Cardiovascular Safety of COVID-19 Treatments: A Disproportionality Analysis of Adverse Event Reports from the WHO VigiBase

Abstract

Introduction

The high mortality of Coronavirus Disease 2019 (COVID-19) highlights the need for safe and effective antiviral treatment. Small molecular antivirals (remdesivir, molnupiravir, nirmatrelvir/ritonavir) and immunomodulators (baricitinib, tocilizumab) have been developed or repurposed to suppress viral replication and ameliorate cytokine storms, respectively. Despite U.S. Food and Drug Administration (FDA) approval, serious cardiovascular adverse events (CVAEs) may not be apparent in initial trials.

Methods

A retrospective analysis of CVAEs linked to five World Health Organization (WHO) recommended COVID-19 therapies was conducted using the WHO VigiBase database from March 2020 to July 2023. Adjusted reporting odds ratios (aROR) with 95% confidence intervals (CI) were calculated to assess CVAE risks.

Results

A total of 276,631 AEs were reported to be associated with COVID-19, of which 13,091 were classified as cardiovascular events. Remdesivir was associated with significantly increased odds of CVAEs, particularly bradycardia (aROR 2.4, 95% CI 2.28–2.52). In contrast, nirmatrelvir/ritonavir and molnupiravir showed reduced CVAEs odds. Among immunomodulators, baricitinib was associated with increased CVAEs risk (aROR 2.31, 95% CI 2.07–2.59), with deep vein thrombosis being the most prominent (aROR 45.34, 95% CI 34.89–58.9), accounting for 38.8% of reported study cases in the database. Also, CVAEs odds were higher during the Omicron period compared to pre-Omicron period.

Conclusions

These findings highlight the importance of continued pharmacovigilance and suggest potential CV safety differences among COVID-19 immunomodulators. Since tocilizumab and baricitinib are similarly indicated for severe patients with COVID-19, further clinical trials are warranted to explore whether tocilizumab represents a safer alternative to baricitinib for these patients. Insights from this study may guide future antiviral repurposing and pandemic preparedness strategies.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40121-025-01225-z.

Key Summary Points

Why carry out this study?

COVID-19’s high mortality required rapid drug approval, but rare cardiovascular (CV) adverse effects (AEs) may not appear in initial trials due to small samples and short durations.

Post-marketing surveillance is critical for detecting CV safety signals of COVID-19 therapies, especially as these drugs may be repurposed for future pandemics.

This study examined whether five World Health Organization (WHO)-recommended COVID-19 therapies are associated with increased CVAE reporting in the WHO VigiBase database.

What was learned from the study?

Remdesivir showed significantly increased CV risk, while nirmatrelvir/ritonavir and molnupiravir were associated with reduced CVAE odds. Baricitinib demonstrated the highest risk of CVAEs with dramatically elevated odds of deep vein thrombosis (adjusted reporting odds ratio (aROR): 45.34).

Larger clinical trials are needed to test if tocilizumab is a safer alternative to baricitinib for patients with severe COVID-19, with insights potentially guiding future antiviral repurposing and pandemic preparedness.

Introduction

The unprecedented death toll from COVID-19 underscores the critical need for safe and effective antiviral treatments to address viral pandemics. Viral diseases such as COVID-19 often consist of an initial infectious phase, followed, in severe patients, by an inflammatory phase [1]. Many drugs have been developed to target the two phases of disease. While efficacy has been well investigated, the safety profiles of antiviral treatment are often less studied, especially for rare but severe complications such as cardiovascular (CV) injury.

During the early stages of viral infection, small molecular antiviral therapeutics are often employed to control viral replication [1]. Examples include nucleoside/nucleotide analogues, which are designed to inhibit virally encoded RNA-dependent RNA polymerase (RdRP) and protease inhibitors, which interfere with the intracellular processing of virally encoded proteins [2]. Remdesivir and molnupiravir are nucleotide analogues originally designed to treat Ebola and influenza, respectively. Both exhibit broad-spectrum antiviral activity [3, 4] and were repurposed for the treatment of COVID-19. To inhibit viral protease, Pfizer developed the oral antiviral combination Nirmatrelvir/ritonavir, which consists of nirmatrelvir, an inhibitor of the SARS-CoV-2 main protease (M^pro), and ritonavir, which slows down the breakdown of nirmatrelvir [1]. All three drugs have shown efficacy against COVID-19 in clinical trials [1, 4]. Remdesivir is currently the only drug approved by the FDA for the treatment of both hospitalised and non-hospitalised patients with COVID-19. Nirmatrelvir/ritonavir and molnupiravir received full and emergency use approval, respectively, as outpatient oral treatment against COVID-19. For severely affected patients with COVID-19, the infectious phase is followed by a hyperinflammatory phase, when the immune system becomes dysregulated to trigger multi-organ damage [1, 5]. To combat this, immune modulators have been repurposed for use against COVID-19. Baricitinib was originally designed against arthritis and inhibits the Janus kinase (JAK) signaling pathway important for cellular inflammation. It also became the first US FDA-approved immunomodulatory treatment for COVID-19 [6]. Tocilizumab is a recombinant humanized anti-interleukin 6 receptor monoclonal antibody that was originally approved to treat patients with rheumatologic disorders and cytokine release syndrome induced by chimeric antigen receptor T-cell therapy [7]. Both are intended to suppress inflammatory signaling and are similarly indicated for patients with severe COVID-19 who require aggressive oxygen support.

Given that multiple drugs are approved for the treatment of patients with COVID-19, the risk of adverse effects is a critical determinant of clinical use. Antivirals, particularly nucleotide analogues, have been shown to induce mitochondrial toxicity, which targets metabolically demanding tissues such as the heart [8]. Importantly, COVID-19 itself can damage the heart [9]. Treatment-induced cardiotoxicity may therefore synergistically worsen cardiac function. Protease inhibitors, although commonly considered to be less cardiotoxic, are known to induce arrhythmias. CVAEs have also been reported for immune modulators such as venous thromboembolism and cardiac arrhythmias [10, 11]. Although these treatments were found to have acceptable safety profiles to warrant FDA approval, post-market monitoring might reveal rare yet serious CVAEs that might not be apparent in initial clinical trials. Furthermore, multiple COVID-19 variants have emerged with different pathogenicity and exert different effects on the heart. The interaction between COVID-19 variants and CV risks of treatment has not been examined. Individual case safety report (ICSR) databases like VigiBase are critical for detecting rare adverse events post-approval due to their global coverage and spontaneous reporting system.

In this work, we conducted a retrospective disproportionality pharmacovigilance cohort study using data from VigiBase, the World Health Organisation (WHO) global database of ICSRs to define the cardiac safety of antiviral and immunomodulatory treatment against COVID-19. We report that of the three antiviral drugs, remdesivir is associated with increased odds of CVAEs, while nirmatrelvir/ritonavir and molnupiravir are associated with decreased odds of the latter. As for the immune modulators, we observed increased odds of CVAEs associated with the use of baricitinib, and substantially higher risk of pulmonary embolism and deep vein thrombosis, while CVAEs are minimally over-represented among the adverse reports associated with tocilizumab.

Methods

Study Design and Data Source

Our study utilized data from VigiBase, the WHO global database of individual case safety report. VigiBase contains over 35 million adverse drug reaction (ADR) reports from approximately 150 countries, collected since 1968 as part of the WHO Program for International Drug Monitoring, and is managed by the Uppsala Monitoring Center (UMC) in Sweden. Drugs are coded according to WHODrug, while ADRs are classified according to MEdDRA terminology (version 26.0). It is structured hierarchically and comprises five levels: system organ class (SOC), high-level group term, high-level term, preferred term (PT) and lower-level term (LLT) [12]. The information in VigiBase is sourced primarily from post-marketing spontaneous cases reported by physicians, pharmacists, and patients, as well as from clinical trial reports and literature reports [13, 14].

The study population comprised adverse event reports in the VigiBase database related to COVID-19. Case reports with drug indications for COVID-19 were extracted from VigiBase for analysis, from the inception of the database to July 2023. Relevant cardiovascular ADR reports for COVID-19 treatments were extracted from the VigiBase database from March 2020 until July 2023. Reports were categorized into case and non-case groups based on the target drugs, which included Nirmatrelvir/ritonavir (Paxlovid^®), Molnupiravir (Lagevrio^®), Remdesivir (Veklury^®), Baricitinib (Olumiant^®), and Tocilizumab (Actemra^®/Roactemra^®). According to a predefined match between the PT and the primary SOC in MedDRA V26.0, we mapped the PTs to their corresponding SOCs for further analysis. As the study aims to focus on investigating CV adverse reports, we extracted all ICSRs reporting ADRs associated with these drugs, filtering for those classified under the SOC of "Cardiac disorders" and "Vascular disorders" according to the MedDRA terminology.

To assess the impact of SARS-CoV-2 variants, we also divided the reported ADRs into those associated with the Omicron variants and those related to pre-Omicron variants, based on the first reporting date (Nov 24, 2021) to the World Health Organization [15].

VigiBase occasionally includes duplicate reports. To address this, the Uppsala Monitoring Center uses statistical algorithms to identify and exclude these potential duplicates from the analysis [16].

Ethical Approval

This article is based on available data in VigiBase (https://who-umc.org/vigibase-search-services/about-vigibase/), and the data is anonymous and is available upon request from said database. This manuscript does not contain any new studies with human participants or animals performed by any of the authors.

The study was approved by the ethics committee of The Chinese University of Hong Kong (CUHK) and New Territories East Cluster (NTEC) (Ref. No. 2023.186). Since only anonymous data was collected, informed consent was waived.

Statistical Analysis

We conducted an analysis to determine whether the proportion of cardiovascular toxicities reported for the study drugs differed from that observed in the control group, which includes the entire extracted COVID-19 database. The top ten most frequently reported PTs for CV ADRs associated with the investigated COVID-19 treatment drugs were identified from the extracted ICSRs. A disproportionality analysis was performed by constructing a 2 × 2 contingency table, relating the observed drug-event combinations to all other drugs and events in the extracted database to calculate the reporting odds ratio (ROR) with a 95% confidence interval (CI) for each adverse event term. Adjusted (a) RORs with 95% CIs were derived using a logistic regression model that accounted for age group and sex.

Disproportionate reporting was further evaluated by calculating the information component (IC) values. The IC measures the disproportionality between the observed and expected number of reports for a drug-event combination.

$$IC={log}_2\frac{O+\alpha 1}{E+\alpha 2}$$

where O represents the number of reports for the drug-event combination, with $\alpha 1\text{and}\alpha 2=0.5$. E is the expected number of events, defined as:

$$E=\frac{\left(\text{no}.\text{of}\text{reports}\text{on}\text{the}\text{drug}\right)\times \left(\text{no}.\text{of}\text{reports}\text{on}\text{the}\text{event}\right)}{\text{total}\text{number}\text{of}\text{reports}}$$

A positive IC indicates that the number of reported cases exceeds the expected number.

Traditional significance thresholds were applied, requiring the lower limit of the 95% confidence interval to be greater than 1 for the ROR and greater than 0 for the IC. A safety signal was deemed present if both methods demonstrated statistical significance [17–19].

To assess the significance of the differences in reporting odds ratios between the pre-Omicron and Omicron periods for baricitinib, remdesivir, and tocilizumab, Z-tests was performed to compare the independent log-transformed odds ratios. Qualitative variables were expressed as counts and percentages. Data processing and statistical analyses were performed using Python and R (R Studio version 2024.04.1 for Statistical Computing).

Results

Overview of AEs

During the study period, a total of 326,655 adverse events related to COVID-19 were identified in VigiBase. After excluding cases with missing data on age, sex, and MedDRA ID, as well as reports prior to January 1, 2020, 276,631 case reports remained for analysis. Among these, 263,540 and 13,091 were categorized as non-CVAEs and CVAEs, respectively (Fig. 1). During the study period, most CVAEs were observed in patients aged 45–74 years (Table 1). Almost all cases were reported spontaneously and originated primarily from the Americas and Europe. Tachycardia was the most frequently reported type of CVAEs, accounting for 17.1% of top ten most frequently reported CVAEs, followed by hypotension (16.5%), and bradycardia (15.5%) (Table 2).

Fig. 1.

Flow diagram for inclusion and exclusion of study. COVID-19 coronavirus disease 2019. ICSRs individual case safety reports, CVAEs cardiovascular adverse events

Table 1.

Demographics and characteristics of ICSRs with CVAEs associated with the five studied drugs in VigiBase during the study period

	Remdesivir	Nirmatrelvir/ritonavir	Molnuparvir	Baricitinib	Tocilizumab	All CVAEs
No. of reports, n (%)	2115 (16.2)	1045 (8.0)	70 (0.5)	352 (2.7)	443 (3.4)	13,091
Regions, n (%)
Americas	1667 (78.8)	830 (79.4)	36 (51.4)	344 (97.7)	316 (71.3)	8971 (68.5)
European region	321 (15.2)	163 (15.6)	29 (41.4)	6 (1.7)	66 (14.9)	3130 (23.9)
Western Pacific	23 (1.1)	52 (5.0)	2 (2.9)	2 (0.6)	32 (7.2)	261 (2.0)
South-East Asia	59 (2.8)	0 (0.0)	0 (0.0)	0 (0.0)	9 (2.0)	271 (2.1)
Eastern Mediterranean	45 (2.1)	0 (0.0)	3 (4.3)	0 (0.0)	19 (4.3)	430 (3.3)
African	0 (0)	0 (0)	0 (0)	0 (0)	1 (0.2)	28 (0.2)
Age groups, n (%)
Under  18 years	71 (3.4)	1 (0.1)	0 (0)	6 (1.7)	2 (0.4)	361 (2.8)
18–44 years	289 (13.7)	142 (13.6)	2 (2.9)	62 (17.6)	89 (20.1)	3001 (22.9)
45–74 years	1296 (61.3)	625 (59.8)	34 (48.6)	243 (69.0)	278 (62.8)	7414 (56.7)
≥ 75 years	459 (21.7)	277 (26.5)	34 (48.6)	41 (11.6)	74 (16.7)	2315 (17.7)
Sex, n (%)
Male	1245 (58.9)	323 (30.9)	24 (34.3)	252 (71.6)	323 (72.9)	6600 (50.4)
Female	870 (41.1)	722 (69.1)	46 (65.7)	100 (28.4)	120 (27.1)	6491 (49.6)
Report type, n (%)
Spontaneous	1996 (94.4)	1022 (97.8)	69 (98.6)	348 (98.9)	399 (90.1)	11,511 (87.9)
Report from study	113 (5.3)	23 (2.2)	1 (1.4)	4 (1.1)	20 (4.5)	1371 (10.5)
Other/unknown	6 (0.3)	0 (0.0)	0 (0)	0 (0)	24 (5.4)	209 (1.6)
Report period, n (%)
Pre-omicron	1622 (76.7)	0 (0.0)	0 (0)	172 (48.9)	290 (65.5)	7226 (55.2)
Omicron	493 (23.3)	1045 (100.0)	70 (100)	180 (51.1)	153 (34.5)	5865 (44.8)

ICSRs individual case safety reports, CVAEs cardiovascular adverse events

Table 2.

Demographics and characteristics of the top ten most frequently reported CVAEs associated with COVID-19 treatments during the study period

	Tachycardia	Hypotension	Bradycardia	Flushing	Hypertension	Palpitations	Cardiac arrest	Atrial fibrillation	Sinus bradycardia	Pallor	Total
Event number, n (%)	1357 (17.1)	1308 (16.5)	1234 (15.5)	910 (11.4)	801 (10.1)	705 (8.9)	577 (7.3)	471 (5.9)	305 (3.8)	280 (3.5)	7948
Regions, n (%)
Americas	905 (66.7)	1016 (77.7)	855 (69.3)	851 (93.5)	586 (73.2)	275 (39.0)	491 (85.1)	365 (77.5)	121 (39.7)	260 (92.9)	5725 (72.0)
European	280 (20.6)	169 (12.9)	255 (20.7)	40 (4.4)	161 (20.1)	304 (43.1)	62 (10.7)	91 (19.3)	148 (48.5)	19 (6.8)	1529 (19.2)
Eastern Mediterranean	123 (9.1)	40 (3.1)	16 (1.3)	13 (1.4)	10 (1.2)	52 (7.4)	11 (1.9)	9 (1.9)	7 (2.3)	1 (0.4)	282 (3.5)
South-East Asia	35 (2.6)	22 (1.7)	63 (5.1)	1 (0.1)	22 (2.7)	35 (5.0)	2 (0.3)	0 (0)	21 (6.9)	0 (0)	201 (2.5)
Western Pacific	13 (1.0)	60 (4.6)	45 (3.6)	5 (0.5)	20 (2.5)	23 (3.3)	6 (1.0)	6 (1.3)	8 (2.6)	0 (0)	186 (2.3)
African	1 (0.1)	1 (0.1)	0 (0)	0 (0)	2 (0.2)	16 (2.3)	5 (0.9)	0 (0)	0 (0)	0 (0)	25 (0.3)
Age groups, n (%)
Under 18 years	48 (3.5)	24 (1.8)	39 (3.2)	28 (3.1)	11 (1.4)	8 (1.1)	9 (1.6)	0 (0)	12 (3.9)	25 (8.9)	204 (2.6)
18–44 years	534 (39.4)	244 (18.7)	229 (18.6)	456 (50.1)	135 (16.9)	277 (39.3)	85 (14.7)	36 (7.6)	64 (21.0)	92 (32.9)	2152 (27.1)
45–74 years	616 (45.4)	756 (57.8)	725 (58.7)	364 (40)	485 (60.6)	364 (51.7)	356 (61.7)	262 (55.6)	184 (60.4)	142 (50.7)	4254 (53.5)
≥ 75 years	159 (11.7)	284 (21.7)	241 (19.5)	62 (6.8)	170 (21.2)	56 (7.9)	127 (22.0)	173 (36.7)	45 (14.8)	21 (7.5)	1338 (16.8)
Sex, n (%)
Male	609 (44.9)	730 (55.8)	731 (59.2)	231 (25.4)	323 (40.3)	222 (31.5)	368 (63.8)	274 (58.2)	172 (56.4)	149 (53.2)	3809 (47.9)
Female	748 (55.1)	578 (44.2)	503 (40.8)	679 (74.6)	478 (59.7)	483 (68.5)	209 (36.2)	197 (41.8)	133 (43.6)	131 (46.8)	4139 (52.1)
Report period, n (%)
Pre-omicron	813 (59.9)	736 (56.3)	839 (68.0)	218 (24.0)	332 (41.4)	240 (34.0)	407 (70.5)	278 (59.0)	198 (64.9)	159 (56.8)	4220 (53.1)
Omicron	544 (40.1)	572 (43.7)	395 (32.0)	692 (76.0)	469 (58.6)	465 (66.0)	170 (29.5)	193 (41.0)	107 (35.1)	121 (43.2)	3728 (46.9)

CVAEs cardiovascular adverse events

The five studied COVID-19 treatment drugs, remdesivir, nirmatrelvir/ritonavir, molnupiravir, baricitinib and tocilizumab contributed to 4025 CVAEs during the study period. The analysis of exposure ORs revealed notable associations between various COVID-19 treatments and incidence of CVAEs, as summarized in Fig. 2.

Fig. 2.

Forest plot of reporting odds ratios for CVAEs associated with five studied drugs during the study period. CVAEs cardiovascular adverse events, ICSRs individual case safety reports, IC/IC_0.25 information component and its 95% credibility interval lower end point,  aROR adjusted reporting odds ratio, CI confidence interval

CVAEs by Treatment

Remdesivir

Our analysis identified 20,373 ICSRs associated with remdesivir, of which 2115 were CVAEs (Fig. 2). The adjusted ROR (95% CI) for CVAEs was calculated to be 2.4 (2.28–2.52), indicating patients who received remdesivir had more than twice the likelihood of experiencing CVAEs compared to those not receiving this antiviral agent. Of the total, 61.3% reports came from individuals aged 45–74 years, and 58.9% of the reported cases were men (Table 1). The most frequently reported events included bradycardia, with 503 cases and an aROR (95% CI) of 7.94 (7.07–8.93) and an information component (IC_0.25) of 2.33. A significantly higher incidence of AEs was observed, including hypotension, cardiac arrest, sinus bradycardia, atrial fibrillation, cardio-respiratory arrest, shock, and pulseless electrical activity were observed in the remdesivir cohort compared to non-remdesivir, showing a significant cardiovascular risk associated with remdesivir treatment (Fig. 3a).

Fig. 3.

CVAEs associated with remdesivir, nirmatrelvir/ritonavir and molnupiravir. Forest plots of reporting odds ratios for ten most frequently reported CVAEs associated with A remdesivir, B nirmatrelvir/ritonavir, and C molnupiravir during the study period. ICSRs individual case safety reports, CVAE cardiovascular adverse events, %CVAE the percentage of a CVAE out of the total number of all CVAEs associated with the drug, IC/IC_0.25 Information component and its 95% credibility interval lower end point, aROR adjusted reporting odds ratio, CI confidence interval

Nirmatrelvir/Ritonavir

CVAEs associated with nirmatrelvir/ritonavir were reported in 1045 cases out of 79,334 ICSRs. The aROR was calculated at 0.20 (95% CI 0.19–0.21; IC_0.25 − 1.93), suggesting a decreased likelihood of CVAEs compared to non-nirmatrelvir/ritonavir treatments. Among the common CV events linked to nirmatrelvir/ritonavir were hypertension and palpitations, with aROR of 0.88 (95% CI 0.75–1.03; IC_0.25 − 0.22) and 0.63 (95% CI 0.53–0.76; IC_0.25 − 0.72), respectively. The IC values for these events were negative, indicating no significant disproportionate reporting (Fig. 3b). Most affected individuals were women (69.1%) (Table 1).

Molnupiravir

Molnupiravir demonstrated a moderate exposure odds ratio for CVAEs, with an aROR of 0.54 (95% CI 0.43–0.69; IC_0.25 − 1.06), indicating reduced risk, but the number of total ICSRs and CVAEs were both low (2417 and 70) (Fig. 2). The top CVAEs were hypotension, palpitations, and atrial fibrillation, but they were not disproportionately over-represented (Fig. 3c).

Immune Modulators

Baricitinib

Baricitinib was associated with a total of 352 CVAEs out of 3329 ICSRs, resulting in an aROR of 2.31 (95% CI 2.07–2.59; IC_0.25 1) (Fig. 2). Most reports involved individuals aged 45 years and older and were predominantly men (see Table 1). Thromboembolic events are notable, with deep vein thrombosis accounting for a significant percentage of all reported CVAEs of baricitinib (27.6%) (Fig. 4a). The aROR for deep vein thrombosis was strikingly high at 45.34 (95% CI 34.89–58.9, IC_0.25 4.5). Other CVAEs were also notable: cardiac arrest (aROR 3.86, 95% CI 2.67–5.6, IC_0.25 1.47), hypotension (aROR 1.53, 95% CI 1.03–2.26, IC_0.25 0.1), pulseless electrical activity (aROR 7.09, 95% CI 3.89–12.93, IC_0.25 1.73), thrombosis (aROR 6.92, 95% CI 3.71–12.91, IC_0.25 1.52), acute myocardial infarction (aROR 4.85, 95% CI 2.46–9.54, IC_0.25 0.92) and cardio-respiratory arrest (aROR 2.81, 95% CI 1.44–5.48, IC_0.25 0.41) were significantly more frequent after baricitinib use, highlighting substantial safety concerns related to the use of baricitinib.

Fig. 4.

CVAEs associated with baricitinib and tocilizumab. Forest plots of reporting odds ratios for the ten most frequently reported CVAEs associated with A baricitinib and B tocilizumab, and pie chart showing counts of C all types of thrombosis and D deep vein thrombosis in CVAEs during the study period. ICSRs individual case safety reports, CVAE cardiovascular adverse events, %CVAE the percentage of a CVAE out of the total number of all CVAEs associated with the drug, IC/IC_0.25 information component and its 95% credibility interval lower end point, aROR adjusted reporting odds ratio, CI confidence interval

Tocilizumab

Our study found 443 CVAEs linked to tocilizumab out of 7719 ICSRs (Fig. 2). The aROR was 1.16 (95% CI 1.05–1.28), suggesting a slight association with cardiovascular risks. The most frequently reported CVAEs were hypotension (aROR 1.48, 95% CI 1.14–1.93, IC_0.25 0.29), cardiac arrest (aROR 2.73, 95% CI 2.03–3.67, IC_0.25 1.14), and bradycardia (aROR 0.89, 95% CI 0.63–1.24, IC_0.25 − 0.49) (Fig. 4b). Deep vein thrombosis, which was the most prevalent CVAEs associated with baricitinib, was not among the top ten CVAEs of tocilizumab.

Thrombosis

Since baricitinib, but not tocilizumab, was associated with an unexpected high risk of thrombosis, we further examined the pattern of thrombosis among patients with COVID-19 (Fig. 4c). We found that baricitinib accounted for 30.29% and 38.80% of cases of thrombosis and deep vein thrombosis reported and is the largest contributor of ICSRs involving these CVAEs.

Analysis During the Omicron and Pre-Omicron Period

Next, we examined if the CVAE profiles differed during the Omicron and pre-Omicron periods. We defined the Omicron period as November 24, 2021 onwards, after the Omicron variant was first reported by the WHO [15], while ICSRs prior to this date were classified as pre-Omicron and comprise the original strain and other main variants of concerns include Alpha (B.1.1.7), Beta (B.1.351), Gamma (P.1), Delta (B.1.617.2) [20].

In the Omicron period, there were 160,578 reported AEs related to COVID-19, including 5865 classified as CVAEs. The pre-Omicron phase recorded 116,053 total adverse events, with 7226 being CVAEs. The demographic and case characteristics of CVAEs reported for studied drugs in pre-Omicron vs. Omicron periods are shown in Table S1.

Both nirmatrelvir/ritonavir and molnupiravir were introduced during the Omicron period; thus no pre-Omicron data was available for analysis. By contrast, remdesivir, baricitinib and tocilizumab were first approved by the US-FDA for emergency use against COVID-19 in May 2020 [21, 22], November 2020 [6, 23], and June 2021 [24, 25], respectively, and this allows us to interrogate the interaction between the variants and CVAEs of these treatments. Comparing ICSRs reported in the two periods, we observed a significant increase in the risk of CVAEs during the Omicron period for all three drugs, while the patterns of CVAEs were broadly similar (Fig. 5). The demographic characteristics of the ICSRs are shown in Table S1. The aROR for CVAEs with remdesivir was 2.42 during the Omicron period, compared to 1.96 (p < 0.001) during the pre-Omicron period. In both periods, bradycardia was the most common, followed by hypotension, cardiac arrest and sinus bradycardia (Fig. 6). Similar trends were observed with baricitinib and tocilizumab, with an increased exposure odds ratio of CVAEs in Omicron vs. pre-Omicron periods. Baricitinib showed an elevated risk in the Omicron period, with the aROR rising from 1.73 to 3.16 (p < 0.001), with pronounced increases in the aROR for deep vein thrombosis (pre-Omicron vs. Omicron: 27.51 vs. 85.29) and cardiac arrest (vs. pre-Omicron vs. Omicron: 2.31 vs. 8.35) cardiac arrest (Fig. 6). Similarly, the CV risk of tocilizumab rose from 0.93 to 1.63 (p < 0.001).

Fig. 5.

Forest plot of reporting odds ratios of CVAEs associated with remdesivir, baricitinib, and tocilizumab in the pre-Omicron vs. Omicron periods. CVAEs cardiovascular adverse events, ICSRs individual case safety reports, IC/IC_0.25 information component and its 95% credibility interval lower end point, aROR adjusted reporting odds ratio, CI confidence interval, p value for z-test for difference of aROR in two periods

Fig. 6.

CVAEs associated with remdesivir, baricitinib, and tocilizumab in the pre-Omicron vs. Omicron periods. Forest plots of reporting odds ratios for ten most frequently reported CVAEs associated with A remdesivir, B baricitinib, and C tocilizumab in the pre-Omicron vs. Omicron period. CVAEs cardiovascular adverse events, ICSRs individual case safety reports, CVAE cardiovascular adverse events, %CVAE the percentage of a CVAE out of the total number of all CVAEs associated with the drug, IC/IC_0.25 information component and its 95% credibility interval lower end point, aROR adjusted reporting odds ratio, CI confidence interval, p value for z-test for difference of aROR in two periods

Discussion

Since COVID-19 emerged as a global pandemic, there have been numerous studies on treatment that can ameliorate this disease. Most studies focus on the efficacy, but safety concerns have been less extensively evaluated. CVAEs are common among patients with COVID-19 because this disease can damage the heart and because CV co-morbidities are very frequent among these patients. Therefore, treatment-related CVAEs might have been overlooked. In this retrospective pharmacovigilance cohort study, we analysed CVAEs associated with five COVID-19 treatments for patients with this disease. Our findings highlight significant differences in the incidence of CVAEs among various antiviral agents and immune modulators. Among small molecular antivirals designed to suppress viral replication, remdesivir was associated with higher odds of CVAEs, with bradycardia being the most common. Conversely, nirmatrelvir/ritonavir and molnupiravir were associated with decreased odds of CVAEs. Among immune modulators designed to control the cytokine storm, baricitinib use carries increased risks of CVAEs. In particular, the risks of thrombotic embolism (deep vein thrombosis and thrombosis) were both substantially increased with baricitinib, while these trends were not observed with tocilizumab. Considering that these two drugs are similarly indicated to treat patients with severe COVID-19, our study highlights potential CV safety concerns specifically associated with baricitinib.

To date, there have been limited reports on the adverse CV effects of nirmatrelvir/ritonavir. Our analysis shows that nirmatrelvir/ritonavir exhibits the smallest OR for CV events, indicating a weak association with CVAEs, consistent with the results of previous studies [26, 27]. Nevertheless, there have been several reported cases of CVAEs associated with the use of nirmatrelvir/ritonavir, including a possible instance of nirmatrelvir/ritonavir-induced bradycardia in a patient with asymptomatic COVID-19 [28, 29]. Extra caution should be exercised when using nirmatrelvir/ritonavir, particularly due to its potential for drug-drug interactions, as ritonavir is a strong CYP3A4 inhibitor [1]. Similarly, our current work and others show that molnupiravir has a lower risk of CVAEs and serious adverse events when compared to other COVID-19 treatments [30, 31]. Despite its favorable CV safety profile, concerns about its potential to promote resistance and mutagenesis raise important considerations for its use [1, 32]. Additionally, molnupiravir has not received FDA full approval, which further questions its clinical application [33]. Nevertheless, the WHO has recommended molnupiravir as an alternative to nirmatrelvir/ritonavir in certain situations, acknowledging its role in the treatment landscape for COVID-19 [32]. Conflicting evidence exists regarding the cardiovascular safety of remdesivir. A recent post hoc safety analysis on a phase 3 study of remdesivir showed no significant association between this drug and risk of CVAEs compared with control in patients hospitalized with moderate or severe COVID-19 [34]. However, our findings show that remdesivir is linked to a higher incidence of CVAEs, such as bradycardia, sinus bradycardia and hypotension, with an overall aROR of 2.4 (95% CI 2.28–2.52), which aligns with previous pharmacovigilance studies that identified significant associations between remdesivir and cardiovascular adverse drug reactions such as cardiac arrest, bradycardia, and cardiogenic shock, as well as numerous reported cases of bradycardia among patients treated with this drug [35–37]. Studies using human pluripotent stem cell-derived and adult primary cardiomyocytes by our group and others also demonstrated the potential of remdesivir to induce injury and alter cardiac function in vitro [36, 38, 39].

While our results on antivirals are broadly consistent with those reported in prior studies, we made the unexpected finding that baricitinib is associated with increased odds of CVAE (aROR 2.31) and substantially higher odds of deep vein thrombosis (aROR 45.3) and thrombosis (aROR 6.92). Furthermore, arterial thrombosis is a frequent cause of acute myocardial infarction, and the latter is also overrepresented among the AEs of baricitinib (aROR 4.85). Contradictory evidence exists regarding the safety profile of baricitinib. A Swedish comparative safety study showed that JAK inhibitors, including baricitinib, were associated with increased incidence of venous thromboembolism among patients with rheumatoid arthritis [10]. Indeed, JAK inhibitors, including baricitinib, carry a black box warning for venous thromboembolism [6]. Conversely, a meta-analysis of patients with rheumatoid arthritis showed that JAK inhibitors, including baricitinib, were not associated with increased CVAEs or venous thromboembolism, although it was noted that a lower dosage (2 mg) of baricitinib was safer than a higher dosage (4 mg). When used in patients with COVID-19, baricitinib was associated with reduced mortality compared to standard of care [40], while the frequencies of serious adverse events and venous thromboembolic events were similar between the two groups. A meta-analysis that examined the efficacy and safety of baricitinib against COVID-19 also revealed improved mortality and similar rates of AEs (including MACE and deep vein thrombosis) in baricitinib vs. control patients [41]. In light of the above studies with diverse results, the dramatically increased odds of deep vein thrombosis were unexpected. Tocilizumab is another immune modulator approved to treat severe COVID-19. Like baricitinib, there is a lack of consensus regarding the safety of tocilizumab. Some studies did not find any correlation between the use of tocilizumab and cardiovascular risk among patients with rheumatoid arthritis [42] or severe COVID-19 [43], or while others showed higher reporting of cardiac failure and embolic and thrombotic events [11]. Here, we found that tocilizumab was associated with only modest or non-significant increases in CVAEs. Under FDA approvals, baricitinib and tocilizumab are both indicated for hospitalized adults with COVID-19 requiring supplemental oxygen, non-invasive or invasive mechanical ventilation, or ECMO, with tocilizumab additionally requiring patients to receive systemic corticosteroids [6, 44]. Therefore, the greatly increased odds of CVAEs and deep vein thrombosis, which were seen with baricitinib but not tocilizumab, could not be explained by differences in disease severity of patient cohorts. Instead, since baricitinib carries a black box warning indicating an ‘increased risk of serious heart-related events, cancer, blood clots, and death for JAK inhibitors that treat certain chronic inflammatory conditions’, patients with risk factors for these disorders are less likely to be prescribed baricitinib [45]. The mechanisms through which baricitinib may induce deep vein thrombosis are not clear. Baricitinib targets the JAK pathway responsible for transmitting inflammatory signals from many cytokines, making the former a powerful immunomodulator. However, the ‘indiscriminate’ targeting of multiple cytokines and chemokines has also been proposed to upset the balance between pro-and anti-thrombotic signaling, with adverse consequences [46]. Severe COVID-19 can induce an acute proinflammatory state that can precipitate thrombosis. In a large cohort study of involving 48 million adults in England and Wales, COVID-19 was associated with a greatly increased incidence of venous thromboembolism [47], and both venous and arterial thromboembolism are associated with increased mortality. It is possible that baricitinib further perturbs endothelial cells, leukocytes, or other components of inflammation to exacerbate thromboembolism.

Since COVID-19 was first detected, multiple variants of this virus have emerged, and they differ in their tropism and pathogenicity. In particular, the Omicron variant is known for its high transmissibility and reduced severity compared to earlier variants such as the wild-type, Alpha, and Delta variants [48]. These variants also differ in their abilities to infect components of the cardiovascular system, and to inflict damage on the heart [39]. In light of the differences, we compared the reporting of CVAEs of COVID-19 treatment in pre-Omicron vs. Omicron periods. To our knowledge, this is the first report to do so. Overall, we found increased odds of CVAEs during the Omicron vs. pre-Omicron periods, which is opposite to the pathogenicity of these variants. The reasons for these differences are not clear. More clinical studies are needed to confirm these results and dissect the mechanisms.

This study's key strength lies in its use of VigiBase, a comprehensive global database of ICSRs. This resource offers a large sample size and diverse data sources, facilitating the identification of rare adverse drug reactions and generating hypotheses regarding drug safety that may not be observed in clinical studies. The inclusion of reports from multiple countries and various types of reporters enhances the generalizability of the findings [16]. However, the study also has several limitations. Its retrospective design and reliance on spontaneous reporting may introduce biases and underreporting. Additionally, the lack of detailed clinical information and potential confounding factors could compromise the accuracy of the findings [16]. Treatment protocols for COVID-19 have evolved over time, which may affect the consistency and comparability of data across different periods [32]. Another limitation is the potential for misclassification of ADRs due to the use of MedDRA coding, which may not always accurately reflect the clinical diagnosis. Moreover, the severity of COVID-19 and the presence of cardiovascular co-morbidities were not specified in the reports, yet these can significantly influence the occurrence of CVAEs. Patients with more severe disease and/with co-morbidities (coronary disease, diabetes etc.) are at a higher risk for adverse outcomes, potentially confounding the relationship between treatment and CVAEs [49]. Furthermore, many patients with COVID-19 are treated with multiple medications simultaneously, and the interactions between these drugs, as well as their combined effects on cardiovascular health, are not fully accounted for in this study. The analysis also did not evaluate the impacts of COVID-19 co-infections with other viruses and medication interactions on adverse events due to data limitations, and this represents another constraint in the analysis.

Conclusions

Our study revealed differences in the CV safety profiles of COVID-19 treatment, and during different periods of the pandemic. These results underscore the need for continued post-marketing surveillance and safety monitoring in patients receiving these treatments, particularly baricitinib. Since tocilizumab and baricitinib have similar efficacy against COVID-19 [50] and are indeed similarly indicated for use among patients with severe COVID-19, further clinical trials are needed to test if tocilizumab might be a safer alternative to baricitinib. Next, the treatments we examined are amendable for repurposing: the remdesivir and molnupiravir exhibit broad-spectrum antiviral activity, while baricitinib and tocilizumab can be used to treat multiple immune disorders. These results may guide the risk–benefit assessment of future pandemics.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Conceptualization: Ellen Ngar-Yun Poon; Methodology: Hoi Ki Cheng, Angel Lai, Maxwell Kwok; Formal analysis and investigation: Hoi Ki Cheng, Maxwell Kwok; Writing and editing manuscript: Hoi Ki Cheng, Maxwell Kwok, Ellen Ngar-Yun Poon; Funding acquisition: Bryan P. Yan, Ellen Ngar-Yun Poon; Supervision: Bryan P. Yan, Ellen Ngar-Yun Poon. All authors read and approved the final manuscript.

Funding

This work was supported by funding from the Health and Medical Research Fund (22210062) from the Health Bureau, Hong Kong SAR, China, to E.N.P. The Rapid Service Fee was funded by the Chinese University of Hong Kong.

Data Availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Conflict of interest

Hoi Ki Cheng is an employee of Pfizer Inc, but the latter did not play any role in the design and execution of this study. All other authors (Angel Lai, Maxwell Kwok, Bryan P Yan, Ellen N Poon) have no conflicts of interest relevant to the content of this manuscript.

Ethical approval

This article is based on available data in VigiBase (https://who-umc.org/vigibase-search-services/about-vigibase/), and the data is anonymous and is available upon request from said database. This manuscript does not contain any new studies with human participants or animals performed by any of the authors. The study was approved by the ethics committee of The Chinese University of Hong Kong (CUHK) and New Territories East Cluster (NTEC) (Ref. No. 2023.186). Since only anonymous data was collected, informed consent was waived.