Introduced in 1980, DSM-III was a major paradigm shift for psychiatry. It provided explicit, well-defined diagnostic criteria. The criteria were polythetic in that not all the criteria need be present to make the diagnosis. It provided a diagnostic system that contained other pertinent information (e.g., it was multiaxial) in which IQ, medical illnesses, stressors, and social functioning could be taken into account. It also necessitated that we look at the patient longitudinally rather than at a moment in time. The explicit and understandable criteria of DSM-III and its successors have been very useful in educating patients. DSM-III spurred a great deal of research, particularly in anxiety and affective disorders, as well as laying the foundation for treatment outcome studies, such as the important Epidemiologic Catchment Area Program (1). In fact, the new DSM diagnostic process has dominated the research, teaching, and contemporary practice of psychiatry (2). The DSM diagnosis has almost become a thing in itself—a certainty of "concrete" dimensions. The DSM diagnosis has become the main goal of clinical practice. DSM-IV, as "allegedly" being more data based, has even assumed the aura of allowing psychiatry to keep pace with the rest of medicine as a "technological triumph" but our current diagnostic process and zeal may also be ruining the essence of psychiatry. It is time to look at what we have wrought and make some midcourse corrections.

It is conceivable that some of the current plight of psychiatry is not due to health care reform alone but is partly our own doing. The zeal with which we and managed care have grasped the DSM process to our bosoms may also be a factor. The current DSM process gives the image of precision and exactness. In fact, many have come to believe that we are dealing with clear and discrete disorders rather than arbitrary symptom clusters. We are now being held to our rhetoric by managed care companies which state that if the patient's symptoms fulfill the criteria for a major depressive episode, then the treatment should be X treatments and Y drugs. At best, we are between Scylla and Charybdis—we no longer want to say that each patient is a unique individual, nor can we honestly say that every case clearly fits diagnostic criteria. All of this apparent precision overlooks the fact that as yet, we have no identified etiological agents for psychiatric disorders. Our diagnoses are nowhere near the precision of the diagnostic processes in the rest of medicine. While there are similar diagnostic processes in medicine, most medical diagnoses are based on objective findings; e.g., cancer is based on structural pathology, pneumonia on a bacterial or viral agent, and hypertension on the numerical deviation from a numerical norm. In only a few areas is the diagnosis based purely on the patient's subjective complaints (headache would be one—and we all know how difficult this diagnosis can be). In psychiatry, no matter how scientifically and rigidly we use scales to estimate the patient's pathological symptoms, we are still doing pattern recognition. We are still making an empirical diagnosis and not an etiological diagnosis (deductive) based on disruptions of either structure or function.

Other issues created by the current use of DSM-IV are more subtle, but they are real and they are important. One, we have lost the patient and his or her story with this process; two, the diagnosis, not the patient, often gets treated; three, surprisingly, the study of psychopathology is almost nonexistent; and four, the strict focus on diagnosis has made psychiatry boring (this never seemed to be a problem in our field before).

At a recent teaching conference, the case of an 18-year-old woman with an eating disorder was presented. She was being treated with behavioral techniques. The resident started his presentation as follows: "This 18-year-old girl became preoccupied with her figure and lost 25 pounds in the three weeks preceding admission." The case discussant stopped the presentation and said it was rare, in her experience, that a patient with an eating disorder would have such an acute weight loss and asked if there was anything else going on. Finally, a nurse noted that a month previously, the patient's father had left home with his secretary, and the patient, at that point, stopped eating and became preoccupied with her weight. Somewhere, the patient's story had gotten lost. What was being treated was a diagnosis and not the patient. What happened here? These were not incompetent clinicians. However, it became clear that in our contemporary psychiatric practice, the patients' stories and the way in which the patients are functioning are not necessary to the diagnostic process. In this case, the symptoms were sought, but not the antecedents or consequences of these symptoms. There are even some recent data that indicate how the symptoms used for the diagnosis color the clinician's perception of the patient's functioning. Roy-Byrne et al. (3) showed that psychiatrists' global ratings of patients' functioning were totally unrelated to detailed nurses' ratings of the same patients' functioning but were highly correlated with their own rating of symptom severity. Halleck (4), in 1988, pointed out that DSM-III seemed to focus the trainee on the diagnostic process and not on the patient. Our view of the patient can become restricted as we are looking for a predetermined set of symptoms. This not only tends to focus our information on what we are looking for, but can allow us to ignore important distinctions between patients. It is important to remember that we are evaluating subjective experiences reported by a patient. Can these self-reported symptoms be effectively evaluated without exploration of their antecedents, consequences, overall context, and fluctuations in intensity over time? Jaspers (5), the great phenomenologist, would say no.

We are not looking at or studying the patient's phenomenology anymore but are looking for the symptoms needed to make the diagnosis. For example, most current residents will look at you blankly when you mention the term thought disorder. In 1979, Andreasen (6) did a seminal study of thought disorder, citing 18 different types of speech or thinking disorders that had been described in schizophrenic patients. At present, our only concern about any type of information processing or thinking defect in schizophrenia seems to be the descriptive term disorganization of speech. We have lost not only our curiosity about how a psychotic patient thinks, but also our abilities to observe. This is a major unanticipated consequence of this most empirical of all diagnostic systems.

We now tend to study how a patient fits a diagnosis, or how groups of patients fit a diagnostic category, not psychopathology. Not only has this led to a boring, voluminous new literature, but it tends to force fit all patients into the diagnostic categories we have, rather than study the varieties of psychopathology. With our current polythetic system and our lack of attention to observational skills, it is possible that we are often comparing two different groups of patients in the same diagnostic category or similar groups in different diagnostic categories. While standardized diagnostic criteria are necessary for research, it is not clear that using polythetic criteria derived by expert opinion is methodologically sound. Two data-based studies in this issue (by Kendler and Gardner and Boris et al.) highlight the arbitrary nature and the varied results that can occur when different polythetic criteria are used.

Another problem that emerges when we rely exclusively on diagnostic categories, particularly with complex patients, is that it is difficult to monitor treatment. It is often clinically useful to monitor a few "target" symptoms, or patient functions, as a measure of treatment efficacy. If we monitor just diagnoses, then the patient either continues to have the diagnosis or not.

Most of us view with some enthusiasm this new diagnostic process as making our clinical practice more "medical" e.g., make the proper diagnosis and the treatment then follows. However, this increased medical approach may have unanticipated consequences. It puts us in the position of being able to "grind out" patients, which has the potential to make psychiatry boring and repetitive. Many clinics and university programs have developed productivity standards based on a psychiatrist seeing four patients an hour for medication management. This process plays into the hands of those who want to regulate practice, in that it reduces our expertise (and the complexity of the patient) to the DSM diagnosis and medication management. It is also possible that medical students, witnessing this assembly line approach to psychiatry, feel that as psychiatrists, they would just be grinding out patients and that if this is what the practice of psychiatry is about, then they might as well grind out patients in a more lucrative or more prestigious specialty.

DSM-III and IV have been major advances, but they provide only part of the information we need. The other part is the story of the patient or his or her narrative. Jaspers emphasized "empathy" as the key to understanding the patient (5). It is vital that we restore this aspect to our diagnostic process. How does the patient's experience resonate with our own? What is it like to be this person? A good clinician moves back and forth from detached observation to empathic probing. Only through such a process can we, for example, distinguish between a belief related to being a member of a strange sect and a mood-congruent delusion. This is different from learning psychotherapy. There are many psychotherapies; but this is how one goes about learning what each patient is experiencing and how this relates to his or her reported symptoms. A narrative story must emerge on how each individual copes and adjusts to his or her life, and from this, a hypothesis of understanding of the patient's problems should develop (other than that the patient has some "biochemical defect") (7, 8). The time has come to merge the empirical psychiatry of DSM-IV with the story and actual observation of the patient. Accurate observation and the story of the patient must be included in our diagnostic processes. All are necessary for the effective care of patients, which, in the long run, is what it is all about.

Address reprint requests to Dr. Tucker, Department of Psychiatry and Behavioral Sciences, University of Washington, 1959 Pacific St., Box 356560, Seattle, WA 98195. The author thanks Peter Roy-Byrne, M.D., Jurgen Unutzer, M.D., and Johan Verhulst, M.D., for their comments regarding this editorial, which is a revision of a lecture given at the Department of Psychiatry at Dartmouth Medical School on Oct. 21, 1996, on the occasion of its silver anniversary.