Sci Rep. 2026 Jul 9. doi: 10.1038/s41598-026-57057-5. Online ahead of print.
ABSTRACT
Cardiovascular-metabolic multimorbidity (CMM) has emerged as a defining clinical and public health challenge of ageing societies, characterised by exponential increases in all-cause mortality and healthcare expenditure. Although the triglyceride-glucose index weighted by waist-to-height ratio (TyG-WHtR) has been validated as a robust composite marker of insulin resistance and central adiposity, existing evidence relies predominantly on static, single-timepoint measurements. This cross-sectional paradigm is inherently unable to capture the cumulative burden of metabolic impairment or phenotypic drift over time. Furthermore, in middle-aged and older populations, the high incidence of all-cause mortality constitutes a substantial competing risk that may bias conventional risk estimates. Whether longitudinal TyG-WHtR trajectories-reflecting the velocity and directionality of metabolic deterioration or recovery-confer incremental prognostic value for CMM within a rigorous competing-risks framework remains unexplored. This study aimed to characterise longitudinal TyG-WHtR trajectory patterns over a four-year exposure window in a nationally representative sample of middle-aged and older Chinese adults, and to evaluate the independent associations between these trajectory patterns and incident CMM risk over a subsequent five-year follow-up period. Specifically, we sought to determine: (1) whether sustained high metabolic burden or deteriorating trajectories confer disproportionately elevated CMM risk; and (2) whether substantial TyG-WHtR reduction (trajectory improvement) among individuals with high baseline burden attenuates the risk of disease progression. Selection bias was rigorously addressed through stabilised inverse probability weighting, and the influence of all-cause mortality on the primary outcome was accounted for within a competing-risks framework. Drawing on the nationally representative, prospective China Health and Retirement Longitudinal Study (CHARLS), we employed a sequential dual-window design to establish the causal temporal ordering of exposure and outcome. The exposure window (2011-2015) utilised data from three consecutive follow-up waves to characterise individual TyG-WHtR trajectories. To transcend the empirical limitations of conventional quantile-based classification, we derived a reference change value (RCV = 6.27%) from biological variation data as an objective classification threshold. Individual trajectories were thereby categorised as improving (relative change < - 6.4%), stable (absolute change ≤ 6.4%), or worsening (relative change > 6.4%), and were further cross-classified with baseline metabolic burden to construct six dynamic exposure patterns. During the outcome window (2015-2020), 3,526 participants free of target conditions at baseline were followed for incident CMM, defined as the new onset of two or more conditions among hypertension, diabetes, cardiac disease, and stroke. To address the survival data structure inherent to an older cohort, we applied both Fine-Gray subdistribution hazard models and cause-specific Cox regression in a dual-modelling framework to adjust for the competing risk of death. The statistical framework simultaneously incorporated stabilised inverse probability of censoring weighting (IPCW) and multiple imputation by chained equations (MICE) to control for non-random attrition and missing covariate bias. Restricted cubic splines (RCS) were used to characterise dose-response relationships and assess non-linearity, and E-values were computed to quantify the sensitivity of findings to unmeasured confounding. Among the 3,526 participants included, six distinct TyG-WHtR longitudinal trajectory patterns were identified. Compared with the low-burden stable group (LL-S), the high-burden worsening group (HH-W) exhibited the highest independent risk of incident CMM (HR 2.36, 95% CI 1.77-3.15). High baseline metabolic burden demonstrated a pronounced metabolic "legacy effect": even among participants whose index improved during follow-up (HH-I), long-term risk remained significantly elevated (HR 1.54), whereas sustained low baseline burden conferred a protective buffering effect against subsequent metabolic fluctuations. After rigorous adjustment for the competing risk of all-cause mortality, the observed associations remained highly robust (subdistribution HR 2.36; cause-specific HR 2.39), effectively excluding survival bias. Dose-response analyses identified a TyG-WHtR value of 3.16 as the clinical inflection point beyond which risk increases non-linearly. Notably, the risk-amplifying effect of the HH-W trajectory was particularly pronounced among participants with normal baseline body mass index (BMI < 24 kg/m²; HR 2.91), suggesting that TyG-WHtR captures early cardiometabolic risk independently of conventional general adiposity measures. This study establishes that the dynamic longitudinal trajectory of the TyG-WHtR index is an independent predictor of incident CMM in middle-aged and older adults. Compared with individuals whose index remained stable, a worsening TyG-WHtR trajectory (RCV > 6.4%) was associated with a significantly elevated risk of incident CMM, an association that remained robust after accounting for the competing risk of death and non-random attrition bias. These findings underscore the superior prognostic utility of monitoring longitudinal TyG-WHtR trajectories over single baseline measurements for identifying individuals at high risk of CMM. In clinical and public health practice, early intervention targeting abnormal metabolic fluctuations beyond the defined RCV threshold warrants prioritisation to decelerate the progression toward multimorbidity.
PMID:42426049 | DOI:10.1038/s41598-026-57057-5