Diabetes Research and Clinical Practice · Nov 11, 2026
Joint triglyceride–glucose index and single-point insulin sensitivity estimator phenotypes and diabetes outcomes across three populations
By Qi H, Chen N, Long M et al.
Aims To assess joint triglyceride–glucose index (TyG) and single-point insulin sensitivity estimator (SPISE) phenotypes and their incremental information for diabetes-related outcomes. Methods We analyzed participants without baseline diabetes from the China Health and Retirement Longitudinal Study (CHARLS; n = 4,933) and English Longitudinal Study of Ageing (ELSA; wave 6, n = 2,703; wave 4 sensitivity), with the National Health and Nutrition Examination Survey (NHANES; n = 10,080) as secondary cross-sectional evidence. Sample medians and literature-based fixed thresholds defined phenotypes. Adjusted logistic, Cox, and survey-weighted logistic models assessed associations and incremental fit. Results High TyG/low SPISE had the largest median-based estimates: CHARLS odds ratio 2.65 (95% confidence interval 2.05–3.45); ELSA hazard ratios 3.57 (2.04–6.25) and 6.41 (3.98–10.33); NHANES odds ratios 6.61 (5.40–8.08) for prevalent diabetes and 4.05 (3.47–4.72) for dysglycemia. Fixed-threshold associations retained the adverse direction. SPISE improved fit beyond TyG, but adding TyG and interaction to SPISE did not improve CHARLS fit (P = 0.251). Conclusions High TyG/low SPISE was associated with diabetes-related outcomes across the three populations. SPISE provided incremental information beyond TyG, with gains varying by comparator and study design.
doi.org/10.1016/j.diabres.2026.113657Open accessPubMed 43014420Biomarkers