Fig. 2

The dose-response association of luteolin intake with all-cause mortality (A) and cardiac mortality (B) among patients with type 2 diabetes mellitus

The dose-response association of continuous luteolin intake (Ln-transformed) with mortality risk was visualized by the restricted cubic spline model (corresponding to the range of luteolin intake: 0.005–9.870 mg/day). Four knots of the spline model were determined at specific distribution percentiles (5%, 35%, 65%, and 95%). The spline model was adjusted for consistent confounding factors, including age, gender, race, income-poverty ratio, body mass index, educational attainment, smoking, drinking, energy intake, physical activity, hypertension, hyperlipidemia, HbA1c level, and use of oral anti-diabetic agents or insulin. For more details on confounding factors processing, refer to Table 2. The non-linearity of the dose-response association was examined by the likelihood ratio test. The Y-axis represents the adjusted HR given the value of luteolin intake compared to the corresponding median. The shadow area depicts the 95% confidence intervals. The density diagram at the top depicts the distribution of luteolin intake (Ln-transformed)