diseases (e.g. CHD and type II diabetes) often seen over a 20-year period for the age group studied in the simulations. The RAFs examined were 0.05, 0.10, 0.25, 0.50, 0.75, 0.90 and 0.95. The strength of the association was varied over HRs of 1.05, 1.10, 1.15, 1.20, 1.30, 1.50 and 2.00. To examine the type I error rates of the two approaches, we simulated under an HR of 1. In each simulation scenario the simulation under the null yielded a type I error of approximately 0.05 for both the Cox and logistic regression models (Supplementary Tables S1 and S2).
