In the simulation study, we assessed the performance of the Cox and logistic regression models (details of these models are given in the Supplementary Text) in terms of bias, precision and statistical power for cohort and case-cohort studies. The computational time needed to perform a GWAS using both models in cohort and case-cohort studies was also assessed using simulated data. We further compared the Cox and logistic regression models using SNP–CHD associations in EPIC-CVD for loci that have previously been robustly associated with CHD.1 All analyses were performed in R (version 3.1.0) on a 24 core Linux server (1.8 GHz, 124Gb RAM).
