Matthew Holt of The Health Care Blog fame partially nails this issue – QUALITY: Why doesn’t evidence-based medicine happen in practice? (permalinks do not work, so scroll to Thursday, Dec. 11).

My conclusion is that no evidence-based guideline will be perfectly applied. Some don’t take into account the human situation of the patient. Meanwhile physicians will find it very hard to do something that their experience tells them is wrong–no matter what the data says.

But of course in the US this is more or less moot, as we don’t have the data.

So he gets right the parts about the difficulty in applying evidence-based guidelines to individual patients. As we (and I am part of a research group that studies such issues) study these issues, one of our greatest challenges comes in defining “ideal” candidates for a drug. For example, we all know that ACE inhibitors decrease mortality in CHF caused by systolic dysfunction. However, ACE inhibitors do have side effects and contraindications to use. Our challenge (and the challenge of any report card study) is to accurately define the denominator which we use to calculate the percentage of patients who achieve the guideline.

Now Matthew is mistaken in thinking that we do not study this in the US. Medicare sponsors many such studies, giving feedback to physicians. We have learned several things about quality.

Quality (as measured by percent compliance with a guideline) varies across indicators. Quality changes across time. More post myocardial infarction patients take a beta blocker now than 5 years ago. Physicians do learn and do adopt changes in practice.

However, changing ones practice occurs for physicians at different speeds. As we get older, we become wary of the latest and greatest. We have seen too many new drugs have major side effects discovered within 2 years after release. We need excellent data to change from therapy that has worked.

I have written about this several times in the past – these two rants are a good start –
On knowledge translation in which I discuss the problem of translating knowledge into practice and art 3 in which I answer a question about why physicians do not adopt change quickly. This link may help also – The Technology Adption Life-cycle . Quoting from my Part 3 rant –

As one studies adoption of new practice, one finds an interesting curve of adoption.

At what point on this curve would you find someone guilty of malpractice. How do we decide when everyone should have adopted an innovation (and I would argue from my example that many still consider NAC an innovation in protecting against dye induced renal failure)?

We should look at the flip side of this curve. What if I am an early innovator of a drug which causes a serious side effect? Am I guilty of malpractice then? Where should I lie on the technology adoption curve?

My point remains that these issues are more complex than simple sound bites make them appear. We are striving to teach physicians to optimize their practice, however they know that optimal practice in 2003 may change in 2005 (e.g.,hormone replacement therapy for preventing coronary artery disease).