A restriction of this study concerns the relatively small number of individuals with a diagnosis of ADHD, CD or AUD, which may have caused a lack of statistical power. However, the present study used a large population Trichostatin A purchase based sample. This enabled us to compare relatively small numbers of diagnosed individuals with large numbers of undiagnosed individuals. The many significant associations as well as the generally narrow confidence intervals suggest that statistical power was sufficient. Previous research among adolescents showed that the three ADHD subtypes (i.e., inattentive, hyperactive, and combined) had different associations with AUD (Elkins

et al., 2007 and Molina and Pelham, 2003). However, due to the small amount of respondents with ADHD in present study we were not able to assess the possible differential contribution of the three ADHD subtypes. Also, we were unable to conduct separate analyses for alcohol abuse and dependence. Only a small group of respondents developed alcohol dependence, which is characterized by different symptoms as well as a higher symptom count than alcohol abuse (number of symptoms occurring within a 12-month period ≥3 in dependence vs. ≥1 in abuse). The associations with ADHD and CD could thus be different for both AUDs. Previous

research suggested, however, that this is not the case (Fergusson et al., 2007). Diagnoses of ADHD, CD, and AUD were based on retrospective reports, as is often the case in population Selleck FRAX597 studies. Retrospective assessment could have resulted in recall bias. However, it is unclear how

this would affect the presented associations. In accordance with earlier research (Kessler et al., 2007), we choose to restrict our sample to respondents aged 18–44, to minimize problems with recall bias. Approaches using multi-informant information could have resulted in other prevalence rates of ADHD as compared to the self reports that were used in present research. However, an earlier comparison between adult self-reports and informant reports of childhood and adult ADHD showed fairly strong associations between the two (Murphy and Schachar, 2000). The use of self-reports in present research seems therefore justified. Notwithstanding GPX6 the potential limitations, this study helps to understand how ADHD is associated with alcohol use (disorder), and how CD affects this association. Replication of the current findings is needed, preferably in longitudinal design, so that the progression from ADHD to CD and subsequent to AUD can be further examined. The current paper treated ADHD, CD, and AUD as separate disorders. However, some studies have suggested that these disorders reflect a general dimension of externalizing behavior (Kendler et al., 2003 and White et al., 2001). Future research should study this dimension and the possibility that current findings of mediation represent a phenotypic phased expression of this partially genetically determined (Hicks et al., 2007, Kendler et al.