Due to potential cross-ancestry differences in linkage disequilibrium patterns and minor allele frequencies, we calculated PRS and analyzed the initial regression models in each ancestry group separately to avoid possible confounding. To maximize sample size and power, we then used a fixed-effects model to meta-analyze these results across the European and African ancestries.38 We corrected for a total of 61 tests (details in online supplementary materials). Using a conservative Bonferroni correction, we consider P-values < .05/61 = 8.2e−4 to be statistically significant.
