In order to carry out downstream analyses, including heritability, gene prioritization, enrichment, and genetic correlation, we performed a fixed-effects meta-analysis for samples of European ancestries (525,197 cases and 3,362,335 controls), using a large single linkage disequilibrium (LD) reference dataset. The consequences of MD phenotyping on the meta-analyses were examined using genomic structural equation modeling (SEM) with a common-factor meta-analysis of the European ancestry summary statistics in genomic SEM13 (Figure S1). Cohorts were first meta-analyzed based on how the MD phenotype was determined: clinical/interview, electronic health record [EHR], questionnaire, or self-report of MD diagnosis. The proportion of total effective sample size contributed by each phenotype definition was 4% clinical/interview, 54% EHR, 14% questionnaire, and 27% self-report. The different phenotype definitions of MD had strong genetic correlations (LD score rg from 0.78 to 0.88). We fitted a common-factor model in genomic SEM and set the clinical/interview phenotype as the primary phenotype by fixing its factor loading to 1 and its residual variance to 0. This factor model was consistent with the data (χ32=4.49, p = 0.213); therefore, we could not reject the
