Ethics and Evidence-Based Medicine: Is There a Conflict?

Erich H. Loewy, MD

In This Article


When the phrase was first coined, "evidence-based medicine" (EBM) simply referred to the use of therapy that had been proved and tested in a rigorous manner to the point of its becoming "state of the art."[1] EBM has not escaped criticism either because of the assumptions it makes or for the way it has come to be used. Indeed, some of its critics have gone so far as to call reason "the most stiff-necked adversary of EBM."[2] If one strictly adheres to the terminology and resultant definition (ie, that "good" medical practice must be based on evidence that "x" is helpful in treating disease "y" for a particularly defined group "B") one can hardly oppose EBM outright. Presumably the only alternative to that definition would be "myth or rumor-based medicine" (quote attributed to Dr. Thomas Frühwald, a friend and colleague from Austria) which is, obviously, not acceptable when true and valid EBM is available and seems appropriate. Nevertheless, as it now stands, if EBM is applicable anywhere, it is applicable at the bedside and certainly not in experimental procedures, the very essence of which is to go beyond what we already know.

I shall use "evidence" here to mean a statistically valid conclusion about a rigorously defined cohort. It is evident that properly trained biomedical statisticians must deal with statistics and that what constitutes a properly defined cohort (say, healthy Chinese males aged 20-30 years) is determined by physicians and ethnologists who are experts in what is being studied and in etymologic data.

Because we communicate in words, utmost rigor is necessary. Much, however, that is touted as "evidence" is evidence for only a particular cohort under particular circumstances and may be entirely inapplicable in another. What is "evidence" for a 22-year-old male is not automatically applicable to a 66-year-old female. What is "evidence" for 22 year-old Chinese males may not be applicable for 22-year-old black females -- simple logic should tell one these things.

We must, I think, be most careful in defining what we will accept as "evidence." Further, there are always "sports," ie, people who, while they fit as nearly as one can tell into a particular group, nevertheless respond differently from their cohorts. EBM speaks about a large statistical group of people and, therefore, cannot simply be mindlessly applied to the individual patient. Unfortunately, EBM is often used inappropriately. In truth, the actual hard evidence (double-blind studies applied to a specific group) is often badly lacking. And because good ethics begins with good facts, the quality of the facts themselves takes on ethical significance.

Patients differ in age, sturdiness, antecedent or concomitant disease, and a host of other pathophysiologic facts. Just as important, they differ in cultural, social, and economic factors. When confronted with an individual patient it is often hard to determine just who fits where. Surely a 25-year-old well-fed Asian college student who does not drink alcohol and who develops pneumonia is not legitimately placed in the same group as a 20-year-old white unemployed person who drinks heavily and whose nutritional intake is inadequate. EBM perforce is based on averages with the inescapable Mrs. Smith who may not fall into the Gaussian curve.

We must, first of all, understand how the concept "normal" is determined. Let us say we desire to set a range of "normal" for BUN. A certain number of allegedly (deemed by statisticians to be adequate for the study) healthy volunteers have their blood taken and inevitably will have a range of BUN going from low to high. The curve will always be the same bell-shaped curve. To be on the conservative side, the lower as well as the higher values are excluded and the rest of what remains is called "normal." This does not mean that the 5% on either side are "diseased"; once again, it simply indicates that while you can use a large number of allegedly well persons to determine a range of "normals," you cannot reverse the process and apply it to the patient being examined. For example, a patient with a BUN of 22 cannot immediately be assumed to have renal disease.

Statistical data are just that: They tell us, for example, whether a given group of people (and here what is meant by "group" is essential to know) do better with drug "A" or with drug "B," but they tell us little about any individual patient actually being treated.

One of the other problems is that EBM began as a helpful memory jogger -- a reminder of what, in general and for the most part, was superior to other ways of proceeding. Unfortunately, EBM has tended to become a straightjacket, and physicians who deviate from such EBM algorithms risk being "called on the carpet." Some EBM was motivated by the desire to provide the best and most modern care available; however, like "quality control," it has had increasingly less to do with the patient's good than with institutional costs and profits. There is nothing wrong with cost control as long as it is labeled as such instead of being used as a euphemism for quality or evidence. But when we as humans allow disingenuousness to become the norm, and are no longer disturbed by lying for the sake of convenience or using euphemisms with a wink and a nod, hypocrisy becomes the norm and society as well as the profession frays. It almost seems that this is the age of euphemisms. In a society that prides itself on a superficiality of "political correctness," this should come as no surprise.

Further, we specifically have to be careful with what we call "evidence-based medicine." Evidence gathered in 20- to 30-year-old athletes cannot be transposed to 50- to 60-year-old women who lead sedentary lives. One cannot extrapolate from adult to infant or from a middle-class American family to Australian Bushmen. Except for the fact that this misapplication inexplicably continues to happen daily, it would not seem to be worth mentioning. How do we define "evidence"? By whose criteria? What constitutes acceptable evidence? Furthermore, individuals often handle drugs differently; one person may metabolize drugs rapidly and therefore need a higher dose or more frequent administration even though EBM would indicate that most people would not. Most elderly people metabolize drugs more slowly, but some do not. When a patient tells his/her physician that s/he has always degraded or excreted drugs rapidly, the answer "Well, you shouldn't; you are over 70" strikes one as absurd -- except that I have actually heard it, or something similar, expressed on several occasions. It is one of the classic ways in which statistics are misused.

Unfortunately, just because something new has been discovered, tested, and found to "work well" does not mean that it is correct. I remember (this will reveal my vintage) when an antimicrobial called "oleandomycin" was discovered and was glowingly reviewed in the literature. A few years later it was found to be worthless in infection and is now rarely used in place of erythromycin in gastric motility disorders. Imagine if EBM would have, at the time when enthusiasm for oleandomycin ran high, included it in their recommendations. There will be "heretics" who simply are skeptical that what we consider to be true is true -- often their ideas go unpublished, are scorned, and sometimes years later are considered to be "revealed wisdom." The story of Helicobacter pylori as causative in a large percentage of peptic ulcer disease should serve as a warning; it did not then fit in with the prevalent "dogma" of medicine and, therefore, could not even be published. Refusing to publish a paper, by the way, may be a highly effective form of censorship. We badly need a "Journal of Negative Results" devoted to experiments that proved a reasonable hypothesis to be untrue. What we painstakingly find not to be true is at least as important as what is thought to be true and may prevent a repetitive trip down a cul-de-sac.

The problem, as usual, is money. Most institutions in which research is done, and most pharmaceutical companies giving grants, have a clause permitting the researcher to publish negative results. In theory, there is no reason why findings that do or do not substantiate the value of the drug being tested should not be labeled as "inferior" or as having significant side effects. After all, the research presumably passed Institutional Review Board (IRB) standards. Another favorite trick is to compare the new drug "X" against the "old drug" or "against a placebo." The reason pharmaceutical companies like placebo trials is that outcomes are far more impressive when the new drug "X" is being tested against a placebo rather than against an established -- albeit only modestly satisfactory -- drug.

However, where there is a will, there is a way. The experimenter is entirely free to publish results that show that the drug being tested is inferior to the currently used drug. The next time the experimenter applies to the company for a new grant, does anyone truly think that the experimenter's honesty and forthrightness will not affect the outcome? Obviously the experimenter is entirely free to publish results that show that the drug being tested is inferior to the standard of treatment. However, journals, which rely heavily on advertising from such companies, may be loathe to risk publishing papers that are unfavorable to their advertisers.

I fully agree that much of what we do in medicine lacks rigorous justification and is based on anecdote. Why do we counsel antibiotic prophylaxis before patients with valvular heart disease have teeth pulled (but not before colonoscopy or tooth cleaning)? Does the Swan-Ganz catheter truly provide us with useful information? There are hundreds of these prejudices or myths in medicine that have become so entrenched that no IRB would permit establishing a control group to measure otherwise and we, therefore, probably shall never know. If, however, we go down the road of EBM we must rigorously define and rigorously apply the criteria we establish, and must be ever aware of the dangers and not merely embrace the advantages. Physicians must have sufficient elbowroom to deviate for good reason from the current EBM.

EBM has to be seen through the filter of personal training and experience precisely because EBM does not and cannot control for all variables, and personal experience provides those variables. It is like the difference between experimental equipoise and personal equipoise. The former refers to statistics, the latter to statistics seen through the eyes of personal experience.[3]

Furthermore, and to me of extreme importance, EBM protocols must, like drugs or food, have an expiration date after which they have to be reexamined in far more than a perfunctory way to make sure they are still legitimate. Reevaluating EBM protocols would, furthermore, be a wonderful opportunity for one part of CME.

Most things we do in medicine today are by no means strictly evidence-based; one hopes that this will change but only after the term "evidence" is strictly applied. But even then, medical treatment consists of a lot more than prescribing drugs or operating on people. Physicians themselves are by their very nature therapeutic tools -- something that cannot be weighed and measured the way EBM would suggest. There is and always will be a difference between Mrs. Smith and Mrs. Claudesky who have the same pathology but are quite different people. "One size fits all" does not work well for clothing; it works not at all in medicine.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.