Only a tiny fraction of the compounds tested for different diseases ever make it to clinical trials. Now a report in Science suggests that the results of even encouraging clinical trials are later refuted with surprising frequency.

Scientists from the University of Ioannina School of Medicine in Greece analyzed published studies, from 1990 to 2004, of promising new drug candidates or medical devices. (A sampling is shown at below.) Of 32 interventions described in these papers, each of which had been cited more than 1,000 times, 13 were later shown not to work or to be less effective than originally thought.