We had information on VTE status and GWAS data for 1,364 cases and 17,628 controls within NHS, NHSII and HPFS. The median number of case subjects by dataset was 87.5 and ranged from 16 in the NHSII breast cancer GWAS dataset (total of 289 individuals) to 417 in the type 2 diabetes GWAS dataset (total of 5,773 individuals). The small number of cases in many individual GWAS data sets led to unstable study-specific association statistics. Restricting to studies with an expected case minor allele count >10 for SNPs with a MAF of 0.05 (i.e. studies with at least 200 cases) reduced the sample size to 417 cases and 5,356 controls. However, within each compiled imputed GWAS dataset, VTE case numbers ranged from 406 (OmniExpress) to 532 (Affymetrix). Thus, combining the individual GWAS datasets into three main datasets enabled association analysis of hundreds of cases rather than tens, leading to more stable estimates in the regression analysis. Platform-specific QQ-plots (S5A–S5C Fig) showed no indication of systematic bias (genomic inflation factor λ = 1.00–1.01). The results from the meta-analysis are shown in
