Keep criteria sets stable
Criteria sets should not be changed except for very good reasons and when there is exceptionally strong supporting evidence. The most important criteria sets have a long track record with substantially constant wording ever since the publication of DSM-III in 1980. They have survived the test of time and have become familiar to generations of clinicians and educators. Any arbitrary changes will be needlessly disruptive.
Especially problematic is the impact of changed criteria sets on research. All the widely used interview instruments have for decades been keyed to the existing DSM criteria items. Changed wording would be enormously disruptive to the conduct of future research and to the interpretation of the existing clinical and epidemiologic literature, which would no longer be consistent with new findings based on the changed diagnostic definitions. The research community will be justifiably upset by a DSM-V that liberally alters criteria sets when there is no clear reason to do so.
The forensic applications of the DSM system can be extremely sensitive to even slight changes in the criteria sets. I have discussed previously how a seemingly trivial “or” for “and” wording change in the DSM-IV paraphilia section had a very unfortunate impact on the civil commitment of rapists.1 It must be anticipated that the legal system will parse DSM-V wording changes much more precisely and idiosyncratically than can ever occur to work group members. Experts in forensic psychiatry will therefore also be very concerned about any changes in DSM-V and should be recruited to vet the wording of every option with a fine tooth comb.
Problems can result even from improvements in the wording of criteria sets. For example, the better written and more easily remembered DSM-IV criteria set for ADHD may have resulted in its overuse—especially by primary care doctors and the general public.
However perilous it is to change existing criteria sets, the risks are much greater still whenever the system adds totally new diagnoses that are at best lightly tested. The potential for false positive epidemics and forensic conundrums are much harder to predict for anything novel. New disorders are best kept in the appendix until they have achieved wide acceptance in the field. The DSM system should always follow, not lead, research and practice. It can never be paradigm shifting on its own weight.
The final caution, if one were needed, is that it is surprisingly difficult to write clean, foolproof criteria items. I know this from frustrating personal experience. Despite many years of effort and practice, I never mastered this highly technical writing skill. Until the actual DSM-V options are publicly posted, it is impossible to judge whether they will meet the necessary standards of precision, clarity, and consistency. However, the obscure writing style displayed in the available conceptual papers about DSM-V do not inspire confidence in this regard.2,3 Moreover, it is troubling that no one working on DSM-V has had any extensive experience in writing diagnostic criteria.
Areas for Innovation
DSM-V should update and greatly improve the tired, old text of DSM-IV-TR. Most in need of exhaustive revision are the text sections on biolological factors, epidemiology, and the developmental, cultural, and gender contributions to diagnosis. But all of the DSM-IV-TR text should be up for grabs. Its current formulaic style fails to convey any of the vividness of actual clinical practice. There could be less rote repetition of the wording of items in the criteria sets and much more illustration with rich clinical examples.
The DSM-V task force has suggested another possible innovation: the reorganization of the grouping of disorders. Obsessive-compulsive disorder might be pulled from the anxiety disorders and placed as the lead of its own section with accompanying spectrum disorders (eg, Tic Disorders, Body Dysmorphic Disorder). The section on Disorders First Diagnosed in Infancy, Childhood or Adolescence might be eliminated altogether (or stripped down) and its component disorders given their own sections or sorted with their closest counterparts in other sections (eg, Separation Anxiety Disorder with the Anxiety Disorders).
I am neutral on these suggested reorganizations—plausible arguments can be made either way. But the point here is that such restructuring is much less risky than changing criteria sets. Similarly, the suggestion to add dimensional ratings to DSM-V has its pluses and minuses, but is more likely to be neglected by clinicians than to cause any serious harm.
Undoubtedly, the most valuable innovation possible for DSM-V would be an integration with ICD11, but this important topic requires another column.