Nomogram for prediction of severe community-acquired pneumonia development in diabetic patients: a multicenter study

Abstract

BackgroundDiabetic patients with community-acquired pneumonia (CAP) have an increased risk of progressing to severe CAP. It is essential to develop predictive tools at the onset of the disease for early identification and intervention. This study aimed to develop and validate a clinical feature-based nomogram to identify diabetic patients with CAP at risk of developing severe CAP.MethodA retrospective cohort study was conducted between January 2019 to December 2020. 1026 patients with CAP admitted in 48 hospitals in Shanghai were enrolled. All included patients were randomly divided into the training and validation samples with a ratio of 7:3. The nomogram for the prediction of severe CAP development was established based on the results of the multivariate logistic regression analysis and other predictors with clinical relevance. The nomogram was then assessed using receiver operating characteristic curves (ROC), calibration curve, and decision curve analysis (DCA).ResultsMultivariate analysis showed that chronic kidney dysfunction, malignant tumor, abnormal neutrophil count, abnormal lymphocyte count, decreased serum albumin level, and increased HbA1c level at admission was independently associated with progression to severe CAP in diabetic patients. A nomogram was established based on these above risk factors and other predictors with clinical relevance. The area under the curve (AUC) of the nomogram was 0.87 (95% CI 0.83–0.90) in the training set and 0.84 (95% CI 0.78–0.90). The calibration curve showed excellent agreement between the predicted possibility by the nomogram and the actual observation. The decision curve analysis indicated that the nomogram was applicable with a wide range of threshold probabilities due to the net benefit.ConclusionOur nomogram can be applied to estimate early the probabilities of severe CAP development in diabetic patients with CAP, which has good prediction accuracy and discrimination abilities. Since included biomarkers are common, our findings may be performed well in clinical practice and improve the early management of diabetic patients with CAP.