Background The prognostic role of N-terminal pro-brain natriuretic peptide (NT-proBNP) in patients with community-acquired pneumonia (CAP) has not been evaluated. The aim of the present study was to investigate whether NT-proBNP level could predict mortality in hospitalised CAP patients.

Methods We performed a structured medical record review of all hospitalised CAP patients from May 2003 to October 2006, and classified patients into the 30-day survival and non-survival group. Data included demographic and clinical characteristics, and laboratory findings including NT-proBNP levels. The APACHE II scores, PSI (pneumonia severity index) and CURB65 (confusion, urea, respiratory rate, blood pressure and aged 65 or more) scores were calculated. Comparisons between survivors and non-survivors were made with χ², non-parametric tests and logistic regression and ROC analysis were used to compare the ability of NT-proBNP (adjusted for age, heart failure and creatinine), APACHE II, PSI and CURB65 to predict mortality.

Results Of 502 patients, 61 (12.2%) died within 30 days. NT-proBNP levels were measured in 167 patients and were significantly higher in non-survivors compared to survivors (median 841.7 (IQR 267.1–3137.3) pg/ml vs 3658.0 (1863.0–7025.0) pg/ml, p=0.019). NT-proBNP was an independent predictor of mortality (adjusted OR 1.53; 95% CI 1.16 to 2.02, p=0.002). The AUC for NT-proBNP was 0.712 (95% CI, 0.613 to 0.812), which was comparable to those of PSI (0.749, p=0.531) and CURB65 (0.698, p=0.693), but inferior to that of APACHE II (0.831, p=0.037). Adding NT-proBNP to APACHE II, PSI and CURB65 did not significantly increase the AUCs, respectively.

Conclusions NT-proBNP level is an independent predictor of mortality in hospitalised CAP patients. The performance of NT-proBNP level is comparable to those of PSI and CURB65 in predicting mortality.