A screenshot of a safety database may show a large number next to a familiar medicine. The number deserves attention, but it cannot explain itself. Before treating a count as a rate, ask what was counted, how the information arrived and whether the number of people exposed is known. These are practical reading questions, not excuses for dismissing a possible harm.
What the reporting system receives
FDA's public dashboard FAQ describes reports of adverse events, medication errors and product-quality complaints. Reports can come from consumers, health professionals and manufacturers. Submission does not require proof that the product caused the reported event. The account may contain important information, but a report's presence in a database is not a completed causal assessment. [Public safety-data limitations]
The denominator problem
A rate needs a defined numerator and denominator for a relevant period. A pile of spontaneous reports does not supply a complete count of every exposed patient or every event that occurred. Reporting may be incomplete; the same episode may be submitted more than once. Publicity, time on the market and other influences can change the likelihood of reporting. FDA therefore cautions against calculating population incidence from these data. [Public safety-data limitations]
Imagine two products with different numbers of users and different reporting histories. Comparing their raw report totals cannot isolate a difference in risk. Dividing a public count by an unrelated estimate of prescriptions does not necessarily repair the problem: prescriptions are not always unique people, exposure time differs and the reporting population may not match the estimated denominator.
A signal is a starting point for investigation
FDA describes using reports alongside other evidence, including active surveillance and additional studies. Reviewers examine the detail and consistency of cases, consider alternative explanations and investigate potential associations. The useful distinction is between information that raises a question and evidence that resolves it. A report may be valuable even when a public dashboard cannot settle the question. [Regulatory surveillance explanation]
“This count does not establish causation” does not mean “the medicine could not have caused harm.” Both an unsupported alarm and an unsupported dismissal go beyond the data.
Read the current safety communication
For practical care, look for FDA's current safety guidance and the information for the actual product. Compounded preparations raise additional questions about product identity, dosing errors and quality; a brand-level dashboard search cannot answer all of them. Our article is about interpreting public evidence, not deciding what to do with an individual symptom. [Regulatory and safety context]
If you are concerned about something happening during treatment, contact the appropriate clinician or pharmacist rather than using a report total to self-triage. A public database is a research resource. It does not replace a person who can assess the event, its timing and your medical circumstances.
Source notes
Checked October 5, 2026. Provider pages document advertised terms. Study summaries identify the scope of our review.
- FDA — AEMS Public Dashboard FAQPublic safety-data limitations
- FDA — interpreting postmarket safety surveillanceRegulatory surveillance explanation
- FDA — concerns with unapproved GLP-1 drugsRegulatory and safety context