Randomized clinical trials are pivotal in determining whether regulatory authorities should approve a new drug for commercial distribution. They may serve to probe into the biology of a disease. They may provide estimates, however rough, of the overall reduction in morbidity or mortality that might be expected if a successful test treatment were to be implemented on a wide scale. But if queried about the most critical role of clinical trials, then most physicians would probably answer that they help guide treatment decisions for the individual patient. And here one runs into a major conceptual obstacle. How can clinical trials actually inform a specific choice of therapy?
Physicians have long recognized that simply assigning a diagnosis does not help much in estimating a patient’s risk of running into future trouble. Most illnesses exhibit a wide range of behaviors, so it seems natural to look to clinical trials for information about specific subgroups of patients in the hope that subgroup-specific estimates of treatment effects, rather than the overall effect, will predict more accurately the effect on specific individuals. In theory, a particular intervention might work better in less sick patients than in very sick ones. On the other hand, if a new effective therapy has certain troublesome or life-threatening toxicities, then the balance of risk and benefit might be more favorable if the patient has a late, rather than early, ...