Table 3.
Comparison between study groups in the level of T2D risk factors in each wave as well as their change from baseline.
| Risk factor | Wave 2 | Wave 3 | Wave 4 | Wave 5 | ||||
|---|---|---|---|---|---|---|---|---|
| Mean (SD) | Change from baseline | Mean (SD) | Change from baseline | Mean (SD) | Change from baseline | Mean (SD) | Change from baseline | |
| Waist | ||||||||
| Intervention | 91.5 (12.0) | 4.2 (3.9, 4.5) | 92.7 (12.4) | 5.4 (5.1, 5.8) | 96.2 (11.8) | 8.9 (8.6, 9.3) | 96.1 (11.8) | 8.8 (8.4, 9.3) |
| Control | 91.6 (12.1) | 4.9 (4.6, 5.1) | 92.3 (12.3) | 5.6 (5.4, 5.8) | 96.2 (11.8) | 9.4 (9.2, 9.7) | 96.3 (11.8) | 9.5 (9.2, 9.8) |
| Between group difference |
− 0.5 (− 0.8, − 0.1) p-value = 0.021 |
− 0.6 (− 1.0, − 0.3) p-value < 0.0001 |
0.0 (− 0.3, 0.4) p-value = 0.875 |
− 0.2 (− 0.6, 0.2) p-value = 0.434 |
− 0.3 (− 0.7, 0.0) p-value = 0.065 |
− 0.5 (− 0.9, − 0.1) p-value = 0.017 |
− 0.5 (− 0.9, − 0.1) p-value = 0.011 |
− 0.7 (− 1.1, − 0.2) p-value = 0.004 |
| Body mass index | ||||||||
| Intervention | 27.6 (4.8) | 1.0 (0.8, 1.2) | 28.0 (4.8) | 1.4 (1.2, 1.6) | 28.5 (6.7) | 1.9 (1.7, 2.2) | 28.7 (5.1) | 2.1 (1.9, 2.4) |
| Control | 27.5 (4.8) | 1.0 (0.9, 1.2) | 27.9 (4.8) | 1.4 (1.3, 1.6) | 28.6 (9.9) | 2.1 (2.0, 2.3) | 28.8 (5.3) | 2.3 (2.1, 2.5) |
| Between group difference |
0.0 (− 0.2, 0.2) p-value = 0.957 |
0.0 (− 0.2, 0.2) p-value = 0.821 |
− 0.1 (− 0.3, 0.1) p-value = 0.504 |
− 0.1 (− 0.3, 0.2) p-value = 0.509 |
− 0.2 (− 0.5, 0.1) p-value = 0.219 |
− 0.2 (− 0.5, 0.1) p-value = 0.251 |
− 0.1 (− 0.3, 0.1) p-value = 0.190 |
− 0.2 (− 0.4, 0.1) p-value = 0.273 |
| Fasting plasma glucose | ||||||||
| Intervention | 4.99 (0.76) | 0.03 (− 0.01, 0.06) | 5.13 (0.97) | 0.17 (0.13, 0.21) | 5.49 (1.14) | 0.53 (0.48, 0.57) | 5.60 (1.38) | 0.64 (0.59, 0.69) |
| Control | 5.13 (0.72) | 0.14 (0.11, 0.17) | 5.09 (0.88) | 0.11 (0.07, 0.14) | 5.47 (1.20) | 0.49 (0.45, 0.52) | 5.60 (1.44) | 0.61 (0.57, 0.65) |
| Between group difference |
− 0.13 (− 0.17, − 0.08) p-value < 0.0001 |
− 0.11 (− 0.16, − 0.07) p-value < 0.0001 |
0.05 (0.00, 0.09) p-value = 0.035 |
0.06 (0.01, 0.11) p-value = 0.019 |
0.03 (− 0.02, 0.08) p-value = 0.211 |
0.04 (− 0.01, 0.10) p-value = 0.133 |
0.02 (− 0.04, 0.08) p-value = 0.539 |
0.03 (− 0.04, 0.10) p-value = 0.358 |
| Post-prandial glucose | ||||||||
| Intervention | 6.1 (2.2) | 0.2 (0.1, 0.3) | 6.2 (2.6) | 0.3 (0.2, 0.4) | 6.7 (3.0) | 0.8 (0.7, 0.9) | 7.2 (3.6) | 1.3 (1.2, 1.4) |
| Control | 6.3 (2.2) | 0.4 (0.3, 0.4) | 6.2 (2.5) | 0.3 (0.2, 0.4) | 6.6 (3.0) | 0.8 (0.7, 0.8) | 7.2 (3.6) | 1.4 (1.2, 1.5) |
| Between group difference |
− 0.1 (− 0.3, 0.0) p-value = 0.021 |
− 0.1 (− 0.3, 0.0) p-value = 0.009 |
0.0 (− 0.1, 0.1) p-value = 0.800 |
0.0 (− 0.1, 0.1) p-value = 0.751 |
0.0 (− 0.1, 0.2) p-value = 0.720 |
0.0 (− 0.1, 0.2) p-value = 0.815 |
− 0.1 (− 0.2, 0.1) p-value = 0.496 |
− 0.1 (− 0.2, 0.1) p-value = 0.513 |
| Triglycerides | ||||||||
| Intervention | 1.74 (1.10) | − 0.09 (− 0.12, − 0.05) | 1.73 (1.08) | − 0.10 (− 0.14, − 0.06) | 1.68 (1.05) | − 0.15 (− 0.20, − 0.11) | 1.71 (1.01) | − 0.11 (− 0.16, − 0.07) |
| Control | 1.80 (1.20) | − 0.02 (− 0.05, 0.01) | 1.78 (1.09) | − 0.04 (− 0.08, − 0.01) | 1.73 (1.12) | − 0.10 (− 0.13, − 0.06) | 1.73 (1.01) | − 0.10 (− 0.14, − 0.06) |
| Between group difference |
− 0.06 (− 0.11, − 0.01) p-value = 0.012 |
− 0.07 (− 0.11, − 0.02) p-value = 0.007 |
− 0.05 (− 0.09, − 0.01) p-value = 0.021 |
− 0.06 (− 0.11, 0.00) p-value = 0.037 |
− 0.05 (− 0.09, − 0.01) p-value = 0.025 |
− 0.05 (− 0.11, 0.00) p-value = 0.061 |
− 0.01 (− 0.05, 0.03) p-value = 0.629 |
− 0.02 (− 0.07, 0.04) p-value = 0.581 |
| HDL-cholesterol | ||||||||
| Intervention | 1.02 (0.27) | − 0.08 (− 0.09, − 0.07) | 1.11 (0.27) | 0.02 (0.01, 0.03) | 1.22 (0.29) | 0.13 (0.11, 0.14) | 1.28 (0.32) | 0.19 (0.18, 0.20) |
| Control | 0.99 (0.26) | − 0.10 (− 0.11, − 0.10) | 1.07 (0.27) | − 0.03 (− 0.04, − 0.02) | 1.23 (0.29) | 0.13 (0.12, 0.14) | 1.26 (0.32) | 0.17 (0.16, 0.18) |
| Between group difference |
0.03 (0.01, 0.04) p-value < 0.0001 |
0.03 (0.02, 0.04) p-value < 0.0001 |
0.04 (0.03, 0.06) p-value < 0.0001 |
0.05 (0.03, 0.06) p-value < 0.0001 |
− 0.01 (− 0.02, 0.00) p-value = 0.173 |
− 0.01 (− 0.03, 0.01) p-value = 0.375 |
0.02 (0.01, 0.03) p-value = 0.001 |
0.02 (0.00, 0.03) p-value = 0.013 |
| Total cholesterol | ||||||||
| Intervention | 4.9 (1.0) | − 0.4 (− 0.4, − 0.3) | 5.0 (1.0) | − 0.3 (− 0.3, − 0.3) | 5.0 (1.0) | − 0.3 (− 0.4, − 0.3) | 5.1 (1.0) | − 0.2 (− 0.2, − 0.1) |
| Control | 5.0 (1.1) | − 0.3 (− 0.3, − 0.3) | 5.0 (1.0) | − 0.3 (− 0.4, − 0.3) | 5.1 (1.0) | − 0.2 (− 0.3, − 0.2) | 5.1 (1.0) | − 0.2 (− 0.2, − 0.2) |
| Between group difference |
− 0.1 (− 0.1, 0.0) p-value = 0.012 |
− 0.1 (− 0.1, 0.0) p-value = 0.010 |
0.0 (0.0, 0.1) p-value = 0.053 |
0.0 (0.0, 0.1) p-value = 0.170 |
− 0.1 (− 0.1, 0.0) p-value = 0.004 |
− 0.1 (− 0.1, 0.0) p-value = 0.015 |
0.0 (0.0, 0.1) p-value = 0.238 |
0.0 (0.0, 0.1) p-value = 0.499 |
| DASH score | ||||||||
| Intervention | NA | NA | 5.9 (1.5) | NA | 6.8 (1.4) | NA | 6.9 (1.4) | NA |
| Control | NA | NA | 5.8 (1.5) | NA | 6.6 (1.5) | NA | 6.7 (1.5) | NA |
| Between group difference | NA | NA |
0.1 (0.0, 0.3) p-value = 0.091 |
NA |
0.2 (0.1, 0.3) p-value = 0.0001 |
NA |
0.2 (0.1, 0.3) p-value = 0.0003 |
NA |
| Energy intake | ||||||||
| Intervention | NA | NA | 2,347 (1,041) | NA | 2,481 (1,025) | NA | 2,432 (1,020) | NA |
| Control | NA | NA | 2,371 (1,066) | NA | 2,490 (1,053) | NA | 2,366 (1,007) | NA |
| Between group difference | NA | NA |
− 17 (− 106, 72) p-value = 0.690 |
NA |
− 2 (− 72, 68) p-value = 0.946 |
NA |
73 (16, 131) p-value = 0.014 |
NA |
| N (%) | Change from baseline (Odds ratio) | N (%) | Change from baseline (Odds ratio) | N (%) | Change from baseline (Odds ratio) | N (%) | Change from baseline (Odds ratio) | |
| Smoking | ||||||||
| Intervention | 424 (12.2) | 0.9 (0.8, 1.1) | 376 (10.9) | 0.8 (0.7, 0.9) | 379 (10.9) | 0.8 (0.7, 1.0) | 384 (11.1) | 0.8 (0.7, 1.0) |
| Control | 878 (15.3) | 1.1 (1.0, 1.1) | 832 (14.5) | 1.0 (0.9, 1.1) | 783 (13.6) | 0.9 (0.8, 1.0) | 768 (13.4) | 0.9 (0.8, 1.0) |
| Between group difference |
0.72 (0.56, 0.91) p-value = 0.008 |
0.88 (0.76, 1.02) p-value = 0.085 |
0.62 (0.48, 0.79) p-value < 0.001 |
0.81 (0.69, 0.96) p-value = 0.014 |
0.74 (0.57, 0.96) p-value = 0.024 |
0.89 (0.74, 1.08) p-value = 0.246 |
0.79 (0.61, 1.04) p-value = 0.092 |
0.93 (0.76, 1.14) p-value = 0.496 |
| Low physical activity | ||||||||
| Intervention | 424 (12.2) | 0.3 (0.2, 0.3) | 376 (10.9) | 0.2 (0.2, 0.2) | 379 (10.9) | 0.2 (0.2, 0.2) | 384 (11.1) | 0.2 (0.2, 0.2) |
| Control | 878 (15.3) | 0.3 (0.3, 0.3) | 832 (14.5) | 0.2 (0.2, 0.2) | 783 (13.6) | 0.2 (0.2, 0.3) | 768 (13.4) | 0.2 (0.2, 0.2) |
| Between group difference |
1.00 (0.84, 1.19) p-value = 0.974 |
0.86 (0.71, 1.04) p-value = 0.113 |
1.10 (0.98, 1.23) p-value = 0.118 |
0.94 (0.81, 1.09) p-value = 0.436 |
0.92 (0.82, 1.03) p-value = 0.135 |
0.79 (0.68, 0.92) p-value = 0.002 |
0.99 (0.86, 1.14) p-value = 0.886 |
0.85 (0.72, 1.01) p-value = 0.064 |
To compare the risk factor levels between the study groups in different waves, generalized estimating equations (GEE) were fitted in a long-form dataset including data from waves 2–5. In each model, the level of risk factor was defined as the outcome and the predictors were defined to be baseline level of risk factor, time-point variable (i.e., waves 2–5), interventions status (control and intervention), the interaction term between intervention status and time-point variable, and potential confounders (i.e., age, sex, area of residence, education level, family history of diabetes, smoking, low physical activity, WC, BMI, systolic blood pressure, diastolic blood pressure, FPG, 2 h-PG, total cholesterol, triglycerides, and HDL-C, and self-reported drug consumption for hypertension, and dyslipidemia). The model also accounted for the clustered nature of data due to repeated measurement using an auto-regressive correlation matrix (the autoregressive process of order 1). The autoregressive correlation matrix was chosen since measurements taken further apart were less correlated than those taken closer together. Logit link function for binary outcomes and identity link function for continuous outcomes were used in the GEE models.
To compare change from baseline for the risk factor levels between the study groups in different waves, GEE models with similar link functions (as described above) and autoregressive correlation matrix were fitted in a long-form dataset including data from waves 1–5. In each model, the level of risk factor was defined as the outcome and the predictors were defined to be time-point variable (i.e., waves 1–5), intervention status (control and intervention), the interaction term between intervention status and time-point variable, and potential confounders (as listed above). The estimated coefficient for interaction terms between intervention status and time-point variable were used to compare study groups in change in risk factor levels from baseline to different waves of the study.
Based on the Bonferroni method, the corrected p-value threshold was 0.00058 for the secondary aim of the study.