Summary
    Background At least 75% of head and neck cancers are attributable to a combination of cigarette smoking and alcohol drinking. A precise understanding of the independent association of each of these factors in the absence of the other with the risk of head and neck cancer is needed to elucidate mechanisms of head and neck carcinogenesis and to assess the efficacy of interventions aimed at controlling either risk factor. Methods We examined the extent to which head and neck cancer is associated with cigarette smoking among never drinkers and with alcohol drinking among never users of tobacco. We pooled individual-level data from 15 case-control studies that included 10244 head and neck cancer case subjects and 15227 control subjects, of whom 1072 case subjects and 5775 control subjects were never users of tobacco and 1598 case subjects and 4051 control subjects were never drinkers of alcohol. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated using unconditional logistic regression models. All statistical tests were two-sided. Results Among never drinkers, cigarette smoking was associated with an increased risk of head and neck cancer (OR for ever versus never smoking = 2.13, 95% CI = 1.52 to 2.98), and there were clear dose-response relationships for the frequency, duration, and number of pack-years of cigarette smoking. Approximately 24% (95% CI = 16% to 31%) of head and neck cancer cases among nondrinkers in this study would have been prevented if these individuals had not smoked cigarettes. Among never users of tobacco, alcohol consumption was associated with an increased risk of head and neck cancer only when alcohol was consumed at high frequency (OR for three or more drinks per day versus never drinking = 2.04, 95% CI = 1.29 to 3.21). The association with high-frequency alcohol intake was limited to cancers of the oropharynx/hypopharynx and larynx. Conclusions Our results represent the most precise estimates available of the independent association of each of the two main risk factors of head and neck cancer, and they exemplify the strengths of large-scale consortia in cancer epidemiology