The Washington University Approach to Psychopathology

As the American Psychiatric Association moves toward the next edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM), questions concerning psychiatric diagnosis and classification will once again come to the forefront. Surprisingly, relatively little has been written for general audiences about one of the most important developments in the history of psychiatric diagnosis: the Washington University approach to psychopathology.

Developed at Washington University in St. Louis during the 1950s, 1960s, and 1970s, this approach sought to establish psychiatry firmly as a branch of medicine by grounding psychiatric diagnosis in empirical research rather than theoretical speculation or professional consensus. Its influence was enormous, although many of its central principles were eventually neglected.

I wish here to briefly describe this approach, which informs my recent proposal with psychiatrist Ronald W. Pies for a validity-based diagnostic hierarchy in DSM-6, forthcoming in Psychiatric Services.

The Historical Background

In the mid-20th century, American psychiatry was dominated by psychoanalysis, which generally placed little emphasis on formal diagnosis. With some notable exceptions, such as Roy R. Grinker Sr. and Otto Kernberg, psychoanalysts tended to classify patients broadly as either neurotic or psychotic, while the question of whether particular syndromes represented distinct diseases or conditions received less attention.

Psychiatry also faced criticism over diagnostic unreliability. David Rosenhan's influential 1973 study, although later discredited (Cahalan, 2019), fueled public skepticism. More compellingly, the U.S.–U.K. Diagnostic Project found that American psychiatrists diagnosed schizophrenia at roughly twice the rate of their British counterparts (Kendell et al., 1971), largely because of differences in diagnostic practice. Psychiatry needed more reliable diagnoses, but reliability alone would not establish whether those diagnoses represented actual conditions.

It was against this backdrop that a group of psychiatrists at Washington University in St. Louis, most notably Eli Robins, Samuel Guze, and George Winokur, developed a fundamentally different approach to psychiatric diagnosis.

Unlike other branches of medicine, psychiatry generally lacks laboratory tests capable of establishing specific diagnoses. The 19th-century anatomical-clinical method linked symptoms to identifiable pathological changes, transforming much of medicine. In psychiatry, however, this method produced relatively few successes, apart from conditions such as neurosyphilis.

The German psychiatrist Emil Kraepelin recognized that symptoms alone could not provide an adequate basis for classification. Many medical diseases produce fever, cough, fatigue, or pain; no one would seriously propose classifying diseases solely on the basis of such symptoms. The same principle applies to psychiatry: Similar symptoms may arise from fundamentally different diseases, while the same disease may present differently across patients or stages of illness. This is a point overlooked by contemporary dimensional and network approaches to psychopathology, which tend to treat symptoms as the primary data of psychiatry.

Kraepelin emphasized another source of evidence: the longitudinal course and outcome of illness. Dementia praecox (later renamed schizophrenia), manic-depressive insanity, and Alzheimer's disease could all involve psychotic symptoms, but their characteristic courses differed. Dementia praecox was chronic or deteriorating; manic-depressive illness was episodic and relapsing; Alzheimer's disease involved progressive cognitive deterioration, ultimately ending in death. By considering the course and outcome of illness, in addition to symptoms, Kraepelin discovered three diseases.

This basic insight remains essential to psychiatric classification. As Goodwin and Guze (1989) later emphasized, "Diagnosis is prognosis." A valid diagnosis should convey something meaningful about the nature and expected course of an illness, not merely describe its presenting symptoms.

The Washington University Approach

In 1970, Robins and Guze published their landmark paper, "Establishment of Diagnostic Validity in Psychiatric Illness: Its Application to Schizophrenia." They proposed that diagnoses be validated through convergence across multiple independent lines of evidence.

Their framework identified five phases: clinical description, laboratory studies, delimitation from other disorders, follow-up studies, and family studies. Clinical description establishes a syndrome's characteristic presentation; laboratory studies seek distinguishing findings; delimitation examines its boundaries with other conditions; follow-up studies assess course and outcome; and family studies investigate genetics and patterns of familial aggregation. Later researchers added other validators, including differential treatment response.

A crucial distinction here is between reliability and validity. A group of clinicians may agree on the criteria for a disorder without establishing that it represents a distinct disease. Nor does the frequent co-occurrence of symptoms establish an independent disease. Diagnostic validity instead depends on the convergence of evidence showing that a condition can be meaningfully distinguished from others.

No single validator is necessarily decisive. Two diseases may share symptoms, genetic markers, or treatment responses while differing in course and other characteristics. Conversely, differences in symptoms do not necessarily establish separate diseases. The objective is to identify distinguishable forms of psychopathology, not simply to achieve diagnostic agreement.

From WashU to DSM-III

The Washington University school influenced Robert Spitzer, who led the development of DSM-III. The Feighner criteria (Feighner et al., 1972) and the subsequent Research Diagnostic Criteria (Spitzer et al., 1978) represented major advances in psychiatric classification, providing explicit criteria that different clinicians could apply consistently. Both projects substantially improved diagnostic reliability. In some ways, this was a major advance: clinical communication and research require agreement about what is being studied. A problem arises, however, when reliably defined categories are mistaken for validated diseases.

DSM-III, published in 1980, built on these advances but prioritized operational definitions and diagnostic reliability over the rigorous validation envisioned by Robins and Guze. Political compromises further shaped the classification, prompting several Washington University psychiatrists to resign from the DSM-III Task Force (Decker, 2013). Although DSM-III substantially improved psychiatric communication and research, it also introduced numerous diagnostic categories that lacked comparable evidence of validity.

How Many Psychiatric Conditions Are There?

The Feighner criteria paper initially described 14 psychiatric conditions, and the Research Diagnostic Criteria expanded this number to 25. In their 1989 textbook Psychiatric Diagnosis, Goodwin and Guze discussed roughly a dozen major syndromes for which they considered the validating evidence strong. Yet DSM-III had already included 265 diagnostic categories.

Applying strict standards of diagnostic validity would require reconsidering many familiar DSM categories. Although a detailed examination is beyond the scope of this article, numerous widely accepted psychiatric diagnoses—including ADHD, most personality disorders, the dissociative disorders, and several depressive and anxiety disorders—would fail to meet conventional standards for validity as independent disease entities (Ghaemi et al., 2022).

Proliferating categories also create the appearance of extensive psychiatric comorbidity. A patient with one underlying condition may acquire several overlapping diagnoses, each describing a different manifestation of the same psychopathology. A classification intended to identify diseases can thus become an inventory of symptoms and syndromes without adequately establishing their distinctiveness.

A Proposal for DSM-6

In our forthcoming paper in Psychiatric Services, Pies and I propose a validity-based diagnostic hierarchy for DSM-6 (Ruffalo & Pies, in press). Diagnoses would be organized according to the strength of evidence supporting them, distinguishing well-validated syndromes from provisional or less-established constructs. Clinicians could continue to describe symptoms and clinically useful patterns without treating every descriptive category as an equally valid disease entity.

This does not mean that dimensional measurement has no place in psychiatry. Dimensions can be useful for assessing severity and describing variations within and across psychiatric syndromes. But they cannot, by themselves, establish whether particular diagnoses represent distinct conditions. That requires the kind of systematic investigation and convergence of evidence that Guze (1970) described as a commitment to "tough-mindedness."

As psychiatry prepares for DSM-6, it would do well to recover this approach. The central task of psychiatric classification is not simply to improve diagnostic agreement or develop increasingly elaborate descriptions of psychopathology, but to determine which conditions actually exist and how they differ from one another. This was the central concern of the Washington University group more than half a century ago, and it remains no less important today.

References