DSM-5 in its proposed form presents a significant shift in the approach to diagnosing personality disorders. The diagnostic criteria outlined in DSM-III and DSM-IV and the introduction of axis II were intended to focus attention on these syndromes in clinical practice and to foster research on their diagnosis, epidemiology, psychobiology, clinical course, and treatment. A diagnostic system should be clinically relevant, encompass the spectrum of personality syndromes seen in practice, facilitate their recognition, and still be simple enough to be used by busy clinicians, including those who do not specialize in the assessment and treatment of personality. At the same time, the diagnostic scheme needs to reflect and support progress in research that leads to increased understanding and better treatment of these illnesses. Regrettably, the proposed system for classifying personality disorders is too complicated, includes a trait-based approach to diagnosis without an adequate clinical rationale, and omits personality syndromes that have significant clinical utility.

The proposed DSM-5 diagnostic scheme for personality disorders is an unwieldy conglomeration of disparate models that cannot happily coexist and raises the likelihood that many clinicians will not have the patience and persistence to make use of it in their practices. The resultant draft criteria encompass 5 levels of personality functioning, 5 personality types, 6 personality trait rating scales, and 4—10 trait rating subscales or facets per trait rating scale.

A clinically useful approach should focus on types of people, not types of ratings scales. The primary unit of diagnosis should be a personality syndrome—a configuration or pattern of functionally interrelated personality processes encompassing cognition, affectivity, interpersonal functioning, behavior, coping, and defense. Mental health professionals think in terms of syndromes or patterns (as recognized by all previous versions of the DSM), not in terms of deconstructed subcomponents or in terms of 30-plus separate trait dimensions to be rated (as in the current DSM-5 proposal). Clinicians see coherent patterns of interrelated processes where untrained persons may see confusion. The diagnostic assessment should also acknowledge gradations of severity, as the proposed revision does (a welcome improvement over DSM-IV). A narcissistically disordered patient may be mildly socially impaired or so severely impaired as to be unable to engage effectively in any type of personal interaction.

The prototype approach proposed for DSM-5 provides descriptions of five personality disorders: antisocial/psychopathic, avoidant, borderline, obsessive-compulsive, and schizotypal. There is empirical support for the usefulness of a prototype approach to diagnosis: research in cognitive science tells us that diagnostic decision making, which is inherently a judgment about category membership, generally relies on prototype matching (1—6). Diagnosticians develop cognitive prototypes of diagnostic syndromes, and they make diagnoses by gauging the match between an individual and a particular prototype. Cognitive prototypes are syndromal constructs that capture many different but interrelated features of a personality syndrome. In this arena, the DSM-5 Personality and Personality Disorders Work Group has been attentive to the need for an approach that can work with, rather than against, the cognitive processes of the clinicians who will use it.

The DSM-5 proposal combines this clinically grounded prototype approach with a second multidimensional assessment model organized around trait dimensions (rating scales) rather than syndromes. Trait models developed within academic psychology and have their origin in research on normal (not clinical) populations. The six proposed dimensions are 1) negative emotionality, which includes "facets" of depression, anxiety, shame and guilt; 2) introversion, which includes withdrawal from social interaction; 3) antagonism, which includes an exaggerated sense of self-importance; 4) disinhibition, which includes impulsivity; 5) compulsivity, which includes perfectionism and rigidity; and 6) schizotypy, which includes odd perceptions and beliefs. These traits represent an effort to synthesize and to make compromises between various data-based models. However, the proposed trait dimensions and facets and the overall assessment system have not been investigated empirically. The resulting model no longer rests on decades of research, which had been the chief rationale for including it.

Moreover, there is good reason to doubt that trait-based systems, even if validated, could be transformed into a clinically useful diagnostic system. While we favor continued research on dimensional trait approaches, the creation of the diagnostic system of DSM-5 must recognize that clinicians find dimensional trait approaches significantly less relevant and useful, and consider them worse, than the current DSM-IV system (7, 8). The combination of a syndromal prototype approach with a second, conceptually distinct dimensional trait approach needlessly complicates the DSM-5 proposal, and we think will render an already underutilized system for personality diagnosis unworkable in clinical practice. Dimensional trait strategies are certainly valuable research tools, particularly in identifying relationships among variables in a population, but these should be provided in a way that does not interfere with the primary clinical utility of the DSM in describing the interrelationships of psychological processes in an individual.

We share a further concern that the five personality prototypes proposed by the Work Group are insufficient to encompass the spectrum of personality pathology seen in the community. Absent from the diagnostic system are narcissistic, paranoid, dependent, and histrionic (historically termed "hysterical") personality disorder. Combinations of the proposed dimensional trait rating scales will not easily yield the omitted syndromes (8), for which a significant amount of empirical data and an even more significant body of clinical wisdom has accumulated over the past decades (9—11).

We strongly advocate that the prototype system be expanded to encompass the range of personality syndromes seen in the community and identified empirically (9—11). Additionally, prototype descriptions that have emerged in empirical research in national samples, written in clinically familiar language, should not be altered by the addition of terminology from dimensional trait models that has not been tested empirically for use in prototype descriptions. We presume that certain personality disorders have been omitted because of limited available research, but absence of evidence is not evidence of absence. For example, narcissistic personality disorder may have received less empirical research attention because samples are hard to obtain outside of clinical practice settings, and because it is difficult to assess via the self-report methods such as questionnaires and structured interviews that are used by most personality researchers. If this rationale is used to exclude specific well-known, widely seen syndromes, disorders commonly seen in practice may disappear from the proposed revision, potentially driving a wedge between clinical reality and officially sanctioned diagnoses.

This commentary represents the consensual, collective views of experienced senior clinicians and researchers of diverse backgrounds who have expertise in personality disorders. We want to assure that personality assumes a more prominent and rightful place in psychiatric diagnosis and treatment. We are concerned that a dimensional trait approach will not work for real-world clinical diagnosis. We are additionally concerned that the DSM-5 proposal includes too few personality disorder prototypes to cover the spectrum of personality pathology seen in practice. We endorse the value of dimensionalizing well-recognized clinical constructs (e.g., severity), but do not agree with the use of non-clinical concepts drawn from academic personality psychology.

Unless the system for diagnosing personality syndromes is recast in terms that mental health professionals recognize as clinically useful and relevant, and unless the methods for making diagnoses are designed with real-world usability and time efficiency in mind, we may see personality increasingly bypassed and neglected in clinical practice, to the detriment of comprehensive understanding of psychiatric patients. In too many psychiatric records today, the entry after axis II seems to be "deferred." We strongly advocate a system that will reverse, not exacerbate, this unfortunate trend.