Predictive accuracy and feasibility of risk stratification scores for 28-day mortality of patients with sepsis in an emergency department.

Abstract

Sepsis is associated with high mortality. Because early therapy has proven to decrease mortality, a risk stratification tool that quickly and easily quantifies mortality risk of patients will be helpful to guide appropriate treatment. We investigated five scores in terms of (a) predicting 28-day mortality and (b) their feasibility for use in the emergency department (ED). We carried out a historical cohort study in the ED of Maastricht University Medical Centre (MUMC). Patients who fulfilled the criteria for sepsis were included if they had been admitted to the hospital by an internist between August 2009 and August 2010. The Mortality in Emergency Department Sepsis (MEDS), Confusion, Urea, Respiratory rate, Blood pressure, age>65 (CURB-65), Acute Physiology And Chronic Health Evaluation II (APACHE II), Rapid Acute Physiology Score (RAPS), and Rapid Emergency Medicine Score (REMS) scores were calculated using ED charts. The primary outcome was total 28-day mortality. Receiver operating characteristic curves and calibration plots were constructed to evaluate predictive accuracy. Feasibility was defined as the proportion of patients for whom all data were available. We included 600 patients, of whom 90 (15%) died within 28 days. Discriminating ability for total 28-day mortality of the MEDS [area under the curve (AUC): 0.82, 95% confidence interval (CI) 0.78-0.87], CURB-65 (AUC: 0.78, 95% CI 0.73-0.83), and APACHE II (AUC: 0.71, 95% CI 0.64-0.79) was the highest, but only the difference between the MEDS and REMS (P=0.007) and the RAPS score (P<0.001) was significant. Both the MEDS and the CURB-65 had higher AUCs for predicting 28-day in-hospital mortality than the other three scores, but this was only significant for the MEDS score compared with the RAPS (P=0.003). Both the MEDS and the CURB-65 underestimated mortality, especially for the higher scores. The MEDS, CURB-65, REMS, and RAPS were most feasible as they could be calculated in more than 96% of patients. The MEDS and CURB-65 scores are the most adequate and feasible tools for the prediction of total 28-day mortality in septic patients presenting at the ED, but they need local recalibration before use in the ED.