Can Validity and Pragmatism Go Hand in Hand? Yes? No? Sometimes?

I am pleased to be able to join the dialogue on science, psychiatric diagnosis, and the DSMs that Dr James Knoll initiated with the posts of Drs Ronald Pies, S. Nassir Ghaemi, and Theodore Pearlman in Psychiatric Times on October 14, 2013.

I come to this exchange via my recent book, The Making of DSM-III: A Diagnostic Manual’s Conquest of American Psychiatry, where I consider (1) The manual’s historical background (American psychiatry after World War II; the work of Emil Kraepelin, the master describer; and the neo-Kraepelinians at Washington University in St. Louis); (2) DSM-III’s prime movers (Robert Spitzer, the Editor, and his Task Force); (3) The construction of the manual and the development of many of its diagnoses. The history of DSM-III (1980) enables us to understand how we got to where we are today with DSM-5, which has aroused so much controversy.¹

I would like to begin by expressing my appreciation of many of the points enunciated by Dr Pies.² Pies’ article is a clear and carefully nuanced profession of the overdetermined-Freud’s useful term-nature of medical science and psychiatry. The issue at stake is the manner in which the physician goes about arriving at a diagnosis and whether certain frequently used processes produce a scientifically valid diagnosis. To this end, Pies asserts, quite correctly, that language matters. “Science” is a problematic word and can be used to mean different things. “Scientific process” is a more useful term, but even it does not always guarantee validity. The scientific process can produce what appears to be a valid diagnosis, but that diagnosis may turn out to be false. The scientific process then produces an alternate diagnosis that may remain valid or may be replaced in its turn. For psychiatrists, “diagnostic validity is always provisional [Pies’ emphasis] and probabilistic.”

Can a valid diagnosis be arrived at if laboratory tests are not available? Of course it can. Pies trenchantly cites the observations of Dr Kurt Kroenke who points out that usually the patient’s history contributes 75% of the information used in making a helpful diagnosis-lab tests generally furnish no more than 10%.³ Pies concludes that total reliance on lab tests to the exclusion of other factors when seeking validity is not science but “scientism.” Interestingly, Eli Robins and Samuel Guze, in their influential 1970 paper on diagnostic validity and its application to schizophrenia, came to their conclusions about “good prognosis” schizophrenia and “bad prognosis” schizophrenia relying mainly on a meta-analysis and family studies, with no laboratory information.⁴

Contrary to what Dr Ghaemi declares, not all revisions of the DSMs have been unscientific if one conceives of psychiatry as a “medical science” that relies on a multitude of factors in order to make a diagnosis. Here I must agree with Pies’ conclusion that psychiatry is not a natural science like physics.

Before I go any further, I want to mention that in a review of my book, Dr Ghaemi emphasizes my “neutrality” in interpreting the development of DSM-III.⁵ I bring this up only as background because at times in his October 14 post, he has made it seem that my work is at one with his critical views of how DSM-III came about.⁶ While there is much to criticize about the negative effects DSM-III has had on psychiatry-which I amply document in my book-I cannot accept his judgment that the manner in which DSM-III was constructed rendered it virtually worthless as a useful classification.

Let me draw on my work on DSM-III to make the argument that a valid diagnosis (ie, one helpful for prognosis and treatment) could be achieved by conscientious and careful investigation alone. This is not to say that the Task Force’s personal opinions never held sway nor that a member’s determined will to produce a diagnosis never occurred. (No classification has ever totally escaped the pressure of personal bias.) But one cannot conclude, as does Ghaemi, that all the work on DSM-III can be reduced to “political wheeling-and-dealing,” words I never use and a judgment I never make. Although the Task Force of DSM-III made many mistakes and created serious problems for the future, I have a healthy appreciation of their efforts to tackle knotty and unavoidable aspects of psychiatric diagnosis, often without laboratory authentication and decisive studies at their disposal.

Moreover, it is impossible to avoid the influences of non-scientific factors on scientific decisions, however desirable it would be that the former should not play a role. In 1973 the APA Board of Trustees at an emergency meeting mandated that DSM-II (1968) be speedily revised in order to show that psychiatry was a part of scientific medicine, as well as to combat the anti-psychiatry movement that was so prominent in the 1960s and ‘70s. The challenge of these two charges was so great as to defy any one verdict on the nature of the process whereby DSM-III was developed. To these two goals must be added the pressure on Robert Spitzer to produce a manual that would improve diagnostic reliability, which everyone acknowledged was abysmally low; on this score psychiatrists were often held in derision. Furthermore, the APA had to satisfy US treaty obligations to make its classification compatible with that of the WHO’s International Classification of Diseases (ICD.) DSM-III carried a weighty historical load of high expectations, difficult tasks, and lofty aims. History forces us to acknowledge the multiplicity of factors involved in the making of DSM-III.

I would like to discuss two diagnoses in DSM-III-Borderline Personality Disorder (BPD) and Schizoaffective Disorder, both having no “hard” evidence to determine their validity, but arrived at by a series of thoughtful and even dogged approaches.

At the time DSM-III was constructed, BPD was not in DSM-II and was considered a “soft” diagnosis by some researchers who were opposed to it. But if we look at its history since 1980, it is clear that it has become a widely recognized disorder and one for which there is new understanding and specific treatment. Spitzer had declared at the outset of the revision of the manual that he would not include a diagnosis in DSM-III unless clear and unequivocal descriptive diagnostic criteria could be developed for it. He was always cognizant that he served two constituencies, researchers as well as clinicians, and that he had to make the diagnostic criteria accurate and specific enough so the researchers could form reliably homogeneous disease groupings with which to conduct studies and then do follow-ups.

When a psychoanalyst at the Menninger Clinic suggested BPD be added to the new manual, Spitzer wrote back to ask whether he could supply specific descriptive criteria that would capture the concept of “borderline personality organization.” The APA Archives contain no response from the analyst, but ultimately Spitzer made the decision that it would be “useful” to include this disorder but that it should be used only if no other personality disorder could be diagnosed. One of the first matters Spitzer addressed was whether “borderline” should be used as a noun or an adjective. Did it describe a full-fledged disorder or did it merely mean a level of severity? Actually, he asked, could there be “borderline” anything? Another issue that Spitzer had to untangle was whether there were episodes of psychosis associated with BPD; the profession did not agree on this question.

Spitzer also wished to make BPD a clearer, firmer entity. To this end, he sought to elucidate two conceptions of “borderline” as put forth by researchers who worked in this area. To start with, he proposed that the “borderline” diagnosis most closely conceived of as “borderline schizophrenia” in DSM-II be called Schizotypal Personality Disorder in DSM-III. Spitzer was attempting to rid the nomenclature of excessive and unwarranted schizophrenia diagnoses. He was heavily influenced by a disquieting US/UK study which had found that both countries’ diagnosing patterns for schizophrenia and affective disorders were alarmingly out of kilter. In the US schizophrenia was over-diagnosed-there were 10 different schizophrenia diagnoses in DSM-II and so-called “latent schizophrenia” contained 4 subtypes. Affective disorders were under-diagnosed. The exact reverse was the pattern in the UK.⁷ In the US, the diagnosis of borderline schizophrenia did not call for a psychotic episode, which made Spitzer question the diagnosis.

Then Spitzer turned to another conception of “borderline,” this one envisaged as borderline personality “organization.” The notion was the creation of the psychiatrist and analyst Otto Kernberg (b. 1928), and Spitzer corresponded with him. It was Kernberg’s conception that Spitzer wanted to make into BPD, but with a different, more descriptive and meaningful name than “borderline.” In this pursuit, he totally failed. The “borderline” name was so deeply entrenched that other designations were soundly rejected.

Spitzer also sought clinical validation for his plan. He sent out a questionnaire to 4,000 APA members “to help firm up our criteria” for “borderline” individuals. The APA members’ answers to the questionnaire on rating these individuals yielded data on 808 patients and 808 controls that Spitzer and a colleague subjected to a variety of statistical measurements.⁸ To be sure, this was no double-blind study. But with the information he had, he concluded that it was possible to find diagnostic criteria to describe both BPD and Schizotypal PD and that there was a measureable difference between the two personality disorders.

Yet Spitzer did not rest. He was determined to answer the question of whether psychosis was a part of BPD, and for help with this he turned to John G. Gunderson, a psychiatrist at McLean who had established a name for himself with a 1975 article, together with Margaret T. Singer. They had reviewed the literature on borderline patients and proposed 6 features that could be used to diagnose borderline patients in the initial interview.⁹ Still on the trail for accurate diagnostic criteria for BPD, Spitzer wanted to know if he should include “brief psychotic experiences.” He asked if Gunderson would please “provide several specific descriptions of such experiences.” The level of detail Spitzer went on to ask for is striking: samples of 6 to 10 such incidents, information on whether the experiences were “limited to therapy or transference distortions, whether delusions or hallucinations were involved, and how long such episodes lasted.”

While it is certainly true that psychoanalysts found personality disorders useful to their interests in character pathology-BPD was first enunciated by an analyst in 1938¹⁰-it is not true, as Dr Ghaemi states, that I “document well” that personality disorders “were almost completely based on psychoanalytic opinion, with hardly any scientific validity literature to support them.” I could not conclude this because there is nothing in the Archives to indicate this. Spitzer and the DSM-III Task Force hardly discussed most of the Personality Disorders, and Spitzer’s meticulous efforts to define BPD do not support Ghaemi’s assertions.

Furthermore, of the 10 members of the DSM-III Personality Disorders Advisory Committee, at least 5 were openly antagonistic to psychoanalysis, and one scathingly so. Aiding the analysts was the last thing on their minds. Also, one member of the committee, Allen Frances, was very dubious about BPD and published a number of papers with the personality disorders expert Thomas Widiger that reflected this. And Spitzer, a research psychiatrist who had left his psychoanalytic training far behind, dumped 4 PDs and then added 5 new ones. There is no evidence that he did this to please the analysts. And further, whenever Spitzer showed signs of meeting some of the psychoanalysts’ concerns, the Task Force always overruled him. (These episodes are in the Archives.)

The second diagnosis I want to discuss is Schizoaffective Disorder. We can all agree that the delineation of this disorder has not primarily come about through rigorous testing, but because it is pragmatically needed. Someday, when the APA’s classification shifts to one based on etiology, perhaps it will disappear. But in the meantime, we need something when the formal division of schizophrenia from bipolar breaks down. (And even Kraepelin wondered towards the end of his life if his division of dementia praecox and manic-depressive illness was sound.) Modern neuroscience has already provided us with tantalizing indications that aspects of the two are closely linked.

Although Spitzer and the Task Force could not construct a hard case for the existence of this disorder, they did the next best thing: debate-nay, wrestle-with the thorny question of whether the diagnosis belonged under Schizophrenia or Affective Disorders and what its diagnostic criteria should be. In this short article there is no possibility of recapitulating the lengthy and informed discussions that went to the heart of diagnostic issues in psychiatry. Readers interested in examining this issue further can consult my book.

In DSM–II, Schizoaffective Disorder had been placed under Schizophrenia and indeed, in the first draft of DSM-III (April 1977) that is where it remained, with 2 sub-types-Depressed and Manic, and diagnostic criteria for each. But in the next draft (January 1978) it was placed in its own separate category with just 1 list of diagnostic criteria, but with an elaborate sub-typing of 9 forms, this latter intricacy being a clear indication of a struggle to describe the disorder. There it did not rest either because William T. Carpenter, on the Multiaxial Diagnosis Advisory Committee, (in 2013 on the Task Force of DSM-5 and also on its Psychosis Work Group) proposed that the diagnosis be situated under “Psychosis Not Elsewhere Classified” along with Schizophreniform Disorder, Brief Reactive Psychosis, and Atypical Psychosis.

The result was the creation of an increasingly heated debate over what had appeared shortly before to be a settled question. Highly intense discussions went on for several months with extremely specific and sophisticated confrontation of the issues involved in further describing the disorder. There were urgent memos and meetings, but a resolution did not emerge. One result though, of the robust examination of where Schizoaffective Disorder belonged and how to describe it, was the decision to strengthen the criteria for Major Depressive Episode, Manic Episode, and Schizophrenia, and this was done.

Ultimately the decision was made to retain Carpenter’s proposal for placement of the disorder but to abandon sub-typing it or developing diagnostic criteria for it. In earlier drafts of DSM-III the description of Schizoaffective Disorder had occupied several pages. It now appeared on a mere ¾ of a page without diagnostic criteria, in spite of Spitzer’s vow never to allow that to happen. The manual contained the statement that future research was needed on whether there is a need for the category, if so how it should be defined, and what is its relationship to Schizophrenia and Affective Disorders. Spitzer dutifully informed the APA’s Council on Research and Development, to which he reported, about the decision and eventually received word that the Council was in agreement with the general approach that had been taken.

In DSM-IV the disorder was categorized under Schizophrenia and Other Psychotic Disorders with a lengthy description and diagnostic criteria. It appears in DSM-5 under Schizophrenia Spectrum and Other Psychotic Disorders with more specific criteria than in DSM-IV in the hope this will increase reliability and reduce the overutilization of the diagnosis. Perhaps it can be said that the history of the placement and description of Schizoaffective Disorder illustrates usefully some of the challenges involved in constructing a psychiatric classification and the approaches that may be involved if lab tests and other studies are not available. It can even be argued that Kraepelin himself would have appreciated the dilemma of the makers of the DSMs.

BPD and Schizoaffective Disorder entered DSM-III not on the basis of laboratory evidence or double-blind studies, but with thoughtful and determined efforts to meet the medical needs of researchers and clinicians and to revise the APA’s classification system in order to achieve specific goals. The Task Force and Advisory Committees members took their charge very seriously, belying a common criticism that DSM-III was created by a group of people who insouciantly tossed around opinions. One may agree or disagree with the DSM-III resolutions of the BPD and Schizoaffective Disorder quandaries. But one cannot easily attribute the eventual outcomes to wheeling-and-dealing. And as Dr Pearlman has pointed out, the DSM Task Forces were “spurred towards seeking a truthful classification of psychiatric illnesses for . . . research and treatment,” mindful that the endpoint for their diagnoses was the succor of human beings with “a level of malfunction causing clinically significant distress or impairment in social, occupational or other important areas of functioning.”¹¹

Far more important as to whether all DSM-III diagnoses were arrived through a “scientific” method was the fact that once constructed they became reified as sharply delineated categories. The makers of DSM-5 were able to partially break down categorization in 2 instances: Autism Spectrum Disorder and Schizophrenia Spectrum and Other Psychotic Disorders. However, the attempt to conceptualize certain aspects of Personality Disorder in dimensional terms ultimately failed. Here I agree most strongly with Dr Ghaemi’s criticism of the APA’s Board of Trustees for rejecting the Personality Disorders Work Group’s recommendation on this score. As all clinicians know, personality disorders run the gamut from mild to severe, and assessing where a patient falls on the continuum helps in prognosis and treatment. Yes, it is time-consuming for the clinician to make a dimensional diagnosis, but such a diagnosis is more valid, a reflection of real life circumstances. On the other hand, dimensional diagnoses may decrease reliability. No one ever said it is easy to be a psychiatrist.

I appreciate having the opportunity to add my remarks to this very focused discussion of psychiatric diagnoses and the DSMs, currently a topic of great controversy.

Acknowledgement: I want to thank Dr Norman Decker for his close reading of this paper. However, the views expressed here are my own.

Disclosures:

Dr Decker is a Professor of History at the University of Houston and an Adjunct Professor of Medical History in the Menninger Department of Psychiatry, Baylor College of Medicine, Houston. She has written widely in the histories of psychiatry and psychoanalysis. In 2007 she received the Eric T. Carlson Award for “Extraordinary Contributions to the History of Psychiatry and Psychoanalysis” from the DeWitt Wallace Institute for the History of Psychiatry, Weill Cornell Medical College, New York City.

References:

1. Decker HS. The Making of DSM-III: A Diagnostic Manual’s Conquest of American Psychiatry. New York: Oxford University Press, 2013.
2. Pies RW. Science, Psychiatry, and Family Practice: Positivism vs. Pluralism. Psychiatric Times. October 14, 2013. http://www.psychiatrictimes.com/articles/science-psychiatry-and-family-practice-positivism-vs-pluralism. Accessed October 31, 2013.
3. Kroenke K. Diagnostic testing and the illusory reassurances of normal results. JAMA Internal Medicine. 2013;173:416-417.
4. Robins E. Guze SB. Establishment of diagnostic validity in psychiatric illness: its application to schizophrenia. Am J Psych. 1970;126:983-987.
5. Ghaemi N. Book Review, The Making of DSM-III: A Diagnostic Manual’s Conquest of American Psychiatry. Acta Psychiatrica Scandinavica. In press.
6. Ghaemi N. Why DSM-III, IV, and 5 are unscientific. Psychiatric Times. October 14, 2013. http://www.psychiatrictimes.com/blogs/couch-crisis/why-dsm-iii-iv-and-5-are-unscientific. Accessed October 31, 2013.
7. Cooper JE, Kendell RE, Gurland BJ, et al. Psychiatric diagnosis in New York and London. Maudsley Monograph Series, No. 20. London: Oxford University Press, 1972. Also see Kendell RE, Cooper JE, Gourlay AJ, et al. Diagnostic criteria of American and British psychiatrists. Arch General Psych. 1971;25:123-130; Kendell RE. Psychiatric diagnosis in Britain and the United States. Brit J Psych. 1975,9:453-461.
8. Spitzer RL, Endicott J. Justification for separating schizotypal and borderline personality disorders. Schizophrenia Bull. 1979;5:95-104.
Also see Spitzer RL, Endicott J, Gibbon M. Crossing the border into borderline personality and borderline schizophrenia: the development of criteria. Arch General Psych. 1979;36:17-24.
9. Gunderson JG , Singer MT. Defining borderline patients: an overview. Am J Psych. 1975;132:1-10.
10. Stern A. Psychoanalytic investigation of therapy in the borderline group of neuroses. Psychoanalytic Quarterly. 1938,7:467-489.
11. Pearlman T. In defense of DSM. Psychiatric Times. October 14, 2013. http://www.psychiatrictimes.com/blogs/couch-crisis/why-dsm-iii-iv-and-5-are-unscientific. Accessed October, 31, 2013.