Let us suppose that we are gods. (Many physicians are said to suffer from this delusion; here it is meant as a thought experiment). As gods, since we know everything, we know that a certain disease exists. Let's call it disease X; in this disease, initial symptoms Y last 2-3 days, followed by other severe symptoms Z, which last weeks to months, eventually leading to death, if repeated, in 5% of persons.
Let us further suppose that doctors, being ignorant human beings rather than gods, mistakenly diagnose that disease only when its initial symptoms last 4 days or longer.
Let us also suppose that no known effective treatments exist yet in the world.
Would the presence or absence of treatments make any difference as to the reality of the disease?
No. The reality of a disease is what it is, irrespective of what treatments may or may not be available, and irrespective of what human beings think.
There are, and always have been, many diseases in medicine for which no treatments have been available, but it has always still been useful for doctors to understand them as well as they can; thankfully, treatments are often later developed.
Limitation in treatment does not justify ignorance of disease.
Enter Allen Frances, the head of psychiatry's current diagnostic system, DSM-IV, who has been vocal in his criticism of much of the next revision, DSM 5. In a recent blog post, he aims his critique at any whiff of risk of expanding the diagnostic definition of bipolar disorder, type II. His main rationale: such patients would receive antipsychotics or mood stabilizers, which have physical side effects.
But here is the question: What is the scientific rationale for leaving matters where they are, at the 4 day criterion for hypomania introduced by DSM-IV? What is the scientific basis for that 4 day cut-off as opposed to 3 or 5 days? Was there ever a scientific basis for it? (In diagnosis research, scientific validation means studies comparing clinical samples on the five independent diagnostic validators of symptoms, course of illness, genetics, treatment response, or biological markers)
The answer is again no. David Dunner, a member of the DSM-IV mood disorders task force, stated as much by admitting that the 4 day definition of hypomania was "arbitrary." He later added that members of the task force wanted to keep the duration criterion high based on fear of overdiagnosis of bipolar disorder, rather than any scientific evidence for that number.
In contrast, in the past decade, numerous diagnostic studies (one is example is here; this extensive research literature was summarized in consensus recommendations of a task force of the International Society for Bipolar Disorders, which I chaired) show that a 2-3 day cutoff is scientifically valid: it differentiates samples with bipolar disorder (based on course of illness, genetics, and treatment response) from samples with non-bipolar depression. So we have reasonable scientific evidence for a shorter cut-off; and our current definition was arbitrary, based on no evidence.
Dr. Frances trots out the old threat of overdiagnosis to ignore a decade of science. He does so despite another decade of research repeatedly showing that bipolar disorder is underdiagnosed in about 40% of persons. In direct comparisons with major depressive disorder (MDD), bipolar disorder is underdiagnosed, while MDD is overdiagnosed. Even in reported claims of bipolar overdiagnosis, as I discuss also in a previous post and recently in the British Medical Journal, the data from those same studies show that bipolar disorder is twice more frequently underdiagnosed than overdiagnosed.
The overdiagnosis fear does not stand up to scientific scrutiny.
We are left with the drug side effects. I agree that antipsychotics are overused but I also think that antidepressants are overused. Why is it better to use antidepressants excessively in persons who do not have an illness that antidepressants treat? Especially when we now know that antidepressants can also worsen that very illness in some persons?
In a recent large analysis of practice patterns in the US, mood stabilizers - lithium, valproate, carbamazepine - are in fact the least frequently used class of psychotropic medications, and the only class whose use has not increased in the last decade.
So the drug fear is also overstated.
Even if we had no drugs, though, the point is irrelevant to the reality of trying to get our diagnoses right. The scientific aim is to diagnose mental illnesses correctly (not too broadly - yes, but also not too narrowly) and to diagnose when mental illnesses are not present.
Dr. Frances is to be thanked for the benefits of DSM-IV, but we should use science to correct and advance it. Of course, politics is unavoidable, and practical considerations are relevant, but we should not ignore legitimate scientific evidence. DSM-IV has been legitimately criticized for many reasons, and so will DSM 5. In my view, those criticisms are most legitimate when aimed at sacrifices of scientific knowledge to political expedience.