Selection of controls
Controls were non-users of HRT or any type of drugs containing oestrogen and/or progesterone at baseline. Controls were matched with cases in one to up-to-three ratio by age and GP practice. The study entry date for controls was the first HRT prescription date of their matched cases.