Comparative effectiveness research in medicine: we need it, but how to get it?

Every patient wants the best care — what is known to work. Certainly nobody wants care that doesn’t work, especially if what doesn’t work carries some risk of its own. But what if we don’t know what treatment is best for a particular condition? Shouldn’t we find out?

It is well known that there are wide variations in medical practice across the country, even for the same conditions. Even in the same region, doctors choose different therapies for the same condition. This is often because the circumstances of patients differ. But what if doctors choose different therapies for no particular reason except personal preference? And what if those therapies, although producing the same outcomes, differ in cost by a large amount?

Enter comparative effective research (CER). The idea is simple: compare two treatments and see which works better. And, if one works only a little better and the cost difference is huge, is the tiny improvement worth the cost. It seems odd to nonphysicians, but this kind research is hard to do, and is in fact rarely done. We often compare some new, experimental treatment with the standard; but it is often hard to compare two standard treatments with each other to see which works better. Why?

A recent editorial in the New England Journal of Medicine examines why CER can be challenging to do. The reason often boils down to money. In the example given in the article, a very expensive drug was being compared with a much cheaper one. Both therapies were covered under the patients’ insurance (in this case Medicare). But one group of patients would have an enormously high out-of-pocket copay fee because the copay is often calculated as a percent of the total bill. How do we deal with that? The guiding principle of a randomized, controlled trial is that the patient groups are similar. It would affect the trial if one group of patients could afford (or was willing to pay) a higher copay than the other.

Another key principle of such trials is that neither the patient nor the evaluating physician knows which group the patient is in — which drug they are receiving. This is called blinding. For experimental drugs blinding is easy; the patient gets an unidentified drug, marked only with a code that will be broken later. But if the drugs being compared are both covered by insurance, the insurance statement shows to the patient what it is.

These problems are solvable. But it is important to realize that comparative effectiveness research, which we very much need if we are to control our exploding medical costs, will not be easy to do in many situations.