Who Decides Whether an Adverse Event Is Related to the Research?

Who gets to decide whether the research caused an adverse event?

A participant develops a medical problem after a research intervention. The site investigator thinks it is probably unrelated. The sponsor sees similar events at several other sites and becomes concerned that there may be a pattern. Whose assessment counts?

There is rarely one person who makes every causality decision for every purpose. The investigator is usually closest to the individual participant and often makes the initial case-level assessment. A sponsor, medical monitor, data monitoring committee, IRB or REC, or regulator may have a different role depending on the study and the decision being made.

This distinction matters because determining that an adverse event occurred and determining what caused it are separate steps. An AE can exist even when its relationship to the research remains uncertain.

The investigator usually assesses the individual event, but causality may be reassessed

Relatedness is an evidence-based judgment, not a label attached automatically

The investigator is usually closest to the individual case

When an event occurs at a research site, the investigator can examine information that may be critical to causality: what happened, when it happened, the participant's underlying conditions, concomitant medications, research procedures, intervention exposure, laboratory findings, and plausible alternative causes.

OHRP's guidance places an important initial responsibility on the investigator. When a local investigator becomes aware of an internal adverse event, the investigator assesses whether it represents an unanticipated problem. That assessment necessarily includes deciding whether the event is related or possibly related to research participation.

In clinical trials, ICH E6(R3) similarly expects investigators to provide causality assessments when reporting serious adverse events to sponsors. The investigator's role is therefore substantive rather than clerical.

Relatedness is usually expressed as probability, not certainty

Causality in research safety is often uncertain. OHRP explicitly recognizes that relatedness determinations commonly fall along a continuum between definitely related and definitely unrelated.

Different protocols may use categories such as related, possibly related, unlikely related, or unrelated. Others may use fewer categories. Researchers should use the terminology defined in the approved protocol rather than inventing their own causality scale.

Timing matters, but “after” does not mean “because of”

Temporal relationship is useful evidence. If a reaction begins minutes after administration of an investigational product, that timing may support a causal hypothesis. Yet sequence alone cannot establish causation.

A participant can develop influenza after receiving an intervention. Someone with diabetes can experience hypoglycemia during a study. A participant can be injured in an unrelated accident while enrolled. OHRP expressly recognizes that adverse events may arise from research procedures, the participant's underlying disease or condition, or circumstances unrelated to either.

The task is therefore to compare competing explanations rather than treating chronology as proof.

Alternative causes matter

A relatedness assessment should consider whether another explanation accounts for the event more convincingly. Relevant possibilities may include the participant's underlying disease, another medication or treatment, an infection, an unrelated injury, or another medical condition.

Importantly, the presence of an alternative explanation does not automatically prove that the research played no role. More than one factor can contribute to an event. OHRP considers an AE related to research when research participation at least partially caused the event, whereas an event caused solely by an underlying condition or unrelated circumstances would generally be considered unrelated.

The intervention is not the only possible research-related cause

Researchers sometimes ask only whether the investigational drug or device caused the event. That can be too narrow.

Research-related harm may arise from other procedures, such as blood collection, biopsies, imaging, withholding or changing treatment, study-specific questionnaires, interviews, or other interventions required by the protocol. OHRP's relatedness framework refers broadly to procedures involved in the research.

This becomes particularly important when evaluating unexpected psychological distress during research. The relevant causal question may concern an interview or questionnaire rather than a medical product.

The sponsor may see evidence the investigator cannot see

A site investigator typically sees one participant or one site's experience. A sponsor may have safety information from many participants, sites, or related studies.

That difference in perspective can change the causality assessment. An event that looks indistinguishable from background disease at one site may become more suspicious when similar events accumulate disproportionately across participants exposed to an intervention.

OHRP explicitly recognizes this possibility. An investigator may initially determine that an event is not an unanticipated problem, while a monitoring entity may later reach a different conclusion after detecting an unexpectedly high frequency of the event.

Different assessments do not necessarily mean somebody made a mistake

An investigator and sponsor may reasonably reach different causality assessments because their evidence differs. A local investigator may know that a participant has several strong alternative explanations. A sponsor may know that an unusual pattern is emerging across sites.

The important response is not to force artificial agreement. The assessments and their rationales should be documented according to the applicable system, and safety decisions should use the evidence available to the party responsible for that decision.

Relatedness can change as new evidence arrives

Causality assessment is not necessarily permanent. Initial information may be incomplete. Laboratory findings may arrive later. A participant's diagnosis may change. Similar events may accumulate elsewhere.

A reasonable initial classification can therefore be revised when stronger evidence becomes available. Safety follow-up is partly designed for precisely this reason.

“Possibly related” can be enough for an important reporting decision

Researchers should not assume that only events proven to have been caused by research matter for oversight.

Under OHRP's unanticipated-problem framework, the relevant criterion is related or possibly related. OHRP defines possibly related as a reasonable possibility that the event may have been caused by the research procedures. Proof is not required.

This distinction affects how causality influences adverse-event reporting. An uncertain but reasonably plausible relationship should not automatically be converted into “unrelated” simply because certainty is unavailable.

The site sees one case, while the sponsor sees a pattern

The investigator's initial assessment was based on the evidence available locally. The later sponsor assessment incorporates information the investigator did not initially possess. Causality assessment can therefore evolve without requiring the fiction that one of the earlier judgments was necessarily careless.

Common mistakes when assessing adverse-event relatedness

Make the causality assessment systematic and revisable

If you are responsible for evaluating an AE, begin with the causality categories and procedures specified by the protocol. Then assess the evidence rather than starting with the conclusion you would prefer the event to have.

Most importantly, do not allow causality debate to delay urgent participant care or time-sensitive adverse-event reporting requirements.

Before assigning an adverse-event relatedness category

Frequently asked questions about adverse-event relatedness

Causality is assessed from evidence and may change as the evidence changes

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.