Table 2.

Outcomes of Breakout Group Discussions

Group 1- What factors should be considered in the development of a Federal policy for access to publicly funded “omic” research data	Outcomes of Discussion: Proposed a pilot of a truly open-source dataset that would allow access to all types of data, including “omic” data. Participants willing to contribute their biospecimens and data would be informed that all data would be made public and that anyone would be able to access them. Participants would be informed about the risks and benefits associated with such data sharing.
	• The pilot would involve multiple datasets stratified by varying comfort levels, phenotypes, and/or potential for social stigma. However, in light of continued budgetary constraints, one focused pilot of an open dataset would be a good start
	• The pilot could assemble a large cohort of individuals with specific phenotypes and track it longitudinally, with broad consents to overcome issues of controlled access
	• Patient advocacy groups, public health departments, and epidemiologists could, among others, serve as partners in recruiting individuals for such a cohort
Group 2- What considerations enter into determining whether “omic” data is identifiable?	Outcomes of Discussion: There was consensus that all data are identifiable, but acknowledgment that re-identification requires a link to a matched sample and that some data are at greater risk for identifiability than others. Although some barriers are necessary to prevent against abuses of data, many barriers restrict scientific discovery. Therefore the group proposed a three-tiered “Russian doll model,”
	• The open-access tier (Tier 1) allows anyone to download data contributed by participants who have been informed of identifiability and risk issues but permit wide access.
	• The general controlled access tier (Tier 2) allows investigators to apply for access to data that fall under Exemption 4 or data for which participants have consented to broad sharing.
	• The restricted tier (Tier 3), which allows no data sharing, includes data from specific population studies in which consent has been given for one research study only.
Group 3- What are appropriate ethical constraints to allowing researchers broad access to “omic” data?	Outcomes of Discussion: Participants are willing to consent to use of their data, but that willingness stems partly from a desire to be involved and to know how their data are helping. Formal mechanisms are not needed to keep participants informed.
	• Transparency should occur throughout the process, not just when participants are consenting to contribute specimens or data.
	• The burden of contact and re-contact is not fully on the research community. Participants can be asked to maintain their correct contact information if they want to receive results.
	• Using identifiable data would facilitate respect for participants and the return of research results. However, such a focus also would limit some current practices and use of retrospective samples.
	• The group also discussed the need to include new technologies into current research processes, e.g. online consent that allows participants to choose the level of information they want to share
Group 4- How can society minimize any risks and maximize any participant benefits of “omic” research?	Outcomes of Discussion: The levels of risk vary across data types. Moreover, the risks for identifiability must be distinguished from the risk for harm—for example, identification of an adopted child's birth parents, risks to someone running for office, inability to obtain health or life insurance, risks to family members etc.
	• Real instances of misuse have related to issues of data security and, so far, re-identification has occurred primarily within the research setting. Therefore, the risks to research participants remain remote.
	• Efforts to address remote risks could minimize the benefits to research participants and the community. Accordingly, studies should minimally report aggregate data.
	• Should use streamlined data access committees to control the release of data and requiring investigators to accept accountability by registering for access to data.
	• In addition, training data are needed to develop and improve bioinformatics.