Table 3.
Mixed model estimate of the effect of atorvastatin on change in HbA1c using all available follow-up HbA1c measures
| Model | Net treatment effect | ||
|---|---|---|---|
| Men (n=1,861) | Women (n=878) | Men and women (n=2,739) | |
| Model 1a | |||
| % | 0.18 (0.11, 0.26) | 0.06 (−0.06, 0.18) | 0.14 (0.08, 0.21) |
| mmol/mol | 2.0 (1.2, 2.9) | 0.6 (−0.7, 1.9) | 1.5 (0.8, 2.3 |
| Model 2 b | |||
| % | 0.18 (0.10, 0.26) | 0.06 (−0.06, 0.18) | 0.14 (0.07, 0.21) |
| mmol/mol | 2.0 (1.1, 2.9) | 0.7 (−0.6, 2.0) | 1.5 (0.8, 2.3) |
| Model 3 c | |||
| % | 0.18 (0.10, 0.26) | 0.08 (−0.04, 0.19) | 0.14 (0.08, 0.21) |
| mmol/mol | 2.0 (1.1, 2.8) | 0.8 (−0.5, 2.1) | 1.5 (0.8, 2.3) |
Data are shown as % (95% CI) or mmol/mol (95% CI)
In the mixed models observations were nested within patient and trial site.
aModel 1 includes fixed effects of baseline HbA1c, age, diabetes duration, sex (when analysing both sexes together) and treatment group, as well as linear and quadratic terms for study time as random effects nested within individuals
bModel 2: Model 1 + intensity of anti-glycaemic drug therapy, systolic and diastolic blood pressure, total cholesterol, HDL-c and triacylglycerols
cModel 3: Model 2 + baseline waist–hip ratio and BMI, time-updated BMI, anti-glycaemic drug intensity and history of major CVD