CD was assessed in all three studies that comprise SAGE using versions of the Semi-Structured Assessment for the Genetics of Alcoholism (SSAGA).31,32 Our primary phenotype was the natural log transformation of the number of CD symptoms [ln(1+x)] that were endorsed (raw symptom counts ranged from 0 to 13). There were 17 individuals missing CD information, yielding a total N of 3963 for analyses. Table 1 shows the distribution of the raw CD symptom counts, split by sex, before transformation. We log transformed the data due to the skew apparent in the symptom counts, and to minimize the impact of outliers at the upper end of the distribution. As expected, males were more likely to endorse CD symptoms than were females; therefore, gender was used as a covariate in all genetic analyses. In addition, we analyzed a dichotomous CD case status variable, because this phenotype and accompanying odds ratios may also be of interest. Case status was defined by the endorsement of ≥3 of the 15 DSM-IV CD criteria under Criterion A, without assessing the requirement of clustering of different symptoms
