Covariates included gender (coded male and female), age at last assessment, generational cohort (dummy coded using the following scheme from Bourdon et al.35: silent [b. prior to 1946], baby boomer [b. 1946 to 1964], generation X [b. 1965 to 1980], millennial [b. 1981 to 1996]); and the first ten within-ancestry principal components. Principal components are often used to address population stratification in PRS analyses36. Similar to other recent studies37–40 we used 10 ancestry principal components as covariates.
