NIH Wants to Rewrite Its 1998 Safety Monitoring Policy: Fewer Automatic DSMBs, Monitoring Plans for All Clinical Research, Comments by January 6
November 4, 2026 · 6 min read
Granted Research Team · Editorial policy
The rule that decides whether an NIH-funded study needs a Data and Safety Monitoring Board was written in 1998. It came before the modern NIH definition of a clinical trial, before single IRB review, before ClinicalTrials.gov results reporting, and before decentralized and pragmatic trials became common. NIH now wants to replace it.
On November 2, 2026, the NIH Office of Science Policy released a draft update to the NIH Policy for Data and Safety Monitoring for public comment. The Guide notice is NOT-OD-27-006, and comments go through an OSP web form. The NIH announcement describes the purpose as reflecting "the evolving clinical research landscape": standardizing monitoring requirements across institutes and reducing administrative work that does not match a study's actual risk.
According to a summary from Harvard Medical School's Office for Academic and Research Integrity, the draft would make four main changes:
- Extend risk-based monitoring requirements from clinical trials to all NIH-supported clinical research.
- Remove the current requirement for Data and Safety Monitoring Boards for all Phase 3 or multi-site clinical trials.
- Require an NIH-approved data and safety monitoring plan tailored to study risk.
- Apply to applications submitted six months after the final policy is published.
The comment deadline is January 6, 2027. Harvard's office is a secondary source, and NIH's own pages were not accessible when we checked, so read the full draft before relying on any single provision.
What the 1998 Policy Does
The current policy has two parts. The 1998 policy requires every NIH-supported clinical trial to have data and safety monitoring "commensurate with risk," and it ties DSMBs to multi-site trials of interventions that carry potential risk, which in practice has meant most Phase 3 trials. Guidance issued in 2000 added a requirement for monitoring plans in Phase 1 and Phase 2 trials. Individual institutes then layered their own rules on top. NCI, NHLBI, NIA, NIDA and others each have distinct DSMB policies, plan templates and reporting schedules.
The result is a system that is both too strict and too loose:
- Too strict: a low-risk, multi-site behavioral trial can be treated as needing a full independent DSMB, with the recurring costs of member honoraria, unblinded statistical reports and meetings that come with it, because the rule is based on phase and site count rather than risk.
- Too loose: observational and mechanistic studies that are not clinical trials, including some with invasive procedures, sensitive data or vulnerable populations, fall outside the policy's explicit requirements and depend on IRB review alone.
- Inconsistent: the same protocol can face different monitoring expectations depending on which institute funds it.
The draft addresses all three problems with one principle: monitoring should match risk, and the plan should show how.
The Four Changes in Detail
1. From clinical trials to clinical research
This is the largest change in scope. NIH's clinical trial definition is the four-question test it adopted in 2014. Clinical research is a broader category that covers observational cohorts, natural history studies, biospecimen studies with identifiable participants, behavioral research and mechanistic human studies.
If the summary is accurate, thousands of NIH-funded projects with no current monitoring-plan obligation would need one. For a minimal-risk survey study, the plan could be a single paragraph naming the PI as monitor and describing how adverse events and confidentiality breaches are reported. For an observational study involving lumbar punctures or imaging with contrast, reviewers will expect more.
2. No more automatic DSMBs
Removing the phase- and site-based trigger does not eliminate DSMBs. It means the decision is based on risk. A Phase 3 drug trial with mortality endpoints will still need one. A multi-site, pragmatic comparative-effectiveness trial of two approved standard-of-care approaches might not. The change shifts responsibility to the applicant: you will have to justify your monitoring level instead of pointing to the trial's phase.
3. An NIH-approved plan
The phrase "NIH-approved" matters. Today, monitoring plans are reviewed unevenly, sometimes at peer review, sometimes by program staff after award, and sometimes barely at all. A requirement for NIH approval suggests a more formal step, which may occur before award or as a term and condition. Applicants should expect program officers to ask for revisions and should budget time for that step between the summary statement and the notice of award.
4. A six-month runway
Applying to applications submitted six months after final publication means no current awards or pending applications are affected. If NIH finalizes in mid-2027, the first affected receipt dates would likely fall in early 2028. Large clinical research programs therefore have about a year to update templates, budgets and institutional procedures.
Who Benefits and Who Bears the Cost
| Group | Likely effect |
|---|---|
| Low-risk multi-site trials (behavioral, pragmatic, implementation) | Cost savings if a full DSMB is no longer automatically required. Budget money moves from DSMB support back to the science |
| Observational and mechanistic human studies | New requirement. Plans needed where none existed, though usually brief for minimal-risk work |
| High-risk interventional trials | Little practical change. DSMBs remain, and plan expectations may become more uniform |
| Institutions and IRBs | Need to coordinate IRB risk determinations with NIH monitoring plans and avoid duplicating reviews |
| Academic coordinating centers and DSMB vendors | Demand shifts from mandatory boards to tiered monitoring services such as independent safety monitors and smaller committees |
How This Fits NIH's Broader Clinical Research Agenda
The draft is one of several clinical research policy changes NIH has made in 2026:
- NOT-OD-26-113 proposes requiring plain-language, summary-level results to be shared with participants in NIH-funded clinical research. It also uses "clinical research" rather than "clinical trials" as its scope.
- The K award consolidation would end independent clinical trials as the core of career development awards.
- NIH has dropped the Data Management and Sharing Plan requirement for K award applicants and is replacing letters of support with letters of collaboration.
- OSP held a stakeholder webinar on DSM modernization on May 13, 2026, which shaped this draft.
The common direction is that NIH is redefining its human-subjects obligations around clinical research as a whole and around proportionality. Investigators running observational studies who have assumed trial-focused policies don't apply to them should plan for that to change.
What to Put in Your Comment
Comments that change final policy are specific and backed by evidence. Topics to address:
- Define the risk tiers. Ask NIH to publish clear criteria, or a decision tool, for when a DSMB, an independent safety monitor or PI-only monitoring is appropriate. Without them, "tailored to risk" will mean whatever an individual program officer thinks it means.
- Set a minimal-risk floor. Ask that minimal-risk research, as determined by an IRB under 45 CFR 46, qualify for a short, standardized monitoring statement rather than a full plan.
- Clarify "NIH-approved." Who approves, at what point, and on what timeline? If approval happens before award, ask for a maximum review period so awards are not delayed.
- Harmonize institute policies. The main benefit of the reform is one standard. Ask NIH to commit to retiring or conforming institute-specific DSMB policies when the final policy takes effect.
- Address ongoing projects. Ask how the policy applies to renewals, competing revisions and multi-year awards whose next application falls after the effective date.
- Provide numbers. If your institution runs DSMBs for low-risk multi-site trials, report the annual cost per board. If you run large observational cohorts, estimate the effort a new plan would require. Specific figures are the most persuasive comments.
What to Do Now
- Clinical trial PIs: list the trials in your portfolio that have DSMBs only because of phase or site count, and note whether you would choose a lighter approach on risk grounds. That list is both useful comment material and future budget planning.
- Observational researchers: start a one-page monitoring plan template now covering adverse-event identification, reporting lines, confidentiality breach procedures and stopping rules where relevant. It will be needed in 2028, and it improves IRB submissions today.
- Research offices: check where DSMB decisions are recorded in your systems. If the decision becomes a risk judgment rather than a rule, you will need documentation showing who made it and why.
- Everyone: submit comments by January 6, 2027 through the OSP form linked from NOT-OD-27-006, and check the notice for any extension.
NIH rarely revisits foundational human-subjects policy. The 1998 policy lasted 28 years, so whatever is finalized from this draft will likely govern monitoring for most of the next generation of NIH-funded clinical research. Commenting now is the main chance to influence it.
Granted helps investigators find NIH funding opportunities and draft application sections, including human subjects and data and safety monitoring plans. Search NIH grants or start a draft with Granted AI.