Drug Trial Snafu: The Mandate for Ignorance

We use the term "mandate for ignorance" to refer to a hidden but, we believe, enormously important problem in the effective management of clinical trials.
It is widely agreed that a placebo trial (or any other drug trial in which patients are assigned at random to two or more treatment possibilities) is unethical unless the researchers can state an honest "null hypothesis"--that is, unless they really do not know which of the treatment arms is better. For example, a placebo trial is unethical if the researchers know that the drug works better than no treatment. But if they do not know, then the study is OK, and they are free to proceed with it.
This ethical standard is obviously intended to safeguard the welfare of patients who become subjects in experimental trials. The idea is that subjects should not have to risk a placebo unless nobody knows whether the drug or placebo is better. But the ganciclovir history shows that this well- meaning ethical standard also has monstrous consequences, because it puts a premium on ignorance.
Why have drug companies become so adamant about refusing compassionate access to AIDS treatments, access commonly available for cancer or other diseases? One reason is to force patients into placebo trials--trials often so poorly designed and poorly publicized that few would enter them voluntarily. But another reason is that drug companies must remain ignorant of how well their own products work, lest that knowledge ethically bar them from conducting the very trials which the FDA requires for approval!
This catch-22 is exactly what happened to Syntex with ganciclovir. And it has terrified other companies into denying compassionate use, and thereby refusing to learn how to use new their drugs effectively except as the result of cumbersome, years-long formal trials.
What should Syntex have done with ganciclovir? Everyone seems to agree that the company should have run placebo trials early--in what is being called "the window of opportunity" to conduct clinical trials--before the public or even the researchers know whether the drug works or not. In other words, the trial must be run before compassionate use--lest the knowledge that the drug does work make it impossible to run the trials. But in practice, trials have usually taken years to set up and conduct. Therefore, compassionate use must be denied for those years.
This mandate for ignorance not only denies compassionate access to patients; it also saps the vitality of the entire clinical-research enterprise. It requires that companies not know that their drug works until after they run formal trials-- not before. But companies have little motivation to invest money in trials unless they believe that the drug does work-- the same belief that would make those trials unethical. So nothing happens.
One consequence is that drug companies refuse to cooperate when physicians want to test new uses for products in development. Besides other problems such as fear of liability if things go wrong, the company could lose its window of opportunity if things go right.
Therefore the mandate for ignorance not only destroys the commercial incentive to run trials, it also forbids the normal exploratory phase of science--which here would be physicians trying experimental drugs in compassionate, well-supported attempts to treat patients. As a result, clinical trials must be designed in a vacuum--based on whatever theories are fashionable, rather than on exploratory experience of what actually does work in practice. The result is irrelevant, indecisive, and ultimately ineffectual clinical trials, when they happen at all.
In a recent editorial in the Journal of the American Medical Association, a former head of the FDA urged a re- examination of "all of the assumptions on which the scientific requirements of the present system (of drug approvals) are based" (November 25, 1988; see also The New York Times, Medical Science section, December 6, 1988.) We hope that this mandate for ignorance will be among the issues addressed.
**************************************************