Three of the terms turning up repeatedly in these reports are not descriptions of what the drug did to somebody. They are descriptions of what somebody did with the drug, and the authors separate the two explicitly — a distinction a product page never makes.
What a record is
A case-product entry in the FDA’s Adverse Event Reporting System where a GLP-1 drug was named the primary suspect, after removing deleted cases, keeping only the latest version of each, and deduplicating. [1] That produced 243,114 records inside a background of 8,995,547 reports covering everything else.
Somebody — a patient, a clinician, a manufacturer — decided each of those was worth submitting. There is no denominator of users anywhere in it.
Two different kinds of thing in one list
Nausea, vomiting, diarrhea and injection-site pain are effects of a drug. Incorrect dose administered, extra dose administered and off-label use are events in the handling of a product.
The authors put it as medication-use terms reflecting use-process rather than conventional adverse-drug-reaction issues, and they build a whole prioritized domain around medication-use and device events alongside the clinical ones — gastrointestinal intolerance, impaired gastric emptying, pancreatobiliary, renal and dehydration events, and hypoglycemia.
The counts by drug are not a comparison
Tirzepatide accounted for 133,100 records, semaglutide 55,619 and dulaglutide 38,406.
That ordering says nothing about which drug is riskier. Reporting volume follows how many people take something, how long it has been on the market, how much press it has had, and who is being prompted to report — which is why a count of reports is not a count of events, and why the authors say explicitly that none of their sources estimates incidence, comparative risk or causality.
What it is actually for
Deciding what to look at next. The authors describe their output as signal prioritization and medication-safety evidence, cross-checked against Canadian reports and against the FDA’s own labeling changes.
Read that way it is useful and modest: a map of where to point a proper study. Read as a league table of harm it is worthless, and it will be read that way, which is the standing fate of every spontaneous reporting analysis that reaches a headline.