Why DSM is mostly false

95.2% false, to be exact

Introduction

A decade ago, the fifth revision of DSM (the  Diagnostic and Statistical Manual) was published by the American Psychiatric Association. This diagnostic system has been called the “Bible” of psychiatry.  The metaphor suggests some cultural realities.  It tends to be worshipped; some view it as the literal truth; it can inspire, but it can be used to suppress dissent. The problem is that DSM can’t be a Bible and science at the same time.

In this article, I examine what the DSM system is and what it isn’t. The brief summary is that it has become more like a religion, and less and less scientific.  (Much of the source evidence for the details of the DSM-III history can be found in this book by Hannah Decker.)

History

The origins of DSM go back to the first edition in 1952, and the second edition in 1968, but those earlier versions had little impact on the US profession, much less the world.  This lack of impact mainly had to do with the fact that American psychiatry mostly was psychoanalytic in orientation.  In psychoanalysis, little importance was given to diagnosis, unlike the medical tradition.  Diagnoses were “labels,” mere shorthand categories for communication. They weren't “real” and they didn’t represent “diseases.”  All psychopathology was about unconscious emotional conflicts, mainly dating to childhood; if the conflicts were normal or mild, they produced “neuroses”; if they were severe, they produced “psychoses.” That was the extent of psychoanalytic nosology. 

The American Psychiatric Association (APA) organized the first two editions of DSM mainly for administrative purposes.  Those who ran mental hospitals needed to label the reasons patients were treated.  Since psychoanalytic theory mostly ignored diagnostic labels, DSM terms often were taken from the alternative medical approach to psychiatry,  popular in parts of Europe, and associated especially with the research of Emil Kraepelin (circa 1900), and other German and French psychiatrists dating back to Philippe Pinel (circa 1800). 

Kraepelin’s tradition

The French nosologists of the 19th century had produced a descriptive tradition with hundreds of diagnoses.  Kraepelin had boiled down those myriad diagnoses to a dozen or two.  The three most common psychiatric diagnoses in state hospitals, according to Kraepelin, were general paralysis of the insane (GPI, which soon was found to be neurosyphilis), dementia praecox (soon renamed schizophrenia), and manic-depressive insanity (MDI).  He also identified a fourth diagnosis of older age dementia. 

Kraepelin’s influence was strong for a while because he helped make sense of a mass of patients with similar symptoms. GPI, schizophrenia, and MDI were more or less indistinguishable in symptoms.  Most patients were psychotic at some point; many had mood episodes sometimes; some got better; some didn’t.  

Kraepelin’s method was simple:  Ignore the myriad symptoms and focus on the course of illness.  At what age did the symptoms start? How long did they last? Did they persist or go away?

Age of onset.  Duration of episodes. Chronicity versus episodicity.  Outcome. These are the hallmarks of Kraepelinian diagnosis. 

One disease began in adolescence, was chronic, and its eventual outcome was poor in most cases.  That’s dementia praecox. Another disease began in adolescence, was episodic, and its eventual outcome was good in most cases.  That’s MDI. Another disease began in middle age, was episodic, and its eventual outcome was poor.  That’s GPI. A final disease began in old age, was chronic, and its eventual outcome was poor in most cases.  That’s dementia of old age.

All the above patients were studied carefully in  brain anatomy by faculty Kraepelin had hired for the departments of psychiatry at Heidelberg and in Munich, where he was chairman.  Among those faculty  were Franz Nissl (father of the “Nissl stain”) and Alois Alzheimer.  In the first two diseases, abnormalities of the brain couldn't be identified.  In the third (GPI), brain abnormalities of a diffuse kind were found. In the fourth, specific plaques and tangles were found, with the condition being named after Alzheimer.  

Kraepelin had one major weakness:  He didn’t have any special treatment to offer for his carefully defined diagnoses. He could only give a prognosis - the patient would improve or not - but he couldn't treat any disease.  He was a “therapeutic nihilist.” 

After Kraepelin

The new psychoanalytic movement, in contrast, proposed to be able to treat almost ever everything and everyone - from the sickest psychotic to the healthiest neurotic. Psychoanalysis was therapeutically optimistic.  

These promises appealed greatly to the psychiatric profession in the early to mid 20th century, especially in the optimistic culture of the United States.

Kraepelin’s influence was strongest in his life, from the 1880s to the 1920s.  When he passed away in 1926, his influence began to wane quickly as Freud’s star rose.  By the 1950s and 1960s, when DSM terms were used, Kraepelin’s phrases were mixed with Freud’s phrases, but it was the Freudian theory that was seen as important. 

A change began to happen in the 1960s and 1970s, though, with the introduction of new psychotropic medications.  The “antidepressants” and “antipsychotics” and lithium seemed to have major effects, far beyond what had been available in the era of Freud and Kraepelin.  In other words, by the 1970s, there were alternative treatments, which seemed effective in practice, and those treatments seemed to fall in line with Kraepelin’s basic diagnostic system of schizophrenia and MDI.  

A rebel group within American psychiatry, headquartered at the Washington University in St. Louis (headed by its chair Eli Robins), began to promote and study psychiatric diagnosis using Kraepelin’s method of focusing on course of illness, along with using new genetic methods.  These psychiatrists were called “neo-Kraepelinian.” They prepared a summary of diagnoses that had valid research evidence: there were 14 such diagnoses, summarized in their classic paper on Research Diagnostic Criteria (RDC).  

At the same time, clinicians were looking for new ways of thinking about diagnoses in their use of the new drugs.  The APA planned a 3rd revision of DSM, and it chose a New York psychiatric researcher to head it, Robert Spitzer. New York was the mecca of world psychoanalysis, and Spitzer had been steeped in it, but was also sympathetic to neo-Kraepelinian ideas. He decided to invite the Washington University crowd to help prepare DSM-III.  The APA leadership, headed by psychoanalytic leaders, didn’t care, because, following Freudian ideology, it didn’t think diagnosis mattered. 

DSM-III: 95.2% false

By the late 1970s, the APA leadership realized that the DSM-III task force was moving from Freud back to Kraepelin’s methods. The majority of American psychiatrists were Freudian in orientation.  The DSM-III changes needed to be approved by a vote of the APA General Assembly, and by the APA Board of Trustees, all overwhelmingly controlled by Freudian clinicians. 

In the final year or two of the DSM-III process, before its ratification by the APA in 1980, Spitzer negotiated hundreds of changes to make the final document acceptable to the Freudian rank-and-file and leadership of the APA. 

In addition to 14 validated diagnoses published in the RDC in 1978, a mere two years later DSM-III came out with 292 claimed diagnoses. There is no metaphysical possibility that 278 psychiatric diagnoses suddenly were discovered in two years. They were invented. This mere fact proves that DSM-III was mostly false - 95.2% false, to be exact. 278 diagnoses were made up in the final years of the process to satisfy the profession; they weren’t true, or at least proven true, and everyone knew it.

By that time, most of the Washington University neo-Kraepelinian members of the task force were disgusted and dropped out of the process. One of them even committed suicide. 

In the end, DSM-III was seen as a radical change from its predecessors, and it was in some ways.  It defined schizophrenia as Kraepelin had, based on the course of illness being chronic and worse, as opposed to the Freudian definition which ignored outcome.  It introduced the term “bipolar” and “major depressive disorder”, which was not how Kraepelin thought of it, but which was derived from the MDI concept. It introduced many other categories which weren't part of the psychoanalytic jargon but were purely descriptive.  It also introduced many categories that were psychoanalytic in concept - such as the dozen “personality disorders.”

In short, DSM-III began as a scientific research project, along the lines of Kraepelin’s tradition, and it ended up becoming a compromise with psychoanalytic theory and with the beliefs and labels used by clinicians in practice. 

“Pragmatism”

In other words, from the very beginning, science and scientific research was not the primary motivator or method of DSM-III.  It had been the ideal for the neo-Kraepelinian group at Washington University, but they were soon excluded and marginalized in the DSM-III process by Spitzer and the Freudian majority of the APA.  Spitzer was explicit about the process:  science couldn't be the main method partly because there was much that wasn't known or studied scientifically, and partly because DSM-III was meant to be useful to clinicians, so it should reflect clinicians’ beliefs. 

The head of the next revision, DSM-IV, has called this process “pragmatism.” This is the main method of DSM revisions.  What is best for the profession?  This decision is made by the APA leaders and by the activists of the APA.  Sociologists call this kind of activity a “social construction.” It’s a cultural creation based on social, economic, and political currents.  It’s not science, by the usual definition of hypotheses and experiments and objectivity.  

The DSM-III process was justified by Spitzer and his supporters as representing progress but being limited by its era. We can’t make a completely scientific document, the DSM-III leaders said, but we can at least take our best science now and combine it with a consensus of the clinical community.  We’ll achieve “reliability”:  everyone will agree on how we define our labels, like a dictionary.  Then with future research, we can change those definitions to make them closer and closer to reality: “validity”.  Reliability was supposed to lead to validity eventually.

DSM-IV

This is what happened with DSM-IV, in 1994, under the leadership of Allen Frances: The whole project became false at its core.

The dictionary became an object of worship, transformed into a Bible. DSM-IV leaders made it explicit that changes should be as few and as minor as possible. The radical changes of  DSM-III could barely be altered. Some small changes were made based on research, but larger changes were rejected on “pragmatic” grounds:  DSM definitions were fine as they were, meeting the needs of clinicians. 

It went unrecognized that the promise that reliability would lead to validity had been thrown aside.  What mattered wasn’t what was true, based on scientific research, but what was useful, based on the pragmatic beliefs of DSM-IV leaders and APA activists. 

DSM-5

14 years elapsed between the third and fourth revisions of DSM.  19 years would pass before the 5th revision was published in 2013.  A whole generation. Psychiatry spent two generations, almost 40 years, with the basic structure and content of DSM-III in 1980.  The pragmatic ideology had prevented major changes in 1994, and the same approach became the rationale for little to no change in 2013.  

Initially, there was interest in some major changes in DSM-5. In particular, over half a century of personality research had supported the concept of personality “traits” or dimensions, rather than “disorders” or categories. The DSM-5 personality disorders task force carefully documented and supported that large amount of scientific evidence.  It recommended dropping the long-used psychoanalytically-based “personality disorder” categories, and replacing them with personality traits (like neuroticism, extraversion, and openness to experience, among others). The DSM-5 leaders and scientific review committee approved this radical change. In the final weeks of the DSM-5 process, before the vote of the APA general assembly, the Board of Trustees of the APA vetoed the explicit recommendations of the personality task force, and kept the personality disorders categories mostly unchanged from the initial definitions of 1980. Traits and dimensions were placed in “Section 3”, which represents alternative definitions that aren’t part of the official categories of DSM-5. 

In other words, pragmatism again took precedence over science. Though personality is the clearest example, there are other examples. Sometimes, a notable amount of scientific research was deemed inadequate, on “pragmatic” grounds, to make small changes (such as the duration of hypomania; it was feared bipolar illness would be overdiagnosed). Other times, despite less scientific research than the above example, major changes were made because of pragmatic beliefs (such as the addition of “disruptive mood dysregulation disorder” in kids, again aimed at discouraging bipolar diagnosis).  

To their credit, DSM-5 leaders took some steps in the direction of accepting science as their main criterion. One example is the removal of the antidepressant exclusion for the diagnosis of acute mania or hypomania.  Previously, based solely on the “pragmatic” preferences of DSM-IV leaders, bipolar disorder couldn’t be diagnosed if manic/hypomanic episodes happened solely in the presence of antidepressant treatment.  Research shows that this observation is common in bipolar disorder (10-50% of cases) but very rare in major depressive disorder (MDD, <1% of cases). In other words, antidepressant-induced mania happens almost exclusively in people who have bipolar illness.   

Other examples of mostly science-based changes included the acceptance of an autism spectrum, and the reconceptualization of mixed states as  occurring in both MDD and bipolar disorder. Other examples of mostly “pragmatic” changes include: the broadening of the ADHD diagnosis to extend the age of onset to 13 and to define it into adulthood; the rejection of a bipolar spectrum; ignoring the concept of affective temperaments; and the rejection of the concept of prodromal schizophrenia. 

Controversy

While these DSM-5 changes and non-changes were developing, many commented from the sidelines.  Much attention was given to the leaders of DSM-III and IV, Spitzer and Frances, who often criticized DSM-5 leaders no matter what the latter did.  If they changed something, like the antidepressant-induced mania exclusion, the prior DSM leaders would criticize the change on “pragmatic” grounds.  Oh no, they’d say, now everyone will get diagnosed bipolar. In the case of this issue, DSM-5 leaders made a change despite the pragmatic criticism. 

In many other cases, DSM-5 and/or APA leaders caved to the external criticism led by the “pragmatic” crowd: most importantly, the personality trait/dimension concept was vetoed at the last second; the prodromal schizophrenia concept was dropped despite extensive scientific evidence in its favor; the ADD concept was broadened; the “disruptive mood dysregulation” concept was invented to discourage bipolar diagnosis.  Ideas like the bipolar spectrum concept were never even briefly considered in any serious way. 

In short, DSM-5 was mostly pragmatic, not scientific, like its predecessors. 

An obstacle to knowledge

Where does this leave us?  We have to accept DSM-5 definitions from a legal and practical perspective. We have to use them for insurance forms, and to protect ourselves against lawsuits.  But we don't have to believe in them.

The same week as the official publication of DSM-5 in May 2013, the head of the NIMH, Thomas Insel, announced that DSM diagnoses would no longer be used for scientific research funded by the NIMH. Despite political pressure and a later joint press release from the NIMH and APA, Insel had told the truth: Science isn’t the primary basis for DSM definitions, and hence it isn’t useful for scientific research. 

If we want to know what genes cause a disease, or the biological processes of an illness, we will not find such genes or biological markers using DSM diagnoses.  This statement is based on 40 years of failed research in psychiatry, but is also based on simple logic:

Why would nature structure its genes and biology based on the “pragmatic” preferences of DSM leaders or APA committees?

In fact, DSM definitions have been an obstacle to scientific progress, because, since they aren't based on our best science, they will steer scientific studies in the wrong direction, with both false positive and false negative findings in genetics, biology, and pharmacology. That's why psychiatry has become  a profession waiting for a miracle that never happens.  

The admission of the NIMH has undercut the scientific legitimacy of the DSM system, and it leaves clinicians in a quandary: Can we use a diagnostic system that isn’t scientifically-based and scientifically-sound?  

If DSM isn’t good enough for research, why should we accept it as good enough for practice?

A Solution?

There is a simple solution: Focus on research - ONLY. Where DSM-5 agrees with the scientific evidence, accept DSM-5 definitions. Where DSM-5 disagrees with our best scientific evidence, so much the worse for DSM-5.

We have proposed a new Clinical Research Diagnostic Criteria (CRDC) for the 21st century. This approach would replace DSM invalid diagnoses with only those diagnoses that have some validity, and with criteria based on validity research, not professional opinion.

We can accept the DSM system as a dictionary, but not as a Bible. There are no Bibles in science, only hypotheses, which we need to be willing to refute, not just accept.  

Even as a dictionary, though, DSM-5 is weak. Its field trials found poor reliability of major categories, like MDD.  (The latter had a very low “kappa” value of 0.25, which is worse than prior field trials in DSM-IV and DSM-III).  In other words we have given up on validity, and explicitly placed science below “pragmatism”, and even so, the same DSM definitions invented in 1980, largely unchanged (as with MDD) after nearly four decades of use, aren’t even as reliable as in 1980. 

The DSM-5 leadership sometimes boasts about a million copies sold in a year, and the fact that the DSM system is used all over the world. This is the case: a large part of the APA budget is based on sales of DSM.  And the world has taken DSM at face value, as if it’s the truth. But clinicians should know, all over the world, that the DSM system is a reflection of the “pragmatic” beliefs of mainstream American psychiatry. DSM is not, first and foremost, a reflection of our best science.

Clinicians would then be free, with this recognition, to accept or reject or modify DSM definitions based on the best research and based on their own experience and their own best judgment.

In each patient case, state the DSM-5 diagnosis. And then state your best provisional clinical diagnosis, based on your own clinical judgment and scientific research of which you might be aware. In other words, make two diagnoses in each case:  what the mainstream profession tells you (DSM-5) and what the scientific literature and your own clinical judgment tell you. Don’t let the former preclude or define the latter.  Use the former for legal and administrative purposes. Use the latter to provide the best clinical care.    

That’s what surgeons do. That’s what cardiologists do. They only use the ICD diagnosis at the end of their appointment to get billing payment. They never even think about it as they are diagnosing and treating their patients. Psychiatrists, as usual, stand apart from the rest of medicine with their unscientific, mostly false DSM system of diagnosis and treatment.

The Bottom Line

• DSM is a social construction, based on “pragmatism” much more so than science. 

• Use DSM administratively, not for best clinical practice. 

(Revised from an original version in the printed Psychiatry Letter, May 2016)