Smoking, drinking and incident tuberculosis in rural India: population-based case-control study

Abstract

To investigate the extent to which smoking and/or drinking can increase the incidence of pulmonary tuberculosis (TB), a population-based case-control study was conducted in rural south India. A total of 1839 males and 870 females treated in 2000-03 by state TB clinics were interviewed at home in 2004-05 about their education, smoking and drinking habits before disease onset. As controls, 2134 men and 2119 women without TB were randomly chosen from case villages and interviewed. Incidence rate ratios (RRs) are from logistic regression, adjusted for age and education. No women smoked or drank. The main analyses are of men aged 35-64 years, 949 cases treated for new pulmonary TB and 1963 controls. In the study, 81.5% of the cases and 55.2% of the controls had ever smoked, yielding a standardized ever- vs never-smoker TB incidence RR of 2.7 [95% confidence interval (CI) 2.2-3.3, P < 0.00001). Among control ever-smokers 96% still smoked, 71% used only bidis (mean 17 per day) and 28% used only cigarettes (mean 7 per day). After additional adjustment for alcohol, this RR was 2.2 (95% CI 1.7-2.7, P < 0.00001), but even among those who had never drunk alcohol the standardized ever- vs never-smoker RR was 2.6 (95% CI 2.0-3.6, P < 0.00001). The corresponding RRs for ever- vs never-drinking were somewhat less extreme: 2.2 (95% CI 1.8-2.6, P < 0.00001) without adjustment for smoking, 1.5 (95% CI 1.2-1.9, P = 0.00004) with adjustment for smoking and 2.1 (95% CI 1.4-3.0, 2P = 0.0001) among those who had never smoked. Among control ever-drinkers, 96% still drank and 99% used only spirits (mean 0.3 l/week). This study of reliably confirmed disease (by the criteria of state TB clinics) demonstrates an increased incidence of pulmonary TB among those who smoke and among those who drink. The effects of smoking after adjustment for drinking were more definite than those of drinking after adjustment for smoking.