We applied confirmatory factor analysis to the covariance matrix generated by LDSC using genomic structural equation modeling (genomic SEM28) with weighted least squares estimation. As in Hatoum et al.5, the indicators were allowed to load freely on a single latent factor (Addiction-Risk-Factor), but updated the OUD26, PAU24, and CanUD25 GWAS. We compared Addiction-Risk-Factor models with DSM-NicDep, PTU, or ICD-TUD as the tobacco-related indicator. The variance of the common latent factor was scaled to 1.0.
