Prognosis of patients hospitalized with community-acquired pneumonia

Our purpose was to determine which clinical features predict short-term mortality in patients with community-acquired pneumonia.

We conducted a prospective multicenter study of 347 patients hospitalized in Pittsburgh (the derivation cohort) and 253 hospitalized and ambulatory patients in Boston (the validation cohort) with clinical and radiographic evidence of pneumonia. Patients in the derivation cohort underwent an extensive microbiologic evaluation including bacteriologic sputum culture, blood cultures, direct fluorescent antibody testing for Legionella species, and serologic testing for Mycoplasma pneumoniae, Legionella species, and Chlamydia TWAR.

The overall mortality was 18% in the derivation cohort and 13.2% in the validation cohort. We identified five independent predictors of mortality in the derivation cohort: pleuritic chest pain (risk ratio, 0.4; 95% confidence interval [CI], 0.17 to 0.99), mental status changes (risk ratio, 2.6; 95% CI, 1.4 to 4.6), a severe vital sign abnormality (risk ratio, 2.1; 95% CI 1.2 to 3.6), neoplastic disease (risk ratio, 5.0; 95% CI, 2.7 to 9.1), and "high-risk" pneumonia etiology (risk ratio, 2.8; 95% CI, 1.6 to 5.0). A mortality index based on these factors accurately classified patients into five risk classes of increasing mortality. In the derivation cohort, the 6-week mortality rates were 0% in class I, 2.9% in class II, 13.1% in class III, 32.7% in class IV, and 89.5% in class V. There was little deterioration in the predictive accuracy of the model when tested in the validation cohort: mortality was 2.2% in class I, 0% in class II, 13.5% in class III, 33.3% in class IV, and 55.6% in class V.

This prognostic classification may help direct triage decisions, assess appropriateness of care, and guide the design and analysis of therapeutic trials in patients with community-acquired pneumonia.