Abstract
OBJECTIVE: The aim of this study was to develop a tool for predicting in-hospital mortality of community-acquired pneumonia (CAP) in patients with type 2 diabetes (T2DM). METHODS: A retrospective study was conducted on 531 CAP patients with T2DM at The First Hospital of Qinhuangdao. The primary outcome was in-hospital mortality. Variables to develop the nomogram were selected using multiple logistic regression analysis. Discrimination was evaluated using receiver operating characteristic (ROC) curve. Calibration was evaluated using the Hosmer-Lemeshow test and calibration plot. RESULTS: Multiple logistic regression analysis showed that age, pulse, urea and albumin (APUA) were independent risk predictors. Based on these results, we developed a nomogram (APUA model) for predicting in-hospital mortality of CAP in T2DM patients. In the training set, the area under the curve (AUC) of the APUA model was 0.814 (95% CI: 0.770-0.853), which was higher than the AUCs of albumin alone, CURB-65 and Pneumonia Severity Index (PSI) class (p<0.05). The Hosmer-Lemeshow test (χ (2)=5.298, p=0.808) and calibration plot (p=0.802) showed excellent agreement between the predicted possibility and the actual observation in the APUA model. The results of the validation set were similar to those of the training set. CONCLUSION: The APUA model is a simple and accurate tool for predicting in-hospital mortality of CAP, adapted for patients with T2DM. The predictive performance of the APUA model was better than CURB-65 and PSI class.