Adverse Event (AE)

An adverse event is any untoward medical occurrence in a patient or trial participant given a product, whether or not it is considered related to that product. The definition is deliberately agnostic about cause. It sweeps in new symptoms, worsening of pre-existing conditions, clinically significant laboratory abnormalities and intercurrent illness with nothing to do with the drug. Trials collect events either unsolicited, by open questioning, or solicited against a prespecified list, and the two methods yield very different numbers from the same population.

Placebo arms are the clearest demonstration. In the 68-week STEP 1 trial of semaglutide for obesity, gastrointestinal events were reported by roughly three quarters of participants on active drug but also by around half of those on placebo, because nausea, constipation and diarrhoea are common in any large cohort watched closely for a year. The informative quantity is the difference between arms, not the number on the drug.

An event count becomes evidence only once it has a comparator, a denominator and an exposure period attached. Twelve reports of pancreatitis means nothing until you know how many people were treated, for how long, and what the background rate is in a matched untreated population. It also explains why single-arm and open-label data inflate: unblinded participants who expect effects report more of them, and no arm exists against which to subtract expectation.

Where this bites is the screenshotted label. A list of events observed in a trial circulates as what the drug does to you, when much of that list is what happens to people over a year regardless. The reverse failure is just as common with unapproved compounds: no structured collection exists, so an absence of reported events reads as a clean record when it only reflects that nobody was asked.