Monday, July 18, 2005

Evidence Based Medicine can never provide all the answers

Evidence Based Medicine (EBM) 's broad "official" definition is the integration of the best evidence with clinical expertise and patient's values. The popular or limited meaning seems to be EBM is the evidence derived from RCTs and meta-analysis and if all fails observational data or case control studies. It is EBM used in that second sense that I maintain will never, can never provide the answers to all of the questions and decisions that arise in the practice of medicine. There are simply far too many questions and issues that there is time, energy or funds to do all the RCTs to answer all of the questions posed. The Herculean task of gathering all the data necessary to characterize unique patients for Bayesian prior probability determinations is likely one that will never be completed. It is worse that that because as new medications, surgical techniques and diagnostic techniques are devised more and more questions are raised. As new modalities appear, previous RCTs may become no longer relevant. RCTs sometimes are contradictory ( remember ALLHAT, and the Australian BP study and now ASCOT). sometimes a RCT concludes something and as time goes by the data is dissected and analyzed further and the once solid conclusion melts away.Case in point was the DIG trial in which, although no morality benefit was evident , fewer hospitalizations were needed and symptoms relief occured. Later upon more analysis we learn that digitalis may not help women and that the patients whose dig levels were in the acceptable blood levels were harmed.Further, a RCT does not provide all the answers . Consider a RCT that shows a new ACE inhibitor decreases stroke and cv disease risk by X %. There are many questions raised by that. To what degree can we go beyond the data to apply the new treatment to patients whose characteristics would have made them not eligible for the trial. How do we use the drug in patients who may have several or many co-morbidities which -for the most part or the entire part-the study subjects did not have.Further, many of the day-to-day mini-decisions physicians make will never likely fall under the analytic eye of the clinical trial or even the coarse grain data gathering of observational data (aka data dredging).These are such things as when to have the patient return for followup under a myriad of endlessly varied conditions. (A lab test is slightly abnormal, when to repeat it, if at all, or do confirmatory tests, if at all,etc. etc.)
Further as I mentioned before the application of group data or population data to individual patients involves more than knowing the bottom line of the article. It is only the beginning of the decision process for a given patient. You cannot expect the average effect to occur in every given patient you are treating. Even in the circumstance when for a given clinical treatment decision there at least one RCT or meta-analysis that is relevant to the issue a clinician still has to apply the data to the particular case and that often involves much more than" give drug x because a RCT showed that x brings about 30% decrease in mortality".
Why bother saying that RCTs and MAs cannot provide all the answers? Am I just saying the obvious. I believe it is worth mentioning for at least two reasons.
1) the medical students of today may plunge into the real world thinking that all the answers will be revealed through EBM-used in the limited definition given above- much as some of the older generation of doctors may have believed that the truth lay in the wisdom imparted to us by our venerated role models 2) some of the medical and some of the lay literature seem to believe that the answer to all of medicine's real or alleged problems lay in EBM. Some of the answer do lie there, but not all and many will never be addressed.
As is true of all human endeavor many times physicians will fall back on various heuristics for guidance and use human judgment to sort it all out.
EBM will provide the answers if we take the official definition to "really" be EBM because sometimes the integration or application of the "best available" evidence is to apply little or no evidence at all, at least not evidence in the sense of RCT's or even case control studies. Here the individual doctor-not the health care team-has to give individual advice to an individual patient. The physician will be aided with ready computer access to the latest information but as Dr. Thomas Giles-president of the American Society of Hypertension- recently said "we'll need a thinking physician and [a] thinking patient to come up with the best regimen for every individual".

No comments: