Toward a Clinically More Useful Model for Diagnosing Narcissistic Personality Disorder

Given the ubiquity of narcissistic behavior in society (1 , 2) and clinical practice, it is striking that so little clinically focused research has been conducted on narcissistic traits and narcissistic personality disorder. Narcissistic personality disorder first appeared in DSM in the third edition. Its inclusion was stimulated in large part by contributions of psychoanalysts such as Kernberg, who continued in the tradition of a conflict-based ego psychology and object relations approach, and Kohut, who opted for a deficit-based theory of personality development postulating that narcissistic pathology resulted from the child’s repeated experiences of parental empathic failure. The unique challenges of treating these patients received considerable attention, which led to an acceptance that expressive or insight-oriented treatment could be effective for many patients with narcissistic personality disorder previously thought untreatable.

Most empirical studies of narcissistic personality disorder have examined epidemiological dimensions, especially prevalence and comorbidity. The prevalence of the disorder has been estimated to range from 0% to nearly 15%, with the most robust studies establishing prevalence rates below 4% (3) . Recently the Wave 2 National Epidemiologic Survey on Alcohol and Related Conditions (4) documented a 6% lifetime prevalence rate (7.7% for men and 4.8% for women) based on 34,653 face-to-face structured interviews that included DSM-IV diagnostic criteria. The study not only documented a higher than anticipated prevalence of narcissistic personality disorder, it also demonstrated considerable psychosocial disability among men. Rates were higher for younger adults and for those who were separated, divorced, widowed, or never married. Significant cultural differences were noted: black men and black and Hispanic women had higher rates compared with Hispanic men and whites of either gender. Comorbidity, an important consideration in clinical practice, was noted in the study, with significant past-year co-occurrence of substance use, major depressive disorder, bipolar I disorder, anxiety disorders, and other personality disorders.

Agreement about the validity of diagnostic criteria for narcissistic personality disorder has been elusive. Many clinicians express confusion about the diagnosis since patients with different personality disorders frequently share common experiences and presentations. Indeed, in preparation for DSM-V, the question of whether to incorporate a dimensional model into the current categorical classification of personality disorders is being debated in the hope of better meeting unique challenges in axis II diagnoses. Limitations of the categorical model include an inadequate research base for some personality disorders (of which narcissistic personality disorder is an excellent example), excessive co-occurrence with other personality disorders, arbitrary and unstable diagnostic boundaries between axis I and II disorders, and the likelihood that one patient will differ from another in the criteria utilized to reach a diagnosis (5) . (The most common personality disorder diagnosis in clinical practice may be personality disorder not otherwise specified, which perhaps reflects a widespread dissatisfaction with the current axis II criteria.)

A dimensional approach, on the other hand, assesses personality traits on a continuum with psychopathology, hopefully yielding an overarching hierarchical model (6) . Such a direction would attempt to define personality disorders through examination and integration of perhaps four or five personality dimensions—as described, to name but two approaches, in the five-factor work of Costa and McCrae (7) and the psychobiological model of Siever and Davis (8) . As an example, the psychobiological perspective on axis II disorders is organized according to the following domains: cognitive-perceptual (consider the dimensional differences between schizotypal versus avoidant personality disorders), impulsivity-aggression (antisocial versus dependent personality disorders), affectivity instability (narcissistic and borderline versus obsessive-compulsive personality disorders), and anxiety-inhibition (avoidant versus antisocial personality disorders). This model attempts to link personality disorders with specific neurotransmitter systems, although robust support for this approach is lacking, except in the relationship between impulsivity and serotonin. An acceptable dimensional model would require consensus regarding the number and specificity of these domains and the helpfulness of this approach in the clinical setting.

The exceptional work by Russ et al. (9) reported in this issue does not focus per se on the categorical versus dimensional question. However, its delineation of the characteristic and distinctive features of narcissistic personality disorder among North American patients in treatment is a significant and highly useful contribution to the diagnostic and treatment literature on the disorder. This approach builds on clinical approaches advocated by a number of prominent psychoanalytic theoreticians, beginning in the 1960s, who recognized that narcissism was not necessarily psychopathological and, indeed, exists along a continuum in everyone. In many ways, Russ et al. refine dimensional aspects of earlier models. They note their preference for a dimensional approach and argue that their new subtype classification is based on prototypes that often vary from patient to patient in presentation of a specific characteristic. In previous work Shedler and Westen (two coauthors of the article) have addressed the limitations of models that fail to describe the complexity of personality as seen in clinical practice. Additionally, they are critical of a model that focuses on behavioral tendencies at the expense of the patient’s internal experiences.

The authors used the Shedler-Westen Assessment Procedure–II (SWAP-II), a 200-item checklist of personality descriptive statements; a DSM-IV personality disorder diagnostic criteria checklist; and personality disorder construct ratings to obtain data from more than 1,200 clinicians nationwide treating patients with personality disorders. Prominent findings include a more clinically sophisticated and accurate categorization of narcissistic personality disorder and a better appreciation of the psychic pain (insecurity, vulnerability, fraudulence) experienced by a preponderance of patients with the disorder. With respect to categorization, this work elucidates three distinct narcissistic personality disorder subtypes: grandiose/malignant, fragile, and high-functioning/exhibitionistic. Traditionally, clinicians have categorized narcissistic personality disorder into two subtypes. Gabbard (10) , for example, describes the oblivious narcissist and the hypervigilant (what others have termed shy) narcissist. The oblivious subtype is characterized by hostility and arrogance, self-centeredness and self-absorption, and little appreciation of the impact of his or her behavior on others. The hypervigilant subtype has much greater psychological vulnerability, disavows the desire to be the center of attention, and constantly scans the world for slights and criticisms. To these descriptions, Russ et al. have added a third prototype, the high-functioning/exhibitionistic patient, which they note has received little empirical support but is well described in the clinical literature. Using Q-factor analysis, the authors find that the grandiose/malignant subtype is characterized by manipulation, minimal appropriate remorse, entitlement, aggression, and quest for power over and control of others. Unlike in the other two subtypes, envy does not appear to be common in the grandiose/malignant subtype. Although treatment outcome data for these three subtypes are nonexistent, the authors note—and most clinicians would agree—that patients with a malignant subtype may have limited treatability since they lack insight and sometimes do not experience symptoms such as anxiety and depression, which might motivate a person to accept treatment. Patients with the fragile subtype, on the other hand, are likely to be those most in pain, since their defenses against feelings of inadequacy frequently fail, often followed by narcissistic rage. Those with the high-functioning subtype are undoubtedly the most treatable, since their grandiosity is often accompanied by a reasonable level of introspection and treatment motivation, although they are not without challenges.

This research provides additional contributions and points to future issues deserving investigation. First, it is a model for conducting generalizable research through clinician participation. The enduring practitioner criticism that many efficacy studies of psychotherapy and psychopharmacology have limited applicability to clinical practice can be addressed through the approach used in this study. Compared with most psychotherapy outcome and process studies, for example, the number of participating experienced mental health professionals in this study (in practice 19.8 years on average) is impressive. The non-theoretically biased data provided by psychiatrists and psychologists over a mean of some 17 months of treatment, moreover, is superior to the single-episode reporting of contact obtained through structured clinical interviews or self-report measures, which have unfortunately served as the basis for most empirical diagnostic studies. The authors’ approach therefore can provide a more comprehensive understanding of patients’ subjective experiences and, by extension, of what might be anticipated within the treatment relationship. Any new approach to the diagnosis of personality disorders must account for the types of patient presentations in clinical practice. Russ and colleagues also amplify their findings through clinical vignettes, an underutilized practice that is of significant help to those treating patients.

Second, their findings clarify some of the confusion between the diagnoses of antisocial personality disorder and narcissistic personality disorder. Much has been written about the malignant narcissistic personality disorder, but often this appears in the forensic literature addressing psychopathy. In one study (11) , exploitation and lack of remorse distinguished antisocial personality disorder from narcissistic personality disorder, and grandiosity distinguished narcissistic personality disorder from antisocial personality disorder, yet both disorders shared narcissistic features, such as arrogance, entitlement, self-centeredness, envy, and sensitivity to criticism. The identification of the malignant subtype is important because the propensity for hostile aggression in this subtype is considerable.

Third, the methodology used by Russ et al. offers a greater ability to document patients’ experiences of themselves and the world around them but which are not often conveyed in the typical DSM-IV approach.

Fourth, this study may provide direction to the exploration of a neurobiological and genetic appreciation of narcissistic personality disorder. In a twin study examining the dimensions of personality disorders, Livesley et al. (12) found genetic dominance effects for intimacy problems, affective lability, and especially narcissism. There has not been any study examining gene-environment interaction in those with narcissistic personality disorder as there has for antisocial behavior. A recent innovative study of borderline personality disorder approached abnormal behavior and social interactions through the use of game theory and neuroimaging. Subjects with borderline personality disorder exhibited distinctive responses in the anterior insulae associated with failure to recognize social norms and to cooperate (13) . The explication of personality disorders from a neurobiological perspective is an important goal in deepening our understanding of their etiology and treatment.

Building on earlier work by Shedler and Westen, the Russ et al. study confirms the attractiveness of a clinician-based research network and the SWAP methodology in elucidating important issues routinely faced in daily practice. Hopefully, this refinement of narcissistic personality disorder subtypes can serve as a model and stimulate more precise gene-environment inquiry in axis II disorders. The authors have reaffirmed the meaningfulness of patient behavior in treating the emotional pain and discomfort experienced by many with axis II disorders.