There seems to be confusion as to what I mean, which is understandable as even some doctors don't really understand EBM.
EBM has revolutionised Medicine, but it has not made it more 'scientific'. Cochrane and Feinstein, two of the framers of EBM, were instrumental in it not being called Scientific Medicine, as some initially wanted it to be called. Evidence-Based practice is increasing, because it is useful and a good thing, but it should be understood in context.
The art of Medicine is not just in bedside manner, but also in which treatments to employ or drug regimens to choose. EBM suggests what is 'Evidence-Based Practice', but it is an heuristic technique. It suggests what should be followed, but by no means says this is therefore correct. As I said, Medical studies cannot be repeated or disprove other studies. Even if a study found just a slight increase in morbidity, ethically it can't be repeated since you are then negatively impacting some participants' health. It is the problem of non-maleficence. Much medical practice thus hinges on flimsy evidence of one or two small studies, so really should not be taken as absolute gospel. You can't just say Evidence says this and slavishly follow the guidelines, for that would make you a poor doctor. The evidence needs to be weighed and assessed based on the patient in front of you, for on occasion you would need to throw in your lot based on weaker evidence or against it.
Sometimes it is clear cut and definitive though, such as with the efficacy of Aspirin or Statins.
Good examples here are with Digoxin, which shows no decrease in mortality by Evidence-Based means, but which is still employed for symptomatic relief. It increases the cardiac output, so in sick patients that require it, by inductive reasoning it should increase life-expectancy, and EBM has not altered this understanding. So there is a lot of disagreement.
Similarly antidepressants, which Kitsch showed mostly had effectivity that could not be confidently said to be beyond Secondary Placebo effect (except Sertraline). The confidence interval is slim. So do you prescribe them or not, as they have significant side effects?
Or in my own field, Cricoid Pressure. This is pressing on the cricoid cartilage during Anaesthesia to prevent passive regurgitation of stomach content into the lungs. It is used worldwide based on 1 study done in pregnant patients. It cannot be repeated, because that showed a protective factor - yet cricoid pressure distorts the airway, making intubation more difficult; requires pressure of 10 mN, which no one really knows if sufficient is given; can cause oesophageal tears if patient actively wretches; 40% of patients the oesophagus is not behind the cricoid anyway; etc. The study was also done in pregnant patients that have decreased lower oesophageal sphincter tone, so we don't even know if it works at all in non-pregnant patients.
So EBM says to give cricoid pressure, but should you? You need to look at each patient and decide if you think these risks outweigh the presumed benefits, or not. Whether this specific patient would benefit in your opinion. In other fields studies could be done, and oblique approaches like cadaver studies can shed light on it, but Evidence alone can't tell you. This is the Art of Medicine.