Abstract
Diabetes is a chronic metabolic disorder that has become a growing global health concern. Waist circumference-glucose index (WyG) is an effective predictor of diabetes; however, its predictive performance in the Japanese population and potential sex-specific differences remain unexplored. This study evaluated the dynamic prediction capability of WyG for diabetes in Japanese individuals, focusing on the differences between sexes. We analyzed data from 15,464 Japanese adults initially without diabetes (7,034 women and 8,430 men). Multivariate Cox regression analysis was used to evaluate the association between WyG and diabetes. Sex-based subgroup analyses were conducted to assess the impact of potential confounders. Sensitivity analysis was conducted to exclude specific populations to test for robustness, and E-values were calculated to evaluate the impact of unmeasured cofounders. Predictive performance was assessed using time-dependent receiver operating characteristic curve analysis. WyG was significantly associated with diabetes in both sexes. This association remained consistent across female subgroups, whereas in males, it was influenced by age and the presence of fatty liver. The area under the curve values ranged from 0.73 to 0.78 in women and 0.73 to 0.77 in men. Sex-specific thresholds (WyG > 8.19 for women, > 8.32 for men) were identified, with greater relative risk elevation for individuals above the threshold. WyG is a reliable predictor of diabetes, with its predictive performance varying across sexes.