The Game of Ghost Writing

When Sen. Charles Grassley raised concerns last month about scientific "ghost writing," in which papers are produced by companies or other parties whose names do not appear as authors, he acknowledged that he was unsure how widespread the practice was. A pair of studies presented Thursday at a scholarly meeting in Vancouver offer evidence of both the depth and the bread of the controversial activity.

The Sixth International Congress on Peer Review and Biomedical Publication, which began Thursday and is sponsored by the Journal of the American Medical Association and the British Medical Journal, featured a series of panels on some of the most vexing issues facing scholarly publishing in the biological sciences, many of which have become bound up with broader concerns about corporate influence and potential conflicts of interest. Among the topics: data sharing and conflicts of interest, "publication bias," and the quality and preparation of peer reviewers.

One panel on Thursday presented a series of studies that examined issues surrounding authorship and contributorship to journals, with a particular focus on authorship that isn't. One study, by a trio of researchers at the University of California at San Francisco, documents one particular instance of ghost writing from the past -- a pharmaceutical company's carefully planned process in the late 1990s to market a drug (Neurontin) that treats nerve pain and some aspects of epilepsy. The scholars, UCSF's Lisa Bero, Jenny White and Maureen Boland, based their study in part on a review of documents from the university's Drug Industry Document Archive.

As laid out by the UCSF authors, a company called Medical Education Systems, Inc., solicited and received a grant of $160,000 from the Parke-Davis pharmaceutical company (now part of Pfizer, Inc.) in 1997 to produce a series of articles to be published in scholarly journals. Medical Education Systems paid the "authors" but played a central role in writing and submitting the articles, the UCSF authors found.

Of the 24 articles that Medical Education Systems proposed producing, the scholars were unable to find any record of 11. Six were published in the journals in which the companies originally sought to place them; seven were published in alternative journals, in one case because the original target rejected the study, Bero said in an interview.

The journals' publishers in most cases had no idea about the heritage of the articles in question. Of the 13 articles that were published, one author disclosed that he or she had received an honorarium from Medical Education Systems and one disclosed receipt of a grant from Parke-Davis; none noted the fact that the companies had played any role in authorship of the studies.

While many of the journals at the time had general policies requiring disclosure of conflicts of interest, fewer than half had statements specifically about the criteria for defining "authorship," and none explicitly prohibited funding for ghost writing, the authors found. Now, a decade later, change has been minimal; virtually all have general conflicts policies, but slightly more than half have set criteria for authorship and just 15 percent explicit bar ghost writing.

"A lot of these journals had authorship policies, but you just don't capture ghost writing unless they have a really explicit policy about it," Bero said. "Unless you ask a really explicit question like, 'Was this article ghost written?' these types of situations will slip through the cracks."

By exploring a situation from a decade ago and showing that the circumstances that helped bring it about have changed little, the study by the UCSF researchers suggests that a campaign like the one it examines could unfold today. Another study presented Thursday shows more directly that the practice of ghost writing has not vanished.

The authors of the study, "Prevalence of Honorary and Ghost Authorship in 6 General Medical Journals, 2009," are editors at JAMA; they surveyed authors of 900 articles published in 2008 in JAMA and five other major journals (Annals of Internal Medicine, Lancet, Nature Medicine, New England Journal of Medicine, and PLoS Medicine) to ask them about the degree to which they had been ghost or "honorary" authors (the latter being those who do not meet the standard definition of having contributed meaningfully to the publication).

The authors use as their starting point data from 1996 -- around the time that Bero and her co-authors examined -- showing that 19 percent of articles had honorary authors, 9 percent had been ghost written, and 2 percent had both honorary and ghost authors. In their 2008 survey, the JAMA researchers reported, 26 percent of articles had honorary authors, 8 percent ghost writers and 2 percent both honorary and ghost contributors.

The authors describe the decline in ghost authorship as "significant," although they acknowledge that the practice is "still a concern." That is probably especially so at the journals where authors reported more significant prevalence of it; 11 percent of authors who'd published in the New England Journal of Medicine, for instance, reported having had ghost writers.

Bero, the UC-San Francisco professor of clinical pharmacology, said that ghost writing won't diminish unless and until journals adopt stronger policies against it. Clearly stated policies, like the journal Pharmacotherapy's requiring authors to say whether their "paper has been ghost written by an individual sponsored by a pharmaceutical, biotechnology or medical company or other entity," are one important step, she said.

But at her session in Vancouver Thursday, Bero said, she reiterated the paper's recommendation that journal editors "verify" the disclosures that authors make, through tools like the Drug Industry Document Archives or the expertise of their peer reviewers or editors. The journal editors at her session "pushed back" against that idea, Bero said, not because they were philosophically opposed but because of concerns about demands on time and resources.