where T is survival time (in years); U is a random variable following a uniform distribution on [0, 1] A is age measured in years; G is genotype coded as 0, 1 or 2 according to the number of risk alleles that the individual has; βA is the log(HR) for a 1 year increase in age; and βG is the log(HR) for a one allele increase in the risk allele. Approximate estimates of βA and λ for CHD (fatal and non-fatal) were obtained by fitting a Weibull regression model with age as the only covariate in males from the EPIC-CVD subcohort. βG was set equal to the logarithm of the HR for the hypothetical SNP in each simulation scenario. The shape parameter (v) was scaled according to the RAF and underlying HR for the hypothetical SNP to ensure a number of cases that approximately reflects the cumulative disease incidence in each simulation scenario.
