01 · The Question
When does one participant's safety problem become a problem with the study?
A participant experiences an unexpected injury. Does the protocol need to change? What if the same event happens twice? What if nobody is injured, but a procedural error exposes participants to substantially greater risk than anticipated?
Not every adverse event requires rewriting the protocol. Participants can become ill for reasons unrelated to research, expected side effects can occur at anticipated frequencies, and some isolated incidents can be addressed appropriately at the individual level.
But a safety event can reveal something more important: the research may be riskier than previously understood, an existing safeguard may not work, a particular group may be especially vulnerable, or participants may be making decisions based on incomplete risk information. At that point, protecting the affected participant is no longer enough. The study itself needs review.
03 · What You Need to Know
The key question is whether the study's original safety assumptions still hold
One adverse event does not automatically require a protocol amendment
An adverse event can occur without revealing anything new about the research. An expected event occurring with the anticipated nature, severity, and frequency may already be adequately addressed by the protocol, consent process, monitoring plan, and available safeguards.
An unrelated illness or injury may likewise require participant care and documentation without implying that the research procedures need modification.
The study-level question is different: does this event change what we know about the risks of conducting the research as currently designed?
A single event can sometimes be enough
Researchers should not wait for a second participant to be harmed merely because only one event has occurred. A single event can be sufficiently informative when it is unexpected, plausibly related to the research, and indicates a substantially greater risk than previously recognized.
OHRP considers an incident, experience, or outcome an unanticipated problem when it is unexpected, related or possibly related to research participation, and suggests that the research places participants or others at greater risk of harm than was previously known or recognized. Such events generally warrant consideration of substantive protocol or consent changes or other corrective action.
The trigger is therefore not a predetermined event count. It is the information the event provides about risk.
A pattern can reveal what individual cases cannot
Other safety problems become apparent only after events accumulate. One occurrence may be indistinguishable from background illness. Several similar events may reveal an unexpectedly high frequency, a vulnerable subgroup, or a relationship to a particular procedure.
This is one reason safety monitoring should consider aggregate information rather than treating every participant incident as an isolated administrative file.
A sponsor, coordinating center, DSMB or DMC, or other monitoring entity may detect a pattern that an individual site cannot see. Once a meaningful signal emerges, the question becomes whether current protections remain adequate.
Unexpected serious research-related harm deserves particular attention
OHRP identifies adverse events that are unexpected, related or possibly related to research, and serious as an especially important subset of unanticipated problems. These events indicate greater physical or psychological risk than previously known and routinely warrant consideration of substantive changes or corrective actions.
A serious adverse event is not automatically an unanticipated problem, because seriousness alone does not establish unexpectedness or relatedness. Once all of those elements converge, however, a study-level response becomes much more difficult to dismiss as unnecessary.
Actual injury is not required before the study may need to change
A safety problem can reveal inadequate protections even when nobody is ultimately harmed.
OHRP's examples include incidents such as a protocol error that exposes a participant to greater risk without producing a detectable adverse effect. Because the incident can still reveal an unexpected research-related hazard, corrective action may be warranted before the feared harm actually occurs.
Participant-level response
Addresses what the affected participant needs, including care, follow-up, documentation, and appropriate reporting.
Study-level response
Asks whether procedures, eligibility, monitoring, consent, enrollment, or other protections should change for current or future participants.
The appropriate change should address the risk that was discovered
There is no single standard amendment for every safety problem. OHRP identifies several possible responses to an unanticipated problem, including modifying inclusion or exclusion criteria, adding monitoring procedures, suspending enrollment, suspending research procedures in currently enrolled participants, revising informed consent to describe newly recognized risks, and providing new risk information to participants already enrolled.
| New safety information |
Possible study-level response |
| A subgroup appears at greater risk |
Revise eligibility criteria, stratify monitoring, or add subgroup-specific safeguards |
| A procedure produces more complications than anticipated |
Modify the procedure, increase monitoring, add stopping criteria, or temporarily suspend it |
| A previously unknown risk becomes established |
Update risk information and determine what current and prospective participants should be told |
| A preventable operational error creates substantial risk |
Correct the process, retrain staff, strengthen controls, and determine whether enrollment or procedures should pause |
| The overall risk-benefit balance becomes uncertain |
Consider temporary suspension while the new evidence is evaluated |
| Unexpected serious harm makes continued research unacceptable |
Suspension or termination may need consideration by the responsible oversight bodies |
New risk information can change informed consent
Participants decide whether to join research based partly on what they understand about its risks. If the risk information materially changes, the consent process may need to change too.
OHRP notes that significant new findings developed during research that may relate to a participant's willingness to continue should, when appropriate, be provided to the participant. Its consent guidance also recognizes that changes in risks can require supplementation or repetition of the consent process.
This can affect both future and already enrolled participants. Revising the consent form for tomorrow's recruits does not necessarily address what today's participants need to know.
Sometimes enrollment should pause while the new risk is addressed
If newly discovered risk information makes the existing consent information inadequate, enrolling additional participants under the old information may be inappropriate.
OHRP has specifically explained in the context of certain NCI-sponsored trials that when new or modified risk information necessitates changes to the description of reasonably foreseeable risks in order to satisfy informed-consent requirements, new participants cannot be enrolled until the revised consent documents and accompanying protocol changes have received the necessary IRB review and approval.
That example should not be converted into a universal rule that every protocol amendment requires suspension of recruitment. The reason for the pause is the significance of the new information and its effect on valid consent.
Immediate hazards are an important exception to waiting for prior approval
Normally, changes to approved research require IRB or REC review before implementation. Participant protection creates an important exception.
OHRP states that changes necessary to eliminate apparent immediate hazards to participants may be initiated before IRB approval. Other proposed changes in response to an unanticipated problem should be reviewed and approved before implementation.
Watch Out
“Urgent safety change” is not a convenient route around prospective ethics review. Implement changes without prior approval only when the applicable framework permits doing so to eliminate an apparent immediate hazard, then follow the required notification and review procedures.
Suspension and termination are protective tools, not automatic punishments
Pausing research does not necessarily mean that misconduct occurred. Sometimes uncertainty itself justifies temporarily reducing exposure while a potentially important safety signal is investigated.
Under the HHS framework, an IRB has authority to suspend or terminate approval of research that is not being conducted in accordance with its requirements or that has been associated with unexpected serious harm to participants. Current OHRP guidance on IRB written procedures also emphasizes consideration of participants already enrolled and orderly management when research is suspended or terminated.
The appropriate action may be narrower than stopping the entire study. Enrollment can pause while follow-up continues. One procedure can be suspended while others remain necessary for participant safety. The response should fit the problem.
Repeated distress can also become a study-level safety signal
Study-wide safety problems are not limited to biomedical injuries. If multiple participants experience unexpected psychological distress during interviews or surveys, the pattern may indicate that questions, sequencing, warnings, interviewer preparation, stopping criteria, or support arrangements are inadequate.
Researchers should not repeatedly manage each case through the participant-distress procedure without asking why the same problem keeps happening.
The risk-benefit assessment remains active throughout the study
Initial ethical approval is based on the evidence available before or at the time of review. New safety information can change that evidence.
OHRP's continuing-review guidance describes ongoing oversight as an opportunity to reassess whether risks remain minimized and reasonable in relation to anticipated benefits and the knowledge expected from the research.
The practical implication is straightforward: “the study was approved” is not an answer to evidence that the study has become materially different from the study that was originally judged acceptable.
04 · A Practical Example
Repeated injuries reveal that a procedure needs more than case-by-case management
Hypothetical Example
A research procedure produces an unexpected pattern of injuries
A study uses a protocol-specific physical assessment that was expected to carry minimal risk. One participant sustains an unexpected injury. The team responds appropriately and investigates. Two additional participants subsequently experience similar injuries associated with the same procedure.
First event
The team treats the participant, documents and reports the event as required, and examines whether an identifiable procedural problem occurred.
Pattern emerges
Similar events make an isolated participant explanation less convincing and raise concern that the procedure itself carries greater risk than anticipated.
Immediate protection
If continuing the procedure poses an apparent immediate hazard, the team follows the applicable process for urgent protective action.
Study-level review
The investigator, sponsor, institution, and relevant oversight bodies evaluate procedure design, staff training, eligibility, monitoring, and the overall risk-benefit assessment.
Protocol response
The study may require modification of the procedure, additional safeguards, revised eligibility, enhanced monitoring, paused enrollment or procedures, or another proportionate response.
Consent response
If the emerging risk is relevant to decisions about participation, current and prospective participants receive appropriate updated information through the approved consent process.
The crucial transition occurs when the evidence no longer supports treating each injury as merely an individual event. The study's safety assumptions themselves have become questionable.
