Table 2. Adjusted hazard ratios (95% CI) for incidence diabetes according to baseline smoking status.
Never smoker | Former smoker | Current smoker | |
---|---|---|---|
n | 23189 | 10162 | 20579 |
Cases | 799 | 568 | 1074 |
Incidence (/1000 person-years) | 9.7 | 16.1 | 14.7 |
Age-adjusted incidence | 10.5 | 14.5 | 15.4 |
Attributable risk 1 | - | 4.0 | 4.9 |
Model 1 2 | 1.00 | 1.20 (1.07–1.34) | 1.31 (1.19–1.44) |
Model 2 3 | 1.00 | 1.16 (1.04–1.30) | 1.34 (1.22–1.48) |
Model 3 4 , 5 | 1.00 | 1.16 (1.02–1.33) | 1.31 (1.16–1.47) |
BMI <23 kg/m2 | |||
n | 12936 | 4570 | 10414 |
Cases | 215 | 149 | 325 |
Incidence (/1000 person-years) | 4.7 | 9.3 | 8.7 |
Age-adjusted incidence | 5.3 | 8.2 | 9.0 |
Attributable risk 1 | - | 2.9 | 3.7 |
Model 1 2 | 1.00 | 1.33 (1.07–1.65) | 1.50 (1.25–1.79) |
Model 2 3 | 1.00 | 1.27 (1.02–1.59) | 1.49 (1.24–1.78) |
Model 3 4 , 5 | 1.00 | 1.22 (0.94–1.59) | 1.37 (1.10–1.72) |
BMI ≥23 kg/m2 | |||
n | 10253 | 5592 | 10165 |
Cases | 584 | 419 | 749 |
Incidence (/1000 person-years) | 16.2 | 21.7 | 21.0 |
Age-adjusted incidence | 16.9 | 20.1 | 22.3 |
Attributable risk 1 | - | 3.2 | 5.4 |
Model 1 2 | 1.00 | 1.12 (0.99–1.28) | 1.28 (1.14–1.43) |
Model 2 3 | 1.00 | 1.11 (0.97–1.26) | 1.26 (1.13–1.41) |
Model 3 4 , 5 | 1.00 | 1.13 (0.96–1.32) | 1.25 (1.09–1.43) |
P for interaction 6 | 0.13 | 0.07 |
CI, confidence interval; BMI, body mass index.
1Difference in age-adjusted incidence rate per 1000 person-years between never smoker and current or former smoker.
2Adjusted for age (years), sex, and worksite.
3Adjusted for all factors in model 1 plus BMI (kg/m2), waist circumference (cm), and hypertension (yes or no).
4Adjusted for all factors in model 2 plus alcohol intake (<23 g ethanol/d or ≥ 23 g ethanol/d), shift work (yes or no), sleeping time (<6 h/d, 6–7 h/d, or ≥7 h/d), leisure-time physical activity (<150 min/wk or ≥150 min/wk), family history of diabetes (yes or no).
5Data were available for 32373 subjects.
6P for interaction for dichotomized variable of BMI (<23 or ≥23 kg/m2) on the association between smoking and type 2 diabetes was based on model 2 and was examined using the likelihood ratio test.