This document discusses best practices for sharing research data. It begins by noting that data sharing is often required by journals, funders, and professional societies. It then provides guidance on several key aspects of planning for data sharing: 1) Do not promise to destroy data or limit its future use or access. Instead, plan to use a reputable data repository. 2) Consider risks of re-identification and invest in de-identification. Get consent to retain and broadly share data that explains the benefits and protections. 3) Work with your IRB to incorporate data sharing plans and get approval for previously collected data where consent was silent on sharing. Factors like feasibility of recontact and sensitivity of data can increase

1. Michelle N. Meyer, PhD JD
Assistant Professor Faculty Co-Director
Associate Director, Research Ethics Applied Behavioral
Insights Team
Center for Translational Bioethics & Steele Institute for Health
Innovation
Health Care Policy
IRBS, ETHICS
& DATA
SHARING

3. Why share data?
• Journal requirements (Science, Nature)
• Funder requirements (NIH, PCORI)
• Society norms (APA)

4. APA Certification of Compliance With APA Ethical
Principles (2001):
After research results are published, psychologists do not
withhold the data on which their conclusions are based
from other competent professionals who seek to verify the
substantive claims through reanalysis and who intend to
use such data only for that purpose, provided that the
confidentiality of the participants can be protected and
unless legal rights concerning proprietary data preclude
their release.

5. Why share data?
• Journal requirements (Science, Nature)
• Funder requirements (NIH, PCORI)
• Society norms (APA)
• Tax payer-funded research
• Enable reanalysis
• Enable secondary research
• Enable well-powered consortium science
• More citations of original research?
• Honor participant motives for participating

6. 1. Planning ahead for
data sharing

7. DON’T promise to destroy data
“After the study is
completed, videotapes will
be destroyed personally by
the investigator with a
sledgehammer.”

8. DON’T promise not to share data
“Only our research
team will have
access to your
data.”

9. DON’T promise limited scope of
research analyses or analysts
“Your data will only
be shared with
competent
researchers.”
“Your data will
(only) be used to
study X.”

10. DO use a data repository
Wicherts et al., The Poor
Availability of Psychological
Research Data for
Reanalysis, Am. Psychol. 61, 726–
728; 2006
• Authors requested data for
reanalysis from 249 studies
published in 141 articles in 4 top
APA journals
• 6 months, 400+ emails,
sometimes detailed descriptions
of study aims, IRB approvals,
signed assurances not to reshare
data & c.v.s
• Obtained 64/249 data sets
(25.7%); 73% of authors did not
share data

11. DO choose a data repository wisely
See Table in AMPPS article
— De-identification services?
— Restricted access options?
— Data citations?

12. DO be thoughtful about
re-identification risks
The U.S. Common Rule:
Data are “identifiable” when “the
identity of the subject is or may readily
be ascertained by the investigator or
associated with the information”

13. DO be thoughtful about re-
identification risks
• Don’t promise or guarantee anonymity
• Risk of re-identification a moving target; often
difficult/impossible to quantify
• Identifiable to investigators not all that matters

14. DO be thoughtful about re-
identification risks

15. DO be thoughtful about re-
identification risks
• Don’t promise or guarantee anonymity
• Risk of re-identification a moving target; often
difficult/impossible to quantify
• Identifiable to investigators not all that matters
• Invest in de-identification experts (write into your
grants!)
• But recognize that risk of re-identification is
sometimes exaggerated
RISK = Magnitude of harm x probability
(technical ability +
incentives)

16. DO get consent to retain & share data
• What is an archive/data repository?
• Why is it important to share data?
• With whom will data be shared?
• How will data be protected (if at all)?
• How might my data be used?
— Blanket consent
— Broad consent
— Reanalysis only
• Do I have a choice in whether my data is shared?
— No: Participation is voluntary; sharing is not
— Yes: Tiered sharing

17. Sample Consent Language: Open Brain Consent
http://bit.ly/2h1v7t1

18. Sample Consent Language: Open Brain Consent
http://bit.ly/2h1v7t1

19. IRB-Approved Consent Language
Researcher: Sean Murphy
Institution: University of Melbourne
Source: http://bit.ly/2g5iYWV (Github)

20. Successful IRB Application Language
Researcher: Sean Murphy
Institution: University of Melbourne
Source: http://bit.ly/2g5iYWV (Github)

21. DO make friends with your IRB/REB

22. DO nudge investigators by
incorporating data sharing plans into
IRB applications & consent templates

23. 2. What about previously
collected data, where consent
was silent about sharing?
Needs to be considered on a case-by-case basis
(consult your IRB)
But generally, when consent was silent, questions
should not be whether to share de-identified data,
but with whom and for what purpose(s)

24. Factors tending to increase the ethical acceptability of sharing
de-identified data without explicit consent
• Not feasible to recontact participants for permission
• Reason to believe sharing would not have affected participants’ enrollment
decision—or, ideally, that they would have endorsed sharing
• Data not especially sensitive (i.e., re-identification unlikely to cause
significant harm)
• Data not especially likely to be re-identified (i.e., low incentives for re-
identification, repository/DUA limits to qualified researchers and prohibits
re-identification, and/or data are technically difficult to re-identify)
• Sharing would not impose new or significantly greater risks (including
privacy risks) not described in the initial consent
• Sharing limited to secondary research purposes that fall within scope of
original consent
• Sharing limited to secondary research purposes participants are not known
to/would be unlikely to object to
• Sharing limited to secondary researchers that participants would be unlikely
to distinguish from original researchers
• Sharing limited to secondary researchers that participants are not known
to/would be unlikely to object to

25. What if my IRB objects to reasonable data data-
sharing?
 Data sharing itself isn’t
“research”
 Research w/existing, “non-
identifiable” data does not
constitute “human subjects”
research (so not regulated
by Common Rule, and
usually outside IRB
jurisdiction)

26. 4. What about existing
data, where the consent
promised no sharing
outside the study team?

27. — Not technically a Common Rule violation (!)
— But likely a protocol violation (so consult your IRB)
— And generally, this will be a difficult ethical case to
make.
— But, some possible exceptions:
1. Data will be shared for reanalysis only to confirm
original results

29. — Not technically a Common Rule violation (!)
— But likely a protocol violation (so consult your IRB)
— And generally, this will be a difficult ethical case to
make.
— But, some possible exceptions:
1. Data will be shared for reanalysis only to confirm
original results
2. No reasonable person could have premised their
enrollment decision on the promise not to share
(e.g., data are completely innocuous—but someone
independent should make this call)

30. Successful IRB Study Amendment Language
Researcher: Jessie Sun
Institution: University of Melbourne
Source: http://bit.ly/2gfm4pJ (Dropbox)