Predictive Factors Associated with In-Hospital Mortality for Patients Across the Sepsis Spectrum

Introduction Sepsis poses a large burden for hospitals across the world owing to its high resource utilization and poor patient outcomes. This project aimed to assess and quantify the importance of sepsis-specific predictors for in-hospital mortality. Several studies have identified comorbidities and clinical laboratory variables associated with mortality among the severe sepsis population, but few have combined the 2 in making a model specific for all sepsis patients. Methods We used clinical and laboratory variables seen at the time of first lactic acid measurement and patient characteristics upon admission to develop a logistic regression model to predict mortality in a sepsis population at 2 area hospitals in Omaha, Nebraska. Results Overall, sepsis-related mortality was 12.9%. Advanced age, a higher number of comorbidities, and leucopenia were positively associated with mortality, whereas preexisting diabetes was negatively associated with mortality. Each 1 mmol/L difference in lactic acid levels was positively associated with 1.38 times higher adjusted odds (95% confidence interval, 1.30-1.48) of mortality after adjustment for other covariates. For every 1% decrease in hematocrit, the adjusted odds of mortality were 1.04 times higher (95% confidence interval, 1.02-1.06). Conclusions We identified a number of sepsis mortality predictors. The negative association between preexisting diabetes mellitus and sepsis mortality requires further investigation.