Using forensics to reveal medical ghostwriting

NEW YORK (Reuters Health) - Medical editors are growing increasingly frustrated with top researchers who sign their names to manuscripts but fail to disclose the contributions of ghostwriters paid for by pharmaceutical companies. In an effort to crack down on the practice -- widely viewed as unethical -- one tech-savvy editor has been turning to data forensics worthy of a crime investigation drama.

The presence of such authors is not uncommon. A survey presented this week at the Sixth International Congress of Peer Review and Biomedical Publication in Vancouver found that 12 percent of research articles in top medical journal studies failed to disclose authors who contributed substantially to the work. Those additional authors sometimes have financial ties to pharmaceutical companies, and critics say the resulting papers often interpret data about those companies’ drugs more favorably.

Frederic Curtiss, editor-in-chief of the Journal of Managed Care Pharmacy, told Reuters Health that data attached to documents by Word has allowed him to discover undisclosed contributors. In one case, for instance, a revised manuscript arrived at his office with four named authors, but when he examined the metadata, he discovered an additional author was making substantial contributions.

When documents are saved in Word, the software attaches additional information, called metadata, which identifies the creator of the document. During the editing process, changes made by additional authors are also sometimes labeled with authors’ names. Curtiss estimates that every third manuscript he receives has metadata that doesn’t match listed authors, which can subsequently result in contributors being added to the acknowledgments, or, rarely, as additional authors.

Curtiss’ journal is not alone. Recently released court documents revealed that drugmaker Wyeth enlisted ghostwriters to pen 26 papers appearing in 18 scientific journals between 1998 and 2005 touting hormone replacement therapy in women. Those papers played up the benefits of the therapy in women for preventing heart disease and dementia while downplaying the potential risks, such as increased risk of breast cancer, heart disease, and stroke.

In response to these revelations, Senator Charles Grassley, an Iowa Republican, has demanded that the National Institutes of Health provide information on their policies regarding ghostwriting.

Every medical journal has its own editorial policies, and Curtiss claims his journal has long had one of the tougher policies around. He requires any authors who contribute more than 1% to the manuscript to be disclosed, and those that contribute more than 25% of a manuscript must be a listed author. “Disclosure does not correct or overcome the quality of the evidence,” Curtiss says, “but it is critical for a reader to interpret the research, so it needs to be accurate and complete.”

Consequently, improving the journal’s disclosure statements has been a chief topic of conversation at every editorial meeting he has attended for the last eight years. These statements have long required authors submit a list of every person who made a contribution to the research, but in the last two years, they have added a seemingly redundant question requiring authors to explicitly vouch that “every contributor” to the research has been identified.

Readers of his journal will primarily be determining which drugs are most cost-effective for inclusion in managed care plans, which means articles can have a significant influence on pharmaceutical profits. However, he cannot always count on authors to provide accurate disclosures, which is why he is often checking for authorship clues in the metadata.

In what is probably a first for the journal, the June issue featured an article where a medical writer shares authorship with a scientist. “Much to their consternation, we listed an author who did not want to be listed,” Curtiss says.

The article was on the Eli Lilly and Company drug Effient (Prasugrel) for preventing atherothrombosis and was prepared with financial support from the pharmaceutical company. The published version of the article lists Sarah Spinler of the University of the Sciences in Philadelphia and Catherine Rees, a senior medical writer for Adis Communications.

According to the published disclosure statement, “Rees performed the majority of data collection and writing of the initial draft, and both authors shared equally in the revision.”

However, Curtiss says Rees did not wish to be a co-author. He would not say how he learned of Rees’s involvement in the manuscript and whether it was disclosed upon submission.

Rees, who is based in Auckland, New Zealand, and is a member of the Australasian Medical Writers Association, confirmed over the telephone that she was not initially listed as an author on the paper but declined to answer any further questions.

Unfortunately, says Curtiss, he is losing the lead in the battle against ghostwriting. Sometimes, the metadata includes only the word “Author” or a number. “Now people are savvy about it,” Curtiss says. “We are getting manuscripts where they have purposely removed data from the identification box.”