Table 2.
The association between smoking status and HbA1c reduction was estimated by generalised estimating equations after propensity score matching.
Model 1: dichotomous smoking status |
Model 2: four groups of smoking status |
||||
---|---|---|---|---|---|
Variables | β (95% CI) | p-value | Variables | β (95% CI) | p-value |
Group: | Group: | ||||
Non-smokers | 0 | Smoking (−/−) | 0 | ||
Smokers | −0.11 (−0.37, 0.16) | .44 | Smoking (−/+) | −0.10 (−0.68, 0.47) | .72 |
Smoking (+/−) | −0.15 (−0.60, 0.31) | .52 | |||
Smoking (+/+) | −0.11 (−0.40, 0.18) | .45 | |||
Time: Baseline | 0 | Time: Baseline | 0 | ||
3 months | −2.60 (−2.81, −2.40) | <.001 | 3 months | −2.62 (−2.85, −2.40) | <.001 |
6 months | −2.73 (−2.94, −2.52) | <.001 | 6 months | −2.74 (−2.96, −2.51) | <.001 |
9 months | −2.63 (−2.84, −2.42) | <.001 | 9 months | −2.64 (−2.87, −2.42) | <.001 |
12 months | −2.68 (−2.88, −2.47) | <.001 | 12 months | −2.70 (−2.92, −2.47) | <.001 |
Interaction of smoking group and time | Interaction of smoking group and time | ||||
Baseline | 0 | Smoking (−/+): Baseline 0 | |||
3 months | 0.33 (0.05, 0.62) | .02 | 3 months | 0.12 (−0.46, 0.69) | .68 |
6 months | 0.31 (0.02, 0.59) | .04 | 6 months | 0.07 (−0.54, 0.69) | .81 |
9 months | 0.25 (−0.04, 0.54) | .09 | 9 months | 0.07 (−0.57, 0.71) | .84 |
12 months | 0.30 (0.01, 0.59) | .04 | 12 months | 0.18 (−0.44, 0.79) | .58 |
Smoking (+/−): Baseline 0 | |||||
3 months | 0.62 (0.12, 1.12) | .02 | |||
6 months | 0.58 (0.06, 1.10) | .03 | |||
9 months | 0.47 (−0.04, 0.99) | .07 | |||
12 months | 0.48 (−0.02, 0.98) | .06 | |||
Smoking (+/+): Baseline 0 | |||||
3 months | 0.29 (−0.03, 0.60) | .08 | |||
6 months | 0.25 (−0.06, 0.57) | .12 | |||
9 months | 0.21 (−0.10, 0.52) | .19 | |||
12 months | 0.29 (−0.02, 0.60) | .07 |
Note. The difference in HbA1c reduction between smokers and non-smokers was estimated by the regression coefficient (β) and 95% confidence interval (95% CI) of smoking status-by-time interaction at each time point. The models were established using the backward elimination method to select control variables, including a family history of diabetes, alcohol drinking, knowledge regarding glycemic control, medication adherence, anti-hypertensive agents, blood pressure, total cholesterol, estimated glomerular filtration rate, and body weight change.
Abbreviation: HbA1c: haemoglobin A1c.