Yesterday, I wrote about Brown and Barlow's (2009) proposed dimensional model of mood and anxiety disorders for the DSM-V. The author's original article was published in Psychological Assessment as part of a special issue on suggestions for DSM-V and reading over their thoughts left me fascinated with the rest of the journal issue. That being said, today I would like to continue yesterday's discussion, but I will shift the focus to a dimensional model of personality disorders, proposed in a paper published in the same issue of Psychological Assessment by Thomas Widiger of the University of Kentucky, John Livesley of the University of British Columbia, and Lee Anna Clark of the University of Iowa (2009).
What is wrong with our current diagnostic system for personality disorders?
Personality disorders, for a variety of reasons, have always been a source of controversy, frustration, and misinformation. The very nature of the idea that a fairly stable and genetically determined aspect of an individual can, in and of itself, constitute a mental illness is troublesome to some people, as they interpret that to mean that psychologists are arguing that some individuals are inherently flawed and need to be changed or, alternatively, that they are simply pathologizing people for being different. This, of course, represents a misunderstanding of the diagnoses, which are intended to recognize long standing patterns of thought, feelings, and behaviors that cause clinically relevant levels of distress and/or impairment. In other words, nobody is looking to criticize anyone for being different, but they are definitely interested in identifying aspects of an individual's life that are causing problems and require help.
Unfortunately, the structure of the personality disorder categories in the DSM are based on very little research and, in many cases, involve vague and/or overlapping symptoms across disorders. In the DSM-IV-TR, personality disorders are divided into three clusters:
Cluster A - Odd/eccentric
Schizoid Personality DisorderSchizotypal Personality Disorder
Paranoid Personality Disorder
Cluster B - Dramatic/emotional
Borderline personality disorderAntisocial personality disorder
Narcissistic personality disorder
Histrionic personality disorder
Cluster C - Anxious/fearful
Obsessive-compulsive personality disorderDependent personality disorder
Avoidance personality disorder
As is stands, however, there is little research supporting the validity of these distinctions and some evidence that actually refutes it (Sheets & Craighead, 2007; Widiger & Mullins-Sweet, 2005). In addition, there are some that argue that schizoid personality disorder does not cause enough distress or impairment to constitute a mental illness (I have even heard some prolific researchers state the case that this disorder does not truly exist), there is compelling evidence that schizotypal personality disorder should be moved to the schizophrenia spectrum section of the DSM, and some evidence that avoidant personality disorder is not fundamentally different from social anxiety disorder (SAD).
Widiger, Livesley, and Clark (2009) further emphasized the problematic structure of personality disorders in the DSM-IV-TR by pointing out that two individuals can meet criteria for antisocial personality disorder (ASPD) without sharing a single symptom. The disorder includes seven possible symptoms and only requires that an individual exhibit three of them in order to meet criteria for a diagnosis. Quite obviously, if two people can have the same diagnosis without sharing a single symptom, the validity of that diagnosis is open to serious question, even when we consider the research on different factors of ASPD and its relationship to psychopathy.
Yesterday, when I discussed the potential value of Brown and Barlow's dimensional model of mood and anxiety disorders, I mentioned my concern that we would no longer be able to apply our research and understanding of empirically supported treatments for those disorders. In the case of personality disorders, this is much less of an issue. Other than dialectical behavior therapy (DBT) for borderline personality disorder (BPD), there is a dearth of research on treating personality disorders and an empty list of empirically supported treatments. In other words, the risk of shifting to a new system is substantially reduced in this case.
All of this brings me to the central point of Widiger, Livesley, and Clark's (2009) article and their proposal for a new system: there is substantial evidence that personality disorders are dimensional constructs rather than categorical entities (e.g., Clark, 2007) and that personality disorders represent extreme deviations of normal characteristics rather than completely foreign sets of symptoms and behaviors (e.g., Blashfield, 1984). I realize that last sentence was a bit of a mouthful, so let me explain that more clearly. It appears that personality disorders are dimensions rather than categories in that everybody has a certain amount of these symptoms/characteristics. In other words, it's not a matter of simply having them or not, but rather a matter of how much of each characteristic you have. Some people have more or less than others and that deviation in degree is at the root of the problematic outcomes associated with personality disorders. Along those lines, personality disorders appear, at least to some degree, to represent extremely high or low amounts of personality characteristics present to some degree in all people (e.g., emotional reactivity). As such, a personality disorder does not represent some foreign presence clouding an individual's character as a tumor alters the functions of an internal organ, but rather represents an amplified or dulled character trait - likely in large part genetically determined - that causes an individual to think and/or behave differently than most other people in such a manner that the difference is distressing and/or impairing.
So what do we do about this?
Widiger, Livesley, and Clark (2009) propose a system not altogether different from that proposed for mood and anxiety disorders by Brown and Barlow (2009). The authors based their model on a combination of three popular measures of personality: the Dimensional Assessment of Personality Pathology - Basic Questionnaire (DAPP-BQ; Livesley & Jackson, in press), the Schedule for Nonadaptive and Adaptive Personality (SNAP; Clark, Simms, Wu, & Casillas, in press), and the NEO Personality Inventory - Revised (NEO PI-R; Costa & McCrae, 1992). All three of these measures attempt to break personality into smaller components and to provide personality profiles for individuals based upon their scores on each subscale.
Using a statistical procedure known as factor analysis, the authors attempted to determine the degree to which these three models of personality could be integrated into a single model, thereby giving us a unified manner in which to understand and measure personality. What they found was a hierarchical model that could potentially serve as a valid measure capable of being integrated into clinical practice in a useful manner.
Essentially, personality in this model is broken down into two levels. The first level features four categories:
- Emotional instability
- Introversion
- Antagonism
- Constraint
When a client first presents to a clinic, Widiger, Livesley, and Clark (2009) suggest that they be assessed on these four categories. Clinicians should look to see if the client exhibits a deviation - either high or low - on any of them. Remember, every individual has some degree of all four of these traits, but significantly high or low levels of any or them could result in problematic outcomes.
If a client does demonstrate a deviation from the norm on one or more of the four categories, the clinician should then follow-up on that particular category with an assessment of level two personality features - subcomponents within each of the four first level categories. The following are examples of level two features, organized within their respective level one categories.
Emotional Instability:
- Anxiousness
- Affective lability
- Narcissism
- Self-harm
Introversion:
- Intimacy problems
- Restricted expression
- Social avoidance
- Stimulus seeking (for low introversion scores)
- Exhibitionism (for low introversion scores)
- Aggression
- Entitlement
- Suspiciousness
- Callousness
Constraint:
- Compulsivity
- Workaholism
- Impulsivity (for low constraint scores)
So, for every client that comes in to a clinic, the clinician will assess their profile on the four first level categories. In this sense, the first level serves as a screener. The clinician administers a self-report questionnaire to see if further assessment is warranted and, in the meantime, develops a rough picture of both the assets and liabilities in the profile. For any client that deviates with a high or low score on one or more of those categories, a further assessment will be conducted to examine the specific facets of the first level category that are causing the deviation.
So what would happen to our diagnostic categories?
If this system were adopted in the DSM-V, the traditional personality disorder categories would no longer be used. Instead, individuals would be assessed for their individual personality profile, which would detail both assets and liabilities. The profile would consist of a number of non-overlapping concepts, each of which has been heavily researched and shown to correlate with distinct outcomes. The specific, level two features (e.g., impulsivity, entitlement, emotional lability) would serve as targets for treatment.
Are these necessarily the best variables to use?
The authors freely admit that their model likely needs some modification. Variables from measures other than the three mentioned above might be useful additions and some of the variables included above might not contribute much use. In this sense, the authors are much more certain about the basic structure and purpose of the model than they are of the specific components.
This is a tricky situation, however, because many different people would have a stake in what actually become integrated into the final model. For instance, as a researcher in the field of "impulsivity" I can tell you without hesitation that the word "impulsivity" is of almost no use, as it represents many different distinct ideas (Whiteside & Lynam, 2001). Much of my research has focused on negative urgency and, as such, if that variable were integrated into the model, this would be great for my career personally and a decision with which I would agree wholeheartedly. At the same time, there are countless other researchers out there with similar thoughts on different variables and no clear cut way to determine which of us makes the best case. As a result, much like with the creation of the current DSM system, we would be left at the mercy of the committee to make that decision for us and, in all likelihood, biases will be involved in the selection of items.
All of that being said, the variables mentioned above are all very well researched and represent strong choices for inclusion in a dimensional model of personality disorder.
What does this mean for treatment?
Treatment research for personality disorders other than DBT for BPD is limited. Additionally, although DBT is highly successful in reducing many components of BPD, including suicidal behavior and non-suicidal self-injury, it does not fully address each symptom of the diagnosis. As such, we are currently faced with a situation in which, regardless of whether or not the system is overhauled, a massive amount of research is needed. Along those lines, implementing this system would require immense amounts of research indicating that individuals with particular personality profiles benefit in clearly identifiable ways from specific treatment protocols. Such research would enable the creation of a new generation - or least an updated version of the current generation - of empirically supported treatments.
Conclusion
Here is what we know: the current diagnostic system for personality disorders is highly flawed in a number of ways. In fact, the flaws inherent in this section of the DSM far outweigh those of just about any other part of the system. For years, many individuals have pushed for a dimensional system for assessing personality and recognition of the fact that symptoms of personality disorder might represent extreme deviations of normal personality characteristics, but no consensus has been reached on a suitable model. Widiger, Livesley, and Clark (2009) presented a thorough and compelling case for a new model based on empirical research, but acknowledge that their model is also imperfect. That being said, there seems to be reason to believe that some form of dimensional model will be incorporated into DSM-V, at least for personality disorders, so it is important to discuss the pros and cons of such a move and to become informed on what this type of model would mean, both for clients and clinicians. Hopefully, you will share your thoughts on this model in particular or on changes needed for personality disorders in general. If you are a clinician, how do you think this will impact your practice? If you are an individual struggling with a personality disorder, to what degree does this model make sense to you? If you are a researcher, what do you think needs to be integrated into this or other models of personality?
If you would like to learn more about the DSM or personality disorders, we recommend the following resources, all of which are available through our online store:
- Cognitive-Behavioral Treatment of Borderline Personality Disorder
by Marsha Linehan
- Personality Disorders: Toward the DSM-V
by William O'Donohue, Katherine Fowler, and Scott Lilienfeld
Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University
