Richard Smith: Advice to the NEJM on dealing with old influential articles with undisclosed COIs

richard_smith_2014At the end of last year JAMA Internal Medicine published a study that showed that the authors of two highly influential papers published in the New England Journal of Medicine in 1967 had undisclosed but important conflicts of interest. The two review papers demolished the idea that sugar might be important in coronary heart disease and supported strongly the hypothesis that saturated fat was the main cause. It now emerges that the authors, all from the Department of Nutrition at the Harvard School of Public Health, had undeclared funding from the Sugar Research Foundation. There have been calls for the papers to be retracted, but the NEJM has dismissed the calls and published nothing about the undeclared conflicts of interest. What should it do?

The NEJM has a simple case for dismissing the calls for retraction. It—and medical journals in general—didn’t start asking that conflicts of interest be declared until 1984. So neither the authors nor the journal did anything wrong by the standards of the time.

I agree that it would be an over-reaction to retract the articles, and to publish a short note declaring that the authors had undeclared conflicts of interest would achieve little and set an awkward precedent in that no doubt many of the papers published in the journal since it started in 1812 have had undeclared conflicts of interest.

The obvious thing to do to my mind would be to publish an article or series of articles that examines the many interesting issues thrown up by this case—and for all I know the journal is planning such a response. Simply ignoring the issue enhances the journal’s reputation for high-handed arrogance.
The most important issue thrown up by the case is the continuing and highly contentious debate over the relative contributions of sugar and saturated fat to coronary heart disease. The articles that failed to disclose the conflicts of interest were crucial in favouring saturated fat as the cause and dismissing sugar, and it now looks as if the conclusions of the articles were at least partly wrong. I’m sure that the readers of the NEJM would appreciate both an unbiased state of the art review on the subject and a re-examination of the arguments of the Harvard authors.

Then, the whole story of saturated fat versus sugar has been tied up with conflict of interest. In crude terms the meat and dairy industry was pitched against the powerful corporations producing and using sugar. The stakes could hardly have been higher, and both sides used the techniques made familiar by the tobacco industry to influence research and academic and public views. Yet interestingly the academic or intellectual conflicts of interest were probably more powerful than financial conflicts of interest. D M Hegsted and F J Stare, two of the authors of the NEJM papers, had committed themselves strongly to the saturated fat side; and their influence extended way beyond the articles in the NEJM to deciding what sort of research should be funded (shutting out those sympathetic to the sugar hypothesis), advising government, and drawing up the first national dietary guidelines. Readers of the NEJM would appreciate some serious analysis of the networks of conflicting interests.

The mechanics of how the Sugar Research Foundation was involved are another interesting issue. The Foundation funded the authors to the tune of $6500 (equivalent to about $50 000 now), but probably more importantly it “set the review’s objective, contributed articles for inclusion, and received drafts.” If the NEJM had known these facts it probably would have declined to publish the article even by the standards of the time. The journal was being used, and the price of being as prestigious as the NEJM is that the unscrupulous (and actually the scrupulous a well) will go to great lengths to exploit that prestige.

And interestingly the authors of the NEJM articles do unusually for the time declare funding from six sources, one of which was “the Special Dairy Industry Board.” It’s as if by declaring a conflict that runs contrary to their main conclusion they will strengthen belief in that conclusion.

The editors of the NEJM could incorporate these specific points around conflict of interest into an article on the journal’s turbulent journey in relation to conflict of interest. Although the journal—along with other members of the International Committee of Medical Journal Editors—required authors to disclose conflicts of interest from 1984 it didn’t do well with disclosing them to readers. A study that I did with Amina Hussein showed that in 1999 the NEJM disclosed conflicts of interest in none of 56 editorials, four of 26 research articles, and none of 113 letters. Other major journals were little better, but we know that most of the authors of research articles and letters would have had conflicts of interest.

The authors of editorials might not have had conflicts of interest because from 1990 the editors of the NEJM refused to publish editorials from authors who had conflicts of interest. Unfortunately this made it difficult to publish editorials on some subjects and also led to authors failing to disclose conflicts of interest, as an investigation by the LA Times showed. When Jeff Drazen, the current editor, became editor of the journal in 2000 he soon allowed editorials to be written by authors with “insignificant” financial conflicts of interest ($10 000 a year), so drawing the fire of one of his predecessors. Drazen later went against the current of the times and suggested that journals had become over concerned with financial conflicts of interest. And we never hear from the NEJM (or other journals) over its massive conflict of interest in making large profits from restricting access to research funded with public money.

Then the NEJM editors might also like to consider the articles that failed to disclose the conflicts of interest in relation to the letter it published in 1980 claiming that only four of 40 000 patients taking powerful painkillers had become addicted. This one paragraph letter has been cited some 600 times and been accused of fuelling the current US epidemic of opioid addiction. In this case the editors have responded, and the 1980 letter is now preceded by this Editor’s Note: “For reasons of public health, readers should be aware that this letter has been “heavily and uncritically cited” as evidence that addiction is rare with opioid therapy.” Might the NEJM editors do something similar for the nutrition articles—not least because they may well have been important in fuelling America and the world’s obesity epidemic.

Finally, the NEJM editors could have contributed their views on the debate raging around the current vogue for removing statues of once famous but now infamous figures from the past—Cecil Rhodes in Cape Town and the confederate soldier in Durham, North Carolina. How far should we go in “correcting” history?

My view is that the statues and the NEJM articles should stand but be used as a focus for debate, and I urge the NEJM editors to do so.

Richard Smith was the editor of The BMJ until 2004.

Competing interest: None declared.

  • Stuart Worsley

    Excellent thought piece on the same day that Stephen Hawking rebuts Jeremy Hunt’s accusation of “pernicious falsehoods” stemming from invalid use of data.

    Why is it that peer review systems can so easily collapse in the face of political pressure? What can we do to assure effective peer review?

  • jhnoblejr

    Richard Smith once more identifies past mistakes and ways to leverage them to make progress.. Yes, the mistakes of the past should become the focus for debate and discussion of remedies. We can measure progress by reference to Richard Smith’s recommended actions.

  • jhnoblejr

    In answer to your question, change human nature. We are an authority-ridden species that runs in a herd. Political regimes count on the dynamic to maintain and extend power . . . witness what is going on all around us.