However, the null finding may be an artifact of statistical collinearity. MDD, bipolar disorder (Skinstad and Swain, 2001), PTSD, panic/agoraphobia disorder, and ASPD (Markon and Krueger, 2005) frequently are comorbid with most of the substance dependence diagnoses assessed in this study. Thus, the strong relationship of those other substance dependence diagnoses with cocaine dependence may have obscured the relationship of ASPD or bipolar disorder with cocaine dependence severity. The finding in previous studies of a relationship between psychiatric morbidity (e.g., PTSD; Parrott et al., 2003) and increased substance dependence severity or impairment was evident in unadjusted bivariate analyses (results not reported here but available from the first author), in which each psychiatric disorder assessed in this study was significantly associated with at least one (and usually several) indices of cocaine dependence severity. This suggests that research may be needed to determine where indirect relationships link psychiatric and substance dependence disorders with cocaine dependence severity (e.g., using structural equation modeling). For example, dependence on substances (e.g., alcohol, cannabis, opioids) that are often used to self-medicate PTSD-related anxiety (Tarrier and Sommerfield, 2003)
