ObjectiveTo evaluate the predictive value of serum C-reactive protein (CRP) for hospital death events in patients with severe community-acquired pneumonia (CAP). MethodsThe clinical data of 202 patients with CAP in Changhai Hospital and Second People’s Hospital of Wuxi between Sep. 2006 and Sep. 2010 were retrospectively reviewed. The clinical and laboratory parameters, including the serum CRP level, white blood cell count, erythrocyte sedimentation rate (ESR) and serum creatine concentration were collected from Hospital Information System (HIS) and Laboratory Information System (LIS). The patients were divided into two groups according to the final death (CAP related complications) or survival of patients in the hospital. The receiver operating curve (ROC) analysis and multivariable logistical model were used to assess the predictive value of CRP on hospital death events. ResultsThe median (interquartile range) serum CRP level of survival patients and patients who died during the hospital stay were 167.00(132.50,208.50) mg/L and 327.00(246.25, 411.50) mg/L, respectively (Z=-7.481，P<0.001). ROC analysis showed that CRP was an effective predictor for hospital death of CAP patients, with the area under curve (AUC) being 0.85 (95%CI: 0.78-0.91). The optimal cot-off value for serum CRP was 230.50 mg/L, with the sensitivity being 0.83(95%CI: 0.76-0.89) and specificity being 0.79(95%CI: 0.65-0.89). Logistic regression analysis showed that, after adjusted for age, serum creatinine and ESR, CRP on admission was still independently associated with hospital death of CAP patients (OR=13.42, P<0.01). ConclusionIncreased CRP is an independent risk factor for hospital death events in patients with severe CAP.