Scientific communication depends on trust. We should be able to believe what we read, and trust that knowledge when we plan experiments and treat patients. Unfortunately, we cannot. Conflicts of interest are ubiquitous; billions of dollars are being earned undeservedly by drug companies through flaws in research, research articles, reviews, and editorials; and many academic careers have also been built on doubtful evidence. The recent stream of books about the corruptive influence of money on health care research and practice, some of which have been written by editors of our most prestigious journals [1–3], illustrates just how bad the situation is.

Part of the problem is that good names give papers credibility. A colleague once told me that in his country it was more important to know the authors than the methods of a research paper, as some professors lent their names to almost anything if they were well paid. I have seen single-authored meta-analyses on drugs presenting sophisticated analyses that went far beyond the capability of the author, without a word about who did the analyses (and presumably even wrote the paper). Similarly, many drug reviews are unlikely to have been written by the authors, as these professors probably have more important things to do than writing book-length drug reviews in sponsored supplements or peripheral journals that few would ever read and that have no impact factor.

Ghost authorship exists when someone has made substantial contributions to writing a manuscript and this role is not mentioned in the manuscript itself [4]. It often occurs simultaneously with its opposite, guest authorship (sometimes called honorary or gift authorship), where the contributions of the named authors are so small, or nonexistent, that they do not merit authorship [5,6].

Court cases that allowed access to industry files have shown that ghost and guest authorship are common, even in our best journals [5,6]. But misappropriation of authorship is dishonest [4] and is regarded as scientific misconduct in some jurisdictions like Denmark, where the law states that misappropriation of author role, either deliberately or through serious neglect, is a type of scientific misconduct [7]. Regrettably, it is rarely detected. The involved parties have a common interest in secrecy, and junior researchers can ruin their careers if they reveal that the professor did not write the papers that bear his or her name. It is still commonly accepted that department chairs claim authorship of all papers emanating from the department, and newspaper articles celebrating a professor's 60th birthday may note that he or she has written more than 500 papers. The professor may have contributed, but almost certainly wrote only a minority of them.

I have some suggestions that might reduce the prevalence of misappropriated authorship:

Most importantly, all journal articles should list the contributions of the authors [8]. This would make it far easier for authors and editors to object before publication, and to document cheating after publication.

To put it simply: if a cook isn't cooking, what is a cook then doing?

Ghostwriting debases fundamental tenets of the medical profession. It violates authors' personal integrity, responsibility, and accountability. More importantly, ghostwriting threatens the very fabric of science and thus the validity of our medical knowledge, and in doing so it jeopardizes patient care.

Denouncing ghostwriting is easy; defining its variants is not. At the extreme, ghostwriting is easy to define. Perhaps the most egregious example is that of a pharmaceutical company's marketing department promoting one of its products by carefully selecting positive reports and deemphasizing the product's risks, and then paying a well-known academic author to submit the paper for publication without attribution [6]. Few would disagree that this behavior is inappropriate, unprofessional, unacceptable, and potentially dangerous. Lesser degrees of ghost involvement in the writing process are also problematic, such as permitting the commercial sponsor of a clinical trial to collect and hold all of the trial's data, providing exclusive statistical expertise and final tables for publication, drafting a complete manuscript of a study it supported, or insisting contractually on the last say on the final manuscript's content and conclusions.

Other kinds of ghost involvement are more ambiguous. Is it acceptable to hire a science writer to interview a physician and write a paper on that subject, which the physician then calls his or her own? Is it appropriate for a scientist in a company to analyze a portion of evidence, write a draft of the information, and then for another author to incorporate that draft into a manuscript without crediting the scientist? Is it acceptable for a physician-researcher to pay someone to do the same? How much help with writing is okay? Where do we draw the line in some of these fringe areas of ghostwriting?

Because many interactions between academics and industry in developing, testing, and reporting on new products are desirable, such authorship definitions are critical. Unfortunately, there is not a great deal of uniformity among journals (the major publishing gatekeepers) in what constitutes an acceptable contribution. The guidelines of the ICMJE and WAME provide a good start [4,10]. They define an author as someone who has made substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data; has drafted the article or revised for important intellectual content; and has given approval of the final version. More stringent requirements may be necessary, however. Some journals require that authors have had full control of the primary data, have carried out the statistical tests themselves (e.g., JAMA), and have created tables and figures themselves. Others require that each author state the specific contributions they have made to the study or to the writing of the manuscript, and some journals publish these (e.g., BMJ, PLoS Medicine).

Other journals review components of study protocols that are vulnerable to manipulation by the sponsor and will only publish clinical trials that are registered in a public database endorsed by the ICMJE. Increasingly, journals require that the role of the study sponsor be transparent in manuscripts, and in addition will not accept papers unless the decision to publish is controlled by the researchers. Finally, one former editor, Richard Smith, thinks that the entire peer review system should be scrapped because of excessive industry influence on publishing [12]. He would replace the current system by an open, Web-based disclosure of protocols, data, and statistical assessment, with publication only of systematic reviews based on the study data [13]. These tactics go some way toward stemming ghost involvement in the publication process.

Any additional “cure” for ghostwriting must take into account its frequency and impact. Information from questionnaire studies suggests that authorship in up to 10% of published papers could be attributed to ghostwriters [14–17], although the fraction in industry-sponsored clinical trials in one study was considerably higher [5]. In trying to understand the prevalence, we are stymied because we just don't know what we don't know. Moreover, most of the data in these studies are self-reported, and the exact frequency of company-inspired writing is well hidden. And finally, the impact of ghostwriting is even more difficult to estimate. For this reason, we must be careful not to impose excessive regulations to solve problems that may not be threatening.

Nonetheless, editors of medical journals could devote more effort to define what constitutes appropriate and inappropriate participation in studies and manuscript preparation, especially by companies with products at stake. At the very least, editors can demand transparency. Who were the trial designers? Who were the trial conductors, the researchers, and the data managers? Who did the statistics? Who wrote the manuscript and who signed off on the final draft? Can all authors take public responsibility for their roles?

Overtly biased ghostwritten articles can cause patient harm; others damage the public's trust in both the pharmaceutical industry and the medical profession. Loss of trust may be ghostwriting's major victim. Neither the industry nor the profession can afford further damage to their reputations. Both should “just say no” to ghostwriting.

Professional medical writers have been recognized by medical journal editors as legitimate contributors to manuscripts [4,10], but concerns remain about “ghostwriting,” namely the failure to disclose such contributions. Professional medical writers, who must be distinguished from ghostwriters [18], could be valuable allies to those determined to eradicate ghostwriting. Professional medical writers have communication expertise and health care knowledge, and abide by ethical guidelines for medical writers. They assist authors to prepare documents (e.g., abstracts, slides, posters, and manuscripts), but ensure that the authors control the content and that appropriate disclosures of funding and involvement are made. We believe that professional medical writers can offer unique, and too frequently untapped, insight into how to address ghostwriting. We offer the perspective of professional medical writers on three strategies that have been considered for tackling ghostwriting.