Smoking and smoking cessation in relation to risk of diabetes in Chinese men and women: a 9-year prospective study of 0·5 million people

Abstract

Background: In developed countries, smoking is associated with increased risk of diabetes. Little is known about the association in China, where cigarette consumption has increased (first in urban, then in rural areas) relatively recently. Moreover, uncertainty remains about the effect of smoking cessation on diabetes in China and elsewhere. We aimed to assess the associations of smoking and smoking cessation with risk of incident diabetes among Chinese adults.

Methods: The prospective China Kadoorie Biobank enrolled 512 891 adults (59% women) aged 30-79 years during 2004-08 from ten diverse areas (five urban and five rural) across China. Participants were interviewed at study assessment clinics, underwent physical measurements, and had a non-fasting blood sample taken. Participants were separated into four categories according to smoking history: never-smokers, ever-regular smokers, ex-smokers, and occasional smokers. Incident diabetes cases were identified through linkage with diabetes surveillance systems, the national health insurance system, and death registries. All analyses were done separately in men and women and Cox regression was used to yield adjusted hazards ratios (HRs) for diabetes associated with smoking.

Findings: 68% (n=134 975) of men ever smoked regularly compared with 3% (n=7811) of women. During 9 years' follow-up, 13 652 new-onset diabetes cases were recorded among 482 589 participants without previous diabetes. Among urban men, smokers had an adjusted HR of 1·18 (95% CI 1·12-1·25) for diabetes. HRs increased with younger age at first smoking regularly (1·12, 1·20, and 1·27 at ≥25 years, 20-24 years, and <20 years, respectively; p for trend=0·00073) and with greater amount smoked (1·11, 1·15, 1·42, and 1·63 for <20, 20-29, 30-39 and ≥40 cigarettes per day; p for trend<0·0001). Among rural men, similar, albeit more modest, associations were seen. Overall, HRs were more extreme at higher levels of adiposity. Among men who stopped by choice, there was no excess risk within 5 years of cessation, contrasting with those who stopped because of illness (0·92 [0·75-1·12] vs 1·42 [1·23-1·63]). Among the few women who ever smoked regularly, the excess risk of diabetes was significant (1·33 [1·20-1·47]).

Interpretation: Among Chinese adults, smoking was associated with increased risk of diabetes, with no significant excess risk following voluntary smoking cessation.

Funding: Wellcome Trust, Medical Research Council, British Heart Foundation, Cancer Research UK, Kadoorie Charitable Foundation, Ministry of Science and Technology, National Natural Science Foundation of China, and China Scholarship Council.

Figure 1

Adjusted HRs of incident diabetes by smoking in men Stratified by age at risk and study area and adjusted for education, alcohol consumption, physical activity, body-mass index, and waist circumference. Analyses examining smoking duration were additionally adjusted for age at baseline. Ever-regular smokers excludes occasional smokers (n=22 435) and ex-smokers who stopped by choice (n=12 950). Tests for trend include all smoking categories. Tests for heterogeneity include only smokers. HR=hazard ratio.

Figure 2

Adjusted HRs of incident diabetes by amount smoked, stratified by levels of adiposity, in men Stratified by age at risk and study area and adjusted for education, alcohol consumption, and physical activity. Ever-regular smokers excludes occasional smokers and ex-smokers who stopped by choice. HR=hazard ratio. BMI=body-mass index. WC=waist circumference. BF%=body fat percentage.

Figure 3

Adjusted HRs of incident diabetes in male ex-smokers by reasons for, and years after, smoking cessation Model 1 was stratified by age at risk and study area and adjusted for education, alcohol consumption, and physical activity. Model 2 was additionally adjusted for body-mass index and waist circumference. Tests for trend include only ex-smokers. HR=hazard ratio.

Figure 4

Adjusted HRs of incident diabetes by smoking in women Stratified by age at risk and study area and adjusted for education, alcohol consumption, physical activity, body-mass index, and waist circumference. Analyses examining smoking duration were additionally adjusted for age at baseline. Tests for trend include all smoking categories. Test for heterogeneity across smoking categories includes all smoking categories. Other tests for heterogeneity include only smokers. HR=hazard ratio.