Risk factors and prediction model for chronic bacterial infection in stable bronchiectasis in Shanghai, China.

BACKGROUND: Chronic bacterial infection (CBI) represents a key feature in patients with bronchiectasis. Therefore, it is of great clinical significance to develop an effective nomogram model for predicting the risk of CBI in stable bronchiectasis, which guides individualized clinical treatment strategies. METHODS: The study enrolled patients in stable bronchiectasis in Shanghai between January 2020 and December 2024. They were categorized into two groups of CBI and without CBI. We used Univariate logistic analysis, LASSO regression and Multivariate logistic analysis to identify predictors associated with CBI. Based on the screened-out risk factors, a nomogram was constructed to predict the risk of CBI in adults with stable bronchiectasis. We used receiver operating characteristic, the area under the curve (AUC) and calibration curve to determine the predictive accuracy and discriminability of nomogram. The decision curve analysis (DCA) was employed to further confirm the clinical effectiveness of nomogram. RESULTS: Multivariate logistic analysis revealed the risk factors of CBI included history of smoking, number of lobes affected ≥3, number of exacerbation in the prior year ≥3, history of hemoptysis in the prior year, CRP, CD3+CD4+T-cell count <500 cells/μL. The AUC was 0.861 (95% CI: 0.825-0.897). We developed a nomogram model. Based on AUC, Hosmer-Lemeshow goodness-of-fit test (p = 0.794), calibration curve, and DCA, we conducted that the model exhibits excellent predictive accuracy, discriminability and clinical effectiveness. CONCLUSION: The nomogram model demonstrated satisfactory discrimination and calibration accuracy, enabling screen patients with stable bronchiectasis at high risk of CBI and facilitating individualized clinical decisions in future clinical practice.