Sunday, February 15, 2015

Research funding and conflicts of interest

The Guardian reports Susan Jebb's response to a BMJ piece about the food industry funding research into food and nutrition topics.  Professor Jebb is chair of the Public Health Responsibility Deal, and the Guardian reports several people who defend her independence, which she herself asserts.   The three articles referenced in the BMJ piece appear to be peer reviewed articles.  The ICMJE Conflicts of Interest form (found via the EQUATOR site) indicates that authors need to declare any associations with entities relevant to the work.  Do peer reviewers need to take this into account?   Nothing I have found indicates they have to.  Do they in fact know who funded the research they are reviewing?   Would it give them information that might help them identify who wrote the research?   They do of course have to critique the methodologies used.   If the funders had any influence over the research, would it show up in the choice of methodology, the presentation of results, or the discussion of those results.  Interestingly, Behind the Headlines and CASP checklists both appear to make no reference to who funded the research being critiqued.

The World Association of Medical Editors have a document about conflict of interests, which describes what authors, editors and reviewers have to do, but reviewers' COI seems to relate to conflicts of interest with the authors, or their field of research, rather than whether they shoudl know or take into account the authors' COI regarding sources of funding.  

EQUATOR has guidance on industry sponsored research, largely about using industry "ghost" writers to produce the papers, but it does include links to guidance for pharmaceutical companies.

Another Guardian piece reports that research into plain cigarette packaging in Australia, which was publicised by the tobacco industry as showing no effect, and which was funded by a tobacco company, has been criticised "on the BMJ website".  Actually, the piece concerned is not on the BMJ website but a peer reviewed article published online in the journal Tobacco Control (which is a BMJ journal).

This research was not peer reviewed.  I (along with at least one commenter on the Guardian site) wondered why not. But, looking for the paper online reveals it is a working paper, published by the Department of Economics at the University of Zurich.  My limited understanding (I am a medical librarian, really!) of economics literature is that "working papers" are an important way to publish research in economics, and are discussion papers, preliminary results, not peer reviewed before publication.  Rather, they are  in effect peer reviewed after publication.

So, the Tobacco Control article is perhaps part of that post publication review.  The Guardian link to it, but you will need to pay or have a subscription to read the full article.   The University of Leicester has a subscription, so I have seen the full thing.   The Tobacco Control paper authors say that the Zurich authors were provided with the data (from an Australian market research company) by the tobacco company in question, which I can't immediately see in their working paper, but their arguments seem really to be with the statistical methods employed by the Zurich authors.  That would perhaps have been picked up by peer reviewers, but is now being picked up by this post publication review of a working paper.

Perhaps accepting funding will restrict the places where you can publish (Tobacco Control, for example, will not accept work funded by the tobacco industry).  I do think that research funders should not be able to influence what the researchers find or report.  That should be made clear to them when the funding is accepted and presumably is the sort of thing that your institution would advise you about when considering whether to accept funding or not.  

However, there is a Cochrane Review (Lundh A, Sismondo S, Lexchin J, Busuioc OA, Bero L. Industry sponsorship and research outcome. Cochrane Database of Systematic Reviews 2012, Issue 12. Art. No.: MR000033. DOI: 10.1002/14651858.MR000033.pub2), which suggests that industry sponsorship does in fact influence results.

More research needed on my part, perhaps. 

No comments: