The Uniformed Sample: Veterans' Genomes and the Stewardship Obligations of a Federal Biobank

The Department of Veterans Affairs holds one of the largest genomic research resources in the United States. The Million Veteran Program, launched in 2011, asks veterans to contribute blood samples and health information so that researchers can study how genes, environment, and military service interact. As of 2026, the program has enrolled more than 800,000 veterans, and its combination of genomic data with VA electronic health records gives it a depth that few civilian cohorts can match.

That scale is a scientific asset. It is also a governance test. Veterans are not ordinary research volunteers. Many depend on the VA for health care, disability benefits, and long-term services. When the same agency that provides those services also holds a veteran's genome, the familiar consent question—will this person understand what they are agreeing to?—becomes a broader custody question: what obligations follow the sample after enrollment, and who decides where it goes next?

A Cohort Built on Trust, and on an Asymmetry

The MVP's success rests on a straightforward bargain: veterans contribute samples and records, and the VA promises responsible research. That bargain is reinforced by the VA's unusual relationship with its patients. For many veterans, the VA is not one provider among many but the central institution of their health and financial lives. Enrollment in a research program can therefore feel continuous with receiving care, even when the consent form plainly separates the two.

That asymmetry does not make MVP consent invalid. It does, however, raise the stakes for how the program explains future uses, commercial partnerships, and the limits of withdrawal. A consent form signed at a VA medical center may be read by a veteran as a promise that the sample will remain within the VA system. If that assumption is wrong, the governance gap is not merely technical. It is a gap between what veterans believe they agreed to and what the program's data-sharing arrangements actually permit.

From Research Consent to Clinical Signal

Genomic research and clinical care are often described as a continuum, but they carry different legal and ethical weights. A variant discovered in a research study may eventually return to a veteran's clinical record. A clinician may use that information in treatment decisions. A research administrator may use it in program evaluation. A benefits adjudicator may be tempted to use it in a disability determination.

Those are not equivalent uses. The first is a return of useful health information. The second is quality improvement. The third would be a category error with serious consequences. Veterans' genomic data should not migrate into benefits adjudication without explicit statutory authority, separate notice, and a meaningful opportunity to object. A firewall between the biobank and the benefits system is not an obstacle to good government; it is a condition of trustworthy government.

The clinical pathway also needs clearer rules. When a research finding is confirmed and returned, who documents it, who explains it, and who bears the cost of follow-up? Veterans should not have to navigate those questions alone, and clinicians should not be asked to improvise a governance system at the point of care.

Partners, Vendors, and the Long Chain of Custody

Federal biobanks rarely operate in isolation. Genomic analysis may involve contracted laboratories, cloud storage, bioinformatics platforms, and academic or industry collaborators. Each link in that chain creates a new custody question. A data use agreement may prohibit re-identification, but de-identified genomic data is not automatically anonymous. A vendor may promise not to sell data, but the agreement may still allow broad internal research or product development.

The relevant policy question is not whether partnerships should exist. They often bring expertise and capacity that the VA cannot replicate alone. The question is whether the terms of those partnerships are visible, auditable, and bounded. At a minimum, agreements should specify what data are shared, for what purposes, for how long, with what security controls, and under what conditions data must be returned or destroyed. They should also give the VA audit rights and require notification of breaches or material changes in use.

The Consent Form Is Necessary but Not Sufficient

No consent document can anticipate every future research question. That is why consent must be paired with ongoing governance. Veterans should be able to learn, in plain language, what categories of research their data support and which partners receive access. Material changes—such as a new commercial use or a shift from research to clinical products—should trigger reconsent or, at the very least, a clear opt-out.

Withdrawal deserves particular attention. A veteran who leaves the program should receive a realistic explanation of what can be removed and what cannot. Deleting a sample is different from deleting derived data, aggregate results, or information already incorporated into a publication. Honest withdrawal language will not satisfy every request, but it will prevent the program from promising more than it can deliver.

A Stewardship Agenda for the Next Phase

A stronger framework can be built without abandoning the MVP's research mission. The Institute recommends six measures:

1. Create a statutory firewall. Genomic data held for research or clinical care should be categorically unavailable for veterans' benefits determinations unless Congress authorizes a specific use with separate notice and consent.
2. Publish a plain-language data-use map. The VA should describe, at a level veterans can understand, the categories of partners that receive MVP data and the purposes for which it is shared.
3. Adopt model data-sharing terms. Agreements should include purpose limitation, audit rights, return-or-destruction clauses, a ban on re-identification, and clear breach notification.
4. Establish independent veteran oversight. A standing board with veteran representation, privacy expertise, and clinical input should review material changes in data use and partnership.
5. Report publicly each year. An annual stewardship report should cover access requests, approvals, denials, breaches, withdrawal outcomes, and partnership categories.
6. Fund return-of-results infrastructure. If genomic findings are worth returning, the VA should fund the counseling, confirmation, and follow-up that make return meaningful.

The MVP was built on a promise that veterans' contributions would serve veterans and the public interest. That promise is still credible, but it cannot rest on consent alone. Federal genomic stewardship requires visible rules, bounded partnerships, and a clear line between care and coercion. Veterans have already done their part. The government's part is to keep custody worthy of the sample.