Evidence-based medicine is, presumably, about practicing using interventions and techniques which have been supported by empirical, clinic-based evidence—and avoiding techniques which are untested or have failed empirical tests. As one advocate explains,
The practice of evidence based medicine means integrating individual clinical expertise with the best available external clinical expertise from systematic research....By best available external clinical evidence we mean clinically relevant research, often from the basic sciences of medicine, but especially from patient centred clinical research into the accuracy and precision of diagnostic tests (including the clinical examination), the power of prognostic markers, and the efficacy and safety of therapeutic, rehabilitative, and preventive regimens (D.L. Sackett et al., British Medical Journal, 1996).
What alternatives are there to basing medical practice on evidence?
In the face of the obvious, there is indeed much of standard medical practice that remains empirically unsupported. To take an example, the standard formula for the expected length of pregnancy is calculated according to Naegele's Rule, which places the due date at 40 weeks from the start of the last menstrual period. This guideline was calculated in 1838 by German physician Franz Carl Naegele based on the belief that a term pregnancy ought to last exactly 10 lunar cycles (a nice, round number). It was not based on empirical data.
Indeed, empirical studies have shown that length of pregnancy is influenced by previous number of births, age, race, and other factors. So, for instance, the duration of pregnancy for white women with no previous births averages 7 days longer than Naegele's rule predicts (Mittendorf, American Journal of Obstetrics and Gynecology, 1990 and 1993).
Where is the harm in this? Standard practice for many obstetricians is to recommend labor induction—which increases risks to mother and baby—in the 41st week of pregnancy. This is the point at which, at least for first time mothers, only half would be expected to go into labor naturally. The most current figures I could find (from 2002) show that induction rates exceeded 50% for some hospitals, and anecdotal evidence is that they are now even higher.