Are Bayes' Days Upon Us? Statistical Methods Could Change the Conduct of Clinical Trials

Extract

A group of statisticians who dislike the null hypothesis as much as struggling graduate students do are teaming up with clinical researchers to redefine the questions that can be asked and answered in clinical cancer trials, using statistical methods first put forth by an 18th century minister.

Rev. Thomas Bayes' theorem had lofty goals from the time it first arose in the 1760s—a mathematician and theologian who published it after Bayes' death thought the theorem could be used to prove the existence of God. Now it is inching into vogue for some medical research. Proponents say Bayesian methods can tell physicians how much better a drug is than another, rather than merely if it works better, and it will ensure better treatments for trial volunteers. But some skeptics argue that pure Bayesian methods aren't appropriate for drug trials because they explore the effectiveness of drugs rather than confirm that the drugs work.