Comparative Cardiovascular Safety Profiles of Atenolol, Bisoprolol, and Nebivolol: A Pharmacovigilance Disproportionality Analysis of the FDA Adverse Event Reporting System

1. Introduction

Cardiovascular disease (CVD) remains the leading cause of death worldwide. CVD encompasses a spectrum of conditions including coronary artery disease (CAD), stroke, hypertension, angina pectoris, myocardial infarction, and rheumatic heart disease [1]. Despite considerable therapeutic advances, the global burden has not abated: CVD accounts for approximately 17.5 million deaths annually, around 32% of all deaths worldwide [2]. The COVID-19 pandemic compounded this burden by disrupting healthcare delivery and worsening cardiovascular risk factors in vulnerable populations, with post-pandemic CVD mortality climbing to an estimated 20.5 million deaths in 2021 alone [3].

This burden has driven sustained efforts to develop, repurpose, and optimize cardiovascular treatments. Among currently available agents, beta-blockers have held a central place in cardiovascular medicine for decades, valued for their efficacy across a range of cardiac conditions [4]. Beyond their cardioprotective role, beta-blockers are also used in the management of anxiety, migraines, and certain tumors [5].

These agents act by competitively inhibiting catecholamine binding at β-adrenoceptors. Their primary cardiovascular effects (reduced heart rate, lower blood pressure and decreased myocardial workload) are mediated chiefly through β1-adrenoceptors in cardiac tissue, with β2-adrenoceptor antagonism in vascular and bronchial smooth muscle contributing to peripheral side effects in less selective agents [5,6].

Atenolol, bisoprolol, and nebivolol represent three successive pharmacological generations within this drug class and are among the most widely prescribed beta-blockers worldwide. Atenolol and bisoprolol are selective beta-1 adrenergic antagonists that reduce heart rate, blood pressure, and myocardial oxygen demand [7,8,9]. Nebivolol is a third-generation agent that combines beta-1 cardioselectivity with nitric oxide-mediated vasodilation [10]. Their labelled indications differ. In the United States atenolol is indicated for hypertension, angina pectoris and acute myocardial infarction; bisoprolol for hypertension; and nebivolol for hypertension only. The heart-failure indication for bisoprolol derives from European approval following CIBIS-II. Comparative real-world cardiovascular adverse event profiles remain incompletely characterized. These three agents were selected because they are among the most widely prescribed beta-blockers globally, have well-characterized pharmacological profiles, and carry sufficient FAERS reporting volumes to support stable disproportionality estimates, whereas other beta-blockers had considerably fewer reports. Known adverse effects include bradycardia, hypotension, heart block, bronchospasm, and fatigue, with severity influenced by patient comorbidities, dosing, and concomitant medications [11]. Despite widespread use of all three agents, systematic pharmacovigilance comparisons across generations in large spontaneous reporting datasets are absent from the literature.

This gap is especially relevant in patients with coexisting conditions such as asthma, diabetes, or peripheral vascular disease, in whom adverse event risk may be considerably higher.

This study compares the most frequently reported cardiovascular adverse events associated with atenolol (Tenormin), bisoprolol (Cardicor or Emcor), and nebivolol (Bystolic), using data from the FAERS database, building on prior FAERS-based disproportionality analyses in cardiovascular pharmacovigilance [12,13].

2. Methods:

2.1. Study Design and Data Source

This retrospective pharmacovigilance study queried the FDA Adverse Event Reporting System (FAERS) to evaluate cardiovascular adverse-event reporting patterns associated with three β₁-selective adrenergic receptor antagonists: atenolol, bisoprolol, and nebivolol. Individual case safety reports (ICSRs) were retrieved using OpenVigil 2.1, a validated pharmacovigilance data-mining platform [12], which provides structured access to FAERS data [13]. OpenVigil 2.1 processes cleaned FAERS data and supports complete-case extraction, drug-name mapping, duplicate checking, data filtering, and disproportionality analysis [12,14]. The extracted ICSRs contained information regarding the suspected drug, indication, reported adverse events, seriousness and clinical outcomes, reporter country, reporter type, patient sex, and patient age, when available [13,14]. The publicly available data did not contain directly identifying patient information. Reports with missing demographic information were retained for the primary disproportionality analysis but were categorized as unknown or missing in descriptive analyses. Atenolol, bisoprolol, and nebivolol received initial FDA approval on August 19, 1981 [15], July 31, 1992 [16], and December 17, 2007 [17], respectively. The analysis included the full post approval reporting period available in OpenVigil 2.1 for each drug through the database cutoff used for extraction. Because OpenVigil 2.1 contains FDA AERS/FAERS data beginning in the first quarter of 2004, the effective observation periods began in 2004 for atenolol and bisoprolol and on December 17, 2007, for nebivolol.

2.2. Procedure

Adverse events were identified using the Medical Dictionary for Regulatory Activities (MedDRA), a standardized international terminology used across the regulatory lifecycle from clinical trials to post-marketing surveillance. MedDRA is structured hierarchically, with Standardized MedDRA Queries (SMQs) grouping related adverse event terms. Initially, SMQs were queried for adverse events related to atenolol (Tenormin), followed by those for bisoprolol (Cardicor or Emcor), and nebivolol (Bystolic), spanning each drug’s available FAERS reporting history through January 1st, 2026 (see below). The primary focus of this study was to assess the cardiac safety profile, particularly in relation to cardiovascular adverse events (CV-AEs). The cardiac adverse events list data has been provided in the supplementary files.

2.3. Statistical Analysis

Signal detection used a case/non-case design, comparing CV-AE reports for each beta-blocker against all other drugs in the FAERS database. Disproportionality was quantified using the reporting odds ratio (ROR)

Calculation of reporting odds ratio (ROR) with test drug versus all other drugs in FAERS database.

	Test drug.	All other drugs in FAERS
Reports of AE of interest	A	B
All other events	C	D

ROR= (A×D)/(B×C), 95%CI＝e^{ln(ROR)±1.96√(1/A+1/B+1/C+1/D)}

The ROR was used to compare the likelihood of CV-AEs being reported with a specific drug (e.g., atenolol) relative to the same events reported for other drugs within the database. A higher ROR indicates a greater association between the drug and the adverse event. A positive signal for the ROR was considered according to Evans’ criteria, which required more than three cases, Chi-square values exceeding 4, a 95% confidence interval (CI) lower limit above 1.0, and an ROR value greater than 2.0. ROR > 1 means the adverse event is more likely to occur with the treatment. ROR < 1 means the adverse event is less likely to occur with the treatment. ROR = 1 suggests no association [15]. Reporting Odds Ratios were computed in SPSS (version 25.0); the Information Component and its 95% credibility interval were computed manually in Microsoft Excel from the observed and expected counts, following the BCPNN method of Bate et al., as SPSS does not provide a native Bayesian confidence propagation neural network (BCPNN) routine [14].

2.4. Information Component (IC) Analysis

Disproportionality was further evaluated using the Information Component (IC), a Bayesian signal detection method commonly applied in spontaneous reporting systems [14]. The IC compares the observed number of reports for a given drug–event pair with the number expected under the assumption of statistical independence, and was calculated as:

$$IC={\mathrm{l}\mathrm{o}\mathrm{g}}_2\left({\displaystyle \frac{N_{\mathrm{O}\mathrm{b}\mathrm{s}}+0.5}{N_{\mathrm{E}\mathrm{x}\mathrm{p}}+0.5}}\right)$$

where N_Obs denote the observed number of reports for the drug–event pair and N_Exp represents the expected number based on marginal totals. A shrinkage factor of 0.5 was applied to stabilize estimates for rare events. Statistical significance was assessed using the lower bound of the 95% Bayesian credibility interval (IC₀₂₅), with IC₀₂₅ > 0 considered indicative of a positive disproportionality signal.

3. Results

3.1. Atenolol

3.1.1. Cardiovascular Adverse Events Data Determined for Atenolol Using FAERS Database

Disproportionality analysis identified significant cardiovascular pharmacovigilance signals for atenolol. The strongest signal was bradycardia (ROR = 10.513; 95% CI: 10.021–11.028), followed by left ventricular hypertrophy (ROR = 6.765; 95% CI: 5.514–8.300), pericardial haemorrhage (ROR = 7.067), ventricular hypertrophy (ROR = 6.999), and complete atrioventricular block (ROR = 6.686). Thirty-eight atenolol cardiovascular drug–event pairs were evaluated in total (Table 1); case counts ranged from 15 to 2,097 unique reports, and the twelve events with the highest Information Component are shown in Figure 1.

3.1.2. Characteristics of Patients and Outcomes

Individual case safety reports for atenolol are summarized in Supplementary Table S1. Bradycardia was the most frequently reported atenolol-associated AE (n = 1,796), followed by myocardial infarction (n = 1,667) and atrial fibrillation (n = 1,251). Among those affected by bradycardia, gender distribution was nearly balanced, with females comprising 46.3% and males 45.5% of cases. Hospitalization was reported in 56.3% of bradycardia cases, with life-threatening outcomes in 5.2% and death in 6.4%. For left ventricular hypertrophy (n = 95), the second highest-ROR event, hospitalization was observed in 57.9% of cases, with death in 5.3% and life-threatening outcomes in 9.5%.

Reports originated predominantly from the United States (32.6%; combining the “UNITED STATES” and “US” reporter-country entries), followed by France (12.6%), Great Britain (9.8%), and Italy (9.1%), with other countries collectively accounting for the remainder. Adverse events were observed across all age groups, with the highest occurrence in patients aged 80–89 (21.4%), followed by those aged 70–79 (20.4%), 60–69 (15.0%), and 90+ (6.1%), confirming that adverse event reports are concentrated in the oldest age groups.

3.2. Bisoprolol

3.2.1. Cardiovascular Adverse Events Data Determined for Bisoprolol Using FAERS Database

Disproportionality analysis identified significant cardiovascular pharmacovigilance signals for bisoprolol. The strongest signals were diastolic dysfunction (ROR = 24.875), atrial tachycardia (ROR = 23.283), and chronic cardiac failure (ROR = 23.222). Thirty-six bisoprolol cardiovascular drug–event pairs were evaluated in total (Table 2); case counts ranged from 147 to 4,154 unique reports, and the twelve events with the highest Information Component are shown in Figure 2.

3.2.2. Characteristics of Patients and Outcomes

Individual case safety reports for bisoprolol are summarized in Supplementary Table S2. Hypotension was the most frequently reported bisoprolol-associated AE (n = 4,154), followed by atrial fibrillation (n = 2,856) and bradycardia (n = 2,764). Among hypotension cases, males comprised 49.0% and females 47.4% of reports. Hospitalization was observed in 70.1% of hypotension cases, with death in 8.5% and life-threatening outcomes in 20.3%. For bradycardia, hospitalization occurred in 50.6% of cases, death in 14.9%, and life-threatening outcomes in 10.7%, with bisoprolol identified as the primary suspect in 88.3% of bradycardia reports. Reports originated most frequently from the United States (40.4% for hypotension; 77.8% for bradycardia), France, Great Britain and Italy. Adverse events occurred across all age groups, with the highest occurrence in patients aged 60–69 (28.1%) and 70–79 (26.6%) for hypotension, and 60–69 (22.5%) and 70–79 (20.4%) for bradycardia.

3.3. Nebivolol

3.3.1. Cardiovascular Adverse Events Data Determined for Nebivolol using FAERS Database

Disproportionality analysis identified significant cardiovascular pharmacovigilance signals for nebivolol. The strongest signals were bradycardia (ROR = 9.619) and coronary artery stenosis (ROR = 9.351), followed by chronic cardiac failure (ROR = 6.924) and sinus bradycardia (ROR = 6.464). Twenty-five nebivolol cardiovascular drug–event pairs were evaluated in total (Table 3); case counts ranged from 25 to 655 unique reports, and the twelve events with the highest Information Component are shown in Figure 3.

3.3.2. Characteristics of Patients and Outcomes

Individual case safety reports for nebivolol are summarized in Supplementary Table S3. Hypertension was the most frequently reported nebivolol-associated AE (n = 655), followed by hypotension (n = 638) and bradycardia (n = 422). Among hypotension cases, females comprised 47.6% and males 43.7% of reports. Hospitalization was observed in 50.3% of hypotension cases and 50.5% of bradycardia cases, with death reported in 12.9% and 12.6%, respectively. Life-threatening outcomes were observed in 9.6% of hypotension cases and 9.7% of bradycardia cases. Nebivolol was identified as the primary suspect in 87.1% of hypotension reports and 88.2% of bradycardia reports. The majority of reports originated from the United States (71.5% for hypotension; 70.6% for bradycardia). Adverse events were reported across all age groups including pediatric patients (0–17 years: 1.4% for hypotension, 1.4% for bradycardia), with the highest occurrence in patients aged 60–69 (23.0% for hypotension; 24.4% for bradycardia) and 70–79 (21.2% for hypotension; 20.4% for bradycardia).

4. Discussion

Clinical trials establish the benefits of beta-blockers, but they enroll selected patients under controlled conditions. The patients who receive these drugs day to day are older, more comorbid and often on polypharmacy [1,2]. Their real-world adverse event experience across pharmacological generations remains poorly characterized [16]. Our analysis of FAERS reports for atenolol, bisoprolol, and nebivolol brings that evidence into sharper focus across 100 cardiovascular drug–event pairs (39 for atenolol, 36 for bisoprolol and 25 for nebivolol). Because a single report may be coded to more than one Preferred Term, case counts are not additive across events.

Atenolol showed the strongest association with bradycardia (ROR = 10.513; n = 1,796) and left ventricular hypertrophy (ROR = 6.765). The bradycardia signal is mechanistically expected, since atenolol competitively blocks sinoatrial node β1 receptors and slows heart rate in a dose-dependent manner [11]. The outcome burden is nonetheless notable: 56.3% of bradycardia reports were associated with hospitalization and 6.4% with a fatal outcome. The predominance of reports in patients aged 70–79 (20.4%) and 80–89 (21.4%) is consistent with established prescribing patterns and further amplifies risk, as older patients exhibit greater susceptibility to hemodynamic compromise from bradyarrhythmias [17]. The LVH signal also deserves separate comment. While antihypertensive therapy is expected to attenuate left ventricular hypertrophy (LVH) through blood pressure reduction, the landmark LIFE trial demonstrated that atenolol was significantly inferior to losartan in reversing LVH despite equivalent blood pressure lowering, suggesting that atenolol does not confer comparable end-organ protection via non-hemodynamic pathways [18]. The remaining signal cluster covers atrioventricular block complete (ROR = 6.686; n = 156), pericardial haemorrhage (ROR = 7.067; n = 54) and ventricular hypertrophy (ROR = 6.999; n = 76). Together these show that atenolol’s pharmacovigilance risk spans both structural and electrophysiological domains, which should inform monitoring decisions in elderly or high-risk patients. These estimates rest on one to two orders of magnitude fewer reports than the bradycardia signal (n = 1,796) and are correspondingly less precise, as the wider credibility intervals in Figure 1 show.

For bisoprolol, the strongest disproportionality signals were observed for diastolic dysfunction (ROR = 24.875; n = 285), atrial tachycardia (ROR = 23.283; n = 158), and chronic cardiac failure (ROR = 23.222; n = 353). Notably, these three highest-ROR signals are not the ones supported by the most reports: bisoprolol’s most frequently reported events were hypotension (n = 4,154) and bradycardia (n = 2,764), both with substantially lower disproportionality (ROR = 5.617 and 13.957 respectively). Signal magnitude and volume of supporting evidence therefore need to be read together rather than interchangeably. The chronic cardiac failure signal needs to be read in context. Bisoprolol is guideline-recommended for heart failure with reduced ejection fraction (HFrEF), as established by the landmark CIBIS-II trial [19], and has also demonstrated mortality reduction in high-risk perioperative patients [20]. The pharmacovigilance signal likely reflects channeling bias: patients with pre-existing or progressive heart failure are disproportionately prescribed bisoprolol, and decompensation events are therefore more likely to be co-reported with this drug [21]. The diastolic dysfunction signal (ROR = 24.875) is mechanistically plausible: bisoprolol’s potent negative chronotropic effect may impair diastolic filling time by excessively slowing heart rate, particularly in patients with heart failure with preserved ejection fraction (HFpEF) or pre-existing diastolic impairment [22]. Among bisoprolol-associated adverse events, hypotension (n = 4,154) showed the highest hospitalization rate in the entire dataset at 70.1%, with death in 8.5% and life-threatening outcomes in 20.3%. Bisoprolol-associated bradycardia (n = 2,764) carried the highest mortality rate across all bradycardia cases in our analysis at 14.9%, compared to 12.6% for nebivolol and 6.4% for atenolol, a difference that may partly reflect bisoprolol’s greater degree of β1-selectivity and potency relative to the other agents studied.

Of the three agents, nebivolol had the lowest overall FAERS reporting burden. Bradycardia (ROR = 9.619; n = 422) and coronary artery stenosis (ROR = 9.351) were its strongest signals, but the absolute case counts were a fraction of those for atenolol (n = 1,796) and bisoprolol (n = 2,764). This pattern fits with nebivolol’s third-generation pharmacology: β1-selectivity combined with nitric oxide-mediated vasodilation attenuates the peripheral vasoconstriction and hemodynamic instability typical of earlier agents [10]. The coronary artery stenosis signal rests on only 29 reports, and its wide confidence interval (95% CI: 6.482–13.490) reflects that sparse evidence base; it most plausibly reflects channeling toward patients with established coronary disease on guideline-directed therapy rather than a drug-attributable effect [23]. Moderately associated events, including sinus bradycardia (ROR = 6.464; n = 51), angina pectoris (ROR = 5.805; n = 129) and ischaemic stroke (ROR = 6.443; n = 98), reflect the cardiovascular comorbidity burden typical of patients receiving nebivolol. Hypotension- and bradycardia-associated mortality (12.9% and 12.6%, respectively) were lower than bisoprolol’s bradycardia mortality of 14.9%, consistent with nebivolol’s more attenuated hemodynamic impact. Bradycardia and hypotension were also reported in pediatric patients (0–17 years; 1.4% of cases for each event). This proportion was closely comparable across all three agents (bradycardia reports aged 0–17: atenolol 1.45%, bisoprolol 1.48%, nebivolol 1.42%), and none of the three carries FDA approval in this age group. The finding is therefore best interpreted as a class-wide off-label reporting pattern rather than a nebivolol-specific signal. Interest in nebivolol’s cardioprotective potential beyond hypertension is also growing, including emerging applications in oncology [24].

Pregnancy and lactation deserve separate consideration, as neither exposure group was captured in our FAERS stratification. Available evidence from systematic reviews indicates that beta-blockers cross the placenta and can cause neonatal bradycardia, hypoglycemia, and respiratory depression; maternal use has also been associated with intrauterine growth restriction in some cohorts [25,26]. Atenolol in particular has been linked to significant intrauterine growth restriction and is generally avoided in pregnancy; labetalol is the preferred first-line beta-blocker in pregnancy, with bisoprolol or metoprolol as cardioselective alternatives when labetalol is not suitable, given comparatively lower rates of fetal growth restriction than atenolol [26]. In the context of lactation, beta-blockers are excreted in breast milk to varying degrees depending on protein binding and renal elimination; atenolol accumulates more substantially than labetalol or propranolol and has been associated with neonatal bradycardia and cyanosis in breastfed infants, highlighting the need for agent-specific guidance [27]. Our FAERS analysis could not stratify reports by pregnancy or lactation status; studies specifically designed around these populations are needed to fill this gap.

Looking across all three drugs, a consistent demographic picture emerges: adverse event burden was concentrated in patients aged 60–89, as expected given the prescribing context of these agents in hypertension, coronary disease, and heart failure. Age-related declines in baroreceptor sensitivity, autonomic reserve, and renal drug clearance each amplify beta-blocker risk in this group [17], and clinicians should calibrate dosing and monitoring accordingly, particularly in patients with renal impairment or diabetes, where hemodynamic instability can escalate rapidly. The high primary suspect attribution rates for bradycardia (atenolol 45.9%; bisoprolol 88.3%; nebivolol 88.2%) are consistent with a drug-related contribution. Primary-suspect designation, however, reflects the reporting clinician’s judgement rather than an assessment of causality, and the atenolol figure (45.9%) is appreciably lower than for the other two agents. Reports came from multiple countries across North America and Europe, suggesting broad geographic applicability. Nebivolol’s substantially lower reporting burden across all metrics supports a comparatively favorable pharmacovigilance profile [23].

Among the three agents, nebivolol showed the most favorable pharmacovigilance signal profile, consistent with its dual β1-selective and nitric oxide-mediated vasodilatory mechanism [10,22]. Bradycardia- and hypotension-associated mortality were lower than for bisoprolol, and the overall reporting volume was substantially smaller. That smaller volume cuts both ways: nebivolol’s most frequently reported event carried 655 cases, against 1,796 for atenolol and 4,154 for bisoprolol, so its disproportionality estimates are the least precise of the three and a genuinely lower reporting burden cannot be separated from lower reporting volume on these data alone. These observations may inform drug selection in older patients with diastolic dysfunction, HFpEF, or high bradyarrhythmia risk, but they should not be taken as proof of clinical superiority. They are hypothesis-generating signals from a voluntary reporting system, and prospective comparative studies are needed before they should drive practice changes.

4.1. Limitations

Several limitations inherent to spontaneous reporting database analyses must be acknowledged. First, FAERS data are subject to significant underreporting, as adverse events are submitted voluntarily by healthcare providers, patients, and manufacturers, and the true incidence of adverse events in the population cannot be derived from report counts alone. The ROR and IC are measures of disproportionate reporting, not measures of absolute risk or causality. Second, confounding by indication is an important consideration throughout this analysis: drugs are prescribed to patients with pre-existing cardiovascular disease, and many of the reported adverse events may reflect the underlying condition rather than a direct drug effect. This is particularly relevant for bisoprolol, which is guideline-recommended for heart failure, making signals such as chronic cardiac failure and diastolic dysfunction highly susceptible to channeling bias. Third, the analysis could not account for concomitant medications, polypharmacy, dose, or treatment duration, all of which may independently contribute to adverse event risk. Fourth, differences in market penetration, prescribing volume, and duration of clinical availability across the three agents influence both absolute report counts and the stability of disproportionality estimates, limiting direct cross-drug comparisons of reporting frequency and signal magnitude. Fifth, FAERS reports may contain duplicate entries despite OpenVigil 2.1’s deduplication procedures, and data quality is inherently dependent on the accuracy of voluntary submissions. Sixth, we did not have access to other large spontaneous reporting systems (e.g., VigiBase) or longitudinal cohort datasets for external validation, and our findings should therefore be regarded as hypothesis-generating pharmacovigilance signals requiring confirmation in independent data sources. Seventh, FAERS does not reliably capture patient ethnicity, precluding ethnicity-specific signal analyses; future studies using datasets with robust ethnicity data are needed to explore potential ethnic differences in beta-blocker safety profiles. Eighth, adverse event reports were not stratified by pregnancy or lactation status; beta-blocker use in pregnant or breastfeeding patients warrants dedicated pharmacoepidemiological investigation, as our FAERS analysis cannot characterize risks in these subpopulations. These limitations notwithstanding, the use of dual signal detection methods (ROR and IC), adherence to Evans’ criteria, and the large sample size accrued over each drug’s full FAERS reporting history strengthen the validity of the pharmacovigilance signals identified.

5. Conclusion

This study identified significant cardiovascular pharmacovigilance signals for all three beta-blockers analyzed, with meaningful differences between agents in signal magnitude, severity, and associated mortality. Bisoprolol demonstrated the highest disproportionality signals overall, particularly for diastolic dysfunction and chronic cardiac failure; these signals are best interpreted in the context of channeling bias, given bisoprolol’s guideline-endorsed role in heart failure management. Atenolol carried substantial bradycardia burden with notable life-threatening outcomes in elderly patients, consistent with its pharmacological profile as a non-vasodilatory beta-1 blocker. Nebivolol showed the most favorable pharmacovigilance profile, with the lowest overall reporting volume and lowest bradycardia-associated mortality. These findings can inform drug selection in older patients with diastolic dysfunction or elevated arrhythmia risk, though they cannot establish causality and should be confirmed in prospective comparative studies. The detection of nebivolol-associated adverse events in pediatric patients, a population for whom the drug carries no FDA approval, is a notable off-label signal that regulatory bodies should investigate. The signal differences identified here can guide more targeted drug selection; all three agents require close monitoring for bradycardia and hemodynamic instability in elderly patients. These results show what large-scale pharmacovigilance adds to clinical trial evidence: a real-world picture of who experiences adverse events, how serious they are, and whether one agent carries a lower burden than another. Controlled trials are rarely designed or powered to answer these questions directly.