Tobacco smoking and bladder cancer in coffee non-drinkers.

Data from a large Italian case-control study confirm that the tobacco related relative risk of bladder cancer is higher in coffee non-drinkers than in coffee drinkers. However, given the correlation between tobacco and coffee, the population attributable risk was similar for coffee drinkers and non-drinkers. Lopez-Abente and Escolar, using data from a Spanish case-control study, suggested that the association between tobacco smoking and bladder cancer may be stronger among coffee non-drinkers than among usual coffee drinkers. 1 This potential favourable interaction has some biological plausibility, and can be related to the inhibition of enzymes involved in the activation of carcinogenic aromatic amines, including cytochrome P4501A2. 2 To provide further information on the issue, we analysed data from a case-control study conducted in Italy between 1985 and 1992, whose main results on tobacco and coffee have been published elsewhere. 3 4 Briefly, a total of 727 histologically confirmed, incident cases of bladder cancer (617 men, 110 women), aged 27–79 years (median age 63), were included in the study. They were recruited in the National Cancer Institute and in a network of other major general hospitals and university clinics in greater Milan and in Pordenone (north eastern Italy). Controls were 1067 subjects (769 men, 298 women), aged 25–79 years (median age 60), admitted to the same network of hospitals of cases for acute, non-neoplastic, non-urological or genital tract diseases (23% were admitted for surgical conditions, 29% for fractures and traumatic conditions, 18% for non-traumatic orthopaedic disorders, and 30% for miscellaneous other illnesses). Less than 3% of subjects approached (cases and controls) refused the interview. Data were collected by trained interviewers using a structured questionnaire, including information on sociodemographic factors, personal characteristics and lifestyle habits, history of selected diseases and of relevant occupational exposures. Questions on coffee drinking included average cups drunk per day and duration of use; questions on tobacco included smoking status (never smoker, ex smoker or current smoker), average quantity smoked per day, and duration of the habit. Information on tobacco and coffee was satisfactorily reliable. 5 We estimated the odds ratios (OR) and their 95% confidence intervals (CI) using unconditional multiple logistic regression, including terms for age, sex and study centre. Further adjustment for education and occupation did not materially change any of the results. Table 1 shows the distribution of cases and controls, the corresponding OR and 95%CI according to tobacco smoking in strata of coffee drinking. Among …