Development and validation of a risk prediction model for respiratory failure in AECOPD patients

Objective To explore the risk factors for respiratory failure (RF) in patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD), construct a prediction model, and evaluate its effectiveness. Methods Retrospective analysis was conducted on the clinical data of 460 AECOPD patients who were admitted to Department of Respiratory and Critical Care Medicine, the Affiliated Hospital of Xuzhou Medical University, from April 2023 to August 2024. According to admission time, the patients were divided into two groups in a 7:3 ratio: a training group (n=323, April 2023 to March 2024) and a validation group (n=137, April 2024 to August 2024). Based on whether respiratory failure occurred, the training group was further divided into an AECOPD group (n=147) and an RF group (n=176). Univariate and multivariate logistic regression analyses were performed to identify independent risk factors, and a nomogram model was constructed. The model's performance and clinical utility were assessed through plotting receiver operating characteristic (ROC) curves to calculate the area under the curve (AUC) and calibration curves, decision curve analysis (DCA), and clinical impact curves (CIC) analysis. Results Diabetes, red blood cell distribution width standard deviation (RDW-SD), prognostic nutritional index (PNI), and neutrophil percentage to serum albumin ratio (NPAR) were identified as independent risk factors for RF in AECOPD patients (P≤0.05). A prediction model was constructed based on these four factors, and a nomogram was plotted. ROC curve analysis showed that the AUC in the training group was 0.839 (95%CI: 0.797-0.881), and 0.806 (95%CI: 0.734-0.878) in the validation group. The calibration curves for both groups were approximated to the ideal curve, and the Hosmer-Lemeshow test (P>0.05) indicated a good consistency between predicted and actual results. Decision curve and clinical impact curves demonstrated that the model has good clinical utility. Conclusions The nomogram constructed based on diabetes, RDW-SD, PNI, and NPAR demonstrates good accuracy and clinical utility in predicting RF. It is a simple and cost-effective method suitable for primary hospitals to assess the risk of RF in AECOPD patients.