AS THE NEW YORK TIMES reported recently, psychiatric epidemiologists from the Harvard Medical School have published studies purporting to demonstrate that some 55 percent of Americans suffer from mental illness in their lifetime. These studies--which cost $20 million, most of it out of the taxpayer's pocket--are based on a survey of 9,282 randomly selected English-speaking subjects over the age of 18 who were seen in their homes by technicians trained to ask specific questions about symptoms believed to indicate mental illnesses. The results led Thomas Insel, director of the National Institute of Mental Health, the studies' primary sponsor, to note that indeed "mental disorders are highly prevalent and chronic." More than half the people of the United States, in other words, have been or are mentally ill. What should we make of this?
Not to put too fine a point on it, we should take the study's conclusions with a huge grain--perhaps a silo would be required--of salt. Diagnostic exaggeration dogs psychiatry today and will not subside until research psychiatrists use ways closer to those of practicing clinicians for recognizing mental disorders and differentiating the serious from the trivial in mental life. Let me explain.
The survey technicians were instructed to fill in a questionnaire by asking the subjects about mental symptoms such as depression and anxiety that they might have experienced in their lives. Such technicians, sticking to the prescribed inventory, essentially act as secretaries, recording what people say they recall from their past. The techs gather no sense of the persons they are meeting--no appreciation of their life circumstances, the issues they have dealt with, what strengths they brought to bear, or what vulnerabilities they overcame, in dealing with the good and bad fortune life brought them. The individual's family, social circumstances, temperament, character, opportunities, successes, and disappointments are all outside the attention of these interrogators.
Instead, the technicians run down their checklist of symptoms with no thought to causes, simply recording a yes or no answer to each. This is not a psychiatric examination; it is barely a census. The assessment does not rest on a trusting relationship, it presumes honesty and openness in the replies, and it assumes that both the subjects and the technicians understand the questions the same way the experts who constructed the inventory did. Finally, by focusing solely on symptoms--indications of disease or disorder--these inventories tend to direct attention to human frailty rather than to human strengths and to emphasize the burdens and obscure the gifts that life has brought these subjects.
At Johns Hopkins, we became aware of these problems after the last national attempt to do a census of the mentally ill--the so-called Epidemiological Catchment Area Study (ECA) of the early 1980s. We followed up similar questionnaires with a complete examination by qualified psychiatrists of a sample of the subjects previously assessed. These examinations produced diagnoses that failed miserably to match those generated by the less thorough and clinically inexperienced technicians. The questionnaires depicted individuals who were distressed but could neither accurately identify the nature of their distress nor make confident claims about any mental impairment. Nothing in the present study indicates that its expanded version of the old questionnaires can do any better at diagnosing the subjects.
But this simply raises the question, Why would anyone dream that an inventory of psychic aches and pains would reliably identify mental impairments and distinguish them from the kinds of mental distresses that are part of every person's life?
In addition to relying solely on respondents' yes or no answers to a checklist, the investigators are committed to employing the official Diagnostic and Statistical Manual of Mental Disorders--Fourth Edition (abbreviated DSM-IV), which bases all psychiatric diagnoses on symptoms and their course, not on any fuller knowledge of the person. It is as if public health investigators studying the prevalence of pneumonia over time in the American population were satisfied to call every instance of a cough with a fever and a mucoid sputum a case of pneumonia.
Internal medicine gave up on symptom-based diagnosis more than a hundred years ago, replacing it with diagnosis that rests on knowledge of pathology and what produces it. Thus, internists no longer speak of coughs as racking, brassy, or productive, but as produced by viral or bacterial infection, allergies, or vascular congestion. They no longer differentiate Tertian, Quotidian, and Continuous Fevers but fevers from infection, neoplasia, dehydration, and so on.
DSM-IV makes no attempt to classify mental symptoms or complaints by cause. As a result, it mingles serious and impairing conditions with other forms of mental distress in one hopeless and scientifically indigestible stew. When this diagnostic method is employed for a census of mental disorders in the citizenry, it ominously exaggerates the incidence and the nature of mental troubles. It leaves the public wondering: If more than 50 percent of Americans have at some point been mentally "impaired," what constitutes a "normal" mental life?
Another way of stating the problem is that DSM-IV is the medical counterpart of a naturalist's field guide--say, Roger Tory Peterson's Field Guide to the Birds. To develop his guide, Peterson asked expert bird watchers what features of shape, coloring, voice, and range they used to distinguish one warbler from another, and he arranged his guidebook accordingly. As a result, bird watchers became more precise in the terms they used to describe what they saw. But as Peterson noted, amateurs relying on the way birds look often confuse varieties with separate species, while ornithologists turn to biology to make more fundamental distinctions.
Similarly, clinical psychiatrists in 1980 wanted to find a way to apply their diagnostic terms consistently. With DSM-IV, they agreed on which symptoms they would use as criteria for each diagnosis, and thus increased their diagnostic consistency. But the best clinicians apply DSM-IV diagnostic terms only after they have fully examined the patient and come to see these symptoms in context. They do not simply run down a checklist of symptoms, count them up, and attach a diagnosis, as did the technicians from Harvard.
Psychiatrists are right now rewriting the diagnostic manual. I believe they will move closer to internal medicine, classifying patients according to what has provoked their symptoms rather than according to the symptoms alone. Only then will scientific and epidemiologic studies in psychiatry improve.
In the meantime, while scientists are working to lift psychiatry beyond the level of a field guide, epidemiologists should stop expending time and money repeating surveys that purport to measure the prevalence of psychiatric disorders but instead only mislead and alarm the public. They should spend their efforts in more productive areas of psychiatric research.
They might, for example, start following people over time, as cohorts with particular life circumstances: They might consider the long-term performance of children with particular classroom-identified dispositions or children exposed to various forms of deprivation or trauma early in life, seeking to discover how these people manage the hurdles they face and which vulnerabilities to mental problems and which resiliencies they manifest in later life. Epidemiologists should attend to studies where patients with particular characteristics--such as temperament, upbringing, or stress--are compared with nonpatients with similar characteristics (so called case-control studies) testing whether these characteristics provoke, protect against, or are incidental to the patients' mental unrest or illness. They should enhance cross-cultural knowledge of how mental impairment, as opposed to mental distress, is expressed by people of differing cultures and exactly what measures help to prevent or treat the case examples.
Analytic studies like these could accomplish much more than descriptive surveys that do little in the long run but exasperate the public and make ephemeral headlines. Along the way, with these more specific studies we would likely discover not that the majority of people are impaired but just how remarkably resilient most of us are and what distinct and wonderful assets most people bring to life. To conduct more of the same kind of empty surveys as are now being done is, I'm afraid, a little crazy--with crazy defined as doing the same thing again and again and expecting a different result.
Paul McHugh is a university distinguished service professor of psychiatry and behavioral science at the Johns Hopkins School of Medicine and former psychiatrist in chief of the Johns Hopkins Hospital.