Introduction

The coronavirus disease 2019 (COVID-19) caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is associated with a high pulmonary burden leading to hypoxemic respiratory failure that may ultimately result in acute respiratory distress syndrome (ARDS)1,2. Moreover, extrapulmonary complications have been described such as cardiovascular (CV) events3,4,5. The intricate pathophysiology remains unclear, and two phases have been proposed. The virus initially infects respiratory epithelial and immune cells through the angiotensin-converting enzyme type 2 receptor, leading to cytotoxic damage. Secondly, proinflammatory damage causes dysregulation of proinflammatory cytokines, alveolar edema, and the formation of a thick hyaline membrane that impairs gas exchange. It is hypothesized that hypoxia, endothelial damage, immune dysregulation, and a transitory hypercoagulable state contribute to thrombosis development1,6,7. Several biomarkers have also been proposed to predict severe COVID-19 or the risk of CV events8,9; whilst antiphospholipid antibodies do not seem to be involved in the hypercoagulable state in the setting of SARS-CoV-2 infection10. Besides, a higher risk of venous thromboembolism (VTE), in particular pulmonary embolism (PE), has been reported; however, the true epidemiology of PE in hospitalized COVID-19 patients remains unclear with a wide range of reported incidence rates from 5 to 85% depending on the clinical scenario3,11,12,13,14. Despite this variability, COVID-associated VTE overall rates appear to be higher than non-COVID-associated VTE (or classic VTE) among critically ill patients (4.9 vs. 1.7%, respectively), and especially for PE (20.6 vs. 6.1%, respectively)11. Moreover, myocardial injury has been reported in 20–30% and biventricular cardiomyopathy has been reported in up to 7–33% of critically ill COVID-19 patients depending on different cohorts; while severe neurologic manifestations such as acute stroke have been reported in up to 6% of those with severe illness4.

Nevertheless, CV complications related specifically to therapies against SARS-CoV-2 have barely been evaluated. This study aims to assess the potential association between treatment options for acute SARS-CoV-2 infection and the risk of developing VTE and other cardiovascular events such as myocardial infarction (MI) or ischemic stroke (IS) within a large national cohort of hospitalized COVID-19 patients.

Materials and methods

Study design

The present study is a retrospective, multicenter, observational study that includes patients hospitalized due to SARS-CoV-2 infection from March 2020 to March 2022 in Spain. The “first wave” was defined as all cases occurring before July 7, 2020, and the “second wave” as cases occurring between July 7, 2020, and December 25, 2020. Due to the low number of patients recorded in this registry from third wave onward, cases occurring after December 25, 2020, were categorized under the variable “other waves”15. The Spanish Society of Internal Medicine sponsored the nationwide SEMI-COVID-19 Registry.

Population of the study

For the SEMI-COVID-19 Registry, an online electronic data capture system (DCS) was developed, which included a database manager along with procedures for verification of data and contrasting information against the original medical record to ensure the best possible quality of data collection. Patient-identifiable data were dissociated and anonymized. Data were collected retrospectively and included more than 300 variables grouped under different headings: inclusion criteria, epidemiological data, personal medical and medication history, symptoms and physical examination findings at admission, wide laboratory panel at admission, diagnostic imaging tests, and complications during the hospitalization. More in-depth information about the registry can be found in previously published works16. Inclusion criteria were age over 18 years old, consent information, and complete data for the outcome variable. Exclusion criteria were denial or withdrawal of informed consent or the presence of missing relevant data related to admission and therapies. From 28,444 patients included in the SEMI-COVID-19 Registry in March 2022, a total of 18,324 patients were included as the final sample. Demographic characteristics, clinical and radiological presentation, laboratory biomarkers upon admission, treatment modalities, complications, and outcomes were analyzed.

Data analysis

The Shapiro–Wilk test determined the normality of continuous quantitative variables. The study reported categorical data as proportions and continuous data as mean and standard deviation (SD) or median and interquartile range, depending on their normality. We used the Student's T-test and analysis of variance (ANOVA) to compare means in 2 or more independent categories, respectively, when the variable followed a normal distribution. The nonparametric alternatives were the Mann–Whitney U and the Kruskal–Wallis test, respectively. The receiver operating characteristic area under the curve (AUC) analysis and the Youden index was used to explore the best cut-off point for predicting adverse outcomes for laboratory biomarkers. Multicollinearity was statistically assessed with the variance inflation factor (VIF) test for all potential confounding factors in the bivariate model, allowing a range for the VIF between 1 and 5, and variables with a VIF > 5 were removed from the model. The bivariate logistic regression test evaluated the potential association between different variables and the adverse outcomes using a binomial logistic regression (crude odds ratio, OR). Multivariate logistic regression analysis (adjusted OR) evaluated the independent association of the variables with a p value < 0.20 obtained in the bivariate analysis. Bivariate and multivariate analyses were performed for myocardial infarction, ischemic stroke, and VTE. All tests were two-sided and the level of statistical significance was set at p < 0.05. The statistical analysis was carried out by R, version 4.2.2, R Core Team, Vienna, Austria (main R packages used: tidyverse, pROC, survival).

Ethical aspects

The SEMI-COVID-19 Registry and related studies were approved by the Institutional Research Ethics Committee of Malaga (Spain) in March 2020 (Ethics Committee code: SEMI-COVID-19 27-03-20). All included patients or their legal representatives gave their informed consent. In cases in which there were biological safety concerns or the patient had already been discharged, informed consent was obtained verbally and noted on the patient's medical record. Patient-identifiable data were dissociated and anonymized to ensure confidentiality and privacy. All experiments were performed in accordance with relevant guidelines and regulations.

Results

The study included 18,324 hospitalized COVID-19 patients, with a mean age of 67.5 (± 16) years. Among the participants, 58.4% (10,701 patients) were male, and 91% (16,674 patients) were Caucasian. The most prevalent comorbidity was arterial hypertension affecting 52.5% (9618 patients), followed by dyslipidemia in 39.5% (7237 patients), diabetes mellitus in 23.7% (1748 patients), atrial fibrillation (AF) in 10.6% (1948 patients), chronic ischemic heart disease in 8.7% (1586 patients), and chronic ischemic stroke in 7.4% (1356 patients). Symptoms at presentation included fever (79%), cough (69.9%), and dyspnea (60.6%). Among therapies most commonly used were low molecular weight heparin (LMWH) at standard prophylactic doses 84.8% (11,912 patients), beta-lactams 66.7% (12,228 patients), systemic corticosteroids 56.9% (10,424 patients), hydroxychloroquine 55.4% (10,146 patients), macrolides 51.4% (9425 patients), lopinavir/ritonavir 38.4% (7044 patients), tocilizumab 11.6% (2128 patients), and beta interferon 6.8% (1802 patients). The overall ICU admission rate was 9.9% (1,819 patients), and the overall mortality rate was 17.4% accounting for 3,183 patients. In the SEMI-COVID-19 Registry, most of the patients recorded were admitted during the first two pandemic waves in Spain (82.2%): 11,852 patients (64.7%) in the first wave period and 3215 patients (17.5%) in the second wave period. Detailed information on other relevant comorbidities, clinical presentation, laboratory findings, and treatments can be found in Table 1. Among the entire sample, 4.08% (748 patients) experienced any CV event, while VTE occurred in 2.78% (510 patients), myocardial infarction occurred in 0.75% (137 patients), and ischemic stroke in 0.55% (101 patients). For laboratory biomarkers, the ROC curve analysis showed the best cut-off points for predicting cardiovascular events: 100 mg/L for C-reactive protein (CRP), 1000 ng/mL for D-dimer, 800/µL for lymphocytes, and 1.5 mg/dL for creatinine.

Table 1 Clinical, laboratory characteristics, and outcomes in our population.
Full size table

Therapies and the risk of VTE events

Bivariate logistic regression analysis for VTE is presented in Table 2. Multivariate analysis was performed for variables with a p value < 0.15 obtained in the bivariate analysis for the risk of VTE events (Table 3). Among all therapies, only the use of tocilizumab (adjusted OR 2.07; 95% CI 1.55–2.74; p < 0.01), corticosteroids (adjusted OR 1.44; 95% CI 1.06–1.97; p = 0.02), and macrolides (adjusted OR 0.58; 95% CI 0.45–0.75; p < 0.01) showed an independent association in the multivariate analysis, adjusted for age, sex, ARDS, PaO2/FiO2 ratio < 300, pandemic waves, ICU admission, CRP > 100 mg/L, other therapies, among other factors.

Table 2 Bivariate logistic regression analysis for VTE risk assessment.
Full size table
Table 3 Multivariate logistic regression analysis* for VTE risk assessment.
Full size table

Therapies and the risk of MI and IS events

Bivariate logistic regression analysis for myocardial infarction and ischemic stroke are presented in supplementary Tables S1 and S2, respectively. There was no independent association between any therapies used against SARS-CoV-2 infection and the risk of developing myocardial infarction (lopinavir/ritonavir [adjusted OR 0.63; 95% CI 0.36–1.05], beta-interferon [adjusted OR 1.63; 95% CI 0.75–3.32], and beta-lactams [adjusted OR 1.75; 95% CI 0.99–3.35]) or ischemic stroke (tocilizumab [adjusted OR 1.60; 95% CI 0.90–3.07]). Multivariate analysis was performed for variables with a p value < 0.15 obtained in the corresponding bivariate analysis for the risk of MI or IS events, respectively. Supplementary Table S3 summarizes the findings of the multivariate analysis.

Discussion

This national multicenter retrospective observational study included 18,324 hospitalized COVID-19 patients, making it one of the largest cohorts to analyze the incidence of CV events associated with therapies against SARS-CoV-2 infection. Risk factors for COVID-19-associated thrombosis are complex and not yet fully understood4,6,17. Within this cohort, VTE was observed in 2.78% (510 patients) with pulmonary embolism accounting for 80.4% (410 patients), followed by myocardial infarction in 0.75% (173 patients), and stroke in 0.55% (101 patients). Regarding VTE events, our findings align with previously published literature where PE is the most common form of VTE in COVID-19 patients, in opposition to classic VTE where the most common presentation is deep vein thrombosis11,12.

Furthermore, only a limited number of studies have assessed the potential role of different therapies against COVID-19 for the development of overall cardiovascular events, including VTE. Sagris et al.18 conducted a meta-analysis of 22 studies involving a total of 8,499 hospitalized COVID-19 patients, of whom 4,635 received immunomodulatory therapy. The meta-analysis aimed to evaluate the potential beneficial role of immunomodulatory therapies such as tocilizumab, anakinra, corticosteroids, and hydroxychloroquine to the risk of VTE (a total of 7,873 patients with 170 VTE events), ischemic stroke (a total of 4,352 patients with 26 ischemic stroke events), and myocardial infarction (a total of 5,438 patients with 23 myocardial infarction events) in hospitalized COVID-19 patients. The analysis did not find a significant association between these therapies and the risk of VTE (OR 0.84, 95% CI 0.61–1.16; I2: 0%), ischemic stroke (OR 1.10, 95% CI 0.50–2.40; I2: 0%), or myocardial infarction (OR 1.06, 95% CI 0.47–2.39; I2: 0%).

This large national cohort showed that the risk of developing VTE events was independently associated with elevated D-dimer levels (adjusted OR 3.64; 95% CI 2.86–4.66), cancer (adjusted OR 1.68; 95% CI 1.11–2.46), and ICU admission (adjusted OR 3.14; 95% CI 2.33–4.23). Among COVID-19 patients, those who received tocilizumab presented a higher risk of VTE (adjusted OR 2.07; 95% CI 1.55–2.74), independently of the use of LMWH at prophylactic doses, elderly age, sex, pandemic waves, presence of cancer, ICU admission, elevated D-dimer levels, or the severity of the disease: inflammation (CRP > 100 mg/dL), PaO2/FiO2 ratio < 300, pulmonary infiltrates, and the presence of ARDS (Table 2). The potential association between the development of new thromboembolic events and the use of tocilizumab remains controversial in the literature19,20. On the one hand, Reid et al.21 conducted a retrospective study of 176 hospitalized patients with moderate to severe COVID-19 who received either tocilizumab or baricitinib therapy, finding higher rates of venous thrombosis in the tocilizumab group compared to the baricitinib group (11.5% vs. 3.5%, respectively), including both PE and DVT. Similarly, Chober et al.22 retrospectively evaluated the effects of tocilizumab in 4,287 patients with SARS-CoV-2 infection who developed a cytokine storm and PE, reporting a higher mortality rate among those receiving tocilizumab therapy (43% vs. 19%), with an independent association demonstrated in a multivariate Cox proportional hazards analysis for mortality (HR 3.34; 95% CI 1.08–10.37). On the other hand, a multicenter retrospective study from Saudi Arabia that used propensity score matching (1:2 ratio) to evaluate 453 critically ill COVID-19 patients (151 of whom [33.3%] received tocilizumab) found no association between tocilizumab administration and the risk of VTE after matching (6.8% vs. 7.7%), with an odds ratio of 0.83 (95% CI 0.39–1.79)23. Furthermore, there is currently insufficient evidence to assert that elevated levels of interleukin-6 are independently associated with an increased risk of VTE24,25. Additionally, in a different clinical context, conflicting evidence has been reported regarding the use of tocilizumab in patients with vascular Behçet's syndrome, which is well-known by the development of thrombosis in the setting of hyperinflammation26. Some studies caution against its use due to the potential risk for new vascular relapses, including DVT, PE, pulmonary artery aneurysm, and other complications27,28. In addition, corticosteroids were also independently associated with the risk of VTE in this cohort (adjusted OR 1.44; 95% CI 1.06–1.97). Although it is not well understood, there is some evidence in the literature suggesting that both endogenous and exogenous corticosteroids might be potential risk factors for thromboembolic events. It is suggested that corticosteroids may lead to an imbalance in the complex pathways of the immune-thrombosis system, inducing a hypercoagulable state that promotes thrombosis29. Therefore, the results of the present study are of great interest, as they may help clarify whether there is a true association between the use of tocilizumab and corticosteroids and the onset of venous thrombotic events.

Furthermore, to address the potential confounding factor of atrial fibrillation in the development of thromboembolic events, AF was included as a variable in the multivariate analysis. The results were interesting, the use of tocilizumab persisted as an independent risk factor for VTE while the presence of AF acted as a protective factor as expected (AF adjusted OR 0.51; 95% CI 0.30–0.82). We hypothesize that most of patients with AF were on long-term anticoagulant therapy, and therefore maintaining anticoagulation during hospitalization would prevent the onset of venous thrombosis event. Besides, the use of macrolides also appeared to present an independent protective role against VTE (adjusted OR 0.58; 95% CI 0.45–0.75). Macrolides may exert anti-inflammatory activity inhibiting the activation of the immune-thrombosis pathways. Interestingly, low systolic blood pressure showed no significant association with VTE in the bivariate analysis (crude OR 1.06; 95% CI 0.59–1.75), neither was the presence of low PaO2/FiO2 ratio < 300 (adjusted OR 0.88; 95% CI 0.67–1.15). Although COVID-associated thrombosis might be associated with venous thrombosis at unusual sites, such as mesenteric or cerebral vein thrombosis30,31; these events were not analyzed specifically in the present study and would require further investigation. Finally, in this cohort, there was no association between any of the therapies used for SARS-CoV-2 infection and the risk of myocardial infarction or ischemic stroke (Table 3).

Our study has several limitations. First, it is a retrospective observational study, and prospective studies are needed to confirm these findings. Second, as a retrospective study, specific information regarding treatment, such as vaccination and other antiviral agents (paxlovid or molnupiravir) for SARS-CoV-2 infection, is not available in this registry. This represents a potential confounding factor when interpreting these results, as cardiovascular complications may have been missed due to the limitations imposed by the pandemic. Third, our study included patients from March 2020 to March 2022, and there was no registry of SARS-CoV-2 variants. Thus, indirect effect and confounding factors were measured by defining Spanish pandemic wave, which represents a potential limitation of the analysis. Fourth, since the study is based on Spaniard populations, there might be a lack of ethnic diversity bias. Nevertheless, this study has several strengths. It is a large multicenter national study that analyzed variables according to real clinical practice, and inclusion and exclusion criteria were established beforehand to minimize information biases.

Conclusions

This large national multicenter study found that hospitalized COVID-19 patients who received tocilizumab and corticosteroids had an independent association with the onset of venous thromboembolism events. Conversely, those who received macrolides appeared to have a reduced risk of thrombotic events. However, none of these therapies were independently associated with an increased risk of myocardial infarction or ischemic stroke. These results should be interpreted with caution due to potential biases, and further studies are needed to confirm these findings.