Publication ethics: not just an industry problem

Much is written about ghostwriting in the medical literature. To be clear, ghostwriting is unethical, and pretty much everyone agrees about that. However, it seems to be a worrying trend that many of those who are most vocal in condemning it are guilty themselves of the practices they condemn.

My last blog post mentioned Senator Grassley’s report on ghostwriting. It appears that he didn’t actually write the report himself, despite putting his name on the front. I have asked him to clarify, but have  had no reply. Draw your own conclusions. It looks like Senator Grassley thinks ghostwriting is evil if other people do it, but it’s OK if he does it himself.

I’ve just read a couple of articles published in the journal Bioethics that are in a similar vein. The first article makes the case that ghostwriting is equivalent to plagiarism. That’s a reasonable point. You could maybe dispute some of the subtleties of that argument, but the main thrust of the argument, that ghostwriting and plagiarism are ethical breaches of similar severity, seems pretty sound to me. However, after that, the article has some serious problems. It makes the point that ghostwritten articles have the potential to be biased and to distort the scientific record. No doubt they do, but it’s ironic in that case that there should be so much biased and inaccurate information in the article.

The main problem with the article is that it (rightly) criticises ghost and guest authorship in very strong terms, but then seems to maintain that this is only a problem of industry sponsored research. Very odd that, when one of the articles it cites to demonstrate how common inappropriate authorship practices are actually investigated researchers funded by the NIH, rather than industry. I’m not aware of any research showing that inappropriate authorship practices are more common in industry than in academia. So to make the totally non-evidence-based leap of faith to say that it’s entirely an industry problem is hardly in the spirit of writing accurate and unbiased literature.

In fact, one piece of research, which looked not at inappropriate authorship, but retraction of research from the published record as a result of misconduct coming to light, found that such retractions were less common in industry sponsored papers. Interestingly, having a paper written by an acknowledged medical writer (not to be confused with a ghostwriter) seemed to protect against misconduct. Taken in the light of that research, the idea that misconduct is somehow only an industry problem seems deeply misguided.

Citation of statistics in that paper showed a rather cavalier attitude as well. One statistic that was cited is that 40% of published articles for a particular drug may be ghostwritten. This statistic comes from a deeply flawed back-of-the-envelope calculation in a previous paper, which used totally different methods for calculating the numerator and denominator from which that percentage was derived, and is therefore likely to be wildly inaccurate (as well as being based on a single drug and publications over 10 years old, and therefore extremely unlikely to be generalisable to the present day). Did the author not bother to check where that statistic came from? If he is worried about biased information in publications, then checking where your statistics come from seems a pretty basic thing to do.

The paper also states, without citing any evidence, that ghostwriting is a “continuing if not growing problem”. If he’d bothered to look up the evidence, he would have found not only my own research, but also research from the editors of JAMA, both of which have shown that ghostwriting is becoming less common.

Perhaps the author should have hired a professional medical writer to help write his paper (with appropriate acknowledgement, of course). That might have helped to weed out some of the biases and come up with a more accurate paper.

The second article criticises industry publications for, among other things, duplicate publication and ‘salami slicing’ (ie publishing the same material in only slightly changed forms multiple times). A bit ironic then, that this article covers remarkably similar ground to 2 previous publications (here and here) by one of the authors. The paper also cites the same 40%-of-papers-on-a-new-drug-ghostwritten statistic (even going so far as to claim that it is “reasonable to conclude” it’s generalisable to all new drugs), and I can see this one becoming widely cited and rapidly gaining the status of a classic urban myth.

Some ways of combating ghostwriting are proposed in both articles, but none seems very practical or likely to be implemented, for example suggesting that all clinical research on drugs be done by government agencies. I believe drug development controlled by government was tried in the Soviet Union, and I don’t believe it was very successful.

It is disappointing that neither paper mentioned one small but very practical and easy to implement step that I and some colleagues published last year: a checklist that journals could incorporate in their instructions to authors. That could be implemented tomorrow if journals wished, and would require very little in the way of extra resources.

Let me be clear again: I’m not defending ghostwriting. It’s unethical. We’d never do it here at Dianthus Medical. I welcome any attempts to crack down on it. However, I don’t think that simply ranting about it and falling prey to just the sort of biases that anti-ghostwriting initiatives are supposed to eliminate is helpful.

About the author

Adam Jacobs

set up Dianthus Medical in 1999. He is an experienced medical writer and statistician, has a PhD in organic chemistry from the University of Cambridge and an MSc in medical statistics from the London School of Hygiene and Tropical Medicine. You can follow him on Twitter @dianthusmed

Leave a comment

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>