On categorical diagnoses in DSM-V: cutting dimensions at useful points?

ABSTRACT

The Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-V) will likely place more emphasis on dimensional representation of mental disorders. However, it is often argued that categorical diagnoses are preferable for professional communication, clinical decision-making, or distinguishing between individuals with and without a mental disorder. For these specific aims, utility-based categories can be created on the basis of a dimensional framework by using cut-points. This article addresses several ideas for combining categorical and dimensional approaches like prototype matching, adding scores of symptom-severity, and introducing utility-based categories in dimensional models. The authors identify alternative objectives for specifying cut-points and describe ways of determining the cut-points accordingly. It is recommended that, for creating standard diagnostic concepts, fixed cutoffs be used, as this promotes accumulative science, but these cutoffs may not be optimal for other clinical decisions because of local base rates and decision-specific (dis)utilities. Receiver operator characteristics curves can facilitate the comparative evaluation of the trade-off between sensitivity and specificity for multiple cut-points and diagnostic rules. The authors advocate a DSM-V that contains both categories and dimensions to serve the multiple and complex aims of utility and validity.