Background: Knowledge regarding the causes of comorbidity between two disorders has a significant impact on research regarding the classification, treatment, and etiology of the disorders. Two main analytic methods have been used to test alternative explanations for the causes of comorbidity in family studies: biometric model fitting and family prevalence analyses. Unfortunately, the conclusions of family studies using these two methods have been conflicting. In the present study, we examined the validity of family prevalence analyses in testing alternative comorbidity models.
Method: We reviewed 42 family studies that used family prevalence analyses to test three comorbidity models: the alternate forms model, the correlated liabilities model, or the three independent disorders model. We conducted the analyses used in these studies on datasets simulated under the assumptions of 13 alternative comorbidity models including the three models tested most often in the literature.
Results: Results suggest that some analyses may be valid tests of the alternate forms model (i.e., two disorders are alternate manifestations of a single liability), but that none of the analyses are valid tests of the correlated liabilities model (i.e., a significant correlation between the risk factors for the two disorders) or the three independent disorders model (i.e., the comorbid disorder is a third, independent disorder).
Conclusion: Family studies using family prevalence analyses may have made incorrect conclusions regarding the etiology of comorbidity between disorders.