Fig 4.

Adjusted odds ratios for different durations of recent and past exposures to hormone replacement therapies in association with breast cancer risk. Odds ratios are with reference to never users and adjusted for smoking status, alcohol consumption, Townsend fifth (QResearch only), body mass index, ethnicity, history of other cancers, oophorectomy or hysterectomy, records of early and late menopause, menopausal symptoms, mammography or scans, family history, comorbidities, other drugs, and years of data. Cases are matched to controls by age, general practice, and index date. Model includes fractional polynomial terms for recent use of oestrogen-progestogen (power 0.5), estradiol-norethisterone (power 0.5), past use of oestrogen-levonorgestrel (power 0.5), and linear terms (1) for all other exposures