Should Peer Review be Confidential?

The current edition of Science magazine has an editorial by Donald Kennedy recommending that peer reviews remain confidential. It recounts the story of a legal action by a drug manufacturing firm seeking access to peer reviews for articles concerning one of its products. The New England Journal of Medicine, which is the subject of the action, guarantees reviewers anonymity, and is resisting showing the reviews to the company.

I spent many years managing peer review processes, and thinking about them and have a few comments.

• I have found it useful to tell reviewers that I would forward the text of their reviews directly to the scientist making the reviewed submission, and that the text should be professional and suitable for professional communication among peers. After all, it is the original author who will have to refute the criticisms, revise his/her work, and/or go back to the bench to do better next time. In fact I tended to edit the reviews to try to assure such professionalism.
• I have also found it useful to offer the reviewers that I would not inform the reviewed of the reviewer identity. Of course, sometimes the identity can be guessed from the nature of the review.
• It is important that if you can not in fact guarantee confidentiality, you should not tell reviewers that have a guarantee. Reviewers are at best underpaid for their work, and more often do reviews pro bono, and for that reason it is especially important not to mislead them.
• Sometimes reviewers feel the need to submit their own data with a review. Doing so provides a basis for the editors to continue an interchange with original authors on their submissions. I can understand that sometimes that data should be held confidential. It might, for example, not yet have gone thru its own peer review process and been published, and thus open to misinterpretation if made public. It might alternatively be commercially proprietary. It seems to me that the reviewer should not be offered confidentiality unless such a justification is made. It also seems to me that the law should allow for confidentiality in selected cases.

The case of epidemiological data about the efficacy and/or safety of pharmaceuticals has an ethical aspect, in that there are public health issues. I wonder if in such cases, the court might appoint a scientific advisory panel to review the confidential data, under bonds of confidentiality, to assure that patients would not be endangered by its being withheld.