Prospective comparison of three validated prediction rules for prognosis in community-acquired pneumonia

Abstract Purpose We assessed the performance of 3 validated prognostic rules in predicting 30-day mortality in community-acquired pneumonia: the 20 variable Pneumonia Severity Index and the easier to calculate CURB (confusion, urea nitrogen, respiratory rate, blood pressure) and CURB-65 severity scores. Subjects and methods We prospectively followed 3181 patients with community-acquired pneumonia from 32 hospital emergency departments (January–December 2001) and assessed mortality 30 days after initial presentation. Patients were stratified into Pneumonia Severity Index risk classes (I–V) and CURB (0–4) and CURB-65 (0–5) risk strata. We compared the discriminatory power (area under the receiver operating characteristic curve) of these rules to predict mortality and their accuracy based on sensitivity, specificity, predictive values, and likelihood ratios. Results The Pneumonia Severity Index (risk classes I–III) classified a greater proportion of patients as low risk (68% [2152/3181]) than either a CURB score <1 (51% [1635/3181]) or a CURB-65 score <2 (61% [1952/3181]). Low-risk patients identified based on the Pneumonia Severity Index had a slightly lower mortality (1.4% [31/2152]) than patients classified as low-risk based on the CURB (1.7% [28/1635]) or the CURB-65 (1.7% [33/1952]). The area under the receiver operating characteristic curve was higher for the Pneumonia Severity Index (0.81) than for either the CURB (0.73) or CURB-65 (0.76) scores (P<0.001, for each pairwise comparison). At comparable cut-points, the Pneumonia Severity Index had a higher sensitivity and a somewhat higher negative predictive value for mortality than either CURB score. Conclusions The more complex Pneumonia Severity Index has a higher disciminatory power for short-term mortality, defines a greater proportion of patients at low risk, and is slightly more accurate in identifying patients at low risk than either CURB score.