Purpose

Summary

The National Institutes of Health (NIH) is issuing this final NIH Policy for Data Management and Sharing (DMS Policy) to promote the management and sharing of scientific data generated from NIH-funded or conducted research. This Policy establishes the requirements of submission of Data Management and Sharing Plans (hereinafter Plans) and compliance with NIH Institute, Center, or Office (ICO)-approved Plans. It also emphasizes the importance of good data management practices and establishes the expectation for maximizing the appropriate sharing of scientific data generated from NIH-funded or conducted research, with justified limitations or exceptions. This Policy applies to research funded or conducted by NIH that results in the generation of scientific data.

Background

Sharing scientific data accelerates biomedical research discovery, in part, by enabling validation of research results, providing accessibility to high-value datasets, and promoting data reuse for future research studies.[1] As a steward of the nation’s investment in biomedical research, NIH has long championed policies that make research available to the public to achieve these goals. For example, the 2003 NIH Data Sharing Policy reinforced NIH’s commitment to data sharing by requiring investigators to address data sharing in applications for large research awards. NIH’s 2014 Genomic Data Sharing (GDS) Policy, initially preceded by the 2008 Genome-Wide Association Studies Policy, set the expectation that researchers share large-scale genomic data, regardless of species, to enable the combination of large and information-rich datasets. In 2016, the NIH Policy on the Dissemination of NIH-Funded Clinical Trial Information (Clinical Trials Policy) further reinforced NIH’s commitment to research participants and the research community by making the results of clinical trials accessible in a timely fashion.

NIH recognizes that its data sharing policy efforts must flexibly evolve to keep pace with scientific and technological opportunities and notes that researchers’ ability to generate, store, share, and combine data has never been greater. To capitalize on these advancements, NIH initiated the development of a more comprehensive data sharing policy alongside its efforts to modernize data sharing infrastructure in its 2015 Plan for Increasing Access to Scientific Publications and Digital Scientific Data from NIH Funded Scientific Research. With policy and infrastructure modernization efforts working in tandem, NIH initiated a stepwise process for seeking feedback from the community to develop a robust data sharing policy capable of reflecting the diversity of its community’s data sharing needs. In 2016, NIH requested public comments on data management and sharing strategies and priorities (NOT-OD-17-015). In 2018, NIH solicited public input on proposed key provisions that could serve as a foundation for a future NIH policy for data management and sharing (NOT-OD-19-014). Using public feedback to inform its thinking, in 2019 NIH released a draft proposal for a future data management and sharing policy in the Federal Register (84 FR 60398).

Along with the Draft Policy proposal, NIH sought feedback on supplemental materials that could help researchers integrate effective data management and sharing practices into research, including “Elements of an NIH Data Management and Sharing Plan” and “Allowable Costs for Data Management and Sharing.” We note that a third document, “Supplemental Information to the NIH Policy for Data Management and Sharing: Selecting a Repository for Data Resulting from NIH-Supported Research,” was developed in response to public comments received on both the Draft Policy and the “Request for Public Comments on Draft Desirable Characteristics of Repositories for Managing and Sharing Data Resulting From Federally Funded Research,” which was released for public comment by the White House Office of Science and Technology Policy (OSTP) to promote consistency across federal agencies and reduce researcher burden (85 FR 3085).

In respect and recognition of Tribal sovereignty, NIH also initiated Tribal Consultation on its Draft Policy proposal, in accordance with the HHS Tribal Consultation Policy and the NIH Guidance on the Implementation of the HHS Tribal Consultation Policy. The NIH Tribal Consultation Report – NIH Draft Policy for Data Management and Sharing provides more detail on the Tribal Consultation process relative to the development of the final DMS Policy and NIH’s response. Briefly, three themes emerged from Tribal Nations’ input: 1) Strengthen engagement built on trust between researchers and Tribal Nations; 2) Train researchers to responsibly and respectfully manage and share American Indian and Alaska Native (AI/AN) data; and 3) Ensure research practices are aligned with the laws, policies, and preferences of AI/AN community partners. NIH intends to continue discussions to ensure appropriate implementation of the DMS Policy as it relates to these communities, and details about some of the implementation planning follows in the discussion below.

Overview of Public Comments

NIH incorporated feedback over the course of several years to develop a data management and sharing policy proposal and released its Request for Comments on the Draft NIH Policy for Data Management and Sharing and Draft Supplemental Guidance on November 8, 2019 (84 FR 60398, comment period closing on January 10, 2020). NIH held a public webinar on December 16, 2019, with over 580 people participating. In response to the Draft Policy, NIH received 203 responses from both domestic and international stakeholders, and the comments are publicly available.[2] The largest group of respondents reported affiliation with universities, followed by nonprofit research organizations, professional associations (tied with “other”), as well as small percentages of respondents affiliated with government agencies, healthcare delivery organizations, and patient advocacy organizations. Respondents typically identified themselves as scientific researchers, while another sizeable section self-identified as “other.” Remaining respondents identified as institutional officials, with smaller percentages self-identified as bioethicists or social science researchers, government officials, patient advocates, and members of the public. NIH considered all feedback in the development of the final DMS Policy, and a discussion of the public comments on topics follows below.

Discussion of Public Comments on the Draft NIH Policy for Data Management and Sharing

Clarifying Expectations for Sharing Scientific Data

Draft Policy: The Draft Policy did not explicitly set a default expectation of data sharing. Rather, it focused on requiring submission of and compliance with a Data Management and Sharing Plan (Plan) that outlines how data will be managed and shared. The Draft Policy also included recognition of that fact that certain factors (i.e., legal, ethical, or technical) may limit the ability to preserve and share data.

Public Comments: While commenters were generally supportive of the overall scope of the Draft Policy, many requested NIH make an explicitly stronger commitment to expecting data sharing from the research community. Suggestions included requiring data sharing and indicating that data sharing should be the default, with well justified exceptions being permitted.

Final Policy: The final DMS Policy does not create a uniform requirement to share all scientific data. Unlike a requirement for submission of Plans, which can be implemented across various funding mechanisms and types of research with little variation, appropriate data sharing is likely to be varied and contextual. Through the requirement to submit a Plan, researchers are prospectively planning for data sharing, which we anticipate will increasingly lead researchers to integrate data sharing into the routine conduct of research. Accordingly, we have included in the final DMS Policy an expectation that researchers will maximize appropriate data sharing when developing Plans. The final DMS Policy retains the Draft Policy’s factors (i.e., ethical, legal, or technical) that may necessitate variations in the extent of scientific data preservation and sharing, and researchers should convey such factors in their Plans. The final DMS Policy has also been modified to clarify these factors are not limited to data derived from human research participants. We believe this will provide the necessary flexibility for researchers to accommodate the substantial variety in research fields, projects, and data types that this expectation will encompass.

Definition of “Scientific Data”

Draft Policy: The scope of which data will be shared relies on the definition of “scientific data.” This term was defined in the Draft Policy as: “The recorded factual material commonly accepted in the scientific community as necessary to validate and replicate research findings, regardless of whether the data are used to support scholarly publications. Scientific data do not include laboratory notebooks, preliminary analyses, completed case report forms, drafts of scientific papers, plans for future research, peer reviews, communications with colleagues, or physical objects, such as laboratory specimens. NIH expects that reasonable efforts will be made to digitize all scientific data.”

Public Comments: Commenters focused on a variety of aspects of the definition of “scientific data.” They suggested that the concept of data quality be included, as data that may otherwise meet the definition but, if uninterpretable, are not of value. Commenters also suggested the definition address null or negative findings (and indicate that these data should be shared). Commenters requested clarification about the sentence that NIH expects reasonable efforts will be made to digitize all scientific data, including whether NIH would cover costs to digitize data that are not collected in digital form.

Final Policy: The final DMS Policy defines Scientific Data as: “The recorded factual material commonly accepted in the scientific community as of sufficient quality to validate and replicate research findings, regardless of whether the data are used to support scholarly publications. Scientific data do not include laboratory notebooks, preliminary analyses, completed case report forms, drafts of scientific papers, plans for future research, peer reviews, communications with colleagues, or physical objects, such as laboratory specimens.” We agree that data quality is an important concept to convey to ensure that scientific data are useful and to prevent data sharing from becoming a perfunctory administrative requirement, but rather one that should be done with the understanding that these data are intended to be used by others. Therefore, we have added to the definition that the data should be of sufficient quality to validate and replicate research findings. Even those scientific data not used to support a publication are considered scientific data and within the final DMS Policy’s scope. We understand that a lack of publication does not necessarily mean that the findings are null or negative; however, indicating that scientific data are defined independent of publication is sufficient to cover data underlying null or negative findings.

We also note that while the final DMS Policy states that scientific data are those of sufficient quality to “validate and replicate,” we anticipate that shared scientific data will be used for a variety of purposes (consistent with applicable laws, policies, and limitations) including subsequent analyses, as suggested in the Purpose section of the final DMS Policy. Therefore, the concepts of validation and replication provide a standard for determining what constitutes scientific data and are not intended to limit uses of shared data.

Finally, we have removed the expectation for digitizing scientific data. We encourage reasonable efforts to digitize data, recognizing that digitizing data may be a technical factor that may limit the sharing of data.

Timing of Submission of Data Management and Sharing Plans

Draft Policy: The Draft Policy proposed the submission of Plans at Just-in-Time for grants.

Public Comments: While we received a range of comments about timing of Plan submission, the majority were opposed to or requested further clarification about Just-in-Time Plan submission. Commenters were concerned about not having sufficient time to develop Plans and expressed concerns about the Plan revision process leading to delays in issuing awards. Others indicated that institutions would want to review Plans because they would ultimately be responsible for compliance, but a Just-in-Time Plan submission would not afford institutions sufficient time. A key practical concern with Just-in-Time Plan submission was difficulty submitting a budget at application that included requests for allowable data management and sharing costs prior to actually drafting the Plan. Commenters who favored submitting Plans at Just-in-Time frequently cited decreased burden on applicants, because with Just-in-Time, only those applicants likely to be funded would be required to submit Plans, rather than all applicants.

Final Policy: The final DMS Policy requires submission of a Plan for extramural grants at application. This approach is more conducive to achieving NIH’s goal of promoting a culture in which data management and sharing are recognized to be an integral component of a biomedical research project, rather than an administrative or additive one. While NIH is aware that this approach places the requirement on the general pool of grant applicants rather than on those likely to be funded, it is precisely this approach of prospective planning for data management and sharing that NIH hopes to promote and that a number of commenters suggested is crucial for ensuring more regular planning for data management and sharing. We were swayed by the logistical concerns expressed in comments, namely how applicants could submit budgets appropriately reflective of data management and sharing when not yet required to submit the Plan that is intended to help them consider these issues. In addition, the concerns about institutions having sufficient time to review Plans and potential logistical challenges in issuing timely awards was persuasive. This approach is also consistent with the 2018 Request for Information on Proposed Provisions of a Draft Data Management and Sharing Policy for NIH Funded or Supported Research, which proposed Plans be submitted with extramural grant applications. The responses to that proposal generally favored Plan submission at the time of application.

Assessment of Plans

Draft Policy: The Draft Policy proposed that NIH Program Staff in the funding NIH ICO assess Plans from extramural grants.

Public Comments: Many commenters supported peer review of Plans, noting their skill and that peer review of Plans would promote a cultural shift in favor of data sharing. Commenters also suggested that NIH Program Staff review may lead to more consistent Plan assessment and decrease peer reviewer burden.

Final Policy: The final DMS Policy maintains NIH Program Staff assessments of Plans’ merits. However, peer reviewers may comment on the proposed budget for data management and sharing, although these comments will not impact the overall score. This approach balances the benefit of consistency afforded by NIH Program Staff review of Plans, review of updates, and compliance monitoring, with the opportunity for peer reviewers to comment on the requests for data management and sharing costs. Over time, and through these reviews, we hope to learn more about what constitutes reasonable costs for various data management and sharing activities across the NIH portfolio of research.

NIH ICO Consistency of Data Sharing Expectations

Draft Policy: The Draft Policy noted that NIH ICOs may supplement the Policy’s expectations for Plans with their own complementary requirements to further advance their specific program or research goals. In addition, the Draft Policy stated the funding NIH ICO may request additional or specific information to be included within Plans to meet expectations for data management and sharing in support of programmatic priorities or to expand the utility of the scientific data generated from the research.

Public Comments: In light of various existing NIH ICO data sharing policies, commenters expressed confusion around having potentially varying expectations in data sharing policy implementation across NIH. There were concerns about insufficient direction to NIH ICOs and around a potentially uncoordinated variety of approaches. Commenters suggested guidance to facilitate NIH ICO consistency and suggested that NIH provide a centralized location of NIH ICO-specific expectations to help researchers navigate variations, particularly when subject to more than one NIH ICO’s data sharing policies.

Final Policy: While the final DMS Policy’s language on this issue has not substantively changed from that of the Draft Policy, we have heard the concerns and intend to address them during the period of implementation planning prior to the DMS Policy’s Effective Date. NIH ICOs can, within certain bounds, meet their scientific, policy, and programmatic goals in different ways. As such, this Policy affords NIH ICOs the opportunity to meet the goals of this Policy in ways that enhance their respective science. However, we intend to promote consistency on some key tenets of the final DMS Policy, such as the requirement for submission of Plans and the timing of their submission. The DMS Policy represents the minimum requirements for the NIH, but NIH ICOs may expect more specificity in Plans. For example, NIH ICOs and Programs may wish to promote, via specific Funding Opportunity Announcements (FOAs) or across their research portfolios, the use of particular standards to enable interoperability of datasets and resources. We are appreciative of the suggestion about how to organize NIH ICO-specific expectations and will be working to ensure clear implementation materials for applicants and awardees.

Data Derived from Human Participants

Draft Policy: The Draft Policy acknowledged the applicability of laws, regulations, guidance, and policies that govern the conduct of research with human participants and how data derived from human participants should be used. It also described that Plans should indicate how human participants and data derived from them would be protected. Finally, the Draft Policy acknowledged that certain factors may limit the ability to share data and proposed that these factors be described in the Plan. Importantly, the Draft Policy did not propose any new expectations for the conduct of research with human participants.

Public Comments: Commenters expressed concerns about how to safeguard participant privacy and confidentiality when sharing data, with some requesting information on de-identification practices. Commenters also requested guidance on best practices in communicating data sharing in informed consent. They also stressed the importance of data sharing to maximize the contributions of those who volunteer to participate in NIH-funded studies. Some pointed to special populations with preferences on data sharing issues, such as AI/AN populations, and asked how sharing of data from these participant populations is expected to be handled.

In addition to the public comments submitted during the comment period, NIH received input from the Secretary's Advisory Committee on Human Research Protections (SACHRP).[3] SACHRP provided a set of recommendations relating to applying the DMS Policy to research with human participants, some of which we have incorporated into the final DMS Policy and are discussed below.

AI/AN communities provided input through various channels, including through letters sent to NIH as part of government-to-government communications. The Tribal Consultation process also led to valuable input that is informing NIH’s implementation efforts, described further below.

Final Policy: As with the Draft Policy, the final DMS Policy does not introduce new requirements for protections for research with human participants. Existing laws (e.g., Certificates of Confidentiality), regulations (e.g., the Common Rule), and policies (e.g., the NIH Genomic Data Sharing Policy) continue to apply. However, through this Policy and associated supplemental information and other activities, NIH promotes thoughtful practices regarding the treatment of data derived from human participants.

In response to public comments and SACHRP’s recommendations on the Draft Policy, we have included in the final DMS Policy three concepts that we believe are important to emphasize for investigators as they think through how to engage prospective participants regarding what is expected to happen with the data they contribute and, downstream, how best to respect these contributions. First, we encourage investigators to consider, while developing their Plans, how to address data management and sharing in the informed consent process, such that prospective participants will understand what is expected to happen with their data. This planning will serve investigators as they develop their Plans, because some of the Plan elements prompt investigators to outline anticipated factors that might affect the ability to share and preserve scientific data, such as any limitations arising from the informed consent process. NIH also intends to develop resources to help researchers and institutions in communicating the intent to share data with prospective research participants. Second, we note that any limitations on subsequent use of data (which may apply to non-human data as well) should be communicated to those individuals or entities preserving and sharing the scientific data. This ensures that factors that may affect subsequent use of data are properly communicated and will travel with the data. Finally, we highlight