Guest Authorship and Ghostwriting in Publications Related to Rofecoxib: Title and subTitle BreakA Case Study of Industry Documents From Rofecoxib Litigation

Authorship in biomedical publication provides recognition while establishing accountability and responsibility. Guest authorship has been defined as the designation of an individual who does not meet authorship criteria as an author.^{1 - 2} It was identified in 16% of research articles, 26% of review articles, and 21% of editorials in a survey of 6 peer-reviewed medical journals,³ in addition to 41% of Cochrane reviews.⁴ Ghostwriting has been defined as the failure to designate an individual (as an author) who has made a substantial contribution to the research or writing of a manuscript.¹ Ghostwriting was demonstrated in 13% of research articles, 10% of review articles, 6% of editorials, and 11% of Cochrane reviews^{3 - 4} ; other research has found similar rates.⁵

Two studies have characterized the practices of guest authorship and ghostwriting using industry documents, one examining practices related to gabapentin by Pfizer Inc and Parke-Davis, Division of Warner-Lambert Company,⁶ the other sertraline by Pfizer Inc.⁷ However, these studies were focused on how the research and publication strategy of the companies was used to promote and market the products. No studies have used internal documents to characterize the role of authorship in collaborations between industry and the medical profession.

Recent litigation against Merck & Co Inc related to rofecoxib provided a unique opportunity to examine the practice of guest authorship and ghostwriting related to the research and promotion of this medication. Our objective was to provide a review using a case-study exploration of court documents, in tandem with a review of the medical literature, to describe the practice of guest authorship and ghostwriting related to rofecoxib.

Documents used for this article are posted at http://dida.library.ucsf.edu.

In the course of the combined trials of Cona vs Merck and Co, Inc (No. ATL-L-3553-05, New Jersey Superior Court, Atlantic City) and McDarby vs Merck and Co, Inc (No. ATL-L-1296-05, New Jersey Superior Court, Atlantic City), millions of documents were made available to and archived in an integrated database maintained by the plaintiff's attorneys. These documents were created between 1996 and 2004 and included Merck internal and external correspondence, reports, and presentations. As consultants to attorneys on the behalf of plaintiffs, we had complete access to all archived documents. One investigator (J.S.R.) searched the database to extract a subset of documents related to authorship (Figure 1). The search was performed using the database keyword search function and included the following search terms: clinical trial, author, authorship, review, manuscript, and publication, along with terms encompassing the names of Merck scientists, the names of academically affiliated authors of clinical trials, the names and numbers of clinical trials, the names of medical publishing companies, and the names of journals. The search identified approximately 20 000 documents that included 1 or more of the keyword terms. Document numbers are an approximation because information within one document may overlap with another, making it difficult to determine the exact number of distinct documents. For example, 1 document may include a string of 2 e-mails, whereas another document may include a string of 5 e-mails, including the prior 2.

One investigator (J.S.R.) searched the documents identified using the authorship keywords to determine if each was actually related to or discussed authorship, examining the document titles and the content within the database. “Related to or discussed authorship” refers specifically to examination for authorship of manuscripts describing nonpharmacological, human participant clinical trial results or scientific reviews (or journal supplements) that included an external, academically affiliated (non-Merck employee) author. Approximately 250 documents were identified, the majority of which were Merck internal correspondence and publication reports, along with external correspondence between Merck and medical publishing companies.

Two investigators (J.S.R. and K.P.H.) reviewed these 250 documents using the principles of grounded theory, an inductive approach in which source material was used to generate ideas rather than to test a preestablished hypothesis.⁸ This method has been applied to study issues at the intersection of litigation and health, particularly with tobacco,^{9 - 10} and more recently with pharmaceutical⁶ products. We first reviewed the documents to identify broad themes reflecting the practice of ghostwriting and guest authorship. Next, pertinent documents were reviewed again by all of the authors, using a negotiated consensus process to reach our final interpretation. This process ultimately generated a single agreed-upon set of themes, as well as documents and quotations to illustrate each theme.

After determining themes, 2 investigators (J.S.R. and K.P.H.) again reviewed these 250 documents to identify articles describing clinical trial results or scientific reviews (including journal supplements) discussed internally within Merck prior to publication that proposed an external, academically affiliated investigator as an author. No documents were excluded as part of this search; all documents were related to 1 or more of the identified manuscripts (most manuscripts were discussed within >1 document). The published articles were subsequently identified via MEDLINE. This search was supplemented with MEDLINE queries for other rofecoxib-related articles authored by academically affiliated investigators identified as first authors within documents, which were found by searching for the author's name and “rofecoxib” or “cyclooxygenase inhibitor.” This search identified 96 published articles. Importantly, identification of these articles does not imply that each was guest authored or ghostwritten; we examined these articles because we believed their discussion within internal documents (or the discussion of specific authors) suggested that Merck was aware of the manuscript prior to publication and perhaps had provided support for the project.

All published articles were categorized as to whether a manuscript's coauthor was affiliated with Merck (ie, a Merck employee), whether the published article included any financial disclosure, whether the published article included a financial disclosure of Merck support, and whether the published article included a financial disclosure of Merck support by at least 1 of the academic authors. For articles in which there was neither a published disclosure of Merck financial support nor a published disclosure of Merck financial support by 1 of the academic authors, other articles by the academic authors published within 2 years of the relevant article were examined to determine if they had disclosed Merck financial support.

This research was deemed exempt from normal review by the Yale University Human Investigation Committee.

Review of internal documents and the published literature revealed 3 key findings related to guest authorship and ghostwriting: the first focused on the publication of clinical trials, the second focused on the publication of review papers, and the last was related to financial support disclosures.

When publishing their own clinical trials (designed, conducted, and sponsored by Merck), documents were found describing Merck scientists often working to prepare manuscripts and subsequently recruiting external, academically affiliated investigators to collaborate on the manuscript as guest authors. For instance, trial 078 (a randomized, double-blind study to investigate whether rofecoxib could delay the onset of Alzheimer disease in patients with mild cognitive impairment) was designed and conducted principally by scientists at Merck. Figure 2 shows the title and author list both from draft¹¹ and published¹² versions of the manuscript describing the trial. Both the title and the authorship were modified to attribute authorship to 3 academically affiliated investigators (first, second, and third authors) on the published article, in addition to the 8 Merck scientists who are attributed authorship on both the draft and published versions of the manuscript (1 Merck scientist is attributed authorship on the draft but not the final manuscript). Of note, only 1 of the 3 academically affiliated investigators who are attributed authorship on the published article was acknowledged in the draft version as a participating investigator in the rofecoxib 078 study group. In an internal e-mail discussing where to publish trial 078 as the draft is circulated, one of the Merck scientists states, “I think you should be the first author since you have done virtually all of the writing.”¹³ Although there are minor differences in language and organization between the draft and final versions of the manuscript (particularly in the abstract, as opposed to the text), the results presented are almost identical, reinforcing that the trial itself and the analyses were complete before the academically affiliated investigators were involved in the manuscript.

This same pattern occurred for the manuscript describing the Merck protocol 901 studies, which compared the efficacy of rofecoxib and naproxen in Asian and European populations. An e-mail written on behalf of members of Merck's publication committee to a Merck scientist states that the European study had been prepared as a manuscript and that a draft was shared with the European authors, in addition to describing 2 Merck employees who will prepare the manuscript describing the Asian study.¹⁴ However, the final publication describes both trials in a single article and lists neither of them as authors.¹⁵

Documents were found describing other examples of Merck recruiting external, academically affiliated investigators to collaborate as guest authors on manuscripts prepared by Merck scientists. The first author of the Assessment of Differences Between Vioxx and Naproxen to Ascertain Gastrointestinal Tolerability and Effectiveness (ADVANTAGE) study¹⁶ described to a New York Times reporter in 2005, “Merck designed the trial, paid for the trial, ran the trial . . . Merck came to me after the study was completed and said, ‘We want your help to work on the paper.’ The initial paper was written at Merck, and then was sent to me for editing.”¹⁷ The academically affiliated authors of the Vioxx GI Outcomes Research (VIGOR) study,¹⁸ in response to an expression of concern by the New England Journal of Medicine,¹⁹ make a point of asserting that no Merck employee or representative was involved in the drafting of their response, but do not discuss who drafted the manuscript and with respect to cardiovascular events allude to not developing the analysis plan, not having access to the data, and not performing the analyses.²⁰ Merck's performance of the analyses was confirmed by the Merck-affiliated authors.²¹

A Merck publications status report identifies several of the early clinical trials conducted by the company that were eventually published for which data were available (or would soon be available), including Merck protocols 010, 029, 033, 034, 035, 040, 041, 044, 045, 050, 058, 068, 072, 085, 088 (VIGOR), 090, 097, 098, 102 (ADVANTAGE), 103, 120, 121, 901, and 902.²² For each of these study protocols, a Merck employee is designated within the report as the author of the first draft of the manuscript. Examining the published articles, the first author is an external, academically affiliated investigator for 16 of 20 articles (some protocols were combined into single articles),^{15 - 16 ,18 ,23 - 35} with the exception of protocols 010, 029, 058, and 072,^{36 - 39} 3 of which had external, academically affiliated investigators listed as authors, but not in the first authorship position. Among these 16 articles, all had 2 or more external, academically affiliated investigators who were attributed authorship (median, 4.5; range, 2-10) and these authors occupied 77% of the first, second, and third authorship positions (37 of 48). Of note, the Merck employee designated to be first author in the Merck publications status report is attributed authorship in 14 of these 16 articles (88%), most often as the final author.

Not all manuscripts were prepared independently by Merck before inviting an academically affiliated author; documents also were found describing Merck contracting with medical publishing companies to have manuscripts prepared. For instance, Figure 3 displays a letter from representatives of Scientific Therapeutics Information to Merck employees presenting the completed draft of a contracted manuscript for rofecoxib protocol 116.⁴⁰ Scientific Therapeutics Information describes itself on its Web site as “a full-service medical publishing group specializing in the development of scientific literature and other resource media with direct application to clinical therapeutics that has been serving members of the pharmaceutical industry and medical associations since 1985.”⁴¹

Documents were found describing Merck employees contracting with medical publishing companies to ghostwrite review manuscripts focused on rofecoxib and subsequently recruiting external, academically affiliated investigators to be guest authors. For example, Figure 4 displays an e-mail from representatives of Scientific Therapeutics Information to Merck employees providing an update on the development and estimated delivery dates for 8 manuscripts related to rofecoxib that the company was preparing, including intended titles, authors, and journals.⁴² Review articles were identified by 7 of the 8 investigators listed in Figure 4, several with titles nearly exactly as proposed.^{43 - 49} In addition, Figure 5 displays a contract for Health Science Communications Inc to provide a 20-page review manuscript with 6 figures or tables intended for a cardiology audience for Merck at a cost of $23 841.00.⁵⁰ Contracts also were identified between these 2 parties to provide review manuscripts intended for nephrology and primary care audiences,^{51 - 52} as well as for other medical specialties. Health Science Communications Inc describes itself on its Web site as “a full-service health marketing communications company committed to the highest quality of service . . . We're there pre-launch, preparing the market for a product's introduction. At launch, we establish the foundation for product uptake.”⁵³

Documents were found demonstrating that medical publishing companies provided near complete drafts of review manuscripts to authors for editing, in addition to managing submissions and revisions. For instance, in preparing one manuscript, representatives from Scientific Therapeutics Information indicate in a publications status report that the first draft was sent to Merck and the company was awaiting comments, but an author needed to be invited.⁵⁴ In another e-mail that discusses an article with which the company was involved, a Scientific Therapeutics Information representative states:

“The .1439 journal article that was submitted to Pharmacotherapy by Dr. William Garnett has been accepted (I believe) with revisions. He has faxed me only the reviewers' comments, but is mailing me the entire packet that they sent to him. He would like us to make the revisions, as he is too busy at the moment to make them himself. According to the proposal (Doc # 66468) there is no mention of whether revisions are included, or can be done for an additional fee.”⁵⁵

Documents also were found demonstrating that medical publishing companies played critical roles in overseeing the development, organization, and manuscript drafting of supplemental issues focused on rofecoxib for journals.^{46 ,49 ,56 - 75}

Documents were found describing Merck compensating investigators with honoraria for agreeing to serve as authors on review manuscripts ghostwritten on their behalf by medical publishing companies. Honoraria varied, ranging from $750 to $2500. One author refused his honorarium from Scientific Therapeutics Information stating, “I really do not feel it is appropriate to be paid for this type of effort.”⁷⁶

There were 96 relevant published articles including 24 clinical trials and 72 reviews (Table). Of the 24 clinical trials, 22 (92%) published a disclosure of Merck's financial support. Of the 72 reviews (38 of which published any financial disclosures, and 34 of which either did not require or did not publish financial disclosures), 36 (50%) published either a disclosure of Merck sponsorship or a disclosure of whether the author had received any financial compensation from the company.

Among 24 nonpharmacological, human participant clinical articles, all 24 included at least one coauthor who was a Merck employee. Nearly all (n = 22) published a disclosure that the trial was supported by Merck, including 7 that also published a disclosure of a financial relationship between Merck and an academic author of the article. No financial disclosure was published for the academic authors of the remaining 17 articles, which may indicate that they did not receive or that they did not disclose receiving financial compensation.

Among 72 scientific review articles, 50 (69%) were solo-authored by an academic physician and 2 (3%) included at least 1 coauthor who was a Merck employee. Of these 72, 21 (29%) published a disclosure that the review was supported by Merck, usually through an unrestricted educational grant, 14 (19%) published a disclosure of a financial relationship between Merck and an academic author of the review article, and 1 (1%) published a disclosure that the author had no financial relationship with Merck. However, for 15 of the 35 review articles (43%) that published a disclosure of Merck support, the disclosure was provided either as part of an introductory editorial describing the supplement's contents or on a separate “disclosure” page; a disclosure statement was not identified when any of the articles were individually accessed electronically [accessed by the authors on February 27, 2008].

Among the 36 of 72 review articles (50%) in which there was no published disclosure of Merck support or a financial relationship between Merck and the academic author, 24 of 36 articles (67%) were authored by at least 1 investigator who published a disclosure of a financial relationship between himself/herself and Merck within another article published within 2 years of the review article. Moreover, several others were authors of clinical trials sponsored by Merck, although no disclosure of a financial relationship between himself/herself and Merck was published. In addition, while none of the review articles from one journal's supplement disclosed financial relationships between Merck and the non-Merck employee authors of the review article, communication between representatives from Scientific Therapeutics Information describes an honorarium offered to the authors in payment for their service.¹²⁸

This case-study review of industry documents related to rofecoxib demonstrates that Merck used a systematic strategy to facilitate the publication of guest authored and ghost written medical literature. Articles related to rofecoxib were frequently authored by Merck employees but attributed first authorship to external, academically affiliated investigators who did not always disclose financial support from Merck, although financial support of the study was nearly always provided. Similarly, review articles related to rofecoxib were frequently prepared by unacknowledged authors employed by medical publishing companies and attributed authorship to investigators who often did not disclose financial support from Merck.

The limited nature of our source material for this case-study review prevented an exact determination of the contributions of recruited authors to the overall design and conduct of the clinical trial and/or the preparation of manuscripts. Although we reviewed in excess of 20 000 documents produced during the consolidated rofecoxib litigation, we were frequently unable to identify versions of manuscript drafts dated before and after external, academically affiliated authors had been recruited. In addition, we cannot exclude contributions by authors made by telephone or in person that would not be identified by reviewing documents obtained through litigation. However, the instances for which we did identify before and after manuscript drafts, such as for protocol 078, we found scant documentary evidence that the recruited authors were involved in the design or conduct of the study or made substantive contributions to the manuscript beyond minor editing. Participating only in minor editing does not meet authorship criteria of the International Committee of Medical Journal Editors (ICMJE).¹²⁹ In addition, we could not determine how often ghostwriting and guest authorship actually occurred, whether the contracted manuscript drafts from medical publishing companies were used, or if the proposed payments (honoraria) were provided. Nevertheless, although we cannot conclude that each of the external, academically affiliated investigators attributed authorship for their respective trial or review article made no substantive contributions to the study design or manuscript preparation, the authorship pattern observed within these documents suggests there was a widespread practice of inappropriately attributing authorship to academic authors and a failure to disclose relevant financial relationships.

Several issues should be considered in evaluating this study. Although every effort was made to present this information objectively and fairly, it is important to note that all of the authors of this article have been compensated for their work as consultants/expert witnesses at the request of plaintiffs in litigation against Merck related to rofecoxib. In addition, relevant documents may not have been identified in our review, despite our use of a systematic method with a comprehensive and exhaustive search strategy to minimize missed documents. However, we believe that while our review may not be sensitive, it was specific. We do not think that we missed documents that would negate the totality of our findings.

Finally, this case-study review is based on documents from a single company related to a single medication. We cannot determine if the authorship pattern we observed for clinical trial and review articles related to rofecoxib also would be observed in articles describing other Merck products or the products of other pharmaceutical companies. However, given the reported prevalence of guest authorship and ghostwriting among the most prestigious medical journals^{3 - 5} and that similar authorship patterns were identified using documents produced during litigation surrounding both gabapentin and sertraline,^{6 - 7} it is reasonable to expect that the authorship practices observed in this case study may be used by other pharmaceutical companies as well. A recent press account seems to confirm as much,¹³⁰ as does the presence of an industry specializing in medical writing.^{41 ,53} Because Merck has traditionally characterized itself and its conduct as among the most ethically appropriate of pharmaceutical companies,¹³¹ perhaps the practices we observed are conservative in comparison with other companies within the industry. Nevertheless, access to industry documents through litigation presents a rare opportunity to explore the relationship between the medical profession and the pharmaceutical industry and has provided valuable insights and findings in the past.¹³²

The medical profession must determine how to interpret and respond to these examples of guest authorship and ghostwriting, conduct that the World Association of Medical Editors has described as dishonest and unacceptable¹³³ and that erodes the ethical foundation of medicine and medical research.¹³⁴ Our case-study review suggests that the practice of inappropriately attributing authorship was common. However, we cannot be certain of the specific actions of individuals, both by those active in academic medicine and those employed by Merck. Perhaps academic authors just permitted themselves to be listed as authors, perhaps they did a substantial amount of editing and simply should have disclosed the actual writer as a coauthor. Moreover, we cannot be certain of the actions of journals. Each journal likely differs in its policies regarding authorship and financial disclosures; we assume that every journal expects that the primary author of an article makes substantive intellectual contributions to the paper, which may include conception of the project, design and conduct of the trial, responsibility for the data and analysis, or drafting of the manuscript, and discloses all other individuals who substantially contributed to the article.

We are hopeful that our findings encourage discussion of ways in which to improve the integrity of research. The medical profession and the pharmaceutical industry should agree that collaborations must be conducted with the highest standards.¹³⁵ We suggest that academic researchers consistently provide to the journals the author contributions for all manuscripts, including original research, meta-analyses, reviews, and commentaries, and disclose relationships and support from all industry sources, regardless of the journal's requirements. Authors who “sign-off” on or “edit” original manuscripts or reviews written explicitly by pharmaceutical industry employees or by medical publishing companies should offer full authorship disclosure, such as, “drafting of the manuscript was done by representatives from XYZ, Inc; the authors were responsible for critical revisions of the manuscript for important intellectual content.” A coordinated oversight strategy involving academic physicians, journal editors, and industry representatives is necessary to discourage both guest authorship and ghostwriting and improve the integrity of the biomedical authorship system.