What FDA Adverse Event Reports Reveal About Tirzepatide Safety in the Real World

The researchers extracted tirzepatide-associated adverse event reports from the FDA Adverse Event Reporting System (FAERS) database covering Q2 2022 through Q3 2023. They used four established disproportionality analysis methods — reporting odds ratio (ROR), proportional reporting ratio (PRR), Bayesian confidence propagation neural network (BCPNN), and empirical Bayes geometric mean (EBGM) — to detect statistically significant safety signals. Results were classified using the MedDRA (Medical Dictionary for Regulatory Activities) coding system.

Tirzepatide (Mounjaro/Zepbound) has rapidly become one of the most prescribed medications in the world, but clinical trials only capture a limited picture of safety. Real-world adverse event data from millions of reports provides a broader view of what actually happens when diverse patient populations use the drug outside of controlled trial settings. Identifying unexpected signals like dosing errors and injection site hemorrhage helps guide prescriber education and patient monitoring, while the paradoxical finding of increased appetite in some patients challenges assumptions about the drug's mechanism.

As GLP-1/GIP agonists like tirzepatide are prescribed to millions of people, post-marketing surveillance becomes critical for identifying safety signals that clinical trials — with their controlled conditions and selected populations — might miss. This type of pharmacovigilance study is a standard tool in drug safety monitoring and has historically led to label changes, black box warnings, and in rare cases, drug withdrawals. For tirzepatide, the findings are largely reassuring but highlight areas needing attention, particularly around patient education on proper dosing and injection technique.

FAERS is a voluntary reporting system subject to reporting bias — adverse events may be over- or under-reported, and reports do not prove causation. The database cannot establish incidence rates since the total number of patients using the drug is unknown. Reports may contain duplicates, incomplete information, or confounding factors from co-medications. The analysis period covers only the first ~18 months after approval when reporting rates tend to be higher.