problems in later life. We note that, unlike COGA, Biobanks lack some variables of interest that limit our ability to evaluate findings in as much depth as is possible in COGA, which includes AUD severity and family history of remission. Biobanks are collections of electronic health records designed for clinical convenience and are commonly used to quickly increase sample sizes to study the genetic and other underpinnings of many diseases, and more effort should be devoted to collect study-related covariates. However, longitudinal and deeply phenotyped research cohorts such as COGA are still critical to study the risk and resilience factors associated with AUD and related conditions. These research cohorts cannot be replaced by Biobanks and their sample sizes, especially those from under-studied populations, should be dramatically increased.
