Variations in organism-specific severe sepsis mortality in the United States: 1999-2008

Abstract

Objective: Recent studies have reported decreased overall severe sepsis mortality, but associations with organism trends have not yet been investigated. This study explored organism-specific severe sepsis mortality trends from 1999 to 2008 in a large hospital-based administrative database.

Design: Secondary data analysis using the Nationwide Inpatient Sample.

Setting: United States hospitals sampled in the Nationwide Inpatient Sample dataset.

Patients: This sample approximates a stratified 20% sample of all nonfederal, short-term, general, and specialty hospitals serving adults in the United States. Severe sepsis hospitalizations and organism-specific causes were identified using predetermined International Classification of Diseases, 9th Revision, Clinical Modification codes.

Interventions: None.

Measurements and main results: Analysis was conducted using descriptive Cox proportional hazards and linear regression trend analysis. Adjustments were made for the influence of demographics, comorbidities, number of organisms, and number of organ failures on hospital mortality. The data for 5,033,257 severe sepsis hospitalizations were examined and revealed decreased in-hospital mortality from 40.0% to 27.8% during the study period. The leading cause of severe sepsis was 51.5% Gram-negative bacteria, followed by 45.6% Gram-positive, 1.7% anaerobic, and 1.2% fungal species. The most common Gram-negative organisms were 39.9% Escherichia coli and 17.6% Pseudomonas. Staphylococcus species (62.2% methicillin-sensitive Staphylococcus aureus and 22.6% Streptococcus) were the most commonly reported Gram-positive organisms. Crude mortality estimates were higher for anaerobic and fungal organisms, 34.5% and 31.4%, respectively. Among Gram-positive bacteria, mortality was highest for methicillin-sensitive S. aureus, 30.9%, whereas Pseudomonas was associated with the highest mortality for Gram-negative septicemia cases, 29.5%. After adjusting for covariates, anaerobes were associated with the highest mortality hazard of 1.31 (95% CI, 1.23-1.40). Methicillin-resistant S. aureus had the highest mortality hazard of 1.38 (1.33-1.44) for Gram-positive organisms, whereas all Gram-negative bacteria had decreased mortality hazards.

Conclusions: We not only confirmed an overall decline in severe sepsis mortality from 1999 to 2008 but also identified previously unreported variations in organism-specific severe sepsis mortality. Gram-negative organisms predominate, whereas anaerobes and methicillin-resistant S. aureus are significant predictors of mortality. Future clinical trials exploring new treatments in severe sepsis should incorporate individual organism trends to elucidate potential effect on mortality.