The spectral fingerprints of a big drug company have once again been found all over academic publications. Documents released last week by a watchdog group based in Washington DC raise concerns about the role of writers paid by GlaxoSmithKline (GSK) in works attributed to psychiatric researchers at a number of US institutions. They add to the drumbeat of allegations in recent years indicating that such ghostwriting — in which articles contain substantial portions written by someone who is not listed as an author — is endemic in the biomedical literature.

The documents were made available as a result of litigation over GSK's antidepressant Paxil (paroxetine) and were pounced on by the Project on Government Oversight, which raised concerns about authorship of a research article, journal editorial and textbook.

The researchers did acknowledge the alleged ghostwriters of the textbook and the editorial in notes, but only for “editorial support”. For the journal article, which appeared in a supplement to Psychopharmacology Bulletin, GSK is thanked for an “unrestricted educational grant”. But the original front page of the manuscript — which the academic author is instructed to remove before submission to the journal — declares that it was prepared by writers from Scientific Therapeutics Information, a company based in Springfield, New Jersey, hired by GSK. The article and textbook discuss the uses of Paxil. The editorial, in Biological Psychiatry, gives an overview of depression as a major and growing public-health problem — which certainly does no harm to a company aggressively marketing an antidepressant.

The academic authors and the American Psychiatric Association, which published the textbook, have strongly denied that the pharmaceutical giant influenced its content. So, too, have the authors of the editorial and the journal article. GSK shareholders, then, may wonder what the company got for its money. The issue here is not that industry-financed experts cannot write useful and unbiased reports, but that their role must be declared in full. It is for readers, not authors, to conclude that there is no conflict of interest.

All the academic authors involved in this case have been recipients of US National Institutes of Health (NIH) funding; all but one still are. The NIH may argue, rightly, that the ghostwritten publications did not use its money. It will also note, correctly, that this in an issue that demands far broader action. Both are beside the point. Money is fungible, and rarely do the studies and intellectual output of senior researchers divide neatly into industry-funded and taxpayer-funded work. If its grantees are not playing by the rules, the NIH is tarred and public trust is damaged. So, how clear are the rules on ghostwriting? A study last year found that just 10 out of 50 top US academic medical centres had explicit, web-accessible policies that prohibit the practice. Another three banned ghostwriting in practice without naming it as such (J. R. Lacasse and J. Leo PLoS Med. 7, e1000230; 2010).

Discussing the issue of ghostwriting a year ago, Francis Collins, the NIH director, said publicly that he was “shocked” that “people would allow their names to be used on articles they did not write, that were written for them, particularly by companies that have something to gain by the way the data is presented”. Many will share that shock, but, unlike Collins, few are in a position to do something about it. The agency is “considering how best to address and ensure” greater transparency and accountability as its grantees develop and author articles, Sally Rockey, NIH chief of extramural research, told Nature in an e-mail last week.

A good start would be for the NIH to require all institutions that take its funds to articulate, publicize and vigorously enforce a clear ban on ghostwriting. Other funders should follow suit. Without such a clear signal, and the willingness to give a ban teeth, this troubling ghost will linger at the feast.