I have not known National Public Radio (NPR) to closely follow the annual meetings of the American Psychiatric Association (APA). Maybe I just missed it, but I haven't noticed Carl Kasell roaming around the 15 conventions I've attended. Hence my surprise to hear a report on May 6, 2008 entitled "Study: Doctors overdiagnosing bipolar disorder." The interviewer described a study presented at the recent APA annual meeting by Brown University psychiatric researchers in which about 50% of patients diagnosed with bipolar disorder by community psychiatrists, upon careful re-examination by research psychiatrists, were deemed to fail to meet DSM-IV definitions of bipolar disorder. Hence, bipolar disorder is overdiagnosed.
This seems straightforward and worthy of reportage; after all, one can then blame the pharmaceutical industry for hyping up this diagnosis, as with so many others, to sell their harmful poisons, creating profits for Wall Street, irrespective of the weight gain, diabetes, and other side effects suffered on Main Street. So went the reporting - albeit with some equal time given later in the piece to researchers on bipolar disorder who cautioned that the illness is difficult to diagnose, and that many patients are also underdiagnosed as having other conditions.
I had the opportunity to see the study described in the piece months before the APA, at another research conference, and I gave the researchers the feedback I am about to write here. It could be I am wrong; or it could be that criticism is hard to hear. Either way, I will risk repeating my view, since I think there is a major scientific mistake here.
What is wrong with the Brown study? It seems like straightforward overdiagnosis. Well, it may represent misdiagnosis, but whether it is overdiagnosis needs to be shown in another way. Overdiagnosis means that it is made more than others: where is the control group which is underdiagnosed, and mistakenly labeled bipolar, by contrast? There was none.
Perhaps more importantly - and this is the critique I made directly to the researchers, to no apparent avail - the study mistook reliability and validity, two terms that need definition. Reliability means (in this case) that two doctors call an illness (say, bipolar disorder) the same thing; what they call it may be right or wrong (their definitions may or may not be right) but at least they agree on what to call it (their definitions). Validity is about whether their definitions are right or not.
This study assessed reliability - to what extent doctors agree - not validity - how frequently clinicians are wrong.
Put another way: This study shows that when people are called bipolar, they do not have it half the time. (The same applies for all psychiatric conditions, see below). But many other studies show that when people actually have bipolar disorder, they are not diagnosed with it about half the time.
This is the problem, then: There is disagreement about diagnosis of bipolar disorder, but it still remains underdiagnosed, not overdiagnosed.
Now the explanation:
Reliability studies start with a group of diagnoses, which may or may not be correct, as with the Brown study. This group of patients was seen as bipolar by clinicians. Then researchers (or a second group of clinicians) reassess the same patients with what is our current gold standard (a research diagnostic interview with DSM-IV criteria). They disagreed about 50% of the time. That looks bad. But the claim that it represents overdiagnosis of bipolar disorder runs aground on the fact that such data also exist with similar results when the initial diagnosis by clinicians is unipolar depression, or schizophrenia, or alcoholism, or obsessive compulsive disorder, or (for that matter) congestive heart failure. In the real world clinical practice of psychiatry (and much of medicine), doctors frequently disagree. Reliability of clinical diagnoses for any psychiatric diagnosis is rarely more than 50%. In one large community-base study (the Epidemiologic Catchment Area study, ECA), reliability of psychiatric diagnoses ranged from 5-35%. Thus, all diagnoses are overdiagnosed!
But that conclusion is mistaken too. To claim the wrong diagnosis (whether over or under), we must claim validity. We must know whether or not the diagnosis is valid, before we can tell whether it is being over or underdiagnosed. We need to start with valid diagnoses of bipolar disorder, and then assess past clinician's diagnoses to see whether they were right - not the other way around, as was done in the Brown study (and indeed in most studies claiming overdiagnosis).
Here is proof of underdiagnosis: a validly diagnosed bipolar sample would have been diagnosed, in part, as having other conditions by past clinicians. Here is proof of overdiagnosis: a validly diagnosed bipolar sample would have been diagnosed, almost always, as having bipolar disorder by past clinicians, and validly diagnosed other conditions (like schizophrenia or unipolar depression) would have been diagnosed, in part, as having bipolar disorder by past clinicians.
That is the way to do it: yet no such study has ever been done that shows overdiagnosis of bipolar disorder. In contrast, a few such studies have been conducted and shown underdiagnosis of bipolar disorder, and overdiagnosis of schizophrenia, unipolar depression, or ADHD. In those studies, only about 40% of persons with bipolar disorder receive that diagnosis despite repeated manic episodes. They are misdiagnosed with the other conditions, receive the wrong medications (antidepressants, stimulants, or antipsychotics instead of mood stabilizers), and lead miserable lives for, on average, a decade, seeing over 3 psychiatrists, before they get correctly diagnosed.
Doctors may call conditions bipolar that are not bipolar, just as they call conditions congestive heart failure that are not congestive heart failure (unreliability), but they also consistently and demonstrably fail to diagnose bipolar disorder when it exists, while diagnosing other conditions (like depression or ADHD) not only in those who have them but in those who have bipolar disorder (underdiagnosis).
My experience supports the scientific literature just described: I've seen about a thousand such patients in the last decade, and I've seen their lives turn around when they get off the wrong drugs and get on the right ones.
This aversion towards bipolar disorder is a matter of some cultural interest. It is an historical fact, worthy of note, that bipolar disorder has generally not been commonly diagnosed. It was first described 150 years ago by French and later German psychiatry (especially Emil Kraepelin, pictured), much as it is now. (NB: There were no functional pharmaceutical companies in that era).
But for much of the 20th century, the most commonly diagnosed mental disorder, by far, was schizophrenia. In the 1950s, for instance, when the first antidepressants were developed, the pharmaceutical industry was relatively uninterested, because schizophrenia was believed to be far more prevalent. A half century of interest in depression has followed - and continues: depression received increasing attention, and a slew of medications were developed and marketed for it.
Bipolar disorder remained an orphan, with a single generic drug - lithium - that was hardly marketed and infrequently used. Until the last decade, other mood stabilizers were not proven or marketed, and now that some attention is being given to them, academics and skeptical clinicians raise concerns. The fact remains, though, that despite being at least as common as schizophrenia (probably more), and perhaps one-third as common as depression, research funds for, and scientific studies about, bipolar disorder represent one-fifth or less of what is spent on, or published in, either schizophrenia or unipolar depression. Perhaps four drugs now qualify as mood stabilizers, compared to more than three times as many antipsychotics or antidepressants respectively. There are about twenty research centers on bipolar disorder in American universities, versus hundreds for schizophrenia or depression separately. The pharmaceutical industry begins research on many drugs in animal models of depression or psychosis, but hardly ever mania; thus drugs are rarely specifically developed for bipolar disorder.
Too much attention would seem to be the last problem with bipolar disorder.
Rather, there seems to be a cultural resistance to the whole concept, whereas depression or even schizophrenia seem to have been more palatable to researchers, clinicians and the public. Also, perhaps the claim of overdiagnosis itself is attractive: people generally want to be told they are less ill, rather than more. And attacks on the pharmaceutical industry, though often valid, easily follow in what seems to have become a sudorific sport.
To sum up: Unreliability, yes (like most psychiatric illnesses); overdiagnosis, no (unlike many other psychiatric illnesses) - a century and a half later, and still counting.