مسائل در ارزیابی و مفهوم اختلالات شخصیت

Abstract This article reviews several current issues associated with the definition and assessment of personality disorders (PDs) as defined in the third and fourth editions of the Diagnostic and Statistical Manual of Mental Disorders (DSM). Specifically reviewed are issues associated with classification, PD conceptualizations, and the assessment of these disorders. DSM PD categories are also reviewed in terms of their psychometric properties. A review of the PD assessment literature suggests that DSM conceptualizations and definitions of PDs are problematic at both conceptual and quantitative levels. This article concludes with suggestions for possible alternative approaches to and modifications of DSM PD assessment.

Conclusions and future directions Two decades after the publication of DSM-III and official recognition of the clinical relevance of PDs, there has been a growing concern as to whether DSM PD concepts are valid and useful. The weight of the evidence, particularly that related to classification and validity issues, suggests that DSM PD concepts are problematic. Dissatisfaction with DSM PD concepts is also mirrored in a recent international survey of psychologists and psychiatrists (Maser et al., 1991). Upon reflection on the research literature, a wholesale abandonment of DSM PD classification is not recommended. However, after 20 years of research, it seems clear that future DSM work groups should give thoughtful consideration to incorporating a variety of modifications in how PDs are conceptualized, defined, and evaluated for inclusion in official nomenclatures. The following sections summarize some suggestions for consideration. The Rationale and Theoretical Basis for Each PD Concept Needs to Be Delineated To constructively engage in the process of empirical evaluation and falsification, some delineation of the rationale or theoretical basis for each of the PDs is required. The provision of a limited listing of features associated with each of the PDs is uninformative as to the underlying concept or interactions that give rise to these features. Absent in current DSM classification is an articulation of the underlying cohesiveness which binds the PD features together. As a consequence, researchers and clinicians are faced with the task of evaluating the utility and validity of DSM PD concepts with little sense of what those concepts represent apart from a handful of associated features. In order for the evaluation of the adequacy of the definitional features of PD concepts to proceed, the disorders themselves need greater description and articulation. Greater Consistency in the Content of PD Criterion Sets Should Be Sought An examination of the definitional features of the various PDs reveals inconsistency in how such conditions are defined. For example, antisocial PD is the only PD where childhood and early adolescent behaviors are included among the criterion features. However, there is no rationale provided in DSM as to why childhood behaviors are more relevant for this condition than other PDs. Similarly, some disorders emphasize cognitive distortions among definitional features (e.g., paranoid), whereas others emphasize interpersonal relations (e.g., schizoid, avoidant, dependent), behavioral habits (e.g., obsessive-compulsive), problems in emotional regulation (e.g., borderline), or a combination of these (e.g., schizotypal, antisocial, histrionic, narcissistic). Furthermore, definitional features within some PDs tap several dimensions, whereas others seem to be based on a single dimension. Relatedly, some criteria seem to represent descriptions of the construct that the PD concept intends to capture, whereas other features seem to describe indicators of the construct (Shea, 1992). Yet, DSM does not provide an account for why some domains of description are more central to the definition of some disorder than others. The Validity of the Axis I and II Distinction Warrants Further Consideration A fair amount of evidence has accumulated that indicates that Axis I and II conditions are not “quasi-independent” as previously suggested. Rather than maintain this false and arbitrary distinction, a more fruitful approach might be to explore patterns of symptom covariation across Axis I and II conditions. Such explorations may reveal that some Axis I and II conditions are linked to more superordinate constructs that may, in turn, enhance our theoretical understanding of psychological disorders as well as the organization of such conditions. The work of Siever and Davis (1991) and Cloninger (1987) represent attempts to link patterns of symptom covariation across Axis I and II conditions in the context of theory which, in turn, can be subjected to empirical evaluation and falsification. Consider the Use of Psychometric and Test Theory Principles for the Evaluation and Inclusion of PD Features and Concepts Skinner (1981) has suggested that classification systems should be subjected to the same standards as those required of psychological tests. Although psychiatric classification and psychological tests have different theoretical assumptions and applications (Blashfield & Livesley, 1991), there are many similarities between the two approaches. Although the interrater reliability and internal consistency of DSM PD concepts are often within acceptable limits (e.g., Blais & Norman 1997 and Zimmerman 1994), temporal stability estimates have not been provided for many of the PDs, and findings that do exist tend to be equivocal. Similarly, the validity of DSM PDs has not been adequately explored. The scant research that does exist suggests poor face validity, poor concurrent validity, and poor discriminant validity. Sensitivity and specificity estimates are also unacceptably poor, which is a particular problem when one considers that the base rates for individual PDs tend to be relatively low. Aspects of test theory (e.g., Anastasi & Urbina 1997, Cronbach & Gleser 1965 and Nunnally & Bernstein 1994) are compatible with psychiatric classification, and may potentially function as a useful model of measurement that can be applied to problematic aspects of PD classification. Blashfield and Livesley (1991) and Skinner (1986) describe how these two different approaches can be complementary, and they review measurement models that combine both approaches. Although the use of literature reviews, data reanalyses, and field trials to inform modifications of disorder criteria (Widiger, Frances, Pincus, Davis, & First, 1991) represent a step towards an empirically-based classification, it is ultimately unsatisfactory as the concepts used as reference criteria for subsequent modification were primarily those concepts defined in accordance with previous editions of DSM. This incestuous approach to the refinement of psychiatric classification only adds to the reification and circularity already present within the DSM system (Hickey, 1998), and makes it virtually impossible for future editions of DSM to substantially evolve beyond its predecessors. Consider the Shift Away from Categorical Representation of PDs to Dimensional or Prototype Representation Consistent with current available data, the notion that PDs exist as discrete categories caused by a single, tangible factor is untenable. Furthermore, dimensional classification of PDs is associated with greater reliability, validity, and more accurately describes distributions of PD symptom features. Widiger 1991 and Widiger & Sanderson 1995 offers examples of how PD diagnostic categories can be transformed to dimensional ratings. Give Serious Consideration as to the Types of Problems Which May Contribute to Poor Concurrent Validity Concurrent validity, as indexed by the degree of diagnostic concordance reached by two different tests (or interviews) that assess the same PD concepts defined by the same criterion sets (i.e., DSM), is poor for DSM PDs (e.g., Perry, 1992). This is a serious problem, as it represents a challenge for the validity of the PDs and raises potential professional dilemmas, such as the utility of applying questionnaire or interview findings to treatment decisions and the labeling of persons. If diagnostic decisions that influence these processes are inconsistent, then it is conceivable that persons may be harmed by such faulty diagnostic judgments. Address Additional Concerns That May Compromise the Validity of Assessment It may be the case that DSM PD definitions, as currently delineated, are actually valid and useful. Problems with concurrent validity, for example, may be the result of how PDs are generally assessed. In a majority of the studies reviewed, PDs were assessed during the course of 1- to 2-hour interviews in which patients were asked about the presence or absence of each of the DSM PD features. Determination of the presence or absence of such symptomatology is based in large part on the verbal reports of the person being interviewed. As noted earlier, there are several features associated with PDs that are inconsistent with the ability for insightful self-reflection or the willingness to openly describe unpleasant aspects of one's experience. Consequently, it may not be the criteria themselves that are problematic, but rather the methodology for assessing them. Employ the Established Nomological Network Associated with PDs in the Refinement of PD Concepts and Definitions The ability to express constructs quantitatively along a dimension or at the intersections of several dimensions provides a foundation for what Cronbach and Meehl (1955) have referred to as the nomological network. The delineations of such networks as they apply to the PDs may illuminate that which accounts for covariation of specific disorder features or the covariation of clusters of disorders. Such networks may also reveal the diversity of dimensions that underlie these disorders, and thus provide a more comprehensive picture of the range or expression of these constructs. Taken together, these networks provide a fertile ground for construct validation, theorizing, and empirical verification and falsification. Consider the Development and Evaluation of a Hierarchical Approach to the Delineation or Classification of PDs As previously noted, the co-occurrence of several personality disorders within the same individual tends to be the rule rather than the exception. Yet, with the possible exception of the PD clustering scheme into three symptomatological classes, the DSM classification system does not provide a framework for conceptualization of the covariations that are observed. The development and evaluation of a hierarchical classification system has advantages as it: (a) reduces comorbidity, (b) reduces the number of diagnostic categories, (c) can illuminate the presence of a superordinate syndrome or construct that accounts for the comorbidity, and in the process, suggest an intermediate level of classification or description. Consider the Horizontal Organization of PD Features One problematic aspect of current PD definitions is that some PD criterion features share large or larger correlations with features associated with other PDs. Similarly, multivariate studies suggest that many diverse PD features load on dimensions that are not representative of a single PD concept. Such observations suggest that DSM PDs cannot be discriminated at the feature level, and provide an explanation for their substantial comorbidity and poor discriminant validity. To reduce comorbidity and enhance discriminant validity, PD descriptive features retained for DSM criterion sets should be those that are: (a) maximally discriminative, (b) most central to the diagnostic concept, and (c) load on a single dimension that defines the PD concept. PD concepts that continue to share substantial overlapping features should then be considered as potential representatives or variants of the same underlying construct and combined in order to avoid the assignment of two or more diagnostic labels for the same underlying condition. As such, future studies should shift focus to the relationships among PD symptoms or trait features and away from diagnostic comorbidity. Resist the Temptation to Revise Official Nomenclatures Within Short Time Frames To establish an empirical basis for a classification scheme, considerable time is required to conduct the necessary research. To publish revisions of classification systems every 7 years seems to render such a task unattainable. A survey by Zimmerman, Jampala, Sierles, and Taylor (1991) of psychiatrists indicated agreement among members of their sample that revisions of DSM are published too soon. Among the problems noted by Zimmerman (1990) associated with the premature publication of DSM revisions are the prevention of the accumulation of replicated research necessary to justify modifications in diagnostic criteria, the expenditure of resources to compare similarities and differences between new and old criteria, and the creation of difficulties in interpreting divergent findings stemming from different criteria sets. Zimmerman (1990) goes on to note that in no other area of medicine are pathological conditions redefined before research studies are conducted on the conditions of interest, or put more succinctly “changes in the nosology should follow, not precede, science” (p. 975).