]. The relative risks of all-cause and cardiac mortality were calculated for luteolin intake tertiles. Kaplan-Meier survival analyses and Cox proportional hazards models were employed to process the survival data. Three statistical models were fitted. The first model adjusted for age , race , and gender . The second model further adjusted for BMI , energy intake , and physical activity in leisure time .
A dose-response association between continuous luteolin intake and mortality risk was visualized by the restricted cubic spline model. Four knots of the spline model were determined at specific distribution percentiles . The likelihood ratio test was used to determine the non-linearity of the dose-response association by comparing the model with and without spline terms.
Subgroup analysis was conducted when patients were stratified by age , gender , race , BMI , energy intake , HbA1c level , and T2DM duration . Interactions were examined by integrating the product term between continuous luteolin intake and stratified factors. To test the robustness, we also conducted several sensitivity analyses. First, the flavonoid database provided the amounts of 29 flavonoids . To avoid the potential effect of remaining flavonoids, we further adjusted other flavonoid intakes in the model. Second, given the underlying luteolin intake outliers, we fitted a new dose-response association after excluding luteolin intake outside the 5th and 95th percentiles.
Table 1 Baseline characteristics based on luteolin intake among patients with type 2 diabetes mellitusDuring follow-up periods , 561 all-cause deaths were documented. Given the sufficient positive outcomes, the sample size was considered adequate based on the criterion of EPV greater than 10.