The Serious and Insurmountable Problems of Evidence Based Medicine1) Something works in medicine. It spreads around the world in months. When patients are doing well, medicine is a highly paid, piece of cake. It is almost like stealing. When patients are not doing well, it is a living hell of time, effort, and extremely low pay for doctors. So the incentives are in the right direction. Doctors may be trusted to want patients to do great. And the loss of work by patients' cures is not a problem due to the shortage of overly busy doctors.
2) A couple of years later, an academic doctor sees this response, designs a study, writes a research proposal, gets funding, carries out the study, writes up the results, waits for its publication. So, 7 years has passed. A number of studies accumulate. A committee reviews them. They enter a textbook, as accepted practice. It has now been longer than 7 years, by the time a guideline is written based on published studies. Meanwhile, the docs are doing almost nothing the way they were 7 years ago. The standard of care has moved on, except in the minds of guideline writers, government officials using guidelines like laws. These officials only want to slow clinical care to save money by piling bureaucratic procedures, and by denying dark skinned people that white people are getting.
3) On the first day of high school statistics class, coin tossing is discussed. That event is described by the binomial distribution statistic.
4) Studies comparing the fractions of responders to a treatment and to a placebo are supposed to represent the larger population. The parametric statistic is used to compare the fractions. The parametric statistic is the one whose formula describes a bell shaped curve, a common distributions of populations. Before carrying out such a test, one must show that 4 assumptions have been fulfilled. The most important is random selection. So any selection bias, such as an exclusion criterion, makes it so that the test is not even allowed to be done, let alone have any validity. All studies have exclusion criteria and violate the central assumption of parametric statistical testing. The populations in these studies do not represent those in the clinical setting. Doctors do not have exclusion criteria in their practices. Imagine excluding suicidal patients from a depression treatment study. That is routinely done in the FDA approval of new anti-depressants. Worthless.
5) Clinical care differs from the comparisons of the fractions in groups. It is closer to coin tossing. Have or not have a diagnosis. Give or not give a treatment. Have a good result or a bad result. The binomial statistic is more appropriate to clinical care than the parametric. Nevertheless one is not allowed to apply parametric statistics to a population best described by a binomial distribution.
6) Dose response curve is ignored. Low doses of radiation are good for the health, for example, as in radiation hormesis. One must delineate the dose response curve of all remedies. Then one must do so in the individual patient, and this is where experience based medicine beats evidence based medicine in outcomes.
So evidence based medicine has problems, 1) delineation by academic professors with half the clinical experience and therefore half the insider knowledge of clinicians; 2) obsolescence; 3) based on wrong statistical application; 4) violation of the rules of statistical testing by exclusion criteria in all studies; 5) misapplication to individual patients (a treatment killed 99% of patients who had it, this patient has done well on it, follow guidelines and stop this effective treatment?); 6) ignorance of the individualized dose-response curve.