Should Practical Consequences Influence DSM5 Decisions? Yes, Of Course

I often get asked if practical consequences should play an important role in DSM5 decisions. It was posed again yesterday in response to my blog "Bipolar II Revisited" which tangentially raised the issue. A reader wondered whether the availability of safer treatments would impact on how best to set the boundary between unipolar and bipolar disorder. My quick answer is a very emphatic yes—but this is so important and fundamental an issue that it deserves its own separate, fuller explanation. In fact, pragmatic concerns must play a central role in shaping any DSM.

Why is this the case? DSM is an official system of classification that has a huge (perhaps excessive) influence on how everything works in the mental health world—who gets diagnosed, how they are treated, who pays for it, whether disability is appropriate, and whether someone can be involuntarily committed, released from legal responsibility, or sue for damages.

DSM also has a diverse influence on public policy- directly or indirectly influencing things as varied as the way scarce treatment and school resources are allocated, the impact of medication on the obesity/diabetes epidemic, and how sexual offenders are (mis)handled in the legal system.

Ever since the introduction of DSMIII, the DSM system has been a great promoter of psychiatric research and the principal means of translating across the clinical/research interface. But, DSM is decidedly first and foremost a clinical document, with its other uses being important, but definitely secondary. As an official diagnostic system, DSM is not meant to place its highest priority on promoting or facilitating the latest in research ideas.

Because it has such a powerful influence on real life (and occasionally even life or death) decisions, DSM can't ignore its practical consequences—intended or unintended. It has to be workaday, trying very hard not to make mistakes that will hurt people, rather than having fancy (but untested) "paradigm-shifting" ideas that almost always wind up doing more harm than good.

This brings us back to the question—if we had safer treatment would this importantly swing the risk/benefit analysis concerning how best to define Bipolar II? Sure it would. If we had a free and risk-free medication that prevented mood swings in unipolar patients, there would be no (or little) cost to opening the floodgates to bipolar diagnosis. However, the corollary to this is that as the treatment costs go up, the threshold to the bipolar diagnosis necessarily must also be set higher.

This answer may well bother any reader who wants to define Bipolar II Disorder the "way it really is" and damn the consequences. Or stating the same point another way—"Let's let science and experts decide the boundaries of bipolar (or any other) disorder." It simply does not work that way.

Our definitions of all the DSM disorders are always and necessarily at least somewhat arbitrary. There is no one right definition. Cutoffs requiring say 5 of 9 criteria could just as plausibly have used 4 or 6 to set the boundary. A cutoff of 5 was chosen as the best balance of sensitivity and specificity—that is, trying to find an acceptable trade-off between the inevitable opposing risks of under- and overdiagnosis. But admittedly, in some situations, 4 might work better (especially if each symptom is severe); in others, perhaps 6 should be required.

There is no scientifically proven, just-one-right, way to diagnose any mental disorder—and don't let any expert tell you differently. Clinical judgment is required in diagnosing any individual patient and Do No Harm common sense is always necessary in establishing any of the DSM thresholds.

Finally, as to the question of how best to make DSM decisions, much has been written about the "validators" of psychiatric diagnosis and how they should influence DSM. The problem is that available information on the validators for most diagnoses is usually equivocal and inconsistent. Validators never reach out, grab you by the throat, and say, "Do it this one way or the science gods shall be displeased.”

To my mind, by far the most important validator is how any decision will help or harm patient care, given the foreseeable circumstances under which it will be used. Let's go back to how this practical, common sense approach works for the boundary between unipolar and bipolar disorder. Start with the facts that there is no biological test to make the distinction and no certain way to know what the appropriate ratio should be among mood disorder patients.

We do know 1 important fact: The ratio of bipolar diagnoses have at least doubled since the introduction of Bipolar II in DSM-IV and the extraordinary drug marketing campaign promoting antipsychotics and mood stabilizers. This has undoubtedly helped some people and harmed others; the exact extent of each is unknown and perhaps unknowable. But my bet is that this is a fad that has overshot—fads always do. I would assume that anyone now presenting with anything suggesting equivocal bipolar disorder is much more likely to be overdiagnosed and overtreated than to be missed. Close, watchful waiting in doubtful cases beats rushing in with potentially dangerous meds.

DSM5 should always take full account of the risks—not just the benefits—of getting a diagnosis; it should factor in the side effects and complications of real world treatments (usually medications) that will follow. Those working on DSM5 must take responsibility for the practical consequences that their decisions will have on peoples' lives.