Letters Data sharing among trialists

What about trialists sharing other study materials?

BMJ 2012; 345 doi: https://doi.org/10.1136/bmj.e8352 (Published 10 December 2012) Cite this as: BMJ 2012;345:e8352
  1. Rik Crutzen, senior researcher1,
  2. Gjalt-Jorn Y Peters, assistant professor2,
  3. Charles Abraham, professor of behaviour change3
  1. 1Maastricht University/CAPHRI, PO Box 616, 6200 MD Maastricht, Netherlands
  2. 2Open Universiteit, Postbus 2960, 6401 DL Heerlen, Netherlands
  3. 3University of Exeter Medical School, St Luke’s Campus, Exeter EX1 2LU, UK
  1. rik.crutzen{at}maastrichtuniversity.nl

Rathi and colleagues provide insight into clinical trialists’ opinions and experiences of sharing clinical trial data.1 We would like to raise two related matters.

Firstly, this study and previous studies cited within are limited to sharing data. We suggest that questionnaires, intervention manuals, analysis scripts (such as syntax files), output files, and other materials that are used to generate study reports should also be shared. Accurate replication requires these additional materials. This is essential for scientific progress because successful replication strengthens our evidence that a given theory, model, or assumption holds, whereas unsuccessful replication is our only means of disproving them.2 Publication of analysis scripts and output files is necessary because researchers make many important choices that are not disclosed in research articles.3

Secondly, the concerns about data sharing raised by Rathi and colleagues’ respondents can mostly be refuted.4 For example, one concern was the ability to publish own research. When a researcher wants to publish several articles about one dataset, publishing the dataset before all articles are published can be risky, because others might conduct analyses that had been planned for later articles and publish them earlier. A solution is not to publish the entire dataset but only those variables described in published articles. This is related to another identified reason: sufficient academic or scientific recognition. In the rat race of “publish or perish,” it might seem that publishing materials (such as data, questionnaires, and intervention manuals) will help others. However, materials are only published after acceptance of the article, so any race has already been won, or lost.

Notes

Cite this as: BMJ 2012;345:e8352

Footnotes

  • Competing interests: None declared.

References

View Abstract