Giving medicine a fair trial

Evidence based medicine requires clinical trials, but these are often hampered by a misunderstanding of the relation between controlled experiment and “routine” health care. The debate in research ethics should now focus on the usability of clinical trial evidence and the fairness of the design and timing of trials.

What do patients want from doctors in the way of treatment? Only what is best for them as individuals. Doctors want and intend the same for their patients. Sometimes there is uncertainty about which treatment is best. Which to choose? And how? Suppose we are considering not one patient but many. Does this make any difference?

Considering large groups of patients, uncertainty comprises both uncertainty about what to do for each individual and uncertainty about the intrinsic merits of the treatments themselves. Uncertainty about the treatments themselves is now generally agreed to be a solid reason to perform a clinical trial. Arguably such a trial is morally necessary. If one of the treatments is actually better, in terms of effectiveness or safety, but we continue in a state where we don't know which then many future patients—including the patient in the waiting room outside—will receive inferior treatment as a result.¹

This conclusion is the source of one of the most misleading prejudices about clinical trials: that they are run to benefit future patients at the expense of present patients. This is not entailed by anything we have assumed. Moreover, there is substantial evidence that people do better in trials (whichever treatment they are assigned to) than in routine treatment and that patients (especially doctors) would prefer to enter a trial than be treated in an “uncontrolled experiment.”^2–5

Conversely, if we are uncertain about the relative intrinsic merits of the treatments, then we cannot be certain about those merits in any given use of one of them—as in treating an individual patient. So it seems irrational and unethical to insist one way or another before the completion of a suitable trial. Thus the answer to the question, What is the best treatment for the patient? is: the trial. The trial is the treatment.⁶ Is this experimentation? Yes. But all we mean by that is choice under uncertainty, plus data collection. Does it matter that the choice is “random”? Logically, no. After all, what better mechanism is there for choice under uncertainty?

All the weight in this argument depends on the nature of the uncertainty. And this uncertainty is essentially relative, on the one hand to a body of existing knowledge (evidence) and on the other to what is meant by effectiveness. Effectiveness is always effectiveness for something. Obviously, we want cures, reductions in mortality and morbidity. But when treatments are imperfect in achieving these ends, or when the ends are more nebulous, there is a task of definition. Who gets to play this game? More complicated still—how can we decide about trade offs of different values, for instance of long (or short) run risks against short (or long) run benefits? It is a well known theorem in social choice theory that no consistent social preference function can be constructed on the basis of individual preferences alone.⁷ We cannot, in other words, decide once and for all which trade offs will be generally accepted. The only general rule is to use treatments that work and not treatments that don't. And by “work” we had better mean something straightforward.

It is commonly said that clinical trials should be directed at endpoints that are relevant to patients, such as quality of life. This is only an effective way of making trials harder to run, take longer to implement, and more difficult for ethics committees to approve and of making their results harder to generalise and apply. Such trials then need to be repeated. The window of uncertainty gets bigger, and the time when patients can place confidence in their doctors' judgments of superiority recedes into the future.

From this we can draw two morals. Firstly, trials should be simple, timely, and well designed to answer well posed questions.⁸,⁹ That is all. They cannot second guess what patients prefer. When I make a choice, I want to know what will happen, what may happen, and what could go wrong. What other people might have chosen in my place is interesting but irrelevant.

Secondly, trials are intrinsically no more threatening than any other form of treatment. In some ways they are better. Standards of care and consent are often superior.⁶ What most people worry about is uncertainty being discussed out loud. Yet medical science and practice never was built on certainty. We have better understanding of uncertainty than ever before, and better methods for grappling with it. The demystification of the doctor, which is supposedly the risk which attaches to admitting uncertainty, has already happened.¹⁰