Several days ago I questioned the term “evidence-based” and implied semantic drift. A regular commenter provided this wonderful link to support my hypothesis – What statins tell us about the mess in evidence based medicine

The problem is actually a straightforward one. Rarely does the evidence provide a clear answer to our question. We can all think of exceptions – rapid percutaneous intervention for STEMI, ACE-I and beta blocker for decreased systolic function, anticoagulation for acute pulmonary embolus, etc.

But often the evidence does not provide a definitive answer. Sometimes the evidence is just that, evidence. Using detective thinking, we really understand the each piece of evidence represents a clue to the mystery. Often, we over value evidence.

Using the detective/mystery framework, evidence generally stimulates hypotheses. We then need more confirming evidence. In many classic mystery novels, movies and TV shows, the evidence seems to point in one direction, but further thought reveals other possibilities.

In medicine we find the same problem.

True EBM devotees evaluate the evidence in the proper way. They value the evidence according to the type of study (RCT, observational, etc.) and evaluate the study’s methods carefully. They examine the preponderance of the evidence prior to giving a recommendation.

We have too many amateur data interpretations. We find a piece of evidence and extrapolate beyond the evidence. We fall in love with our interpretation, and ignore evidence that does not fit, or the lack of adequate evidence.

EBM (using these initials implies the proper use of evidence) has great promise, but cannot answer all our questions. But too often, amateurs (here I am using the phrase pejoratively to refer to those who do not carefully consider all the evidence) invoke the term “evidence-based” to add value to their own expert opinion. We physicians must learn to distinguish between true EBM and inaccurate terminology usage.

How does this impact patient care? Since many medical societies advance performance measures, we would expect careful use of EBM, however, too often we get eminence based medicine masquerading as EBM. Let me use the example of HgbA1c targets. The evidence for a target of HgbA1c < 7 came from a very well controlled study. Those in favor of these targets extrapolated from a carefully constructed study population to all diabetic patients. They did not understand the limitations of the evidence.

This problem is common. The ACP Performance Measurement Committee reviews submitted measures and evaluates them in a very careful EBM process. The committee gives a thumbs down to many more measures than they endorse. The ACP web site has these evaluations available – ACP Performance Measure Recommendations. This high rate of “disapproval” makes my point. We must all be skeptical of the phrase “evidence-based” unless we know that the person using the term truly understands all the implications and requirements.