Maintaining the integrity of the scientific record

We editors of medical journals worry that we sometimes publish studies where the declared authors have not participated in the design of the study, had no access to the raw data, and had little to do with the interpretation of the data. Instead the sponsors of the study—often pharmaceutical companies—have designed the study and analysed and interpreted the data. Readers and editors are thus being deceived. Editors are also concerned that the declared authors might not have ultimate control over whether their studies are published. That decision may rest with the funders of the research—perhaps a government department or a pharmaceutical company—which could mean that results unfavourable to the funders are suppressed. This distorts the scientific record and again deceives readers, allowing them to read only favourable results. Editors have taken steps to counter the problem by revising the uniform requirements for manuscripts submitted to biomedical journals of the International Committee of Medical Journal Editors, and changing editorial practices

It's hard to know how often such problems arise, but they occur against a background of increased entanglement of academia with industry.^1–5 Contract research organisations have developed to help pharmaceutical companies conduct their trials, and these organisations need to develop relationships with doctors to recruit patients.² Twenty years ago investigators outside companies designed trials, but now companies or the contract organisations are more likely to write the protocols. Control lies in the commercial rather than in the academic or public sector, and “companies may design studies likely to favour their products.”² There are many ways in which they may do so.³ The companies may then analyse the data, “providing the spin ... that favors them.”² Theoretically the BMJ's policy on contributorship should expose the problem of authors not designing the trial and analysing the results, because we ask for each author's contribution to be made explicit.⁶ But we remain anxious—I can think of one example from a few years ago of a study published in the BMJ that had a single author but whose data had not been analysed by him.⁷ That study produced unexpectedly positive results but was severely criticised in the correspondence columns.⁸

Problems concerning control of publication have had a much higher profile. Drummond Rennie, a deputy editor of JAMA, has told the now famous story of how Boots went to great lengths to try to suppress a study that showed that its product levothyroxine was not superior to its competitors' products.⁴ The authors came from the University of California, San Francisco, which insists, wisely, that its academics keep control of publication of their papers. Unfortunately in this case the authors did not. The head of the sponsored research office of Massachusetts General Hospital estimates that about 30-50% of contracts submitted by companies have unacceptable clauses on publication that must be renegotiated.² A survey of over 3300 members of life science faculties in 50 universities found that a fifth had had publication of study results delayed by more than six months at least once in the past three years.⁹ One reason for this delay was to slow the dissemination of undesired results. Certainly there seems to be a proliferation of stories of companies suppressing publication,¹⁰,¹¹ despite forceful arguments that failure to publish amounts to research misconduct.¹²

This initiative by journal editors should not be seen as an attack on the pharmaceutical industry. Almost all new drugs are developed by the industry, and many companies have high ethical standards and will see no problem in complying with the new policies. Pharmaceutical companies become successful not through dubious publication or marketing policies but by developing important new drugs. And other groups—including hospitals and governments—are keen to control publication. NHS researchers in England had to fight a major battle at the end of the 1980s to stop the government inserting a clause into contracts that would have given it control of publication.¹³

The journals that are members of the International Committee of Medical Journal Editors, including the BMJ, will now routinely require contributors to disclose details of their own and their funders' roles in the study. We will ask contributors to sign a statement that they accept full responsibility for the conduct of the study, had access to the data, and controlled the decision to publish. If authors cannot satisfy us on these points we will not publish. In this way we hope to contribute to maintaining and improving the integrity of the scientific record.