Table 7.
Changes in intermediate patient outcomes over the study period
| Intermediate outcomes | Baseline | Year 1 | Year 2 | Mean differences or Risk Ratios (95% CI)† Year 1 vs baseline | Mean differences or Risk Ratios (95% CI)† Year 2 vs baseline |
| Mean or percentage (95% CI) | |||||
| HbA1c | |||||
| Mean HbA1c level (%) | 9.3 (8.8, 9.8) | 8.9 (8.3, 9.4) | 8.9 (8.6, 9.3) | -0.4 (-0.7, -0.1) | -0.4 (-0.7, -0.1) |
| HbA1c < 8% | 37 (28, 46) | 40 (30, 50) | 46 (40, 52) | 1.18 (0.79, 1.61) | 1.43 (1.03, 1.82) |
| HbA1c < 7% | 19 (13, 24) | 21 (13, 29) | 28 (22, 34) | 1.16 (0.58, 2.09) | 1.74 (1.11, 2.50) |
| Blood pressure (mmHg) | |||||
| Mean systolic BP | 130 (127, 133) | 131 (128, 135) | 130 (126, 133) | 2.3 (-0.6, 5.2) | -0.2 (-3.2, 2.8) |
| Mean diastolic BP | 79 (77, 82) | 79 (77, 82) | 79 (76, 81) | 0.4 (-1.6, 2.3) | -0.3 (-2.1, 1.4) |
| BP < 140/90 | 65 (58, 72) | 59 (51, 67) | 67 (61, 73) | 0.84 (0.66, 1.02) | 1.04 (0.90, 1.17) |
| BP < 130/80 | 33 (25, 41)) | 33 (23, 43) | 29 (22, 36) | 0.97 (0.69, 1.29) | 0.80 (0.57, 1.10) |
| Total cholesterol (mmol/L) | |||||
| Mean total cholesterol level | 4.9 (4.7, 5.1) | 4.9 (4.6,5.2) | 4.9 (4.6, 5.3) | -0.01 (-0.2, 0.2) | -0.06 (-0.3, 0.2) |
| Total cholesterol < 5.5 | 73 (66, 80) | 73 (64, 82) | 73 (63, 84) | 1.00 (0.77, 1.17) | 1.02 (0.82, 1.16) |
| Total cholesterol < 4.0 | 22 (16, 28) | 24 (17, 31) | 30 (19, 41) | 1.22 (0.68, 1.97) | 1.58 (0.95, 2.35) |
| ACR | |||||
| Median ACR level * | 18.0 (3.7, 63.9) | 19.6 (3.5, 83.2) | 18.7 (4.8, 67.2) | P = 0.49 | P = 0.32 |
| ACR ≤ 3.4 | 21 (13, 29) | 24 (13, 35) | 18 (13, 23) | 0.93 (0.42, 1.78) | 0.76 (0.35, 1.46) |
| 3.4 < ACR ≤ 34 | 41 (33, 49) | 33 (24, 42) | 43 (35, 51) | 0.73 (0.43, 1.11) | 1.05 (0.72, 1.40) |
| ACR > 34 | 38 (29, 47) | 43 (30, 56) | 39 (30, 48) | 1.60 (0.92, 2.16) | 1.13 (0.67, 1.63) |
* Figures in this row are medians (interquartile ranges). P values indicate statistical significance based on Wilcoxon-Mann-Whitney test for non-parametric group comparison.
† Calculated by using multilevel regression models with adjustment for health centre clustering and repeated measurements within the same individuals, and by converting odds ratios into risk ratios when appropriate using a published formula. While the data for duration of diabetes were not sufficiently complete to allow for adjustment in the analysis (29% of participants had no date of diabetes diagnosis documented in medical records), additional adjustment for age did not significantly change the results. Therefore, we presented the data with no adjustment for age and duration of diabetes in this table. Mean differences or risk ratios significant at 0.05 level are shown in bold.