Imputed genotypes for GENOA studies were tested for association with opioid addiction case–control status, adjusting for sex, age, genotype principal components and in some cases recruitment site and other study-specific covariates. For all cohorts except deCODE, association testing was performed using the score test meta-analysis model of rvtests68. For COGA, which is a family study, an empirically determined kinship matrix was used with rvtests to account for relatedness. For deCODE, data were analyzed using logistic regression treating disease status as the response and imputed genotype counts as covariates. Other available individual characteristics that correlate with disease status were also included in the model as nuisance variables (sex, age, county of origin), blood sample availability, and an indicator function for the overlap of the lifetime of the individual with the time span of phenotype collection) using previously described methods69. To account for inflation due to population stratification and relatedness, test statistics were divided by an inflation factor (1.10) estimated from linkage disequilibrium score regression (LDSR)70.
