Non-traditional lipid indices beyond the C-reactive protein-triglyceride glucose index and incident cardiovascular disease: a 2015 landmark analysis in China Health and Retirement Longitudinal Study (CHARLS) participants with operationally defined cardiovascular-kidney-metabolic syndrome (CKM) stages 0–3

The C-reactive protein-triglyceride-glucose index (CTI) integrates inflammatory and insulin-resistance-related information, whereas non-traditional lipid parameters (nTLPs) summarize complementary features of atherogenic lipid metabolism. Whether nTLPs provide information beyond CTI when repeated measurements, shared components, selection, and predictive optimism are considered remains incompletely characterized. We performed a 2015 landmark analysis in CHARLS participants who were free of reported clinical CVD through the landmark and classified within operational CKM stages 0–3. CTI was evaluated alongside the atherogenic index of plasma (AIP), non-HDL cholesterol, Castelli risk index I (CRI-I), Castelli risk index II (CRI-II), and the lipoprotein combine index (LCI). In the revised analysis, the primary inferential family comprised five time-weighted dual-high versus dual-low contrasts for incident CVD with Benjamini-Hochberg false-discovery-rate control. We also compared alternative exposure summaries, decomposed CTI into TyG and hs-CRP components, quantified the incremental contribution of each nTLP beyond CTI, and examined competing risk, interval ascertainment, selection, censoring, missing data, calibration, and internally validated prediction. Of 17,708 Wave-1 respondents, 4,420 met the landmark, paired-biomarker, covariate, and follow-up requirements; 827 developed reported incident CVD. The mean intermeasurement interval was 3.98 ± 0.10 years. CTI showed similar associations when represented by the 2011 value (HR per SD 1.14, 95% CI 1.07–1.23), 2015 value (1.14, 1.06–1.23), two-visit mean (1.17, 1.09–1.25), or time-weighted summary (1.19, 1.11–1.27), whereas the change score was not clearly associated with CVD. Across the five primary dual-high contrasts, adjusted HRs ranged from 1.45 to 1.64 and remained significant after FDR correction; competing-risk and post-landmark censoring-weighted estimates were similar. Formal interaction was generally limited. In mutually adjusted models, CRI-II retained the clearest association beyond CTI (HR per SD 1.13, 95% CI 1.05–1.21; FDR P = 0.003). Against a clinical benchmark, CTI and its directly estimated components produced very similar optimism-corrected discrimination (C-index approximately 0.635–0.636). Higher CTI and more atherogenic lipid profiles identified a higher-risk phenotype in this selected 2015 landmark cohort. The elevated hazards in dual-high groups appear to reflect concurrent metabolic-inflammatory and lipid burden more than statistical interaction, while CRI-II retained modest independent information beyond CTI. The two-visit time-weighted metric behaved similarly to the two-visit mean, and internally validated prediction gains were small. These findings support a component-oriented interpretation of CTI-nTLP phenotypes and provide a basis for evaluation in cohorts with adjudicated outcomes and external validation. Not applicable.