Science and Technology CommitteeWritten evidence submitted by Stephen Senn BA, MSc, PhD, CStat

Declaration of Interest

1. I consult regularly for the pharmaceutical industry. I maintain a full declaration here The views expressed here are my own and should not be ascribed to any organisations with whom I am associated.

Statement of Experience

2. I am an experienced medical statistician who has worked for the National Health Service, the Swiss pharmaceutical industry and has held two chairs at British universities, including one in Pharmaceutical and Health Statistics at University College London (1995–2003).1


3. I have long maintained that data from clinical trials sponsored by the pharmaceutical industry should be available not only to regulatory agencies but also to patients and prescribers. For example, in an article entitled Statistical Quality in Analysing Clinical Trials published in 2000 I wrote:

“The results that were needed to convince a regulator are precisely those that in an ideal society we would expect subscribers and reimbursers to want also. No sponsor who refuses to provide end-users with trial data deserves to sell drugs.”[1] (p26).

4. Nevertheless I am dismayed that in inviting comment on this important issue the Science and Technology Committee (STC) has been prepared to become part of the publicity machine of Bad Pharma,[2] a badly researched and highly biased, if well-written, book by Dr Ben Goldacre that is misleading in many respects, in particular in its discussion of drug regulation. For example a paper[3] that Goldacre cites as proving bias of FDA panellists in favour of applications in which they have an interest shows the opposite of what he claims[2] (p126). If the way in which STC understands scientific issues with important policy implications is through reading misleading and inaccurate polemics, this is a sad reflection on the place of science in British public life.

5. The book is quite wrong to imply that the regulators do not do a good job. They do a much better job than the medical press and it is necessary for the Select Committee to appreciate this in order for it to understand that the medical press cannot be any part of an effective solution to the problem of missing data. Despite Goldacre’s assertions to the contrary (see his claim on p34), the medical press is biased against negative studies. (See my recent papers[4,5] on the subject to understand how Goldacre has misunderstood the relevant literature.) The medical press is also very slow to retract incorrect and misleading articles, including those containing false data, as the recent scandal involving Duke University clearly illustrates. (See Baggerly and Coombes for a full exposé of this story.[6]) Furthermore published articles rarely provide the data that enable a thorough check of their claims. Even reviewers are not provided these data, as I know having reviewed for such journals for many years. Thus any solution to the problem of missing clinical trials data should not involve the medical press.

6. Furthermore, responsibility for publishing results is divided between authors and editors. Although it is a necessary condition for a trial to be published for authors to prepare a paper and submit it, it is not sufficient, since any given journal may refuse it. A system is necessary in which those who apply for permission to run a clinical trial are the publishers. In this way they can be made entirely responsible for the successful conclusion of their publishing obligation.

7. Thus, rather than relying on journals, I propose that a web-based system of publishing trial results should be used. In fact we should move towards a system where the results of clinical trials are always made available on the internet and this becomes the primary means of communication, with medical journals limited to publishing commentaries. Many of the leading medical journals have high rejection rates and also embargo presentation of results of a trial accepted for publication by them until the journal publishes them. The combination of these two features adds delay and uncertainty to the business of publishing.

Specific Recommendations

8. For regulatory trials as part of the drug regulatory process sponsors should be required to provide a publishing plan as to how the data will be made available on the internet to all interested parties. This should be part of the dossier submitted to the regulator. Many research councils make similar requirements that a dissemination plan be part of any grant application.

9. It should be part of the drug regulatory process to make sure that this plan is considered adequate. In other words in addition to the Quality, Safety, and Efficacy requirements there should be a Dissemination requirement.

10. Marketing approval for a drug should not be granted until the sponsor has demonstrated that the publishing plan has been fulfilled.

11. Mere publishing of an article in the medical press should not be considered an adequate alternative to fulfilment of a publishing plan by publication on the internet.

12. For non-regulatory trials. As part of any submission to an ethical committee for clinical trial approval any submitting party should be required to provide:

(a)A publishing plan with an undertaking to make the results available on the internet.

(b)A statement of all previous applications that have ever been made to any ethical committee with an explanation as if and how any previous obligations have been met.

13. On no account should the medical press be regarded as part of the solution to the problem of missing data. A system needs to be developed that is completely independent of medical journals.

February 2013


1. Senn, S J, Statistical quality in analysing clinical trials. Good Clinical Practice Journal, 2000. 7(6): p. 22–26.

2. Goldacre, B, Bad Pharma: How Drug Companies Mislead Doctors and Harm Patients2012, London: Fourth Estate. 430.

3. Lurie, P, et al, Financial conflict of interest disclosure and voting patterns at Food and Drug Administration Drug Advisory Committee meetings. JAMA: the journal of the American Medical Association, 2006. 295(16): p. 1921–8.

4. Senn, S J, Misunderstanding publication bias: editors are not blameless after all F1000Research, 2012. 1(59).–59/v1

5. Senn, S J, Authors are also reviewers: problems in assigning cause for missing negative studies. F1000Research, 2013. 2(17).–17/v1

6. Baggerly, K and K Coombes, Deriving chemosensitivity from cell lines: Forensic bioinformatics and reproducible research in high-throughput biology. The Annals of Applied Statistics, 2009. 3(4): p. 1309–1344.

1 Details of my qualifications and experience are available at

Prepared 16th September 2013