We fully support open access to clinical trial data. At the same time, it is important to avoid the unintended consequence of creating barriers to delivery systems’ and providers’ participation in embedded research. Different solutions will be required for sharing data that supports embedded research compared to those developed for conventional randomized trials. A central issue is who decides which uses are allowable. To motivate organizations to opt in to embedded research, a framework that delineates what can automatically be shared (questions that refine or deepen the original research question, such as sub-sets or secondary outcomes) from what must be considered by a panel of stakeholders (re-purposing the data for a new hypothesis) would be useful. Uses of the data that require provider attributes would require agreement from the organizations that originally agreed to participate in the research. Furthermore, this framework could ensure the data can be used to support the public good without jeopardizing the individuals or organizations whose data are at risk of inappropriate use.

For certain secondary analyses, making limited datasets available through public or private archives may be worthwhile. However, analyses of distributed datasets that remain in the possession of clinical organizations will likely require these organizations to execute analyses for investigators. It will be necessary to ensure appropriate technical infrastructure to support such work. This infrastructure and the personnel to support it will incur substantial costs.

In our article in Annals of Internal Medicine, we encourage researchers planning and leading embedded research to consider the same questions asked of the NIH Collaboratory Trials:

• “What data could be shared by the least restrictive mechanism, a public archive open to any interested user?
• What additional data could be shared using a more restrictive mechanism (private archive, public or private data enclave)?
• Would the scientific or public health benefit of sharing additional data justify the additional effort to establish a more restrictive data sharing mechanism? (Simon et al. 2017)"

Data sharing policies for embedded research that support important research can be developed with a concerted effort of all stakeholders, and we must balance potential for harm with the ethical imperative to be more transparent and share data. The development of data sharing plans that are the most useful will be the least restrictive while providing appropriate protection for participant privacy, health system privacy, and scientific integrity.