Safety evaluation of baloxavir marboxil: analysis and discussion utilizing real adverse events from the FAERS database

Abstract

Background

As a novel anti-influenza agent, baloxavir marboxil lacks real-world safety data in large populations. Therefore, this study aimed to investigate adverse drug events (ADEs) associated with baloxavir marboxil by analyzing the Food and Drug Administration Adverse Event Reporting System (FAERS) database.

Methods

Adverse event reports involving baloxavir marboxil were extracted from the FAERS database spanning the fourth quarter of 2018 to the third quarter of 2023. Demographic characteristics and reporter profiles were analyzed to characterize the exposed population. A disproportionality analysis was performed using four validated pharmacovigilance algorithms: reporting odds ratio (ROR), proportional reporting ratio (PRR), Bayesian Confidence Propagation Neural Network (BCPNN), and multi-item gamma Poisson shrinker (MGPS). These complementary approaches were employed to detect, prioritize, and validate potential safety signals.

Results

Analysis of 8,824,675 ADE reports from the FAERS database identified 1,654 cases (0.19%) associated with baloxavir marboxil. Pediatric patients (< 18 years) exhibited the highest ADE reporting rate. Geospatial analysis revealed marked clustering, with 98.97% of reports originating from the United States (63.2%) and Japan (35.77%). We detected 47 significant safety signals spanning 27 System Organ Classes (SOCs), including established reactions such as pneumonia (n = 90) and vomiting (n = 77). Novel signals emerging from the analysis comprised hemorrhagic diathesis (n = 3), rhabdomyolysis (n = 25), hepatic dysfunction (n = 13), and cardiorespiratory arrest (n = 7). Notably, bleeding-related events (e.g., ischemic colitis, IC025 = 5.03) and neurological complications (e.g., febrile delirium, IC025 = 9.12) demonstrated statistically significant associations.

Conclusion

This pharmacovigilance study identifies previously undercharacterized safety signals associated with baloxavir marboxil, including hemorrhagic complications, liver dysfunction, rhabdomyolysis, and life-threatening cardiorespiratory events. Pediatric populations and patients on anticoagulants may require heightened monitoring. While these findings provide critical pharmacovigilance insights, our study is inherently constrained by the spontaneous reporting system, which introduces potential underreporting, reporting biases, and confounding factors. Future research could employ more rigorous prospective study designs, integrating clinical trials and epidemiological studies, to more accurately assess the safety risks of baloxavir marboxil.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40360-025-00940-0.

Introduction

Influenza is a viral acute respiratory infection with global prevalence. It imposes significant health, economic, and medical burdens worldwide [1, 2]. Most influenza cases manifest as self-limiting upper respiratory symptoms. However, high-risk groups—including adults ≥ 65 years, chronic cardiopulmonary patients, and immunocompromised individuals—experience disproportionately severe outcomes [3, 4]. A retrospective cohort study shows 9-fold higher hospitalization odds in elderly (≥ 65 years) versus 5–17 year-olds (OR = 9.4, 95%CI 8.8–10.1) [5]. Comorbidities elevate this risk by 2–3 times [5]. A meta-analysis of 63 global studies estimates 5.7 million annual influenza-related lower respiratory hospitalizations. While younger adults have higher absolute numbers, older adults exhibit 5-fold greater age-standardized rates, indicating increased biological vulnerability to severe complications and mortality [6].

Influenza vaccination is the most cost-effective prevention for high-risk groups. However, limitations like low coverage, waning immunity in elders, and viral evolution (antigenic drift/shift) reduce efficacy. These constraints emphasize the critical need for antiviral therapies [7, 8].

Neuraminidase inhibitors (NAIs) like oseltamivir and zanamivir have critical limitations in influenza treatment. Key issues include subtype-dependent efficacy, emerging resistance (e.g., NA-H274Y mutation against oseltamivir), strict reliance on early administration, and reduced potency against strains such as H7N9 [9, 10]. Combined phenotypic and genomic analyses show low global NAI resistance rates (0.5% in 2018–2019; 0.6% in 2019–2020). However, these data cannot eliminate risks from subtype variations, treatment delays, or viral evolution [11]. Collectively, these limitations underscore the urgent need for next-generation antiviral strategies to address both current and emerging influenza threats.

Baloxavir marboxil (Xofluza) overcomes key limitations of existing influenza treatments. This novel antiviral shows potent activity against influenza A/B and oseltamivir-resistant strains. Its single-dose oral regimen improves adherence and accessibility compared to multi-dose therapies like oseltamivir [12]. After oral intake, the prodrug rapidly converts to active baloxavir acid in intestinal cells, blood, and liver. The metabolite specifically inhibits the PA endonuclease—a viral enzyme critical for cap-dependent transcription. This mechanism blocks viral replication with high specificity [13, 14].

The clinical efficacy of baloxavir marboxil is supported by its global regulatory approvals. It received initial approval in Japan (February 2018) and the U.S. Food and Drug Administration (FDA) (October 2018) for acute influenza in patients aged ≥ 12 years. Subsequent authorizations include the European Medicines Agency (EMA) approval (January 2021) for treatment and post-exposure prophylaxis, and China’s National Medical Products Administration (NMPA) approval (April 2021) for uncomplicated influenza in the same age group [15, 16]. Phase 3 trials (CAPSTONE-1/2) and post-marketing data indicate a favorable safety profile for baloxavir marboxil. However, real-world evidence gaps persist. Current studies focus on short-term outcomes in immunocompetent populations, while critical areas—including drug interactions, special groups (immunocompromised/elderly patients), and long-term risks—remain understudied. Given these knowledge gaps and its recent market authorization (2018–2021), sustained pharmacovigilance is essential to systematically assess long-term safety, resistance evolution, and risks in underrepresented populations.

The FAERS is a database established to assist the FDA in compiling information regarding adverse drug events (ADEs) and medication errors. It collects reports of ADEs and medication errors from both within and outside the United States, enabling the FDA to effectively monitor and address concerns related to drug safety [17]. This database provides crucial real-world data, facilitating a comprehensive assessment of drug safety and risk profiles during medication usage.

This study employs advanced data mining methodologies to statistically analyze real adverse reaction signals of baloxavir marboxil post-market release. The primary objective is to offer clinical insights for prudent drug usage and to mitigate the associated risks of adverse reactions.

Methods

Data source

The FAERS has been publicly accessible since 2004, compiling adverse event reports from healthcare professionals, drug manufacturers, patients, and other stakeholders globally. Recognized for its extensive data volume and standardized reporting structure, FAERS serves as a critical resource for international pharmacovigilance research. For this study, data spanning from the fourth quarter of 2018 to the third quarter of 2023 were extracted and analyzed using R version 4.3.2. Most reports originated in the US and Japan (baloxavir marboxil’s early approval regions in 2018), reflecting high clinical adoption. Only 1.03% came from other countries, indicating limited post-marketing data from regions with delayed approvals or low prescription rates.

Data mining algorithm

Preliminary data processing involved the removal of duplicate reports and the retention of the latest report based on date for cases with identical caseid in the DEMO table. Drug names were standardized utilizing the Medex_UIMA_1.8.3 system. Reports pertaining to adverse reactions suspected to be primarily associated with baloxavir marboxil were extracted, covering clinical characteristics such as gender, age, reporting region, reporter, reporting time, and outcome.

In this study, we utilized a disproportionality analysis commonly employed in pharmacovigilance studies to identify potential signals between ADEs. We employed a combination of ROR [18], PRR [19], BCPNN [20], and MGPS [21] algorithms. The methodological rationale for selecting ROR, PRR, BCPNN, and MGPS lies in their complementary strengths for pharmacovigilance signal detection: PRR offers simplicity and rapid screening of high-frequency events through proportional comparisons but risks bias when drugs alter baseline adverse event rates, while ROR mitigates such bias by modeling data as a case-control study to estimate relative risk through strategic control selection, albeit requiring meticulous exclusion of confounders and being sensitive to reporting heterogeneity. BCPNN enhances specificity by handling sparse data via Bayesian probability adjustments and quantifying uncertainty with information components (IC), though its computational complexity demands expertise and may oversmooth low-frequency signals, whereas MGPS prioritizes rare event detection through shrinkage stabilization of estimates, excelling in longitudinal risk monitoring but underperforming for high-frequency events due to overcorrection. Integrating these methods leverages their complementary strengths: PRR and ROR provide rapid initial screening and cross-validation of common signals, BCPNN enhances specificity by incorporating Bayesian uncertainty and multi-variable analysis, and MGPS refines detection of rare events while minimizing noise. This multi-algorithm approach mitigates individual limitations, improves signal robustness, and reduces both false positives and false negatives through statistical triangulation. Specific algorithms and formulas are detailed in the appendix (Methodology.docx file).

Signal filtering and categorization

Beyond deduplication, we addressed inherent FAERS data biases through algorithmic compensation and threshold optimization: PRR and ROR adjusted for denominator distortions by excluding disease-related confounders to isolate drug-specific signals, while MGPS and BCPNN incorporated shrinkage techniques and Bayesian priors to stabilize estimates of rare adverse events, thereby mitigating noise and enhancing the reliability of low-frequency associations through probabilistic adjustments. Preferred terms (PTs) with a minimum report count of ≥ 3 were selected for further analysis. We utilized the Medical Dictionary for Regulatory Activities (MedDRA) version 26.1 to encode, classify, and localize signals using PTs and System Organ Classes (SOCs). This enabled a focused analysis of specific SOCs relevant to adverse event signals. PTs categorized as “No Adverse Event” were excluded from the study results. The detailed date mining process is presented in Fig. 1.

Fig. 1.

The flow diagram of selecting baloxavir marboxil-related ADEs from FAES database

Results

Basic characteristics of baloxavir marboxil-related ADEs

From the fourth quarter of 2018 to the third quarter of 2023, the FAERS database recorded 8,824,675 adverse event reports. Of these, 1,654 listed baloxavir marboxil as the primary suspected drug for ADEs. Sex data were missing in 21.64% of reports and age data in 35.91%, reflecting incomplete reporting. baloxavir marboxil was primarily prescribed for influenza (66.57% of cases). The proportion of ADE reports was slightly higher in female patients compared to males, with rates of 41.90% and 36.46%, respectively. Analysis of age distribution revealed a significantly higher frequency of reported ADEs among individuals under 18 years old (25.70%) compared to other age groups, followed by the 18 to 45 age group (13.54%). The United States accounted for 59.79% of reports and Japan for 39.18%. Serious adverse events of baloxavir marboxil mainly included hospitalization, life-threatening events, death, disability, congenital anomalies, and unknown outcomes. Among these, hospitalization was the most common serious adverse event, accounting for 33.78% of cases, followed by death and life-threatening events, which accounted for 11.06% and 7.25% of cases, respectively. Additionally, unknown severe medical events represented 46.56% of cases. Details are provided in Table 1.

Table 1.

Basic information on ADEs related to baloxavir marboxil from the FAERS database

Variable	Total
Year
2018	10(0.60)
2019	779(47.10)
2020	640(38.69)
2021	39(2.36)
2022	72(4.35)
2023	114(6.89)
Sex
Female	693(41.90)
Male	603(36.46)
*Unknown	358(21.64)
Age_yr	32.00(11.00,60.00)
Age_yrQ
<18	425(25.70)
18 ~ 45	224(13.54)
45 ~ 65	188(11.37)
65 ~ 75	85(5.14)
>=75	138(8.34)
*Unknown	594(35.91)
wt	67.75(54.48,81.93)
Reporter
Physician	624(37.73)
Consumer	578(34.95)
Pharmacist	365(22.07)
Other health-professional	86(5.20)
*Unknown	1(0.06)
Reported countries
United States	989(59.79)
Japan	648(39.18)
Other	17(1.03)
Route
Oral	929(56.17)
Other	725(43.83)
Outcomes
Other serious	379(46.56)
Hospitalization	275(33.78)
Death	90(11.06)
Life threatening	59(7.25)
Disability	9(1.11)
Congenital anomaly	2(0.25)
tto	1.00(0.00,2.00)
ttoQ
<7	565(71.79)
7 ~ 28	43(5.46)
28 ~ 60	2(0.25)
>=60	4(0.51)
Unknown	173(21.98)
Indications
Antiviral prophylaxis	11(0.67)
Covid-19	22(1.33)
Influenza	1101(66.57)
Others	25(1.51)
Product used for unknown indication	493(29.81)
*Unknown	2(0.12)

*Unknown includes cases where variable was not reported or ambiguously documented

Detection of baloxavir marboxil-related signals

Signal identification based on SOCs

Baloxavir marboxil-associated ADEs involved 22 SOCs, as illustrated in Fig. 2. The most frequent SOCs were injury, poisoning, and procedural complications (n = 846; ROR = 2.49, PRR = 2.11, IC = 1.08, EBGM = 2.11), general disorders and administration site conditions (n = 732; ROR = 1.26, PRR = 1.20, IC = 0.27, EBGM = 1.20), gastrointestinal disorders (n = 373; ROR = 1.41, PRR = 1.36, IC = 0.45, EBGM = 1.36), nervous system disorders (n = 245; ROR = 0.95, PRR = 0.95, IC=-0.07, EBGM = 0.95), and infections and infestations (n = 237; ROR = 1.26, PRR = 1.24, IC = 0.31, EBGM = 1.24). While some SOCs aligned with labeled adverse reactions, others revealed unlabeled risks, such as metabolic and nutritional disorders (n = 55; ROR = 0.82, PRR = 0.82, IC=-0.28, EBGM = 0.82), musculoskeletal and connective tissue disorders (n = 52; ROR = 0.28, PRR = 0.29, IC=-1.76, EBGM = 0.29), renal and urinary disorders (n = 49; ROR = 0.67, PRR = 0.67, IC=-0.57, EBGM = 0.67), and cardiac disorders (n = 38; ROR = 0.54, PRR = 0.55, IC=-0.87, EBGM = 0.55). Details are outlined in Table 2.

Fig. 2.

The cases of ADEs of baloxavir marboxil at the SOC level in FAERS database

Table 2.

The signal strength of ADEs of baloxavir marboxil at the SOC level in FAERS database

SOC	Case reports	ROR (95% CI)	PRR (95% CI)	chisq	IC (IC025)	EBGM (EBGM05)
Injury, poisoning and procedural complications	846	2.49(2.3, 2.69)	2.11(1.99, 2.24)	560.4	1.08(0.97)	2.11(1.97)
General disorders and administration site conditions	732	1.26(1.16, 1.37)	1.2(1.13, 1.27)	30.79	0.27(0.15)	1.2(1.12)
Gastrointestinal disorders	373	1.41(1.27, 1.57)	1.36(1.23, 1.5)	39.53	0.45(0.29)	1.36(1.25)
Nervous system disorders	245	0.95(0.83, 1.08)	0.95(0.84, 1.07)	0.62	-0.07(-0.26)	0.95(0.85)
Infections and infestations	237	1.26(1.1, 1.44)	1.24(1.1, 1.39)	11.88	0.31(0.12)	1.24(1.11)
Skin and subcutaneous tissue disorders	160	0.78(0.67, 0.92)	0.79(0.68, 0.92)	9.37	-0.34(-0.57)	0.79(0.69)
Respiratory, thoracic and mediastinal disorders	134	0.85(0.71, 1.01)	0.86(0.74, 1.01)	3.44	-0.23(-0.47)	0.86(0.74)
Psychiatric disorders	102	0.53(0.44, 0.65)	0.55(0.45, 0.67)	40.84	-0.87(-1.16)	0.55(0.46)
Investigations	75	0.36(0.28, 0.45)	0.37(0.3, 0.46)	84.45	-1.43(-1.75)	0.37(0.31)
Immune system disorders	73	1.73(1.37, 2.18)	1.71(1.35, 2.16)	21.82	0.77(0.44)	1.71(1.41)
Metabolism and nutrition disorders	55	0.82(0.63, 1.07)	0.82(0.64, 1.06)	2.18	-0.28(-0.66)	0.82(0.66)
Musculoskeletal and connective tissue disorders	52	0.28(0.22, 0.37)	0.29(0.22, 0.38)	92.96	-1.76(-2.16)	0.29(0.23)
Renal and urinary disorders	49	0.67(0.5, 0.89)	0.67(0.51, 0.88)	7.9	-0.57(-0.97)	0.67(0.53)
Cardiac disorders	38	0.54(0.39, 0.75)	0.55(0.4, 0.75)	14.56	-0.87(-1.33)	0.55(0.42)
Eye disorders	33	0.5(0.35, 0.7)	0.5(0.36, 0.7)	16.77	-1(-1.49)	0.5(0.38)
Hepatobiliary disorders	31	1.09(0.76, 1.55)	1.09(0.77, 1.55)	0.22	0.12(-0.38)	1.09(0.81)
Vascular disorders	31	0.47(0.33, 0.67)	0.48(0.34, 0.68)	18.28	-1.07(-1.57)	0.48(0.35)
blood and lymphatic system disorders	22	0.37(0.24, 0.57)	0.38(0.25, 0.57)	23.16	-1.41(-2)	0.38(0.26)
Ear and labyrinth disorders	12	0.84(0.48, 1.48)	0.84(0.48, 1.48)	0.37	-0.25(-1.04)	0.84(0.52)
Reproductive system and breast disorders	10	0.45(0.24, 0.84)	0.46(0.25, 0.86)	6.56	-1.13(-1.99)	0.46(0.27)
Neoplasms benign, malignant and unspecified (incl cysts and polyps)	8	0.06(0.03, 0.12)	0.06(0.03, 0.12)	123.3	-4.06(-5.01)	0.06(0.03)
Pregnancy, puerperium and perinatal conditions	7	0.54(0.26, 1.14)	0.54(0.26, 1.14)	2.68	-0.88(-1.88)	0.54(0.29)

Signal detection based on PTs

To improve specificity and reduce misclassification, we organized 47 PTs that met the criteria of four algorithms based on signal strength (IC025 value) in descending order. Some PTs demonstrated notably high signal strength, with delirium febrile showing the highest IC025 signal strength, along with other unexpected adverse reactions such as ischemic colitis, rhabdomyolysis, disseminated intravascular coagulation, and cardio-respiratory arrest.

As detailed in Fig. 3, the top 10 PTs by reported frequency were off-label use (n = 378; ROR = 7.05, PRR = 6.36, IC = 2.67, EBGM = 6.36), intentional product use issue (n = 278; ROR = 35.88, PRR = 32.97, IC = 5.04, EBGM = 32.81), pneumonia (n = 90; ROR = 5.28, PRR = 5.16, IC = 2.37, EBGM = 5.16), vomiting (n = 77; ROR = 3.41, PRR = 3.35, IC = 1.74, EBGM = 3.35), loss of consciousness (n = 36; ROR = 6.16, PRR = 6.11, IC = 2.61, EBGM = 6.10), anaphylactic reaction (n = 35; ROR = 12.09, PRR = 11.97, IC = 3.58, EBGM = 11.95), urticaria (n = 35; ROR = 4.07, PRR = 4.04, IC = 2.01, EBGM = 4.04), rhabdomyolysis (n = 25; ROR = 14.05, PRR = 13.95, IC = 3.80, EBGM = 13.92), seizures (n = 25; ROR = 3.16, PRR = 3.14, IC = 1.65, EBGM = 3.14), and abnormal behavior (n = 23; ROR = 18.13, PRR = 18.01, IC = 4.17, EBGM = 17.96). Details are provided in Table 3.

Fig. 3.

The cases of ADEs of baloxavir marboxil at the PT level in FAERS database

Table 3.

The signal strength of ADEs of baloxavir marboxil at the PT level in FAERS database

SOC	PT	Case reports	ROR (95% CI)	PRR (95% CI)	chisq	IC (IC025)	EBGM (EBGM05)
Psychiatric disorders	Delirium febrile	6	1564.02(641.2, 3814.99)	1561.2(646.27, 3771.43)	7544.55	10.3(9.12)	1259.23(597.14)
Nervous system disorders	Febrile convulsion	4	91.09(33.93, 244.51)	90.98(34.15, 242.41)	351.06	6.49(5.21)	89.74(39.28)
Gastrointestinal disorders	Colitis ischaemic	16	53.57(32.72, 87.73)	53.32(32.67, 87.04)	814.81	5.73(5.03)	52.89(35.01)
Injury, poisoning and procedural complications	Intentional product use issue	278	35.88(31.73, 40.58)	32.97(29.31, 37.08)	8595.98	5.04(4.86)	32.81(29.6)
Skin and subcutaneous tissue disorders	Erythema multiforme	14	35.61(21.04, 60.28)	35.46(20.89, 60.2)	466.4	5.14(4.41)	35.28(22.71)
Injury, poisoning and procedural complications	Product administered to patient of inappropriate age	19	32.58(20.73, 51.2)	32.4(20.64, 50.85)	575.37	5.01(4.38)	32.24(22.09)
Infections and infestations	Pneumonia bacterial	14	24.11(14.25, 40.79)	24.01(14.14, 40.76)	307.64	4.58(3.85)	23.93(15.41)
Gastrointestinal disorders	Melaena	22	19.97(13.12, 30.38)	19.84(13.15, 29.94)	392.54	4.31(3.71)	19.78(13.92)
Respiratory, thoracic and mediastinal disorders	Lower respiratory tract congestion	4	30.08(11.26, 80.38)	30.05(11.28, 80.07)	111.81	4.9(3.63)	29.91(13.14)
Psychiatric disorders	Abnormal behaviour	23	18.13(12.02, 27.34)	18.01(11.93, 27.18)	368.65	4.17(3.59)	17.96(12.74)
Immune system disorders	Anaphylactic shock	19	14.95(9.52, 23.48)	14.87(9.47, 23.34)	245.3	3.89(3.26)	14.84(10.17)
musculoskeletal and Connective tissue disorders	Rhabdomyolysis	25	14.05(9.47, 20.83)	13.95(9.43, 20.65)	299.99	3.8(3.24)	13.92(10.01)
Nervous system disorders	Altered state of consciousness	17	15.18(9.42, 24.45)	15.1(9.43, 24.17)	223.44	3.91(3.24)	15.07(10.11)
Immune system disorders	Anaphylactic reaction	35	12.09(8.66, 16.87)	11.97(8.58, 16.7)	351.64	3.58(3.1)	11.95(9.04)
Renal and urinary disorders	Cystitis haemorrhagic	4	19.86(7.44, 53.01)	19.83(7.44, 52.84)	71.32	4.31(3.04)	19.78(8.69)
Gastrointestinal disorders	Enterocolitis	5	14.8(6.15, 35.61)	14.78(6.12, 35.7)	64.09	3.88(2.73)	14.75(7.07)
Blood and lymphatic system disorders	Disseminated intravascular coagulation	7	13(6.19, 27.31)	12.98(6.16, 27.34)	77.23	3.7(2.69)	12.95(6.96)
Gastrointestinal disorders	Ileus paralytic	3	17.19(5.53, 53.42)	17.18(5.51, 53.55)	45.59	4.1(2.68)	17.14(6.64)
Blood and lymphatic system disorders	Haemorrhagic diathesis	3	16.45(5.3, 51.12)	16.44(5.27, 51.24)	43.4	4.04(2.62)	16.4(6.35)
Injury, poisoning and procedural complications	Off label use	378	7.05(6.33, 7.84)	6.36(5.77, 7.01)	1737.13	2.67(2.51)	6.36(5.81)
Injury, poisoning and procedural complications	Prescribed underdose	14	9.21(5.45, 15.57)	9.17(5.4, 15.57)	101.87	3.2(2.46)	9.16(5.9)
General disorders and administration site conditions	Face oedema	8	10.64(5.31, 21.3)	10.61(5.34, 21.07)	69.57	3.41(2.46)	10.6(5.93)
skin and subcutaneous tissue disorders	Drug eruption	9	9.83(5.11, 18.91)	9.8(5.13, 18.71)	71.05	3.29(2.39)	9.79(5.66)
Nervous system disorders	Loss of consciousness	36	6.16(4.44, 8.56)	6.11(4.38, 8.53)	153.92	2.61(2.14)	6.1(4.64)
Infections and infestations	Pneumonia	90	5.28(4.28, 6.51)	5.16(4.24, 6.28)	303.25	2.37(2.07)	5.16(4.33)
Injury, poisoning and procedural complications	Medication error	13	7.07(4.1, 12.19)	7.04(4.07, 12.19)	67.38	2.82(2.06)	7.04(4.46)
Infections and infestations	Bronchitis	22	6.05(3.97, 9.2)	6.01(3.98, 9.07)	91.95	2.59(1.99)	6.01(4.23)
Hepatobiliary disorders	Hepatic function abnormal	13	6.7(3.89, 11.56)	6.68(3.86, 11.56)	62.73	2.74(1.98)	6.67(4.23)
Psychiatric disorders	Delirium	12	6.82(3.87, 12.02)	6.8(3.85, 12)	59.3	2.76(1.98)	6.79(4.22)
Investigations	International normalised ratio increased	6	7.9(3.55, 17.61)	7.89(3.53, 17.62)	36.05	2.98(1.91)	7.88(4.03)
Vascular disorders	Cyanosis	5	7.71(3.21, 18.55)	7.7(3.19, 18.6)	29.13	2.94(1.79)	7.69(3.69)
Nervous system disorders	Encephalopathy	8	6.14(3.07, 12.29)	6.13(3.09, 12.17)	34.29	2.61(1.67)	6.12(3.42)
Skin and subcutaneous tissue disorders	Urticaria	35	4.07(2.92, 5.68)	4.04(2.9, 5.64)	80.22	2.01(1.54)	4.04(3.06)
eye disorders	Swelling of eyelid	4	6.96(2.61, 18.58)	6.96(2.61, 18.54)	20.39	2.8(1.53)	6.95(3.06)
Vascular disorders	Shock	6	5.73(2.57, 12.76)	5.72(2.56, 12.78)	23.34	2.51(1.44)	5.71(2.92)
Gastrointestinal disorders	Vomiting	77	3.41(2.72, 4.27)	3.35(2.7, 4.16)	127.91	1.74(1.42)	3.35(2.77)
Nervous system disorders	Depressed level of consciousness	10	4.74(2.55, 8.82)	4.73(2.53, 8.86)	29.39	2.24(1.39)	4.72(2.81)
Gastrointestinal disorders	Lip swelling	8	4.96(2.48, 9.93)	4.95(2.49, 9.83)	25.21	2.31(1.36)	4.95(2.77)
Nervous system disorders	Facial paralysis	4	6.16(2.31, 16.43)	6.15(2.31, 16.39)	17.25	2.62(1.35)	6.15(2.71)
Cardiac disorders	Cardio-respiratory arrest	7	4.31(2.05, 9.05)	4.3(2.04, 9.06)	17.75	2.11(1.1)	4.3(2.31)
Nervous system disorders	Seizure	25	3.16(2.13, 4.68)	3.14(2.12, 4.65)	36.54	1.65(1.09)	3.14(2.26)
Investigations	Blood creatine phosphokinase increased	5	4.54(1.89, 10.92)	4.54(1.88, 10.97)	13.77	2.18(1.02)	4.53(2.18)
Psychiatric disorders	Abnormal dreams	4	4.82(1.81, 12.86)	4.82(1.81, 12.84)	12.1	2.27(1)	4.82(2.12)
Psychiatric disorders	Hallucination	13	3.3(1.91, 5.69)	3.29(1.9, 5.7)	20.72	1.72(0.96)	3.29(2.08)

Discussion

Demographic characteristics of baloxavir marboxil-related ADEs

The data from this study indicates that among patients, there’s a slightly higher proportion of ADE reports in females compared to males after using baloxavir marboxil (41.9% vs. 36.46%). This may reflect both increased reporting tendencies (e.g., drug use, healthcare engagement, societal roles) and biological susceptibility, necessitating adjusted analyses to disentangle true risk from reporting bias [22]. In terms of age distribution, the proportion of reported ADE cases among individuals under 18 years old was significantly higher than in other age groups (25.70%), followed by the 18-45-year-old group (13.54%). Among children, influenza virus infection rates are highest, along with increased hospitalization needs and greater risks of secondary bacterial infections and flu-related complications. Severity correlates inversely with age [3, 23, 24]. The phase 3 miniSTONE-2 trial reported predominantly mild-to-moderate gastrointestinal reactions in pediatric patients, likely linked to influenza infection rather than the drug itself [25]. Subsequent subgroup analyses found no significant outcome differences between children aged 5–11 and miniSTONE-2 participants [26]. A Japanese meta-analysis comparing children under 6 and 6–12 years old reported similar overall ADE rates but higher infection incidence in the younger cohort, particularly under 2 years, likely due to immune immaturity and infection dynamics [27]. Furthermore, data reports highlight the United States and Japan as the top two countries in terms of reported cases, collectively representing 98.97% of the total reported cases. This prevalence may be attributed to the drug’s initial market availability in Japan and the United States, with research data potentially sourced from the U.S. FAERS. Moreover, there may be a correlation between the emphasis on drug safety and the priorities set by individual nations.

Detection of established ADEs associated with baloxavir marboxil

Signal detection from our study revealed a spectrum of ADEs categorized under various SOCs, many of which are already documented in the drug label. Noteworthy SOCs encompassed infections and infestations (e.g., bronchitis, pneumonia), psychiatric disorders (e.g., delirium, abnormal behaviour, hallucination), general disorders and administration site conditions (e.g., face oedema), immune system disorders (e.g., anaphylactic reaction, anaphylactic shock), gastrointestinal disorders (e.g., vomiting, melena, colitis), and skin and subcutaneous tissue disorders (e.g., urticaria, erythema multiforme, drug eruption).

Safety studies concerning baloxavir marboxil primarily rely on two randomized, double-blind, Phase III clinical trials: CAPSTONE-1 and CAPSTONE-2. The latter incorporates data from patients at high risk of developing influenza-associated complications. Data analysis indicates a lower occurrence of adverse reactions with baloxavir marboxil compared to the oseltamivir group. The most frequently reported adverse reactions include diarrhea, nausea, bronchitis, sinusitis, and headache [28, 29]. Research on over 3000 patients after the marketing of baloxavir marboxil indicates that 11.15% of patients experienced ADEs within 7 days of starting the medication. Nearly all ADEs were documented in the drug’s labeling and were of mild to moderate severity. Most patients recovered within 3 days [30]. The most common reactions were gastrointestinal symptoms and headache, with a prevalence similar to the findings of the CAPSTONE-1 trial [28]. The study also found that psychiatric disorders, such as delirium, febrile seizures, and abnormal behavior, had a higher incidence among patients under 12 years old compared to adolescents and adults. Five cases met the criteria for serious adverse reactions, yet all resolved within 3 days without long-term consequences. In contrast, other studies only reported cases of sleep disturbances in pediatric patients [31]. Pediatric populations exhibit heightened vulnerability to neuropsychiatric ADEs due to immature blood-brain barriers and immune systems, which may amplify drug-related neurological effects. Following its market release, there have been reports of drug rash, erythema multiforme, and urticaria. This highlights the importance of staying alert to allergic reactions during medication use. Timely management is essential if allergy-like symptoms occur.

In the category of infections and infestations, labeling pneumonia influenzal and influenza as ADEs is considered misreporting, where reporters mistakenly identify the diseases themselves as adverse reactions. In the classification of injury, poisoning and procedural complications, most reports are categorized as off-label use and intentional product use issue. Currently, there is no evidence demonstrating the effectiveness of baloxavir marboxil against pathogens other than influenza viruses, so.

prescriptions should be issued based on the results of pathogen-specific testing. Furthermore, there are reports of product administered to patient of inappropriate age, prescribed underdose and medication error. Pediatric research data shows that among patients under 5 years old, the frequency of developing drug-resistant viruses post-treatment is highest across all age groups. Therefore, some countries’ drug labels do not approve usage in children under 5 to mitigate the risk of potential viral mutations leading to public health concerns [25, 31–33].

Identification of novel ADEs signals associated with baloxavir marboxil

Bleeding events

This study detected bleeding-related signals (e.g., haemorrhagic diathesis, cystitis haemorrhagic, disseminated intravascular coagulation, international normalised ratio increased, and colitis ischaemic) potentially linked to baloxavir marboxil. These observations are supported by clinical case reports: A 62-year-old Japanese woman developed ischemic colitis with rectal bleeding post-treatment, likely due to baloxavir’s metal ion chelation impairing intestinal mucosal perfusion [34]. Another case involved a warfarin-treated patient with elevated PT-INR post-treatment, possibly due to protein-binding competition or flu-induced vitamin K deficiency, highlighting the need for anticoagulation monitoring [35].

However, causality remains debated due to confounding factors. Severe influenza can independently trigger endothelial injury, cytokine-driven hyperinflammation, and DIC, irrespective of antiviral therapy [36–38]. For instance, Okayama described hemorrhagic colitis in an influenza A patient without antiviral exposure, implicating viral pathology in bleeding manifestations [37]. A Japanese cohort study reported similar bleeding rates across treated and untreated influenza patients, with marginally higher risk in untreated groups, suggesting no inherent anticoagulant effect of antivirals [39]. Methodological limitations, including the FAERS database’s lack of detailed temporal information and reporting bias favoring severe outcomes (e.g., DIC), complicate signal interpretation. Consequently, the observed bleeding signals may arise from interplay between baloxavir’s pharmacological properties and influenza pathology. While causality remains unestablished, precautionary strategies are recommended, including: (1) intensified monitoring in high-risk cohorts (e.g., anticoagulant users) through frequent assessments; (2) clinical evaluations that systematically incorporate drug exposure timelines, influenza severity, and competing etiologies to mitigate attribution bias; (3) Future prospective studies are required to clarify the causal relationship between drug metabolism mechanisms and bleeding events.

Cardiovascular events

The current analysis identifies a potential association between baloxavir marboxil and severe cardiovascular events, including shock (n = 6), cyanosis (n = 5), and cardiac or respiratory arrest (n = 7), all linked to poor clinical outcomes and mortality risk. While influenza-related complications (e.g., hypoxia, metabolic acidosis, and septic shock) may indirectly contribute to cardiovascular collapse. Drug-specific risks are suggested by label warnings of anaphylaxis-related arrest. Drug-related allergic reactions accounted for 58.8% of fatal anaphylaxis cases [40]. Tanaka demonstrated that 4.5% of antiviral drug-associated allergic reactions resulted in severe outcomes, including a 2.5% mortality rate [41], implying drug-induced hypersensitivity exacerbates cardiovascular risks.

However, several limitations complicate causal inference. First, the absence of pre-existing cardiac comorbidity data (e.g., arrhythmias, heart failure) in FAERS obscures the distinction between drug effects and underlying disease progression. Second, systemic immune activation triggered by viral infection itself may independently induce anaphylaxis or shock, confounding the attribution of adverse events to the drug. Furthermore, no direct pathophysiological link between the drug and cardiac arrest has been established.

To resolve these uncertainties, future studies should focus on high-risk cohorts with pre-existing cardiovascular conditions, employing continuous ECG monitoring and serial troponin measurements to establish temporality. Concurrently, systematic comorbidity data collection and mechanistic studies integrating immunology and cardiovascular pathophysiology are essential to differentiate drug-specific effects from viral or synergistic mechanisms.

Liver dysfunction

This study identified 13 cases of hepatic dysfunction, consistent with observations from Phase I trials in which elevated ALT levels were reported in three healthy participants (one case was definitively linked to infectious mononucleosis rather than drug exposure). The drug undergoes hepatic metabolism predominantly via UGT1A3 [42], with a minor contribution from cytochrome P450 3A4 (CYP3A4)-mediated oxidation [43–45]. These pharmacokinetic properties, particularly its reliance on hepatic pathways, suggest a potential metabolic burden that may predispose to hepatotoxicity. However, current evidence remains inconclusive due to important missing information. First, FAERS pharmacovigilance data lack the resolution to distinguish drug-induced hepatotoxicity from influenza-related hepatic stress. Systemic inflammation during active infection can transiently elevate transaminases through viral hepatitis effects, while the absence of baseline liver function precludes differentiation between de novo injury and exacerbation of pre-existing conditions. Second, while baloxavir’s UGT1A3-dependent pharmacokinetic profile reduces risks of CYP450-mediated drug-drug interactions, this metabolic specificity renders conventional hepatotoxicity prediction models less applicable.

Rhabdomyolysis

Rhabdomyolysis, characterized by muscle tissue breakdown and elevated serum creatine kinase (CK) levels, arises from diverse etiologies including excessive exercise, medications, metabolic myopathies, electrolyte imbalances, toxins, and infections. While influenza is the most common viral trigger, acute myositis with concurrent rhabdomyolysis remains a rare complication of influenza infection [46–50]. Although no definitive causal relationship has been established between baloxavir marboxil and rhabdomyolysis, existing pharmacovigilance data highlight the need for vigilance regarding potential muscle injury during its use. Notably, pharmacological mechanisms or interactions with other myotoxic agents (e.g., statins, NSAIDs) may indirectly contribute to myocyte damage, necessitating dynamic CK monitoring to mitigate irreversible kidney injury. However, confounding factors complicate risk assessment: influenza itself can directly invade skeletal muscle, inducing CK elevation, while the absence of baseline CK data in adverse event reporting systems obscures pre-existing muscle injury. Thus, future studies must rigorously control for viral effects, concomitant medications, and baseline CK levels to clarify baloxavir marboxil’s safety profile.

Limitations

The analysis of baloxavir marboxil safety using the FAERS database is inherently constrained by systemic limitations that warrant careful interpretation. The self-reporting nature of FAERS introduces risks of incomplete, underreported, or selectively biased data, exemplified by missing gender (21.64%) and age (35.91%) information, alongside a substantial proportion of reports (34.95%) originating from non-healthcare professionals, which may skew data toward severe or subjectively perceived adverse events.

Geographic bias further limits generalizability, as 98.97% of reports originate from the U.S. and Japan, raising concerns about extrapolation to other populations due to regional prescribing practices, ethnic pharmacodynamic variability, or healthcare reporting disparities.

Crucially, confounding factors remain inadequately addressed: concomitant medications (e.g., anticoagulants, NSAIDs) commonly used in influenza management may independently contribute to observed adverse events (e.g., bleeding, hepatic dysfunction), yet FAERS lacks structured fields for dose, duration, or temporal relationships to rigorously evaluate drug-drug interactions. Simultaneously, influenza-induced systemic effects—such as coagulopathy, inflammation, and organ stress—overlap with adverse events attributed to baloxavir, complicating causal attribution. The absence of detailed clinical data (e.g., comorbidities, disease severity) and control groups precludes robust adjustment for these confounders, reducing findings to statistical associations rather than evidence of causation.

Furthermore, FAERS’ focus on spontaneous, acute reports inherently omits long-term safety assessments, leaving delayed or chronic risks uncharacterized.

While the study identifies potential safety signals, these limitations together highlight that observed correlations cannot quantify risk magnitude, establish causality, or require long-term controlled studies to distinguish the drug’s specific effects from other influencing factors. Given the relatively recent introduction of baloxavir marboxil to the market, prospective studies with rigorous designs, incorporating active surveillance, matched controls, and biomarker assessments, are imperative to validate these signals. Large-scale pharmacoepidemiological studies and international collaborations are also needed to address regional biases and enhance generalizability.

Conclusion

This pharmacovigilance study utilizing the FAERS database has identified critical safety signals associated with baloxavir marboxil, including hemorrhagic events, hepatic dysfunction, rhabdomyolysis, and life-threatening cardiorespiratory complications. These findings augment the existing safety profile of baloxavir marboxil by highlighting risks that may be underrecognized in clinical trials, particularly in pediatric populations and patients with comorbidities such as cardiovascular disease or those on anticoagulant therapy. The study underscores the importance of post-marketing surveillance in detecting rare or delayed adverse events and reinforces the need for clinicians to balance antiviral benefits against potential risks, especially in complex clinical scenarios where influenza-related complications and drug effects may overlap.

However, the observational nature of FAERS data imposes limitations, including an inability to control for confounding factors. While the detected signals suggest plausible drug-related risks, causality remains uncertain due to potential confounding by viral pathophysiology or reporting artifacts. Future prospective studies and pharmacoepidemiological analyses are warranted to validate these signals and establish causal relationships. Additionally, expanding data collection to underrepresented regions and vulnerable populations (e.g., immunocompromised individuals) will enhance generalizability. Collectively, these efforts will clarify the safety landscape of baloxavir marboxil, inform risk mitigation strategies, and optimize its role in influenza management amid evolving antiviral resistance and therapeutic needs.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Xiaolong Lai and Liuyin Jin: Methodology, Writing – original draft. Xiaolong Lai and Liuyin Jin: Formal analysis, Methodology, Writing – original draft.: Data curation, Writing – review & editing. Liuyin Jin: Data curation, Writing – review & editing. Yixia Zhou: Data curation, Writing – review & editing. Yang Li: Validation, Writing – review & editing. Lindan Sheng: Supervision, Writing – review & editing. Jianjiang Fang and Guomin Xie: Funding acquisition, Supervision, Writing – review & editing.

Funding

This study was supported by Zhejiang Health Science and Technology Project (2021KY1034). The study was financially supported by the grants from the Project of NINGBO Leading Medical & Health Discipline (Grant No.2022-F05).

Data availability

The dataset supporting the conclusions of this article is available via the public FDA Adverse Event Reporting System database found at https://www.fda.gov/drugs/questions-and-answers-fdas-adverse-event-reporting-system-faers/fda-adverse-event-reporting-system-faers-public-dashboard. The'FAERS Public Dashboard' option was then selected prompting to its home page. The'Search' tab was then selected, and additionally, our dataset comes from the public database, which can be found at https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html.

Declarations

Ethics approval and consent to participate

Not applicable. This study was deemed non-human subject related research.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.