Dimensional Versus Categorical Classification of Mental Disorders in theFifth Edition of the Diagnostic and Statistical Manual of Mental Disordersand Beyond: Comment on the Special SectionTimothy A. Brown and David H. BarlowBoston UniversityThe value of including dimensional elements in the Diagnostic and Statistical Manual of MentalDisorders (DSM) has been recognized for decades. Nevertheless, no proposals have been made forintroducing dimensional classification in the diagnostic system in a valid and feasible manner. As aninitial step in this endeavor, the authors suggest introducing dimensional severity ratings to the extantdiagnostic categories and criteria sets. Although not without difficulties, this would begin to determinethe feasibility of dimensional classification and would address some limitations of the purely categoricalapproach (e.g., failure to capture individual differences in disorder severity, and clinically significantfeatures subsumed by other disorders or falling below conventional DSM thresholds). The utility ofincorporating broader dimensions of temperament and personality in diagnostic systems beyond the fifthedition of the DSM is also discussed.Keywords: diagnostic classification of mental disorders, dimensional versus categorical assessment ofpsychopathology, risk factors for mental disorders, comorbidity of mental disorders, Diagnostic andStatistical Manual of Mental DisordersThe purpose of this special section is to promote the develop-ment of dimensional models of diagnostic classification and ap-proaches to research on psychopathology. Although each article inthis series highlights the importance of incorporating dimensionalelements in the formal nosology (the fifth edition of the Diagnosticand Statistical Manual of Mental Disorders [DSM–V] and be-yond), little was said about how this might be accomplished infuture editions of the DSM (but see Widiger, Costa, & McCrae,2002; cited in Widiger & Samuel, 2005). This reflects a longstand-ing predicament. For nearly 30 years, researchers have acknowl-edged the potential utility of such systems (e.g., Barlow, 1988;Kendell, 1975; Maser & Cloninger, 1990; Widiger, 1992). Overthis considerable time span, however, no strong proposals haveemerged with regard to exactly how dimensional classificationcould be introduced in the DSM. Investigators involved in thepreparation of DSM–IV considered and rejected the adoption of adimensional classification in part because “there is yet no agree-ment on the choice of the optimal dimensions to be used forclassification purposes” (American Psychiatric Association[APA], 1994, p. xxii). With the possible exception of the Axis IIdisorders (Widiger & Samuel, 2005), DSM–V may be bound forthe same destiny.As was the case over a decade ago when the DSM–IV waspublished, a tremendous amount of empirical and conceptualgroundwork is needed to understand how dimensional classifica-tion can be validly and practically realized in the DSM. Coredimensions must be identified, measured, and validated. As mostresearchers would concur that the current categorical model ofclassification should not be abandoned entirely (e.g., Brown &Barlow, 2002; Krueger, Markon, Patrick, & Iacono, 2005; Widiger& Samuel, 2005), it must also be determined at what level dimen-sional elements are best incorporated into the diagnostic system(e.g., dimensional severity ratings for the existing DSM disorderconstructs; dimensional assessment of higher order constructs, notcurrently recognized by the DSM, that reflect putative risk factorsfor families of disorders and account for their high rate of comor-bidity; cf. externalization: antisocial behavior and substance usedisorders, Krueger et al., 2005; trait negative affect–neuroticism:anxiety and mood disorders, Brown, Chorpita, & Barlow, 1998;Clark, 2005; Watson, 2005). An equally daunting challenge is thedevelopment of a dimensional assessment system that is widelyagreed upon by DSM investigators and that can be practically andreliably implemented by both clinicians and researchers.The limitations of a purely categorical approach to diagnosticclassification are widely documented. For example, in our workwith the anxiety and unipolar mood disorders, we have encoun-tered many problems with DSM’s categorical diagnostic system(see Brown & Barlow, 2002, for a review). A diagnostic reliabilitystudy of the DSM–IV anxiety and mood disorders (Brown, DiNardo, Lehman, & Campbell, 2001) found that for many catego-ries (e.g., social phobia, obsessive–compulsive disorder [OCD]),diagnostic disagreements less often involved boundary issues withother formal disorders but were primarily due to problems indefining and applying a categorical threshold on the number,severity, or duration of symptoms. This threshold problem ismanifested in various ways. It can be seen in diagnostic disagree-Timothy A. Brown and David H. Barlow, Center for Anxiety andRelated Disorders, Boston University.Correspondence concerning this article should be addressed to TimothyA. Brown, Center for Anxiety and Related Disorders, Boston University,648 Beacon Street, 6th Floor, Boston, MA 02215-2013. E-mail:[email protected] of Abnormal Psychology Copyright 2005 by the American Psychological Association2005, Vol. 114, No. 4, 551–556 0021-843X/05/$12.00 DOI: 10.1037/0021-843X.114.4.551551ments where both raters concur that the key features of a disorderare present but disagree as to whether these features cause suffi-cient interference or distress to satisfy the DSM–IV threshold for aclinical disorder (common with social phobia and specific phobia).The problem is also evident in the high rates of disagreementsinvolving “not otherwise specified” (NOS) diagnoses (both ratersagree on the presence of clinically significant features of thedisorder, but one rater does not assign a formal anxiety or mooddisorder diagnosis because of subthreshold patient report of thenumber or duration of symptoms; common with generalized anx-iety disorder [GAD] and major depressive disorder [MDD]). Asimilar problem is at the root of diagnostic disagreements involv-ing MDD versus dysthymia (core features of clinically significantdepression are observed by both raters, but disagreement occurswith regard to the severity or duration of these symptoms). Finally,whereas dimensional ratings of the severity of MDD symptoms arereliable (r ⫽ .74), the DSM–IV categorical severity specifiers ofthis disorder are