This paper reports the results of two separate case-control studies of breast cancer which included questions on exposure to tobacco and alcohol. One study used hospital cases and hospital controls while the other looked at cases identified during mammographic screening compared with normal screenees. Both studies used the same questionnaires and methods to obtain information. The main purpose of the studies was to investigate the relationship between breast cancer and certain aspects of fertility and contraception (McPherson et al., 1987). These two studies provide a unique opportunity to relate differences in calculated relative risks for breast cancer to the way cases and controls were selected.
Subec an mwtBetween 1980 and 1984, 998 married women aged 25-59 years, newly presenting with breast cancer at eight hospitals in London and Oxford, and 118 women aged 45-69, diagnosed at the mammographic breast cancer screening clinic in Edinburgh, were interviewed by specially trained nurses. For each London and Oxford patient a married, age-matched (within the same 5-year age group) control was selected at random from female patients in the same hospital who were judged to have conditions which were not associated with breast cancer or with contraceptive practice. The controls in Edinburgh were randomly selected from among the normal screenees. The response rate among women asked to take part approached 100% in both studies.The same questionnaire was used to collect the data in the two studies. As well as information on cigarette smoking and alcohol use (number of cigarettes smoked and alcoholic drinks drunk daily before onset of current illness (or corresponding period for screening controls), age at starting to smoke, history of ever being a regular smoker), data were obtained on socioeconomic status, reproductive variables (including age at menarche, age at menopause, age at first term pregnancy, number of children, details of oral contraceptive use) and other potentially confounding variables (including family history of breast cancer, weight and height).The data in both studies were analysed using a matched pairs multiple logistic method which yielded relative risks for Br.