Toward a Multidimensional Model of Personality Disorder Diagnosis: Implications for DSM–5

Abstract

This article outlines a model of personality disorder (PD) diagnosis that combines clinically useful constructs from the Diagnostic and Statistical Manual of Mental Disorders (DSM) with assessment procedures that maximize reliability and clinical utility while minimizing problems associated with threshold-based PD classification. I begin by addressing limitations in the current DSM conceptualization of PDs: excessive comorbidity, use of arbitrary cutoffs to distinguish normal from pathological functioning, failure to capture variations in the adaptive value of PD symptoms, and inattention to situational influences that shape PD-related behaviors. The revisions proposed by the DSM–5 Personality and Personality Disorders Work Group help resolve some of these issues, but create new problems in other areas. A better solution would be to employ a multidimensional model of PD diagnosis in which clinicians (a) assign a single dimensional rating of overall level of personality dysfunction, (b) provide separate intensity and impairment ratings for each PD dimension, and (c) list those personality traits—including PD-related traits—that enhance adaptation and functioning. Preliminary evidence bearing on the multidimensional model is reviewed, and broader clinical and empirical implications of the model are discussed.

Notes

Although reducing the total number of categories might help reduce PD comorbidity, the impact of PD syndrome reduction on comorbidity rates is in part a function of which PDs are eliminated. If those with the highest comorbidity rates are removed, overall comorbidity should indeed decrease, but if those with average or below average comorbidity rates are removed, comorbidity among the remaining syndromes might remain unchanged, or increase. In this context it is worth noting that the PDs proposed for retention in DSM–5 include some with relatively high comorbidity rates (e.g., borderline), and those proposed for elimination have comparatively modest comorbidity with other Axis II syndromes (e.g., histrionic; see Ekselius et al., 1994; Oldham et al., 1992; Zimmerman, Rothchild, & Chelminski, 2005).

To illustrate this dilemma, consider the researcher who hopes to compare the relative impact of early parental neglect on the subsequent development of schizoid PD and borderline PD. Using the DSM–5 PPD Work Group model, the researcher must use indexes of early neglect to predict categorical or threshold-based borderline PD diagnoses, then use these same indexes of neglect to predict a series of dimensional trait ratings associated with schizoid PD. Even if such a comparison is statistically feasible, it is conceptually uninterpretable.

In situations wherein PPRS and GAF scores are to be contrasted directly, the scales can be rendered equivalent by (a) multiplying PPRS scores by two, or (b) converting both PPRS and GAF ratings to standardized Z scores (see Rosenthal, 1991).

A skeptic might argue that inclusion of adaptive personality features in a diagnostic record is inappropriate because diagnoses are intended to identify areas of dysfunction and suggest possible interventions. Information regarding adaptation and healthy functioning can be useful in this regard as well. Consider, for example, a medical patient with a diagnosis of elevated blood glucose. If this patient also has normal blood pressure (an adaptive feature of physical functioning), medication options are available to treat this patient's elevated blood glucose that would not be available if the patient was hypertensive. The same is true of personality pathology: Knowing that a narcissistic person has good impulse control opens avenues for possible intervention that would not be available if that narcissistic patient showed poor impulse control.