December 9, 2016: Data Sharing in Pragmatic Trials

December 9, 2016: Data Sharing in Pragmatic Trials


Data Sharing in Pragmatic Trials: Balancing Transparency and Health System Privacy


Greg Simon, MD, MPH, Senior Scientific Investigator, Group Health Research Institute

Lynn DeBar, PhD, MPH, Senior Investigator, Kaiser Permanente Center for Health Research

Bill Vollmer, PhD, Senior Investigator, Kaiser Permanente Center for Health Research

Edward Septimus, MD, Medical Director, Infection Prevention and Epidemiology Clinical Services Group, Hospital Corporation of America


  • Data sharing; Data analysis; Transparency; Privacy; Healthcare systems; Pragmatic clinical trials; Embedded trials; Research ethics; Data use agreements; HIPAA; Human subject protection
    Key Points
  • While patient privacy is an essential consideration when implementing data sharing, there are also potential privacy concerns for healthcare systems:
    - Disclosure of proprietary business information
    - Increasing potential liability due to gaps in care of individuals
    - Reputational risk regarding perceived gaps in overall quality
  • The 3 levels of data are (1) what investigators can touch: original health system data; (2) what investigators can use: limited by consent or waiver; and (3) what investigators can share: further limited by regulation (and possibly health system policy).
  • Different data-sharing structures, from least to most restrictive include:
    - Public archive, where data is sent to any user
    - Private archive, where data is sent to approved users for any purpose or only for approved purposes (governed by a data use agreement)
    - Public enclave, where any user may query the data
    - Private enclave, where some users may query the data (executed automatically or moderated)

Discussion Themes

What are the distinctions in privacy protection for individual patients, healthcare systems, and pharmaceutical sponsors of research?

Could more data be shared using a more controlled method? How can we assure long-term availability of knowledgeable staff to support such a resource?

One recommendation is to establish supervised, private data-sharing enclaves to allow outside investigators to perform analyses without taking control of the data.

We need more evidence about the utility of archived data so we can build the case for secondary use. Which types of datasets have been reused by unrelated investigators and yielded useful info?

For More Information

Read more about HIPAA guidance for deidentification and privacy at

#DataSharing, #pctGR
@PCTGrandRounds, @Collaboratory1, @PCORnetwork