Association of the triglyceride-glucose index/high-density lipoprotein cholesterol ratio, a novel insulin resistance marker, with incident diabetes mellitus: a large-scale retrospective cohort study

Objective Data on the relationship between the triglyceride-glucose index/high-density lipoprotein cholesterol ratio (TyGHR) and diabetes mellitus (DM) onset remain limited. This study investigated the association between TyGHR and incident DM and assessed its predictive utility. Methods A total of 8,789 adults who received routine health examinations at Kuichong People’s Hospital, Shenzhen, between 2018 and 2023 were included in this retrospective cohort analysis. The association between TyGHR and incident DM was assessed using Cox proportional hazards models. Potential dose-response patterns were explored using restricted cubic splines. The robustness of the findings was examined through stratified and sensitivity analyses. Conventional receiver operating characteristic (ROC) curves and time-dependent ROC analyses were used to evaluate the predictive performance of TyGHR for incident DM over a follow-up period of up to 5 years. Results After adjustment for potential confounders, TyGHR was positively associated with incident DM (HR = 1.108, 95% CI: 1.046–1.173). A nonlinear association was observed, with an estimated statistical threshold of 7.27. Below this threshold, each one-unit increase in TyGHR was associated with a higher risk of incident DM (HR = 1.296, 95% CI: 1.103–1.524). Time-dependent ROC analysis showed relatively consistent predictive performance from 2.0 to 5.0 years, with AUCs ranging from 0.7351 to 0.7402. Sensitivity analyses yielded similar results. Conclusion TyGHR was independently associated with incident DM and showed a nonlinear pattern with an estimated threshold of approximately 7.27. It also demonstrated relatively stable predictive capacity over 2 to 5 years of follow-up, suggesting its potential value as a simple indicator for early DM risk stratification and prevention.