Transparency and trust: Figure for ghost written articles was misquoted

EDITOR—Editor's choice in the issue of 23 October on transparency and trust seems to perpetuate a misleading press citation of my testimony to a House of Commons Select Committee last month.1 The original statement, supported by the transcript, was that 50% of the articles dealing with therapeutics were ghost written, not 50% of all articles.2 3

I, like most readers, almost instinctively shrink from a claim that anything like 50% of the articles, even those on therapeutics alone, are ghost written in journals such as the BMJ, New England Journal of Medicine, JAMA, and the Lancet. But equally instinctively, most readers if asked to estimate how many of the key articles on their drugs, and this means articles in major journals, pharmaceutical companies are likely to have had a determining role in writing, would probably come up with figures close to 100%. If the question is in what proportion of articles on therapeutics in major journals do the apparent academics hold the raw data and are able to share that data if needed, the answer in many estimates will not be much greater than 0%.

Abbasi usefully brings out a point made in the select committee's meeting, that the key problem with ghost writing is not the medical writing itself but the issue of transparency. When there is reason to believe that the articles that result from the ghost writing process do not offer a fair representation of the underlying data there is a problem. Otherwise ghost writing poses much less of a problem.