Semaglutide vs Tirzepatide vs Liraglutide: A FAERS/AEMS Adverse Event Comparison (2026)
The three most-prescribed GLP-1 receptor agonists in the United States — semaglutide, tirzepatide, and liraglutide — all share a similar mechanism, all carry overlapping safety warnings, and all show up frequently in FAERS/AEMS adverse event reports. But their actual reported safety profiles are not interchangeable. This article compares them side-by-side using AEMS data across the 2021 Q2 – 2026 Q1 window, with a focus on the dimensions that matter to clinicians and patients trying to choose between them.
Data window for this analysis: 2021 Q2 through 2026 Q1 (20 quarters). This window captures the major GLP-1 prescribing wave (the Wegovy weight-management approval in June 2021, the tirzepatide launches in 2022 and 2023, the cultural inflection point in 2023-2024). Reports submitted before 2021 Q2 are not included in the case counts. This particularly affects liraglutide, which was first approved in 2010 and has substantial pre-window reporting history that's outside our analysis.
Before getting into the data, two pieces of context:
The brand-name mapping. All three molecules have multiple brands:
- Semaglutide = Ozempic (diabetes), Wegovy (weight management), Rybelsus (oral, diabetes)
- Tirzepatide = Mounjaro (diabetes), Zepbound (weight management)
- Liraglutide = Victoza (diabetes), Saxenda (weight management)
For safety signal purposes, the molecule is what matters; FAERS/AEMS data aggregates across brand names. A reader searching "Ozempic vs Mounjaro" and a reader searching "Wegovy vs Zepbound" should both find the same underlying comparison.
The mechanism. Semaglutide and liraglutide are pure GLP-1 receptor agonists. Tirzepatide is a dual GLP-1 / GIP receptor agonist (it activates a second incretin hormone receptor in addition to GLP-1). This dual mechanism contributes to tirzepatide's stronger efficacy on weight loss in head-to-head trials and may also explain some differences in side-effect profile.
Headline numbers
AEMS case counts for the 2021 Q2 – 2026 Q1 window:
| Drug | Cases in window | Brand names |
|---|---|---|
| Tirzepatide | 144,078 | Mounjaro, Zepbound |
| Semaglutide | 66,321 | Ozempic, Wegovy, Rybelsus |
| Liraglutide | 7,686 | Victoza, Saxenda |
The three drugs were not all on the market at the start of the window. Liraglutide was first approved in 2010 (Victoza) and 2014 (Saxenda) — substantial pre-window reporting exists in the database but is excluded here. Semaglutide was approved as Ozempic in December 2017; the Wegovy weight-management indication followed in June 2021, right at the start of our window. Tirzepatide was first approved as Mounjaro in May 2022, fully inside our window, with Zepbound following in November 2023.
Two things to notice immediately:
Tirzepatide has more cases than semaglutide despite being newer. This is a function of prescription volume — tirzepatide's launch coincided with peak public attention to GLP-1 drugs, and Mounjaro/Zepbound prescriptions ramped faster from launch than Ozempic/Wegovy did. Tirzepatide also has, as we'll see, a substantial volume of dosing-error reports that inflate its case count.
Liraglutide has the smallest in-window case base despite the longest market history, because most of its prescribing peaked before 2021 Q2 and many patients have since migrated to the newer drugs. Its smaller in-window count also makes its disproportionality numbers more vulnerable to reporting clusters — a point we'll return to below.
The gastrointestinal comparison: semaglutide leads
GI reactions are the dominant safety theme for all three drugs. They're also the dimension where the three differ most clearly. Per 1,000 cases:
| GI reaction | Semaglutide | Tirzepatide | Liraglutide |
|---|---|---|---|
| Nausea | 171.5 | 107.8 | 150.7 |
| Vomiting | 122.8 | 53.6 | 113.2 |
| Diarrhoea | 96.2 | 61.2 | 83.5 |
| Constipation | 77.4 | 39.7 | 61.8 |
| Decreased appetite | 70.8 | 21.6 | 42.0 |
| Abdominal pain | 47.1 | 19.2 | 58.6 |
| Abdominal pain upper | 44.4 | 22.4 | 39.2 |
| Dyspepsia | 28.4 | 16.9 | 23.8 |
Semaglutide consistently shows the highest GI rates per case across most reactions: nausea (171 vs 108), vomiting (123 vs 54), diarrhoea (96 vs 61), constipation (77 vs 40). The per-case ratio versus tirzepatide is around 1.6x to 3x higher across the GI cluster.
This is not the same thing as saying semaglutide is "worse" pharmacologically — the per-case rates depend on what else is in the case. Tirzepatide's denominators include a huge volume of dosing-error reports (more on this in a moment) that don't include GI events. If you adjust by excluding pure-dosing-error cases, the GI gap between semaglutide and tirzepatide narrows.
What you can say confidently: the GI profile of GLP-1 cases reaching AEMS is more dominated by digestive reactions for semaglutide than for tirzepatide. Whether that reflects a true pharmacologic difference (tirzepatide's GIP component possibly buffering GI effects), a difference in titration practice, or a difference in what gets reported alongside what — that takes deeper analysis to disentangle.
The clinical takeaway: if you're a patient or prescriber choosing between semaglutide and tirzepatide and GI tolerability is your primary concern, the comparative AEMS data supports the head-to-head trial finding that tirzepatide tends to be GI-better-tolerated than semaglutide at comparable doses, though both produce substantial GI effects.
Tirzepatide's signature pattern: injection and dosing errors
The clearest single feature of tirzepatide's AEMS profile is that it's dominated by use-related reports, not pharmacologic ones:
| Reaction | Tirzepatide rate/1k | Semaglutide rate/1k | Liraglutide rate/1k |
|---|---|---|---|
| Incorrect dose administered | 212.4 | 8.8 | 4.9 |
| Injection site pain | 99.9 | 11.8 | 7.2 |
| Extra dose administered | 64.8 | 2.9 | 1.8 |
| Injection site haemorrhage | 38.9 | 6.9 | 1.9 |
| Injection site erythema | 35.3 | 2.3 | 4.2 |
| Accidental underdose | 32.8 | 0.05 | 2.1 |
| Product dose omission issue | 29.3 | 9.6 | 22.8 |
| Injection site bruising | 24.2 | 5.7 | 8.2 |
| Injection site pruritus | 19.0 | 0.5 | 3.6 |
"Incorrect dose administered" at 212 per 1,000 cases for tirzepatide — more than 1 in 5 reports — is extraordinary. For semaglutide the same reaction is at 8.8/1k, a 24x difference. This is not a pharmacologic finding. It's a product-design finding.
Tirzepatide is delivered via Eli Lilly's KwikPen or a single-dose vial, depending on formulation and indication. The pen's interface, dose-selection mechanism, and refill workflow have generated a substantial volume of usage-error reports — including incorrect doses administered (often related to dose escalation), extra doses (often from confusion about timing or pen reuse), and accidental underdoses (failure to deliver the full intended amount).
Injection site reactions for tirzepatide (pain, erythema, bruising, pruritus) are also higher per case than for the other two drugs. This pattern — pen-related errors plus injection site reactions — points at the device/formulation interface, not the molecule.
The clinical implication: tirzepatide's "true" pharmacologic safety profile may actually be more favorable than its AEMS numbers suggest, once you discount the device/dosing-error layer. Conversely, in real-world use, the device matters; patients and prescribers should not underestimate the practical impact of dosing complexity.
A caution about the liraglutide data
Looking at liraglutide's top reactions, a pattern emerges that doesn't fit a GLP-1 mechanism:
| Reaction | Liraglutide rate/1k | Deaths reported with this reaction |
|---|---|---|
| Anxiety | 60.4 | 123 |
| Dyspnoea | 54.8 | 120 |
| Condition aggravated | 49.6 | 123 |
| Asthma | 46.2 | 122 |
| Hypothyroidism | 43.5 | 122 |
| Wheezing | 42.3 | 118 |
| Therapeutic product effect incomplete | 42.9 | 120 |
| Pain in extremity | 41.5 | 121 |
| Full blood count abnormal | 40.2 | 121 |
| Loss of personal independence in daily activities | 40.1 | 113 |
| Sleep disorder due to a general medical condition | 39.0 | 120 |
| Lung disorder | 38.4 | 120 |
| Pulmonary fibrosis | 38.1 | 122 |
| Neurological symptom | 38.1 | 121 |
Several red flags here.
First, the reactions themselves: asthma, pulmonary fibrosis, wheezing, lung disorder, neurological symptom. Liraglutide is a GLP-1 agonist — there's no plausible mechanism by which it would cause asthma or pulmonary fibrosis in a meaningful way. These aren't on the label, aren't supported by the clinical trial literature, and don't fit the known pharmacology.
Second, the death counts are nearly identical across these unrelated reactions: 118-123 deaths reported for each of asthma, dyspnoea, pulmonary fibrosis, condition aggravated, sleep disorder, lung disorder, hypothyroidism, and pain in extremity. This is the statistical fingerprint of a single reporting cluster — a batch of cases that share a similar event list and death attribution, repeatedly reported under different MedDRA preferred terms.
What's likely happened: a single source (a litigation campaign, a registry submission, or a coordinated reporting effort) deposited a few hundred cases into AEMS with similar multi-event profiles. Against liraglutide's small case base (7,686 cases), those cluster reports dominate the rates.
This is one of the well-known failure modes of FAERS/AEMS analysis. Disproportionality scores can be inflated by clustered reports without representing any population-level pharmacologic effect. For liraglutide, the GI signals near the top of its list (nausea, vomiting, diarrhoea, constipation, impaired gastric emptying) are consistent with the GLP-1 mechanism and probably reflect real drug effects. The respiratory/neurological cluster lower in the list is much more likely to be a reporting artifact.
The practical implication for this comparison: comparing semaglutide and tirzepatide on dimensions where their data is similar in quality is straightforward. Comparing either against liraglutide on the respiratory or anxiety dimensions is not — those liraglutide numbers don't reflect a comparable signal. We've flagged this rather than silently using suspect numbers.
A useful analyst habit: always look at the case-level patterns, not just the aggregate disproportionality numbers. Reporting clusters look like aggregates until you decompose them by source, date, and event-pair structure.
Less common but watched signals
A few specific reactions worth comparing across the three drugs:
Gastroparesis (impaired gastric emptying)
| Drug | Rate per 1,000 cases | Total cases |
|---|---|---|
| Liraglutide | 85.6 | 658 |
| Semaglutide | 68.7 | 4,557 |
| Tirzepatide | 14.9 | (lower) |
Counterintuitive finding: per case, liraglutide has the highest gastric emptying impairment rate, slightly above semaglutide and roughly 6x above tirzepatide. Most public coverage focuses on semaglutide because of its huge absolute case count, but per case the older drug shows up stronger.
For tirzepatide's much lower rate, the explanation may be partly the GIP component (which has different effects on gastric motility than GLP-1 alone) and partly the denominator dilution from the dosing-error cases discussed above.
Intestinal obstruction
| Drug | Rate per 1,000 cases |
|---|---|
| Semaglutide | 25.9 |
| Liraglutide | 17.7 |
| Tirzepatide | 5.3 |
Semaglutide leads, and the gap to tirzepatide is large (about 5x). This is consistent with the September 2023 FDA label update that added ileus and intestinal obstruction warnings to Ozempic specifically.
Decreased appetite
| Drug | Rate per 1,000 cases |
|---|---|
| Semaglutide | 70.8 |
| Liraglutide | 42.0 |
| Tirzepatide | 21.6 |
This one is interesting because decreased appetite is partly an intended effect — these are weight-loss drugs. Higher rates reflect either (a) more effective appetite suppression or (b) more reporting of an expected effect as an adverse event. Semaglutide's rate being roughly 3x tirzepatide's may partly reflect the latter; patients and prescribers may be more accustomed to reporting it for the more-publicized drug.
Off-label use
| Drug | Rate per 1,000 cases |
|---|---|
| Semaglutide | 96.1 |
| Tirzepatide | 63.3 |
| Liraglutide | 53.7 |
About 1 in 10 semaglutide cases include "off-label use" as a reported reaction. For tirzepatide it's about 1 in 16; for liraglutide about 1 in 19. This is the weight-loss-prescribing reality showing up in the data: semaglutide had Wegovy (the weight-loss indication) approved in 2021 but the Ozempic brand has been heavily prescribed off-label for weight loss, which is what shows up as "off-label use" reports.
For tirzepatide, the dual-brand structure (Mounjaro for diabetes, Zepbound for weight loss, both approved) means there's less off-label prescribing because the on-label indication exists for both purposes. The lower rate is consistent with that.
Most divergent reactions: where the drugs differ most
Looking at reactions where the three drugs show the largest spread in per-case rates surfaces the differentiating features clearly:
Tirzepatide >> Semaglutide: - Incorrect dose administered: 212.4 vs 8.8 (24x) - Injection site pain: 99.9 vs 11.8 (8.5x) - Extra dose administered: 64.8 vs 2.9 (22x) - Injection site erythema: 35.3 vs 2.3 (15x) - Accidental underdose: 32.8 vs 0.05 (655x)
The device/dosing pattern dominates.
Semaglutide >> Tirzepatide: - Product use in unapproved indication: 45.7 vs 0.3 (153x) - Wrong technique in product usage process: 40.6 vs 8.6 (4.7x) - Weight decreased: 55.7 vs 15.0 (3.7x) - Decreased appetite: 70.8 vs 21.6 (3.3x) - Intestinal obstruction: 25.9 vs 5.3 (4.9x)
The off-label / weight-loss / GI-pharmacology pattern dominates.
These are the dimensions that actually distinguish the drugs in real-world reporting. Most other reactions are present for all three at broadly similar rates.
What this comparison can and can't tell you
It can tell you:
- Where the drugs differ in their real-world adverse-event reporting patterns
- Which reactions are class effects (present for all three) and which are drug-specific
- Where device/formulation matters more than molecule (tirzepatide's dosing errors)
- Which patient experiences are driving the public conversation (semaglutide's off-label use, tirzepatide's injection complexity)
It cannot tell you:
- Whether the drugs are equally effective (that's a different question; head-to-head trials show tirzepatide produces more weight loss than semaglutide at comparable doses)
- True incidence rates of any reaction (there's no denominator of total patients treated)
- Whether the same patient on different drugs would have different outcomes (only randomized trials can answer this)
- Whether the differences in reporting reflect true pharmacologic differences or differences in how the drugs are prescribed, taught, packaged, and discussed
For an individual choosing between these drugs with a clinician, the AEMS comparison is one input among several. The clinical trial data, the prescribing information, the patient's individual risk factors, and the practical considerations (cost, insurance coverage, supply availability) all matter at least as much.
How we analyzed this
Same methodology as our single-drug Ozempic article:
- Data source: FDA AEMS public quarterly data files, 2021 Q2 through 2026 Q1.
- Drug identification: Map each report to a canonical molecule (semaglutide, tirzepatide, liraglutide), handling brand names, misspellings, and combination products.
- De-duplication: Merge multi-source reports of the same case.
- Rate per 1,000 cases: Calculate per drug as (cases with reaction) / (total cases for drug) × 1,000.
- Most-divergent ranking: For each reaction reported by at least 3 cases per product, compute the ratio of highest to lowest per-1k rate across the three drugs.
You can run this same comparison yourself on our analytics page, including any pair or triple of drugs you want to look at. Add or remove drugs from the comparison and the rate matrix updates against the latest AEMS data.
Limitations specific to comparison analyses
In addition to the general FAERS/AEMS limitations (underreporting, no denominator, reporting bias) covered in the Ozempic article, comparison analyses have their own:
Differential reporting attention. Drugs that are in the news get more reports. Semaglutide has had unusual media attention through 2023-2026; tirzepatide has had less; liraglutide has had relatively little. Comparing them assumes the reporting rate is similar across them, which it isn't.
Differential reporter mix. Semaglutide cases come from a more diverse set of reporters (manufacturers, clinicians, patients, lawyers, pharmacies) than older drugs whose reporting is more clinician-driven. Different reporters report different things.
Reporting cluster contamination. As discussed above for liraglutide, a single cluster of cases can dominate the rates for a smaller-case-base drug. This is partly mitigated by looking at case-level patterns, but not eliminated.
Time-window mismatch. All three drugs were available for at least part of the 2021 Q2–2026 Q1 window, but their cumulative exposure differs. Tirzepatide had no reports before 2022; liraglutide had years of pre-window reports already in the database.
When using comparison analyses for clinical decision-making, treat them as hypothesis-generating, not definitive. They surface patterns; they don't establish that one drug is "safer" than another in any individual patient.
Run this comparison yourself. Every table and rate in this article comes from our analytics platform at AdverseEvent.ai. Free, no signup. Add any two-to-ten drugs to a comparison and the side-by-side rate matrix and divergent-events analysis update against the latest AEMS data. We built the tool so clinicians, researchers, journalists, and patients can compare drug safety profiles directly rather than relying on second-hand summaries.
What to do with this comparison
If you're a clinician choosing between these drugs:
- All three are GLP-1 receptor agonists with similar warning labels. The class-level safety profile (GI effects, pancreatitis risk, thyroid C-cell tumor warning, gallbladder events) applies to all of them.
- For GI tolerability, tirzepatide tends to look better per case than semaglutide; trial data supports this.
- For injection workflow, semaglutide and liraglutide tend to look simpler than tirzepatide; the dosing-error data supports this.
- Individual patient factors usually matter more than the AEMS comparison.
If you're a patient deciding between brands:
- The FDA-approved label for each drug is the authoritative source on safety warnings.
- The comparative AEMS data is real-world reporting, which includes the noise from off-label use, counterfeit products, and device complications. It is not the same as the controlled-trial safety profile.
- Your prescriber's experience with these drugs in your specific clinical situation is the most important input.
If you're a researcher or journalist:
- The data quality differences across the three drugs are real and matter for interpretation. Don't average them as if they were equivalent.
- The most informative single-pair comparison is probably semaglutide vs tirzepatide (similar case-base scale, similar reporting attention). Liraglutide's data has structural issues that complicate three-way comparisons.
- The "device/dosing error" dimension for tirzepatide is under-discussed in the literature relative to its prominence in the data. There's probably a paper there.
Further reading:
- How Ozempic (Semaglutide) shows up in FAERS/AEMS — single-drug companion analysis
- FDA AEMS Public Dashboard — official data source
- Ozempic prescribing information
- Mounjaro / tirzepatide prescribing information
- Victoza / liraglutide prescribing information
- How to fill out FDA Form 3500A — for healthcare professionals reporting adverse events to FDA