03 · What You Need to Know
“Indefinite” Is a Retention Decision, Not the Absence of One
There is no universal maximum storage period for research
Research retention requirements vary considerably. A clinical trial may have regulatory record-retention requirements. A longitudinal cohort may need data for decades. A biobank may deliberately be designed as a permanent scientific resource. A small study may have little justification for retaining identifiable working files after its research and verification needs have ended.
Researchers should therefore resist universal claims such as “all research data must be destroyed after five years” or “research data should always be kept forever.” Neither is a general research-ethics rule.
The Common Rule can accommodate indefinite storage
The U.S. Common Rule's broad-consent provisions provide a clear example of a framework that can permit indefinite retention. Broad consent for storage, maintenance, and secondary research use of identifiable private information or identifiable biospecimens must describe the period for which the materials may be stored and maintained, and that period may be indefinite.
The period during which the information or biospecimens may be used for research may likewise be indefinite.
That does not mean the Common Rule says every repository should retain material forever. It means indefinite storage can be an explicitly disclosed and authorized arrangement under that regulatory pathway.
Indefinite storage should not mean undefined purpose
There is an important distinction between retaining a resource indefinitely for a defined long-term research program and retaining personal information forever because somebody might conceivably find a use for it later.
Indefinite retention
No predetermined destruction date is set, but the storage remains tied to an articulated research or repository purpose, governance system, and applicable authorization.
Undefined retention
Information or specimens are kept without a defensible purpose, review structure, or meaningful criteria for determining whether continued storage remains justified.
The first can be legitimate in appropriate circumstances. The second is much harder to defend.
Philippine privacy rules draw this distinction particularly clearly
The Philippine Data Privacy Act requires personal information to be retained only as long as necessary for the purposes for which it was obtained, for legal claims, legitimate business purposes, or as otherwise provided by law. It also permits personal information collected for other purposes to be processed for historical, statistical, or scientific purposes and, in cases laid down in law, stored for longer periods with adequate safeguards.
The Act's implementing rules go further. They provide that personal data processed for historical, statistical, or scientific purposes may, in cases laid down in law, be stored for longer periods subject to appropriate organizational, physical, and technical security measures. They also state that aggregated data or information kept in a form that does not permit identification may be retained longer than necessary for the original purpose.
Crucially, the implementing rules state that personal data shall not be retained in perpetuity merely in contemplation of a possible future use that has yet to be determined.
For Philippine researchers, “we might use it someday” is therefore not a sufficient retention policy for identifiable personal data.
Consent should accurately describe the expected duration
If researchers intend to store identifiable information or biospecimens indefinitely, participants should not be given the impression that destruction will occur automatically when the original study ends.
The Common Rule's broad-consent provisions expressly require disclosure of the storage and research-use periods, which may be indefinite. HHS biospecimen guidance likewise identifies duration of specimen and information use as an important part of consent for future research.
Researchers should use language participants can understand. “Indefinitely” means there is no predetermined end date. It should not be disguised behind a comforting phrase such as “for a little while longer” when the repository actually intends permanent retention.
Long-term storage creates continuing security obligations
Every additional year of storage extends the period during which data can be breached, accessed inappropriately, transferred incorrectly, or exposed by technological change.
Security controls that were adequate when a repository opened may not remain adequate twenty years later. Encryption becomes outdated, personnel change, vendors disappear, institutions merge, storage platforms migrate, and new linkage resources can make old datasets more identifying.
Indefinite retention therefore implies indefinite stewardship, including periodic review of security and access arrangements. “We kept the server running” is not a governance model.
Identifiability can influence how long retention remains justified
Long-term retention of directly identifiable information can create different risks from long-term retention of appropriately non-identifiable research data. Where continued identification is unnecessary, reducing or removing identifiers may lower long-term risk.
This should not be used to evade participant withdrawal or explicit restrictions. The ethical problem with treating de-identification as a universal solution remains relevant to retention decisions.
Biological samples require physical stewardship as well as informational stewardship
Long-term biospecimen storage involves freezer systems, backup power, temperature monitoring, inventory management, contamination controls, specimen integrity, disaster planning, and decisions about finite storage capacity.
Samples can also degrade over time or become unsuitable for particular analyses. An institution should not describe storage as scientifically valuable indefinitely if it has no realistic capacity to preserve specimen quality.
Retention decisions therefore involve scientific utility as well as ethics and privacy.
Repositories should periodically ask whether continued retention remains justified
An indefinite storage authorization does not require an institution to keep every sample or dataset forever. Material may become scientifically unusable, duplicative, legally restricted, financially unsustainable, or inconsistent with the repository's mission.
Repositories should have governance procedures for disposal, destruction, transfer, or archival preservation when circumstances change.
This is especially important when an institution closes a repository or can no longer provide the safeguards promised to participants.
Retention and future use are related but different decisions
Permission to keep a specimen does not necessarily mean permission to perform every future analysis on it. A repository might legitimately retain material while requiring separate review before each secondary project.
Similarly, whether a stored sample can support an unanticipated future study depends on consent scope and governance rather than merely on the fact that the sample was lawfully retained.
Withdrawal becomes more consequential when storage has no end date
If participants authorize indefinite retention, they should understand what withdrawal can accomplish later. Can remaining identifiable specimens be destroyed? Can future collection stop? Can data already shared be recalled? Will completed analyses remain?
Long-term storage without a clear repository withdrawal process risks turning an open-ended authorization into an effectively irreversible one.
Research integrity can justify retaining some records after a project ends
Researchers may need to retain records to verify analyses, respond to questions, investigate research integrity concerns, satisfy funder or journal requirements, comply with law, or preserve the evidentiary basis of published work.
Those purposes should be distinguished from retaining identifiable participant information indefinitely for unspecified new research. A research-integrity archive can sometimes be designed with restricted access or reduced identifiability rather than remaining an active secondary-research dataset.
Data and samples do not necessarily need the same retention schedule
A physical specimen may lose scientific utility while derived sequence data remain valuable. Conversely, researchers may decide to delete identifiable working data while preserving a specimen under controlled repository governance for approved future assays.
Retention plans should therefore identify the lifecycle of each resource rather than assigning one date mechanically to “the study.”
Watch Out
Do not write “data will be stored indefinitely for possible future research” merely because no one has decided when to delete them. Indefinite retention should be an intentional, governed arrangement supported by an applicable purpose and authorization, not a euphemism for having no retention policy.
07 · A Quick Checklist
Before Keeping Research Data or Samples Indefinitely
Confirm that indefinite retention is deliberate and defensible:
Identify the specific research, repository, regulatory, archival, or integrity purpose that justifies continued storage.
Check the retention requirements and limits imposed by applicable law, ethics approval, funders, institutions, repositories, and contracts.
Verify that participants were accurately informed about long-term or indefinite storage where consent is the relevant authorization.
Distinguish permission to retain material from permission to use it for a particular future study.
Determine whether continued identifiability is necessary and proportionate to the research purpose.
Maintain security, access controls, auditability, and specimen-quality systems appropriate for long-term stewardship.
Review periodically whether the scientific value and purpose still justify continued retention.
Maintain a workable withdrawal process for material that remains identifiable and subject to withdrawal.
Establish procedures for secure destruction, transfer, or archival preservation if the repository closes or the retention basis ends.
