Tirzepatide safety in type 2 diabetes: a disproportionality analysis of adverse events using the FDA FAERS database

Abstract

This study evaluated adverse events reported with tirzepatide, a dual GIP/GLP-1 receptor agonist approved for type 2 diabetes and obesity, using real-world data from the FDA Adverse Event Reporting System. A disproportionality analysis was conducted on reports from May 2022 to the fourth quarter of 2024. Reports were deduplicated, normalized using standardized medical terminology, and analyzed using four disproportionality analysis algorithms. Significant signals required meeting all four methods’ criteria with at least three cases. Among 20,350 adverse event reports (68.0% female; median age 50.4 years), 105 significant adverse events were identified. Common events included gastrointestinal disorders (nausea and diarrhea) and injection-site reactions. The strongest signals were injection-site coldness and belching. Known risks such as pancreatitis (190 cases) and hypoglycemia (115 cases) were confirmed. Novel signals included upper respiratory infections and postmenopausal hemorrhage. The median onset time was 26 days, with 50% of events occurring within the first month. Older adults (65 years or older) experienced earlier onset (12 versus 31 days, significant difference). This analysis is consistent with known gastrointestinal and pancreatic risks of tirzepatide from prior clinical studies and identifies potential new safety concerns, underscoring the need for vigilant monitoring, particularly during initial treatment phases.

Introduction

Diabetes is a chronic metabolic disorder, characterized by persistent hyperglycemia, which can induce various tissue and organ damages and functional impairments, including cardiovascular, renal, ocular, and dermal soft tissue complications, posing a severe threat to human health (1, 2, 3, 4). Current estimates indicate that approximately 537 million adults (aged 20–79 years) worldwide have diabetes. This figure is projected to rise to 643 million by 2030 and 783 million by 2045 (5). Among diabetic patients, over 90% have type 2 diabetes (6). Obesity is also one of the most prevalent chronic diseases, affecting approximately 650 million adults globally (https://www.worldobesity.org/resources/resource-library/world-obesity-atlas-2022, accessed April 10, 2024). Excessive obesity and type 2 diabetes impose a substantial economic burden and are major contributors to global morbidity and mortality (7, 8).

However, diabetes currently lacks a curative treatment, and pharmacological interventions are generally the primary therapeutic strategy (9). Tirzepatide is a dual agonist of the glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide-1 (GLP-1) receptors, administered subcutaneously once weekly. As the first GLP-1/GIP receptor agonist, tirzepatide was approved by the FDA in 2022 for improving the glycemic control in patients with type 2 diabetes (10). It has also been approved by the FDA for chronic weight management, suitable for adult patients with obesity (BMI ≥30 kg/m²) or overweight (BMI ≥27 kg/m²). In January 2025, the FDA further approved tirzepatide (marketed as Zepbound) for moderate-to-severe obstructive sleep apnea in adults with obesity, making it the first drug approved for this indication (11). As a novel antidiabetic drug, tirzepatide is increasingly being used in clinical practice, and evaluating its relative efficacy and safety in treating type 2 diabetes and obesity is of significant importance. It has demonstrated unprecedented efficacy in glycemic control and weight reduction. Five clinical trials (SURPASS 1–5) involving subjects with type 2 diabetes have shown improvements in glycated hemoglobin (HbA1c), fasting blood glucose, and body weight when tirzepatide was used as monotherapy or in combination with metformin, sodium–glucose cotransporter-2 inhibitors (SGLT2is), and/or basal insulin (12, 13, 14). Notably, recent evidence suggests that tirzepatide may facilitate diabetes remission, with a meta-analysis demonstrating >16-fold increased odds of achieving normoglycemia (HbA1c ≤5.7%) compared to controls (15). However, conclusions regarding its safety remain controversial. Over the past few years, the safety of tirzepatide has been evaluated in monotherapy and combination therapy trials, and it is considered to be as safe as GLP-1RAs (16, 17). Nevertheless, conflicting opinions persist (12, 18). Moreover, post-marketing safety data from large-scale real-world studies are still limited. Recent pharmacovigilance studies using FDA Adverse Event Reporting System (FAERS) data have examined tirzepatide safety in mixed T2DM/obesity cohorts. However, none isolated pure T2DM populations, and most included secondary/concomitant drug roles, potentially confounding signal attribution (19, 20, 21, 22, 23). Our study addresses these gaps by characterizing AE profiles using rigorous disproportionality analysis. This study provides a T2D-specific analysis with updated data through Q4 2024, enhancing relevance to diabetic care. Critically, our analysis excluded compounded tirzepatide formulations due to unstandardized composition and potential safety concerns, focusing solely on FDA-approved products.

This study aims to characterize and delineate the features of tirzepatide-related adverse events using the FAERS database, exploring the types, frequencies, severities, and relative risks of adverse events induced by tirzepatide. Through this study, we hope to provide more comprehensive and specific information regarding the safety of tirzepatide.

Materials and methods

Data source and study design

The FAERS database is a crucial public resource maintained by the FDA for collecting adverse event (AE) reports and medication error reports associated with approved drugs (24, 25). We conducted a pharmacovigilance study on all adverse events reported with tirzepatide based on the FAERS database. Using the trade name or generic name of tirzepatide as keywords, we retrieved report data from the FAERS database from its market launch (May 2022) to the four quarter of 2024. The selected role code was ‘PS’ (primary suspect) and the indication was limited to ‘type 2 diabetes’. We deduplicated the reports obtained from the FAERS database by selecting the most recent FDA receipt date (FDA_DT) when the case identifier (CASE ID) was identical. If both CASE ID and FDA_DT matched, the report with the highest primary identifier (PRIMARY_ID) was retained, and reports with identical values for fields such as gender, age, country, event date, adverse reaction, and indication were also identified as duplicates. Figure 1 summarizes the inclusion/exclusion workflow.

Figure 1.

Study selection flowchart.

Adverse event terminologies were normalized to the preferred terms (PTs) using the Medical Dictionary for Regulatory Activities (MedDRA) version 27.0, facilitating consistency and comparability (26). AEs were categorized and delineated depending on the system organ class (SOC) for a structured overview of AE characteristics. A target reports/non-reports study employing a disproportionality approach was then conducted to identify signals for tirzepatide at both PT and SOC levels, in comparison to other medications, using R software version 4.2.0 and Microsoft Excel 2021.

Statistical methods

Descriptive analysis was employed to present the characteristics of all AE reports reported with tirzepatide. Disproportionality analysis methods, such as the reporting odds ratio (ROR), proportional reporting ratio (PRR), Bayesian confidence propagation neural network (BCPNN), and multiple gamma poisson shrinker (MGPS) (27, 28, 29, 30), were utilized to detect potential safety signals. In this study, we employed four methods – ROR, PRR, BCPNN, and MGPS – for AE signal detection to reduce the likelihood of false-positive signals. A potential risk signal is identified when all four algorithms indicated a positive signal, defined as follows: the frequency of the adverse event occurrence is ≥3, the lower limit of the 95% confidence interval (CI) for the ROR is >1, the PRR is ≥2 with a chi-square value ≥4, the lower limit of the 95% CI for the information component (IC) is >0, and the lower limit of the 95% CI for the Empirical Bayes Geometric Mean (EBGM) is ≥2.

A log-rank test was conducted to compare the differences in the onset time among different groups. The time to onset of adverse events was calculated as the interval between the start date of tirzepatide therapy and the onset date of the adverse event. Cumulative distribution curves were generated to visualize the time-to-onset patterns. Subgroup analyses were performed by age, gender, and treatment duration. Erroneous reports (e.g., negative time intervals or implausible dates) were excluded from this analysis.

Result

General characteristics

The clinical characteristics of tirzepatide-related AEs are presented in Table 1. A total of 20,350 AE reports were detected from the FAERS database. Regarding gender, AEs reported with tirzepatide were more frequently reported in females (68.0%) than in males (17.5%). The average age of the reported cases was 50.4 years. Dividing patients into four age groups using 18, 65, and 75 years as cutoffs, most patients were aged 65 years or younger (n = 6,379, 31.3%). Most reports originated from the United States (n = 20,191, 99.2%), and most reports were submitted by consumers (n = 19,069, 93.7%). Other serious consequences (n = 802, 51.4%) were the most frequently reported serious outcomes.

Table 1.

Clinical characteristics of reports with tirzepatide from the FAERS database.

Clinical characteristics		Frequency	Percentage
Gender
	Male	3,555	17.5%
	Female	13,835	68.0%
	Missing	2,960	14.5%
Age
	Mean (SD)	50.4	12.6
	Median (min, max)	50	(16, 90)
	Missing	12,939	63.6%
Age group
	0–17	2	0.01%
	18–64	6,377	31.3%
	65–74	813	3.40%
	≥75	219	1.08%
	Missing	12,939	63.6%
Country
	USA	20,191	99.2%
	JP	81	0.40%
	AE	57	0.28%
	SA	12	0.06%
	Other country	9	0.04%
Case priority
	Non-expedited	19,118	93.9%
	Expedited	952	4.68%
	Direct	224	1.10%
	30-day	56	0.27%
Reporter type
	Consumer	19,069	93.7%
	Other health-professional	710	3.49%
	Physician	351	1.72%
	Pharmacist	208	1.02%
	Lawyer	1	0.01%
	Missing	11	0.05%
Outcomes
	Hospitalization: initial or prolonged	611	39.1%
	Death	55	3.52%
	Life-threatening	38	2.43%
	Disability	28	1.79%
	Required intervention to prevent permanent impairment/damage	26	1.67%
	Congenital anomaly	1	0.06%
	Other serious	802	51.4%

Signal of preferred terms

The number of PT signals detected for tirzepatide varies across four different algorithms, with a total of 105 disproportionate PTs identified by all four algorithms. Supplementary Table S1 (see the section on Supplementary materials given at the end of the article) shows 105 PTs that met all four algorithm criteria at the PT level. The ROR method detected 160 signals; the PRR method detected 141 signals; the BCPNN method detected 147 signals; and the MGPS method detected 105 signals. Figure 2 depicts the number of signals and AEs reported with tirzepatide as the primary suspect (PS) at the SOC level. We found that tirzepatide induced 30,885 AEs across 20 organ systems. The signals reported with tirzepatide were primarily concentrated in the SOCs of general disorders and administration site conditions, and gastrointestinal disorders. AE reports were mainly focused on the SOCs of injury, poisoning and procedural complications, and general disorders and administration site conditions.

Figure 2.

Figures depicting the number of tirzepatide signals and reports at the SOC level in the FAERS database.

Among the 105 significantly disproportionate PTs, 52 PTs were reported in the product label or clinical trials, as shown in Table 2. These PTs were primarily concentrated in the general disorders and administration site conditions (n = 28) and gastrointestinal disorders (n = 13) SOCs. Injection site pain (n = 2,793), nausea (n = 2,519), injection site hemorrhage (n = 1,267), and diarrhea (n = 1,111) were reported in substantial numbers. Injection site coldness (EB05 = 42.61), injection site injury (EB05 = 39.09), and hunger (EB05 = 23.73) exhibited the strongest signal strengths. In patients receiving tirzepatide treatment, reports of pancreatitis (n = 190, EB05 = 6.85), hypoglycemia (n = 115, EB05 = 4.44), gallbladder disorder (n = 38, EB05 = 3.48), cholecystitis (n = 18, EB05 = 2.00), diabetic retinopathy (n = 9, EB05 = 3.08), biliary colic (n = 8, EB05 = 2.08), and medullary thyroid carcinoma (n = 3, EB05 = 3.55) were observed.

Table 2.

Signal strength of adverse events that appear in the drug label of tirzepatide (n = 52) at the preferred term (PT) level that fit four algorithms simultaneously.

Preferred terms (PT)	n	RORL0.05(−)	PRR	IC-2SD	EB05
Injection site pain	2,793	16.05	15.68	3.78	13.83
Nausea†	2,519	5.26	5.21	2.28	4.88
Injection site hemorrhage	1,267	24.62	25.39	4.36	20.82
Diarrhea†	1,111	2.33	2.43	1.18	2.27
Vomiting†	958	3.34	3.51	1.69	3.24
Injection site erythema	896	15.37	16.15	3.76	13.80
Decreased appetite†	706	4.18	4.45	2.01	4.04
Constipation†	665	4.34	4.63	2.06	4.19
Injection site bruising	572	12.35	13.29	3.48	11.35
Injection site pruritus	503	13.41	14.54	3.58	12.25
Abdominal pain upper	390	2.70	2.97	1.40	2.65
Dyspepsia	335	4.97	5.51	2.24	4.80
Abdominal discomfort	324	2.31	2.57	1.18	2.28
Injection site mass	294	10.66	11.93	3.26	9.93
Injection site rash	281	13.29	14.93	3.54	12.18
Injection site swelling	280	5.62	6.30	2.41	5.41
Injection site urticaria	232	13.29	15.13	3.52	12.17
Abdominal distension	214	2.90	3.31	1.49	2.85
Injection site injury	211	54.13	63.16	4.94	39.09
Flatulence	206	5.43	6.21	2.35	5.24
Pancreatitis*	190	7.18	8.28	2.72	6.85
Gastrointestinal disorder*	187	2.81	3.24	1.45	2.76
Hunger	187	24.21	28.17	4.19	20.69
Gastroesophageal reflux disease*	158	3.06	3.58	1.56	3.01
Injection site reaction	153	4.00	4.68	1.93	3.90
Hypoglycemia*	115	4.58	5.50	2.10	4.44
Impaired gastric emptying	78	11.99	15.10	3.23	11.03
Injection site discomfort	71	7.62	9.67	2.69	7.23
Injection site irritation	70	10.57	13.47	3.07	9.81
Injection site warmth	58	6.16	8.01	2.40	5.90
Gallbladder disorder*	38	3.56	4.92	1.66	3.48
Injection site coldness	34	70.64	108.95	3.52	42.61
Injection site discoloration	34	3.65	5.13	1.67	3.56
Injection site paresthesia	33	21.86	31.72	3.31	18.41
Injection site induration	27	3.45	5.05	1.56	3.36
Injection site hypersensitivity	24	7.74	11.71	2.36	7.26
Starvation	23	30.65	48.82	3.06	23.73
Appetite disorder	22	2.75	4.21	1.25	2.70
Injection site vesicles	22	3.53	5.39	1.54	3.43
Injection site inflammation	21	4.24	6.55	1.72	4.10
Cholecystitis*	18	2.03	3.23	0.85	2.00
Injection site scar	18	5.86	9.43	1.97	5.58
Injection site indentation	12	3.71	6.60	1.35	3.59
Injection site laceration	11	22.54	43.85	2.09	17.88
Injection site hypoesthesia	10	4.48	8.47	1.39	4.29
Diabetic retinopathy*	9	3.18	6.18	1.06	3.08
Biliary colic*	8	2.12	4.29	0.65	2.08
Injection site scab	5	4.78	11.84	0.80	4.48
Lack of satiety	4	8.78	25.14	0.61	7.65
Hypoglycemic unconsciousness*	4	3.23	8.83	0.41	3.08
Medullary thyroid cancer*	3	3.79	12.26	0.13	3.55
Injection site macule	3	7.28	24.51	0.19	6.37

^*PTs listed are present in the warnings and precautions section of the product label.

^† PTs listed are among the most common adverse reactions in the prescribing information.

Among the 105 significantly disproportionate PTs, the remaining 53 PTs represent not previously reported findings related to tirzepatide-associated AEs, as shown in Table 3. These PTs were primarily concentrated in the investigations (n = 14) and injury, poisoning and procedural complications (n = 12) SOCs. Incorrect dose administered (n = 5,864), off-label use (n = 3,145), and underdose (n = 974) were reported in substantial numbers. Belching (EB05 = 33.35), incorrect dose administered (EB05 = 31.87), and underdose (EB05 = 30.16) exhibited the strongest signal strengths. We categorized signals into high, moderate, and low priority based on: severity, frequency, mechanistic plausibility and clinical utility. High priority signals are highlighted as critical for immediate clinical action, such as optimizing prescribing protocols and investigating off-label safety. Moderate priority signals are framed as manageable through patient education and monitoring. Low priority signals are interpreted with caution due to small sample sizes and lack of mechanistic evidence, emphasizing the need for larger studies to validate associations.

Table 3.

Signal strength of adverse events that are not previously reported findings of tirzepatide-related AEs (n = 53) at the preferred term (PT) level that fit four algorithms simultaneously.

Preferred terms (PT)	n	RORL0.05(−)	PRR	IC-2SD	EB05	Clinical relevance tier
Incorrect dose administered	5,864	45.74	40.81	4.99	31.87	High
Off label use	3,145	3.49	3.43	1.70	3.26	High
Extra dose administered	974	38.17	40.05	4.87	30.16	High
Inappropriate schedule of product administration	862	3.38	3.57	1.71	3.28	Moderate
Weight decreased	593	2.74	2.95	1.42	2.69	Moderate
Blood glucose increased	500	5.34	5.79	2.35	5.14	Moderate
Eructation	422	43.25	47.76	4.92	33.35	High
Injury associated with device	231	10.50	11.95	3.23	9.80	Moderate
Blood glucose decreased	229	10.14	11.54	3.18	9.48	Moderate
Dehydration	210	2.56	2.93	1.32	2.52	Moderate
Accidental underdose	204	24.35	28.13	4.21	20.81	Moderate
Product administered at inappropriate site	192	9.09	10.48	3.03	8.56	Moderate
Increased appetite	149	12.53	14.78	3.40	11.53	Moderate
Glycosylated hemoglobin increased	135	7.36	8.73	2.73	7.01	Moderate
Product prescribing error	108	2.75	3.32	1.41	2.71	Moderate
Feeding disorder	106	5.48	6.64	2.33	5.29	Moderate
Blood glucose abnormal	72	5.11	6.46	2.20	4.94	Moderate
Accidental overdose	58	2.60	3.37	1.30	2.56	Moderate
Food craving	46	13.80	18.70	3.19	12.44	Moderate
Glycosylated hemoglobin decreased	29	15.39	22.68	3.01	13.59	Moderate
Lipase increased	27	5.24	7.70	2.03	5.03	Moderate
Wrong patient received product	27	6.23	9.17	2.21	5.93	Moderate
Blood glucose fluctuation	24	3.23	4.85	1.46	3.16	Low
Gastrointestinal sounds abnormal	23	4.07	6.16	1.70	3.94	Low
Sleep disorder due to general medical condition, insomnia type	20	2.36	3.67	1.05	2.32	Low
Frequent bowel movements	19	7.39	11.76	2.19	6.93	Low
Fluid intake reduced	17	3.01	4.88	1.29	2.94	Low
Menstrual disorder	17	2.47	3.99	1.07	2.42	Low
Glycosylated hemoglobin abnormal	14	5.26	9.01	1.73	5.01	Low
Food aversion	12	5.85	10.49	1.71	5.53	Low
Pancreatic enzymes increased	12	9.18	16.62	1.98	8.38	Moderate
Counterfeit product administered	11	6.72	12.40	1.73	6.28	Low
Product after taste	10	3.05	5.73	1.09	2.97	Low
Allodynia	9	6.60	13.02	1.54	6.15	Low
Postmenopausal hemorrhage	9	3.39	6.60	1.11	3.28	Moderate
Vomiting projectile	9	3.18	6.18	1.06	3.08	Low
Product tampering	8	3.97	8.07	1.15	3.80	Low
Abdominal rigidity	7	2.05	4.35	0.56	2.01	Low
Thirst decreased	7	7.62	16.58	1.33	6.96	Low
Gastric pH decreased	6	4.62	10.54	0.98	4.36	Low
Intercepted product selection error	6	6.74	15.56	1.11	6.19	Low
Incorrect product formulation administered	6	3.95	9.00	0.90	3.77	Low
Lymph node pain	6	2.23	5.03	0.54	2.18	Low
Motion sickness	6	2.23	5.03	0.54	2.18	Low
Thyroid hormones increased	6	2.48	5.60	0.62	2.42	Low
Early satiety	5	2.38	5.80	0.46	2.31	Low
Oligomenorrhea	5	3.20	7.86	0.63	3.07	Low
Skin sensitization	5	3.65	8.98	0.69	3.48	Low
Blood insulin increased	4	3.72	10.21	0.46	3.52	Low
Body mass index decreased	3	2.48	7.91	0.02	2.38	Low
Application site wound	3	5.71	18.86	0.19	5.15	Low
Thyroid cyst	3	2.56	8.17	0.03	2.46	Low
Upper respiratory tract infection	3	3.79	12.26	0.13	3.55	Low

Onset time of events

Time to onset was calculated by subtracting the event start date from the therapy start date, and cumulative distribution curves were used to present event onset information for different groups. The time to onset of tirzepatide-related AEs was collected from the database. After excluding false reports, only 1,022 AE reports (5.0% of total) included the time of onset, with more than 50% of the related AE cases occurring within the first month of initiating tirzepatide treatment, with a median time to onset of 26 days (interquartile range (IQR) 7–83 days). The cumulative distribution of adverse event onset times for the subgroups is illustrated in Fig. 3. The median onset time for females is 28 days, while for males, it is 21.5 days. Gender may not affect the median time to onset of tirzepatide-related AEs (P = 0.1379). The median time to onset in the ≥65 years age group was significantly shorter than in the <65 years group (12 vs 31 days, P < 0.001). For cases during the dose-stabilization period (>20 weeks), the median time to onset was 58 days longer than for cases during the dose-escalation period (≤20 weeks) (70 vs 12 days, P = 0.008). Different dosing regimens also influenced the median time to onset of tirzepatide-related AEs (P < 0.001), with smaller doses resulting in shorter median onset times. When stratified by dose (5 mg vs 10/15 mg), the median onset time was shorter in the 5 mg group (35 days (IQR 7–71)) compared to higher doses (143 days (IQR 14–213); P <0.001), aligning with the overall trend.

Figure 3.

Cumulative distribution of adverse event onset times for the subgroups.

Discussion

Previous studies have primarily focused on clinical trials and literature analyses of tirzepatide (31, 32), with few articles addressing the latest real-world research. Based on large-scale real-world data from the FAERS database, a key global pharmacovigilance resource, we collected real-world safety reports and conducted a pharmacovigilance analysis of tirzepatide. The objective was to analyze novel and meaningful adverse reaction signals to generate hypotheses for future studies, which may contribute to the ongoing safety assessment of tirzepatide.

AEs reported with tirzepatide were more frequently reported in females (68.0%) than in males (17.5%). This may be related to increased tirzepatide use among females for weight loss purposes (31). This gender distribution may reflect differential tirzepatide prescribing patterns. The median time to onset was similar between females (28 days) and males (21.5 days), with no statistically significant difference (P = 0.1379). However, this aggregate analysis across diverse adverse events may mask gender-specific variations in individual AE types. Our results indicate that patients aged 65 years and older accounted for a smaller proportion of reports (4.48%), and the median time to onset in the ≥65 years age group was significantly shorter than in the <65 years group (12 vs 31 days, P < 0.001). A study by Kiyosue et al. indicated that although the incidence of adverse events with tirzepatide did not appear to be high in participants aged 65 years or older, the rate of discontinuation due to adverse events was higher (32). Simultaneously, this analysis showed that tirzepatide had similar safety profiles across age subgroups among East Asian clinical trial participants, with no significant differences in the incidence of adverse events across the various age groups, and most were comparable to the overall population (32). The shorter median onset time in older patients observed in our study, despite similar AE incidence across age groups in clinical trials (32), may indicate heightened sensitivity or earlier symptom reporting in this population. However, this finding should be interpreted cautiously due to potential confounding factors, such as comorbidities or concomitant medications. Clinicians should be aware that some elderly individuals receiving tirzepatide treatment may experience treatment discontinuation due to adverse events. The most common AEs leading to discontinuations were nausea, decreased appetite and decreased weight. As tirzepatide is a newly approved drug with significant public interest for weight loss, stimulated reporting bias may inflate AE frequencies (particularly for non-serious events such as injection site reactions) due to heightened patient and media awareness. This could partially explain the high proportion of consumer-submitted reports (93.7%).

According to the disproportionality analysis, the most frequently reported and significant signals at the SOC level were general disorders and administration site conditions, and gastrointestinal disorders. These adverse events primarily included injection site reactions, nausea, diarrhea, and vomiting, which are generally documented in the tirzepatide product label (https://uspl.lilly.com/mounjaro/mounjaro.html#pi, accessed April 10, 2024). In clinical trials involving tirzepatide, the most commonly reported adverse events were gastrointestinal-related, consistent with our findings. These events occurred at mild-to-moderate severity, primarily during the dose-escalation period, and decreased over time, especially after reaching the maintenance dose (33, 34, 35, 36). Our study found that the median time to onset for cases during the dose-stabilization period (>20 weeks) was 58 days longer than for cases during the dose-escalation period (≤20 weeks). The prolonged median AE onset time in the dose-stabilization period could be influenced by attrition bias. As participants experiencing significant AEs during dose-escalation were more likely to withdraw, the stabilized cohort inherently comprised a subpopulation with demonstrated tolerance to initial doses. Consequently, the apparent temporal shift in AE occurrence may not directly indicate enhanced safety of tirzepatide during stable dosing, but rather reflect the survival of a pre-selected group. In addition, delayed AE onset does not preclude severe outcomes, necessitating long-term monitoring. Theoretically, while tirzepatide acts on the GLP-1 receptor, its activation of the GIP receptor does not directly influence gastrointestinal motility and secretory function, potentially not increasing the incidence of gastrointestinal adverse events. A study by Karagiannis et al. found that tirzepatide was associated with a higher incidence of gastrointestinal adverse events (such as nausea, vomiting, and diarrhea) compared to placebo or basal insulin, particularly with the 10 and 15 mg doses showing increased odds of vomiting and diarrhea; however, the odds of gastrointestinal events were comparable between tirzepatide and GLP-1 RAs (37). GLP-1 acts as an inhibitor of gastric and pancreatic motility and maintains postprandial glucose stability (38). Therefore, we should recognize that gastrointestinal adverse events are the most common, which may not only contribute to the weight loss effect achieved by reducing appetite but also be a source of discomfort for patients.

Among all adverse events, those involving warnings and precautions from the product label remain noteworthy. Our findings revealed risk signals associated with thyroid C-cell tumors, pancreatitis, hypoglycemia, and acute gallbladder diseases, consistent with the tirzepatide prescribing information. In a 2-year clinical carcinogenicity study with plasma exposure, tirzepatide induced dose-dependent and treatment duration-dependent increases in the incidence of thyroid C-cell tumors in male and female rats. However, it remains unclear whether tirzepatide causes human thyroid C-cell tumors, including medullary thyroid carcinoma, and no cases of medullary thyroid carcinoma were reported in clinical trials (39). Nonetheless, our results yielded a signal for medullary thyroid carcinoma (n = 3, EB05 = 3.55), suggesting that tirzepatide may induce medullary thyroid carcinoma in humans. Tirzepatide is contraindicated in patients with a personal or family history of medullary thyroid carcinoma. Our results also revealed a signal for pancreatitis (n = 190, EB05 = 6.85), with adjudicated cases of pancreatitis reported in clinical trials (39). After initiating tirzepatide treatment, patients should be observed carefully for signs and symptoms of pancreatitis (including persistent severe abdominal pain, which may radiate to the back, and which may or may not be accompanied by vomiting). If pancreatitis is suspected, tirzepatide should be discontinued, and appropriate management initiated. Whether tirzepatide itself causes hypoglycemia remains controversial, with earlier opinions suggesting that tirzepatide monotherapy may not induce hypoglycemia (40, 41). However, some studies have acknowledged this risk (42), and our findings also yielded a significant signal for hypoglycemia. Clinical observations indicate that patients receiving tirzepatide in combination with insulin secretagogues (such as sulfonylureas) or insulin may be at an increased risk of hypoglycemia, including severe hypoglycemia (40, 41). Collectively, these findings suggest that the hypoglycemic risk associated with tirzepatide should not be entirely disregarded. In addition, we identified signals related to acute gallbladder diseases (cholecystitis, biliary colic, and gallbladder disorders). In placebo-controlled clinical trials of tirzepatide, 0.6% of patients treated with tirzepatide and 0% of patients treated with placebo reported acute gallbladder disease. Tirzepatide’s components, including GIP and GLP-1 RA, have been associated with gallbladder or biliary tract diseases. GLP-1 has been shown to impair gallbladder motility and contractility by inhibiting the secretion of cholecystokinin, a hormone involved in fat digestion and absorption, which may contribute to the development of gallbladder or biliary tract diseases (43). Moreover, recent reports have implicated GIP in gallbladder relaxation (44). A study by Zeng et al. found that tirzepatide increased the risk of gallbladder or biliary tract diseases in patients with T2D and obesity, for gallbladder or biliary disease, and the composite of gallbladder or biliary disease was significantly associated with tirzepatide compared with placebo or basal insulin (45). These findings may explain the observed gallbladder disease risk signals in our study.

We identified some important safety signals not included in the tirzepatide prescribing information. The most frequently reported signal was related to off-label use (n = 11,457), highlighting the need for strict adherence to the prescribing information when using tirzepatide clinically. Belching, a gastrointestinal signal not mentioned in the label, exhibited the strongest signal strength. Furthermore, signals for serious adverse events such as upper respiratory tract infections and postmenopausal hemorrhage emerged with tirzepatide. The precise mechanism of action of tirzepatide on these AEs and the potential underlying association have not been fully explored, warranting further clinical investigation. Due to the limited sample size and lack of updated real-world evidence, these conclusions may be subject to bias. Most AE reports have not been clinically verified nor had causality established. In summary, as more reports are submitted, the safety signal spectrum of tirzepatide may evolve over time.

Our results indicate a median time to onset of 26 days, with most cases occurring within the first month after initiating tirzepatide treatment, although this analysis was limited to only 5% of total reports with available onset data. These findings suggest that we should pay particular attention to adverse events during the first month, enabling early identification of adverse events induced by tirzepatide therapy. While this highlights the importance of early monitoring, clinical vigilance should prioritize adverse events by their severity and clinical relevance (e.g., gastrointestinal disorders, pancreatitis) rather than solely by temporal patterns. Different dosing regimens also influenced the median time to onset of tirzepatide-related AEs. Yu et al. conducted a meta-analysis in patients with diabetes to evaluate the optimal dose of tirzepatide for treating type 2 diabetes. Regarding safety, tirzepatide 5 mg > 10 mg = 15 mg, and they recommended tirzepatide 5 mg as the preferred initial dose for patients with type 2 diabetes to maximize the minimization of adverse events while lowering blood glucose and reducing body weight (46). Studies have shown that higher doses of tirzepatide are more likely to be associated with discontinuation due to AEs; drug discontinuation due to AEs was highest with the 15 mg dose of tirzepatide (10%) (47). The observed shorter median onset time at lower doses may reflect early treatment discontinuation due to initial intolerance or heightened awareness of mild symptoms during dose initiation. This contrasts with the lower overall AE incidence at lower doses reported in clinical trials (46, 47), which likely reflects better long-term tolerability. Thus, while lower doses may be associated with earlier AE onset, they remain preferable for minimizing the overall AE burden.

Based on the FAERS database, this study mined and analyzed adverse reaction signals for tirzepatide, exploring tirzepatide-related gastrointestinal disorders and specific adverse events, and other meaningful adverse events, to provide a reference for improving the safety of clinical drug use. Our investigation encountered several inherent limitations. The FAERS database is a spontaneous reporting system, with 93.7% of reporters in this study being patients. This can lead to bias stemming from concerns and fears regarding the relationship between tirzepatide and adverse events (AEs), rather than actual medication-related AEs. Therefore, the findings and interpretations presented in this study should be considered preliminary evidence rather than definitive conclusions. In addition, the FAERS database has its own limitations, including underreporting, duplicate reports, and incomplete case information. The lack of data on underlying conditions and concomitant medications limits our ability to describe the complete treatment regimen for tirzepatide solely based on adverse event data, without considering patients’ potential diseases or concurrent treatments. Consequently, this prevents us from conducting a comprehensive causal analysis. Finally, the information in the reports used for analysis is often not medically validated, making it impossible to identify the complete target population for tirzepatide or to calculate the incidence of AEs accurately. In addition, the high proportion of missing values across key variables (e.g., onset time available in only 5% of reports, incomplete dose and age data) limits the robustness of our analyses and generalizability of findings. Furthermore, disproportionality analysis is inherently hypothesis-generating and cannot establish causality or quantify risk. It serves to identify potential safety signals for further validation through rigorous epidemiological studies. Given these factors, our findings should be interpreted as preliminary evidence, necessitating cautious clinical interpretation. Given these factors, our report may likely be flawed, necessitating a cautious approach to interpreting the AEs associated with tirzepatide. However, despite some limitations of the FAERS database in pharmacovigilance research, comprehensively characterizing AE signals for tirzepatide and identifying some not previously reported AE signals may provide a foundation for further clinical research on tirzepatide. Continued monitoring of tirzepatide safety is warranted.

Our pharmacovigilance analysis of the FAERS database comprehensively and systematically revealed the safety signals and timing of adverse event occurrence reported with tirzepatide. We identified multiple post-marketing safety signals consistent with clinical trials, and other reported events that require further regulatory investigation to determine their significance. Novel significant AEs, such as belching, upper respiratory tract infections, and postmenopausal hemorrhage, also emerged. For better utilization of tirzepatide, further cohort studies and long-term clinical studies are needed to validate these findings and interpret its safety profile.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the work reported.

Funding

This work did not receive any specific grant from any funding agency in the public, commercial, or not-for-profit sector.

Author contribution statement

Zhenpo Zhang contributed to the study conception, design, and data collection from the FAERS database. Jiangxiong Li, Jingping Zheng and Yankun Liang performed data curation, statistical analysis, and visualization. Lin Ma and Ling Su supervised the research, validated the results, and provided critical revisions to the methodology. The first draft of the manuscript was written by Zhenpo Zhang and Jiangxiong Li, with subsequent revisions and intellectual input from Jingping Zheng, Yankun Liang, Lin Ma, and Ling Su. All authors reviewed and approved the final manuscript.

Ethics approval

Not applicable. The FDA Adverse Event Reporting System is a spontaneous reporting system, the publicly available data are anonymized, and therefore, obtaining consent to participate is not applicable. The present pharmacovigilance study was conducted using a public database of spontaneous reports. Given the use of deidentified data, ethical approval was not necessary.