Here, we propose an adjustment for index event bias in GWAS of subsequent events. The main insight is that confounder effects are approximately constant across SNPs and can be estimated by regressing SNP effects on prognosis on SNP effects on incidence. We illustrate our approach in a GWAS of survival with idiopathic pulmonary fibrosis (IPF) and re-analyse a GWAS of Crohn’s disease prognosis.
