Objective: The optimal prognostic model for community-acquired pneumonia (CAP) remains unclear. In this study, we sought to identify independent predictors of 30-day mortality in patients with CAP and to determine whether adding specific prognostic factors to each of the two clinical prediction scores could improve the prognostic yield.
Methods: This retrospective study involved 797 CAP patients who had been hospitalized at a tertiary referral center. The patients were categorized into two groups: those who survived and those who had died on or before 30 days after admission. Select clinical parameters were then compared between the two groups.
Results: During the 30-day period, there were 72 deaths (9%). We constructed two models for a multivariate analysis: one was based on a high CURB-65 score (3-5) and the other on a high pneumonia severity index (PSI) class (V). In both models, a high CURB-65 score or a high PSI class, along with the presence of dyspnea, high Eastern Cooperative Oncology Group (ECOG) performance status (3-4), and a low serum albumin level, were independent predictors of 30-day mortality. In both the CURB-65-based and PSI-based models, the addition of dyspnea, high ECOG performance status, and hypoalbuminemia (<3 g/dL) enhanced the prognostic assessment, and subsequently, the c-statistics calculated with the use of three- or four- predictor combinations exceeded 0.8.
Conclusion: In addition to the CURB-65 or PSI, the clinical factors of dyspnea, the ECOG performance status, and serum albumin level may be independent predictors of 30-day mortality in CAP patients. When combined with the CURB-65 or PSI, these parameters provide additional evidence for predicting poor prognoses.