Body Mass Index, Cardiometabolic Treatment Patterns and Multifactorial Risk‐Factor Control in Chinese Adults With Type 2 Diabetes: A Multicenter Cross‐Sectional Study

Figures

• 

  FIGURE 1. Observed target achievement by BMI category. Points show the observed proportion of participants achieving each target and vertical lines the 95% Wilson confidence interval, among participants with the relevant observed component (denominators under each panel: Underweight/normal weight/overweight/obesity). Connecting lines are visual guides only. (A) HbA1c < 7.0%; (B) blood pressure < 130/80 mmHg; (C) LDL‐C < 2.6 mmol/L; (D) composite 3B target, restricted to participants with all three observed components (n = 6683).

• 

  FIGURE 2. Center‐clustered adjusted association of BMI categories with composite 3B target achievement. Squares indicate pooled adjusted odds ratios and horizontal lines 95% confidence intervals from exchangeable‐correlation logistic GEE across 20 multiply imputed datasets (20 iterations per chain), clustered by recruitment‐center code with robust standard errors. Models adjusted for age, sex, diabetes duration, smoking, alcohol use, education, physical activity, and standardized macrovascular history; normal weight (diamond) was the reference. Obesity versus normal weight was the primary comparison of this secondary analysis; other p values are nominal.

• 

  FIGURE 3. Baseline medication use by BMI category. Cells show the observed percentage of participants with recorded use of each of the 10 medication classes consistently available in the adult analytic dataset; denominators are the full adult BMI groups (n in column headers). Rows are grouped as glucose‐lowering, blood‐pressure‐lowering, and lipid‐lowering/antithrombotic classes. Medication classes are not mutually exclusive.

• 

  FIGURE 4. Adjusted associations of continuous BMI with target attainment using restricted cubic splines. Curves show pooled adjusted odds ratios (log scale) and shaded bands pointwise 95% confidence intervals relative to BMI 24 kg/m2 (vertical dotted line), from center‐clustered GEE models pooled across 20 imputed datasets with the covariates of Figure 2. Triangles mark the knots (5th, 35th, 65th, and 95th percentiles: 20.28, 24.17, 26.84, and 32.70 kg/m2). Curves span the 0.5th to 99.5th BMI percentiles, and the histogram strip beneath each panel shows the observed BMI distribution of the analytic cohort (0.5 kg/m2 bins). D1 joint tests in each panel compare the spline with a model omitting BMI (overall) and with a linear BMI term (nonlinear). (A) Composite 3B target; (B) HbA1c < 7.0%; (C) blood pressure < 130/80 mmHg; (D) LDL‐C < 2.6 mmol/L.