January 18, 2012

Changing Diagnostic Criteria for Personality Disorders: How Much Should Data Matter?

Author Picture

Mark Zimmerman, MD

Bayside Medical Center, Providence, Rhode Island


Like it or not, ready or not,DSM-5 is coming. Criticisms of the process have been voiced. Concerns about particular changes have been raised. Conceptual critiques have been written (and have been published or are forthcoming). Nonetheless, field trials are underway. The DSM-5 is scheduled for arrival in 2013.

Revising the DSM seems, in part, to be a political effort. And, as in politics, promises are made that are not kept. As with the run-ups to prior revisions of the manual, “the field” has been reassured that changes in criteria would be grounded in science. Yet, despite assurances that only data-driven modifications would be made, with each new edition of the DSM, we have witnessed repeated instances of changes being made in the absence of data demonstrating that the new criteria are superior to the prior. Perhaps this is nowhere else better exemplified than in the personality disorder (PD) section. Criteria have been added, removed, and rewritten, without evidence that the new approach is better than the previous one.

When it comes to revising the official diagnostic classification system, the guiding principal should be that criteria should not be altered in the absence of research demonstrating that the new approach is more valid or more clinically useful or, preferably, both. Just how much data are necessary to justify a change can be debated, but, because of the potential clinical, social, and public health implications, a change in diagnostic criteria should be firmly supported by scientific study.

Through the years, there have been many critiques of the DSM-III, DSM-III-R, and DSM-IV approaches to classifying the PDs. These critiques have identified problems of diagnostic overlap, the lack of a clear boundary between normality and abnormality, the failure to take into account findings from normal personality research, and the lack of diagnostic stability over time. Some of these criticisms have been invoked by the DSM-5 Work Group for Personality and PDs as justification for the proposed changes. However, identification of problems with the existing criteria is insufficient to warrant a change. The new criteria need to be compared to those they are intended to replace and shown to be better in some way. If superiority cannot be demonstrated, why disrupt the field with new criteria?

The DSM-5 Work Group recommended a reformulation of the PD section. While research has been undertaken to examine the clinical utility, reliability, and validity of the proposal, no research is planned to compare the new approach to the DSM-IV approach. It is difficult to criticize a proposal when there is no published research on it. Of course, this lack of study itself is cause for concern. Can you imagine a change in the official, sanctioned definition of hypertension or hypercholesterolemia in the absence of multiple studies examining the implications of such a change?

And what if some of the criticisms of the DSM-IV PD diagnostic criteria that have been invoked by the Work Group lack empirical support? As part of the Rhode Island Methods to Improve Diagnostic Assessment and Services (MIDAS) project, my research group has examined some of the issues that were central to the Work Group’s justification for changing the criteria. In one of these reports,1 we compared different scoring conventions for producing dimensional PD diagnoses in 2,150 psychiatric outpatients. We noted that DSM-IV can be considered to already accommodate a quasi-dimensional system insofar as individuals who do not meet the threshold for diagnosis can be noted to have traits of the disorder, and we found that the DSM-IV 3-point rating convention was as valid as scoring methods using more finely graded levels of severity. We therefore concluded that these findings argued against changing the current DSM-IV diagnostic approach, and we advocated instead for increased recognition that DSM-IV already includes a valid dimensional rating.

To address the problem of excessive comorbidity, the DSM-5 Work Group initially recommended reducing the number of specific PD diagnoses from 10 to 5 by eliminating paranoid, schizoid, histrionic, narcissistic, and dependent PDs. They later recommended that narcissistic PD be retained (without citing any new data supporting their reversal). Amazingly, they cited no data on the impact this change might have on the prevalence of PDs, the validity of PD diagnoses, or the rate of comorbidity. An analysis2 of the same MIDAS project database found that eliminating 5 PDs would, indeed, reduce comorbidity, although comorbidity would not be eliminated. Moreover, compared with patients without a PD, the patients with either a retained or excluded PD had greater psychosocial morbidity, and there was little difference in psychosocial morbidity between patients with either a retained or excluded PD. This finding suggested that the reduction of comorbidity could come with a cost of false-negative diagnoses.

To be sure, there are problems with the DSM-IV approach to the classification of PDs. However, if diagnosis is an important clinical endeavor, if the avoidance of potential unforeseen negative consequences resulting from a change in diagnostic criteria is important, and if the empirical foundation supporting changes to the diagnostic system is important, then it is premature to adopt the DSM-5 Work Group’s proposal to revise the Personality Disorders section.

Financial disclosure:Dr Zimmerman had no relevant personal financial relationships to report.


1. Zimmerman M, Chelminski I, Young D, et al. Does DSM-IV already capture the dimensional nature of personality disorders? J Clin Psychiatry. 2011;72(10):1333–1339. Abstract

2. Zimmerman M, Chelminski I, Young D, et al. Impact of deleting 5 DSM-IV personality disorders on prevalence, comorbidity, and the association between personality disorder pathology and psychosocial morbidity. J Clin Psychiatry. 2012;73(2):202–207. Abstract.​

Category: Personality Disorder
Link to this post:
Related to "Changing Diagnostic Criteria for Personality Disorders: How Much Should Data Matter?"

17 thoughts on “Changing Diagnostic Criteria for Personality Disorders: How Much Should Data Matter?

  1. The most scientific basis for reducing the number of PD diagnoses is to perform factor analytic studies on the 10 PDs from DSM4 and from criterion referenced PD measures such as the IPDE. The factors which emerge as robust would tell us the number of REAL PDs that exists clinically.
  2. I’m concerned about creating “new” personality disorders or eliminating “unsubstantiated” ones since in my limited experience (40 years) few clinicians are truly experienced or qualified to even provide a proper Axis II diagnosis. The thought of clinician rushing to try to have a client “fit” a certain PD frightens me. Axis II diagnosis’ can carry a very adverse effect on a client, especially if it’s later discovered to be unfounded. Likewise, thinking that certain Axis II diagnosis’ can be treated more effectively soley by the use of medication could prove to be fool hearty in the long term.
  3. I’ve just reviewed the DSM5 PD section. It appears to me that they have combined an “object relations” type of approach, with a 5 factor personality dimensional trait approach. The latter is empirical, the former is not. The reasoning for reducing the number of PDs is beyond me.
  4. i wonder if the fundamental criteria for these problems is predicated primarily on how those labeled as such relationally impact others rather than being grounded in the phenomenological. specifically, given that we accept that early adverse experiences strongly influence the manner in which self relates to self, others and the world and given also that trauma predisposes persons to maladaptively relating, why are we not conceptualizing many, if not all of these difficulties as being trauma related and thus reconceptualize them more compassionately and approach them in treatment as such.
  5. Thank you Dr. Zimmerman for taking on the committee. I hope you will also consider that narcissistic PD (NPD), antisocial PD, (ASPD) and to some extent borderline PD (BPD) affect others including children and spouses perhaps as much as they affect the individual. Therefore our definitions and descriptions have to be amenable to public education.

    The term antisocial, sociopath and psychopath all should be formally defined even if it is recommended these terms fall into disuse. In my opinion, science is not served by the DSM ignoring the fact there is a legitimate scientific society- Society for the Scientific Study of Psychopathy, yet the term is not addressed well by the DSM. The criteria for ASPD fail to gather data on the interpersonal and emotional deficits that exist in this group of patients.

  6. As a follower of the Allen Frances (chairman, DSM-IV task force) line of criticism, as well as majoring in psychology pre-med and providing psychotherapy for patients rather than the current unfortunate trend towards “10 minute med checks,” I wonder who’s read Millon or Shapiro or even Gabbard these days? Not to mention (gasp!) Freud, Jung, Adler, Sullivan, and numerous other mentionables in our history? Personality disorders are theoretical constructs to help us understand our patients and how to help them, not just to label them for the chart and insurance companies. Squabbling over esoteric statistical analyses to validate or invalidate the concepts might just primarily serve the purpose of making the psychiatrist more comfortable with primarily throwing meds at the problem. A past Am.J. Psychiatry article was titled “Is psychiatry losing it’s mind?” Meanwhile the committee is spinning over such concepts as “Hebephilia” and contemplating other modern-age nosologies for compulsive internet use, “porn addiction,” etc. I’m disappointed that they won’t delay for another year to consider my favorite, “texting while applying make-up on the freeway driving disorder.” To close, consider the international psychiatric community and, you know, the ICD classification system (which was to be coordinated with the DSM, but alas, the delays, the dawdling). I’d be interested to hear commentaries from our world colleagues. So there you go. Like, you know , like, factor analyze this editorial rant, my friends.
  7. I forgot to compliment Mark on his comments and suggestions/reminders regarding the need to remember what psychiatry is really all about. I could argue the evidence for cognitively-based theories and interventions as well, but empathy and compassion (with proper boundaries) are fundamental. “And oh, by the way, you’re a dependent histrionic obsessive-compulsive, here’s prescriptions for your Prozac, Xanax, Depakote, and Abilify, see ya back in 4 months” misses the conceptual appreciation of this line of thought.
  8. Maybe we need to develope a questionnaire that has good validity and reliability for diagnosing personality disorders and that would be the standard instument we all use. Something like the MMPI perhaps. Just an idea.
  9. Dr. Zimmerman’s concerns are valid. Any comments/reaction from the Working Group on the MIDAS results? On the other hand, can someone help me understand the politics thought to be behind non-data-driven changes in this area?
  10. Varudeyam keep your ignorant opinons to yourself, this discussion is not intended to act like your Facebook page or a sounding board for personal gripes with the government. At least not in the context of making broad generalizations that you very obviously have no education about.
  11. I will tell you my opinion why: the current DSM, as you know, grounds diagnosis in the phenomenological while apparently presenting itself as atheoretical (there is no section on etiology in the DSM!). This of course, as we all know, is not true. The unspoken theory that drives the DSM project since at least 1980 is a massive biomedical paradigm. Introducing a psychological or social conceptualization is consequently anathema to the current construction of mental illness as basically a biomedical illness in the present day United States. Alas, the dark side of the force….
  12. I had a robust laugh at the title of your post. Alas, the chieftains of the DSM project are caught in the spell of a biomedical model, and what a powerful spell indeed. A sign of the times….a sign of the times. But all hope is not lost….even Darth Vader overthrew his master and took off his mask.
  13. If we are to go with X pd’s and y dimensions tehn a pd-assessment wil take a 2 day asessment with 30 tests/questionnairs.
    Again the psychiatrists do not know the amount and psychologists can moderate their dimensions.
    Let’s keep the DSM-IV pd’s in the wake of the DSM-V and do some proper research, and I mean a couple of thousands from various countries, spheres, and grind the data down to comprehensible dimensions.

  14. Allen Frances, M.D. has written concise criticisms of the upcoming DSM. If he is accurate, psychiatry as a profession will sink even lower in respect, legal importance and income. As a clinician, the easiest solution to the matter is to make ALL Axis II diagnoses “Personality Disorder NOS” and ignore the new DSM. If we do not use criteria reflective of the psychological dynamics (as in more classic therapies) of REAL treatment, DSM-V is irrelevant.
  15. From what I’ve read so far I’m thoroughly disgusted with the state of DSM-V. Cynics will conclude it’s just a way of generating new textbook, course and lecture revenues. I would be happier if the entire DSM-V effort were COMPLETELY DROPPED and we waited until there were better reasons to discard DSM-IV TR.

Leave a Reply


Browse By Author



Browse By Author