A Warning Sign on the Road to DSM-V: Beware of Its Unintended Consequences

June 27, 2009

We should begin with full disclosure. As head of the DSM-IV Task Force, I established strict guidelines to ensure that changes from DSM-III-R to DSM-IV would be few and well supported by empirical data. Please keep this history in mind as you read my numerous criticisms of the current DSM-V process. It is reasonable for you to wonder whether I have an inherently conservative bias or am protecting my own DSM-IV baby. I feel sure that I am identifying grave problems in the DSM-V goals, methods, and products, but it is for the reader to judge my objectivity.

We should begin with full disclosure. As head of the DSM-IV Task Force, I established strict guidelines to ensure that changes from DSM-III-R to DSM-IV would be few and well supported by empirical data. Please keep this history in mind as you read my numerous criticisms of the current DSM-V process. It is reasonable for you to wonder whether I have an inherently conservative bias or am protecting my own DSM-IV baby. I feel sure that I am identifying grave problems in the DSM-V goals, methods, and products, but it is for the reader to judge my objectivity.

Much of our effort in developing DSM-IV centered on avoiding possible misuses of the system. We established a rigorous 3-stage procedure of empirical documentation to filter out mistakes. This consisted of systematic and extensive literature reviews, data reanalyses, and field testing conducted under well-controlled conditions and in a wide variety of settings.1-3 The null position was always to keep things stable: any change had to meet a high burden of empirical proof and risk-benefit analysis.

The work on DSM-IV was transparent and widely inclusive.4 We knew how important it was to get as many critical comments as possible to assist us in spotting pitfalls and blind spots. To this end, we enlisted the help of more than 1000 advisors, seeking particularly those opinions most opposed to the changes being considered. To recruit as many comments as possible from users at large, we also prepared a regular and widely distributed newsletter and journal column.5

There was explicit accountability for decision making on all changes. We published many articles to establish the methodology of the DSM-IV empirical review, to indicate ways of judging the value and risks of “innovations,”5 and to determine the pluses and minuses of the particular diagnostic changes that were under review.1,2,6-9 In midstream, we published a widely distributed DSM-IVOptions Book: Work in Progress10 that contained the alternative criteria sets proposed for every disorder. This gave everyone a chance to join us in evaluating each decision for change in DSM-IV.

After DSM-IV was completed, we published 4 sourcebooks, laying out in great detail the process and rationale for all the decisions that had been made, as well as their empirical support.11-14 Our goal throughout was to ensure that everyone would understand precisely what we were doing and how we were going about it. There was explicit accountability for decision making on all changes.

Why did we go to all this trouble in preparing DSM-IV and why should DSM-V undergo a sharp midterm correction to provide equivalent safeguards by becoming far more transparent, explicit, and conservative? I believe that the work on DSM-V has displayed the most unhappy combination of soaring ambition and weak methodology. First we will explore the excessive ambition, because it has encouraged an excessive tolerance for risk taking.

The DSM-V goal to effect a “paradigm shift” in psychiatric diagnosis is absurdly premature. Simply stated, descriptive psychiatric diagnosis does not now need and cannot support a paradigm shift. There can be no dramatic improvements in psychiatric diagnosis until we make a fundamental leap in our understanding of what causes mental disorders. The incredible recent advances in neuroscience, molecular biology, and brain imaging that have taught us so much about normal brain functioning are still not relevant to the clinical practicalities of everyday psychiatric diagnosis. The clearest evidence supporting this disappointing fact is that not even 1 biological test is ready for inclusion in the criteria sets for DSM-V.

Fortunately, the NIMH is now embarked on a fascinating effort to effect the real paradigm shift of basing diagnosis on biological findings. Unfortunately, this is years (if not decades) from fruition.

So long as psychiatric diagnosis is stuck at its current descriptive level, there is little to be gained and much to be lost in frequently and arbitrarily changing the system. Descriptive diagnosis should remain fairly stable until, disorder by disorder, we gradually attain a more fundamental and explanatory understanding of causality.

Indeed, there has been only 1 paradigm shift in psychiatric diagnosis in the past 100 years-the DSM-III introduction in 1980 of operational criteria sets and the multiaxial system.15,16 With these methodological advances, DSM-III rescued psychiatric diagnosis from unreliability and the oblivion of irrelevancy. In the subsequent evolution of descriptive diagnosis, DSM-III-R and DSM-IV were really no more than footnotes to DSM-III and, at best, DSM-V could only hope to join them in making a modest contribution. Descriptive diagnosis is simply not equipped to carry us much further than it already has. The real paradigm shift will require an increase in our knowledge-not just a “rearrangement of the furniture” of the various descriptive possibilities.

Part of the exaggerated claim of a paradigm shift in DSM-V is based on the suggestion that it will be introducing dimensional ratings and that this will increase the precision of diagnosis. I am a big fan of dimensional diagnosis and wrote a paper promoting its use as early as 1982.17 Naturally, I had hoped to expand the role of dimensional diagnosis in DSM-IV but came to realize that busy clinicians do not have the time, training, or inclination to use dimensional ratings. Indeed, the dimensional components already built into the DSM system (ie, severity ratings of mild, moderate, and severe for every disorder and the Axis V Global Assessment of Functioning scale) are very often ignored. Including an ad hoc, untested, and complex dimensional system in an official nomenclature is premature and will likely lead to similar neglect and confusion.18

If the potential gains of DSM-V are extremely modest, the potential risks are great and largely unrecognized. Making changes in the diagnostic system is never cheap. Just as with an individual patient, the first consideration in revising the diagnostic classification must always be to “do no harm”-and the harm inflicted by changes in the DSM diagnostic system can come in many, and usually unexpected, forms.

The most obvious cost is the significant burden to the field of having to learn and adapt to any changes included in DSM-V. This will be borne by all clinicians, educators, administrators, and especially mental health researchers.19 Changes that frivolously require new diagnostic instruments (or result in findings that are not comparable over studies and over time) waste money, slow progress, and make it far more difficult to translate research findings into clinical practice. Any “innovations” made in DSM-V should be clear and proven winners or they will not be worth the high overhead cost inherent in any change.

There is also the serious, subtle, and ubiquitous problem of unintended consequences. As a rule of thumb, it is wise to assume that unintended consequences will come often and in very varied and surprising flavors. For instance, a seemingly small change can sometimes result in a different definition of caseness that may have a dramatic and totally unexpected impact on the reported rates of a disorder.20 Thus are false “epidemics” created. For example, although many other factors were certainly involved, the sudden increase in the diagnosis of autistic, attention-deficit/hyperactivity, and bipolar disorders may in part reflect changes made in the DSM-IV definitions. Note this.

This serious unintended consequence occurred despite the fact that careful field testing of the DSM-IV versions of 2 of these disorders had predicted no substantial differences in their rates as measured by DSM-III, DSM-IV, and International Classification of Diseases (ICD)-10 criteria.21,22 The crucial lesson here is that even careful field testing is never completely accurate in predicting what will happen when the system is eventually used in the actual field.

This issue becomes particularly relevant when one considers the skillful pressure likely to be applied by the pharmaceutical industry after the publication of DSM-V. It has to be assumed that they will attempt to identify every change that could conceivably lead to a marketing advantage-often in ways that will not have occurred to the DSM-V Task Force. To promote sales, the companies may sponsor “education” campaigns focusing on the diagnostic changes that most enhance the rate of diagnosis for those disorders that will lead to the increased writing of prescriptions. As I will discuss, there is a great risk of many new “epidemics” based on changes suggested for DSM-V.

This risk is accentuated further by the fact that the field testing for DSM-V will receive no support from the NIMH. The necessary resources will not be available to measure the impact of suggested changes on the reliability and reported rates of diagnoses in the widely varied settings in which DSM-V will be used. In addition, because no DSM-V Options Book or first draft is being produced, the “DSM-V field trials” are not really field trials at all-they are no more than primary data collections that will have little to say about how the final draft of DSM-V will perform in the field. DSM-V decision making regarding changes will therefore be flying fairly blind.

A further problem is that almost everyone responsible for revising the DSM-V has spent a career working in the atypical setting of university psychiatry. This type of clinical experience is restricted to highly select patients who are often treated in a research context. It is a basic tenet of clinical epidemiology that research results and clinical experience derived from tertiary-care settings often do not generalize well when the diagnostic system has to be applied routinely in a more population-based manner.23,24

Unintended consequences are most unpredictable and consequential in forensic settings. Years after the DSM-IV was completed, we learned about the enormous unintended impact of a seemingly slight wording change we made only for technical reasons in the section on paraphilias. A misreading of our intentions in making the change had led to great confusion25-with forensic evaluators using the diagnosis of paraphilia not otherwise specified to justify the sometimes inappropriate lifetime psychiatric commitment of rapists who had no real mental disorder.

The likelihood that DSM-V will suffer from many such serious unintended consequences is enhanced by the way it is being done. “There are no constraints on the degree of change,” according to a telling quote from Dr David Kupfer, chairman of the DSM-V Task Force.26 The work groups have been instructed to think innovatively about their disorders but have received little guidance on the systematic methods of conducting literature reviews.6 It is a fundamental error of the DSM-V process that it has not put a priori methods in place to provide standards for making changes and for instructing work group members on how to do a careful risk-benefit analysis of each proposal.

Even a cursory review of some of the suggestions for DSM-V clearly illustrates the painful surprises that can inadvertently creep into a system if there are no careful doorkeepers to evaluate the risks of change. I will discuss some representative types of problems for purposes of illustration, but most of the suggested changes for DSM-V will likely have problems of one sort or another.

Undoubtedly, the most reckless suggestion for DSM-V is that it include many new categories to capture the subthreshhold (eg, minor depression, mild cognitive disorder) or premorbid (eg, prepsychotic) versions of the existing official disorders. The beneficial intended purpose is to improve early case finding and promote preventive treatments. Unfortunately, however, the DSM-V Task Force has failed to adequately consider the potentially disastrous unintended consequence that DSM-V may flood the world with tens of millions of newly labeled false-positive “patients.” The reported rates of DSM-V mental disorders would skyrocket, especially because there are many more people at the boundary than those who present with the more severe and clearly “clinical” disorders. The result would be a wholesale imperial medicalization of normality that will trivialize mental disorder and lead to a deluge of unneeded medication treatments-a bonanza for the pharmaceutical industry but at a huge cost to the new false-positive patients caught in the excessively wide DSM-V net. They will pay a high price in adverse effects, dollars, and stigma, not to mention the unpredictable impact on insurability, disability, and forensics.

In my experience, experts on any given diagnosis always worry a great deal about missed cases but rarely consider the risks of creating a large pool of false positives-especially in primary care settings. The experts’ motives are pure, but their awareness of risks is often naive. Psychiatry should not be in the business of inadvertently manufacturing mental disorders. The clinching argument against including subthreshold and prodromal “disorders” is that they are supported only by thin literatures and will not have extensive field trials to predict the extent of the false-positive risks, particularly in primary care settings. I am convinced that none of the proposed subthreshold or premorbid suggestions should be converted to official diagnoses of mental disorder in DSM-V. Each should instead be included in an appendix of suggested disorders that require more research and testing.

Another DSM-V innovation would create a whole new series of so-called behavioral addictions to shopping, sex, food, videogames, the Internet, and so on. Each of these proposals has the potential for dangerous unintended consequences by inappropriately medicalizing behavioral problems; reducing individual responsibility; and complicating disability, insurance, and forensic evaluations. None of these suggestions is remotely ready for prime time as an officially recognized mental disorder.

Getting as much outside opinion as possible is crucial to smoking out and avoiding unforeseen problems. We believed that the more eyes and minds that were engaged at all stages of DSM-IV, the fewer the errors we would make. In contrast, DSM-V has had an inexplicably closed and secretive process.27 Communication to and from the field has been highly restricted. Indeed, even the very slight recent increase in openness about DSM-V was forced on to an unwilling leadership only after a series of embarrassing articles appeared in the public press.28-30 It is completely ludicrous that the DSM-V work group members had to sign confidentiality agreements that prevent the kind of free discussion that brings to light otherwise hidden problems. DSM-V has also chosen to have relatively few and highly select advisors. It appears that it will have no Options Book to allow wide scrutiny and contributions from the field.

The secretiveness of the DSM-V process is extremely puzzling. In my entire experience working on DSM-III, DSM-III-R, and DSM-IV, nothing ever came up that even remotely had to be hidden from anyone. There is everything to gain and absolutely nothing to lose from having a totally open process. Obviously, it is much better to discover problems before publication-and this can only be done with rigorous scrutiny and the welcoming of all possible criticisms.

I have little confidence that the DSM-V leadership will do the kind of careful risk-benefit analysis of each proposed change that is necessary to avoid damaging unintended consequences.

My concerns arise from the following:

• Their ambition to achieve a paradigm shift when there is no scientific basis for one.

• Their failure to provide clear methodological guidelines on the level of empirical support required for changes.

• Their lack of openness to wide scrutiny and useful criticism.

• Their inability to spot the obvious dangers in most of their current proposals.

• Their failure to set and meet clear timelines.

• The likelihood that time pressure will soon lead to an unconsidered rush of last-minute decisions.

This is the first time I have felt the need to make any comments on DSM-V. Even when the early steps in the DSM-V process seemed excessively ambitious, secretive, and disorganized, I hoped that I could avoid involvement and believed that my successors deserved a clear field. My unduly optimistic assumption was that the initial problems of secrecy and lack of explicitness would self-correct and that excessive ambitions would be moderated by experience. I have decided to write this commentary now only because time is running out and I fear that DSM-V is continuing to veer badly off course and with no prospect of spontaneous internal correction. It is my responsibility to make my worries known before it is too late to act on them.

What is needed now is a profound midterm correction toward greater openness, conservatism, and methodological rigor. I would thus suggest that the trustees of the American Psychiatric Association establish an external review committee to study the progress of the current work on DSM-V and make recommendations for its future direction.

References:

References

1. Frances AJ, Widiger TA, Pincus HA. The development of DSM-IV. Arch Gen Psychiatry. 1989;46:373-375.
2. Frances A, Pincus HA, Widiger TA, et al. DSM-IV: work in progress. Am J Psychiatry.1990;147:1439-1448.
3. Widiger TA, Frances AJ, Pincus HA, et al. Toward an empirical classification for the DSM-IV. J Abnorm Psychol. 1991;100:280-288.
4. Widiger TA, Trull TJ. The scholarly development of DSM-IV. In: Costa e Silva JA, Nadelson CC, eds. International Review of Psychiatry. Vol 1. Washington, DC: American Psychiatric Press; 1993:59-78.
5. First MB, Vettorello N, Frances AJ, Pincus HA. Changes in mood, anxiety, and personality disorders. Hosp Community Psychiatry. 1993;44:1034-1036, 1043.
6. Widiger TA, Frances AJ, Pincus HA, Davis WW. DSM-IV literature reviews: rationale, process, and limitations. J Psychopathol Behav Assess. 1990;12:189-202.
7. Frances AJ, First MB, Widiger TA, et al. An A to Z guide to DSM-IV conundrums. J Abnorm Psychol. 1991;100:407-412.
8. Widiger TA, Frances AJ, Pincus HA, et al. Toward an empirical classification for the DSM-IV. J Abnorm Psychol. 1991;100:280-288.
9. Kline M, Sydnor-Greenberg N, Davis WW, et al. Using field trials to evaluate proposed changes in DSM diagnostic criteria. Hosp Community Psychiatry. 1993;44:621-623.
10. Task Force on DSM-IV. DSM-IV Options Book: Work in Progress. Washington, DC: American Psychiatric Association; 1991.
11. Widiger TA, Frances AJ, Pincus HA, et al. DSM-IV Sourcebook. Vol 1. Washington, DC: American Psychiatric Association; 1994.
12. Widiger TA, Frances AJ, Pincus HA, et al, eds. DSM-IV Sourcebook. Vol 2. Washington, DC: American Psychiatric Association; 1996.
13. Widiger TA, Frances AJ, Pincus HA, et al, eds. DSM-IV Sourcebook. Vol 3. Washington, DC: American Psychiatric Association; 1997.
14. Widiger TA, Frances AJ, Pincus HA, et al, eds. DSM-IV Sourcebook. Vol 4. Washington, DC: American Psychiatric Association; 1998.
15. Spitzer RL, Williams JB, Skodol AE. DSM-III: the major achievements and an overview. Am J Psychiatry. 1980;137:151-164.
16. Klerman GL. The significance of DSM-IIIin American psychiatry. In: Spitzer RL, Williams JB, Skodol AE, eds. International Perspectives onDSM-III. Washington, DC: American Psychiatric Press; 1983:3-26.
17. Frances AJ. Categorical and dimensional systems of personality diagnosis: a comparison. Compr Psychiatry. 1982;23:516-527.
18. Frances AJ. Dimensional diagnosis of personality-not whether, but when and which. Psychological Inquiry. 1993;4:110-111.
19. Zimmerman M. Why are we rushing to publish DSM-IV? Arch Gen Psychiatry. 1988;45:1135-1138.
20. Regier DA, Kaelber CT, Rae DS, et al. Limitations of diagnostic criteria and assessment instruments for mental disorders. Implications for research and policy. Arch Gen Psychiatry. 1998;55:109-115.
21. Volkmar FR, Klin A, Siegel B, et al. Field trial for autistic disorder in DSM-IV. Am J Psychiatry. 1994;151:1361-1367.
22. Lahey BB. DSM-IV field trials for oppositional defiant disorder and conduct disorder in children and adolescents. In: Widiger TA, Frances AJ, Pincus HA, et al, eds. DSM-IVSourcebook. Vol 4. Washington, DC: American Psychiatric Association; 1998:661-686.
23. White KL, Williams TF, Greenberg BG. The ecology of medical care. N Engl J Med. 1961;265:885-892.
24. Green LA, Fryer GE Jr, Yawn BP, et al. The ecology of medical care revisited. N Engl J Med. 2001;443: 2021-2025.
25. First MB, Frances A. Issues for DSM-V: unintended consequences of small changes: the case of paraphilias [published correction appears in Am J Psychiatry. 2008;165:1495]. Am J Psychiatry. 2008;165: 1240-1241.
26. Roen S. Psychiatrists rewriting the mental health bible. Los Angeles Times. May 26, 2009. http://www.latimes.com/news/nationworld/nation/la-sci-mental-disorder26-2009may26,0,3081443.story. Accessed June 24, 2009.
27. Spitzer RL. DSM-V: open and transparent? Psychiatr News. 2008;43(14):26.
28. Carey B. Psychiatrists revise the book of human troubles. New York Times. December 18, 2008. http://www.nytimes.com/2008/12/18/health/18psych.html. Accessed June 24, 2009.
29. Grossman R. Psychiatric manual’s update needs openness, not secrecy, critics say. Chicago Tribune. December 29, 2008. http://www.ajc.com/services/ content/health/stories/2008/12/29/mental_manual_critics.html. Accessed June 24, 2009.
30. Lane C. Wrangling over psychiatry’s bible. Los Angeles Times. November 16, 2008. http://www.latimes.com/news/opinion/commentary/la-oe-lane16-2008nov16,0,5678764.story. Accessed June 24, 2009.