Background: Patients with community-acquired pneumonia (CAP) comprised 35.6% of the overall phase 3 Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis (PROWESS) study and 33.1% of the placebo arm. We investigated the use of CURB-65, the Pneumonia Severity Index (PSI), and Acute Physiology and Chronic Health Evaluation II (APACHE II) prediction scores to identify the CAP population from the PROWESS placebo arm at the greatest mortality risk. Methods: Patients were classified as having CAP if the lung was the primary infection site and the patient originated from home. The abilities of CURB-65, PSI, and APACHE II scores to determine the 28-day and in-hospital mortality were compared using receiver operator characteristic (ROC) curves and the associated areas under the curve. Results: PROWESS enrolled 278 patients with CAP in the placebo arm. The areas under the ROC curves for PSI = 5, CURB-65 ≥3, and APACHE II ≥25 for predicting 28-day (c = 0.65, 0.66, and 0.64, respectively) and in-hospital mortality (c = 0.65, 0.65, and 0.64, respectively) were not statistically different from each other. The 28-day mortality of patients with a PSI score of 5, CURB-65 ≥3, and APACHE II ≥25 was 41.6%, 37.9%, and 43.5%, respectively. Conclusions: Despite early diagnosis and appropriate antibiotic therapy, conventionally treated CAP with PSI = 5, CURB-65 ≥3, or APACHE II ≥25 has an unacceptably high mortality. In this study, PSI, CURB-65, and APACHE II scoring systems perform similarly in predicting the 28-day and in-hospital mortality; however, differences in the categorization of severe CAP were observed and there was a significant mortality in patients with a CURB-65 <3 and PSI <5. © SAGE Publications 2011.
