Stressful life events and personality disorders in etiology of alcohol use disorder: a twin study using a novel biometric version of the statistical mediation model

Alcohol use disorder (AUD) is a common psychiatric disorder that induces a heavy burden on public health. It represents a failure of acceptable social behavior and of maintaining moderation. DSM-IV Antisocial and Borderline... [ view full abstract ]

Alcohol use disorder (AUD) is a common psychiatric disorder that induces a heavy burden on public health. It represents a failure of acceptable social behavior and of maintaining moderation. DSM-IV Antisocial and Borderline Personality Disorders (PDs) and associated conduct problems and impulsive behaviors are known risk factors of AUD. Existing evidence indicates that childhood stressful life events (SLEs) play a role in the development of all these disorders. While it may seem likely that PD traits mediate at least some of the effects of childhood SLEs on adulthood AUD, demonstrating causality has turned out difficult because of possible genetic confounding. The statistical mediation model assumes lack of such confounding, and is frequently used to assess hypotheses regarding causal chains of events. Here we extend the model to a biometric version that can both test the assumed lack of confounding and estimate the mediation effects. The extended model is then applied to interview and survey data on PDs, AUD, and SLEs on 2801 Norwegian Twins drawn from the Norwegian Institute of Health Twin Panel data. Our findings indicated that Antisocial PD and its precursor, childhood conduct disorder, mediated the (primarily shared environmental) effects of childhood SLEs on AUD, with no indication of confounding. The mediation path via Borderline PD was confounded, whereas self-harming impulsivity (a component of Borderline PD) again mediated SLE effects on AUD. The total phenotypic effect of SLE count on AUD was ~0.24 standard deviations in AUD liability (CI = 0.11–0.37), more than half of which was mediated by the PD traits. In conclusion, development of antisocial and impulsive traits may explain a large part of the AUD risk related to childhood adversity, whereas the role of borderline traits in this etiologic pathway remains unclear.