Evaluation of a new TyG indicator, TyHGB, in predicting non-alcoholic fatty liver disease: evidence from two independent populations.

BACKGROUND: As non-alcoholic fatty liver disease (NAFLD) becomes increasingly common and affects population health, simple and effective screening tools for NAFLD are essential. The triglyceride high-density cholesterol-glucose body index (TyHGB), a novel metabolic index, has shown promise for predicting metabolic disorders. The objective of this research was to assess TyHGB's predictive capability for NAFLD across two distinct cohorts. METHODS: This retrospective study utilized data obtained from two independent cohorts: a Chinese hospital cohort (n = 181,241) and the National Health and Nutrition Examination Survey (NHANES) cohort (n = 3,286). TyHGB was computed according to the following equation: triglyceride level/high-density lipoprotein cholesterol level + 0.7 × fasting blood glucose level (mmol/L) + 0.1 × body mass index (kg/m²). NAFLD was diagnosed using ultrasonography. The TyHGB-NAFLD association was evaluated using multivariable logistic regression analysis, restricted cubic spline methodology, and receiver operating characteristic (ROC) curves. To establish the added predictive capacity of TyHGB over the triglyceride-glucose (TyG) index, net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were determined. RESULTS: In both cohorts, participants with higher TyHGB values demonstrated a significantly higher prevalence of NAFLD. After full adjustment for potential confounders, the odds ratios for NAFLD on comparing the highest versus lowest quartiles of TyHGB were 12.22 (95% confidence interval [CI]: 10.73-13.93) in the Chinese cohort and 3.17 (95% CI: 2.25-4.46) in the NHANES cohort. Restricted cubic spline modeling demonstrated significant nonlinear associations between TyHGB values and NAFLD risk in both populations (P for nonlinearity < 0.001). TyHGB demonstrated superior discriminative ability for NAFLD in comparison with the TyG index, with area under the curve (AUC) values of 0.8410 versus 0.7995 in the Chinese cohort and 0.6492 versus 0.5952 in the NHANES cohort (both P < 0.001). Adding TyHGB to the baseline prediction models significantly improved risk discrimination, with greater improvements than those achieved by adding TyG (NRI: 0.0183, 95% CI: 0.015-0.0217; IDI: 0.0067, 95% CI: 0.0057-0.0076; both P < 0.001). CONCLUSION: TyHGB demonstrated robust and superior performance in predicting NAFLD in comparison with the established TyG index across two diverse populations. Since TyHGB only requires routinely measured clinical parameters, it represents a practical and equitable tool for early NAFLD risk stratification across diverse healthcare settings, potentially reducing health disparities in liver disease detection and enabling cost-effective prevention strategies at the population level.