ORIGINAL CONTRIBUTION
The DISPARITY-II Study: Delays to Antibiotic Administration in Women With Severe Sepsis or Septic Shock Tracy E. Madsen, MD, and Anthony M. Napoli, MD
Abstract Background: Early antibiotics reduce mortality in patients with severe sepsis and septic shock. Recent work demonstrated that women experience greater delays to antibiotic administration, but it is unknown if this relationship remains after adjusting for factors such as source of infection. Objectives: The objective was to investigate whether gender and/or source of infection are associated with delays to antibiotics in patients with severe sepsis or septic shock. Methods: This was a retrospective, observational study in an urban academic emergency department and national Surviving Sepsis Campaign (SSC) database study site. Consecutive patients age 18 years and older admitted to intensive care with severe sepsis or septic shock and entered into the SSC database from October 2005 to March 2012 were included. Two trained research assistants, blinded to the primary outcome, used a standardized abstraction form to obtain patient demographic and clinical data, including the Sequential Organ Failure Assessment (SOFA) scores and comorbidities. Time to first antibiotic and presumed source of infection were extracted from the SSC database. Univariate analyses were performed with Pearson chi-square tests and t-tests. Linear regression was performed with time to first antibiotic as the primary outcome. Covariates, chosen a priori by study authors, included age, race, ethnicity, source of infection, SOFA score, and lactate. Results: A total of 771 patients were included. Women were 45.3% of the sample, the mean age was 66 years (95% confidence interval [CI] = 65.1 to 67.5 years), 19.4% were nonwhite, and 8% were Hispanic. Mean time to first antibiotic was 153 minutes (95% CI = 143 to 163 minutes) for men and 184 minutes (95% CI = 171 to 197 minutes) for women (p < 0.001). The urinary tract was source of infection for 35.2% of women (95% CI = 30.2% to 40.3%) versus 23.7% (95% CI = 19.6% to 27.8%) of men. Pneumonia was present in 46.9% of men (95% CI = 42.1% to 51.7%) versus 35.8% (95% CI = 30.8% to 40.8%) of women. The mean time to antibiotics in women was longer than in men (adjusted odds ratio [aOR] = 1.18, 95% CI = 1.07 to 1.30), even after adjusting for age, race, ethnicity, presumed source of infection, SOFA score, and lactate (p = 0.001). Those with pneumonia compared to other infections received antibiotics faster (aOR = 0.73, 95% CI = 0.66 to 0.81). There was no significant association between other sources of infection and time to antibiotics in either univariate or multivariate analysis. Conclusions: Women experience longer delays to initial antibiotics among patients with severe sepsis or septic shock, even after adjusting for infectious source. Pneumonia was associated with shorter times to antibiotic administration. Future research is necessary to investigate contributors to delayed antibiotic administration in women. ACADEMIC EMERGENCY MEDICINE 2014;21:1499–1502 © 2014 by the Society for Academic Emergency Medicine
S
evere sepsis and septic shock are associated with high mortality rates in both men and women. Similar to conditions such as acute myocardial
infarction,1 women with severe sepsis and septic shock are treated less aggressively compared to men.2 Women are less likely to receive early goal-directed therapy
From the Division of Women’s Health in Emergency Care (TEM), Department of Emergency Medicine (TEM, AMN), Alpert Medical School of Brown University/Rhode Island Hospital, Providence, RI. Received March 11, 2014; revision received April 18, 2014; accepted April 29, 2014. Funding for this project was provided by the Department of Emergency Medicine at Brown University/Rhode Island Hospital. The authors have no relevant financial information or potential conflicts to disclose. Supervising Editor: Shariar Zehtabchi, MD. Address for correspondence and reprints: Tracy E. Madsen, MD; e-mail: [email protected].
© 2014 by the Society for Academic Emergency Medicine doi: 10.1111/acem.12546
ISSN 1069-6563 PII ISSN 1069-6563583
1499 1499
1500
(EGDT)2 and other measures of care, including central venous catheters.3 The Surviving Sepsis Campaign (SSC) bundle, including administration of antibiotics within 3 hours, is associated with decreased sepsis mortality,4 and for every hour in antibiotic delay, mortality increases by 7.6% for those with septic shock.5 Our previous work showed that women are less likely to receive antibiotics within 3 hours,6 but it was unclear whether this difference was due to an increased prevalence of urinary tract infections (UTIs) in women.7 The primary objective of this study was to investigate gender as a predictor of delays to antibiotic administration in severe sepsis or septic shock. Our secondary objective was to investigate infectious source as a contributor to gender disparities in time to antibiotic administration, if such a disparity existed. METHODS Study Design This was a single-center, retrospective, observational cohort study. This study was approved by the institutional review board. Study Setting and Population The study included adults 18 years or older with severe sepsis or septic shock who were seen at a large (>100,000 yearly visits), urban, academic emergency department (ED) between October 2005 and March 2012. Adults admitted to the medical intensive care unit (ICU) with severe sepsis or septic shock8 were entered into the SSC database, becoming part of the national SSC data set, and were included in the study. Patients not admitted to the ICU, and those receiving comfort measures only during the first 24 hours, were excluded. Study Protocol Data regarding arrival time, ED antibiotic administration, and infectious source were entered into the SSC database during the ICU admissions by quality assurance (QA) nurses; data entry procedures are standardized according to the national SSC. The Sequential Organ Failure Assessment (SOFA) score was used as a measure of illness severity.9 Sources of infection, determined by QA nurses using the physicians’ admission notes, were grouped into categories designated by the SSC:4 pneumonia, UTI, abdominal, meningitis, skin, bone, wound, catheter, endocarditis, device, and other. Variables were created for patients with more than one documented source or with unknown sources. Two trained research assistants used standardized data abstraction forms to abstract data pertaining to study covariates from patient charts. For data elements collected by the research assistants, interrater reliability (IRR) testing was performed. This study was conducted using recommended methodologies including abstractor training and monitoring, abstraction forms, regular meetings, and IRR.10 The primary outcome was time to antibiotics (the interval in minutes from ED arrival to administration of
Madsen and Napoli • GENDER AND ANTIBIOTIC DELAYS IN SEPSIS
first antibiotic), consistent with the SSC quality marker. Covariates included race, ethnicity, age, SOFA score, ED lactate, and presumed source of infection. Data Analysis Means, medians, and proportions were used to describe data. t-tests and chi-square tests were used for bivariate analyses. Linear regression was used for multivariate analyses; no data were censored. Model covariates were chosen a priori and entered directly into models; given lack of prior literature on predictors of time to antibiotics in sepsis, potential covariates were based on scientific plausibility and study hypotheses. Models were evaluated using tests of multicollinearity, homoscedasticity, and normality of residuals. Following model diagnostics, the outcome variable time to antibiotics was log-transformed, and model diagnostics were repeated to ensure model fit. Model coefficients were exponentiated and reported as adjusted odds ratios (aOR). The final model included gender, age, race, ethnicity, pneumonia, UTI, abdominal or skin source, SOFA, and lactate. Other sources of infection were not included because of their low frequencies and lack of significance with time to antibiotics. Variables indicating multiple sources or unknown sources did not improve model fit, so were not included in the final model. The familywise Type I error rate was 0.05. Analyses were completed using Stata 12.1. RESULTS A total of 771 patients were included; the mean age was 66.3 years (95% confidence interval [CI] = 65.1 to 67.5 years), and 19.5% were nonwhite. Baseline characteristics including age, race, and ethnicity were similar by gender. Women had slightly lower median SOFA scores (6, interquartile range [IQR] = 4 to 8 vs. 7, IQR = 5 to 9). IRR ranged from 0.82 to 0.97. Sources of infection also differed by gender: fewer women had pneumonia (35.8%, 95% CI = 30.8% to 40.8%) than men (46.9%, 95% CI = 42.1% to 51.7%), but more women had UTIs (35.2%, 95% CI = 30.2% to 40.3%) than men (23.7%, 95% CI = 19.6% to 27.8%). More women had multiple infectious sources (16.3%, 95% CI = 12.5% to 20.2%) than men (11.4%, 95% CI = 8.3% to 14.4%). Frequency of all other sources of infection was similar by gender. Unadjusted, the mean time to antibiotics was 31 minutes longer for women (184 minutes, 95% CI = 171 to 197) than for men (153 minutes, 95% CI = 143 to 163 minutes; p
The DISPARITY-II Study: Delays to Antibiotic Administration in Women With Severe Sepsis or Septic Shock Tracy E. Madsen, MD, and Anthony M. Napoli, MD
Abstract Background: Early antibiotics reduce mortality in patients with severe sepsis and septic shock. Recent work demonstrated that women experience greater delays to antibiotic administration, but it is unknown if this relationship remains after adjusting for factors such as source of infection. Objectives: The objective was to investigate whether gender and/or source of infection are associated with delays to antibiotics in patients with severe sepsis or septic shock. Methods: This was a retrospective, observational study in an urban academic emergency department and national Surviving Sepsis Campaign (SSC) database study site. Consecutive patients age 18 years and older admitted to intensive care with severe sepsis or septic shock and entered into the SSC database from October 2005 to March 2012 were included. Two trained research assistants, blinded to the primary outcome, used a standardized abstraction form to obtain patient demographic and clinical data, including the Sequential Organ Failure Assessment (SOFA) scores and comorbidities. Time to first antibiotic and presumed source of infection were extracted from the SSC database. Univariate analyses were performed with Pearson chi-square tests and t-tests. Linear regression was performed with time to first antibiotic as the primary outcome. Covariates, chosen a priori by study authors, included age, race, ethnicity, source of infection, SOFA score, and lactate. Results: A total of 771 patients were included. Women were 45.3% of the sample, the mean age was 66 years (95% confidence interval [CI] = 65.1 to 67.5 years), 19.4% were nonwhite, and 8% were Hispanic. Mean time to first antibiotic was 153 minutes (95% CI = 143 to 163 minutes) for men and 184 minutes (95% CI = 171 to 197 minutes) for women (p < 0.001). The urinary tract was source of infection for 35.2% of women (95% CI = 30.2% to 40.3%) versus 23.7% (95% CI = 19.6% to 27.8%) of men. Pneumonia was present in 46.9% of men (95% CI = 42.1% to 51.7%) versus 35.8% (95% CI = 30.8% to 40.8%) of women. The mean time to antibiotics in women was longer than in men (adjusted odds ratio [aOR] = 1.18, 95% CI = 1.07 to 1.30), even after adjusting for age, race, ethnicity, presumed source of infection, SOFA score, and lactate (p = 0.001). Those with pneumonia compared to other infections received antibiotics faster (aOR = 0.73, 95% CI = 0.66 to 0.81). There was no significant association between other sources of infection and time to antibiotics in either univariate or multivariate analysis. Conclusions: Women experience longer delays to initial antibiotics among patients with severe sepsis or septic shock, even after adjusting for infectious source. Pneumonia was associated with shorter times to antibiotic administration. Future research is necessary to investigate contributors to delayed antibiotic administration in women. ACADEMIC EMERGENCY MEDICINE 2014;21:1499–1502 © 2014 by the Society for Academic Emergency Medicine
S
evere sepsis and septic shock are associated with high mortality rates in both men and women. Similar to conditions such as acute myocardial
infarction,1 women with severe sepsis and septic shock are treated less aggressively compared to men.2 Women are less likely to receive early goal-directed therapy
From the Division of Women’s Health in Emergency Care (TEM), Department of Emergency Medicine (TEM, AMN), Alpert Medical School of Brown University/Rhode Island Hospital, Providence, RI. Received March 11, 2014; revision received April 18, 2014; accepted April 29, 2014. Funding for this project was provided by the Department of Emergency Medicine at Brown University/Rhode Island Hospital. The authors have no relevant financial information or potential conflicts to disclose. Supervising Editor: Shariar Zehtabchi, MD. Address for correspondence and reprints: Tracy E. Madsen, MD; e-mail: [email protected].
© 2014 by the Society for Academic Emergency Medicine doi: 10.1111/acem.12546
ISSN 1069-6563 PII ISSN 1069-6563583
1499 1499
1500
(EGDT)2 and other measures of care, including central venous catheters.3 The Surviving Sepsis Campaign (SSC) bundle, including administration of antibiotics within 3 hours, is associated with decreased sepsis mortality,4 and for every hour in antibiotic delay, mortality increases by 7.6% for those with septic shock.5 Our previous work showed that women are less likely to receive antibiotics within 3 hours,6 but it was unclear whether this difference was due to an increased prevalence of urinary tract infections (UTIs) in women.7 The primary objective of this study was to investigate gender as a predictor of delays to antibiotic administration in severe sepsis or septic shock. Our secondary objective was to investigate infectious source as a contributor to gender disparities in time to antibiotic administration, if such a disparity existed. METHODS Study Design This was a single-center, retrospective, observational cohort study. This study was approved by the institutional review board. Study Setting and Population The study included adults 18 years or older with severe sepsis or septic shock who were seen at a large (>100,000 yearly visits), urban, academic emergency department (ED) between October 2005 and March 2012. Adults admitted to the medical intensive care unit (ICU) with severe sepsis or septic shock8 were entered into the SSC database, becoming part of the national SSC data set, and were included in the study. Patients not admitted to the ICU, and those receiving comfort measures only during the first 24 hours, were excluded. Study Protocol Data regarding arrival time, ED antibiotic administration, and infectious source were entered into the SSC database during the ICU admissions by quality assurance (QA) nurses; data entry procedures are standardized according to the national SSC. The Sequential Organ Failure Assessment (SOFA) score was used as a measure of illness severity.9 Sources of infection, determined by QA nurses using the physicians’ admission notes, were grouped into categories designated by the SSC:4 pneumonia, UTI, abdominal, meningitis, skin, bone, wound, catheter, endocarditis, device, and other. Variables were created for patients with more than one documented source or with unknown sources. Two trained research assistants used standardized data abstraction forms to abstract data pertaining to study covariates from patient charts. For data elements collected by the research assistants, interrater reliability (IRR) testing was performed. This study was conducted using recommended methodologies including abstractor training and monitoring, abstraction forms, regular meetings, and IRR.10 The primary outcome was time to antibiotics (the interval in minutes from ED arrival to administration of
Madsen and Napoli • GENDER AND ANTIBIOTIC DELAYS IN SEPSIS
first antibiotic), consistent with the SSC quality marker. Covariates included race, ethnicity, age, SOFA score, ED lactate, and presumed source of infection. Data Analysis Means, medians, and proportions were used to describe data. t-tests and chi-square tests were used for bivariate analyses. Linear regression was used for multivariate analyses; no data were censored. Model covariates were chosen a priori and entered directly into models; given lack of prior literature on predictors of time to antibiotics in sepsis, potential covariates were based on scientific plausibility and study hypotheses. Models were evaluated using tests of multicollinearity, homoscedasticity, and normality of residuals. Following model diagnostics, the outcome variable time to antibiotics was log-transformed, and model diagnostics were repeated to ensure model fit. Model coefficients were exponentiated and reported as adjusted odds ratios (aOR). The final model included gender, age, race, ethnicity, pneumonia, UTI, abdominal or skin source, SOFA, and lactate. Other sources of infection were not included because of their low frequencies and lack of significance with time to antibiotics. Variables indicating multiple sources or unknown sources did not improve model fit, so were not included in the final model. The familywise Type I error rate was 0.05. Analyses were completed using Stata 12.1. RESULTS A total of 771 patients were included; the mean age was 66.3 years (95% confidence interval [CI] = 65.1 to 67.5 years), and 19.5% were nonwhite. Baseline characteristics including age, race, and ethnicity were similar by gender. Women had slightly lower median SOFA scores (6, interquartile range [IQR] = 4 to 8 vs. 7, IQR = 5 to 9). IRR ranged from 0.82 to 0.97. Sources of infection also differed by gender: fewer women had pneumonia (35.8%, 95% CI = 30.8% to 40.8%) than men (46.9%, 95% CI = 42.1% to 51.7%), but more women had UTIs (35.2%, 95% CI = 30.2% to 40.3%) than men (23.7%, 95% CI = 19.6% to 27.8%). More women had multiple infectious sources (16.3%, 95% CI = 12.5% to 20.2%) than men (11.4%, 95% CI = 8.3% to 14.4%). Frequency of all other sources of infection was similar by gender. Unadjusted, the mean time to antibiotics was 31 minutes longer for women (184 minutes, 95% CI = 171 to 197) than for men (153 minutes, 95% CI = 143 to 163 minutes; p
