Objective: To assess the predictive performance for 1-yr mortality of the previously derived and validated Mortality in Emergency Department Sepsis (MEDS) score.
Design: Prospective cohort study.
Patients: Consecutive adult (aged > or =18 yrs) emergency department patients presenting to an urban, tertiary care, university hospital were eligible if they had a clinically suspected infection as indicated by the decision to obtain a blood culture. The enrollment period was between February 1, 2000, and February 1, 2001. Of 3,926 eligible patient visits, 3,762 (96%) were enrolled and 3,102 unique first visits were analyzed.
Interventions: None.
Measurements and main results: A total of 667 patients (21.5%) died within 1 yr. The unadjusted 1-yr mortality rates for the MEDS risk groups were: very low risk, 7%; low risk, 20%; moderate risk, 37%; high risk, 64%; very high risk, 80%. Using a Cox proportional hazard model that controlled for age, sex, and Charlson co-morbidity index, the 1-yr hazard ratios compared with the baseline very low-risk group were: low risk, 2.2 (1.7-2.9); moderate risk, 3.5 (2.7-4.6); high risk, 6.7 (4.9-9.3); and very high risk, 10.5 (7.2-15.4). The groups were significantly different (p < .0001).
Conclusions: Although the score was initially derived for 28-day in-hospital mortality, our results indicate that the MEDS score also predicts patient survival at 1 yr after index hospital visit with suspected infection. The score needs external validation before widespread use.