Abstract
SARS-CoV-2 infection and BNT162b2 mRNA vaccination carry distinct cardiovascular risk profiles, yet direct comparative evidence across all immunological exposure groups and both sexes remains limited. Using the TriNetX Research Network (December 2020–December 2024), we stratified 30.3 million individuals into four mutually exclusive cohorts: uninfected/unvaccinated controls (G1), infected/unvaccinated (G2), vaccinated-only (G3), and hybrid immunity (G4). Fifty prespecified cardiovascular, cerebrovascular, and mortality outcomes were evaluated across four temporal windows (0–3, 3–6, 6–9, and >9 months) with analyses stratified by biological sex. SARS-CoV-2 infection was associated with 3- to 5-fold increases in cardiovascular events during the acute phase, including myocarditis (males: RR 4.44; females: RR 5.59) and all-cause mortality (males: RR 4.53), with risks persisting beyond nine months. BNT162b2 vaccination conferred 65-76% reductions in major adverse cardiovascular events (0–3 months). Post-infection vaccination (hybrid immunity) provided an additional 36–38% MACE reduction; males exhibited late pericarditis elevation beyond nine months. Completing the two-dose primary series maximally reduced mortality (by 77%) and myocarditis (by 62%) versus single dosing. In this US cohort, SARS-CoV-2 infection confers substantially greater and more sustained cardiovascular risk than BNT162b2 vaccination across all comparisons and both sexes, consistent with a favorable cardiovascular risk-benefit profile for vaccination.
Subject terms: Cardiology, Diseases, Immunology, Medical research, Microbiology
Introduction
The COVID-19 pandemic infected more than 770 million individuals, resulted in over 7 million deaths worldwide, and remains a global health threat1. Beyond its acute respiratory manifestations, SARS-CoV-2 infection increases the risk of cardiovascular morbidity, with survivors facing heightened risk of major adverse cardiovascular events (MACE), including myocardial injury, arrhythmias, thrombosis, and stroke, sequelae recognized as major contributors to both short- and long-term mortality2–5. These events reflect convergent pathophysiologic mechanisms: direct viral invasion of cardiomyocytes via ACE2 receptors6, systemic cytokine-driven inflammatory vascular injury7, and the induction of a persistent hypercoagulable state8. The introduction of mRNA vaccines in December 2020 altered the pandemic’s trajectory, demonstrating approximately 95% efficacy against symptomatic COVID-19 of any severity and >98% efficacy against severe disease9,10. However, rare cardiovascular adverse events following vaccination, particularly myocarditis in young males, have complicated risk-benefit assessments and contributed to vaccine hesitancy and rethinking vaccination strategies11–13.
This has ignited critical debates, necessitating a precise and data-driven reassessment of the absolute and relative cardiovascular risks associated with the primary modes of immunity: natural infection, vaccination alone, and hybrid immunity (defined as infection followed by subsequent vaccination). Despite the urgency of this clinical question, the existing literature is constrained by several key limitations. Prior studies have frequently focused on infection or vaccination in isolation, employed short follow-up periods, utilized heterogeneous outcome definitions, and rarely included robust, large-scale cohorts with hybrid immunity13–16. Crucially, the temporal evolution of risk and the influence of sex-specific factors or specific vaccine dosing strategies on long-term cardiovascular outcomes remain incompletely characterized. Addressing these limitations could further inform vaccination recommendations.
To provide the necessary high-resolution evidence, we conducted a large real-world, sex-stratified comparative analysis to date, leveraging a federated health network encompassing 30 million individuals. Our primary objective was to systematically quantify the incidence of cardiovascular, cerebrovascular, thrombotic, metabolic, and mortality outcomes across all clinically relevant exposure groups, unvaccinated infection, vaccination alone, and hybrid immunity, compared with infection- and vaccine-naïve controls. Temporal trajectories of risk were delineated over four post-exposure intervals (0–3, 3–6, 6–9, and >9 months). Secondary analyses examined the impact of vaccine dose strategies, including primary, monovalent, and bivalent series, on incremental risk reduction. Our study outcomes can inform vaccination policy, risk communication, and long-term patient care in the post-pandemic era.
Results
Study population
Between December 2020 and December 2024, 30,323,720 patients met the inclusion criteria (Fig. 1). Sex-specific group sizes and mean follow-up durations (22–26.5 months) for each pairwise comparison are summarized in Supplementary Tables 3–10.
Fig. 1. Study flowchart, cohort selection, and outcomes.

The flowchart summarizes TriNetX Research Network cohort identification from July 15 to 30, 2025, including eligibility and exclusion criteria, classification into control, COVID-only, vaccine-only, and hybrid exposure groups, planned pairwise comparisons stratified by sex and exposure-time window, and the 50 prespecified cardiovascular, cerebrovascular, lipid, biomarker, healthcare utilization, and mortality outcomes. Created using BioRender (biorender.com).
Infection effect (COVID-only vs Control)
SARS-CoV-2 infection in 2 million unvaccinated people led to pronounced acute and subacute increases in cardiovascular and cerebrovascular risk compared to 27 million Control subjects. Myocarditis risk rose 4.44-fold in males and 5.59-fold in females; myocardial infarction nearly doubled; and acute coronary disease and stroke risks tripled. Pulmonary embolism increased fourfold in males and threefold in females. All-cause mortality was substantially higher: 2.7% in infected males vs 0.6% controls; 1.5% in infected females vs 0.4% controls. Most risks progressively across follow-up intervals. Myocarditis, mortality, and some adverse outcomes remained elevated through >9 months (Fig. 2 and Supplementary Tables 3 and 4).
Fig. 2. Sex-stratified risk of selected outcomes: COVID-only vs Controls.

Relative risk (RR) with 95% confidence intervals comparing infected, unvaccinated individuals (Group 2) with infection- and vaccine-naive controls (Group 1) across four post-infection intervals: 0–3 months (acute phase), 3–6 months (subacute phase), 6–9 months, and >9 months. The red dashed line indicates no effect (RR = 1). Blue circles and lines represent females; red circles and lines represent males. Error bars denote 95% confidence intervals.
Vaccine effect in uninfected (Vaccine-only vs Control)
In the 1.36 million Vaccine-only group, MACE was 76% lower in males and 65% lower in females compared to Controls, with sustained reductions in myocardial infarction, stroke, arrhythmias, and mortality. Benefits persisted for 3–6 months and returned to baseline after nine months, with no excess cardiovascular or inflammatory risk (Fig. 3 and Supplementary Tables 5 and 6).
Fig. 3. Sex-stratified risk of selected outcomes: vaccine-only vs Controls.

Relative risk (RR) with 95% confidence intervals comparing vaccinated-only individuals (Group 3) with infection- and vaccine-naive controls (Group 1) across four post-vaccination intervals: 0–3, 3–6, 6–9, and >9 months. The red dashed line indicates no effect (RR = 1). Blue circles and lines represent females; red circles and lines represent males. Error bars denote 95% confidence intervals.
Vaccine effect in infected (hybrid vs COVID-only)
Among 155,715 with hybrid immunity, post-infection vaccination reduced MACE by 36% in males and 38% in females versus COVID-only patients. Most acute phase benefits waned by 6–9 months; however, in females, the reduction in cerebrovascular risk persisted beyond that of cardiac outcomes. Event counts for some endpoints were low, warranting cautious interpretation (Fig. 4 and Supplementary Tables 7 and 8).
Fig. 4. Sex-stratified risk of selected outcomes: hybrid vs COVID-only.

Relative risk (RR) with 95% confidence intervals comparing individuals with hybrid immunity (infected then subsequently vaccinated, Group 4) with infected-only individuals (Group 2) across four post-vaccination intervals: 0–3, 3–6, 6–9, and >9 months. The red dashed line indicates no effect (RR = 1). Blue circles and lines represent females; red circles and lines represent males. Error bars denote 95% confidence intervals.
Vaccination versus infection (vaccine-only vs COVID-only)
Directly comparing 1.4 million Vaccine-only and 2 million COVID-only subjects, vaccination conferred dramatically superior safety: myocarditis risk reduced by over 90%, pulmonary embolism by up to 90%, and mortality by around 90%. Benefits persisted for nine months in females but waned slightly in males (Fig. 5 and Supplementary Tables 9 and 10).
Fig. 5. Sex-stratified risk of selected outcomes: vaccine-only vs COVID-only.

Relative risk (RR) with 95% confidence intervals comparing vaccinated-only individuals (Group 3) with infected-only individuals (Group 2) across four post-exposure intervals: 0–3, 3–6, 6–9, and >9 months. The red dashed line indicates no effect (RR = 1). Blue circles and lines represent females; red circles and lines represent males. Error bars denote 95% confidence intervals.
Metabolic, biomarker, and healthcare utilization outcomes
SARS-CoV-2 infection was associated with marked metabolic derangement: hypertriglyceridemia risk increased by 33% in males and 54% in females, alongside a 23% increase in low HDL in females. Cardiac stress biomarkers such as troponin I and BNP were elevated 2.6-3.1 times over controls. Infected and unvaccinated individuals had substantially higher intensive care utilization, with mechanical ventilation required 3.5–3.9 times more often and ECMO-VV usage 13.6 times higher than in controls. In contrast, vaccination led to 60–69% reductions in dyslipidemia and 67–79% lower cardiac biomarker levels and was associated with more than 80% fewer ICU admissions (Supplementary Tables 3–10).
Sex-stratified cardiovascular and cardiovascular risk patterns
Following infection, females had higher risks for infarction (RR 2.12 vs 1.92 in males), myocarditis (5.59 vs 4.44), arrhythmias (1.63 vs 1.45), and lipid derangement (1.54 vs 1.33); while males suffered more thrombotic events (4.12 vs 3.15) and mortality (4.53 vs 4.04). Vaccination provided a stronger short-term benefit for males, but females maintained more durable cerebrovascular and mortality reductions. In the hybrid group, males exhibited late pericarditis (RR 1.88 vs 0.83) while females retained cerebrovascular protection >9 months (stroke RR 0.59 vs 0.68) (Supplementary Tables 3–10).
Vaccine dosing effects
Two-dose primary BNT162b2 series reduced mortality by 77%, myocarditis by 62%, pericarditis by 58%, myocardial infarction by 53%, and stroke by 48% versus a single dose. Additional monovalent doses diminished mortality by 27%, hospitalization by 42%; bivalent doses added a further 23% reduction in severe disease, yet incremental vascular risk reductions plateaued after the primary series (Supplementary Table 11). These dose-comparison estimates are susceptible to immortal-time bias, because recipients of a second or additional dose necessarily survived event-free to the later administration date; the benefit attributed to series completion is therefore likely overstated and is presented as hypothesis-generating.
Temporal trend analysis
As depicted in Supplementary Table 12, weighted meta-regression identified statistically significant temporal trends in 39 of 48 outcome-comparison-sex combinations (81%). In the COVID-only vs Control comparison, infection-associated cardiovascular risk showed a significant reduction of relative risk toward the null across all six outcomes in both sexes (p < 0.05 for all), most pronounced for mortality (p < 0.0001) and pulmonary embolism (p < 0.001). In the vaccine-only vs Control comparison, vaccine-conferred protection significantly diminished over time for five of six outcomes; myocarditis was the only outcome without significant waning (males p = 0.17, females p = 0.59), indicating a more durable protective effect against inflammatory cardiac injury. In the vaccine-only vs COVID-only comparison, the relative safety advantage of vaccination narrowed significantly for most outcomes, though RRs remained consistently below 1.0 throughout all follow-up intervals. In the hybrid vs COVID-only comparison, significant waning of post-infection vaccination benefit was observed for mortality and pulmonary embolism in both sexes; myocarditis and stroke showed no significant temporal trend.
Discussion
In this large, US-based federated cohort of over 30 million individuals, SARS-CoV-2 infection was associated with substantial and, for most outcomes, sustained increases in cardiovascular and cerebrovascular risk, including myocarditis, infarction, thromboembolism, arrhythmias, and stroke, while BNT162b2 vaccination was associated with lower cardiovascular event rates than naïve controls during the primary follow-up window, with no excess cardiovascular risk detected within that period. The data identify infection as the primary vascular hazard, while vaccination offers significant benefit. Hybrid immunity provided further short-term protection, with late sex-specific patterns.
Natural SARS-CoV-2 infection was associated with three to fivefold increases in cardiovascular and cerebrovascular events, including myocarditis, myocardial infarction, pulmonary embolism, arrhythmia, and stroke, during the initial three months, with risks remaining elevated through nine months for most outcomes. All-cause mortality in males exceeded 4.5-fold in the acute phase and remained elevated for more than nine months (RR 1.11). Stroke, however, was no longer significantly elevated beyond nine months in either sex. These findings are supported by mechanistic evidence from autopsy and registry studies linking acute infection to persistent endothelial injury, chronic immune activation, and thrombo-inflammatory dysregulation17,18, and point to COVID-19 infection acting as a systemic vascular disorder that extends beyond the acute respiratory phase19. Notably, the cardiovascular risk attributed to infection was greater than the risk conferred by major chronic conditions such as diabetes (RR = 1.8) or hypertension (RR = 1.6)20. Our cohort’s excess mortality and adverse event rates are at least as high as those reported in other large U.S. studies (mortality RR: 3.9–4.5)21,22 and surpass recent European estimates (approximately 2.8-fold)23, likely reflecting differences in virus variants (Delta and Omicron versus ancestral strains), comorbidity burden (42% obesity prevalence in U.S. vs 28% in Europe), longer follow-up duration than prior reports, and evolving clinical management strategies. These findings suggest a prolonged period of post-infection cardiovascular surveillance in US populations to assess for potential adverse cardiovascular outcomes maybe warranted24,25.
In contrast, BNT162b2 vaccination in previously uninfected individuals led to 65–76% reductions in MACE and 61–66% lower mortality, with parallel reductions in myocardial infarction, inflammatory heart disease, and ischemic stroke. Protective benefits were most evident during 0–3 months and declined by 6–9 months. Beyond 9 months, all-cause mortality in the vaccine-only group exceeded that of naïve controls (males RR 1.22, p < 0.001; females RR 1.40, p < 0.001), likely reflecting a waning of vaccine effect and return to baseline risk of the Vaccine-only group. Without propensity score matching and assessing for potential confounding variables differentiating the Vaccine-only from the naïve controls this remains only a proposal. Rare vaccination-associated myocarditis was observed far less frequently than post-infection cases (a 4.2-fold lower risk in males, 2.3-fold lower in females), consistent with a favorable benefit-risk profile for BNT162b2 vaccination in this population, supporting its role as a cardiovascular risk-reduction strategy26.
Mechanistically, infection triggers direct viral invasion of cardiomyocytes via ACE2 receptors6, provoking cytokine-mediated inflammation with interleukin-6 levels 10- to 100-fold higher than those after vaccination7 and promoting cardiac-reactive autoantibodies27. Vaccination, by contrast, elicits controlled immunological responses with limited systemic inflammation, no evidence of sustained endothelial injury, and strong formation of immune memory even after breakthrough infection, mitigating the cascade that drives cardiovascular complications28. Although spike protein has been detected in rare myocarditis cases following mRNA vaccination, myocarditis has also been observed following vaccination with non-mRNA platforms, suggesting that spike protein may not be the lone causal mechanism29.
Dose–response analysis demonstrated that completing the initial two-dose series was associated with the largest observed cardiovascular risk reductions (mortality reduction by 77%, myocarditis by 62%, myocardial infarction by 53% vs. single dose), whereas additional doses showed less incremental lowering of adverse cardiovascular outcomes, but sustained risk reductions associated with the primary vaccination series. This plateau effect diverges from the antibody-centric response but parallels influenza vaccination, suggesting cardiovascular benefit is primarily linked to moderating inflammation rather than preventing all infection30–32. Meta-analytic evidence across multiple viruses, including influenza and SARS-CoV-2, further supports the concept that vaccination-driven reduction in vascular risk depends on both individual susceptibility and immune priming dynamics, rather than simple dose escalation33. Thus, strategies for ongoing COVID-19 vaccination may increasingly resemble annual influenza immunization34, with additional doses recommended for older adults, those with comorbidities, and other high-risk groups.
Hybrid group, vaccination after infection, was also associated with further MACE and stroke reductions, consistent with additive benefit due to boosting pre-existing immunity and endothelial stabilization35,36. The observed delayed pericarditis risk in males beyond six months requires further study and may reflect confounding, as individuals receiving vaccination after infection may be those with more severe initial illness following natural infection or underlying comorbidities. These findings highlight the need for extended clinical surveillance in this subgroup. This estimate must be interpreted cautiously: because the hybrid cohort was indexed at vaccination and could be constituted only from patients who survived event-free from infection to vaccination, the comparison is subject to immortal-time bias that inflates apparent early protection. The attenuation and reversal of the mortality relative risk over follow-up, from below 1.0 in the acute window to above 1.0 beyond nine months, is consistent with such a time-zero artifact rather than a durable causal benefit, and these findings are therefore hypothesis-generating.
Sex-specific findings are especially relevant: females displayed higher relative rates of MACE, arrhythmia, dyslipidemia, and stroke post-infection, while males exhibited higher mortality and thromboembolic risk. Vaccination produced greater initial benefit for males and more durable effects for females, consistent with distinct hormonal and immunological factors, such as estrogen-mediated suppression of systemic inflammation37,38 and testosterone-driven pro-inflammatory bias39. The delayed pericarditis seen in males with hybrid immunity may reflect cumulative inflammatory burden and tissue-resident memory T-cell responses40,41.
Our data are consistent with COVID-19 vaccination as a primary tool for cardiovascular risk reduction. For patients with established cardiovascular disease or risk factors, clinical decisions regarding additional vaccine doses should be made in consultation with current guidelines and individual patient context, informed by the known kinetics of waning immunity, variant evolution, and the need for vaccine reformulation, and trends in SARS-CoV-2 transmission and disease, while surveillance for late cardiac complications, particularly pericarditis in males with prior infection, should be extended. Personalized, sex-aware risk communication and follow-up protocols are warranted.
Strengths of our study include its scale, temporal and sex stratification, and comprehensive multiorgan outcomes in a real-world setting. Several limitations should be noted. First, this study is observational and did not use propensity score matching, so confounding by indication cannot be excluded; all findings should be interpreted as associations rather than causal effects. Second, two comparisons, post-infection vaccination (Hybrid vs COVID-only) and vaccine dose-response (two-dose vs single-dose; multiple vs single booster), are also subject to immortal-time bias, a long-recognized pitfall in pharmacoepidemiology42, because eligibility for the more-exposed group required event-free survival to a later index date that was not shared by the comparison group. This structurally favors the more-exposed group during the acute window; the corresponding estimates should be regarded as hypothesis-generating until confirmed with landmark or time-varying-exposure analyses. Similarly, the large early risk reductions seen in the Vaccine-only versus Control comparison are too pronounced to be explained by a direct vaccine effect alone and likely arise in part because healthier, more health-seeking people are more likely to be vaccinated; conversely, the much higher mortality after infection may have limited the accrual of later non-fatal events in that group.
The study period (2020–2024) encompasses ancestral, Delta, and Omicron variant eras; because variant-specific cardiovascular risk profiles differ and were not adjusted for, risk estimates reflect a mixture of era-specific effects that cannot be disaggregated in this design. Analyses are restricted to BNT162b2; findings may not be generalizable to other vaccine platforms. For outcomes with fewer than ten events in a group, the TriNetX platform replaces the true count with ten to protect patient privacy. This means relative risks for uncommon outcomes may be slightly overestimated and should be interpreted with caution. Simultaneous stratification by both age and sex was not feasible due to small outcome counts in subgroups; age-related effect modification, therefore, remains uncharacterized. The requirement for a minimum of two healthcare encounters, applied uniformly across all four cohorts, may modestly over-represent individuals with higher healthcare utilization with comorbidities or complications; findings should be interpreted in this context. To isolate the cardiovascular effects of infection and vaccination, individuals with comorbidities known to independently influence cardiovascular outcomes (including recent influenza infection or vaccination, active malignancy, significant obesity, chronic viral infections, and substance use disorders) were excluded; findings therefore apply to individuals free of these concurrent risk factors and should not be generalized to populations with these conditions. The cohort closed in December 2024 to ensure adequate follow-up; applicability to current settings should account for evolving variants, updated vaccine formulations, and changes in population immunity since the study period. The TriNetX network draws on voluntary institutional participation rather than probability sampling; this does not preclude valid inference within the represented population, but limits generalizability to health-system-engaged individuals in the United States.
Future work should clarify sex-dependent immune and endothelial mechanisms, variant- and vaccine-specific risks, and the optimal timing between infection and vaccination to minimize recurrence. The mechanisms underlying myocarditis across vaccine types, and the delayed pericarditis seen in males with hybrid immunity, require targeted investigation to clarify immunological causes and improve risk prediction. These sex-specific differences translate to actionable clinical recommendations: extended cardiac surveillance in males following hybrid immunity, sex-specific evaluation protocols after SARS-CoV-2 infection, and consideration of sex-stratified booster interval scheduling. Longitudinal studies, including imaging and biomarkers, are needed to refine prognosis and determine if subclinical vascular injury leads to long-term atherosclerosis or waning vaccine protection.
In this large US-based real-world cohort, SARS-CoV-2 infection was associated with substantially elevated cardiovascular risk across most outcomes and both sexes, while BNT162b2 vaccination was associated with lower cardiovascular event rates during the primary follow-up window. Protective effects declined beyond 6 months, and all-cause mortality in the vaccinated group exceeded that of naïve controls after nine months, consistent with waning immunity and return to baseline risk, which could have been higher in the vaccine recipients. These findings support COVID-19 vaccination as a cardiovascular risk-reduction strategy in health-system-engaged populations and inform vaccination recommendations and utilization in clinical practice.
Methods
Study design and data source
We conducted a retrospective cohort study using the TriNetX Research Network (accessed July 2025), a federated health platform of de-identified electronic medical records from 108 voluntarily participating United States healthcare organizations, comprising academic medical centers, community hospitals, and multi-specialty health systems, representing over 150 million patients. The network provides longitudinal data, including demographics, diagnoses, procedures, medications, laboratory values, and clinical observations. Institutional participation is voluntary, and patient data are contributed through existing EHR workflows at participating sites; the effective sampling frame, therefore, consists of individuals who accessed care at participating US institutions, representing a health-system-engaged population rather than a probability sample of the general US population. Ethical approval was not required as only de-identified data were analyzed.
Study population and cohort definitions
Eligible individuals had documented biological sex, self-reported race, and a minimum of two healthcare encounters between December 2020 and December 2024. SARS-CoV-2 infection was defined by positive PCR/antigen testing or ICD-10 codes (U07·1, U07·2). Vaccination status was ascertained from electronic health records, state immunization registries, and pharmacy databases. We excluded patients with clinical trial enrollment; chromosomal abnormalities; major congenital heart disease; baseline known cardiovascular or cerebrovascular disease; active/recent malignancy (<12 months); recent substance use disorders (<3 months); body mass index ≥40 kg/m2; Eastern Cooperative Oncology Group (ECOG) performance status ≥2; recent live vaccine receipt (<3 months); influenza infection or vaccination (<12 months); chronic hepatitis B or C; HIV; tuberculosis; or prior non-Pfizer COVID-19 vaccination (Supplementary Table 1).
Patients were stratified into four mutually exclusive cohorts: (G1: Control)—naïve, no documented infection or vaccination; (G2: COVID-only)—documented infection without subsequent vaccination; (G3: Vaccine-only)—BNT162b2 (Pfizer-BioNTech) vaccination without prior or subsequent documented infection; (G4: Hybrid)—documented SARS-CoV-2 infection followed by subsequent vaccination ≥14 days post-recovery. The index date was defined as the first positive test (G2), first vaccine dose (G3 and G4), or a healthcare encounter date within the matched study window (G1).
Because the Hybrid cohort was indexed at the first vaccine dose (administered ≥14 days after recovery, and frequently weeks to months after infection) whereas the COVID-only comparator was indexed at the infection date, follow-up for these two groups did not begin at a common time-zero; individuals could enter the Hybrid cohort only by surviving event-free from infection to vaccination, an interval of immortal person-time expected to bias the Hybrid-versus-COVID-only comparison toward apparent early benefit. The same structure applies to the dose-response analysis, in which receipt of a second or subsequent dose required event-free survival to a later index date. These two comparisons are therefore interpreted below as susceptible to immortal-time bias.
Outcome definitions
Fifty predefined clinical outcomes were evaluated in eight categories: cardiovascular events, cardiac arrhythmias, inflammatory heart disease, heart failure and cardiomyopathy, thrombotic events, cerebrovascular events, lipid abnormalities, cardiac biomarkers, healthcare utilization, and all-cause mortality. Only new-onset outcomes occurring after the index date were considered; those with prevalent endpoints were excluded from time-specific analyses. Detailed outcome definitions and codes are presented in Supplementary Table 2. Risk analyses were conducted for four prespecified periods following index events: 0–3 months (acute), 3–6 months (subacute), 6–9 months (intermediate), and >9 months (persistent effects).
Statistical analysis
Four pairwise comparisons were conducted: (1) COVID-only vs Control; (2) Vaccine-only vs Control; (3) Hybrid vs COVID-only; and (4) Vaccine-only vs COVID-only. All analyses were stratified by biological sex. Propensity score matching was not undertaken due to the large cohort size (>30 million individuals). Simultaneous stratification by age and sex was precluded by small numbers of outcomes per subgroup, which prevented reliable age-stratified analyses. Each post-exposure time window (0–3, 3–6, 6–9, and >9 months) was analyzed as an independent query with the outcome observation period restricted to that interval. Only new-onset events within each window were captured; events occurring in a prior interval were not carried forward, preventing cumulative counting. The cohort denominator remained constant across all four intervals. Minor variation in cohort sizes between pairwise analyses reflected differences in query timing and data synchronization inherent to the federated platform; the minimum cohort size across all intervals was used for consistency.
Minor variation in cohort sizes between pairwise analyses reflected differences in query timing and data synchronization inherent to the federated platform; the minimum cohort size across all intervals was used for consistency.
For outcomes with fewer than ten events in a cohort, TriNetX privacy policy reports a value of “10” instead of the true count, which may inflate event rates and relative risks for these rare outcomes; zero events are accurately represented as “0”. Affected cells are flagged in supplementary tables and should be interpreted with caution.
Two-sided chi-square tests were used to compare event proportions between groups. Relative risks (RRs) were calculated as the ratio of event rates between comparison groups, with 95% confidence intervals (CIs) and p-values generated within the TriNetX integrated analytics environment; the precise computational implementation is proprietary and not publicly documented. Risk differences (RDs) were calculated as pExposed − pComparator, with 95% CIs by the Wald method: RD ± 1.96 × √[pA(1 − pA)/nA + pB (1 − pB)/nB]. To formally assess temporal trends, weighted meta-regression of log(RR) was performed against time interval midpoints (1.5, 4.5, 7.5, and 12.0 months), using inverse-variance weights (1/SE2, where SE = [ln(upper CI) − ln(lower CI)]/(2 × 1.96)); a significant regression slope (p < 0.05) defined a significant temporal trend. Results are presented as event counts, percentages, RRs, RDs, and 95% CIs. Statistical significance was defined as p < 0.05. No formal correction for multiplicity was applied across the large number of comparisons; significant associations are therefore exploratory and should be regarded as hypothesis-generating.
Secondary analyses evaluated vaccine dosing regimens, comparing outcomes among recipients of a complete two-dose primary series versus a single dose, and among those receiving one versus multiple booster doses (monovalent or bivalent). Data were accessed and analyzed between July 1–30, 2025; visualizations were produced with RStudio (version 2025.05.1).
Artificial intelligence tools
During manuscript preparation, large language models were utilized as editorial assistance tools to reduce word count in compliance with journal guidelines. The following AI technologies were employed: Claude Sonnet 4.5 (Anthropic, Inc.), ChatGPT-5 (OpenAI), and Perplexity Pro (auto-selected optimal model). These tools were used exclusively for text condensation, rephrasing, and searching recent relevant articles. The authors critically reviewed, edited, and validated all AI-assisted content. No AI tools were used for data analysis, interpretation, study design, or generation of original scientific content. The authors assume complete responsibility for the accuracy, integrity, and originality of all submitted material.
Supplementary information
Acknowledgements
We acknowledge the healthcare organizations contributing de-identified electronic health record data to the TriNetX Research Network and the patients whose data made this research possible. No funding was received for this research.
Author contributions
Conceptualization: E.A.T., H.A., and S.T.; Methodology: E.A.T. and D.B.; Software: E.A.T. and D.B.; Investigation: E.A.T., D.B., and M.H.H.; Formal analysis: E.A.T. and D.B.; Validation: all authors; Writing—Original Draft: E.A.T. and M.H.H.; Writing—Review & Editing: All authors; Supervision: H.A. and S.T.; Project Administration: H.A. and S.T.
Data availability
Data access is available through the TriNetX research network under appropriate data use agreements and institutional approvals. All aggregate outcome data generated and analyzed in this study, including event counts, proportions, relative risks, risk differences, and 95% confidence intervals for all 50 outcomes across all four cohort comparisons, both sexes, and four temporal windows, were exported from the TriNetX integrated analytics environment and are reported in full in Supplementary Tables 3–12. Cohort definitions and outcome codes are provided in Supplementary Tables 1 and 2. Individual-level patient data cannot be shared publicly due to HIPAA regulations and TriNetX institutional data governance agreements. Researchers seeking to conduct equivalent analyses may apply for institutional access to the TriNetX Research Network at https://trinetx.com. The complete exported aggregate dataset from the TriNetX platform is retained by the corresponding author and will be made available to the journal upon reasonable request and subject to applicable data governance requirements.
Code availability
All statistical analyses were performed within the TriNetX integrated analytics environment, a proprietary federated platform; no custom analytical code was developed. Data visualizations were produced using RStudio (version 2025.05.1); the visualization scripts are available from the corresponding author upon reasonable request. TriNetX platform documentation is available at https://trinetx.com.
Competing interests
S.J.T. was an investigator in the phase 3 trial of the BNT162b2 COVID-19 vaccine and has participated on data safety monitoring boards, independent data monitoring committees, and in consultative and advisory capacities for several industry sponsors of COVID-19 vaccines. He has been compensated for his time. The remaining authors (E.A.T., D.B., M.H.H., and H.A.) declare no competing interests.
Footnotes
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
These authors contributed equally: Hani Aiash, Stephen J. Thomas.
Contributor Information
Eman A. Toraih, Email: toraihe@upstate.edu
Stephen J. Thomas, Email: ThomStep@upstate.edu
Supplementary information
The online version contains supplementary material available at https://doi.org/10.1038/s41541-026-01528-3.
References
- 1.Organization W. H. World Health Statistics 2025. WHO. (https://data.who.int/dashboards/covid19/cases?n=o).
- 2.Leite Barbosa, N. et al. Prevalence and characteristics of respiratory and cardiovascular sequelae in post-COVID-19 syndromes: a scoping review. Expert Rev. Respir. Med.19, 935–953 (2025). (In eng). [DOI] [PubMed] [Google Scholar]
- 3.Denegri, A. et al. Cardiovascular complications of COVID-19 disease: a narrative review. Diseases 10.3390/diseases13080252 (2025). [DOI] [PMC free article] [PubMed]
- 4.Wang, W., Wang, C. Y., Wang, S. I. & Wei, J. C. Long-term cardiovascular outcomes in COVID-19 survivors among non-vaccinated population: a retrospective cohort study from the TriNetX US collaborative networks. EClinicalMedicine53, 101619 (2022). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Zuin, M., Mazzitelli, M., Rigatelli, G., Bilato, C. & Cattelan, A. M. Risk of ischemic stroke in patients recovered from COVID-19 infection: a systematic review and meta-analysis. Eur. Stroke J.8, 915–922 (2023). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Lindner, D. et al. Association of cardiac infection with SARS-CoV-2 in confirmed COVID-19 autopsy cases. JAMA Cardiol.5, 1281–1285 (2020). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hu, B., Huang, S. & Yin, L. The cytokine storm and COVID-19. J. Med. Virol.93, 250–256 (2021). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ackermann, M. et al. Pulmonary vascular endothelialitis, thrombosis, and angiogenesis in Covid-19. N. Engl. J. Med.383, 120–128 (2020). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Thomas, S. J. et al. Safety and efficacy of the BNT162b2 mRNA Covid-19 vaccine through 6 months. N. Engl. J. Med.385, 1761–1773 (2021). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Polack, F. P. et al. Safety and efficacy of the BNT162b2 mRNA Covid-19 vaccine. N. Engl. J. Med.383, 2603–2615 (2020). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Oster, M. E. et al. Myocarditis cases reported after mRNA-based COVID-19 vaccination in the US from December 2020 to August 2021. J. Am. Med. Assoc.327, 331–340 (2022). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Das, B. B., Moskowitz, W. B., Taylor, M. B. & Palmer, A. Myocarditis and pericarditis following mRNA COVID-19 vaccination: what do we know so far? Children (Basel) 10.3390/children8070607 (2021). [DOI] [PMC free article] [PubMed]
- 13.Marshall, M. et al. Symptomatic acute myocarditis in 7 adolescents after Pfizer-BioNTech COVID-19 vaccination. Pediatrics 10.1542/peds.2021-052478 (2021). [DOI] [PubMed]
- 14.Karimi, R., Norozirad, M., Esmaeili, F., Mansourian, M. & Marateb, H. R. COVID-19 Vaccination and cardiovascular events: a systematic review and Bayesian multivariate meta-analysis of preventive benefits and risks. Int. J. Prev. Med. 10.4103/ijpvm.ijpvm_260_24 (2025). [DOI] [PMC free article] [PubMed]
- 15.Meslé, M. M. I. et al. Estimated number of lives directly saved by COVID-19 vaccination programmes in the WHO European Region from December, 2020, to March, 2023: a retrospective surveillance study. Lancet Respir. Med.12, 714–727 (2024). (In eng). [DOI] [PubMed] [Google Scholar]
- 16.Tobaiqy, M., MacLure, K., Elkout, H. & Stewart, D. Thrombotic adverse events reported for Moderna, Pfizer and Oxford-AstraZeneca COVID-19 vaccines: comparison of occurrence and clinical outcomes in the EudraVigilance database. Vaccines (Basel) 10.3390/vaccines9111326 (2021). [DOI] [PMC free article] [PubMed]
- 17.Raman, B., Bluemke, D. A., Lüscher, T. F. & Neubauer, S. Long COVID: post-acute sequelae of COVID-19 with a cardiovascular focus. Eur. Heart J.43, 1157–1172 (2022). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kuchler, T. et al. Persistent endothelial dysfunction in post-COVID-19 syndrome and its associations with symptom severity and chronic inflammation. Angiogenesis26, 547–563 (2023). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mostafa, R. H. & Moustafa, A. Beyond acute infection: molecular mechanisms underpinning cardiovascular complications in long COVID. Front. Cardiovasc. Med.11, 1268571 (2024). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yusuf, S. et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet364, 937–952 (2004). (In eng). [DOI] [PubMed] [Google Scholar]
- 21.Al-Aly, Z., Xie, Y. & Bowe, B. High-dimensional characterization of post-acute sequelae of COVID-19. Nature594, 259–264 (2021). (In eng). [DOI] [PubMed] [Google Scholar]
- 22.Xie, Y., Xu, E., Bowe, B. & Al-Aly, Z. Long-term cardiovascular outcomes of COVID-19. Nat. Med.28, 583–590 (2022). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Raisi-Estabragh, Z. et al. Cardiovascular disease and mortality sequelae of COVID-19 in the UK Biobank. Heart109, 119–126 (2022). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Aljadah, M. et al. Clinical implications of COVID-19-related endothelial dysfunction. JACC Adv.3, 101070 (2024). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Khan, F. et al. Long-term risk of stroke after transient ischemic attack or minor stroke: a systematic review and meta-analysis. J. Am. Med. Assoc.333, 1508–1519 (2025). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Patone, M. et al. Risks of myocarditis, pericarditis, and cardiac arrhythmias associated with COVID-19 vaccination or SARS-CoV-2 infection. Nat. Med.28, 410–422 (2022). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Halushka, M. K. & Vander Heide, R. S. Myocarditis is rare in COVID-19 autopsies: cardiovascular findings across 277 postmortem examinations. Cardiovasc. Pathol.50, 107300 (2021). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Chaudhary, N., Weissman, D. & Whitehead, K. A. mRNA vaccines for infectious diseases: principles, delivery and clinical translation. Nat. Rev. Drug Discov.20, 817–838 (2021). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Yonker, L. M. et al. Circulating spike protein detected in post-COVID-19 mRNA vaccine myocarditis. Circulation147, 867–876 (2023). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wohl, A. & Leibowitz, R. Protection by a Fourth Dose of BNT162b2 against Omicron in Israel. N. Engl. J. Med.386, 2441 (2022). (In eng). [DOI] [PubMed] [Google Scholar]
- 31.Goel, R. R. et al. mRNA vaccines induce durable immune memory to SARS-CoV-2 and variants of concern. Science374, abm0829 (2021). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Turner, J. S. et al. SARS-CoV-2 mRNA vaccines induce persistent human germinal centre responses. Nature596, 109–113 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Kawai, K., Muhere, C. F., Lemos, E. V. & Francis, J. M. Viral Infections and risk of cardiovascular disease: systematic review and meta-analysis. J. Am. Heart Assoc.14, e042670 (2025). [DOI] [PMC free article] [PubMed]
- 34.Udell, J. A. et al. Association between influenza vaccination and cardiovascular outcomes in high-risk patients: a meta-analysis. J. Am. Med. Assoc.310, 1711–1720 (2013). (In eng). [DOI] [PubMed] [Google Scholar]
- 35.Mroueh, A. et al. COVID-19 promotes endothelial dysfunction and thrombogenicity: role of proinflammatory cytokines/SGLT2 prooxidant pathway. J. Thrombosis Haemost.22, 286–299 (2024). [DOI] [PubMed] [Google Scholar]
- 36.Ćwilichowska-Puślecka, N. et al. Understanding the long-term interplay of SARS-CoV-2 immune and inflammatory responses with proteases in COVID-19 recovery: a longitudinal study. Front. Immunol.16, 1517933 (2025). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Gebhard, C., Regitz-Zagrosek, V., Neuhauser, H. K., Morgan, R. & Klein, S. L. Impact of sex and gender on COVID-19 outcomes in Europe. Biol. Sex. Differ.11, 29 (2020). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Straub, R. H. The complex role of estrogens in inflammation. Endocr. Rev.28, 521–574 (2007). (In eng). [DOI] [PubMed] [Google Scholar]
- 39.Gubbels Bupp, M. R. & Jorgensen, T. N. Androgen-induced immunosuppression. Front Immunol.9, 794 (2018). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Fink, A. L., Engle, K., Ursin, R. L., Tang, W. Y. & Klein, S. L. Biological sex affects vaccine efficacy and protection against influenza in mice. Proc. Natl. Acad. Sci. USA115, 12477–12482 (2018). (In eng). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Klein, S. L. & Flanagan, K. L. Sex differences in immune responses. Nat. Rev. Immunol.16, 626–638 (2016). (In eng). [DOI] [PubMed] [Google Scholar]
- 42.Suissa, S. Immortal time bias in pharmaco-epidemiology. Am. J. Epidemiol.167, 492–499 (2008). (In eng). [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data access is available through the TriNetX research network under appropriate data use agreements and institutional approvals. All aggregate outcome data generated and analyzed in this study, including event counts, proportions, relative risks, risk differences, and 95% confidence intervals for all 50 outcomes across all four cohort comparisons, both sexes, and four temporal windows, were exported from the TriNetX integrated analytics environment and are reported in full in Supplementary Tables 3–12. Cohort definitions and outcome codes are provided in Supplementary Tables 1 and 2. Individual-level patient data cannot be shared publicly due to HIPAA regulations and TriNetX institutional data governance agreements. Researchers seeking to conduct equivalent analyses may apply for institutional access to the TriNetX Research Network at https://trinetx.com. The complete exported aggregate dataset from the TriNetX platform is retained by the corresponding author and will be made available to the journal upon reasonable request and subject to applicable data governance requirements.
All statistical analyses were performed within the TriNetX integrated analytics environment, a proprietary federated platform; no custom analytical code was developed. Data visualizations were produced using RStudio (version 2025.05.1); the visualization scripts are available from the corresponding author upon reasonable request. TriNetX platform documentation is available at https://trinetx.com.
