Bra Wearing Not Associated with Breast Cancer Risk: A Population-Based Case–Control Study

Study population

We used data from a population-based case–control study involving postmenopausal women living in the three-county Seattle-Puget Sound metropolitan area (King, Pierce, and Snohomish Counties). The study was originally designed to evaluate differences in risk factors for the two most common histologic subtypes of breast cancer, invasive ductal carcinoma (IDC) and invasive lobular carcinoma (ILC). The details of subject recruitment and data collection have been published previously (3). Briefly, cases were women between 55 and 74 years of age first diagnosed with invasive breast cancer between January 1, 2000, and March 31, 2004, while residing in the Seattle-Puget Sound area. The Cancer Surveillance System, the region's population-based cancer registry also participating in the Surveillance, Epidemiology, and End Results program of the National Cancer Institute, was used to identify cases. All ILC cases (identified using ICD-O histology codes of 8520, 8522, and 8524) and a random sample of 25% of the IDC cases (identified using ICD-O histology code of 8500) were targeted for recruitment to enroll equal numbers of ILC and IDC cases. A total of 1,044 out of 1,251 eligible cases were interviewed (83%), consisting of 454 IDC and 590 ILC cases. A common control group, frequency matched 1:1 to the ILC cases within 5-year age groups, was selected from the general population of women living in the three-county area by random-digit dialing. We called a total of 29,735 telephone numbers of which 9,876 were verified as residential. Of these residential numbers, 87% were successfully screened for study eligibility. Four hundred and sixty-nine out of 660 telephone-screened eligible controls (71%) completed the interview.

Data collection

The study protocol was approved by the Fred Hutchinson Cancer Research Center Institutional Review Board and written consent was obtained from all study participants. In-person interviews were conducted with both cases and controls to collect information on reproductive history, body size, medical history, family history of cancer, use of hormonal replacement therapy, other potential breast cancer risk factors, and demographic characteristics. Women were asked a series of structured questions to assess lifetime patterns of bra wearing, including bra cup sizes and band sizes, age at which they started regularly wearing a bra, whether they wore a bra with an underwire, number of hours per day, and number of days per week they wore a bra at different times in their life, and if these patterns ever changed. Through these questions, we quantified both lifetime and recent bra-wearing habits. Data on bra-wearing habits were limited to those practiced before each participant's reference date. Date of breast cancer diagnosis was the reference date used for cases, and for controls, a reference date was assigned to reflect the distribution of reference dates among the cases.

Statistical analysis

Various bra-wearing characteristics were categorized on the basis of their distributions in our study population, including bra cup sizes 1 year before the reference date (A, B, C, and D or above), current and lifetime average hours/day wore a bra (categorized into quartiles based on the control distribution), ever and current use of an underwire bra (yes/no), current average hours/day wore a bra with and without an underwire (categorized into quartiles based on the control distribution), and age first began regularly wearing a bra (12 years or younger, 13–14 years, and 15 years or older). There was one participant who reported that she never wore a bra and she was excluded from the analysis. There were seven women who did not currently wear a bra and they were included in our lifetime bra wearing analyses but excluded from the analyses of current bra-wearing habits.

We used polytomous logistic regression to estimate odds ratios (OR) and their associated 95% confidence intervals (CI) comparing IDC and ILC cases with controls (4). P values for trend were computed across categories of bra-wearing duration and age first began wearing a bra. All analyses were conducted using Stata/SE version 13.1 (StataCorp LP). All models were adjusted for age at the reference date (5-year categories), reference year (continuous), and county as controls were frequency matched to cases on these three factors. Several covariates were evaluated as potential confounders, including race/ethnicity, education, annual household income, body mass index (BMI) at age 30 and BMI 1 month before the reference date, bra band size, age at first full-term pregnancy, use of hormone therapy, types of menopause, parity, family history of breast cancer, and mammogram screening in the past 2 years. BMI at the age of 30 was used as a proxy for women's weight status at a young age due to the lack of weight and height information for women when they were adolescents. None of the above variables changed the ORs of interest by more than 10% and, thus, none were included in the final statistical models. In addition, none of the covariates were found to modify the relationships between various bra-wearing characteristics and breast cancer risk based on likelihood ratio tests, including bra cup sizes, BMI, use of hormone therapy, and mammogram screening in the past 2 years (all P_interaction > 0.05). A sensitivity analysis restricted to women who did not change bra-wearing habit during their lifetime was performed and the results were similar to those based on the whole sample (data not shown).