Too soon to marketBMJ 1998; 316 doi: https://doi.org/10.1136/bmj.316.7126.228 (Published 17 January 1998) Cite this as: BMJ 1998;316:228
Statistical aspects of research done outside pharmaceutical industry could be improved
- Stephen Senn, Professor of pharmaceutical and health statistics (email@example.com)a
- aDepartment of Epidemiology, Department of Statistical Science, University College London, London WC1 6BT
- bDepartment of Medicines Management, Keele University, Keele, Staffordshire ST5 5BG
- cCentre for Hip Surgery, Wrightington Hospital, Wigan, Lancashire WN6 9EP
- dUniversity Hospital, Queen's Medical Centre, Nottingham NG7 2UH
- e10908 Piney Meeting House Road, Potomac, MD 20854, USA
Editor-Dent and Hawke write: “The unpublished data seen by the licensing authorities have not been scrutinised by the scientific community and may not have been peer reviewed, which limit their suitability for use in prescribing and funding decisions. The same concerns also apply to their use in licensing decisions.”1 They thus repeat a common prejudice. The truth, however, is quite the opposite of what they believe.
The following seven standards apply to licensing applications from the pharmaceutical industry.2 (1) The method of analysis used will have been prespecified in the trial protocol. (2) The analysis and planning of the trial will have been performed with and by a qualified statistician. (3) All aspects of the trial will have been carried out under “good clinical practice,” covered by detailed standard operating procedures. (4) The reviewer will have access to all data. (5) Site monitoring inspections will have been carried out. (6) Every trial will have been reviewed in detail by professional reviewers. (7) Every aspect of the trial will have been handled by highly qualified staff specifically trained for the task in question.
The BMJ is a leader in the field, and I do not question its excellence. Nevertheless, as someone who has occasionally reviewed papers for it for their statistical soundness, I doubt that most of the papers that it publishes, if from clinical trials outside the industry, would satisfy the drug regulator as regards points (1) to (5) above. About points (6) and (7) one might argue.
Dent and Hawke's concern about the way in which the results of clinical research filter through to prescribing doctors is valid (but applies equally to results of research outside the pharmaceutical industry). One solution would be for reports of clinical trials to be available on the world wide web for …