Trial Adverse Events: Denominators, Severity, and Follow-Up

What This Page Answers

When a clinical trial reports that “12% of participants had a side effect,” that number is only useful if you know three things: what group is in the denominator, how severe the events were, and how long participants were tracked. This page walks through how to find and check those three elements in a trial report, using the reporting standards trials are supposed to follow.

The Evidence Ladder: How Adverse Event Reporting Is Supposed to Work

Randomized trials that follow standard reporting practice are expected to include an adverse events summary built around a few core elements. Understanding the ladder helps you spot when a report is incomplete.

Step 1: The Denominator

Every adverse event count needs a “number of participants at risk” attached to it — the group that could have experienced the event. ClinicalTrials.gov’s results-reporting structure requires this explicitly: adverse event tables list the number of participants affected and the number of participants at risk for each event, by study arm, because the two numbers together are what make a percentage meaningful.

A number without a stated denominator (“adverse events were rare”) is not verifiable. A number with a denominator (“14 of 220 participants in the treatment arm”) is.

Step 2: Serious vs. Non-Serious, Reported Separately

Standard practice is to report adverse events in separate tables: one for serious adverse events, and one for other (non-serious) adverse events that occur above a set frequency threshold — commonly 5% in any study arm, per ClinicalTrials.gov’s results structure. A death count is reported separately again, in an all-cause mortality table, regardless of whether the death was believed to be related to the trial intervention.

The CONSORT Harms guidance (the extension to the main CONSORT reporting checklist that governs how trial results should be written up) similarly calls for the proportion and frequency of serious and non-serious events to be reported both cumulatively and separately, since a participant can experience more than one event.

Step 3: Timing — During vs. After the Intervention

CONSORT Harms guidance distinguishes events that occurred during the active intervention period from those that occurred during follow-up after it ended (for trials involving a procedure, this distinction is drawn between the intra-operative and post-operative periods). A report that collapses “during” and “after” into one number makes it harder to judge whether an effect faded, persisted, or only showed up later.

Separately, ClinicalTrials.gov’s Adverse Events module requires trials to state the specific time frame over which adverse events were collected. Summary reporting happens at the end of the study, not in real time while the trial is ongoing — so a report should tell you the collection window, not just a snapshot from partway through.

Known vs. Unknown

Known: Standard adverse event reporting structure calls for a stated denominator (participants at risk), separate reporting of serious and non-serious events, an all-cause mortality count, and a defined collection time frame.
Known: Non-serious events are typically only tabulated once they exceed a frequency threshold (commonly 5%) in at least one study arm — meaning rarer non-serious events may not appear in the summary table at all.
Unknown from this page alone: Whether a specific trial you’re reading actually followed this structure. Adherence to harms-reporting standards varies considerably across published trials, so the presence of a results table doesn’t guarantee it contains all four elements above.
Unknown from this page alone: Causality. Adverse event tables report events that occurred during the trial, not a determination of which events were caused by the intervention versus coincidental. Reports should separately flag events considered “reactions” (a reasonable possibility of a causal link), but not every report does this consistently.

Practical Checklist: Reading a Trial’s Adverse Event Section

Find the denominator for each percentage. If a report gives a rate with no stated group size, note that as a gap.
Check whether serious and non-serious events are reported in separate tables, not blended into one “side effects” list.
Look for the all-cause mortality count specifically — it should be reported on its own, even if no deaths were considered related to the intervention.
Identify the adverse event collection time frame. If it isn’t stated, that’s a gap worth flagging, not filling in with an assumption.
Note whether the report distinguishes events occurring during the intervention from events occurring during follow-up after it.
Check for a stated frequency threshold on non-serious events (e.g., “events occurring in more than 5% of participants”). This tells you the table is a partial list of non-serious events, not a complete one.

Bringing Questions to a Clinician or Research Team

If you’re trying to use a trial’s adverse event data to inform a real decision — about a treatment, a study you’re considering joining, or a product whose marketing cites a trial — the checklist above is a starting point for questions, not a substitute for individualized guidance. A clinician or the study team can tell you things a published summary table can’t: how the results apply to your specific situation, what the trial’s inclusion/exclusion criteria mean for whether it’s comparable to you, and what to watch for based on your own health history.

About This Article

This is general educational information about how clinical trials report adverse events. It is not medical advice, and it does not evaluate the safety or effectiveness of any specific product, treatment, or trial. If you are weighing a treatment decision or considering joining a study, talk with a qualified healthcare provider or the study’s research team. If you are experiencing a medical emergency, contact local emergency services immediately.

By ClinicalStudyConnect.com Research Desk. Last updated September 23, 2026.