Despite the strengths of the study, certain limitations should be noted. Some of our Yale-UPenn sample was recruited for studies of drug dependence rather than alcohol dependence, while the SAGE sample was primarily recruited for alcohol dependence. Although we adjusted for these comorbidities, greater substance dependence comorbidity in the Yale-UPenn sample may have contributed to the differences in findings with respect to the SAGE sample. We note that many prior studies have not taken note of substance use comorbidity at all, and the assumption that there is less comorbidity in those cases where it is not reported, is not necessarily warranted. We also acknowledged the proportion of sex and races are slightly different between two datasets. In the meta-analysis, we analyzed associations separately in EAs and AAs, and adjusted sex and other demographics.
