Clinical trials in cancer

Randomized controlled trials (RCTs) in cancer are expensive, slow, and often preclude other exploratory investigations that might be valuable. Two types of RCTs should not be conducted. When the results of exploratory studies are so dramatically different from all prior experience, RCTs should be done to improve upon the new finding but not to allocate some patients to inferior control arms to see enough contemporaneous failures to "prove" the superiority. The other preclusion are RCTs that aim for so small or unimportant a likely benefit that the outcome will not be worth the effort. A study not worth doing is not worth doing well. RCTs in cancer are best used to establish moderate differences, suggested by exploratory trials, that have a real chance of advancing the field. If the entry criteria and end points are sufficiently explicit, e.g., biopsy proof of cancer and death, observer bias will not be a major factor.