If the academic career has its heroic obligations, as William James once said, these must include occasionally defending the truth, despite personal risks and without personal rancor.
It is in this spirit that I comment on the debates around DSM-5 involving Allen Frances, MD, whose views seem to elicit sympathy from many unhappy with the DSM system (the 4th edition of which Dr Frances led). In the course of some exchanges with him in writing and on blogs, it seems to me that this is his guiding idea: “Pragmatism” should be the central basis of DSM-5. Or, in his own words, “To my mind, by far the most important [diagnostic] validator is how will any decision help or harm patient care, given the foreseeable circumstances under which it will be used.”1
At one level, this is not unreasonable: there are important practical aspects to DSM. It is used not just by doctors to treat patients, but—unfortunately—by lawyers to sue doctors; by patients to diagnose themselves; by academics to teach students . . . in short, by the whole culture. Long ago, some argued that we should have 2 handbooks: one you might call the “real” DSM, which would have our best science and our most advanced knowledge and would be used for research and practice. The other you might call the “practical” DSM, which would be how we would define things so that lawyers would not bring frivolous lawsuits and students would not misunderstand, and so on. Unfortunately, our one DSM tries to be all things to everyone, and we are left with having to prioritize.
Which matters most: pragmatism or science?
Dr Frances’ answer: common sense. But one might ask, whose common sense? First and foremost, Dr Frances says, “Do no harm.” If expanding a diagnosis, for example, bipolar disorder type II, would lead to use of more harmful drugs, such as neuroleptics, then don’t expand it. If adding a new diagnosis, for example, sexual disorders, would lead to more lawsuits, then don’t add it. Now this makes some sense, especially when one thinks of curbing lawyers. This approach would be fine if there was no truth to the matter. The problem is simple and serious: this is psychiatric gerrymandering. We have one person, or a small committee of self-appointed überexperts in DSM, overseeing the rest of the world’s experts in the science of each diagnosis, nipping and tucking proposed criteria so that our diagnostic experts will be able to feel that good outweighs harm as of 1994 (for DSM-IV). Even if one accepts this approach, one problem is that the balance of good versus harm in 1994 is not the same as in 2004 or 2014.
Look how internal medicine handles it: they have plenty of lawsuits, profit-seeking pharmaceutical companies, and lots of dangerous drugs. But they don’t make their diagnostic definitions reliant primarily on “how will any decision help or harm patient care, given the foreseeable circumstances under which it will be used.” Here are 2 parallel timelines.
The first time line, lasting 56 years:
• 1892: DSM-I for internal medicine, The Principles and Practice of Medicine, by William Osler. The causes of pneumonia are unknown, the treatments ineffective, the outcome mortal; but Osler describes the many features of pneumonias in careful, honest detail.
• 1948: 56 years, and 16 editions, later, and 28 years after his death, the last version of Osler’s textbook comes out the same year as the first randomized clinical trial of any drug, streptomycin(Drug information on streptomycin) for tuberculosis pneumonia.
The second time line, lasting 58 years to date:
• 1952: DSM-I for psychiatry, an administrative document for classifying hospitalizations.
• 1968: DSM-II, a slim document of neuroses and psychoses, largely ignored by practitioners, and mostly used for insurance documentation.
• 1980: DSM-III, a doubling of diagnoses, the use of the word “disorder” to avoid saying what is disease and what is not, but the first attempt to base diagnoses on scientific research (as Osler had done).
• 1994: DSM-IV, another growth of diagnoses by about a third, with small changes based on science and many other changes based on “pragmatism.”
• 2013: DSM-5, planned.
Which timeline had the greatest practical results? Without Osler’s clinical detail and scientific rigor in diagnosis, the antibiotic revolution would not have been so easily applied to medical practice. Think about it: we have plenty of drugs in psychiatry, but we do not seem to know which patients should get them, and which should not. Some of us use them practically indiscriminately; others hardly at all—depending on our personal tastes, or the preferences of our patients. We boomerang between pharmaceutical libertinism and puritanism. Because of the “pragmatic” approach, we have been doomed, for an entire generation, to failure in biological and treatment studies. No wonder, when one realizes that DSM-IV was specifically and consciously designed to satisfy the personal views of its writers as to what is harmful, rather than to try to match up reasonably well to nature or to our knowledge of disease.
Here is a key example of this false pragmatism. An arbitrary criterion introduced into DSM-IV was not based on scientfic evidence but rather on the “pragmatic” concern that bipolar disorder (BD) should not be overdiagnosed. DSM-IV suggests that antidepressant-induced mania should not “count” as BD; instead, it should be labeled substance-induced mood disorder. Most clinicians have followed this approach as if this condition was part of the diagnosis for major depressive disorder (MDD). Reasonable scientific evidence disproves this assumption. A systematic review of the results of 109 stud-ies that included 114,521 patients found that antidepressant-induced mania was 2.6 times higher with BD than with MDD (15.3% vs 6.0%, respectively).2
Dr Frances writes: “The scientific literature is usually limited, never easy to generalize to the real world and always subject to differing interpretations.”3 If scores of studies with over 100,000 patients are still “always” subject to “differing interpretations,” then what kind of scientific data would ever be plausibly acceptable? The best-designed studies find even more of a difference: The STAR*D study has not yet even published manic switch rates in its over 3000 MDD patients treated with antidepressants because hardly anyone got manic, which means that much less than 1% of the total sample experienced antidepressant-induced mania. In contrast, studies of BD, including the Systematic Treatment Enhancement Program for Bipolar Disorder, have generally found antidepressant-related manic switch rates, at the lowest, in about 5% of patients, ranging to up to 50% of patients with some agents.4
Another example: Dr Frances apparently strongly opposes reducing the cutoff for hypomania from 4 days to 2 or 3 days. Yet there was never a single study that established 4 days as a scientifically valid criterion for hypomania; this was just the commonsense opinion of those who wrote DSM-IV. There are now 2 well-conducted studies that show that less than 4 days adequately picks out BD from MDD based on the standard diagnostic validators, both from private practice or community settings, not the ivory tower academic setting (which Dr Frances frequently presumes is the case). In one study by the late Franco Benazzi,5 246 persons with bipolar II disorder were compared with 178 persons with MDD, using the DSM-IV Structured Clinical Interview, and differences were assessed based on the 4-day definition of hypomania versus 2 to 3 days. There was no difference in diagnostic validators (such as early age at onset, depressive recurrence, atypical feature specifier, depressive mixed state and bipolar family history) for the shorter versus longer hypomania duration definitions. Four days does not pick out anything diagnostically different from 2 to 3 days.
These results have been replicated in other studies, including the longest prospective study in mood disorders, that of Dr Jules Angst, who has followed patients for over 40 years in the Zurich cohort study. Angst was the person whose research was key to the whole definition of MDD as separate from BD in DSM-III in 1980. Angst also finds that brief hypomania, defined as 1 to 3 days in duration, is no different from longer hypomania in any diagnos-tic validators (such as family history of mood disorders, age at onset of illness, suicide attempts, treatment history, legal problems, depressive phenomenology, and hypomanic symptom phenomenology).6,7
Why were Dr Angst’s data—from the exact same study by the exact same researcher—central to making the major changes for MDD in DSM-III, but unworthy of being considered at all for much smaller changes for DSM-5 now? The dichotomy between science and pragmatism is false, as long as science meets reasonable standards of quality, as in the above studies.
Let me add that if scientific studies refute these data, then I will accept them. In the absence of such studies, though, the “pragmatic” (ie, personal) opinions of self-appointed DSM überexperts should not be the final word. A critic may say these studies have problems X, Y, and Z. Fine. All studies have limitations, but this does not mean that all data are useless. Show me studies C and D to refute A and B. Otherwise, studies A and B will be the best evidence we have, and no one’s personal opinions should be seen as more correct than those data. Certainly, sometimes data are too limited and studies too flawed, in which case opinion rules; but Frances takes this approach even in cases where good studies and legitimate data exist. (A 40-year prospective cohort study is as good as any nosological study supporting practically any DSM-IV diagnosis.) This attitude, if perpetuated from DSM-IV to DSM-5, would only ensure that our diagnostic system will continue to reflect the opinions of its writers and be a consciously artificial system, dooming all biological and treatment research, and ensuring that we will never improve our diagnoses based on scientific research.
The idea of pragmatism has an honorable genealogy. Invented by the 19th century philosopher and mathematician Charles Sanders Peirce, pragmatism meant that we should judge the truth or falsity of our ideas not by some inherent quality but by their results in experience; a theory is tested in an experiment and the results of the experiment verify or falsify the theory. This is what Peirce—and later philosophers William James and John Dewey—meant by pragmatism. They did not set up pragmatism as an alternative to science. They did not say our “pragmatic” opinions should outweigh science. They all believed science has the highest priority, and they said so, knowing fully that science has its limits, that it can often be wrong—more: that it is always wrong to some extent. And that it can be misused, as in some of the pharmaceutically related research. And that researchers have their pet ideas and blind spots (but so do nonresearchers). The beauty of science is that error is built into the process: in science, truth is corrected error, in Peirce’s phrase. Scientific work has no fear of being wrong, as long as one seeks to tell the truth as clearly and honestly as possible.
The couch-pragmatic approach—one that does not deign to conduct research, much less value it—involves a disregard for what is true, even when the research evidence provides reasonable evidence for what is true, at the expense of what we happen to think is useful. This is pure utilitarianism, not pragmatism—a bland disregard for truth. Dr Frances makes an exaggerated general premise—that science is always dubious—and applies it where he wants: BD is a favored target, but he rarely writes about decreasing the diagnostic range of MDD, a diagnosis he greatly broadened in DSM-IV. Why? Apparently because he thinks antipsychotics do more harm than antidepressants, on a “pragmatic” level. But this gerrymandering based on which drugs we like or don’t like is no way to develop a diagnostic system that actually gets us closer to understanding and effectively treating diseases. (Unless one thinks there are no diseases in psychiatry, which is a larger topic worth separate analysis.)
I have my own criticisms of DSM-5 (such as the idea of “temper dysregulation disorder,” which ignores much relevant scientific data against it and which is being pushed, pragmatically, as a way to avoid diagnosing BD in children). But I will base my critiques on scientific merits first, pragmatic aspects second—not the other way around. Why Dr Frances criticizes DSM-5—his method—is the problem, not the act of criticism. His method is to summarize perception of harm from treatments; this approach will leave us stuck in a dead end, where each revision of DSM is mere opinion, and no evolution ever occurs, as in internal medicine, toward real knowledge of disease, and real cure.
Psychiatry sits in the same place scientifically as internal medicine did at the end of the 19th century. If we are to experience the advances that internal medicine achieved, we would do well to study and follow the example of historical success. We should work vigorously to describe our psychiatric syndromes to the best of our current scientific ability, much as Osler did for medicine in his textbook. If that work is done objectively and honestly, then such definitions will promote and link to future advances in neuroscience and treatment, unlike our past “pragmatic” diagnoses. This approach to medicine is exactly what Hippocrates supported against the “pragmatic” doctors who preceded him (in the school of Cnidus, against which Hippocrates set up his school of Cos).
First do no harm? Of course, but how do you get there? The humblest and most practical approach is the hard work of science, with an honest respect for its results. This is what Hippocrates taught,8 an emphasis on knowing disease, and treating those diseases one could treat, not an abstract caution about treatment, and certainly not the postmodernist view that all criteria are arbitrary, so let’s set them up based on our personal beliefs about what produces the best outcomes in 1994 or 2013. We need to follow what Hippocrates taught, not co-opt the outcome of his philosophy while ignoring his method about how to get to that outcome; otherwise, what emerges is merely an empty slogan.
In sum, contrary to the historical meaning of Hippocratic teachings and the philosophical meaning of pragmatism, we have here a couch-pragmatism, where, from a recumbent posture, one can ignore research studies because scientific data are always dubious, but one’s own “pragmatic” opinions are self-evident.
Dr Frances' Response: Dr Ghaemi's arguments have been made--and I think effectively refuted--in several previous postings in different venues. Rather than repeat myself, please see the Bulletin of the Association for Advancement of Philosophy and Psychiatry.
Note: Information on the Bulletin can be found at Dr Phillips' article online at Psychiatric Times.