A Ghost Appears

The opening scene of Shakespeare’s great tragedy Hamlet takes place at night outside Elsinore Castle. Rumors of a ghost have the sentries, Bernardo and Francisco, on edge. Francisco, apparently mistaking Bernardo for an intruder, calls out, “Stand and unfold yourself,” a phrase that could just as well be addressed to the ghost of Hamlet’s father, who will shortly appear. Shakespeare may be implying that the person (or ghost) must have wrapped himself in layers of fabric in order to conceal something important, an identity or secret that will only be revealed when he obeys the command to “unfold yourself.” Indeed, the shocking secrets that the ghost soon unfolds to Hamlet will drive the course of the play, and lead to the ruin or death of most of the protagonists. As Marcellus comments in Scene IV of Act I, “Something is rotten in the state of Denmark.”

Haunted Research

Now fast-forward four centuries to the April 16, 2008, issue of the Journal of the American Medical Association (JAMA). Four physicians—Joseph S. Ross, MD, MHS; Kevin P. Hill, MD, MHS; David S. Egilman, MD,MPH; and Harlan M. Krumholz, MD, SM—publish an article entitled, “Guest authorship and ghostwriting in publications related to rofecoxib: A case study of industry documents from rofecoxib litigation.” As paid consultants to the plaintiffs’ attorneys, Dr. Ross and colleagues had access to a mass of internal Merck documents that were subpoenaed in connection with litigation concerning adverse effects of Vioxx. They found evidence that articles reporting clinical trials of Vioxx were often authored principally by individuals other than the ultimate first or second author of the published paper. Academics were recruited to be the lead authors as part of a planned strategy that involved ghostwriting by either Merck employees or contracted medical writers. In parallel, review articles were initiated by Merck, directly or through medical writing companies, and were ghostwritten. Recruited authors received honoraria, and such relationships were not fully disclosed. An online database of the documents that were analyzed was made available to allow independent validation of the authors’ conclusions. Unfortunately, several members of the ACR are among the physicians implicated.

The editors of JAMA clearly view the paper of Ross et al. as highly important. In a lead editorial with the sweeping title, “Impugning the integrity of medical science: The adverse effects of industry influence,” Catherine D. DeAngelis, MD, MPH, JAMA editor-in-chief, and Phil B. Fontanarosa, MD, MBA, executive deputy editor, pull no punches.

They call the behavior of at least some of the physicians identified by Ross et al. “unprofessional and demeaning to the medical profession and to scientific research.” They then enunciate a series of proposals to rescue the integrity of clinical research and medical education, clarify the roles of medical writers, and penalize authors whose disclosures are incomplete. An especially interesting position of Drs. DeAngelis and Fontanarosa is that pharmaceutical companies should be prohibited from providing support for medical education, considering that the April 16th issue of JAMA contains several pages of drug ads, including the back cover of the journal. Notwithstanding this one paradox, their message is clear: Something is rotten in the state of medicine.

These publications in JAMA have attracted extensive media coverage. The New York Times headlined its article on this subject, “Merck wrote drug studies for doctors,” and quoted Dr. Ross as saying that “it almost calls into question all legitimate research that’s been conducted by the pharmaceutical industry with the academic physician.” Another article published on ChicagoTribune.com was entitled, “Drugs are hyped by ghostwriters, not sound science.” And now Senator Chuck Grassley (R-Iowa) has launched an investigation.

When Is a Ghost Not Really a Ghost?

The article by Dr. Ross and colleagues deserves careful examination. It does clearly document examples of ghostwriting and inappropriate ordering of authorship, examples that I was able to confirm upon searching the public database of documents that Ross et al. analyzed. However, when I probed this database with the names of some of the ACR members whose articles were implicated, I found disturbing discrepancies between the available data and the conclusions of the JAMA article. For example, regarding a review article on COX-2 inhibitors authored by a single ACR member, the only document in the database that contains this author’s name is a grant request (from someone unrelated) that cites an article by this author as a reference. And this grant request had been submitted to a different pharmaceutical company a decade ago, and had nothing to do with rofecoxib or any other anti-inflammatory drug. Why this document was even in the database is not at all obvious. And why this ACR member’s review article was listed by Ross et al. in a table with the daunting title, “Published Financial Disclosures among 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,” is equally not apparent.

In another case, searching the database with the name of an ACR member who was the single author of a review article described as mentioned in documents “demonstrating that medical publishing companies played critical roles in overseeing the development, organization, and manuscript drafting of supplemental issues focused on rofecoxib” yielded no hits. Likewise, a search using the title of the article yielded no hits. While the supplement in which this article appeared may have had industry sponsorship, no evidence is available to suggest a role for Merck in this specific article.

These examples result from searching the database used by Ross et al. for only a small fraction of the authors and articles that they implicated in the JAMA article, which suggests to me that additional individuals may also have been incorrectly identified as being involved in ghostwriting and related practices. The New York Times quoted Steven Ferris, PhD, a New York University psychiatry professor as stating that the allegation that his work was ghostwritten was “simply false.” Presumably, any errors in the study of Ross et al. were inadvertent, and due to a flawed strategy for analysis of the voluminous documentation that was collected for Vioxx litigation. But these examples beg the question: How many of the accusations are true and how many are not?

What Do We Do Now?

Let me be clear about my own views on ghostwriting: it’s an unacceptable practice that has no place in the scientific and medical literature. The same goes for manipulation of data to falsely enhance perceptions of the efficacy and safety of pharmaceuticals. And there is, unfortunately, little doubt that these practices are not unusual. The medical profession has become far too comfortable with “that affable familiar ghost which nightly gulls him with intelligence,” mentioned by Shakespeare in Sonnet 86.

So, given the importance of these issues, surely Ross et al., as well as the JAMA editors who highlighted their article and extended its conclusions, should have understood the necessity to ensure that their facts were entirely correct—100% correct. Their study, amplified by JAMA’s commentary, is not just a general discussion of the prevalence and undesirability of ghostwriting. It also has the effect of seriously compromising the integrity and professionalism of every single one of the authors who was cited, not to mention potentially making each of them the target of a Congressional investigation. Casting a broad net that traps the innocent with the guilty can reasonably be viewed an act of negligence, even recklessness that—just like ghostwriting—has no place in a medical journal. As in Shakespeare’s plays, good intentions aren’t good enough. After encounters with his father’s ghost, Hamlet, in his zeal to achieve vengeance, accidentally murders the wrong person: the innocent, albeit bombastic, courtier Polonius. Hamlet’s tragic error occurs because he fails to exercise enough care to ascertain the identity of his victim, and instead mistakes him for Hamlet’s evil uncle, who had murdered Hamlet’s father, usurped his throne, and married his widow.

Any damage that’s been done in the pages of JAMA will need to be repaired. If an original article based on laboratory research was found to contain two or three erroneous experiments due to a flawed data collection and analysis strategy, the likely outcome would be retraction of the whole article until the entire body of work could be repeated using a better protocol. It will be up to the editors of JAMA to decide whether that standard should also apply to the type of study done by Ross et al. At a minimum, sincere and prominent corrections and apologies will need to be extended to all those who have been inappropriately tainted.

Meanwhile, physicians, scientists, and their professional organizations should think seriously about the next steps we need to take. Why not upgrade authorship standards for clinical trials, review articles, and, indeed, all of the biomedical literature so that accusations of manipulation and ghostwriting would never be correct? If the ACR’s code of conduct is not sufficiently explicit on these issues, shouldn’t we revise it? In my view it would be a mistake to let any errors in the JAMA article deter us from a serious look in the mirror about these important issues.

Disclosure: My columns aren’t ghostwritten, but I am grateful for review by and input from various ACR staff members and volunteers, most notably Tammy J. Tilley, ACR’s senior director of communications, and Mark Andrejeski, ACR’s executive vice president.

Dr. Fox is president of the ACR. Contact him via e-mail at [email protected].