Authors should make their data availableBMJ 2001; 323 doi: https://doi.org/10.1136/bmj.323.7320.1069/a (Published 03 November 2001) Cite this as: BMJ 2001;323:1069
- Douglas G Altman (), professor of statistics in medicine,
- Christopher Cates, general practitioner
- ICRF/NHS Centre for Statistics in Medicine, Institute of Health Sciences, Oxford OX3 7LF
- Bushey, Hertfordshire WD2 2NN
EDITOR—Hutchon discussed the desirability of publishing the raw data used in medical research articles, and Eysenbach and Sa have outlined some of the difficulties. 1 2 The least that should happen at the moment is that authors should be able to make their data available on a journal's website. One journal that already does this is Clinical Chemistry, which will include data supplements in the material sent to reviewers.3 Journals should encourage authors to post their data on the website.
One area where there should be neither debate nor difficulty is that of systematic reviews. Readers of the review ought to have access to the numerical results of the primary studies being reviewed to allow the analyses to be checked and for other analyses to be investigated. Also, this enables readers to examine the actual results rather than the authors' aggregation or summary; for example, they can assess the variation in event rates across the studies, without which an odds ratio is impossible to interpret.
We have each separately had the experience of authors of systematic reviews published in the BMJ refusing to release these data, in one case to enable the data to be used in an educational article. Such obstruction is worrying and suggests that some suspicion is appropriate where none had existed at the time of the request. Given that in almost all cases the results will already be in the public domain, we can see no valid excuse for not including the data in the report or making them available electronically. The missed opportunity is even clearer in those cases where an extended version of a paper appears on the web page but the trial results are still not given, although in this case the authors have made the data available to us.4
The BMJ and other journals should insist that authors of systematic reviews adhere to one of the key recommendations the QUOROM statement—namely, to present simple summary results for each treatment group in each trial, for each primary outcome.5