How arbitrary description of mental illness messes up public policy
In May, the American Psychiatric Association (APA) issued the latest revision of its official compendium of mental illnesses, the Diagnostic and Statistical Manual of Mental Disorders 5. The DSM, since its debut in 1952, has been the savior and the curse of psychiatry. The oft-revised manual offered a skeptical public the promise of rigorous classification of abnormal behavior and psychological distress into distinct disease entities, proffering science in the place of storytelling. But as with every previous revision, this newest update has thrown the field of psychiatry into turmoil. Clinicians are struggling to master new diagnoses. Researchers are piecing together the new disorders with the old literature. And insurance companies are wondering how much cash newly minted mental disorders will legally require them to dole out for their customers.
Allen Frances, former head of the Duke University School of Medicine's psychiatry department and a man The New York Times once called "perhaps the most powerful psychiatrist in America," was chair of the APA task force for DSM-IV (issued in 1994). Frances has become the loudest
and most influential public voice questioning the DSM's latest revamp. "The fact that we had a descriptive system only revealed our limitations," he says. "If you believe that labels are only labels, you don't want to keep changing the language arbitrarily. It just confuses everybody."
Frances has been warning everyone who will listen that the newest DSM revision will turn even more of human suffering into mental illness and thus into grist for the pharmaceutical mill. I once asked him whether he thought a good definition of a mental disorder would establish a bright boundary line that would definitively sort the sick from the unusual and keep psychiatry in its proper place.
"Here's the problem," Frances said. "There is no definition of a mental disorder."
I mentioned that that hadn't stopped him from putting one into the DSM-IV. "And it's bullshit," he said. "I mean you can't define it." Frances came to regret saying this to me after I published it in a January 2011 Wired article. He worried that his offhand comment would give aid and comfort to the enemies of any sort of psychiatric care whatsoever, while all he meant to do was to question the attempt to create airtight definitions of mental illnesses.
Doubts like Frances', on the part of both respected professionals and confused laypersons, are inevitable given the way the APA conducts nosology, the science of disease classification. The DSM defines mental disorders by listing their symptoms. People who have the symptoms of a given diagnosis—in DSM parlance, people who meet the criteria—can be given the diagnosis if a clinician believes their suffering is clinically significant—an assessment that is largely left up to the doctor. Since the descriptive method first appeared, in the DSM-III (1980), it has provided psychiatrists with a common language for discussing their patients and given millions of patients access to care they believe is vital for coping with their life problems. Since 1980, however, many critics have pointed out that these disease categories do not exist in nature, at least not in the same way that streptococcus and influenza do.
If the DSM is not a map of an actual world whose contours can be independently confirmed, then opening up old arguments or starting new ones is an invitation to chaos. With each revision of the DSM comes the potential for instability and discord that cannot be settled by turning to the microscope or the computed tomography (CT) scanner. Knowing this, Frances says the goal of a DSM revision should be to stabilize an inescapably fragile system rather than to perfect it—or, as he put it to me, "loving the pet, even if it is a mutt."
The mutt has certainly caused some mischief. Among its more prominent detractors was Steven Hyman, who in 1996 became the head of the National Institute of Mental Health (NIMH). A neurogeneticist by training, Hyman hadn't thought much about nosology before taking over at NIMH. It "seemed a bit like stamp collecting," he once wrote in a 2010 article in the Annual Review of Clinical Psychology, "an absorbing activity perhaps, but not a vibrant area of inquiry."
But then he realized that the DSM was "a critical platform for research." Its categories and criteria were the basis of decisions made by journal editors, grant reviewers, regulators, and the Food and Drug Administration, which meant that scientists were bound to frame their proposals in the DSM's language. "DSM-IV diagnoses controlled the research questions they could ask, and perhaps, even imagine," he wrote.
"The tendency [is] always strong," John Stuart Mill wrote in 1869, "to believe that whatever receives a name must be an entity or being, having an independent existence of its own." To Hyman, who quoted Mill approvingly, this tendency had led all the stakeholders in nosology—scientists, regulators, editors, doctors, drug companies, and, of course, patients—to take the labels not as arbitrary descriptions but as the names of actual diseases. They had, at least according to Hyman, reified what were intended only as concepts.
And this was no mere abstract concern. "It became a source of real worry to me," Hyman said, "that as Institute director I might be signing off on the expenditure of large sums of taxpayers' money for…projects that almost never questioned the existing diagnostic categories despite their lack of validation." The DSM, Hyman concluded, had "created an unintended epistemic prison," and anyone with a stake in the mental health treatment system was trapped inside.
The Termite-Riddled Foundation
While he was at the National Institute of Mental Health, Hyman had occasion to confide his reservations to at least one colleague: Steven Mirin, then the APA's medical director. On a weekend afternoon in the summer of 1998, the two were eating lunch by the side of Mirin's swimming pool in the D.C. area when Mirin asked Hyman if the NIMH would give the APA money to get the next revision of the DSM up and running.
Mirin's request for taxpayer money to kick-start a project from which a private organization would profit was not as untoward as it might seem. After all, the DSM is indispensable to public health, and the NIMH had helped fund the DSM-IV. Nonetheless, and despite their friendship, Hyman said no. He told Mirin that a revision was premature, not only because the ink was barely dry on DSM-IV but also because psychiatrists had yet to come up with a better way to chart the landscape of mental illness.
All they could do, Hyman thought, was continue creating and refining concepts that would then be mistaken for real disease entities, further trapping psychiatry in its epistemic prison. Until someone figured out how to fashion an escape plan, Hyman didn't think there was much point to another revision. After all, you don't remodel a house when the foundation is infested with termites.
Mirin didn't fight back—mostly, he says, because he didn't disagree. "The DSM was a system based on descriptive criteria influenced by experts in the field," Mirin tells me. "They had lots of opinions, but these couldn't necessarily be validated. It's one thing to guess and another to biopsy a tumor or to measure an enzyme."
Even so, as much as America's leading psychiatrists knew the DSM was flawed, they didn't have anything to replace it with. "I realized that it got me nowhere to criticize the DSM because that did not offer a constructive alternative," Hyman tells me. "In fact, given the way the DSM had controlled the imagination of scientists, there was little information with which to see beyond it."
Hyman may have been anguishing about psychiatry's predicament, but Mirin wasn't losing any sleep over the fact that his profession was stuck guessing about categories that didn't really exist. "I don't recall feeling particularly tortured about it," he says. "The DSM was essential to being paid for treatment. Without its methodology, payers would see mental illnesses as figments of a provider's imagination." It was also essential to the APA's finances. After all, Mirin tells me, "coming down the mountain with the Ten Commandments sure sells a lot of books."
Too Much Mental Illness
In 2002 Congress was considering a bill requiring insurers to pay for mental health care at the same level as other medical services, across the entire range of DSM-IV diagnoses. Parity, as this mandate was called, had been implemented in a handful of states, often only for mental disorders considered by insurers to be biological in origin, such as schizophrenia and bipolar disorder.
President George W. Bush endorsed the bill, but it seemed likely to sink into the mud of the legislative process, in part, Mirin thought, because The Washington Post—"the hometown paper of every member of Congress," as he puts it—had twice come out against parity. Mirin took it upon himself to convince the Post to change its mind.
Mirin arrived on September 2, 2002, expecting an hour with a single editor, and was pleasantly surprised to find six editors and a reporter willing to give him nearly 90 minutes of back and forth. The journalists did their best to torment Mirin with the discrepancy between the DSM's authority and the science behind it, asking questions, as people present at the meeting recalled, such as "How do you diagnose mental illness?" and "How do you tell if it's real?" and "Do you have a science base like the rest of medicine does?"
Mirin brought with him an expert on diagnostic questions named Darrel Regier, whom he had recently hired to head up the APA's research arm, the American Psychiatric Institute for Research and Education. Mirin had recruited Regier from the NIMH, where he had risen to the rank of vice admiral in the Public Health Service. Regier, an epidemiologist as well as a psychiatrist, had been measuring the levels of mental illness in the population since the earliest days of DSM-III. And what he had seen didn't inspire confidence.
As the head of the NIMH's Epidemiological Catchment Area (ECA) team, Regier supervised a group of researchers who, starting in 1980, fanned out across five U.S. cities armed with a questionnaire keyed to the diagnostic criteria in DSM-III. They asked 20,000 people, selected to reflect the general population, about their worries and their sadness, about whether they heard voices, about how they slept and ate. They tabulated the results and in 1984 began to release them in a series of journal articles.
The findings were stunning. In any given year, more than 20 percent of Americans qualified for a DSM-III diagnosis. Nearly one-third of the population—80 million people, according to the 1990 census—would have a mental illness during their lifetimes. And the sick among us were really sick. Sixty percent of those diagnosed with a mental illness had a comorbid disorder, meaning they qualified for at least two diagnoses. Ninety-one percent of people with schizophrenia had at least one other diagnosis, as did 75 percent of people with a depressive disorder. Fifteen percent had three or more diseases. More than half the people with a drug-related diagnosis, such as cannabis abuse, also had at least one more diagnosis.
Even more alarming, only 19 percent of the afflicted had sought help for their troubles, a number that dropped to 13 percent in the cases where only one diagnosis was warranted. It seemed that America had an enormous but undiagnosed and untreated public health problem whose effects on productivity, family life, and the body politic were unfathomable.
This potential fivefold increase in the size of the market for psychiatry wasn't so much an embarrassment of riches as a plain old embarrassment. Even accounting for the fact that epidemiological studies, in which researchers go out looking for trouble, almost always yield bigger numbers than studies that rely on numbers gleaned from doctors' offices and hospitals, the results beggared imagination. They also cast doubt on the DSM itself. The questions had to be asked: Was the problem in the minds of the people or in the methods of the doctors? Did the DSM-III make it too easy to turn people's everyday troubles into disease?
Two decades after his research, Regier thinks the answers are all too clear. "We just don't have good thresholds for identifying what we would consider mental disorders," he told me in 2010.
Persuading the Post
In the DSM, if you have five of the nine symptoms of depression, you have the same "major depressive disorder" as a person with all nine. This parallels other kinds of medical diagnosis: If you have a small Stage 1 tumor in your lung, you have the same disease as someone who is about to die from lung cancer.
But as Regier's numbers made clear, doctors using DSM checklists were all too likely to find disease everywhere they looked. There was no governor in the mechanism, no way to say this person was sick and that one was simply unhappy, nothing like the CT scan that shows a persistent cough to be caused by a lung tumor. A doctor who diagnosed, say, strep throat entirely on the basis of symptoms, rather than availing himself of the quick and easy lab test for the disease, was practicing bad medicine. But a doctor who diagnosed depression only on the basis of symptoms was practicing standard psychiatry.
The comorbidity rates—the frequency with which people qualified for more than one diagnosis—were another embarrassment. Here again, Regier said, the ECA studies pointed not so much to a sick population as to a flawed set of criteria. Robert Spitzer, the leader of the DSM-III revision in 1980, had anticipated the possibility of multiple diagnoses, and in the introduction to that book he suggested that there was a hierarchy of mental illness, that some disorders only had a narrow range of symptoms while others contained multitudes. Schizophrenia, for instance, was far more encompassing than major depression, so clinicians confronted with a patient presenting symptoms of both were advised to render only a schizophrenia diagnosis on the assumption that the low mood was part of the more comprehensive disorder.
Regier countered that this amounts to a claim that depression itself is "just noise," of no inherent interest or value in understanding the patient or his disorders. The ECA team found that people with symptoms of both schizophrenia and depression were different from people with only schizophrenia in many ways. Ignoring their depression meant failing to paint a complete diagnostic picture and losing "an enormous amount of data" about mental health. "The ECA blew the hierarchy out of the water," Regier says proudly. "It just didn't make any sense when we started looking at the data."
Such concerns led the APA to abandon the hierarchy model of DSM-III-R (a limited 1987 reworking of the manual). But the real problem, Regier tells me, was not the approach but something much more basic: the idea that DSM disorders are discrete diseases that exist in nature in the same way as cancer and diabetes. To Regier, this is the original sin, the flaw that accounts for the high rates of both prevalence and comorbidity. "It makes it seem like an anxiety disorder doesn't have any mood symptoms and a mood disorder doesn't have any anxiety symptoms," he said. "But it isn't that simple. It's just not the way people present."
But it is the way the DSM continues to present mental illness; indeed, that neat separation is the signal innovation of the DSM-III. Fortunately for Mirin and Regier, by the time of their fateful meeting with the Post editors, they'd turned their skepticism into a strategy.
"We walked them through how we understood mental illness, and what our thoughts were about diagnosis and the DSM," Mirin recalls. Not, of course, their thoughts about the book's failure to correspond to clinical reality or about the way the categorical approach trapped diagnosticians in a tautological loop (which, after all, were highly technical matters, known and understood only by experts), but rather their thoughts about the troubles reported in the daily paper that might make the average editor skeptical: the shifting sands of psychiatric diagnosis, the prevalence rates, the frequent and repeated revisions of the nosology, the disorders that came and went with dismaying regularity. These they readily acknowledged but then turned to their advantage.
The problem, they said, wasn't that psychiatry was inexact when compared with the rest of medicine but rather that the rest of medicine was nowhere near as certain as it was cracked up to be. The glucose levels that constitute diabetes, the cholesterol counts that call for treatment, the blood pressure that qualifies as hypertension—these numbers had all changed over time as well, and after no small amount of wrangling. To hold psychiatry to a more stringent standard was unfair, making victims of doctor and patient alike.
It was the perfect approach for an audience of journalists. "They were smart people," Mirin says. "They were sophisticated enough to understand that what their doctor told them about hypertension was not carved in stone, either." If the Post's editors noticed the intellectual sleight of hand at work here, the way these leading psychiatrists were distancing themselves from the claims to certainty that had allowed the DSM to rescue psychiatry from the pseudoscience precipice, they didn't say, at least not in print. Perhaps they were afraid they'd seem unsophisticated.
Either way, six days after the meeting, the paper came out in favor of parity, Congress passed a limited version of the bill, and mental health professionals everywhere rejoiced. Six years and many editorials later, parity became the law of the land with the Mental Health Parity and Addiction Equity Act. Mirin and Regier's strategy succeeded. They had spun the dross of diagnostic uncertainty into gold.
Overselling the Prevalence of Mental Illness
Maybe you think Mirin and Regier were just trying to have it both ways, in which case you might still think those prevalence numbers are a little fishy. You might wonder what would happen if 25 or 30 percent of the population exceeded the standard glucose or blood pressure thresholds, but only 19 percent of that group—about 5 percent of the population—ever got sick enough to show up at a doctor's office. Wouldn't that state of affairs cast doubt on whether those numbers measure disease at all, raising the possibility that the whole idea of hypertension or type 2 diabetes had been cooked up an industry eager to sell treatments?
You might also point out that, regardless of whether they are diseases in themselves or only risk factors, blood pressure and glucose levels can at least be measured with a high degree of certainty. And you might then ask whether it's really fair to compare conditions such as high glucose and blood pressure to mental illnesses, whether telling a patient he has hypertension for which he should take diuretics is really the same kind of intervention as telling him he has a chemical imbalance that antidepressants will correct.
Darrel Regier would have an answer for some of those questions. You think those numbers are high? Well, he would tell you, you should see the Midtown Manhattan Study. Regier started his training with the people who ran that project, which began in 1952, took 10 years to complete, and was conducted exactly where you think it was. Researchers canvassed 1,911 Manhattanites and concluded, according to Regier, that 85 percent of the population had a mental illness.
The Midtown Manhattan Study is a talking point for most defenders of the DSM. They cite that 85 percent number as evidence that even if the DSM is an imperfect document, and even if it catches an improbable number of people in its diagnostic net, at least it's better than what we had in the bad old days. We in the mental health business call this a "downward comparison," and we sometimes recommend one to our patients to help them put their problems in perspective. "Yes, it's true your wife left you for your next-door neighbor," you might say, "but at least your kids won't have to commute as far as most children of divorce."
It's usually a pretty lame intervention, as that example indicates, and to the extent that it's successful, it's often because you are bolstering the patient's self-esteem by pointing out that he's better off than the next guy. Schadenfreudian therapy, you might call it.
You can't exactly blame psychiatrists for grasping at this straw. But some straws are flimsier than others. Take that 85 percent number. The Midtown Manhattan Study team, which was headed by sociologist Leo Srole, never said 85 percent of its subjects were mentally ill. In fact, Srole and his colleagues had not set out to diagnose New Yorkers at all. Indeed, they wrote, they didn't want to use the DSM or any other diagnostic system because they were "designed for classifying full-blown pathology" and didn't do such a good job of defining it in the first place.
So rather than ask about symptoms of mental illnesses, they asked about childhood fears of thunder and strangers, and current attitudes toward drinking and gambling. They solicited subjects' worries about the atom bomb and old age. They asked whether they thought people talked behind their backs or if they wondered whether "anything is worthwhile" or if they believed that "most people think a lot about sex." They paid attention to whether the interviewees were sloppy or neat, nervous or relaxed, if they were facetious, dull, or rambling. They attempted, in other words, to capture the everyday experience of the average citizen and to determine how much psychological suffering it entailed.
To make their assay, Srole's team devised a six-point classification of symptom formation. People at the "healthy extreme" got a zero. At the next two stages were people who had "emotional disturbance without apparent constriction or disability," and the last three ratings "span[ned] the morbidity range of the mental health spectrum." These were the people whose symptoms had "crippling effects on the performance of…daily life roles"—as close to a working definition of mental illness as Srole's team ventured.
Like most mainstream psychiatrists and sociologists at the time, Srole and his colleagues subscribed to the Freudian notion that between the dynamics of our psyches and the demands of civilization, virtually everyone was bound to be troubled, and that, as they wrote, "mental illness and mental health [differed] in degree rather than in kind." So when it turned out that something like 85 percent of the subjects (the actual number is 81.5 percent) scored more than a zero, the researchers could have been no more surprised than the audience at a Woody Allen movie would be to discover that most Manhattanites are at least a little neurotic.
But Srole and his team were not making the ridiculous claim that 85 percent of us are mentally ill. Rather, they were reporting the unremarkable finding that if you sit down with people and ask them about their emotional lives, you will find that most of them will confess to some difficulty. The abnormal people in this study, as in most studies, were those at the ends of the spectrum—the 15 percent who claimed to be free of psychic turmoil and the 23.4 percent who scored in the morbid range.
This latter finding, Srole noted, was a bit of a surprise. It was more than double the rate of mental illness found in an earlier study of Baltimoreans. But it was fully one-third less than the number of people who would, 25 years later, turn up as mentally ill in Regier's ECA study, and surely nowhere near 85 percent.
Regier knows that Srole's study did not really conclude that 85 percent of its subjects were mentally ill. Indeed, he cited the 23 percent figure accurately in a 1978 report to a presidential commission on mental health. He also knows about the Baltimore study, and about a National Institutes of Health study conducted by his mentor, the psychiatrist Michael Shepherd, which found a prevalence rate of around 15 percent. That, in fact, was the figure he used in his report to the president: On any given day, he wrote, about 15 percent of Americans were mentally ill.
Regier thinks Shepherd's study is "a classic" and the 15 percent figure "wonderful." It's easy to see why he didn't lead his presentation to the Post editors with his conviction, based on a quarter century of studying the subject, that prevalence rates had been inflated by the DSM—published, remember, by his employer. Or why he wouldn't volunteer that the categorical approach to diagnosis did not reflect the reality of mental illness. He and Mirin had a guild to represent, and telling the editors that parity laws would force insurers to pay for the treatment of people who didn't necessarily have illnesses would probably not have been the most convincing way to sell national mental health policy.
This article is adapted by arrangement with Blue Rider Press, a member of Penguin Group (USA) Inc., from The Book of Woe by Gary Greenberg, Copyright © 2013 by Gary Greenberg.