There are several limitations to this analysis. First, primarily alcohol-dependent subjects in treatment, their relatives and control families were enrolled into this study. This might explain the rather high rate of alcohol and substance dependence diagnosis in this but may also indicate that a severely comorbid sample with bipolar disorders and several comorbidities in these analyses. Second, several previous investigations included first-episode manic patients to overcome potential bias caused by the number of affective episodes and chronic course of bipolar disorders. The COGA sample included inpatient subjects with and control subjects without alcohol dependence. The enrolees were not first-episode bipolar subjects, compared to other samples [43]. Thus, individuals in a different stage of their disease and more chronic bipolar disorder individuals were recruited. Chronic patients are reported to have a higher rate of previous affective episodes and other comorbidities which in turn increase the likelihood of future affective episodes. When these individuals at different stages of their disease are investigated, it may be more difficult to identify other course predictors and to evaluate the influence of a comorbid disorder on prognosis of bipolar disorder.
