Variable selection
The covariates were selected based on their importance identified from past research, and expert knowledge within the team.29The participants’ baseline characteristics were extracted from the latest records before the study entry date and included age at HRT, birth year, type 2 diabetes, osteoporosis, peripheral arterial or vascular disease (PAD/PVD), coronary heart disease (CHD), hypertension and its treatments, hypercholesterolaemia, oophorectomy or hysterectomy, smoking, body mass index (BMI), and deprivation status. Information about parity (the number of previous births) and age at menopause was not included as it was not reliably recorded in the health records. Medical conditions were identified using the corresponding Read codes which are available online at ClinicalCodes.org.30