The Book of Woe: The DSM and the Unmaking of Psychiatry

by Gary Greenberg

  But nearly forty years after Robins and Guze proposed these validators—four decades in which criteria for inclusion and exclusion were written and rewritten, tests conducted, families studied, and patients followed—Darrel Regier told the Psychiatric News that “validity tests . . . have not lived up8 to the expectations of Robins and Guze.” Even after Virginia Commonwealth University behavioral geneticist Kenneth Kendler added another validator9—differential response to treatment, on the questionable grounds that a response to an antidepressant, for example, confirmed that the patient had MDD in the first place—the patterns that had emerged were weak and confusing. Indeed, by 2010 Kendler himself was still complaining that “the [diagnostic] categories10 in use have been heavily influenced by expert opinion, which . . . has been heavily influenced by a priori factors.”

  Psychiatrists had evidently been fooled again. The new a priori—Spitzer’s principle that mental disorders could be classified by their criteria—has led them “to consider our major diagnostic categories11 to be obvious and even ‘natural,’” Kendler wrote, when, in fact, they were “fuzzy constructs that shift when viewed in different ways.” These benighted psychiatrists had fallen for their own ruse. They had forgotten that the DSM was fashioned by experts, which meant that the diagnostic categories tended to reflect the a priori principles of those experts—which, of course, the rank and file largely shared. “We cannot develop a progressive scientifically based nosology shaped by a single expert-driven conception of psychiatric illness, no matter how wise its advocate,” Kendler wrote. Evidently, it is one thing for the public to believe the experts, and quite another, at least in the view of one key opinion leader, for the experts to believe themselves.

  But Kendler wasn’t ready to get rid of the experts and replace them with, say, a WikiDSM. Nor was he joining with the “critics of psychiatric diagnoses12,” who, so he wrote, insisted that “there is no truth out there” and who would simply give up on the DSM. Indeed, he became a member of a DSM-5 work group and eventually the head of a committee reviewing its scientific soundness. But like Frances, he thought revisers took on a “heavy and conservative burden.” They had to avoid the kind of bruising battles in which “different constituencies in psychiatry . . . vie with each other for influence and control,” and, upon seizing power, “reshape the nosologic system in their own a priori image.” The result, he feared, would be “wide fluctuations between different systems with divergent theoretical perspectives and no net progress.”

  Kendler didn’t point out, because he didn’t need to since his intended audience knew it in their bones, that DSM-III was about as wide as a fluctuation can get. Bob Spitzer seized power, wrenched psychiatry away from its Freudian principles, and reshaped it in his own image. He might have gotten away with imposing this profession-saving paradigm shift by sheer force of will, but it was a desperate measure taken in desperate times, and not one you would want to repeat—especially if you want your revisions to lead to a closer approximation of the reality of mental illness and your constituency to have more faith in you than they do in other institutions known more for their thrashings over divergent theoretical perspectives than their net progress, like the United States Congress.

  That goal is possible, Kendler wrote, but only if changes are made slowly and carefully. In fact, if you are careful enough, a good outcome is nearly guaranteed, thanks to the process of epistemic iteration, a concept Kendler borrowed from mathematics, where it is defined as:

  A historic and scientific process13 in which successive stages of knowledge . . . build in a sequential manner upon each other. . . . When correctly applied, the process of epistemic iteration should lead through successive stages of scientific research toward a better and better approximation of reality in a “spiral of improvement.”

  It’s easy to see the appeal of this promise. Kendler himself is the researcher who reported that when Walter Cassidy, the psychiatrist who first proposed diagnostic criteria for depression, was asked why he set the threshold at six out of ten symptoms, he responded, “It sounded about right14.” If you know that your origins are murky and your tools blunt, and yet you want to claim that you are nonetheless heading toward clarity, then it behooves you to put your money on a “wonderful property of iteration15” and its “capacity to get to the real solution regardless of the starting point.”

  And if you know that your nosology has gyrated wildly through the years, that it has been buffeted by history, its a priori principles brought to light and debunked and hidden away again, if you know that the doubt thus kindled will become your enemies’ weapon, if you know that you and your allies must be able to “assure ourselves16 that each revision of our manuals contains improvements on its predecessor,” then what better talisman to carry into battle than epistemic iteration, with its nearly magical power to ensure that even as you are making your mistakes, truth is all along accumulating, that those fluctuations are really only what Kendler calls “wobbly iterations17,” that every day in every way your map of our suffering is getting better and better, and that sooner or later, the experts, well versed in expertise, will produce a DSM that, as Kendler puts it, “asymptotes to a stable and accurate18 parameter estimate”?

  But psychiatry is not mathematics. The way we suffer, unlike the way numbers behave, changes with time and circumstance, and experts’ opinions of what ails us change the way we think of ourselves and our travails. Kendler insists that mental illnesses must exist in nature. Rewind the tape of history to the dawn of civilization, he says—about ten thousand years ago, when our biological apparatus would have taken shape but history had yet to make us into who we are—and start again. Record the result, and do this a thousand times. While it is likely that in each iteration you will see psychiatrists arriving at different criteria by which they know our mental illnesses, you will also see the illnesses themselves, just as you are sure to see diabetes and strokes and broken bones.

  Of course, just because Kendler says that’s what will happen, it’s not necessarily what would happen. Indeed, the thought experiment falters when you consider that Kendler borrowed the scenario from paleontologist Stephen Jay Gould19. In the original version, Gould rewound the tape to the beginning of earth’s history to point out that given all the accidents, the asteroid crashes and ice ages and tectonic shifts, it was very unlikely that human life as we know it would emerge in any of those do-overs. His rhetorical point was that it was folly to assume that the long arc of evolution bends toward anything in particular, that only Whiggishness or presentism or some kind of Voltairian optimism—not to mention a huge dose of species-level narcissism—would allow a person to claim that we are the inevitable culmination of creation, let alone that our endeavors lead inherently toward progress.

  “We follow the tape forward20 until modern science and medicine develop,” Kendler writes, but this is the whole lesson of Gould’s experiment: We cannot know if anything like modern science and medicine will develop. We cannot simply assume that people will come to identify their subjective troubles as mental, much less place them in doctors’ hands. Gould figured that each new spool would be entirely different, not simply an instant replay.

  If Kendler has assumed his conclusions here, it is because he has to assume that mental illness exists in the same way as diabetes and strokes, and that the only alternative is to believe “there is no truth out there.” He has to believe these things because he is a psychiatrist, and only a notion of historical inevitability can justify the enormous hubris, the inescapable a priori principle of psychiatry: that our psychological suffering is medical—which, as our reel has unspooled, means located in bodily processes gone awry. Go to the tape, Kendler seems to be saying, and you will find that doctors’ failures so far, including the faulty DSM that urgently needed replacing, are just detours on the road to Parnassus and not a disastrous wrong turn.

  Psychiatry is not inevitable. It’s not baseball, either. Those wobbly itera
tions aren’t just bad calls that frustrate a batter and raise a crowd’s ineffectual, if vocal, ire. They are diagnoses that change people’s lives, that render homosexuals unfit for employment, that subject children to untested and powerful drugs, that encourage patients to think of their troubles as chemical imbalances rather than meaningful signs of something gone wrong in their lives. If you go to see a psychiatrist, you probably don’t enter the office as if you were going into Fenway Park. You don’t think you are about to take part in a game whose rules are arbitrary if venerable and negotiable. You don’t expect the number of symptoms that add up to depression to change like the strike zone has or diagnoses to enter the DSM like the designated hitter was added to the rulebook. If you’re anything like me, as much as you might like baseball, you expect more from your doctors than that.

  You might also think there is a difference between the current DSM and the current Official Rules of Major League Baseball—and not just that one costs $95.89 on Amazon, while the other can be downloaded for free and, in its 2011 edition anyway, has an excellent picture of Curtis Granderson crossing home plate on its cover. You might well believe that one compiles the laws of nature and the other the rules of the game, and if you do, it is in part because we all want to believe that someone somewhere can understand and help us when we suffer. But it is also because psychiatrists—with their scientific-looking DSM, with their assurances about epistemic iteration, with their talk of chemical imbalances and their medications to treat them—have spent the better part of the past four decades telling you, and acting like, it is so.

  • • •

  Darrel Regier’s August 2009 interview with the Psychiatric News was part of a concerted effort to counter Frances’s warnings that the DSM-5 was headed for disaster. “The ‘disastrous result21’ in most clinicians’ and researchers’ minds would be for DSM to continue on the same path it has been on for 30 years,” he told the News. That’s why his task force was proposing “significant revisions,” and why the new manual, he wrote in an American Journal of Psychiatry article, would “attempt to address22 the consequences of continuing to use the original . . . hierarchical structure of ‘pure’ diagnostic categories.” Frances and Spitzer’s DSM was simply too bad to be left alone.

  Regier may have been making a wholesale critique, but he was quick to say that “a wholesale revision was not in the cards.” What Regier had in mind was nothing as radical as casting the categories to the wind. Instead, he said, diagnostic labels and criteria would be joined by dimensional measures. The problem with the categorical approach was that it forced clinicians into binary decisions. Did the patient have OCD or MDD, ADHD or BDNOS? Too often the result was more than one diagnosis—the comorbidity problem—or a diagnosis that didn’t really capture the clinical picture. And then there were the “patients with clinically significant distress and impairment,” whose symptoms spanned many diagnoses without reaching a diagnostic threshold in any one. People could be a little depressed, a little anxious, a little obsessive-compulsive, but still in a lot of distress. Wouldn’t it make sense to be able to identify, measure, and study these cross-cutting symptoms as dimensions of pathology without necessarily attributing them to a particular category?

  “The single most important precondition for moving forward to improve the clinical and scientific utility of DSM-V will be the incorporation of simple dimensional measures,” Regier wrote. Their “prominent use” would be “one, if not the major, difference between DSM-IV and DSM-V.” The revision would focus on measuring the severity of symptoms, both within and across diagnoses. This alone would not entirely spring psychiatry from its epistemic prison, but, especially once researchers could “establish better syndrome boundaries23” and “identify continuous measures of the constituent symptoms,” he told me in an e-mail, they would be able to determine the “statistically valid cutpoints between normal and pathological.” Getting to this goal, of course, would require categories based on “a full understanding of the underlying pathophysiology of mental disorders.” But even if this was a long way off, it was important to take the first step now. “Delaying the introduction of dimensional measures in this version of DSM will simply retard both clinical and research progress,” he wrote.

  But it wasn’t as if DSM-IV lacked dimensional measures. Some diagnoses, such as depression24, had severity specifiers, in which clinicians rated a patient’s disorder from zero to five and reported this as the last digit of the diagnosis code. And the DSM-IV also provided a Global Assessment of Functioning25 scale, which clinicians were supposed to use to indicate a patient’s functioning on a scale of zero to one hundred. But, as we clinicians soon found out, when we rated a patient’s functioning too high or his severity too low, insurance companies used these numbers as their own kind of statistical cutpoint—to cut off benefits. So we quickly adapted—inflating the ratings or ignoring them altogether, not telling unless we were asked.

  Regier needed a more comprehensive and scientific approach to dimensional assessment than what the DSM-IV offered, but this presented a problem. Very few severity tests for DSM-IV diagnoses had been developed and validated. Nor did he have a huge literature to draw on for measuring cross-cutting symptoms. The National Institutes of Health had created PROMIS26 (Patient Reported Outcomes Measurement Information System) to gather information on how patients thought they were faring, but while this offered a way to determine, say, how anxious a person was or how well he was sleeping, Regier could not say how this kind of data would be integrated into the diagnostic system. To get dimensional measures in place for the DSM-5 required a “bottoms-up27 [sic] approach for instrument development,” he wrote, but the group that had been assigned to dive into that task started its work only in January 2009. Field trials, in which those tests would presumably be studied, were slated to begin in summer 2009, so they had had just a few months to put together their tests. To critics, this seemed like an indication of disarray. “If they really want to do dimensional assessment28,” Michael First told me, “they should wait the five or ten years it would take for the scales to be ready.”

  But Regier did not think it was necessary to have all this nailed down before the book was published. “We don’t expect the DSM-529 to be perfect or etched in stone for the ages,” he told me in an e-mail. This expectation, after all, was the central trouble with the previous DSMs; designed to look scientific, they had proven too easy to reify. And even if the DSM-5 was not going to be anchored in the bedrock of neurobiology, even if it could not fulfill the promise of paradigm change with which it had been born, still it could achieve one thing: the ratcheting back of expectations for the revision, and for psychiatric nosology itself. Diagnostic criteria “are intended30 to be scientific hypotheses, rather than inerrant Biblical scripture,” he wrote. And the DSM was not scripture. It was a “living document.”

  It was a clever rhetorical move. Regier had turned the reification of the DSM into just another of those “epistemic iterations” that Kendler wrote about, a wobbly step on the way to the truth. The categorical approach had served its purpose, and now it was time to back away from it, and from the misguided fundamentalists who took the diagnoses literally. It was time to pirouette on the back of those new dimensional measures into the “spiral of improvement.”

  Chapter 9

  Throughout the summer and fall of 2009, Allen Frances kept up his attack. He became a regular contributor to the Psychiatric Times, blogging, sometimes in every issue, about what he thought the APA was doing wrong. By the time he posted a blog titled “Advice to DSM V1,” the APA was probably not in the mood to take it.

  “There is no magic moment when it becomes clear that the world needs a new edition of the DSM,” he reminded his successors. A revision of the ICD scheduled for fall 2012 had reportedly been delayed—and, as he also pointed out, because the codes used by the DSM-5 came from the ICD, a new DSM that preceded a new ICD would soon have outdated codes. Wouldn’t it be
better to coordinate the release of the DSM and the ICD, and in the bargain give the revision the time it deserved?

  Getting out from under deadline pressure was the key to saving the DSM-5, Frances thought, and this now became his mission. Unintended consequences were the unknown unknowns of the APA’s revision campaign, and haste only increased their likelihood and severity. Even a good diagnostic change could make unexpected trouble. For instance, he wrote, the DSM-IV had fashioned new criteria for ADHD. They were clearer and easier to remember, and in the field trials the new criteria identified 15 percent more2 kids with ADHD than the old criteria had—an acceptable result, Frances thought at the time. But once the criteria went into wide use, the actual increase was 28 percent3—an outcome he attributed in part to the ease with which doctors, as well as parents and teachers, could apply the simplified criteria.

  Of greatest concern to Frances, however, was the task force’s failure to take “the most important step in the development of any DSM”: to release a complete first draft. Without one, experts outside the work groups could not weigh in on either the proposed revisions or the procedures by which they would be tested, which meant that the field trials might use a faulty methodology to investigate faulty diagnoses comprising poorly written criteria—a compounding of errors that would, he was sure, turn out to be the “fatal flaw” of DSM-5.