For many reasons, defining and classifying psychiatric disorders has always generated debate, disagreement, and controversy. This has certainly been illustrated by the development and publication of sequential editions of the Diagnostic and Statistical Manual of Mental Disorders (DSM) (American Psychiatric Association [APA], 1952). The imminent publication of the fifth edition of the DSM (DSM-5) this year carries on the tradition. Why do we need diagnoses? In general clinical practice, making a diagnosis is a key step in understanding the natural course of a disorder, selecting an appropriate treatment for the disorder, and predicting its response to treatment.
Diagnostic proposals can be evaluated in two main ways: reliability and validity (Kendler, 1990). Reliability refers to the overall consistency of making a diagnosis. This is typically evaluated by determining whether different clinicians using the same criteria can consistently agree on the diagnosis, and whether the same clinician using the criteria at different times will consistently make the same diagnosis. Validity is a much more complex concept to define and to establish. In science, validity generally refers to the extent to which “something” accurately reflects or represents “the nature of reality” (Kendell & Jablensky, 2003, p. 5). For psychiatric disorders, the validity of a diagnosis should represent not just an abnormal or pathological entity that can be distinguished from normal by a natural boundary, but that also can be separated from other disorders by natural boundaries. Kendell and Jablensky (2003) argued that there is little evidence that most psychiatric diagnoses are valid because they are defined by syndromes that have not been demonstrated to have natural boundaries. However, they also believe that diagnoses can be of importance if they provide nontrivial information about prognosis and likely treatment outcomes or if they provide testable hypotheses about biological and social correlates. The reliability and validity of diagnoses are not one and the same, although establishing reliability is usually a necessary step before being able to evaluate and determine validity.
Prior to the 1970s, the lack of reliability of psychiatric diagnoses was a problem in clinical practice but was especially limiting for conducting clinical research. In 1970, Robins and Guze (1970) proposed a method of using five external validators for establishing the validity of psychiatric diagnoses. Their five criteria were: (a) clinical description (symptom profiles, demographic characteristics, precipitating factors); (b) laboratory studies (chemical, physiological, radiological, anatomical, psychological); (c) delimitation from other disorders (exclusion criteria); (d) follow-up studies (determining whether the original clinical presentation persists or changes over time); and (e) family studies. This approach was used by a clinical research group at Washington University in St. Louis, Missouri, to formulate the “Feighner diagnostic criteria” for use in psychiatric research (Feighner et al., 1972). This work in turn formed the basis for the development of the Research Diagnostic Criteria (RDC) (Spitzer, Endicott, & Robins, 1978). In modified form, the RDC were adopted for use in the third edition of the DSM (DSM-III) (APA, 1980).
The Feighner criteria, the RDC, and DSM-III were all commonly used for many years for conducting clinical research, especially pharmacotherapy clinical trials. After the publication of the fourth edition of the DSM (DSM-IV) in 1994, investigators gradually migrated toward the use of the DSM-IV (APA, 1994) criteria in clinical research studies. The use of reliable standardized diagnostic criteria (from any of these classification systems) was an important tool for conducting clinical trials of approved and investigational drugs. One concern, however, is that most diagnoses likely represent heterogeneous syndromes, rather than “pure” or homogeneous disorders. This illustrates the relative lack of diagnostic validity. Heterogeneity within seemingly reliable clinical diagnoses is strongly implied by high rates of comorbid conditions, as well as by the findings from neurobiological research (genetics, brain imaging, cognitive science, and neurophysiology). Clinical and neurobiological heterogeneity within diagnoses may account for a significant proportion of the disparate findings from clinical trials. If this is the case, the effect of a drug may be obscured in randomized controlled trials because the heterogeneous syndrome is composed of “drug-responsive” and “drug-nonresponsive” characteristics.
Because drugs “act” on the brain, they may be used as biological probes. Kendler (1990) included treatment response among his list of external validators that could be considered important for psychiatric disorders: family history, demographic correlates, biological and psychological tests, environmental risk factors, concurrent symptoms that are not part of the diagnostic criteria being assessed, treatment response, diagnostic stability, and, course of, illness.
As part of the development of the DSM-5, the Diagnostic Spectra Study Group examined whether large clusters of diagnoses could be identified by shared external validators (including treatment response) rather than by symptom profiles alone (Andrews et al., 2009). Most DSM-IV disorders were allocated to one of five clusters as a starting premise. Teams of experts then reviewed the literature to determine within-cluster similarities on 11 predetermined validating criteria. The 11 external validators to be considered when grouping related disorders into a cluster were: (a) shared genetic risk factors; (b) familiality; (c) shared specific environmental risk factors; (d) shared neural substrates; (e) shared biomarkers; (f) shared temperamental antecedents; (g) shared abnormalities of cognitive or emotional processing; (h) symptom similarity; (i) high rates of comorbidity; (j) course of illness; and (k) treatment response. A summary of how treatment response fit within these clusters is discussed below.
The final clusters were “neurocognitive” (identified by neural substrate abnormalities), “neurodevelopmental” (identified by early and continuing cognitive deficits), “psychosis” (identified by clinical features and biomarkers for information processing deficits), “emotional” (identified by the temperamental antecedent of negative emotionality), and “externalizing” (identified by the temperamental antecedent of disinhibition). Among the “neurocognitive” cluster disorders (delirium, dementias, amnestic, and others), there is no shared response to treatment. Among the “neurodevelopmental” cluster disorders (mental retardation; learning, motor skills, and communication disorders; and pervasive developmental disorders), there also is no shared response to treatment. The “psychosis” cluster disorders (schizophrenia and related psychoses, bipolar disorders, and schizotypal personality disorder) showed some shared response to treatments. The “emotional” cluster disorders (unipolar depression, dysthymia, generalized anxiety, panic, phobias, obsessive–compulsive disorder, body dysmorphic disorder, hypochondriasis, posttraumatic stress disorder, adjustment disorders, somatoform disorders, and avoidant personality disorder) mostly respond to similar treatments. Finally, among the “externalizing” cluster disorders (substance-related disorders, antisocial and borderline personality disorders, impulse control disorders, and conduct disorder), there is little shared response to treatment.
The National Institute of Mental Health (NIMH) Research Domain Criteria (RDoC) project (NIMH, 2011) is an approach that might be more relevant to understanding the pathophysiology of mental disorders and ultimately may be more relevant to treatment selection. The RDoC project is intended to establish a framework for creating research classifications that reflect functional dimensions stemming from translational research on genes, circuits, and behavior and to “develop, for research purposes, new ways of classifying mental disorders based on dimensions of observable behavior and neurobiological measures” (NIMH, 2011, para. 1). The goal of the RDoC is to shift research away from categorical diagnoses to focus on dysregulated neurobiological systems as the organizing principle for selecting study populations (Insel et al., 2010).
The preliminary RDoC working draft has identified five major domains of functioning, each containing multiple and more specific constructs: (a) “Negative Valence Systems” domain (including constructs for fear, distress, and aggression); (b) “Positive Valence Systems” domain (including constructs for reward seeking and learning and habit formation); (c) “Cognitive Systems” domain (including constructs for attention, perception, working memory/executive function, long-term memory, and cognitive control); (d) “Systems for Social Processes” domain (including constructs for separation fear, facial expression regulation, behavioral inhibition, and emotional regulation); and (e) “Arousal/Regulatory Systems” domain (including systems involved in sleep and wakefulness). For each domain (or for specific constructs within a domain), the patient population selected for research would be based not on diagnoses, but on criteria for the functional domain/construct. Indeed, patients with different clinical diagnoses might very well be recruited for inclusion. The “units of analysis” (levels of investigation) envisioned by the RDoC are genes, molecules, cells, circuits, physiology, behavior, self-reports, and theoretical/hypothetical paradigms. As can be seen, these “units of analysis” range from the lower fundamental level of genes to higher broader levels of assessment, and investigators may choose to study a particular domain/construct at one or more levels of analysis. By generating information from multiple levels of analysis for specific functional domains/constructs, the RDoC project hopes to develop a better understanding of the links between biology and behavior, which will advance our understanding of mental disorders and improve the development and use of treatment interventions.
Current diagnostic systems have not optimally assisted the search for disorder-specific pathophysiological mechanisms and biological and cognitive markers (McGorry & van Os, 2013). By extension, diagnoses do not provide the precision or specificity that clinicians ideally would like when selecting medication. Despite laudable efforts to integrate neurobiological findings, the DSM-5 is not a radical departure from previous diagnostic classification systems (Kupfer & Regier, 2011), and it is therefore unlikely to provide clinical information that better informs treatment selection. However, the changes in the DSM-5 also are not inconsistent with the objectives of the RDoC. It is anticipated that future revisions to the DSM-5 will be based on the findings from the RDoC project and similar efforts (Kupfer & Regier, 2011). After the official publication of the DSM-5, I will have the opportunity to discuss some of these issues in an interview with the DSM-5 Task Force chair, David J. Kupfer, MD, and I will communicate his thoughts in a future article later this year.
- American Psychiatric Association. (1952). Diagnostic and statistical manual of mental disorders (1st ed.). Washington, DC: Author.
- American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author.
- American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.
- Andrews, G., Goldberg, D.P., Krueger, R.F., Carpenter, W.T., Hyman, S.E., Sachdev, P. & Pine, D.S. (2009). Exploring the feasibility of a meta-structure for DSM-V and ICD-11: Could it improve utility and validity?Psychological Medicine, 39, 1993–2000. doi:10.1017/S0033291709990250 [CrossRef]
- Feighner, J.P., Robins, E., Guze, S.B., Woodruff, R.A. Jr.. , Winokur, G. & Munoz, R. (1972). Diagnostic criteria for use in psychiatric research. Archives of General Psychiatry, 26, 57–63 doi:10.1001/archpsyc.1972.01750190059011 [CrossRef] .
- Insel, T., Cuthbert, B., Garvey, M., Heinssen, R., Pine, D.S., Quinn, K. & Wang, P. (2010). Research Domain Criteria (RDoC): Toward a new classification framework for research on mental disorders. American Journal of Psychiatry, 167, 748–751. doi:10.1176/appi.ajp.2010.09091379 [CrossRef]
- Kendell, R. & Jablensky, A. (2003). Distinguishing between the validity and utility of psychiatric diagnoses. American Journal of Psychiatry, 160, 4–12 doi:10.1176/appi.ajp.160.1.4 [CrossRef] .
- Kendler, K.S. (1990). Toward a scientific psychiatric nosology. Strengths and limitations. Archives of General Psychiatry, 47, 969–973 doi:10.1001/archpsyc.1990.01810220085011 [CrossRef] .
- Kupfer, D.J. & Regier, D.A. (2011). Neuroscience, clinical evidence, and the future of psychiatric classification in DSM-5. American Journal of Psychiatry, 168, 672–674. doi:10.1176/appi.ajp.2011.11020219 [CrossRef]
- McGorry, P. & van Os, J. (2013). Redeeming diagnosis in psychiatry: Timing versus specificity. Lancet, 381, 343–345. doi:10.1016/S0140-6736(12)61268-9 [CrossRef]
- National Institute of Mental Health. (2011). NIMH Research Domain Criteria (RDoc). Retrieved from http://www.nimh.nih.gov/research-funding/rdoc/nimh-research-domain-criteria-rdoc.shtml
- Robins, E. & Guze, S.B. (1970). Establishment of diagnostic validity in psychiatric illness: Its application to schizophrenia. American Journal of Psychiatry, 126, 983–987.
- Spitzer, R.L., Endicott, J. & Robins, E. (1978). Research diagnostic criteria: Rationale and reliability. Archives of General Psychiatry, 35, 773–782 doi:10.1001/archpsyc.1978.01770300115013 [CrossRef] .