Peer review is on every life scientist?s mind lately, it seems. One of the main complaints I heard while researching the February cover story is that the process is inherently difficult to investigate scientifically. Each journal has a somewhat unique system for reviewing papers, and each paper will have a unique journey through a journal?s reviewing machinery.

But I?ve learned that even though peer review has obvious imperfections, it?s the best system we?ve got, and simply complaining about what?s wrong with it doesn?t help matters much. Investigating peer review is difficult, sure, but does that mean we shouldn?t even try?

Kirby Lee, an assistant professor of clinical pharmacy, and his colleagues at the University of California, San Francisco, have spent a year and a half collecting data about the peer review process at The Lancet, Annals of Internal Medicine, and BMJ. The data, also reported in Nature, stem from interviews with editors, peer review comments from more than 1,000 papers, and audio and video recordings of editorial meetings where editors discussed the papers.

Lee told me that his team planned on spending six months collecting data, but had to significantly extend the enrollment period "because the acceptance rate was so low." In fact, out of all the papers included in the study, the journals published only 68.

Some other interesting findings:

-There appeared to be no bias towards statistically significant findings, although Lee cautioned that the same trend may not be true at other journals, or at the same journals during a different time period.

-Major changes editors tend to make between acceptance and final publication include toning down authors? conclusions, and ensuring that the manuscripts disclose funding sources and conflicts of interest.

-The methodological quality is higher in accepted papers than rejected ones.

Lee said that the study was originally going to include JAMA, but the journal dropped out without providing a clear reason why. He and his colleagues decided to conduct the study to investigate complaints of publication bias, which can skew the research record and affect meta-analyses. They focused on biomedical publishing because what?s published can lead to changes in the way medicine is practiced.

In the meantime, Lee and his colleagues continue to analyze the hours of tapes collected from editorial meetings, and have already submitted one paper describing their research, currently under review.

"It?s funny," he said. "We?ve done this study on peer review, and now we?re at the mercy of peer reviewers."