Dr. Edelman writes that ADVANTAGE could not have been a seeding trial because the Annals editors decided to publish it, which would imply that the trial had scientific value. We knew about the VIGOR (Vioxx Gastrointestinal Outcomes Research) trial (1), which was published in 2000 and compared Vioxx with naproxen in patients with rheumatoid arthritis. VIGOR showed that Vioxx had significantly fewer gastrointestinal side effects than naproxen. We published the ADVANTAGE trial (2) because it framed the same findings in a different way: It showed that the margin between the new drug and the comparator—also naproxen—was narrow (5.9% discontinuation rate with Vioxx vs. 8.0% with naproxen). We thought this information could help physicians decide which drug to use to start treatment. We thought the scientific value of the finding was modest. We thought its expository value was considerable.

What was the company's prime motivation for ADVANTAGE? Although the company denies that ADVANTAGE was a seeding trial, we remain convinced that a principal purpose was to promote the adoption of Vioxx by physicians who participated in the trial. First, we interpret the language of the company's internal messages—excerpted in the article by Hill and colleagues but available in full online—to mean that a principal purpose was to market Vioxx. We believe that most readers would agree with this interpretation. Second, it is our opinion that the company would not incur the expense of a trial involving 5500 patients and 600 physicians just to confirm what VIGOR trial had already shown for patients with rheumatoid arthritis: that patients with osteoarthritis would tolerate Vioxx better than naproxen, the primary hypothesis listed in the study protocol distributed at an ADVANTAGE Study Investigators Meeting in 1999 (3).

Was it important to know if ADVANTAGE was a seeding trial? We think so. If it was a seeding trial, patients agreed to participate without knowing something that might have led them to decide against participating—that one of the company's principal motivations for the trial was to accustom their physicians to using Vioxx. Is the rationale for a study important? We think it is, because the main reason that patients agree to participate in clinical research is for the potential benefit to themselves and to mankind (4). We believe that many patients would not participate in a trial if they knew that it tested a hypothesis that had already been convincingly addressed and that a principal study objective was to market a new drug.

In the end, the discussion about seeding trials is a discussion about maintaining trust in a system—clinical trials of new drugs—that allows a few people to benefit millions by agreeing to participate in a human experiment. We argue that participants need to know everything that might affect their decision to participate. And so does everyone else whose cooperation is necessary to do the research.