Hazard ratios of breast cancer when data from the randomized WHI study are analyzed as a fictive observational study.

Model 1 is a regression model with 3 year follow-up (scenario 1); model 2 is the same model assuming that 10% starts using hormones every year in the non-user group (scenario 2); in model 3 (scenario 3) it is assumed that the incidence rates in the fictive study were identical to that observed for all 16 608 women of the WHI trial. In parenthesis we also give HR when assuming that 5% start using hormones every year in the non-user groups.Hazard ratios of breast cancer when data from the randomized WHI study are analyzed as a fictive observational study.