RECENT ADVANCES in the development of the classification of mental disorders through the adoption of a basically "atheoretical," empirically based, descriptive approach have enhanced the reliability, coverage across diverse settings, and clinical and research utility of explicit diagnostic criteria and specified diagnostic algorithms.^1,2 However, the diagnosis of specific disorders is still heavily dependent on diagnostic interviews rather than pathognomonic markers, and convincing clinical and nosological validation is lacking in terms of prognostic value and stability, family and genetic findings, and laboratory findings.³

The goal of broader coverage of current psychopathological classification systems has naturally induced more, sometimes artificial, comorbidity than previous diagnostic systems based on questionable nosological and hierarchical assumptions.⁴ The magnitude of comorbidity between disorders has been consistently highlighted by both epidemiological and clinical studies, irrespective of the type of diagnostic assessment interview used. For example, in the Epidemiological Catchment Area Study⁵ and the National Comorbidity Survey,⁶ the odds of a single disorder increased the odds of a second, and the presence of 3 or more disorders was far more common than expected by chance. Further analyses in adults and adolescents^4,7 demonstrated that patterns of comorbidity were caused by neither a methodological artifact nor help-seeking bias, but might have diagnosis-specific implications for severity, impairment,⁸ course, and outcome.⁹