Basing psychiatric classification on scientific foundation: problems and prospects.

To examine whether and how the classification of mental disorders can be based on research, we evaluate the relevance of psychiatric science to the major questions in classification. We conclude that most studies cannot inform the validity of diagnostic categories because they are constrained by the classification through a top-down diagnostic approach. Analyses of relationships between diagnostic categories suggest that most interdiagnostic boundaries in current classifications lack validity. Likewise, genetic studies show that the susceptibility to mental illness is at most partly disorder-specific. Neuroimaging research is uninformative due to unsystematic single-diagnosis studies, use of super-healthy controls, and publication bias. Treatment research suggests moderate specificity in several areas of psychopathology (e.g. lithium for bipolar disorder), but lack of specificity is the rule (e.g. the broad indications of serotonin-reuptake inhibitors). In summary, evidence from multiple lines of research converges to indicate that current classifications contain excessively large numbers of categories of limited validity. Dimensional classification will not solve the problem because the number of dimensions is as uncertain as the number of categories. Psychiatric research should discard the assumption that current classification is valid. Instead of diagnosis-specific investigations, studies of unselected groups assessed with bottom-up approaches are needed to advance psychiatry.