All of the data preparation and descriptive analyses were conducted using SAS. STATA was used for logistic regression analyses examining the association between EAU and self-harm variables. First, self-harm was characterized by examining the overlap between SA and NSSI, its association with EAU (adjusting for family structure), as well as methods of SA and NSSI among those with and without a history of EAU. Correlations between self-harm and EAU and other clinical covariates were then examined to identify variables to include in subsequent analyses. The prevalence of EAU and clinical covariates was also reported in individuals who endorsed self-harm versus those individuals who denied a lifetime history of self-harm. We then examined the prevalence of self-harm across four groups: 1) twins concordant for early onset alcohol use (i.e., prior to age 15), 2) twins concordant for not using alcohol before the age of 15, 3) individuals from discordant twin pairs who reported early onset alcohol use, and 4) individuals from discordant twin pairs who did not drink before age 15. Prevalence of clinical covariates across these groups was also examined,
