The Myths, Abuses and Pseudoscience Surrounding “Evidence Based Medicine”

There are 3 types of lies: Lies, Damn Lies and Statistics (Mark Twain). In approximately the last 10 years there’s been great interest in the concept of Evidence Based Medicine as a unifying concept for teaching, evaluating data and practicing medicine. The concept goes back at least 150 years to France with renewed interest recently for specific purposes. Although evidence-based medicine (EBM) has applicability in all areas of clinical medicine I will focus my commentary as it primary relates to Psychiatry and Clinical Psychology. After all, what a great title, Evidence-Based Medicine! Is there anyone who could argue that the practice of medicine should not be based on the highest levels of evidence? Would anyone suggest that non-evidence-based medicine might be superior? The main argument regarding EBM comes in its relationship to what has been referred to as expert opinion (EO). The concept of EO is where the repository of clinical expertise comes from, a succession of expert clinicians and their overall experiences and knowledge. Certainly these concepts can coexist, and for the most part they do. Nevertheless, there are those within the healthcare industry who strongly push for the superiority, perhaps even exclusivity, of EBM being the unique path to medical proficiency.

The evidence of EBM can come from many sources, but the vast majority accepted by those who champion it’s supremacy come from either controlled clinical trials or large meta-analyses. Controlled clinical trials represent the type of experimental study submitted to the FDA for an approval for a new medication. They’re done with a great deal of precision. They use very special criteria for diagnoses, severity of disease, inclusion and exclusion criteria for entering a patient into a study. They are run for defined periods of time such as eight weeks for the use of a new antidepressants in treating unipolar depression or a16 week module of cognitive behavioral therapy for the treatment of panic disorder. Dosages whether of medicines or therapy are determined previous to the initiation of the study, as is what is constituted as a cure or positive outcome. Many subjects interested in joining such clinical trials are denied because they do not quite meet criteria for inclusion. Meta-analyses are methods of combining data from numerous concluded studies supposedly treating a given illness with some common therapy.

EO would also rely on things such as clinical trials but in a much less rigid manner. The best arbiter of treatment in a difficult case is generally the most intelligent and experienced clinician who has encountered large amounts of patients during his career and therefore can counsel his/her junior colleagues. Clinicians undoubtedly do not rely on one person but in many people who may or may not have formally been their teacher. I personally thank Gerald Klerman, Giovanni Cassano, David Sheehan, Athanazio Kukopoulos and Tom Hackett who in some particular ways have been mentors to me.

In EBM there are actually those who would take a position that if there is no research data behind a certain treatment that it is somehow improper to use that treatment. Interestingly this is a position rarely supported by clinicians such as psychiatrists, psychologists, or nurse practitioners. Those supporting such a stance are usually associated with the insurance industry, the government or hospital administrators. Some may actually believe that if there is no data behind a certain treatment it should not be used. Unfortunately, I’ve become jaded enough myself to believe that their support on EBM is more based on their ability to cut costs and deny the scope and depth of our treatments, especially when costly.

For example. A 44-year-old woman presents with her fifth episode of unipolar depression. She has some passive suicidal ideation but overall has shown some progress over the past four weeks since starting the medication regimen. You’ve treated her with supportive psychotherapy and Fluoxetine 40 mg per day. In the past she’s been very difficult to treat and has gone on to make two significant suicide attempts. She seems to only tolerate Fluoxetine among the many antidepressants. She lives a good distance from your clinic and it would be difficult for her to come in five or six times a week for transcranial magnetic stimulation and you would rather avoid advancing her treatment to ECT. Having had patients similar to this one you feel the next best thing to do is to increase her Fluoxetine to 60 mg per day. You do so and enter it into the record. The next day you get a visit from a hospital administrator who questions your use of 60 mg of Fluoxetine in a patient with unipolar depression. He does so stating that there is no data supporting Fluoxetine at 60 mg being superior to Fluoxetine at 40 mg in unipolar depression (although there is data that supports it being useful in the treatment of OCD, and also safe for the patient at this dosage). The hospital administrator has a smug self-righteousness about himself suggesting that he is “in the know.” Now there can be a lot of features deserving of comment, but let me focus on this one. No one has ever done a systematized, placebo-controlled, double-blind, random assignment study of 60 mg of Fluoxetine in the treatment of unipolar depression. In fact, I would stipulate, especially from the point of view of somatic therapies, that the majority of treatments that we use have no strict, perfectly congruent underlining research supporting them. Clinicians certainly as part of a good standard of care do not just invent treatments or follow their own idiosyncratic ideas, but they most heavily rely on their experience and the experience of their teachers and colleagues.

Also, the nature in which we define psychiatric diagnoses does not divide categories into a neat entities without overlap with other conditions. From the 1980s to the present, the evolution of DSM three to DSM-V was done by consensus of experts as to what best constitutes a given illness recognizing the inherent heterogeneity and overlap of these conditions. The bulk of our diagnostic system was truly not based on the accumulation of controlled data. This further points out how EBM alone is inadequate to deal with the real life patients coming to us for help. Such real-life patients also generally look very different from those that get enrolled into clinical trials. This is for many reasons, but the main one is the use of inclusion and exclusion criteria in our scientific research. If we have a new study for depression, we will generally exclude anyone who is currently alcoholic, having uncontrolled endocrine disease, seizure disorder, cognitive decline, etc. This is because clinical trials search out the prototypical patient of a given category so as not to make the error of enrolling those who do not really have the disorder under study or who may have some factors that biases the sample into being particularly difficult to treat.

I have spent my career primarily designing conducting and analyzing controlled clinical trials. There is no doubt that they are necessary. But their meaning can be abused or misunderstood. Also, there are those that feel that if you can put some statistics to your clinical reasoning it is somehow more sophisticated and true. Statistics nevertheless remain among the most easiest of entities manipulated to serve one’s utilitarian purposes. Such purposes are not always on the side of the patient. I will therefore refer you back to the quote by Mark Twain at the beginning of this article.

In conclusion I would state that both EBM and EO are important in the development of clinicians and the practice of clinical psychiatry and psychology. “But buyer beware.” There will be those trying to influence your practice whose understanding of the clinical and therapeutic issues are overly simplistic or generated by something other than giving the best clinical care.

You can reach Dr. Deltito by email at

Have a Comment?