The diagnostic boundary between Major Depressive Disorder and Bipolar II Disorder is one of the most difficult and also one of the most important in psychiatry. Difficult (and inherently at least somewhat arbitrary) because the distinction rests completely on how one defines a hypomanic episode. Important because the choice of diagnosis determines very different treatment approaches that are likely to lead to radically different outcomes and side effects.

DSM-5 is considering a further expansion of Bipolar Disorder—but unfortunately has not done a careful calculation of its risks and benefits. To remedy this deficit, I invited the assessment of Professor Mark Zimmerman of Brown University. He has collected and analyzed systematic data on a very large number of patients and has thoughtfully reviewed the available literature. Dr Zimmerman writes:

Critics of DSM-IV worry that its requirement of at least four days to define a hypomanic episode is too long and suggest lowering it to two days—so as to pick up individuals currently excluded from the bipolar diagnosis. Their claim is based on clinical and epidemiological findings suggesting that people with subthreshold levels of bipolar pathology differ from unipolar depression in comorbidity, personality, family history, and longitudinal course.

Expanding the diagnostic criteria for bipolar disorder carries with it the potential benefit of picking up true bipolar patients who are currently missed. Advocates of lowering the duration threshold emphasize the costs of missed diagnoses—under-prescription of mood stabilizing medications, an increased risk of rapid cycling, and increased costs of care.

But a more balanced approach toward the question of where to set the diagnostic threshold also recognizes the adverse consequences of false positive diagnoses—especially the over-treatment with unneeded medications and consequent exposure to weight gain and its medical risks.

Keep in mind that in clinical practice patients are followed over time and clinicians do not rigidly adhere to the DSM-IV diagnostic rules. While under-diagnosis due to insufficient duration is a theoretical possibility, it is likely that patients with subthreshold hypomanic episodes will, during the course of treatment, ultimately be diagnosed with bipolar disorder and treated accordingly.

False negative diagnoses are thus much easier to correct than false positive diagnoses, which tend to be long lasting and difficult to undo. It is always more difficult to take away the bipolar diagnosis once it is made than to add the bipolar diagnosis once a new hypomanic episode occurs. The patient with a false positive diagnosis of bipolar disorder who is doing well on a mood stabilizer is unlikely ever to have the mood stabilizer discontinued or the diagnosis corrected—even if the medicine is totally unnecessary and is causing harmful weight gain. The absence of recurrent hypomanic episodes is incorrectly viewed as treatment success.

Any lowering of the bipolar threshold should be supported by strong evidence from prospective follow-up studies that individuals with subthreshold bipolarity are really at high risk for developing bipolar disorder. The four available studies indicate that although subthreshold bipolarity is a risk factor for the future emergence of bipolar disorder, the vast majority of individuals do not develop bipolar disorder during the 10 to 20 years of follow-up. These findings provide no real support for reducing the duration requirement.

The strongest evidence to support expanding the definition of bipolar disorder would of course be the demonstration that mood stabilizers are helpful in subthreshold presentations. But there is not a single controlled study of the efficacy of mood stabilizers in this situation. It makes no sense to reduce the bipolar threshold in the absence of controlled research establishing that this will improve treatment efficacy, especially since we know that the change will add a large side effect burden.

There is already a substantial false positive problem with the existing DSM-IV criteria that require four days duration for hypomania. Lowering the threshold for hypomanic episodes even further will likely increase the over-diagnosis and over-treatment of bipolar disorder. Most disturbing, there are no studies of the potential impact the shorter duration requirement would have on diagnosis and outcome in real-world clinical practice.

It must also be noted that it is often extremely difficult to diagnose a true hypomanic episode. This is particularly the case when substance use is part of the picture. And someone who is depressed a lot may confuse periods of normal mood with being high. Also transient episodes of emotional lability in borderline personality disorder may be confused with hypomanic episodes.

Shortening the duration to only two days will make the elusive diagnosis of a hypomanic episode even less reliable.

It is best to be cautious and to have much stronger evidence before making a change that will have unknown and possibly very harmful consequences.

Thank you, Dr Zimmerman, for providing this well thought out and thorough risk/benefit analysis.

Across the board, DSM-5 proposals consistently fail on 4 counts: (1) their exclusive emphasis on eliminating missed diagnosis with a concomitant lack of concern about false positive over-diagnosis; (2) their neglect of risks when considering benefits; (3) their indifference to historical expectations regarding reliability, and, (4) the lack of empirical support for suggestions that can have profound real life consequences.

There has been a doubling in the ratio of bipolar to unipolar depression since DSM-IV. Some of this growth was the anticipated result of our introducing Bipolar II as a new diagnosis in DSM-IV. Some was caused by massive drug company marketing with resulting loose diagnostic and prescription habits—especially in primary care practice.

DSM-5 should be promoting more careful diagnosis of Bipolar Disorder, not a further reckless expansion.