Introduction

A wealth of data in the medical scientific literature underscores the considerable potential for infections and sepsis, including viral sepsis such as COVID-19, to induce long-term cardiovascular complications1. This elevated rate of major adverse cardiovascular events (MACE) includes nonfatal AMI, acute heart failure (AHF), nonfatal stroke, and cardiovascular death2. Recent data suggest that even sepsis survivors without preexisting cardiovascular disease exhibited a heightened risk of subsequent cardiovascular events3. A meta-analysis of 27 studies examining sepsis as a potential long-term risk factor for cardiovascular disease revealed that the relative risk magnitudes associated with sepsis were comparable to those of established risk factors like hypertension, dyslipidemia, and diabetes mellitus. Remarkably, this elevated risk persisted significantly for at least five years after hospital discharge4.

While the acute and long-term complications of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), affecting the respiratory system, have been extensively documented in COVID-195,6, the associated cardiovascular complications have recently become a major focus of research7,8,9. However, most studies have focused on the short-term acute effects of SARS-CoV-2 to date (such as myocardial injury, arrhythmias, acute myocardial infarction (AMI), and venous thromboembolism (VTE))10, or relied on large administrative database with sometimes limited clinical detail11,12.

Given the recency of the disease, there is a growing body of research investigating longer-term complications potentially associated with COVID-19. Mainly utilizing databases, these studies assess clinical outcomes at one year, exploring aspects such as mortality rates and organ-specific complications13,14,15,16. One study even reported that while patients with COVID-19 did not exhibit an increased risk of cardiovascular death at one year compared to a control cohort, they did demonstrate an elevated risk of all-cause mortality, acute thromboembolic events, VTE, and severe cardiac arrhythmias at one year17. However, most of these studies were conducted using large databases with limited clinical information regarding the patients or their medical history, making it challenging to perform a detailed, granular analysis of these findings.

We hypothesized that patients admitted to the intensive care unit (ICU) due to a more severe form of COVID-19 are at a higher risk of developing MACE compared to those admitted to medical wards (MW). The primary endpoint of this study was MACE-free survival defined as the duration from hospital admission to the occurrence of the event of interest, according to the initial severity of the disease. Secondary objectives were to identify independent risk factors for MACE.

Material and methods

Patients and database

This study was a retrospective analysis conducted at a single academic center, Strasbourg University Hospital, Nouvel Hôpital Civil, France. The study period encompassed the first wave of the COVID-19 pandemic, from March 2020 to May 2020.

All consecutive adult patients (> 18 years old) admitted to Strasbourg University Hospital during the study period with confirmed COVID-19 diagnosis were included. No patient exclusions were applied. Confirmation of COVID-19 was based on a positive reverse-transcriptase-polymerase-chain-reaction (RT-PCR) assay performed on a nasopharyngeal swab specimen. In both groups the severity off illness was consistent with the WHO Clinical Progression Scale for COVID-1918 : all ICU patients fulfilled the criteria for “hospitalized: severe disease” and all MW patients met the criteria for “hospitalized: moderate disease”. Anonymized data from all patients were collected via our institution’s electronic health record. A complete set of clinical, biological, and imaging data were recorded. SOFA and Simplified Acute Physiology Score II (SAPS II) severity scores were routinely recorded in ICU departments as part of standard clinical practice. However, these scores were not commonly collected in other medical departments. Given the exceptional nature of the pandemic, a special authorization for clinical and biological research was obtained at the institutional level and was approved by the Ethics Committee of the University Hospital of Strasbourg (NCE-2020-51). Data were collected as part of the COVID-HUS registry (NCT04405726) and Seneshock study (NCT03559569). Procedures were followed in accordance with the ethical standards of the responsible committee on human experimentation (institutional) and with the Helsinki Declaration.

Study follow up

An anonymized and pre-defined electronic Case Report Form was used to gather all pertinent data. The data fields were structured to capture information related to cardiovascular risk factors, medical conditions, medications, and clinical outcomes. A 12-month follow-up was conducted by phone with the patient and/or their general practitioner as part of the institution’s established COVID-19 patient outcome protocol, implemented at the onset of the pandemic (COVID-HUS). Cardiovascular complications were verified by retrieval of medical reports. Mortality cases were cross-referenced using a French national database called the “Match ID” website, and the cause of death was ascertained through direct communication with the attending practitioner in September 2022.

Study outcomes

The primary endpoint of this study was MACE-free survival defined as the duration from hospital admission to the occurrence of the event of interest, according to the initial severity of the disease. MACE, as defined in our analysis, encompassed cardiovascular mortality, resuscitated cardiac arrest, AMI categorized into ST-segment elevation myocardial infarction and non-ST-segment elevation myocardial infarction, stroke, and AHF19,20. In accordance with the latest guidelines, AHF was defined as the sudden onset or rapid exacerbation of heart failure symptoms and signs, either in patients with de novo heart failure or those with chronic heart failure. Patients were diagnosed with AHF if they presented with dyspnea secondary to acute pulmonary edema, compatible auscultatory findings, evidence of right heart failure, suggestive medical imaging findings (chest X-ray or CT scan), and elevated natriuretic peptide biomarkers (such as BNP or NT-proBNP). These criteria necessitated immediate intervention and typically resulted in rapid clinical improvement21,22. Patients who died from other causes were censored at the date of death. Moreover, secondary objectives were to identify risk factors independently associated with the occurrence of MACE in the population through multivariable analysis.

Statistical analysis

Categorical variables are reported as frequencies with percentages. Continuous variables are expressed as median (interquartile range [IQR]). An inverse probability of treatment weighting (IPTW) approach was used to balance the baseline characteristics between the two groups (patients admitted to MW or ICU). A propensity score was estimated using a multivariable logistic regression model. The dependent variable was the admission service, and the independent variables were all the variables that may confound the exposure-outcome relationship (age, sex, arterial hypertension, obesity, smoking status, MI, peripheral artery occlusive disease (PAOD), diabetes, atrial fibrillation (AF), VTE, malignancy). Then, the stabilized weights were computed to create a pseudo-population (weighted cohort) in which the distribution of baseline covariates is independent of treatment assignment. We used absolute standardized differences (ASD) to assess the comparability of the baseline covariates between the two groups. Covariates with ASD < 0.1 denote non-meaningful imbalance23. For each time-to-event outcome, we performed univariable weighted Cox regression model to estimate the hazards-ratio (HR) of the group admitted in ICU with its 95% confidence interval (CI). A confounder-adjusted survival curve was performed using the weighted Kaplan-Meier estimator. In the multivariate model, traditional cardiovascular risk factors were incorporated (e.g. age, smoker status, history of AMI, PAOD, AF), all comorbidities are recorded as preexisting conditions prior to the index hospitalization. A P-value < 0.05 was considered statistically significant. All the analyses were performed using R software. R Core Team, 2021 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Baseline characteristics

Among 340 patients admitted to Strasbourg University Hospital with COVID-19 during the study period, 138 were admitted to the ICU. Among them, 32 died, one due to cardiovascular death and 31 due to other causes. Of the 202 patients hospitalized with COVID-19 in the Medical Ward, 11 died during hospitalization none from cardiovascular cause and no patients experienced any MACE during their stay.

Due to missing data for 8 patients mainly regarding medical history and consequently impacting the weighting score (6 ICU patients and 2 MW patients), the final analysis included 332 patients (132 in ICU and 200 in MW). A detailed flowchart and study timeline are presented in Fig. 1. Regarding the evaluation criteria, i.e. MACEs and patient outcome, we were able to obtain all information, with no data missing.

The median age in the ICU population was 60 years (IQR 51.0, 70.0) compared to 65 years (IQR 53, 76.7) in the MW. We observed moderate covariate imbalances between patients admitted to the MW and ICU in the original population, with absolute standardized differences ranging from 0.1 to 0.25. Specifically, the ICU population was younger, predominantly male, had a higher incidence of obesity and smoking, fewer medical histories of VTE, active malignancy, AF, stroke, and PAOD compared to the MW patients (Table S1). After inverse probability of treatment weighting (IPTW) using stabilized weights, all baseline characteristics were well balanced between the two groups and had absolute standardized difference below 0.1 (Table 1) (Fig. S1).

Table 1 Baseline characteristics in weighted population by propensity score (in the pseudo-population) AH, arterial hypertension; AMI, acute myocardial infarction; ICU, intensive care unit; ASD, absolute standardized difference; MW, medical ward; PAOD, peripheral arterial obstructive disease; AF, atrial fibrillation; VTE, venous thromboembolism. *Absolute standardized differences (ASD) were used to assess the comparability of the baseline covariates between the two groups. Covariates with ASD < 0.1 denote non-meaningful imbalance.
Full size table

For ICU patients, the median SAPS II score is 49.5 (IQR 38, 64). With respect to disease severity, all patients admitted to the medical ward met the ‘Hospitalized: moderate disease’ criteria according to the WHO Clinical Progression Scale for COVID-19, while all ICU patients met the criteria for ‘Hospitalized: severe disease.

Outcomes

The ICU population experienced a significantly higher risk of MACE during the first year, compared to the MW population (HR 16.2 [5.7–45.8] p < 0.001) (Table 2). A weighted Kaplan–Meier curve is presented in Fig. 2, highlighting that most events occur mainly within the first 3 months following the SARS-CoV-2 infection. More precisely, the ICU population had a significantly higher risk of stroke at 1 year compared to MW population (HR 12.1 [2.6–55.8], p < 0.001) and AHF (HR 14.5 [3.3–63.1], p < 0.001), but also death from any cause (HR 3.1 [1.8–5.4], p < 0.001) (Table 2). There were only 3 cardiovascular deaths in the entire cohort, 4 AMI, and 2 resuscitated cardiac arrests in COVID patients.

Table 2 Primary endpoint in the weighted population (after propensity score). AHF, acute heart failure; ICU, intensive care unit; MW, medical ward; HR, hazard ratio; MACE, major adverse cardiovascular event.
Full size table

Four patients from the MW group presented with a MACE after being discharged from the hospital. In the ICU group, 23 patients presented with a MACE, with 4 events occurring during ICU hospitalization and 19 after ICU discharge. Among these 19 events, 9 occurred during post-ICU hospitalization, and 10 events occurred after discharge to home.

In univariable analysis, several factors appear to be associated with the occurrence of MACE. Smoking status seems to be linked to an increased risk of MACE, as well as a history of AMI, PAOD, and hospitalization in ICU. Only smoking status (HR 4,3, CI1,7,8,9,10, p = 0.002), history of PAOD (HR 6.6, CI (1.5–29.7), p = 0.002), and hospitalization in ICU (HR 13.9, CI (4.8–40.4), p < 0.001) remained independently associated with the occurrence of MACE in multivariable analysis (Table 3), as shown in the forest plot (Fig. 3). The hazard ratio for patients hospitalized in ICU in the multivariable model is consistent with that of the IPTW approach.

Table 3 Factors associated with the occurrence of MACE in univariate and multivariate analysis. AF, atrial fibrillation; AMI, acute myocardial infarction; CKD, chronic kidney disease; CRD, chronic respiratory disease; HR, hazard ratio; ICU, intensive care unit; PAOD, peripheral artery occlusive disease; VTE, venous thromboembolism.
Full size table

Discussion

In our study, severity of COVID-19 was associated with an increased risk of MACE at 1 year, specifically stroke and acute heart failure. We observed that events primarily occurred within the first 3 months following infection, especially among the most severe patients (ICU), but also up to the end of the one-year follow-up period.

Our data align closely with those of the study conducted by Renda et al., which involved a prospective cohort of 296 patients hospitalized for SARS-CoV-2 infection24. Despite a mean follow-up duration of only around 6 ± 2 months, their study exhibits notable similarities with our results. The primary endpoint was death from all causes, with co-primary endpoints including the occurrence of MACEs such as AMI, stroke, pulmonary embolism, AHF, hospitalization for cardiovascular causes, or cardiovascular death. Results from their study revealed an all-cause mortality rate of 4.7%, with age identified as an independent predictor of mortality. Additionally, MACE occurred in 7.2% of patients, with AHF and AF emerging as significant predictors of MACE occurrence following adjustment for sex (adjusted hazard ratio 2.6, 95% confidence interval 1.05–6.52). However, it’s important to note that the follow-up duration in their study was almost half the duration of our study, with fewer patients, and most notably that authors did not compare COVID-19 patients according to disease severity. A study with a brief follow-up of 6 weeks compared 81 COVID-19 patients according to severity, focusing on echocardiography parameters such as left ventricular (e.g., left ventricular ejection fraction) and right ventricular function (e.g., tricuspid annular plane systolic excursion)25. However, it did not reveal any significant echocardiographic differences between groups, nor did it find disparities between patients in MW versus those in ICU settings. Nonetheless, the absence of evidence in comparison to our findings could be attributed to the small sample of patients and the very short follow-up duration.

Our results are also consistent with scientific literature data from large retrospective databases. Indeed, one study utilizing a large national United States health administrative database revealed that within the 4 months following SARS-CoV-2 infection, individuals who had COVID-19 experienced roughly a 2.5-fold increase in the hazard ratio for AHF or cardiac arrhythmia compared to those who had not been infected26. The same findings were also observed in England’s healthcare system database, revealing that patients hospitalized with COVID-19 were approximately three times more likely than uninfected individuals to experience major adverse cardiovascular event (including AHF, AMI, stroke, and arrhythmia) within eight months of their hospitalization27. Another recent study utilized data from US healthcare organizations comprising over four million individuals who had undergone a COVID-19 test and compared rates of cardiovascular outcomes between groups with and without COVID-19 infection14. Results showed that individuals with COVID-19 had an almost 1.8-fold higher risk of experiencing MACE (which includes AMI, ischemic stroke, hemorrhagic stroke, AHF, ventricular arrhythmia, and sudden cardiac death) in the subsequent 12 months after COVID-19. However, akin to studies relying on extensive administrative databases, misclassification bias and, in this instance, survivor bias cannot be entirely mitigated, since the follow-up was initiated 30 days after the COVID-19 test.

Several underlying pathophysiological mechanisms linking COVID-19 and the development of cardiovascular diseases are proposed28. Evidence suggests that SARS-CoV-2 can lead to infection and direct viral cytotoxic damage to the human myocardium, thereby significantly compromising cellular structure and function, ultimately leading to cell death29. Indeed, myocardial injury, observed in a proportion of patients who recovered from troponin-positive COVID-19 up to two months after hospitalization, was accompanied by evidence suggesting possible ongoing localized inflammation, as assessed by cardiovascular magnetic resonance30. Furthermore, SARS-CoV-2 infection can trigger a cytokine storm, resulting in the overproduction of various pro-inflammatory cytokines and chemokines, including IL-6, TNF-α, IL-1β, and IL-10. These cytokines can inflict significant damage to cardiomyocytes, impairing cell viability, structure, and function. Moreover, they can induce endothelial activation, thrombin generation, platelet aggregation, and sustained endothelial dysfunction (up to 6 months)31, promoting a shift towards a procoagulant state32. In a recent study, researchers demonstrated that adding SGLT2 inhibitors to plasma from COVID-19 patients mitigated the harmful effects on endothelial cells33. This suggests that SGLT2 therapy could be a promising approach to restoring vascular homeostasis and preventing cardiovascular complications in severe COVID-19 patients. This hypercoagulable state induced by COVID-19 may persist even after recovery, heightening the susceptibility to venous thromboembolic events like acute pulmonary embolism34 and stroke35. Mechanisms involving the dysregulation of the renin-angiotensin-aldosterone system and the kinin-kallikrein system have also been proposed36 and observed in COVID-1937. COVID-19-induced premature vascular senescence could explain an accelerated atherogenesis process leading to long-term MACE1. Finally, the damage to the respiratory system, which can be irreversible in some cases, is expected to compromise lung function and increase systemic hypoxic stress, leading to ischemic heart disease, including coronary artery disease, acute coronary syndrome, and ischemic stroke38. All these cardiovascular complications are now established components of Long COVID-1939. In severe cases of COVID-19, these complications, in addition to those associated with ICU stay, could further increase the overall risk of adverse outcomes40. Accordingly, a “call to action” was recently issued, given the large number of patients affected by the COVID-19 pandemic. This emphasizes that it is a historic opportunity to enhance our understanding of chronic diseases associated with large-scale infections—an unprecedented moment in human history that must not be overlooked41. Considering the ICU population more specifically, these cardiovascular complications might also be considered as part of the post-intensive care syndrome, which encompasses the acquired, multi-organ, long-lasting sequelae, observed in survivors of critically ill patients and has recently attracted attention in the medical community42,43. On another note, given the limited treatment options available during the first wave of the pandemic, most patients received homogeneous symptomatic care without corticosteroids, as robust evidence, such as the results of the RECOVERY study44, was not yet available.

Strengths and limitations

To our knowledge, this study, although retrospective and single-center, is one of the few to assess cardiovascular complications in COVID-19 survivors based on clinical observations collected at the time of hospitalization and then at 12 months post-discharge (verified by retrieval of medical reports) from the patient and/or the attending physician, rather than relying on large databases, which often suffer from missing data45,46. This approach allows for the evaluation of not only the occurrence of these complications but also the differences in damage according to the initial severity of COVID-19 infection. Moreover, the IPTW approach and the multivariable model provided similar estimations of the increased risk of MACE at one year for severe COVID-19 patients, thereby strengthening the robustness of our study. Our study has however some limitations that should be acknowledged. First, the retrospective single-center nature of the study limits its generalization. The study period was limited to the first wave of the COVID-19 pandemic due to the availability of complete and standardized data for this population. Subsequent waves were not included due to variations in treatment protocols and missing data for key variables. Despite the use of IPTW to balance known risk factors of MACE between the two groups, unmeasured confounding cannot be ruled out. The follow-up assessment was performed by telephone interviews, only the patients themselves or their general practitioners were interviewed to limit bias as much as possible. Besides, given a follow-up at one-years a memory bias cannot be ruled out. Moreover, vaccination status was not known, although this condition is highlighted in scientific literature as a protective factor against cardiovascular disease onset47,48. In addition, cardiological parameters were not monitored (transthoracic echocardiography, magnetic resonance imaging) to assess the frequency of underlying cardiac abnormalities that may be asymptomatic in our population. On another note, given the data available and now well established, it is possible that sepsis itself, especially in intensive care patients who are more severely ill than patients in conventional ward is associated with MACE, independently of the underlying cause of sepsis due to the effects of systemic inflammation, immune dysregulation and multi-organ failure observed in septic shock patients3,4. In addition, disease severity was not specifically and precisely quantified using a score such as the SOFA5, which would have enabled a more accurate comparison between the two groups. Finally, specific ICU related factors such as the use of vasoactive therapy, hemodynamic instability or even hypoxia may have contribute to myocardial injury or promoted cardiovascular complications. These parameters were not systematically collected in our cohort, and their possible contribution could not be captured due to the retrospective design of our study.

Conclusion

Severity of COVID-19 appears as an independent risk factor for major adverse cardiovascular events at one year, with a notable increase in stroke and AHF incidence observed among ICU patients compared to those in MW. The risks and burdens of cardiovascular disease were more pronounced among individuals who experienced severe COVID-19, necessitating hospitalization in the ICU, and might be considered part of post-intensive care syndrome. Care pathways for individuals who survived the acute episode of COVID-19 should prioritize attention to cardiovascular health and disease, particularly for those who experienced severe COVID-19. Extended follow-up periods may provide valuable insights into the long-term implications of COVID-19 on cardiovascular health.

Fig. 1
figure 1

Study flow-chart. ICU, intensive care unit; MW, Medical ward.

Full size image
Fig. 2
figure 2

MACE-free survival probability between ICU patients and MW patients. ICU, intensive care unit; MW, Medical ward.

Full size image
Fig. 3
figure 3

Forest plot of the multivariate analysis showing factors independently associated with the occurrence of MACE. Covariates included in the multivariate analysis represent preexisting medical history prior to index hospitalization for COVID-19. AF, atrial fibrillation; AMI, myocardial infarction; ICU, intensive care unit; MACE, major adverse cardiovascular event; PAOD, peripheral artery occlusive disease.

Full size image