Genetics, Neuroscience, and Mental Illness
MOST AMERICAN PSYCHIATRY remains firmly committed to a biologically reductionist view of mental disorder. Yet the hunt for the physical roots of mental disturbance has not led to the decisive breakthroughs its enthusiasts were convinced it would. The result is a profession facing an existential crisis, one that in my view cannot be resolved by doubling down on its wager that mental disturbance is nothing but brain disease.
The bet on the body is not a novel strategy. As we have seen, late nineteenth-century alienists were convinced that mental illness represented a form of degeneration, of biological evolution run in reverse. Defective heredity was what accounted for the explosion of insanity that accompanied the birth of the asylum, and simultaneously explained (and explained away) the profession’s inability to produce much in the way of cures. In subsequent decades, the rediscovery of Gregor Mendel’s work transformed vague notions of inheritance into a new science of genetics. Particularly in Germany, in the Munich laboratory established by Emil Kraepelin and run by Ernst Rüdin, a series of family and twin studies purported to give scientific substance to the idea that mental illness was inherited. It was a seductive idea that attracted substantial financial support from the Rockefeller Foundation throughout the 1930s, notwithstanding the awkward fact that Rüdin was an ardent Nazi and a key architect of Hitler’s program to slaughter the mentally ill. After the war, except among a handful of psychiatrists, the association of genetic theories of mental illness with mass murder led to the marginalization of such work—a development that also reflected the growing dominance of psychoanalysis.1
Eric Kandel, a Nobel Prize–winning psychiatrist, recalls that in 1965, during his psychiatric and psychoanalytic training at the Massachusetts Medical Center, the major teaching hospital of the Harvard Medical School, he and several of his fellow residents “tried to recruit a psychiatrist in the Boston area to speak about the genetic basis of mental illness. We could find no one; not a single psychiatrist in all of Boston was concerned with or even had thought seriously about that issue.”2
Genetic research began to reemerge as part of the general revival of biological psychiatry in the 1970s and 1980s. Along with neuroscience, it appeared to be the most promising route to understanding major mental illness. The earlier twin and family studies, conducted by eugenicists who knew in advance what conclusions they wished to reach, had been riddled with methodological errors that cast serious doubt on their scientific credibility.3 Now new studies were conducted, supposedly avoiding these errors.4
The centuries-old folk belief that madness ran in families seemed to be confirmed by these new studies. People whose relatives had been diagnosed with serious mental illness were shown to have a much higher risk of having other relatives enter treatment for such disorders. Renewed efforts were made to examine monozygotic twins and compare their susceptibility to schizophrenia or manic-depressive illness and major depression to the incidence among dizygotic twins. Among genetically identical twins, concordance for schizophrenia (i.e., both twins being diagnosed with the disorder) was about 50 percent. Among dizygotic twins, the risk that both would develop schizophrenia was much lower, but, even here, these twins appeared to be about twice as likely to develop schizophrenia as ordinary siblings.5 Using mostly Scandinavian data, which provided access to the health records of entire populations, some of these studies extended the analysis to adopted children, showing that their risk of developing schizophrenia closely resembled that of their biological families, not their environmental families.
Skeptics might note that the well below 100 percent concordance among monozygotic twins demonstrated that biology was not destiny and that families seem to “inherit” many things that are clearly not prompted by their biology (such things were religious beliefs, television viewing habits, and political preferences, for example). This should remind us that asserting “heritability” is not the same as proving that disorders are biologically determined. Family studies failed to provide any insight into the molecular genetic mechanisms that might lie behind the purely correlational data they produced.6 Still, the conviction that genetics might provide a compelling window into the biology of mental disorders grew among many research psychiatrists in the 1990s. The elucidation of the genetic origins of Huntington’s chorea in 1993 encouraged the belief that the genetics of schizophrenia, depression, and other major varieties of psychiatric disorder would soon become clear.7
The National Institute of Mental Health (NIMH) and other funding agencies began to funnel money into this line of research. The development of the polymerase chain reaction (PCR) technique in 1985, and its commercial licensing in 1989, created a promising new tool for genetic analysis, by permitting the easy creation of millions or billions of copies of short segments of a target piece of DNA. The race to decode the complete human genome was entering its final stages—its successful completion was announced in April 2003. The genetic basis of several severe forms of mental illness was something that previously could only be inferred from family, twin, and adoption studies. Now, optimists were captivated by the idea that the newly deciphered human genome might at last reveal the genetic roots of their patients’ troubles. This paralleled the conviction of some early twentieth-century alienists that the discovery of the syphilitic origins of General Paralysis of the Insane would soon lead scientists to uncover the infectious origins of other forms of insanity.
The neuroscientists who made up a large fraction of academic psychiatry were meanwhile pursuing their own pathway to heightened understanding of the biological basis of mental disorder. Alongside the massive expansion of work on neurotransmitters and the operations of the brain, research projects facilitated by connections to psychopharmacology and advances in imaging technology were spurring attempts to map brain activity and brain structures in novel ways. If schizophrenia, mania, and major depression were disorders of the brain—now an article of faith in academic psychiatry—here was another set of medical technologies that might soon uncover the etiology of the major forms of mental disturbance, dispelling the fog of confusion that had long enveloped the field. There were reports of shrinkage and structural abnormalities in the brains of schizophrenics, and when the technique of functional magnetic resonance imaging (fMRI) arrived in the early 1990s, it was proclaimed a way to decipher brain activity in real time, giving hope that it would display the differences between mad and sane brains in living technicolor. (fMRI imaging measures brain activity by detecting changes in cerebral blood flow, using these data as an indirect measure of the activation of neurons.)
CONFIDENT THAT THESE ADVANCES in neuroscience and genetics were on the brink of revealing the biological bases of a host of mental disorders, leading figures in psychiatry began to press for a new revision of the Diagnostic and Statistical Manual of Mental Disorders (DSM). This time, they claimed, the purely symptomatic approach to diagnosis that had formed the basis for the three editions since 1980 would be supplanted by a completely different approach. The process was launched at a conference jointly sponsored by the American Psychiatric Association (APA) and the NIMH in 1999. The director of the NIMH, Steven Hyman, was determined to “introduce neuroscience into diagnosis,” and this was an approach the researchers launching the process endorsed.8 This suggested a sharp break with “the heuristic and anachronistic” approaches that had found their way into previous editions of the DSM, for “as new findings from neuroscience, imaging, genetics and studies of clinical course and treatment response emerge, the definitions and boundaries of disorders will change.”9 The categories that formed the basis for previous versions of the manual had been based on symptoms, with all the emphasis placed on ensuring that psychiatrists reached the same diagnosis when confronted by the same patient. Now, validity would take center stage. Rather than using categories that might or might not identify diseases that existed in nature, the emerging understanding of the biological basis of mental illness would drive the process and transform the ways mental illness was categorized. Henceforth, the would-be architects of DSM 5 asserted, the way forward was “to recognize the most prominent syndromes that are actually present in nature.”10 Their version of the DSM, they promised, would seek to provide “scientific hypotheses rather than inerrant Biblical scripture”—a not-so-subtle jab at the pretensions of their predecessors.11
Their confidence that they could reconstruct the whole basis of psychiatric diagnosis, and do so by 2012, proved misplaced. The conviction that the combination of PCR testing and the decoding of the human genome would rapidly produce revelations about the genetic basis of the whole spectrum of psychiatric disorders—uncovering the reasons mental illnesses seemed to run in families and revealing just how schizophrenia, bipolar disorder, and major depression reflected someone’s genetic code—simply collapsed.12 Science couldn’t deliver the evidence it had promised. The massive resources that had been poured into uncovering the genetics of mental disorder produced more frustration than enlightenment. Or, rather, the enlightenment genetic researchers produced threatened to undermine the standing and legitimacy of the “diseases” that psychiatrists had believed in for more than a century—concepts that had woven themselves deeply into the ways the lay public had been taught to view mental illness. To acknowledge that the distinctions between schizophrenia and bipolar disorder were spurious, that those constructions might have to be abandoned—but with what to put in their places?—threatened the very foundations of the psychiatric profession’s claims to expertise. If such fundamental building blocks of the psychiatric universe crumbled on close inspection, what was left?
IN THE EARLY DAYS OF THE RENEWED EMPHASIS on genetics, there were periodic claims to have discovered the Holy Grail, a genetic basis for schizophrenia. But such assertions have repeatedly failed the test of replication.13 The University of North Carolina geneticist Patrick Sullivan used genetically realistic simulated data to show how easily positive associations could be derived from the kinds of statistical techniques these scholars relied on “merely by the play of chance.” As often as 96.8 percent of the time, when what seemed to be compelling or intriguing linkages surfaced, they turned out to be false positives.14 Genetic researchers had identified a series of candidate genes that seemed likely contenders to explain the genetic roots of schizophrenia and major depression, and they produced papers that purported to document such linkages. Those findings have failed to survive, products of precisely the statistical artifacts to which Sullivan had pointed.15
As research proceeded, it became impossible to keep believing that there was a single gene for schizophrenia or for any other form of major psychiatric disorder.16 The field as a whole, as Kenneth Kendler put it, had to absorb some “painful lessons” and acknowledge that “despite our wishing it were so, individual gene variants of large effect appear to have a small to non-existent role in the etiology of major psychiatric disorders.”17
The search for a genetic basis for mental illness came to be recognized as extremely complex and elusive. Researchers more and more relied on an approach that scoured the entire sequence of the human genome, looking for correlations between particular stretches of genetic code and any susceptibility to schizophrenia, bipolar disorder, or major depressive disorder. So-called genome-wide association studies (GWAS) look as broadly as possible for anomalies associated with these disorders. Such studies make no prior assumptions about the etiology of the disorder under study and treat every genomic variant equally, examining hundreds of thousands of sites looking for possible linkages. It’s a brute-force technique applied to a whole spectrum of psychiatric disorders, and by now researchers have accumulated data covering tens of thousands of patients. Whereas family studies inevitably conflated genetic and environmental influences, and scientists had difficulty separating the two, GWAS labored under no such handicap.
As studies of this sort accumulated, hopes that genetic research would uncover clear pictures of an underlying biology of mental disorders quickly faded. It was not just that the maximal claim—that schizophrenia, for example, was a Mendelian disorder—was quickly shown to be false, but even an alternative hypothesis, that “a substantial proportion of the [hypothesized] genetic signal could have been concentrated in a few large-effect genes,” was soon rejected.18 In the vast majority of cases, what materialized was a muddle. Crunching the data has shown that hundreds of genetic variants may (or may not) contribute to the diagnosis of a particular case. Each of these potential variations is individually of small effect and may be present without giving rise to the disease. Many carriers of these genetic variations fail to exhibit signs of mental disorder, suggesting that these are polygenic risk factors for mental illness, not differences that inevitably or even probably lead to schizophrenia or affective disorders. Besides, even when aggregated together, the genetic variants uncovered by GWAS account for a tiny percentage of the variance. When the International Schizophrenia Working Group published its findings in 2014, it had looked at 36,989 patients and 113,075 controls, and it asserted that it had found 108 loci associated with schizophrenia. Taken together, however, these 108 sites accounted for a total of 4 percent of the variance in the diagnosis of schizophrenia.19 Subsequent collaborative work has expanded the number of genetic loci to as many as 270, and by incorporating this expanded array, these loci account for around 7.7 percent of the observed variance.20
Each of these variations contributes to risk, but only in a very minor way, and in total, this huge array of factors accounts for a small fraction of the risk for schizophrenia. A person may harbor any number of these genetic variations without ever developing mental illness. Conversely, “despite its high heritability, the majority of individuals with schizophrenia do not have a first-degree relative with schizophrenia.”21 The same pattern holds with other mental disorders. For example, a collaborative study of major depressive disorder, “the largest yet conducted,” fared poorly, finding nothing that reached genomewide significance and acknowledging that the work was “unable to identify robust and replicable findings.”22
Genes are not fate, and the thousands of alleles that contribute a small additional risk of illness do not operate “in a simple deterministic manner.”23 Developmental and environmental factors play a crucial role in whether the “nudge” of these alleles manifests itself in mental disorder.
A recent systematic review of the GWAS data on schizophrenia examined the various genetic polymorphisms that had previously been identified as likely to be associated with the disease. With minor exceptions, the authors found that “schizophrenia candidate genes are no more associated with schizophrenia than random sets of control genes,” such as those for height or type 2 diabetes. Their conclusion was stark and unambiguous: “Our results suggest that the statistical rationale for further prioritization of these genes is weak.… [O]ur results do not support the original hypotheses involving the most-studied candidate genes, and thus provide no reason to believe that rare variants in these genes, or the trans-elements that regulate them, will be particularly relevant to schizophrenia.”24
The pattern was identical when the research on major depression was scrutinized: “As a set, depression candidate genes were no more associated with depression phenotypes than noncandidate genes.” That meant, Richard Border and his colleagues concluded, “that the large number of associations reported in the depression candidate gene literature are likely to be false positives.”25
Despite extensive funding and vast teams of scientists cooperating to generate reams of data, in the words of two leading psychiatrists, Rudolf Uher and Michael Rutter, “Molecular genetic studies of psychiatric disorders have done a lot to find very little. In fact, in the era of genome-wide association studies, psychiatric disorders have distinguished themselves from most types of physical illness by the absence of strong genetic associations.”26 Steven Hyman, who did so much as director of NIMH to fund these studies and is now the director of the Stanley Center for Psychiatric Research at the Broad Institute and a professor of stem cell and regenerative biology at Harvard, has spoken of how the early confidence that “we might identify causal mutations” has dissolved into uncertainty. Instead, he suggested that it was incumbent to acknowledge that “the genetic, epigenetic, and other environmental risks of psychopathology are etiologically complex and heterogeneous.” It was now clear that “the many risk genes that might contribute to schizophrenia are not transmitted together across generations.” These findings required stem-cell researchers to take account of “a plethora of other disconcerting observations.”27
NOT ONLY DID GENETICS fail to deliver the promised results but such findings as it did produce threatened to destabilize the psychiatric universe from another direction. For rather than revealing one set of vulnerabilities for schizophrenia and others for manic-depressive disorder or major depressive disorder, most such risks as it identified seemed common to a whole range of mental disorders.28 “The high degree of genetic correlation … among attention deficit hyperactivity disorder (ADHD), bipolar disorder, major depressive disorder (MDD), and schizophrenia,” the international Brainstorm Consortium reported, provides important “evidence that current clinical boundaries do not reflect distinct underlying pathogenic processes.… This suggests a deeply interconnected nature for psychiatric disorders.”29
Biology, it seems, predisposed some to a heightened liability to mental disorder, but most of that liability is not disorder-specific. Modern geneticists would recoil from the language the late nineteenth-century alienist Henry Maudsley used to summarize his beliefs about the inheritance of madness, but their own chastened assessment of the relations between genetics and mental disorder are, mutatis mutandis, not so very distant from his: “It is not that the child necessarily inherits the particular disease of the parent … but it does inherit … a constitution in which there is a certain inherent aptitude to some kind of morbid degeneration … an organic infirmity which shall be determined in its special morbid manifestations according to the external conditions of life.”30
Or perhaps, and this was the more disturbing implication, the distinctions psychiatrists derived from symptoms lacked validity and required rethinking.31 The logic, such as it was, of the various DSM editions was to divide mental illness into a series of discrete and specific “illnesses,” each of which was supposed to constitute a more or less homogeneous entity. That had been one of the major attractions of Robert Spitzer’s system for the pharmaceutical industry, because such constructed “facts” (periodically added to as necessary) provided the underlying conditions it needed to convince the FDA of the value of its products.32 Further entrenching the DSM hegemony, its distinctions led the FDA to embrace the notion that psychiatric disorders were strictly comparable to other illnesses. By requiring drug companies to test and advertise their products as treatments for specific DSM diagnoses, the agency inadvertently contributed to the reification of its categories.33
To the extent that one can derive useful conclusions from recent work on the genetics of mental illness, two seem to stand out: what we have learned strongly supports the adoption of a dimensional rather than categorical view of psychiatric disturbances; and its findings demonstrate that psychotic patients are not genetically different in kind from the rest of us. The latter point is a sharp corrective—indeed, a reversal—of the conclusion many drew from the research of early twentieth-century psychiatric genetics, that the mentally ill were biological degenerates. Those studies from a century ago heightened the stigma attached to mental illness. Perhaps the recognition that the sane are not fundamentally different from those suffering from serious forms of mental disturbance will have the opposite effect, though I can’t say I’m sanguine this will be the case.
Some optimists have suggested that one way to connect genetic findings to clinical applications would be to derive polygenic risk scores from the existing data and use these findings to guide interventions. Aggregating all the small genetic risks that recent research has uncovered can quite cheaply be used to provide a measure of heightened risk for future mental illness, just as it can be used to assess risks of developing (say) type 2 diabetes. The 10 percent of people with the highest such score for potentially developing schizophrenia, for example, appear to face three times the risk that confronts the general population. The authors suggest that because there is some data that suggest that recreational drug use exacerbates the risk of future psychosis, at-risk individuals could then be counseled to avoid such exposure.34
Unfortunately, however well intentioned such suggestions might be, major problems would follow any such move. Epidemiological statistics suggest that one out of every one hundred people will be diagnosed as schizophrenic over the life course. Granting for the sake of argument the claim that psychiatric genetics shows that among a select subpopulation the risk is three times as large, for every one hundred people who are informed of their heightened susceptibility, only three will ultimately be diagnosed as schizophrenic. It scarcely requires much imagination to understand the anxiety and other problems such information would create or to grasp the heightened possibilities of stigma and exclusion such people would then face.35
Perhaps further advances in genetics may provide guidance about treatment choices if such knowledge helps predict responsiveness to medications and susceptibility to their side effects, to guide the treatment of future patients. One can also put an optimistic gloss on the way genetic findings have forced a reassessment of existing diagnoses in psychiatry. The Brainstorm Consortium makes precisely these suggestions. The most recent genetic findings, it argues, “are promising steps toward reducing diagnostic heterogeneity and eventually improving the diagnostics and treatment of psychiatric disorders.”36 Once more, these are promissory notes, speculations that may or may not prove well founded in years to come. Their discounted current value is very small, especially given the dishonored promissory notes that litter psychiatry’s past.
NEUROSCIENCE HAS PROVED similarly unhelpful to date. Much excitement originally arose from a series of findings comparing the brains of schizophrenic patients and controls. These demonstrated greater volumes of brain shrinkage in the patients and were initially hailed as “proof” that schizophrenia was a brain disorder. Setting aside concerns about the comparability of the healthy controls, whose brains may have been affected by the fact that they are “typically living more active lives, interact with more people, are wealthier and have more control over their environment”—all factors likely to positively affect brain structure and functioning—a yet more powerful confounding factor had been overlooked in the excitement: the psychiatric patients had all been taking large quantities of antipsychotic drugs over the years before brain volume was being measured.37 And as a whole series of subsequent studies, both animal and human, would show, there are clear links between the ingestion of these drugs and transformations in the structures of the brain.38 Rather than being the product of their disease, changes in these patients’ brains were probably an artifact of their treatment. Robin Murray, a leading British researcher, puts it more strongly: “It is clear that high-dose antipsychotics contribute, not to the subtle brain changes present at the onset of schizophrenia, but to the subsequent progressive changes thereafter.” Unfortunately, Murray adds, “some psychiatrists refuse to accept the evidence and cling to the nihilistic view that there is an intrinsically progressive schizophrenic process, a view greatly to the detriment of their patients.”39
Among the neuroscience community, much excitement attended the arrival of functional magnetic resonance imaging (MRI) technology. Such scans, which, as I’ve noted, measure blood flow in the brain as an indirect proxy of its neuronal activity (albeit with a lag), have proved of some value in basic research, but their clinical utility in psychiatry to date has been essentially nonexistent.40 It will surprise no one—not even the immaterialist philosopher George Berkeley—to learn that when I move, speak, think, or experience an emotion, these are correlated with changes in my brain. But such correlations tell us nothing about the causal processes involved. Nor do we possess any means to translate “heightened activity” in different regions of the brain to the contents of people’s thoughts. Weak inferences based on statistical averages of gross changes in brain function in large groups of subjects provide little guidance when it comes to the individual subject, and they are a poor substitute for tying together behavioral symptoms and brain defects in a robust causal fashion. Such esoterica served only to widen the chasm separating academic researchers from the clinicians dealing directly with the problems posed by mental illness.
In a review of the possible clinical utility of brain imaging for major psychiatric disorders, Jonathan Savitz and his colleagues reached an unsurprising conclusion: “There are currently no brain imaging biomarkers that are clinically useful for establishing diagnosis or predicting treatment outcomes in mood disorders.”41 This, along with the fading hopes of breakthroughs from genetics and imaging, forced the DSM 5 Task Force to abandon its grand plans and to settle for one more round of revisions to the elephantine device that was the DSM, relying once more on symptoms to guide diagnosis. Hyman, who had done much to underwrite the shift to biology, neuroscience, and genetics, had by 2007 concluded that “it is probably premature to bring neurobiology into the formal classification of mental disorders.”42
In 1886, the American alienist Pliny Earle lamented that “in the present state of our knowledge, no classification of insanity can be erected on a pathological basis, for the simple reason that, with but slight exceptions, the pathology of the disease is unknown.… Hence … we are forced to fall back upon the symptomatology of the disease.”43 Nearly a century and a half later, nothing, it seems, had substantially changed.
THE PROBLEMS ASSOCIATED WITH THE CREATION of the new manual were far from over. On the contrary, they multiplied in the years that followed, spilling over into the public arena in ways that threatened to discredit the whole enterprise. When leaks from Task Force deliberations suggested that its members were contemplating reining in the boundaries of autism—a diagnosis whose prevalence had exploded in the preceding twenty years—a buzz saw of protest ensued.44 Parents, whose access to social and educational services for their children were dependent on the diagnosis, reacted with fury. The Task Force beat a hasty retreat. It nominally abolished the previous diagnosis of Asperger’s syndrome, but, in reality, it simply relocated it into a new category called autism spectrum disorder. That made sense given that the Task Force concluded that the existing dividing lines were hard to justify and difficult to draw, but patients’ families were clearly concerned that insurance companies and government entities might stop paying for services for those at the milder end of the spectrum. To head off further protests, the Task Force issued a “clarification”: “Individuals with a well-established DSM-IV diagnosis of autistic disorder, Asperger’s disorder, or pervasive developmental disorder not otherwise specified should be given the diagnosis of autism spectrum disorder.”45
Far more serious was a campaign of denigration directed at the whole DSM enterprise that came not from outsiders, but from the very heart of the profession. It was led by the two previous directors of the enterprise, Spitzer and Allen Frances. Their sustained assault delayed the appearance of the new manual by a year, to May 2013, and inflicted major damage on its reputation. Their objections had two main foci: the secrecy with which they claimed that the revisions were being undertaken by people with deep ties to the pharmaceutical industry; and the threat that the new manual would drastically expand the range of emotions and behaviors that were seen as forms of psychiatric pathology, transforming the normal vicissitudes of everyday life into “illnesses” requiring psychiatric intervention and treatment.
Critics pointed out that the ties between those constructing the diagnostic manual and the pharmaceutical industry had become even tighter. Fifty-seven percent of the DSM IV Task Force had financial ties to industry; for DSM 5 that percentage rose to 80 percent. In the subsection dealing with mood disorders, the largest market for psychiatric services, that number had reached 100 percent. There was ample reason to worry about the conscious and unconscious biases this situation might introduce. To cite one example, meta-analyses of the reported outcomes of drug trials show a powerful relationship between the identity of the trial’s sponsor and the published results of what purport to be double-blind trials. Trials sponsored by drug companies consistently produced more favorable outcomes than those with independent sponsorship, a pattern also seen when one drug company pitted its product against those of its competitors.46
Spitzer was particularly exercised by the secrecy surrounding the development of the new manual. He had discovered that the members of the various working groups responsible for the revisions had been forced to sign nondisclosure agreements, ensuring that the sorts of political debates that lay behind the final text would remain hidden from view and from outside scrutiny.47 In my view, these complaints were something of a red herring. The deliberations of Spitzer’s Task Forces had likewise taken place behind closed doors, where horse-trading and arm-twisting were the norm.48 As a good politician, Spitzer did periodically discuss the progress of the manual with different audiences, but the same can be said of those working on DSM 5, where the availability of the internet also allowed work groups to post draft documents and to solicit outside comments, which were often extensive. But while the objection about secrecy may have been misplaced or overblown, there can be no doubting its effectiveness as a public relations tactic.49
Frances’s concern, shared by others, was the vast expansion of the psychiatric imperium that would result if some leaked proposals about changes to the manual were eventually implemented.50 He warned:
DSM-5 will turn temper tantrums into a mental disorder.… Normal grief will become Major Depressive Disorder.… The everyday forgetting characteristic of old age will now be misdiagnosed … creating a huge false positive population of people.… Excessive eating 12 times in 3 months is no longer just a manifestation of gluttony and the easy availability of really great tasting food. DSM-5 has instead turned it into a psychiatric illness.… DSM-5 has created a slippery slope by introducing the concept of Behavioral Addictions that eventually can spread to make a mental disorder of everything we like to do a lot.… Many millions of people with normal grief, gluttony, distractibility, worries, reactions to stress, the temper tantrums of childhood, the forgetting of old age, and “behavioral addictions” will soon be mislabeled as psychiatrically sick.51
With Spitzer increasingly sidelined by his advancing Parkinson’s disease, it was Frances who took the lead in amplifying their critique. In online blogs, repeated critiques in the profession’s monthly newsletter, the Psychiatric Times, and op-eds and media interviews, he relentlessly promulgated his criticisms.52
Joining forces with the president of the APA, Alan Schatzberg of Stanford, the leaders of the DSM 5 Task Force struck back with an ad hominem attack. Spitzer’s and Frances’s criticisms, they insisted, should simply be discounted. DSM 5 was just as scientific as its predecessors. Objections stemmed from disreputable motives that neither critic had disclosed. “Both Dr. Frances and Dr. Spitzer have more than a personal ‘pride of authorship’ interest in preserving the DSM-IV and its related case book and study products. Both continue to receive royalties on DSM-IV associated products.”53
In a further attempt to stifle criticism, the DSM 5 leadership took steps to ensure that “all the people at the top of the previous DSMs were completely excluded from their deliberations.”54 It was a strategy that proved wholly counterproductive. The media seized on such a public demonstration of splits within the profession—“a dispute that puts the Hatfield-McCoy feud to shame” as a blog in the Psychiatric Times put it—and so, far from squashing the controversy, the attack on Spitzer and Frances only amplified it.55 Critics took particular delight in pointing out that Schatzberg had accused Frances of being corruptly influenced by $10,000 a year in royalties, while he himself had just been forced to relinquish his position as principal investigator on an NIMH grant over a conflict of interest running into the millions of dollars.56
Behind the scenes, the process of assembling the new DSM was dissolving into chaos. Spitzer and Frances had exercised an iron control over the manuals they oversaw, ensuring at least a continuity of approach. Their successors, David Kupfer and Darrel Regier, had adopted a hands-off approach to the various work groups, which responded to their uncertain mandate in often-conflicting fashion. As one participant complained, “I get aggravated with Kupfer and Regier sometimes, where I want to say, ‘For God sakes, you have to tell us how many dimensions we can have.’ I mean these are things where you really need someone to make the decision about what the parameters are so that you can work. These guys are just way too open and flexible for us.”57 Some opted to loosen the criteria for particular diagnoses (most notably the group working on major depressive disorders), while others took the opposite tack.
As Allan Horwitz has recorded in DSM: A History of Psychiatry’s Bible, an invaluable guide to the construction, contents, and problems surrounding the various editions of the DSM, the result was alarm at the highest levels of the APA. Concerned with the lack of progress on the manual, and the disorganization and the controversy surrounding its work, the board of trustees of the APA appointed an oversight committee to oversee the work groups, and then, a year later, “a Scientific Review Committee that was independent of the DSM revision structure to review all the proposed changes to the manual and make recommendations directly to the APA President and Board of Trustees.”58
While Kupfer and Regier had by now abandoned the idea that they could completely transform the foundations of the DSM, they still clung to the hope that they could substitute a dimensional for a categorical approach to the definition of the various mental illnesses. That is, they wanted to incorporate the idea that these disorders, rather than being sharply distinguished from mental health, fell along a range, varying with respect to intensity, severity, and duration. Here, too, with the partial exception of autism spectrum disorder, where there was a broad agreement that sufferers were arrayed on a continuum, their efforts to introduce major changes to the DSM failed. Many individual working groups simply ignored their directives. Then, at the annual meeting of the APA in 2012, the clinicians who formed the vast majority of the profession openly rebelled. For them, the dimensional approach threatened to be unwieldy or even impossible to administer in practice. There was alarm among those who feared that insurance companies would deny reimbursement for cases that fell on the milder end of the spectrum. They recognized, as the researchers who formed virtually the entire membership of the Task Force clearly had not, that “diagnostic categories make mental disorders seem more real to the public, physicians in other medical specialties, insurance companies, and federal regulators.”59 Accordingly, when the assembly of the APA convened in May 2012, it unanimously voted to consign the talk of dimensions to an appendix of the manual, safely insulated from the sections clinicians used in their daily practice.60
DSM 5 (as it now called itself—the Roman numerals replaced by Arabic numbers to permit easy interim additions and revisions, like the modifications of software) was scheduled to be unveiled at the APA annual meeting in May 2013, a year late. It not only fell far short of the ambitions its progenitors had originally set for it, but also endured years of controversy and even ridicule.
The promise to create professional consensus on diagnosis came under threat. Spitzer had used a statistical measure of agreement among clinicians (kappa) that takes into account the possibility that concordance could occur simply by chance.61 He used kappa to measure interrater agreement in field trials of his new manual to document the heightened reliability the DSM III produced. When the DSM 5 Task Force conducted its own field trials, it used the same statistic. When they reported their findings, Frances immediately cited them as further evidence of the defects of their work. Comparing the two sets of data, he pointed out that kappa fell from 0.81 in the DSM III trials for schizophrenia to only 0.46 in the new trials, and for major affective disorders the results were worse: kappa here declined from 0.80 to 0.25.62
At first blush, that seems a devastating critique. Defenders of the DSM 5, however, rightly pointed out that there were crucial differences in the methods used in 1980 and 2012. Spitzer used two interviewers who had been highly trained in the use of the new manual, and he had them examine the same patient at the same time. It was an artificial test of reliability of the DSM categories that (one is tempted to say by design) increased the chances of interrater agreement. By contrast, the DSM 5 field trials employed briefly trained interviewers who performed separate evaluations at different points in time (between four hours and two days apart), an approach almost guaranteed to produce less agreement, albeit one that much more closely resembled the results to be expected when the manual was deployed in the real world.63
That difference in methodology makes direct comparisons between the results of the two misleading.64 Those involved in the DSM 5 field trials asserted that their aim was “to provide the greatest possible assurance that those with a particular disorder will have it correctly identified (sensitivity) and those without it will not have it mistakenly identified (specificity).”65 Yet with the exception of major neurocognitive disorder and post-traumatic stress disorder, few trials showed even a “fair” degree of agreement among clinicians. Even those in charge of the 2012 trials acknowledged that the degree of agreement they found for major depression was “questionable,” and for schizophrenia, the results were marginally better but scarcely cause for celebration. With the quest for validity abandoned, the new manual failed even to deliver intraprofessional agreement.66
That was not the worst of it. The doyens of NIMH, who had helped launch the revolution DSM 5 was supposed to bring, and who served, alongside the drug companies, as the joint paymasters of academic psychiatry, delivered a damning and all-too-public verdict on what the profession had wrought. They had refused to help underwrite the reworking of the manual, as they had in the case of prior revisions, pointing out that the APA was making millions of dollars from selling it and thus did not deserve public subsidy. Hyman had co-convened the conference that had launched the revision process. Now the director of Harvard’s Stanley Institute for Psychiatric Research, Hyman did not mince words in an interview with the New York Times: “[DSM 5 is] totally wrong in a way [its authors] couldn’t have imagined.… What they produced was an absolute scientific nightmare. People who get one diagnosis get five diagnoses, but they don’t have five diseases—they have one underlying condition.”67
His successor at NIMH, Thomas Insel, was even more savagely dismissive. On April 29, 2013, a week before DSM 5 was officially published, he complained publicly that “the final product involves mostly modest alterations of the previous edition.” That was not intended as a compliment. “In the rest of medicine,” he suggested, “this would be equivalent to creating diagnostic systems based on the nature of chest pain or the quality of fever.… [S]ymptom-based diagnosis, once common in other areas of medicine, has been largely replaced in the past half century as we have understood that symptoms alone rarely indicate the best course of treatment.” DSM 5 set itself up as psychiatry’s Bible, he reflected, but “biology never read that book” and “patients with mental disorders deserve better.”68
It was simply astonishing, he averred, that psychiatrists should practice in this fashion. Most of his psychiatric colleagues “actually believe [that the diseases they diagnose using the DSM] are real. But there’s no reality. These are just constructs. There is no reality to schizophrenia or depression.”69 Henceforth, Insel announced, NIMH would alter its approach to studying mental illness, since “we cannot succeed if we use DSM categories as the ‘gold standard.’ … That is why NIMH will be re-orienting its research away from DSM categories.”70 In particular, he suggested, “we might have to stop using terms like depression and schizophrenia, because they are getting in our way, confusing things.”71 One sees what motivated such a statement (and it must have been greeted with glee by the Scientologists), but the phrasing was distinctly unfortunate. The labels may need to go (with who knows what consequences for psychiatry’s reputation), yet the distress and pathology those traditional labels seek to capture will not disappear with them.
Insel’s comments incited a furor, with both medical and scientific journals and the mass media hastening to report his skepticism.72 He had hoped to use the controversy to advance his own pet project, something he called research domain criteria (RDoC), an attempt to install a research framework based on biology, and more particularly on a nebulous notion that related mental illness to a mysterious something called “brain circuits.” But RDoC was not ready for prime time. No other entity engaged in psychiatric research endorsed his hobbyhorse, and it has faced fierce criticism since Insel stepped down as NIMH director in 2015.73 Meanwhile, the political problem Insel and Hyman had created urgently needed a solution, or at least a public relations response. The latter duly arrived: Insel huddled with Jeffrey Lieberman, the president of the APA, and on May 14, 2013, they issued a joint press release.
Closing ranks, they argued that DSM 5 and the International Classification of Diseases “remain the contemporary consensus standard for how mental disorders are diagnosed and treated.… Patients, families, and insurers can be confident that effective treatments are available and the DSM is the key resource for delivering the best available care.” To justify their earlier statements and position, they then clarified, “What may be realistically feasible today for practitioners is no longer sufficient for researchers.” Hence, the need for RDoC. That, coupled with a promise to continue “to work together … improving outcomes for people with some of the most disabling disorders in all of medicine,” represented the best the two could do by way of damage control, passing over in silence perhaps the most damaging of Insel’s remarks, the assertion that conditions such as schizophrenia and depression were no more than artificial constructs that ought to be discarded.74