On the Road to DSM-V

By David J. Kupfer, M.D; Emily A. Kuhl, Ph.D.; William E. Narrow, M.D., M.P.H.; and Darrel A. Regier, M.D., M.P.H.  

The Diagnostic and Statistical Manual of Mental Disorders (DSM), first published in 1952 and considered the foremost text for mental health specialists to recognize and diagnose mental illnesses, is undergoing revision. Almost 15 years’ worth of research and scientific advances since the fourth edition of the manual (DSM-IV) was released has made it evident that the system used by tens of thousands of psychiatrists, psychologists, social workers, primary care physicians and psychiatric researchers does not adequately mirror what they see in the clinic or laboratory.

The manual provides an index of psychiatric disorders (such as anxiety, mood and eating disorders) categorized by their core symptoms. Disorders are paired with numerical codes, based on the International Classification of Diseases, that are entered into the patient’s medical record and used for medical record keeping, reimbursement from insurance companies and to help researchers compile statistical data. Each listing includes a description of the illness and its features, followed by a listing of symptom criteria required for diagnosis. Clinicians use the DSM’s system of five diagnostic “axes” to evaluate the condition of interest as well as the presence of personality disorders and mental retardation, psychosocial and environmental problems (such as marriage difficulties or the presence of physical abuse) and general medical conditions. The fifth axis allows for a rating of how a patient is functioning overall (such as at work or at school).

Although DSM is the pre-eminent resource for psychiatric diagnosis in this country, it does have flaws. For example, DSM-IV categorizes disorders by shared features or symptoms; it classifies major depressive disorder and bipolar disorder as mood disorders because they have several overlapping diagnostic criteria. But people may show signs of multiple syndromes, their symptoms may range from mild to debilitating, and some people have unique presentations that do not fit the diagnostic mold. The manual’s system accounts for none of these situations.

DSM-IV, like its predecessors, fails to capture the diversity and complexity within disorders that make them both fascinating and perplexing to understand and to treat. Advances in neuroscience, epidemiological data and evidence from clinical care all underscore the fact that psychiatric diagnoses are not neatly compartmentalized entities with clearly defined boundaries. Although other areas of general medicine have their own ambiguities and difficulties, they convey a greater sense of certainty through diagnostic tests, laboratory values (such as levels of blood glucose or cholesterol), radiology scans and other quantitative measurements. Psychiatric states are not as reliably measured.

While such ambiguous boundaries make for enriching and challenging detective work for psychiatric clinicians and researchers, they do not lend themselves to an efficient method of organizing and conceptualizing diagnoses. Even the most seasoned psychiatrists apply DSM-IV guidelines with some level of uncertainty, and non-psychiatric professionals—who frequently have little training in psychiatric disorders and even less time to conduct a thorough clinical evaluation—are at an even greater disadvantage. Neurobiology, imaging and genetics are providing a more tangible, quantifiable understanding of psychiatric disease, but the state of the science in the early 1990s, when DSM–IV was released, did not allow its editors to incorporate these advances. Because DSM remains our primary source for diagnosis, the new edition must be able to reference the rapidly emerging scientific research and incorporate such findings when the empirical foundation supports them.

The Development of DSM

To understand why DSM-IV fails to represent the complex nature of psychiatric disorders, such as the commonly seen mixture of anxiety and depression, we first need to understand the evolution of our psychiatric diagnostic system.

In the late 1960s (the era of DSM-II), disagreement between clinicians and researchers about how to diagnose and treat psychiatric disorders correctly was growing. In response, a group of psychiatrists at Washington University in St. Louis set out to develop a revised set of diagnostic criteria based on a review of nearly 1,000 published articles and existing data. Project leaders Eli Robins, M.D., and Samuel Guze, M.D., allowed one of their young residents, John Feighner, M.D., to be the first author on a paper published in 1972.¹ The paper discussed 15 disorders whose descriptions were based on criteria that the authors believed could be corroborated by future research, thereby enhancing the validity or legitimacy of those criteria. Two years earlier, Robins and Guze had published a set of validity criteria that practitioners could use to test diagnoses.² Under these criteria, disorders are considered valid if they separate clearly from others, follow a predictable clinical course, aggregate in families and eventually have distinct laboratory tests. The “Feighner criteria” introduced in the 1972 paper explicitly identified disease symptoms and durations, a stark contrast to the vague descriptors of DSM-II. The detailed and explicit nature of the criteria made it possible for clinicians to identify similar symptom patterns in patients in different settings, thereby increasing the consistency and reliability of psychiatric diagnosis.

Shortly thereafter, the National Institute of Mental Health (NIMH) initiated the Collaborative Psychobiology of Depression Study. The participating investigators wanted to ensure that all patients who were to be enrolled in the study had the same essential syndromes. Robins, the Washington University psychiatry department chair and one of the study’s lead investigators, expanded the Feighner criteria to create the Research Diagnostic Criteria (RDC).³ The investigators developed a standardized patient interview tool called the Schedule for Affective Disorders and Schizophrenia (SADS)⁴ to be sure that patients who entered the NIMH study truly met Robins’ criteria.

The RDC became the basis for DSM-III, radically shifting the method of diagnosis from a system that used Freudian theories of causation to one based on organizing diseases according to similarities in symptoms and their duration. (DSM-IV continued this system.) For instance, the classification of anxiety in DSM-II is subdivided into forms of neuroses, including anxious, hysterical and hypochondrical, each accompanied by a text description that describes symptoms but without identifying explicit criteria (hysterical neurosis, for example, is “characterized by an involuntary psychogenic loss or disorder of function”). Using the RDC, the new approach in DSM-III provided explicit symptom and duration criteria for each disorder without implying either Freudian or biologically based theories of cause.

DSM-III showed that greater reliance on explicit criteria drastically improved diagnoses’ dependability and consistency. However, it introduced a system in which a “higher-order” disorder subsumed all “lower-order” disorders in the following hierarchy: traumatic/infections brain diseases, schizophrenia, manic-depressive disorder, major depression, anxiety disorders, somatization disorder (multiple unexplained physical symptoms), substance use and personality disorders. Under this system, patients could not be simultaneously diagnosed with both a higher-order and lower-order disorder, as the hierarchy implied that symptoms of the lower-order disorder would also be found in the dominant disorder, making a dual diagnosis unnecessary and redundant. Immediately after the release of DSM-III, a large, NIMH-supported epidemiological study used DSM-III diagnostic criteria to identify prevalence rates of mental disorders in community, hospital and institutionalized populations.⁵ This study demonstrated that a strict implementation of the proposed hierarchical restrictions would suppress a great deal of descriptive clinical information because most individuals who met criteria for one disorder also did so for a second or third—but only one could be diagnosed.⁶ This finding suggested that the hierarchical approach obscures the true complexity of some psychiatric disorders and, by obfuscating important targets for clinical research, could be hindering the development of appropriate treatments. The revised DSM-III (DSM-III-R) partially abandoned this hierarchy but resulted in a large number of patients diagnosed with multiple disorders—a problem that persists in DSM-IV.

Treating patients who have multiple psychiatric diagnoses poses a significant challenge. For example, recent analyses⁷ have shown that people with non-psychotic major depression occurring alongside anxiety or substance use did not respond as well to treatment as those with depression alone. A majority of the major-depression patients in these analyses exhibited such co-occurrences, including about 10 percent who had both anxiety and substance use disorders. In addition, a recent study of primary care patients who had reported severe depression, anxiety or somatization disorder revealed that more than half had been diagnosed with at least two of these.⁸ Patients with all three disorders had significantly more difficulty maintaining physical health and social relationships, adding to the complexity of finding the most appropriate treatment for them.

Despite these and other findings suggesting that “pure” disorders are rare, the DSM-III classifications describe such disorders, easily distinguishable from one another and from healthy behaviors.⁹ The structure of DSM-IV perpetuates this misperception. Further, DSM-IV has not entirely abandoned the DSM-III hierarchy; for example, a psychiatrist following the criteria today can diagnose neither attention-deficit/hyperactivity disorder in the presence of autism nor generalized anxiety disorder if it occurs exclusively in the presence of a mood disorder.

DSM-IV compartmentalizes diagnoses into strict categories that do not reflect the most common symptom patterns that actually appear in patients. For example, the criteria for Major Depression fail to reflect the potential co-occurrence of anxiety symptoms, which appear in more than 50 percent of patients with depression. If a patient fails to meet full criteria for Major Depression and has significant anxiety symptoms, which together cause significant distress and impairment, the diagnosis would fall under the “not otherwise specified” (NOS) category. People who receive such diagnoses do not officially meet criteria for any specific DSM disorder, yet their symptoms may be severe and they may have so much difficulty with ordinary relationships and daily activities that they warrant attention and possibly treatment. Conditions that fall just short of diagnostic requirements, mixed disorders, and those with uncommon or unusual symptoms all may land in the NOS category. Because clinical trials of psychotherapy or medication are conducted only for DSM-defined disorders, an NOS diagnosis makes it difficult for a physician to choose an appropriate evidence-based treatment.

The Utility of Diagnostic Dimensions

One of the more promising pathways out of the categorical conundrum that the DSM revision task force is addressing is a “dimensional” approach—one that enables clinicians to consider distinctive aspects that differ significantly within a disorder (such as symptom severity), as well as the presence of symptoms that are outside the “pure” disorder definitions (such as anxiety and somatic symptom levels for patients with depression). This method incorporates variations of features within a disorder (its severity and whether it is acute or chronic, for example) rather than relying on the answers to simple yes or no questions to arrive at a diagnosis. Dimensions also can be used to examine features of other diagnoses. For example, if DSM-V provided for clinicians’ documentation of certain symptom dimensions in all patients—such as sleep/wake functioning, cognition, mood and anxiety symptoms, substance use and psychosis—the result would be a more useful and realistic representation of the patients’ clinical status than that of the current method. The dimensional approach also helps reduce the need for multiple diagnoses, provides background explanation for an NOS diagnosis, clarifies the presence and severity of individual symptoms and informs treatment planning. Capturing variations would increase DSM-V’s clinical utility and support a systematic, measurement-based approach to monitoring symptoms and their severity when making decisions about treatment.¹⁰

Researchers already have shown such measurement-based care approaches to be both feasible and useful in primary care and mental health settings.¹¹ For instance, investigators found that psychiatrists readily accepted and used one such dimensional measure, the nine-item depression scale of the Patient Health Questionnaire (PHQ-9). Further, psychiatrists reported that they used results of the questionnaire when making treatment decisions for patients with major depression during approximately 40 percent of patient visits. These decisions included changing the dosage of an antidepressant; adding, introducing or switching antidepressants; initiating or increasing psychotherapy; and engaging in additional suicide risk assessments.¹¹ Although practitioners can certainly make such treatment decisions without guidance from measurement-based care methods, these dimensional assessments provide objective data that guide clinical decisions and that might otherwise be overlooked in busy, real-life clinical settings. Notably, 42 percent of the patients in the sample also had an anxiety, substance use or other psychiatric disorder. This indicates that even in real-world settings, where patients’ symptoms vary and time is at a premium, dimensional assessments can assist in diagnosing disorders as well as measuring severity, predicting outcomes and planning treatment.

The expectation that the Robins and Guze validity criteria would lead to the validation of disorders as distinct entities has gone largely unfulfilled. For example, recent advances in neuroscience demonstrate that the classification structure of DSM-IV that strictly separates schizophrenia and bipolar disorder may not be scientifically valid. Family, twin and adoption studies have repeatedly demonstrated an overlap in many genetic markers associated with a susceptibility to these two disorders,¹² This parallel suggests a possible shared genetic vulnerability to some symptom patterns that belies these disorders’ current representation as diagnostically distinct entities. Although they have some notable clinical differences, their strict separation in DSM-IV only encourages clinicians and researchers to persist in conceptualizing them as fundamentally discrete. The use of dimensions also likely will shed light on how other now-separate disorders with shared multiple-gene vulnerabilities and shared symptoms have a tendency to cluster, thus raising important questions about what the overarching structure of DSM-V should be.^13,14

These multiple-gene susceptibility findings lend further support to a reorganization of DSM that moves away from a strict, categorical, “yes/no” approach that was more consistent with the previously prevailing but now obsolete idea that most mental disorders could be linked to a single gene. Basing DSM in part on findings from neurobiological studies is one proposal. For example, studies of genetic variations within groups of people with schizophrenia and of twins where only one sibling has schizophrenia indicate that this disorder arises from a complex combination of genetic and environmental factors.¹⁵ People with schizophrenia may have relatives who do not meet DSM-IV diagnostic criteria for the disease but nonetheless exhibit abnormalities consistent with it—both neurobiological (such as thinning of the brain’s cortex) and neuropsychiatric (such as impaired cognitive abilities).¹² The presence of overlapping conditions, such as bipolar disorder and schizophrenia—or depression and anxiety—has led many experts to suggest that concurrent symptom patterns that cross existing diagnostic boundaries may constitute more aptly named syndromes, such as “affective-laden schizophrenia” or “anxious depression.”

To reorganize categories in DSM-V to better reflect the state of the science, we will have to do more than simply revise individual diagnostic criteria in DSM-IV. Realistically, given the state of the science, psychiatry as a field and of DSM users in general are not yet ready for a drastic overhaul of DSM’s organization. As such, DSM-V revision experts are examining whether specific indicators can inform and validate the grouping of disorders while maintaining much of the existing categorical framework.

Eleven such indicators are currently under examination. Indicators¹⁶ shared by multiple disorders include:

• neural substrates (brain structures that underlie a behavior or psychiatric state),
• genetic risk factors,
• specific environmental risk factors,
• biomarkers (such as a feature on a brain scan that denotes disease presence or progression),
• temperamental antecedents (characteristics of temperament that increase risk for a particular disorder),
• abnormalities of emotional or cognitive processing,
• treatment response,
• familiality (common occurrence within a family),
• symptom similarity,
• course of illness
• and high comorbidity (the co-occurrence of two or more conditions).

These indicators can serve as evidence-based guidelines to inform decision-making about how to cluster disorders to maximize their validity and therefore their utility for clinicians.

Regrouping psychiatric disorders will enable future researchers to enhance our understanding of the origins and common disease processes (pathologies) among disorders. It will also provide a base for future changes that reflect advances in the underlying science: Data can be re-analyzed over time to continually assess the groupings’ validity. Thus, after DSM-V is published, changes to the volume will occur only to the extent that future discoveries in neurobiology, genetics, epidemiology and clinical research support them.

One Vision: A Psychiatric “Review of Systems”

To advance clinical practice and to provide a framework for future testing of the standards for diagnosing mental disorders, the forthcoming DSM-V criteria need to better reflect the true nature and scientific underpinnings of psychiatric disorders while preserving their link to previous diagnostic conventions.¹⁰ An important strategy for achieving these objectives involves the integration of previously described dimensional measures with the current criteria that define mental disorders. By recommending patient self-report screening methods that cut across multiple diagnostic areas, the DSM-V will facilitate a more systematic review of multiple symptom domains (brain or mental systems). This approach is comparable to general medicine’s review of systems, which resembles casting a fishing net that simultaneously captures everything at once and nothing in particular. In general medicine, this broad review process is crucial for detecting pathological changes in different organ systems when creating a comprehensive diagnosis and treatment plan.

Similarly, DSM-V will provide recommendations for guiding practitioners through a systematic review of multiple brain or mental systems to prevent a premature focus on the most obvious symptoms—thereby helping mental health practitioners conduct a more comprehensive mental status assessment. When an initial screening reveals one or more symptoms in different domains, patients and clinicians will be directed to follow up with a more intensive dimensional evaluation of symptom severity, followed by a clinical judgment about whether the symptoms are sufficient for a mental disorder diagnosis. This combined dimensional and categorical diagnostic approach will also help clinicians and research investigators establish a database of better-classified syndromes for future clinical, epidemiological and neurobiological study.

Even though the DSM comprises diagnoses, we must remember that its construction is not so much about pathology as it is about people. Our aim with DSM-V, first and foremost, is to improve patient care. The unique features that a patient brings to an assessment—family background, life experiences, social functioning and relationship history—are as important as the symptoms themselves; without the relevant personal information, a physician observing symptoms alone may not make a correct diagnosis. The science behind DSM-V should therefore serve to strengthen, not to overshadow, clinical care by connecting the most recent scientific findings to the objective information each clinician and patient brings to diagnosis and treatment. If, for example, Tourette’s syndrome shares observable symptoms and underlying biomarkers with obsessive-compulsive disorder, clustering the two in DSM would encourage clinicians to look for tics, a common symptom in Tourette’s but not in obsessive-compulsive disorder.

Research gains in recent years will advance the scientific validity and clinical utility of DSM-V, scheduled for publication in May 2013. As new findings from neuroscience, imaging, genetics and studies of clinical course and treatment response emerge, the definitions and boundaries of disorders will change. Perhaps the most important characteristic of DSM-V is that it will be a living document with a support system for a continuous review and revision process.

Psychiatry at Stalemate

By Paul R. McHugh, M.D. 

The Diagnostic and Statistical Manual of Mental Disorders (DSM) of the American Psychiatric Association (APA) has, from its first edition (DSM-I) of 1952, represented the official taxonomic enterprise of American psychiatry: identifying, delimiting and classifying the diverse mental and behavioral disorders that psychiatrists aim to treat or prevent. Although the series was launched primarily to collect statistical information on mental disorders, with its third edition (DSM-III), published in 1980, this previously descriptive enterprise took a new and prescriptive turn and began directing psychiatric diagnostic practice. Serious challenges to the legitimacy of the discipline—many from within the profession itself—provoked this new effort. They ranged from claims that disorders were not properly differentiated or aptly treated to assertions that mental illnesses were social fabrications of psychiatrists—“myths,” in fact. None of these accusations could be brushed aside given that, at the time, diagnostic agreement between two psychiatrists about the same patient was hardly better than chance.¹

In the face of such foundational disarray, Robert Spitzer, the editor of DSM-III, proposed to make psychiatrists more “scientific” (and thus more medically legitimate) by prescribing for them, through the new edition, a better way of identifying and distinguishing mental disorders. Psychiatrists, he said, should use what he dubbed an “operational,” “a-theoretical” method that could, by identifying each mental disorder from its symptomatic presentation, steer clear of the conflicts over those explanatory theories of mind dividing psychodynamic and neurobiologically oriented psychiatrists on which much of the disarray rested. With awesome diplomatic and political skill Spitzer persuaded the profession to accept this new approach, and now for a generation American psychiatrists have employed DSM’s symptom-based method in practice, teaching and research. DSM-IV, published in 1994 with minor changes, followed the same path, so that hereafter I shall refer to the pair as DSM-III/IV.

DSM-III/IV helped resolve the turmoil that fractured psychiatric discourse in the 1970s by getting all, psychoanalysts as well as neurobiologists, to concentrate on the few things they could agree about: the symptoms patients displayed and how those symptoms naturally emerged over time. This approach succeeded so well that arguing today about “dynamic” and “biological” psychiatry is an anachronism. And no one seriously suggests that mental disorders are “myths,” especially given that the diagnostic consistency of DSM-III/IV improved the efficacy of both psychological and pharmacological treatments.

And yet all is not well with psychiatry under this new dispensation. Many difficulties have emerged directly from it; they are serious enough to challenge the usefulness of any revision of DSM that does not make a significant move to resolve them.

Most of these problems derive from the ad hoc nature of DSM-III/IV and its glorification of process over substance. It aims only to enhance diagnostic consistency. It does not speak to the nature of mental disorders or distinguish them by anything more essential than their clinical appearance. Not a gesture does it make toward the etiopathic principles of cause and mechanism that organize medical classifications and carry, for physicians and patients alike, the promise that rational, effective, cause-attacking therapies will eventually replace symptom-focused, palliating ones.

I and others contend that this symptom-based approach, having accomplished its original purpose of settling the discord within psychiatry, should now gradually but resolutely be supplanted. What we needed in 1980 is not what we need now, a generation later. In fact, today’s pressing issues are those produced by DSM-III/IV.

The most fundamental of these problems is that DSM regularly fails to distinguish between conditions with similar symptomatic appearances such as between ordinary sadness and clinical depression, as Allan Horowitz and Jerome Wakefield have recently and thoroughly documented .² This failure derives directly from the inattention of DSM-III/IV to distinguishing the generative causes of either normal or abnormal mental states.

But the method and purposes of DSM are so aligned (and the profession so accustomed to them) that practitioners and editors alike resist suggestions for revising a new edition in ways more substantial than tinkering with the criteria and expanding the collection of certified conditions. Psychiatry has become a field bridled by its own method and needs to fight its way free.

A Classificatory Dead End

The symptom-based diagnostic method of DSM differs fundamentally from the way doctors classify and distinguish medical disorders. Internists differentiate disorders according to how they display—through their symptoms, signs and laboratory data—some functional disruption of bodily design, such as the pump function of the heart in congestive heart failure, the filtering function of the kidney in renal diseases or the gas-exchange capacities of the lungs in pulmonary disease.

Medical classifications of this sort—properly identified as generative in that they build upon concepts of cause or mechanism generating the conditions—evolve and improve over time, not simply because they follow progress in the natural sciences (physiology in particular), but crucially because they interact with and enrich such sciences with information from physicians who recognize diseases as “experiments of nature” revealing of mechanisms behind the symptoms and their course. For example, William Harvey not only used the experimental method with animals to demonstrate the circulation of the blood but also pointed to the features of human congestive heart failure to demonstrate just what symptoms and signs appear when that circulation begins to falter. Because this scientific partnership between medicine and biology became so successful, time-honored but historically separate modes of thought—the “healing” tradition of medicine and the “natural history” tradition of biology—today merge as aspects of “life science,” to our benefit.

No symptom-based, appearance-driven diagnostic system such as DSM can do the same for psychiatry and neuroscience. It is too passive in relation to knowledge; it awaits discovery rather than suggesting approaches to it; it remains satisfied with consistent analytic definitions of mental disorders at the expense of any synthetic grasp of them from their origins and generation. In all these ways DSM-III/IV has produced what no one ever thought possible—a psychiatry that’s boring.

A close look at its actual classificatory operations reveals why, resistant to change, DSM has reached a dead end.

DSM-III/IV, like any way of classifying things, must deal with two issues in confronting their diversity—here, of the mental disorders. The first is the demarcation, definitional issue whereby individual disorders are identified. The second is the separation, differentiation issue whereby the defined disorders are separated from one another. DSM-III/IV insists on addressing both of these issues “a-theoretically”—that is by an approach that attempts to be “neutral with respect to theories of etiology” (DSM-IV p. xviii).

DSM-III’s a-theoretical approach to the demarcation/definitional issue was simple; it “asked the experts.” Spitzer selected psychiatrists he believed to be expert students of each disorder and solicited definitions and diagnostic rules from them. He, with his fellow editors, refined these suggestions and chose for the official diagnostic algorithms the symptoms easiest to elicit from patients in an interview and to use with consistency in practice.

This definitional exercise now tends to go the other way. “Experts”—many unfortunately with a vested interest (financial, political, legal, ideological) in gaining an official “stamp” certifying the existence of a particular mental condition—now beat on DSM’s editorial door for the inclusion of their favorite malady in the manual. Because no more objective criterion than clinical testimony can be employed to challenge an admission to the DSM catalog, these “experts” cannot be denied if they are a sizeable “lobby” and bring with them a set of user-friendly diagnostic symptoms for the condition they want listed.

This a-theoretical approach to the demarcation/definition issue has produced several problems. First, some DSM conditions are defined with such inclusive symptomatic criteria that they gather in too many patients and so are diagnostically overused to the detriment of refined assessment and treatment. Such DSM-III/IV diagnoses as Bipolar Disorder and Attention Deficit Disorder, for instance, are now so frequently applied, and so many people are inappropriately receiving Prozac or Ritalin, that these diagnoses and medications have become household words to the American public and material for late-night television comedians.

Second, many varieties of the same disorder are separated in the DSM because it emphasizes trivial distinctions in symptom expression. For example, patients given diagnoses of narcissistic personality disorder, histrionic personality disorder, or borderline personality disorder are all unstable (‘high neuroticism”) extraverts who tend to be disagreeable. The specific diagnostic label they receive depends more on what feature a doctor chooses to emphasize than upon anything psychologically distinct or critical to their treatment. A patient who seeks a second opinion across town may well receive one of the other labels—and it would be just as correct, according to the DSM.

Finally, because DSM lacks any other way of judging what fits as a legitimate psychiatric condition but must accept what “experts” champion, it grows in size with each edition, becoming ever more impressive in its list of diagnoses even as it remains ever so humble in its explanations of them.

DSM-III/IV also manages the separation/differentiation issue of classification a-theoretically. The editors employ an old and familiar method of drawing distinctions as demonstrated by the decision trees in Appendix A of both DSM-III and IV. These depict the traditional “downward” method for dividing large classes into progressively smaller sets by using a sequence of dichotomizing (yes/no) questions. For example, the decision tree recommended for the specific diagnosis of an anxious patient uses a sequence of such dichotomizing questions as “due to medical condition,” “cued by situation” or “response to traumatic event” until a yes picks out, say, generalized anxiety disorder or acute stress disorder as fitting the patient of interest.

This method is traditional in being first formally described by Aristotle. But it resembles nothing so much as the children’s “Twenty Questions” game, wherein a player, by means of a sequence of yes/no questions (“animal or mineral,” “warm blooded or cold blooded,” “feathered or furred” and so on), ultimately identifies the object the other player has in mind. It also is the standard method of naturalists’ field guides such as Roger Tory Peterson’s Field Guide to Birds of North America. It works when the sole aim is identification and when the dichotomizing questions are easy enough to answer in the field, as Peterson’s many fans will testify.

But clinicians have several problems with this method of demarcation. Because the dichotomizing questions that ultimately determine a diagnosis are to some extent arbitrary, the method is vulnerable to abuse when advocates interested in producing a given result devise a way of inserting their own distinctions in the sequence. The best example of this is the artificial distinction drawn in the DSM between “conversion disorders” (where patients display pseudo-neurological phenomena of a physical kind, such as paralysis or sensory loss) and “dissociative disorders” (where patients display pseudo-neurological phenomena of a psychological kind, such as amnesia or multiple personalities). A diagnostic distinction between these two expressions of illness-imitating, attention-seeking behavior implies that they are different in some essential way. In reality they are behaviors of the same nature in that both are provoked by suggestion, display symptoms that can attract contemporary clinical attention and services, and, not that infrequently, are exhibited by the same patient.

Another critical problem is that this downward method of differentiation in psychiatry operates with so little information—certainly none of a psychological or neuroscientific kind—that it confounds those symptomatic expressions that are primary and essential to a disorder with those that are secondary and adaptive, such as the depressive and paranoid reactions shared by many disorders. The method hides this diagnostic and therapeutic complication by emphasizing the consistency of its final decisions.

The result is the situation we have today. A process aimed at producing diagnostic consistency has not only generated several practical problems of its own but has reached a dead end where the only route of escape is the one that method categorically rejects: the re-introduction of concepts of cause and mechanism—theories, in fact—into the diagnostic reasoning of the discipline.

However, at a meeting of the Johns Hopkins Department of Psychiatry in 2008, Michael First of Columbia University, who has had senior editorial responsibility for DSM, told us that the editors all agree that despite the increase in psychiatric research that followed the publication of DSM-III in 1980, nothing has emerged in the 30 years since that permits us to diagnose any condition in the DSM by the medically traditional etiopathic—cause or mechanism—approach. Thirty years with a field guide and nothing on the horizon offering another way. And, yet the editors of DSM-V say it must come forth as “Son of DSM-IV.”

Surely one can wonder about the wisdom of this advice. Can DSM-V offer us nothing to provide a better conceptual grasp of mental disorders or, at a minimum, suggest—in the form of reasonable hypotheses based on psychological and neuroscientific evidence—their nature, mechanism or cause?

Just consider the point of view of a patient who has received a DSM-III/IV diagnosis. What does he take away on learning from his doctors that his distressful state of mind “satisfies criteria” for Major Depression? Should he presume that he is afflicted with a disease—something he has—or should he think of his problem as an emotional state or reaction to something he encountered? Or should he strive to realize that the problem is a propensity for low spirits tied to his personality—something he is—or should he consider it a state of mind produced by how he is behaving—something he is doing? The diagnostic label he has received makes none of this clear to him.

It’s all so tautologous. You’re miserable, seek medical attention and are told you’ve got Major Depression! How much more do you know? And what’s the most rational treatment now that you have a label? From where is the help to come?

A Modest Suggestion

Official psychiatry is at stalemate. It must produce a new edition (DSM-V) soon to fit the World Health Organization’s schedule for updating the International Classification of Diseases (ICD), used worldwide for diagnostic and clinical purposes, and for epidemiological studies of disease prevalence and death rates. But currently, most revision proposals either amount to little more than tinkering within the DSM symptom-based diagnostic system or are too uncertain of outcome to be encouraged. Many psychiatrists fear that any classificatory differentiation based on views about the generation of psychiatric disorders will restart the war between the dynamic and biological schools that DSM-III settled.

But surely DSM-V’s editors can take some tentative steps toward classifying psychiatric conditions by what underlies them—particularly if these steps are based on modes of thought ever implicit in much of psychiatric practice and research. Simply making explicit what has been implicit would be progress.

A grouping of disorders, not by their symptomatic similarity but in families that share a causal, generative nature, would introduce into DSM the etiopathic principle fundamental to medical classifications. The DSM already employs something rather like this idea in that it separates disorders that derive from intelligence and temperament into its “Axis II” grouping and so distinguishes them from all other disorders, which it clusters on “Axis I.”

The discipline would surely advance if DSM specifically separated those disorders that represent breakdowns in the mind’s design and indicate brain disease (such as dementia, delirium and schizophrenia) from those that represent disturbed expressions of the mind’s design in the form either of behavioral misdirections (anorexia nervosa, alcoholism and sexual paraphilias) or emotional responses to distressful life encounters (adjustment disorder, post-traumatic stress disorder and bereavement). Such a reorganization of the catalogue would not require abandoning the familiar DSM-III/IV diagnostic algorithms. Rather, it could simply be superimposed upon them. And so, consistency of diagnosis would be retained even as the possibility of eventually resting diagnoses upon generative mechanisms would be foreshadowed.

By encouraging clinicians to think of mental disorders as clustering in families, official psychiatry would prompt them to study, debate and ultimately seek out implications tied to the generative processes being proposed as the bases of the clusters— processes that are proposed to either evoke or sustain the conditions, that rest sometimes on cerebral changes and sometimes on life circumstances and that, if confirmed, will inform rational treatment and prevention.

Such a modest rearrangement and thematic reorganization of disorders would persuade psychiatrists to think much more “contextually” about their patients and their patients’ disorders. This would prompt them to reject symptom checklists as diagnostic instruments in favor of more thorough assessments of their patients that specifically consider their social environments and developmental trajectories, wherein lie the generative sources and differentiating information about many of their disorders.

An official classificatory system should do more than name and list disorders. It should organize them in ways that propose the modes of study most likely to explain them. Such revision of DSM would bring direction back into psychiatric thought, practice and research—indeed it would impel psychiatrists in such a direction by its implications. Psychiatrists should not be satisfied—especially after 30 years—with a process that runs on the hope that diagnostic consistency alone will eventually translate into explanations. This approach has failed for more than a generation to deliver discoveries that can amend it.

Psychiatry may not have what it takes to form a unified “theory” of mental disorders, but it has concepts with enough credibility to indicate that certain disorders differ in their fundamental nature and that these differences are sufficient to influence treatment decisions and to suggest the most apt ways of study. If DSM-V turns out to be nothing more than a tinkered-with version of DSM-IV, many will view it as a failure of nerve.

Perhaps more than anything else, such a simple classificatory rearrangement—by encouraging reflection on the part of psychiatrists and patients alike about the generation of mental disorders (and mental states that are not disorders)—would bring back to psychiatry what we see DSM-III/IV having taken away: a body of assumptions and debatable issues to be ever thought over, contemplated for their implications and sifted through in discerning approaches to better practices and critical research. Regaining this essentially contemplative aspect of the discipline will bring renewed investigative energy to psychiatry and, far from unimportantly, make it fascinating again.