| Intervention, mean outcome (95% CI) (clusters n = 13) | Control, mean outcome (95% CI) (clusters n = 13) | Crude control-intervention difference (95% CI); P valueb | Adjusted control-intervention difference (95% CI); P valueb | |
|---|---|---|---|---|
| Change in SBP, mmHgc | -25.23 (-29.86 to -20.61) | -9.41 (-21.24 to 2.24) | 15.82 (3.60 to 28.04); 0.01 | 12.63 (0.68 to 24.57); 0.04 |
| Change in DBP, mmHgc | -18.18 (-22.12 to -14.25) | -8.62 (-15.00 to -2.24) | 9.57 (2.39 to 16.74); 0.01 | 7.58 (0.61 to 14.55); 0.04 |
| Hypertension control, %d | 69.56 (57.09 to 82.03) | 35.79 (15.44 to 56.13) | -34pp (-56 to -11); 0.01 | -30pp (-53 to -6); 0.02 |
| Change in HbA1c, %c | -0.70pp (-2.02 to 0.61) | -1.71pp (-2.63 to -0.79) | 1.01pp (-2.53 to 0.52); 0.18 | 0.59pp (-0.29 to 0.17); 0.60 |
| Glycaemic control, %e | 33.03 (10.11 to 55.94%) | 42.79 (27.22 to 58.36%) | 10pp (-17 to 36); 0.45 | 3pp (-2 to 87); 0.19 |
| Change in total serum cholesterol (mg/dL)c | -27.69 (-31.69 to -23.69) | 0.52 (-9.85 to 10.89) | 28.21 (17.37 to 39.06); 0.001 | 22.52 (15.89 to 29.16); 0.001 |
| Treatment adherencef | 73.11 (56.34 to 89.88) | 18.60 (6.63 to 30.57) | -54.51 (-74.13 to -34.88); 0. 00 | -53.57 (-77.68 to -29.46); 0.00 |
| Smoking cessation | 3.08 (0.78 to 5.38) | 4.17 (1.74 to 6.62) | 1.10 (-2.08 to 4.28); 0.48 | 0.03 (-1.37 to 1.43); 0.97 |
DBP = diastolic blood pressure. pp = percentage points. SBP = systolic blood pressure. aArm-specific mean outcomes and their 95% confidence intervals are themselves based on cluster-level summary (mean/proportion) outcomes. bAll control minus intervention differences (that is, intervention effect estimates) are based on analysis of crude/covariate-adjusted cluster-level summary (mean/proportion) outcomes. cAll change outcomes are calculated as outcome at endpoint minus outcome at baseline. dHypertension control defined as endpoint systolic BP ≤140 mmHg; eGlycaemic control defined as endpoint HbA1c ≤7%. fTreatment adherence was defined as patient having attended ≥5 treatment visits. All differences for percentage outcomes are on the absolute scale as percentage points. All analyses use only complete cases.