Sunday, October 12, 2003

Stopping a clinical trial before it's complete.

The New York Times editorialized today about a recent decision to end a clinical trial of a breast-cancer treatment because the results were looking so good, the researchers wanted to get the apparently beneficial treatment out to a larger number of women than just those who were enrolled in the research:
The study was testing the value of administering the drug letrozole to older women who had already taken tamoxifen to ward off a recurrence for about five years, the point at which tamoxifen loses its power. The results exceeded expectations. When compared with a placebo, letrozole cut the recurrence rate nearly in half.
As the Times correctly points out, "[t]here seems little doubt that a trial must be terminated if it is harming the participants"; indeed, that is precisely what data safety monitoring boards do. But, as the editorial observes:
By halting the study in midstream, the researchers made the new treatment available to the placebo group and to many thousands of other cancer survivors as well. The downside is that early termination kept them from determining whether the treatment actually saved lives, how long women should keep taking the drug or the likelihood of such adverse effects as osteoporosis and cardiovascular problems.
The National Breast Cancer Coalition has criticized the National Cancer Institute's decision to halt the clinical trial, and their paper makes a lot of sense. The National Cancer Institute's explanation for its decision, with a link to a Q&A page on the letrozole study, is here. The researchers' article, which is scheduled to appear in the New England Journal of Medicine on November 6, was posted to the journal's website on October 9 "[b]ecause of its potential therapeutic implications." [Abstract available here without subscription]
posted by tommayo, 8:37 AM

Health care law (including public health law, medical ethics, and life sciences), with digressions into constitutional law, poetry, and other things that matter