Cardiovascular Diabetology · Nov 10, 2026
Associations of cumulative C-reactive protein–triglyceride glucose index and cumulative functional limitation burden with incident cardiovascular disease among individuals with stages 0–3 cardiovascular–kidney–metabolic syndrome: a nationwide prospective cohort study
By Yu Z, Zhang Z, Zha J et al.
Background Among adults with cardiovascular–kidney–metabolic (CKM) stages 0–3, whether cumulative metabolic-inflammatory burden and self-reported functional limitations provide complementary cardiovascular disease (CVD) prognostic information remains unclear. Methods This China Health and Retirement Longitudinal Study (CHARLS) cohort included 4,048 participants with a 2015 landmark. Cumulative C-reactive protein–triglyceride glucose index (CTI) used 2011 and 2015 measurements; seven-task functional limitation burden used 2011, 2013, and 2015 scores and participant-specific trapezoidal integration. Incident physician-diagnosed heart disease or stroke was assessed in 2018 and 2020. Associations used mutually adjusted Cox and alternative event-time models. Internal prediction used 10-fold cross-validation and 500 full-pipeline bootstrap resamples. Results During follow-up, 778 participants developed CVD. Per 1-SD increment, primary mutually adjusted hazard ratios (HRs) were 1.11 (95% CI 1.03–1.20) for CTI and 1.30 (1.22–1.39) for functional limitation burden. Functional limitation HRs remained 1.30 and 1.31 in complementary log-log and interval-censored models, whereas CTI estimates were attenuated to 1.07 (0.99–1.15; P = 0.073) and 1.07 (1.00–1.16; P = 0.063), respectively. Adding functional limitation burden to conventional factors plus CTI increased the C-index by 0.0208 (95% CI 0.0102–0.0335) and integrated time-dependent area under the curve (iAUC) by 0.0230 (0.0105–0.0389). The integrated Brier score (IBS) difference, evaluated model minus comparator, was − 0.0014 (− 0.0025 to − 0.0005), indicating a small absolute reduction in prediction error. Adding CTI after functional limitation burden showed no clear additional improvement. Conclusions Functional limitation burden showed consistent adjusted associations and internal prediction gains with a small absolute IBS reduction. CTI was event-time sensitive. AUC-specific superiority, clinical net benefit, and improved outcomes were not established. External validation and calibration are needed. Graphical Abstract