Cautionary Note on Data Sharing | Circulation: Cardiovascular Quality and Outcomes

Abstract:  Should data from clinical studies be routinely shared? A growing chorus of thoughtful scholars,1,2 backed by public policy3 and supportive requirements from academic journals,4 argue for making deidentified data sets from clinical trials and prospective cohort studies routinely available to outside investigators for use in secondary analyses.5 Indeed, rigorous empirical work demonstrates patients’ willingness to have their data shared,6 while industry and other stakeholders increasingly participate in these efforts.7,8 The pro-data sharing position celebrates potential efficiencies in the expansion of data use and the advancement of transparency while also resting on a powerful, normative ethical claim: data ought to be shared out of respect for the contributions of human subjects and public funding (when applicable).6 However, this article will argue the opposite from an ethical lens: making primary data routinely available from clinical studies does not always serve patients or public health and may even be detrimental in certain circumstances. We build our case first by carefully defining data sharing, questioning its connection to key goals such as research efficiency and transparency, and reviewing arguments both for and against data sharing in research practice.

From the body of the paper: "This potential misuse of public data and complex reactions to published findings are challenging to control and, ultimately, may mislead rather than serve patients’ interests. Our central ethical concern about secondary analyses is thus that, by definition, these papers will be tangential to the design and rationale of the original project and may be subject to selective reporting—either relentlessly positive or alarmingly negative—alongside plausible systemic type I errors from dredging one data set repeatedly....

As with other areas of research and science, we need to balance its potential benefits with true risks, such as inappropriate secondary analyses and a disincentive for conducting future research in new populations and through new data collection methods."