The article in this issue by Rottman and colleagues (1) is timely. It significantly informs the current debate about how to improve the representation of personality disorders in DSM by comparing two proposed alternative models of assessment on their clinical utility. The limitations of DSM categorical conceptualizations of personality disorders are well known: excessive co-occurrence among disorders, extreme heterogeneity among patients receiving the same diagnosis, arbitrary diagnostic thresholds for the boundaries between pathological and “normal” personality functioning, and inadequate coverage of personality psychopathology such that the diagnosis of personality disorder not otherwise specified (PDNOS) is the most common. The findings by Rottman et al. underscore the need for any proposed revision of the approach to personality psychopathology in DSM to proceed with caution and with appropriate deference to its ultimate purposes—namely, to increase clinical utility and thereby improve patient care (2).
Kupfer et al. (3) argued that the categorical approach to the diagnosis of mental disorders, including personality disorders, needs reexamination. No laboratory marker has been found to be specific for any DSM-defined syndrome. Epidemiologic and clinical studies show high rates of comorbidity, within and across axes, and short-term diagnostic instability. Also, a lack of treatment specificity for individual disorders is the rule rather than the exception. Whether mental disorders should be represented by sets of dimensions of psychopathology, rather than by multiple categories, was identified as one of seven basic nomenclature issues needing clarification for DSM-V.
Rounsaville et al. (4) elaborated, “There is a clear need for dimensional models to be developed and their utility compared with that of existing typologies in one or more limited fields, such as personality. If a dimensional system performs well and is acceptable to clinicians, it might be appropriate to explore dimensional approaches in other domains” (p. 13). Thus, personality disorders have become a test case for dimensional approaches to the diagnosis of mental disorders in DSM-V.
Dimensional models, however, are unfamiliar to clinicians trained in the medical model of diagnosis, in which a single diagnostic concept is used to communicate a large amount of clinical information about a patient’s problems, indicated treatment, and likely prognosis. Proponents of dimensional models point out that some clinical phenomena in medicine, such as blood pressure, have continuous distributions but lead to a meaningful categorical diagnosis (viz., hypertension) once cut-points with significance for morbidity and need for treatment are established. And, closer to home for psychiatry, meaningful cut-points based on progressive degrees of functional impairment exist for extreme (low) values of intelligence.
Based on theories of cognitive science, the study by Rottman et al. was designed to compare the clinical utility of so-called “variable-centered” or trait-based versus “person-centered” or syndromal dimensional approaches to characterizing personality disorders. Previously, Widiger and Simonsen (5) reviewed 18 alternative proposals for dimensional models of personality disorders. The proposals included 1) dimensional representations of existing personality disorder constructs, either by criteria counts or by diagnostic prototypes, 2) dimensional reorganizations of diagnostic criteria, 3) integration of axes II and I by means of common psychopathological spectra, and 4) representation with trait-based dimensional models of general personality functioning. Whether they are variable-centered or person-centered is a major difference between dimensional approaches to personality disorder diagnosis. In variable-centered approaches, multiple individual personality traits are rated on the degree to which they describe a given patient. In person-centered approaches, a patient is compared to a description of a prototypic personality type, and the degree of the descriptive match is rated on a dimensional scale. Trait-based dimensional models of personality psychopathology make the co-occurrence of personality disorders and their heterogeneity more rational because they include multiple continuous dimensions on which all people vary. The configurations of dimensional ratings describe each person’s personality profile. Consequently, many different multidimensional configurations are possible. Trait dimensional models were developed to describe the full range of personality, so it should be possible to describe anyone. Prototype matching allows for both dimensional and categorical representations and parallels diagnosis in other areas of medicine with varying degrees of syndrome severity.
Each of these approaches, represented by the trait-based Five-Factor Model (FFM) of general personality functioning (6) and by dimensional prototypes derived from the Shedler-Westen Assessment Procedure (SWAP-200) (7) and DSM-IV, respectively, were compared in the Rottman et al. study. In terms of the FFM, personality disorders, in general, would be characterized by high neuroticism; a specific personality disorder, such as borderline personality disorder, would also be characterized by low agreeableness and low conscientiousness (8). The FFM has been used extensively to study the stability of personality across the lifespan, gender differences in personality, childhood temperament, and the relationship of personality to important life outcomes (9). Rating of descriptive prototypes of personality styles and disorders are dimensional alternatives to trait-based descriptions. A patient with suspected borderline personality disorder would be compared to a narrative description of a personality style that encompasses prototypic borderline personality characteristics. In previous studies, findings on the clinical utility of the two approaches have been variable. Both dimensional alternatives have been rated as more useful than the current DSM categorical system. The SWAP prototypes have been found to be significantly related to functional outcomes and treatment response (7). Prototypes have also been found to be user-friendly and have received high approval ratings from clinicians when compared to trait-based systems, including the FFM (10).
If we assume that a dimensional approach has much to offer in improving the understanding and diagnosis of personality psychopathology, which dimensional system should it be? Here the article by Rottman et al. makes a significant contribution. They found that clinicians made fewer correct diagnoses of personality disorders and more incorrect diagnoses when given ratings of patients on a list of the 30 facet traits of the FFM than when given prototype descriptions based on either the SWAP or DSM-IV criteria. If clinicians are unable to recognize common clinical syndromes using a new trait-based system, how can it be more clinically useful? And in fact, on most questions about clinical utility, including about treatment planning and prognosis, the prototype systems were rated as superior. According to the authors, these findings indicate that personality traits in the absence of clinical context are too ambiguous for clinicians to interpret: although it may be possible to describe personality disorders in terms of the FFM, mentally translating personality traits back into syndromes or disorders is cognitively challenging.
In 2007, the Personality and Personality Disorders Work Group was appointed to consider the future of personality disorder assessment and classification in DSM-V. Key questions were articulated to inform potential revisions: What is the core definition of a personality disorder that distinguishes it from other types of psychopathology? Is personality psychopathology better described by dimensional representations of diagnostic categories or by extremes on dimensions of general personality functioning than by the categories themselves? Is a separate axis II for personality assessment valuable; and what is the clinical importance (for risk, treatment, or prognosis) of assessing personality or personality disorders in other diagnostic domains, such as mood, anxiety, substance use, or eating disorders?
The current proposal under consideration for the DSM-V assessment of personality and personality disorders attempts to capitalize on the strengths of several dimensional models that have been offered as solutions to the problems posed by categories. It consists of five parts: 1) an overall rating of personality (self and interpersonal) functioning ranging from normal to severely impaired, 2) prototype descriptions of major personality (disorder) types, 3) a personality trait assessment, on which the prototypes are based but that can also be used to describe major personality characteristics of patients who either do not have a personality disorder or have a personality disorder that does not conform to one of the prototypes, 4) generic criteria for personality disorder consisting of severe deficits in self-differentiation and integration and in the capacity for interpersonal relatedness, and 5) measures of adaptive functioning. The specifics of this hybrid model and the integration of its parts are being informed by ongoing literature reviews and will be refined through secondary analysis of existing data sets and by field trials using clinicians evaluating real patients.
Naturally, the challenge for the work group is to formulate a system that allows for meaningful representation of a patient’s personality characteristics and psychopathology most pertinent for clinical care, while not taxing the clinician with an excessively complicated or burdensome assessment that would inhibit its use. In light of the finding of Rottman et al. that clinicians tend to think more readily in terms of prototype matching than in terms of disparate collections of trait attributes, perhaps descriptive traits should be presented in the context of prototype narratives. Linking traits to clinical prototypes would essentially “telescope” the clinician’s view into a primary focus on a more discrete and relevant set of traits and, thus, take advantage of both dimensional assessments—prototype and trait—in the system. Clearly, much more work will be done to determine how best to assess personality and personality disorders in DSM-V.