Psychiatric comorbidity and treatment seeking. Sources of selection bias in the study of clinical populations.

Population studies have shown that the co-occurrence of psychiatric disorders increases the likelihood of treatment seeking. This leads to a biased estimation of the prevalence of comorbidity in clinical samples, and this overestimation can be attributed to two different sources of selection bias. Using data from a population survey of psychiatric disorders, in which 3258 residents of Edmonton, Alberta, Canada, were interviewed with the Diagnostic Interview Schedule, we assessed the extent of each of these two mechanisms. The first source of selection bias is the mathematical bias known as Berkson's bias and arises from the fact that an individual affected with two psychiatric disorders can seek treatment for either one or the other of these disorders. The second source of selection bias is clinical and results from the changed probability of seeking a treatment for a specific disorder because of the existence of a comorbid disorder.