NIH Data Sharing Policy An Overview

Why a Data Sharing Policy?

The National Institutes of Health (NIH) has long recognized that the value of biomedical research is amplified when data generated with federal funding are made widely available. Sharing data accelerates scientific discovery, promotes reproducibility, reduces unnecessary duplication of effort, and maximizes the return on public investment. The NIH Data Sharing Policy, first introduced in 2003 and refreshed several times since, establishes a baseline expectation that investigators will make their data accessible to the broader research community, unless a valid reason for restriction exists.

Evolution of the Policy

• 2003: Initial policy focused on large-scale genomic data.
• 2007: Extension to all types of data generated with NIH funding.
• 2020: Incorporation of the NIH Final Policy for Data Management and Sharing (often called the Data Management Plan requirement) that applies to all new grant applications.
• 20232024: Clarifications on controlled access data, expectations for metadata standards, and alignment with the FAIR principles (Findable, Accessible, Interoperable, Reusable).

Key Requirements for Researchers

1. Data Management Plan (DMP)

Every new NIH grant application must include a Data Management Plan that addresses the following elements:

• Types of data to be generated or collected.
• Standards and formats that will be used.
• Policies for data preservation and sharing, including timelines.
• Roles and responsibilities for data stewardship.
• Budget considerations for data management activities.

2. Timely Sharing

Data should be shared no later than the date of publication (or the end of the grant period, whichever occurs first). For large datasets, a reasonable embargo period may be requested (up to 12 months) to allow the primary investigators to publish initial findings.

3. Use of Repositories

NIH expects data to be deposited in a trusted, publicly accessible repository that:

• Provides a persistent identifier (e.g., DOI, accession number).
• Supports appropriate metadata standards for the discipline.
• Ensures longterm preservation and security.

Popular repositories include NCBI, Dataverse, Figshare, and disciplinespecific archives such as the BioProject portal.

4. ControlledAccess Data

When data involve privacysensitive human subjects, NIH requires that they be stored in controlledaccess repositories (e.g., dbGaP, the European GenomePhenome Archive). Researchers must submit a data use agreement (DUA) that specifies who may request access and under what conditions.

5. Documentation & Metadata

Good documentation is essential for reuse. The DMP should describe the metadata standards (e.g., MIAME for microarray data, DICOM for imaging) and provide clear data dictionaries, codebooks, and any software needed to interpret the data.

6. Compliance Monitoring

NIH staff may request evidence of compliance (e.g., repository accession numbers) during progress reports or at the time of the final grant closeout. Failure to share data as promised can affect future funding eligibility.

Exemptions & Special Cases

When Sharing May Be Restricted

NIH acknowledges that certain circumstances justify limited sharing:

• Privacy & Confidentiality: Data that could identify participants without appropriate deidentification.
• Intellectual Property (IP): Data that are patenteligible or protected by commercial agreements.
• National Security: Information subject to export controls or other legal restrictions.
• Ethical Considerations: Studies involving vulnerable populations where broader sharing might increase risk.

In these cases, investigators must detail the limitation in the DMP and provide a justification that is reviewed by the funding institute.

Data That Are Not Covered

The policy does not apply to:

• Published aggregate results that are already publicly available.
• Data generated before the start date of the award (unless otherwise stipulated).
• Data that are exempt under the Common Rule when specific waivers are granted.

Practical Tips for Successful Implementation

Start Early

Incorporate the DMP into the grant proposal narrative from day one. Identify the appropriate repository, verify its dataformat requirements, and budget for any required storage fees.

Engage Your Institutions Library or Data Services

Many universities provide data stewardship support, including help with metadata standards, selection of repositories, and compliance monitoring. Leverage these resources to avoid lastminute surprises.

Automate Where Possible

Use tools such as Research Objects or workflow managers (e.g., Snakemake, Nextflow) to generate reproducible pipelines that automatically produce both data and accompanying documentation.

Document the Why Not Just the What

Explain the scientific rationale behind the data collection, any preprocessing steps, and how variables were derived. Future users will appreciate the context.

Plan for LongTerm Preservation

Choose repositories that guarantee at least a 10year preservation policy. If the chosen repository is disciplinespecific, confirm that it has a sustainability plan.

Monitor Access Requests

For controlledaccess datasets, designate a data steward who can review DUA applications, ensure compliance with consent terms, and keep records of who accessed the data and when.

Helpful Resources

• NIH Data Sharing Policy (official site)
• NIH Final Policy for Data Management and Sharing (PDF)
• Gene Expression Omnibus (GEO) repository for functional genomics data.
• Database of Genotypes and Phenotypes (dbGaP) controlledaccess human data.
• FAIR Principles Overview
• Research Data Alliance (RDA) guidelines and community standards.
• Dataverse Project opensource repository platform.