But what about the specific issue of disclosure? How could disclosures have been improved in this case so that readers would not have felt that they were being hoodwinked? If the journal had published the authors' financial relationships with Cyberonics, as well as with any other company involved in the antidepressant market, would this have been enough?
Remember, the purpose of disclosure is to give readers the information they require in order to make their own decision about the potential for bias. Typical disclosure statements such as "Dr X discloses that he has consulted with Companies A, B, and C" are a good start, but in many cases they do not tell the whole story. The whole story, in fact, is that different companies pay experts different amounts of money to do different things. Further, in many cases, the identified authors of journal articles are not the actual authors; they merely review and edit the work of an industry-paid professional writer, as was the case in the review article on VNS.
Thus, an argument can be made for 2 new disclosure requirements: authorship disclosures and compensation disclosures.
The problem of ghostwriting
It is clear that a significant proportion of major journal articles—both original research and review articles—are ghostwritten by writers who are funded by the pharmaceutical industry.9 The most well-documented example of this in psychiatry surfaced in the context of a lawsuit against Pfizer. Researchers identified all articles published on the topic of Zoloft (sertraline) from 1998 to 2000. Of the 96 published papers, 55 (57%) were ghostwritten by a medical information company, and 41 (43%) were not. Among the ghostwritten articles, 100% reported clinical results favorable for Zoloft, but only 44% of nonghostwritten articles reported similar findings.10
This study implies that approximately half of all major journal articles published in the field of psychiatry are ghostwritten and that these articles are more likely to present information that is favorable to the sponsoring company, bringing up the possibility of prosponsor bias. Thus, it seems crucial for readers to be fully informed about the authorship of all articles in addition to author affiliations.
Ghostwriting is not necessarily a corrupt practice. Busy researchers would find their productivity hobbled if they had to personally write the first draft of an article for every study they conducted. Experienced medical writers are experts at synthesizing scientific data and communicating it to a wider medical community.
The issue, as always, comes down to the need for full disclosure. In the VNS review, the medical writer who wrote the first draft was mentioned at the end of the paper in an acknowledgment section sandwiched between the article and the references. It stated: "We thank Sally Laden for editorial support in developing early drafts of this manuscript. We maintained complete control over the direction and content of the paper. Preparation of this report was supported by an unrestricted educational grant from Cyberonics Inc."3
To the uninformed, this acknowledgment implies that Laden had an unspecified and presumably minor editorial role, but that the identified authors actually wrote the paper. An "unrestricted educational grant" connotes a process in which authors are paid to write anything they want to write, without any content requirements from the funding company.