The case-control study has been described in detail in previous reports (20
). In brief, cases were women with a first primary breast cancer diagnosis identified from population-based cancer registries in Wisconsin, Massachusetts, and New Hampshire according to protocols approved by institutional review boards at each site. Eligible cases included women who resided in Wisconsin, Massachusetts (excluding metropolitan Boston), or New Hampshire, were ages 20 to 74 years at diagnosis, had a listed telephone number, and were verified by self-report to have a driver’s license for those less than 65 years of age. Approximately 80% of eligible case women were successfully interviewed. Within each state, controls frequency-matched to cases within 5-year age strata were randomly selected from lists of licensed drivers if less than 65 years of age, or a roster of Medicare beneficiaries compiled by the Centers for Medicare and Medicaid Services (formerly the Health Care Financing Administration) if 65 years of age and older. To be eligible as a control, a woman was required to have a listed telephone number, no personal history of breast cancer and, if less than 65 years of age, a driver’s license. Approximately 76% of eligible case subjects were successfully interviewed. The present analysis is based on the 5,082 case women (3,944 invasive/1,138 in situ
) and 4,501 control women interviewed between 1998 and 2001 during the era of the study when questions on tea consumption were incorporated in the study interview.
Information on tea consumption was ascertained in a structured telephone interview. Women were asked how often, on average, they consumed a cup of black or green tea, not herbal tea, approximately 5 years before the breast diagnosis in cases, or a comparable reference period in controls. Women were permitted to report their usual consumption per day, week, month, or year. The interview also elicited information on known and suspected breast cancer risk factors including reproductive and menstrual history, alcohol consumption, physical activity, height, weight, exogenous hormone use, family history of breast cancer, screening mammography history, and demographic characteristics.
Among the 5,082 cases and 4,501 controls that were asked to report usual tea consumption, data were missing for 23 cases and 15 controls. After excluding these women, a total of 5,059 cases and 4,486 controls were included in the present analysis, of whom 98% were Caucasian. Unconditional logistic regression was used to estimate odds ratios (OR) and Wald 95% confidence intervals (CI) for the risk of breast cancer associated with tea consumption. All models included terms for referent age (in 5-year categories) and study site (Wisconsin, Massachusetts, or New Hampshire). We considered potential confounding by established breast cancer risk factors including menopausal status/age at menopause, parity/age at first birth, body mass index, recency of postmenopausal hormone use, education, alcohol consumption, physical activity, history of benign breast disease, and history of screening mammogram in the 5 years before the reference age. However, results were essentially unchanged after further adjustment for these factors and only age/residence-adjusted results are shown. Tests for linear trend were done by including an ordinal variable for cups per day (0, <1, 1<2, 2<3, 3 or more) in logistic regression models that also included terms for age and state of residence, and in multivariate models, all other covariates.