outcomes that have been collected for individuals in multiple cohorts. In particular, the accumulation of GWAS data within large cohorts with rich environmental and outcome data creates new opportunities to assess novel hypotheses. In addition, cohort studies provide unique opportunities to prospectively assess biomarker-disease associations, thereby minimizing bias due to reverse causation or treatment effects. However, “borrowing” GWAS data between traits is not straightforward. Known issues that can cause bias include technical artifacts due to different genotyping platforms, differences in imputation accuracy and ascertainment bias. Thus, careful data management, imputation procedures and quality checks are needed. Furthermore, if the secondary trait is rare, there will be low phenotypic variability within each GWAS dataset. For example, we observed fewer than 100 VTE cases within the majority of individual GWAS, compared to more than 400 cases within each combined dataset.
