Introduction. Over the past several decades, incidence rates of non-Hodgkin lymphoma (NHL) have increased dramatically throughout the world among both men and women, and the responsible factors have not been identified. Cigarette smoking and alcohol consumption have been proposed as two factors that could potentially alter the risk of NHL. However, previous epidemiologic investigations of smoking, alcohol, and NHL have yielded inconsistent results, partially due to relatively small numbers of NHL cases, particularly when considering the subtype of NHL. Therefore, we pooled original data from three case-control studies, representing a total of six SEER sites, as part of the International Consortium of Investigators Working on Lymphoma Epidemiologic Studies (InterLymph) to examine the strength and consistency of the associations between smoking, alcohol, and NHL. Methods. Original data from three similarly designed, population-based case-control studies were obtained and uniformly coded, resulting in a pooled dataset comprising 3513 cases (1721 males, 1792 females) and 4290 controls (2225 males, 2065 females). Statistically significant heterogeneity of the odds ratios between studies was not observed; thus, joint fixed-effects multivariate logistic regression models were used to compute pooled estimates of the odds ratios. Results. When compared to non-drinkers, consumption of at least 12 drinks per year of alcohol was significantly inversely associated with NHL (OR=0.75; 95% C.I. 0.67, 0.85). Results varied slightly by type of alcohol and gender (men: beer OR=0.82, wine OR=0.93, liquor OR=1.08; women: beer OR=0.90, wine OR=0.80, liquor OR=0.92). Risk of NHL was reduced with increasing intensity of alcohol consumption. Compared to non-drinkers, people who drank < 1 drink, 1 drink, 2-3 drinks, 4 drinks, and 5 or more drinks per day had estimated risks of NHL of OR=0.72, 0.71, 0.85, 0.67, and 0.63, respectively (p for linear trend < 0.01). When compared to never smokers, people who ever smoked had only a slight increased risk of NHL (OR=1.09; 95% C.I. 0.98, 1.22). These results did not vary by age at initiation, intensity, duration, or lifetime pack-years of smoking. Although previous studies of NHL have suggested that the impact of smoking may vary by NHL subtype, our preliminary classification of NHL subtypes suggests that smoking does not alter the risk of follicular and diffuse NHL subtypes. Conclusions. The data from this pooled analysis suggest that consumption of alcohol is inversely associated with risk of NHL, while cigarette smoking does not significantly increase the risk of NHL. With additional data in the pooling project, we will estimate effects for individual NHL subtypes.

[Proc Amer Assoc Cancer Res, Volume 45, 2004]