What Counts as an Adverse Event

Part of Research Nurse (RN) Training: Recognizing Adverse Events.

About 2 minutes. Free, in your language, no account needed.

Why this matters

The definition of an adverse event is much wider than most people expect. Applying it narrowly is the most common safety failure at site level.

If you only record what you believe the drug caused, the safety database describes your beliefs rather than what happened.

The definition

An adverse event is any untoward medical occurrence in a participant who has received a study intervention. It does not have to be related to the intervention.

Read that twice. Relatedness is not part of the definition.

A broken ankle from falling off a ladder is an adverse event. A cold is an adverse event. A flare of a condition the participant already had is an adverse event.

What is not an adverse event

Usually, the condition being studied when it behaves as expected, if the protocol says so. Some protocols capture disease progression separately.

Usually, a pre-existing condition that has not changed. It belongs in medical history instead.

Planned procedures that were scheduled before the participant entered the study, though the reason for them may be reportable.

Always check the protocol. These exclusions vary and you should not assume them.

Start and stop dates

An adverse event has an onset date and, usually, a resolution date.

Onset is when the participant first noticed it, not when they told you. Ask. "How long have you had this?" is a data question as much as a clinical one.

Ongoing events are recorded as ongoing. They are followed until they resolve or the protocol says follow-up ends.

In practice

A participant mentions they had a headache for two days last week and took paracetamol.

That is an adverse event. Onset last week, resolved, mild, treated with paracetamol, and the paracetamol is a concomitant medication.

They did not think it was worth telling you. Your job is to make it easy for them to, and to record it once they do.

What this looks like when it goes wrong

A site records only events it considers drug-related. Over a year it reports very few.

The sponsor sees a clean safety profile. When a signal later appears in other studies, this site's data cannot help confirm or refute it, because the ordinary events that would have formed the background rate were never captured.

Filtering at source destroys the comparison.

Key takeaways