 2011chestjournal.chestpubs.orgDownloaded from