Let me start with my background. Several years ago I was the President of the Society for Medical Decision Making. I have taught Basic Diagnostic Testing many times – and helped develop a course on that subject. I have written about Receiver Operating Charactertic Curve analysis and Likelihood Ratios. My CV includes cost-effectiveness analyses and decision analyses.

Given that background, I remain cautious about how we should view evidence based medicine. A recent blog entry at Health Care Renewal – The Post-Modernist View: Evidence-Based Medicine as Fascism? – discusses a critique of evidence based medicine. My post today will critique EBM in a slightly different way.

Physicians must learn the principles of evidence based medicine. We need to understand how to evaluate the literature, especially randomized controlled trials. We need to read critically.

However, we should not expect that EBM will solve all the problems in medicine. As I read EBM reviews, I often see biases. To do a review, the “expert” must evaluate the quality of the data. This evaluation is, unfortunately, a bit subjective. Often the “expert” has to make judgements using an implicit utility structure.

Let me give an example. We currently have two major guidelines on the care of adult pharyngitis – Principles of appropriate antibiotic use for acute pharyngitis in adults (ACP-ASIM) and Practice guidelines for the diagnosis and management of group A streptococcal pharyngitis (IDSA) . Given the same problem and the same data, the two group propose different recommendations.

As one analyzes guidelines, we almost always find prejudices, because we cannot (and I would submit should not) quantitate every feature of medical care. The two above guidelines clearly have different vantage points. The ACP-ASIM guideline gives primacy to patient symptoms, while the IDSA worries most about “unnecessary antibiotics”. Each group uses the evidence to promote their world view.

The Cochrane Collaboration Review of pharyngitis looks somewhat different – Antibiotics for sore throat.

Available data suggests that antibiotics are of limited use for most people with sore throats

Sore throats are infections caused by bacteria or viruses, affecting mostly children and young adults. People usually recover quickly anyway (usually after three or four days), although some people have complications. The most serious possible complication is rheumatic fever, a disease affecting the heart and joints. Antibiotics can reduce bacterial infections, but communities build resistance to these drugs, and they can cause diarrhoea, rash and other adverse effects. The review of trials found that antibiotics shorten the illness by an average of about one day. They can reduce rheumatic fever in communities where this complication is common.

So which evidence based analysis should we use? I have picked this subject, because I have read the literature on sore throats for 25 years. These analyses frame an interesting philosophical debate in medicine – how do we balance patient short term suffering with theoretical public health concerns? But where does that leave one as a practicing physician?

The problem that I see with overzealous promotion of EBM is that is removes context from decision making. I had the good fortune to know one of the giants of Alabama Internal Medicine – John Burnum. Late in his career, he taught medicine at our Tuscaloosa campus. I would go there periodically to take morning report. We spent some time discussing medicine, and developed a catch phrase about patient presentations. One cannot discuss a patient’s case without acknowledging the context of that patient. I cannot think about the word context without remembering Dr. Burnum.

The point that I want to make (after a bit of rambling) is that EBM must be only one factor that we consider in our decision making. We must always consider the context of the patient.

What makes medicine so challenging is that patients vary significantly. Patients do not easily fit into simple definitions. We must temper evidence by the context of the patient.

Now we hear calls for more evidence on diagnostic testing – Diagnosis—the next frontier

The message I take from this is that while evidence based treatment is well on the way to being sorted out, evidence based diagnosis is still in the dark ages. This week’s journal suggests that things are beginning to change. In her editorial Sharon Straus states what should be a self evident truth: “When making a diagnosis in patients who are already ill we should be able to draw on evidence about the accuracy of diagnostic tests” (p 405). Yet just how far we have to go is shown by a study from Susan Mallett and colleagues of reporting and review methods used in systematic reviews of diagnostic tests for cancer (p 413). Lousy methodology means that “even these apparently evidence based studies are flawed,” comments Straus. Relief is at hand: those repositories of evidence based treatments—Clinical Evidence (published by the BMJ Publishing Group) and the Cochrane Library—are turning their attention to diagnosis.

Now I will object to two ideas here. First, I do not believe that we are well on the way to sorting out treatment. We have some good information, but patients are too complex to always fit into a simple framework. Second, I believe that we already have good information on many diagnostic tests. The problem with diagnostic tests is that they only give accurate information in the context of the patient. We must be able to estimate prior probabilities to evaluate any test. We must know which tests to order. We must think of a diagnosis or we will be unlikely to order the right test.

I applaud those who work to develop the field of evidence based medicine, but I caution them to not over sell the promise of EBM. EBM represents another tool in medicine. We should pay attention to the findings, and then apply them to our patients. But we must always apply these, and any other tools in the context of the patient. If we try to overemphasize the science without considering the art, then we are failing our patients.

Related posts:

1. More on premature closure and mystery novels
2. Mystery novels and diagnosis
3. More on safety and root cause analysis
4. Intuition versus evidence
5. The importance of patient volume for learning

Related posts brought to you by Yet Another Related Posts Plugin.