Association between serum chloride and all-cause mortality in bronchiectasis: a retrospective cohort study and predictive model establishment based on machine learning

Bronchiectasis remains a chronic respiratory condition with substantial disease burden. While chloride dysregulation plays a role in its pathogenesis, the effect of serum chloride on clinical outcomes remains unclear. This study aims to investigate the association between serum chloride and all-cause mortality in bronchiectasis, and further develop machine learning-based prediction models. Data were extracted from the MIMIC-IV (v2.2) database, and a total of 1,520 participants were included and stratified by serum chloride concentration percentiles. The primary and the secondary outcomes were 365-day and 28-day all-cause mortality, respectively. The association between serum chloride and outcomes was explored using Kaplan-Meier curves, Cox regression analyses, and restricted cubic splines (RCS) analysis. Subgroup analyses were performed to assess the robustness of the findings. Furthermore, prediction models were established based on machine learning algorithms. RCS analysis revealed a significant U-shaped association between serum chloride and all-cause mortality ( P < 0.001), with both extremes of serum chloride concentration correlated with increased mortality risk. Multivariable Cox regression demonstrated that patients in Q2, with serum chloride concentration of 100–104 mmol/L, had the lowest risk of mortality (28-day adjusted hazard ratio [HR] 0.42, 95% CI 0.25–0.68, P < 0.001; 365-day adjusted HR 0.49, 95% CI 0.38–0.63, P < 0.001). The Random Forest algorithm achieved superior predictive performance with serum chloride emerging as a top-ranked important feature. Serum chloride is significantly correlated with all-cause mortality in bronchiectasis, it may serve as a valuable biomarker in clinical risk stratification and prognostic assessment.