The combined C-reactive protein–triglyceride–glucose and frailty index in relation to incident cardiometabolic multimorbidity across cohorts in China and England
Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Background Cardiometabolic multimorbidity (CMM) is a growing burden among middle-aged and older adults. The C-reactive protein–triglyceride–glucose index (CTI) reflects inflammatory–metabolic dysfunction, whereas the frailty index (FI) captures physiological vulnerability. Whether CTI-FI is consistently associated with incident CMM and adds predictive value across populations remains uncertain. We evaluated associations of CTI-FI with incident CMM and its incremental predictive value. Methods Adults aged ≥ 45 years were analyzed in the China Health and Retirement Longitudinal Study (CHARLS; 2011–2020) and the English Longitudinal Study of Ageing (ELSA; 2004–2018). CTI-FI and cumulative CTI-FI (cumCTI-FI) were evaluated in separate cohorts. Adjusted cumulative incidence curves, Cox regression, and restricted cubic splines assessed associations; C-statistics and reclassification measures quantified added predictive value. Predictive performance was compared across eight machine-learning models developed in CHARLS: CatBoost, XGBoost, Extra Trees, random forest, recursive partitioning, support vector machine, k-nearest neighbors, and neural network; the selected models were externally validated in ELSA. Subgroup and sensitivity analyses assessed consistency across participant groups and analytic choices. Results Among 7,108 participants, adjusted cumulative incidence increased across CTI-FI quartiles in both cohorts (all P < 0.001). Per-unit adjusted HRs for CTI-FI were 1.48 (95% CI, 1.37–1.60) in CHARLS and 1.31 (1.17–1.47) in ELSA; Q4-versus-Q1 HRs were 3.40 and 2.62. Nonlinearity was observed for CTI-FI and cumCTI-FI in CHARLS and for cumCTI-FI in ELSA. In CHARLS, the CTI-FI association was stronger among participants aged < 60 years, married, or without hypertension. Adding CTI-FI increased C-statistics from 0.723 to 0.759 in CHARLS and 0.745 to 0.767 in ELSA; NRI improved significantly in both cohorts (0.40 and 0.35), and IDI improved significantly in CHARLS. cumCTI-FI yielded the highest C-statistics (0.763 and 0.768), and CTI-FI showed greater discrimination than CTI or FI. XGBoost had the highest mean cross-validated AUC (0.700); external AUCs were 0.689 and 0.695. SHAP ranked CTI-FI and cumCTI-FI as leading predictors. Conclusions CTI-FI was associated with incident CMM and improved prediction beyond conventional factors in both cohorts. By integrating inflammatory–metabolic dysfunction and frailty-related vulnerability, CTI-FI supports multidimensional risk assessment. Repeated assessment may strengthen early identification and targeted CMM prevention.
Source: Research Square
Language: English
Subfield: Epidemiology
Field: Medicine
Domain: Health Sciences
Open Access status: green
Awards: G4546817175, G6330284575
OpenAlex ID: W7225905385
