The change of triglyceride-glucose index may predict incidence of stroke in the general population over 45 years old

Stroke has been found to be highly correlated with the triglyceride-glucose (TyG) index. The relation between the TyG index changes and stroke, however, has seldom been reported, and current researches mentioning the TyG index concentrate on individual values. We aimed to investigate whether the level and the change of TyG index was associated with the incidence of stroke.Sociodemographic, medical background, anthropometric and laboratory information were retrospectively collected. Classification was conducted using k-means clustering analysis. Logistic regressions were to determine the relationship between different classes with changes in the TyG index and incidence of stroke, taking the class with the smallest change as a reference. Meanwhile, restricted cubic spline regression was applied to examine the links of cumulative TyG index and stroke.369 (7.8%) of 4710 participants had a stroke during 3 years. Compared to class 1 with the best control of the TyG Index, the OR for class 2 with good control was 1.427 (95% CI, 1.051-1.938), the OR for class 3 with moderate control was 1.714 (95% CI, 1.245-2.359), the OR for class 4 with worse control was 1.814 (95% CI, 1.257-2.617), and the OR for class 5 with consistently high levels was 2.161 (95% CI, 1.446-3.228). However, after adjusting for multiple factors, only class 3 still had an association with stroke (OR 1.430, 95%CI, 1.022-2.000). The relation between the cumulative TyG index and stroke was linear in restricted cubic spline regression. In subgroup analysis, similar results were shown in participants without diabetes or dyslipidemia. There is neither additive nor multiplicative interaction between TyG index class and covariates.A constant higher level with worst control in TyG index indicated a higher risk of stroke.