Fallacy of Relying on Evidence Based Medicine

Contrary to common belief, medical treatments classified as "evidence based"  don't work for everyone and non-evidence based treatments frequently work for some patients. Recently, parents who believed that their children's autistic symptoms improved while taking an experimental medication were dismayed when the drug trial was discontinued because the medicine's efficacy was questioned. (A Drug's Bitter End, NY Times, June 7, 2013) 
     Pharmaceuticals designed to treat autism are especially difficult to evaluate because the condition is multifactorial and patients may have different conditions that cause similar symptoms. At the most simple level, lower abdominal pain caused by appendicitis and the pain caused by a ruptured ovarian cyst might feel the same to the patient and even have similar findings on examination, but the treatment for these conditions is different: If the two conditions were combined in the same study, appendectomy for appendicitis might not pass the test for statistical significance. 
     If we think of the unifying symptom of autism as a severe learning disability for socialization (especially an inability to interpret social cues), it is easy to understand that children who have this symptom as an isolated finding might require different treatment from those who have the same symptom as part of a global intellectual delay.
     As a general principle, it is good policy to prefer a treatment that is "evidence based" over one that is not because it is reassuring to know that scientific studies have confirmed efficacy for the majority of patients. In many cases, if a treatment does not meet this standard, insurance companies and government agencies will deny payment. Unfortunately, the term is frequently misinterpreted: When a treatment is classified as "evidence based" because it works 70 percent of the time, 30 percent of the treated patients will either receive no benefit or possible get worse - sometimes because of the treatment. Conversely, if a treatment only works 30 percent of the time, it will not be classified as evidence based, but the 30 percent of patients who might have been helped will simply be out of luck.
     No one wants to see useless therapies replace those with proven efficacy, but  patients and physicians should be aware of the limitations of relying solely on treatments that are described as evidence based. There are times when being "too scientific" will not be in a patient's best interest.