RCII-Frailty-Sarcopenia Burden and Incident Chronic Lung Disease: A Prospective Cohort Study of Chinese Middle-Aged and Older Adults Based on CHARLS

<title>Abstract</title> <p> <bold>Background:</bold> The residual cholesterol inflammation index (RCII) integrates remnant cholesterol burden with systemic inflammation. Whether RCII combined with frailty and sarcopenia identifies middle-aged and older adults at increased risk of incident chronic lung disease (CLD) remains unclear. <bold>Methods:</bold> We included CHARLS participants aged 45 years or older, free of self-reported CLD at baseline (2011), with at least one follow-up lung disease assessment through 2020. The main exposure was the RCII-Frailty-Sarcopenia (RCII-FS) burden score, defined as the unweighted count of three baseline components: high RCII (RCII ≥cohort median), frailty (frailty index ≥0.25), and sarcopenia status (0-3 components). Cox proportional hazards models were fitted in three stages. We assessed proportional hazards using Schoenfeld residuals, evaluated restricted cubic splines, multiplicative and additive interactions, Fine-Gray competing-risk models, multiple-imputation sensitivity analysis, calibration, Brier score, decision curves, and bootstrap-corrected prediction metrics. <bold>Results:</bold> Among 7,703 eligible participants, 1,204 incident CLD events occurred over a median follow-up of 8.3 years. Cumulative incidence increased progressively with RCII-FS burden. Compared with participants with no risk components, the fully adjusted HRs were 1.19 (95% CI: 1.04-1.37) for one factor, 1.67 (95% CI: 1.38-2.01) for two factors, and 2.06 (95% CI: 1.39-3.05) for three factors. Modelled ordinally, each additional RCII-FS component was associated with a 28% higher hazard of CLD (HR 1.28, 95% CI: 1.17-1.39; P&lt;0.001). The proportional hazards assumption was not violated (global Schoenfeld test P=0.737). Results were directionally consistent in Fine-Gray competing-risk models (subdistribution HR 1.22, 95% CI: 1.12-1.33), age-spline sensitivity analysis (HR 1.27, 95% CI: 1.17-1.39), shifted-log RCII analysis, and missing-data sensitivity analyses. Adding burden to conventional predictors provided a very small increment in discrimination. <bold>Conclusions:</bold> Higher RCII-FS burden was associated with incident CLD in this CHARLS-derived cohort. The association was directionally consistent across sensitivity analyses, but its predictive increment was smal. Its main value appears to be epidemiologic risk stratification rather than strong stand-alone prediction. </p>