I began my training in psychiatry in 1967, when Diagnostic and Statistical Manual (DSM)-I was still the diagnostic system in use in the United States. The next year, DSM-II appeared and no one really much noticed or cared. DSM-I and DSM-II were remarkably boring little books that neither seemed to have little relevance to our understanding of patients nor any influence on decisions how best to treat them. The two predominant models in psychiatry were psychodynamic and social psychiatry and neither relied very much on diagnosing a patient's presenting symptoms (Andreasen, Reference Andreasen2007).
Then, in the early 1970s, psychiatry was rocked by an existential crisis that threatened its newly won and much prized place among the medical specialties. Two damaging and highly publicized studies zeroed in on its Achilles heel of diagnosis to challenge psychiatry's usefulness, integrity, and common sense.
Rosenhan (Reference Rosenhan1973) formulated the devilishly clever experiment of sending eight healthy volunteers, or ‘pseudopatients’, to the admission offices of psychiatric hospitals, where they feigned having heard voices. All were admitted, whereupon each ‘ceased simulating any symptoms of abnormality… the pseudopatient behaved on the ward as he “normally” behaved’ (Rosenhan, Reference Rosenhan1973, pp. 251–252). Nonetheless, all eight were kept in hospital, for between 7 and 52 days. Many other patients caught on to the fact that the pseudopatients were feigning illness, but the psychiatrists did not. This dismal result, reported in Science, triggered a controversy that called into question whether psychiatrists had common sense and could be trusted with even the most obvious diagnostic decisions (Spitzer, Reference Spitzer1975; Wilson, Reference Wilson1993).
Simultaneously, the US–UK Diagnostic Project raised the same question and also returned with a decidedly negative answer (Kendell etal. Reference Kendell, Cooper, Gourlay, Copeland, Sharpe and Gurland1971). Psychiatrists from Great Britain and the United States were shown the same videotape interviews of patients, but drew opposite conclusions from them, with the British diagnosing mood disorder, whereas the Americans saw schizophrenia. The Brits were right, but the point taken was that psychiatry spoke an unintelligible Babel of languages with, as the authors politely put it, ‘important implications for transatlantic communication, and indeed for international communication in general’ (Kendell etal. Reference Kendell, Cooper, Gourlay, Copeland, Sharpe and Gurland1971).
Luckily, psychiatry was saved at the 11th hour when a new and more systematic method of making diagnoses was devised. The absence of objective biological tests required some other means of reducing the idiosyncrasy and subjectivity of descriptive psychiatry. The answer was to provide very explicit criteria sets for defining the different mental disorders. These were first laid out in 1972 in the Feighner criteria (Feighner etal. Reference Feighner, Robins, Guze, Woodruff, Winokur and Munoz1972), then in the Research Diagnostic Criteria (Spitzer etal. Reference Spitzer, Endicott and Robins1978), and finally were exported to general clinical practice in DSM-III (American Psychiatric Association, 1980).
DSM-III was successful beyond anyone's wildest dream (Mayes and Horwitz, Reference Mayes and Horwitz2005) – in fact, far too successful for its own good and for the good of psychiatry, our patients, and society. The successive editions of the DSM have played a very constructive role when used with caution and modesty, but cause problems when reified or used beyond their competence. Intended as a simple guide to clinical practice, DSM-III was soon described and worshipped as the Bible of psychiatry (Angell, Reference Angell2011) and given more weight in many arenas than it could safely carry. We will briefly summarize its uses and harms in each of these arenas.
Clinical
DSM-III played an essential role as lingua franca in improving clinical communication and providing the needed foundation for treatment planning and guideline development (Frances and Egger, Reference Frances and Egger1999). But there were negative unintended consequences. In many centers, the complexity and individuality of human nature was reduced to a rote checklist psychiatry that noted only what was common to patients and ignored what made them different; moreover, criteria created by fallible experts were treated as hard and fast rules that required no exercise of clinical judgment.
Research
DSM-III had the profound beneficial effect of dragging psychiatry out of the research wilderness and making it a research darling, in most universities the second largest research enterprise after internal medicine. Unfortunately, though, the fascinating advances in basic neuroscience have failed to explicate the psychopathology as laid out in DSM-III and its successors. The DSM disorders are too heterogeneous and complex to allow for easy research elucidation (Gillihan and Parens, Reference Gillihan and Parens2011). In the United States, the National Institute of Mental Health in its Research Domain Categories project is now picking simpler targets, not DSM syndromes (NIMH, 2011). Some have even argued that too close an adherence to DSM has retarded research advances (Hyman, Reference Hyman2010).
Epidemiology
DSM criteria sets have been the necessary foundation for all population-based studies, but have been badly misused in reporting and understanding their results. Lay interviewers, who are necessary as cost savers in large surveys, cannot judge the clinical significance of the symptoms they elicit. Reports on community rates of various psychiatric disorder have a consistent and systematic toward overstating prevalences that is rarely addressed in discussing the results (Kessler etal. Reference Kessler, McGonagle, Zhao, Nelson, Hughes, Eshleman, Wittchen and Kendler1994, Reference Kessler, Chiu, Demler, Merikangas and Walters2005). This leads to wild swings in reported rates that are misinterpreted as being real changes rather than methodological blips.
Education
I was taught a rounded biopsychosocial model that DSM-III helped to destroy. Its descriptive criteria approach lends itself well only to the ‘bio-’ model; the ‘psycho-’ model is too inferential to allow for reliability and the ‘socio-’ model, dealing only with the interpersonal arena, is not reducible to collating symptoms in the individual. Training in many centers has become almost completely reductionistic and biological, focusing only on eliciting symptoms and prescribing medication, with little attention to individual and family psychotherapy (Sowers, Reference Sowers2005).
Drug companies
Successive editions of the DSM have become unwitting tools in Pharma's aggressive and well-financed campaign to sell pills to people who do not need them. The misleading marketing messages of ‘disease awareness campaigns’ are that psychiatric diagnosis is easy, that mental disorders are common and unrecognized, and that they are because of a chemical imbalance and always require a pill solution (Spence, Reference Spence2012). Drug companies have no direct influence on DSM decisions, but they are skillful and ruthless in exploiting them. Selling pills promotes diagnostic inflation and diagnostic inflation promotes selling pills in a vicious cycle (Batstra and Thoutenhoofd, Reference Batstra and Thoutenhoofd2012; Frances, Reference Frances2013).
Insurance
In the United States, reimbursement usually depends on the patient's first having received a DSM diagnosis (Hyman, Reference Hyman2010). This is a serious mistake that harms the individual and winds up costing more in the long run. Unnecessary and inaccurate diagnosis results when clinicians are forced to make judgments on the basis of insufficient information gathered in what are often brief interviews on what may be the worst day of the patient's life. It would be highly desirable, clinically and eventually much cheaper to allow a period of extended evaluation before a diagnosis is required for reimbursement. Most people with problems in everyday living that are now misdiagnosed as mental disorder will return to their equilibrium with watchful waiting, time, normalization, advice, support, and brief psychotherapy. They will do better without a premature diagnosis that too often leads to long-term and costly medication.
Forensic
DSM criteria are widely used and accepted in courtroom proceedings because they offer needed clarity on diagnostic questions that otherwise would be impossible to answer reliably. However, despite the cautions provided in the DSM, the manuals are often misunderstood and misused in the courts; psychiatric concepts do not always map well to legal ones and the DSM writing style is insufficiently precise for legal purposes (Slovenko, Reference Slovenko2011). The most egregious examples occur after mass murders and assassinations when psychiatrists line up on both sides of the diagnostic question – is the killer mad or bad? – canceling each other out (Applebaum, Reference Applebaum2013).
School systems
The raging epidemic of autism is partly fueled by its role as a gatekeeper to enhanced school services (Liu etal. Reference Liu, King and Bearman2010). This is bad for optimal allocation of school services and bad for accurate diagnosis of autism. Whenever a benefit is closely attached to the presence of a psychiatric disorder, that disorder will be diagnosed loosely and the benefit will not be applied optimally (Zimmerman etal. Reference Zimmerman, Galione, Ruggero, Chelminski, Dalrymple and Young2010). Educational services should be allocated on the basis of educational evaluations, and not on an unreliable clinical diagnosis not developed with an educational purpose in mind.
Disability
When unemployment rises, the rates of depression and disability increase (Gili etal. Reference Gili, Roca, Basu, McKee and Stuckler2013). Requiring a psychiatric diagnosis for disability often entrenches both the diagnosis and the disability, typically for life. It would be better and cheaper to provide more financial support and job training for everyone who has been displaced, rather than closely coupling benefits to mental disorder.
Conclusion
Getting a psychiatric diagnosis is often a life-changing event – for better or for worse. It deserves the care exercised in picking a spouse or buying a house; the commitment to it may last as long and have a profound impact. It is impossible to be a complete clinician without a good command of a criteria-based diagnostic system (either DSM or ICD). But being a complete clinician requires a great deal more than the rote application of DSM.
DSM can play a very useful role in clinical work, research, epidemiology, education, insurance and medication decisions, and in the allocation of school services and disability benefits. But it has become too dominantly determinant in too many societal decisions that are outside its proper competence. If administrative and financial decisions are to be made wisely, we need to reduce their close coupling with diagnosis and this will also increase the accuracy of diagnosis.
There is one other obvious conclusion. Psychiatric diagnosis has become far too important a societal issue to be left in the hands of the American Psychiatric Association. The experts working on any one part of DSM always strive prematurely to expand their pet area of interest. This has contributed to the past diagnostic inflation that DSM-V now threatens to turn into hyperinflation (Frances, Reference Frances2012).
In preparing DSM-IV, we aimed to be very conservative and to tame expert exuberance by having an extremely high threshold for change (Frances etal. Reference Frances, First and Pincus1995). To succeed, suggestions had to be supported by rigorously conducted literature reviews, data reanalyses, and field trials and also had to achieve unanimous approval from the Task Force. We rejected 92 proposals and accepted only two – Asperger's and bipolar II disorder – both of which led to unpredicted fads.
In contrast, DSM-V made a premature attempt to create a paradigm shift in psychiatric diagnosis (Kupfer etal. Reference Kupfer, First and Regier2002). Striving to do too much, it has introduced new disorders (Batstra and Thoutenhoofd, Reference Batstra and Thoutenhoofd2012) and reduced criteria for existing ones (Zisook etal. Reference Zisook, Corruble, Duan, Iglewicz, Karam, Lanouette, Lebowitz, Pies, Reynolds, Seay, Katherine Shear, Simon and Young2012) in a way that will further complicate psychiatric diagnosis and reduce its credibility (Dayle Jones, Reference Dayle Jones2012). The best approach for Irish psychiatry is to ignore simply DSM-V.
New diagnoses can be more dangerous than new drugs and need a careful vetting beyond the capacities of any one discipline or association (Frances, Reference Frances2012). A new structure for monitoring and revising psychiatric diagnosis needs to be developed. And we need to contain the excessive influence of psychiatric diagnosis on decisions outside its proper competence.
Forty years ago, we surmounted a crisis of confidence in psychiatric diagnosis by creating the DSM system of criteria-based definitions. Now, we face a new crisis of confidence that will be surmounted only if we ensure that DSM diagnoses are tamed and kept within their proper sphere.
Conflicts of Interest
Dr Frances has written two books critical on DSM-V, Saving Normal and Essentials of Psychiatric Diagnosis.
