EditorialPursuing treatments that are not evidence based: How DSM IV clarifies, how it blinds psychiatrists to issues in need of investigation
Introduction
Evidence based treatments are the gold standard. Each has demonstrated superiority to placebo that could not have been due to chance. Unfortunately, this doesn’t guarantee effectiveness. Even with treatment compliance, many patients do not return to their premorbid selves. Their ailment may last for years. A given percentage is not helped at all. Typically, 30–50% of depressed patients [1], [2], [3] will not respond, and among non-responders only 23.5–28% are helped by a second medication [4]. The relatively common failure of evidence based treatments to achieve remission is not unique to depression. Frustrating both doctor and patient, similar results are found throughout the full spectrum of DSM IV disorders.
This is not surprising. The science of psychiatry is still young, its conclusions necessarily preliminary. Psychiatry has not found its penicillin, a drug that will succeed 99% of the time in eliminating strep throat, because it kills the germ causing the illness. DSM IV diagnoses are operational definitions, the best attempt by committees of experts to group manifestations of psychopathology into “disorders”. This cataloguing is not the same thing as understanding cause and effect. We have not yet discovered the etiology of any DSM IV diagnosis. The true fruits of science usually await this knowledge.
Sometimes, psychiatric medications work as miraculously as penicillin. The patient returns a few weeks after beginning a medication and is amazingly restored. However, this does not happen as often as we’d like, and when it does, we cannot always explain what has occurred. Neuropharmacology is a sophisticated science. Medication can be designed having specific effects on specific neurotransmitters. Extraordinary technological advances enable us to see what parts of the brain are effected by the drugs that scientists have designed. But, in contrast to penicillin, we are reduced to guessing how the various medications work with a given disorder. Our success rate reflects this.
What is the clinician to do when the promised results of treatment are limited? He has no choice. Decisions about therapy must be made with or without needed hard knowledge. In a similar vein, there is a more general problem over and beyond inadequate efficacy for a given diagnosis. While researchers have the luxury of choosing patients who comply exactly with DSM IV criteria, in the real world of clinical practice, patients’ presenting symptoms are not always as cooperative. Whether a patient’s symptoms exactly correspond, or not, to DSM IV criteria, clinicians must treat patients as they exist. Otherwise they will be treating only a fraction of those needing care.
Trying to remain within a scientific paradigm, once the first treatment (and perhaps a second or third) has failed, well-trained clinicians can become very “creative” when assigning a diagnosis. Presumably, once that decision is made, further treatment can be guided by evidence based protocols. However, reaching the documented 60–70% success rate for the disorder in question is unlikely since the liberties taken with diagnostic criteria leaves the clinician very far from scientific ideals. And this is not taking into account a different strategy, being guided by expert consensus protocols, treatment recommendations for a given diagnosis based on a polling of “experts” rather than empirical evidence.
In the journals these problems may be tackled in a similar way. Fuzzy diagnoses are encouraged by imaginatively broadening diagnostic criteria (for example spectrum disorders). Or new criteria are created, as has been done, for example, with bipolar disorder in children. This diagnosis has apparently captured the imagination of a good many practitioners. There has been a 40-fold increase in this diagnosis over the course of a decade [6], despite its speculative nature. There is scant evidence that adult bipolar meds work for children. Nor is it known whether what is being called childhood bipolar disorder is the real thing (i.e. longitudinal studies documenting that these children are indeed at the beginning of a life long cycle of mania and depression).
Those advocating newer broader criteria for what constitutes a disorder may reasonably argue for broadened diagnostic criteria in a future DSM V. The expanded diagnosis may indeed represent progress, forwarding our understanding of a given disorder. But, all too often, new scientific knowledge is not the basis for the expansion. It is the need to find a diagnosis so as to adhere to a science based protocol, even though scientific reliability has been necessarily violated by a loose application of diagnostic requirements.
Besides controversies about specific diagnoses it is useful to step back and reconsider what, at first glance, seems like a simple issue. What do DSM IV diagnoses represent? As already noted, when DSM III, and later, DSM IV were released, the authors acknowledged the distinction between diagnoses based on etiology or pathophysiology, and the operational definitions APA committees of experts voted on. Operational definitions have a specific purpose in research. Amidst uncertainty, they allow researchers to move forward, confident that they are talking about the same thing. On the other hand, not paying heed to this issue, assuming that DSM IV diagnoses are more than they are, reifying them, can lead to conclusions that make no sense whatsoever, when viewed in the light of later understanding of what is actually occurring.
To illustrate, let us consider congestive heart failure (CHF) as a model. An important aspect of CHF treatment is to focus on a manifest symptom such as edema, especially pulmonary edema and pleural effusions. The strain on the heart from excessive fluid demands the use of diuretics (which, of course, do not act on the heart but on the kidneys). Unencumbered by fluid in the lungs, shortness of breath and orthopnea will improve. Treatment addresses pathophysiology without taking etiology into account.
The actual diseases that brought about the heart failure are once removed from the focus of treatment. Heart failure might be due to muscle damage from an MI or a viral cardiomyopathy. There might be valvular damage from rheumatic fever, or subacute bacterial endocarditis, and so forth through a long list of diseases that can damage the heart. A clinician focusing on the presenting symptoms of CHF is proceeding rationally because regardless of etiology, when the heart does not do its job – when it is not pumping blood powerfully enough – the final common pathway manifested by the accumulated fluid may be of more immediate concern than the underlying cause of the illness. One hundred years ago CHF could be described without an understanding of its many causes, and that description remains therapeutically relevant today, even after we can now better understand underlying causality.
But there is a problem illustrated by CHF as a diagnosis, which is relevant to the way we use DSM IV today. For argument’s sake, let us say our knowledge base remained at the turn of the 20th century and hypothyroidism was causing the heart failure of a given patient. Thyroid extract would fail miserably when tested in a larger population of patients whose “disorder” had been defined as “congestive heart failure.” It might worsen in particular, the illness of those patients whose CHF was, for instance, caused, or exacerbated, by atrial fibrillation due to hyperthyroidism.
Viewed from the perspective of the heart failure diagnosis, the few patients it helped might be described in individual case studies, but would be rightly dismissed as anecdotal, if proponents argued that it should be used as a general treatment for congestive heart failure. Yet, the fact remains that it is exactly the correct treatment for CHF due to hypothyroidism.
My point is obvious. As reasonable as evidence based treatment protocols for symptom-defined “disorders” might seem to be in psychiatry, they can be nonsensical compared to what is possible when a true understanding of etiology can be used to provide rational care.
Analogous to CHF, schizophrenia might be five or seven or 12 different diseases all with the same common final pathway. The heart is basically a pump. Thus, many different illnesses can manifest themselves as the same “disorder” when its pumping action is defective. Similarly, there may be limited numbers of ways that the brain can malfunction when its higher integrative functions are not working properly, or are overpowered by extremely forceful primitive emotional currents. There might be a genetic-caused schizophrenia-like illness with 100% (or 25%) penetrance, a virally based illness, nutritional etiologies (besides known vitamin deficiency induced psychosis.). Some have suggested fetal damage. There might even be (dare I say it) an illness primarily caused by detrimental child rearing. At one time, a syphilitic psychosis resembling schizophrenia used to fill up mental hospitals, so let us not ignore bacterial causality. And what will we learn about prions?
The point is that any number of diseases sharing common symptoms could mistakenly be lumped together. It is an inherent shortcoming of any system based on symptoms alone. Each of the DSM IV disorders could be several illnesses that appear similar to each other but are, nevertheless, not the same illness. Indeed, a person with no symptoms at all might be a closer match to a person with symptoms, than someone with very similar symptoms. For example, we now know that someone with early latent syphilis (two years in to the disease) showing no signs of illness (once again, for arguments sake, before VDRLs were available) should be treated for the disease regardless of symptoms. Someone with optic neuritis (a common manifestation of neurosyphilis) who is not infected by treponema pallidum, would not be helped by penicillin. There is nothing perplexing here. Our understanding of cause and effect allows clear thinking.
Families with schizophrenia running through them have an abundance of schizotypal disorder [5]. Once again, not a surprise; some day in the future we might find certain patients now diagnosed with panic disorder, or obsessive compulsive disorder, appropriately grouped, from the standpoint of etiology, with some schizophrenics. On the other hand, someone with a biological predisposition, say, towards excessive fearfulness (anxiety), or passivity, might have manifest symptoms that veer towards phobias, avoidant personality, obsessive compulsive disorder, panic disorder, or any one of a number of DSM IV disorders, the differences between them attributable to family or culturally learned defenses, parental models of coping, individual trauma, or individual psychodynamics. For example a person using counter phobic defenses (say sky diving) to “choose” and attempt to master their risks for themselves, will present very differently than a frightened individual passively yielding to that fear, or trying to gain a sense of control through repetitive OCD rituals, or dependent on the protection of others, or on idealized love, or through submersion in cults. And medications that treat “anxiety” might be helpful in a large number of disorders that vary greatly from each other in manifest symptoms, but might have in common the fact that the particular symptoms (shaped by other factors) were fundamentally a response to genetically based greater quantities of “fear.”
Are they all the same disease, or as some would have it, are they part of a spectrum? Using the common effectiveness of a given treatment might give hints about etiology, but, not necessarily. Syphilis, pneumococcal pneumonia, and strep throat, are very different diseases, with symptoms that do not resemble each other, but they are all bacterial and are all handled very nicely by penicillin, just as SSRIs are effective for all kinds of different illnesses. To play with the analogy a bit more, strep throat is not really the same disease as rheumatic fever, although the same germ is involved in both diseases. The key point is that current DSM IV disorders are not likely to be equivalent to each other as separate, equal diagnostic entities once we understand the complexity of their underpinnings.
That there are gray areas in making a diagnosis is perfectly legitimate. The concept of spectrum disorders is also legitimate. The problem is in the application of the idea. Implicit in the decision to stretch the diagnosis, to include a patient that does not meet criteria for that disorder, is the belief that the disorder exists in the same sense as an infection or tumor or diabetes. It is not just an operational definition. It is real in the same way as say, a child presenting with fever, achiness, vomiting and diarrhea, without complaining of a scratchy throat, may nevertheless be suffering from strep throat. A clinician who claims a child or adult “really” has bipolar disorder, or ADHD despite not meeting defined criteria, assumes something is going on related to their actually having a specific disease process. Something is wrong with their brain circuitry, or neurotransmitters, or genetics that justifies the diagnosis even if it is not presenting itself in the usual way, or cannot be demonstrated. And a medicine is going to be effective because it somehow attacks the fundamental pathological process. While clinicians sometimes succeed when they try one medication after another on a trial basis, justifying this by a series of guesses about the “real” diagnosis, attributing a diagnosis does not necessarily add to treatment success. Since, for the most part, we do not understand how the medications work for a given diagnosis, we have not added very much to our approach.
The problem of reifying diagnoses extends beyond trying to classify patients who do not meet full DSM IV criteria. Some disorders are problematic not only because they are spectrum extensions of a diagnosis. From a common sense perspective the cluster of symptoms defining the “disorder” is unlikely to be a disease, as that term is commonly understood. For example, it is possible to operationally define the cluster of symptoms that constitutes oppositional defiant disorder. While certainly something is troubling about the behavior, and in theory it may be studied objectively, is classifying this a “disorder” necessarily the best way to conceptualize it?
Merely labeling a syndrome of behavior a disease creates problems. Here is a quote from Educational Horizons Spring 1996: “Once upon a time parents who lacked the courage and/or interest necessary to set limits and impose responsibilities were thought to produce lamed and defiled children. “Spoiled brats” was the common lexicon. Happily, this benighted notion no longer enjoys currency. We now know that a child’s upbringing may really have little to do with “brattiness.” Children behaving like “spoiled brats” are often really suffering from an illness known as oppositional disorder.”
Is this what we meant to do? Are we really going to solve the problem of oppositional and defiant children by calling it an illness? Are doctors “experts” in this area? Any parent who has raised a child with minor or major problems, knows that understanding their family’s and children’s issues in its specifics is more likely to bring results than the crude portrayal that emerges from the use of the broad brush strokes of diagnostic perspectives.
There are other issues. A perfect illustration of how using the evidence based model is inadequate can be found in the use of Herceptin for those breast cancer patients who have too much HER2 protein. Herceptin does nothing for the others. Statistical studies of its efficacy for all breast cancer patients would be meaningless if everyone with the diagnosis were studied. Similarly, the evidence based treatment model (based on diagnosis) can lead to cruder than necessary formulations as illustrated by the finding that Risperidone is effective for major depression when added to standard antidepressants treatments [7]. Is this due to decreased anxiety in an agitated depression (since anxious depressed patients do not respond as well to treatment [8])? Or is there an inherent antidepressant quality possessed by Risperidone? Would Risperidone be as helpful for a depressed patient with anergia or hypersomnia or psychomotor retardation? Or, would it make things worse? This question is not addressed by the research because the study design is locked into the evidence based model of diagnosis rather than the alleviation of specific symptoms. (It is noteworthy that the FDA seemingly considered this diagnosis perspective necessary for a formal listing of treatment indications.)
There are many other problematic issues to raise about evidence based medicine’s reliance on diagnostic psychiatry. However, we will leave this topic in order to return to the main point of this paper, that as laudable as a scientific stance is, we should unapologetically acknowledge the obvious. The inadequacy of our hard knowledge means clinical psychiatry is, and for the foreseeable future will remain an art as well as a science. There is value in formulating treatment rationales that are not statistically validated protocols based on diagnosis, but are, nonetheless, sensible.
Section snippets
The use of medications within a psychological/clinical context paradigm as opposed to a strictly diagnostic (evidence based) perspective
Herman Van Praag’s classic “Nosological Tunnel Vision in Biological Psychiatry. A plea for a functional psychopathology” (1990) [9] warned that exclusively focusing on diagnosis can blind clinicians to other useful ways of approaching patients’ difficulties. Many clinicians limit their focus almost exclusively to the treatment of DSM IV defined symptoms. The rest of the patient’s complaints may be considered chaff, and therefore, should not be a treatment concern. With this perspective, fifteen
References (18)
- et al.
Treatment of SSRI-resistant depression: a meta-analysis comparing within-versus across-class switches
Biol Psychiat
(2008) - et al.
Switching antidepressants after a first selective serotonin reuptake inhibitor in major depressive disorder: a systematic review
J Clin Psychiat
(2006) - et al.
Residual symptoms in depressed patients who respond acutely to fluoxetine
J Clin Psychiat
(1999) - et al.
Recent progress in pharmacological and non-pharmacological treatment options of major depression
Curr Pharm Des
(2006) - et al.
The roscommon family study. III. Schizophrenia-related personality disorders in relatives
Arch Gen Psychiat
(1993) National trends in the outpatient diagnosis and treatment of bipolar disorder in youth
Arch Gen Psychiatr
(2007)Risperidone for treatment-refractory major depressive disorder: a randomized trial
Ann Intern Med
(2007)- et al.
Difference in treatment outcome in outpatients with anxious versus non-anxious depression: a STAR*D report
Am J Psychiat
(2008) Nosological tunnel vision in biological psychiatry. A plea for a functional psychopathology
Ann NY Acad Sci
(1990)
Cited by (2)
What does it mean to have enough evidence?
2013, Journal of Psychiatry and NeuroscienceThe iacam awards-celebrating excellence
2010, Journal of Indian Association for Child and Adolescent Mental Health