Online Data Supplements: • Online methods. • Supplementary Table S1. A selection of previous studies aiming to estimate the effect of oral contraceptive use on risk of breast cancer. • Supplementary Table S2. Identification of cancer cases in the UK biobank database. • Supplementary Table S3. Identification of covariates in the UK biobank database and their Odds ratio [OR] and 95% confidence interval [CI]) on odds of cancer. • Supplementary Table S4. Fixed (time-independent) associations of oral contraceptives, during and after use, as analysed by Cox regression modelling. • Supplementary Table S5. Fixed (time-independent) associations of oral contraceptives during and after use, as analysed by Cox regression modelling with age as primary time scale (sensitivity analyses). • Supplementary Table S6. Time-dependent associations of oral contraceptives during use, as analysed by Cox regression modelling. • Supplementary Table S7. Time-dependent hazard ratios of oral contraceptives after last use. • Supplementary Figure S1. Sensitivity analyses for the associations between oral contraceptive use and cancer in the Cox regression. • Supplementary Figure S2. Discrepancy between self-reported data and cancer register. • Supplementary Figure S3. Power calculations. • Supplementary Figure S4. Distributions for age at menopause. • Supplementary Figure S5. Sensitivity analyses for different follow-up years after oral contraceptive discontinuation.
ARTICLE ABSTRACT
Oral contraceptive use has been suggested to influence the risk of breast, ovarian, and endometrial cancer. The purpose of this study is to clarify the time-dependent effects between long-term oral contraceptive use and cancer risk. We performed an observational study in 256,661 women from UK Biobank, born between 1939 and 1970. Information on cancer diagnoses were collected from self-reported data and from national registers until March 2019. Cumulative risk of cancer over the timespan of the study, as measured by the OR, and instantaneous risk, as measured by the HR, were assessed using Logistic and Cox regression analyses, respectively. The odds were lower among ever users, compared with never users, for ovarian cancer [OR = 0.72; 95% confidence interval (CI), 0.65–0.81] and endometrial cancer (OR = 0.68; 95% CI, 0.62–0.75), an association that was stronger with longer use (P < 0.001). Increased odds were seen for breast cancer in women when limiting the follow-up to 55 years of age (OR = 1.10; 95% CI, 1.03–1.17), but not for the full timespan. We only found a higher HR for breast cancer in former users immediately (≤2 years) after discontinued oral contraceptive use (HR = 1.55; 95% CI, 1.06–2.28), whereas the protective association for ovarian and endometrial cancer remained significant up to 35 years after last use of oral contraceptives. Given the body of evidence presented in our study, we argue that oral contraceptives can dramatically reduce women's risk of ovarian and endometrial cancer, whereas their effect on lifetime risk of breast cancer is limited.
These results enable women and physicians to make more informed decisions considering oral contraceptive use, thus constituting an important step toward personalized medicine.
