Cardiac Risks of COVID-19 Vaccine vs Infection: Eman Toraih, MD, PhD

Questions about the cardiac safety of COVID-19 vaccination have persisted for years, but they are rarely asked alongside an equally important comparison: what happens to the heart after SARS-CoV-2 infection itself. Eman A. Toraih, MD, PhD, of the department of cardiovascular perfusion at SUNY Upstate Medical University, set out to answer both questions in the same population.

She is the corresponding author of a recent study published in Vaccine that compared myocarditis, pericarditis, and mortality across more than 4 million adolescents and young adults, categorized by documented infection, vaccination, both, or neither. In an interview with The American Journal of Managed Care^® (AJMC^®), Toraih discussed what prompted the study, what a population of this size made possible, and how clinicians should weigh a large relative risk reduction against a very small absolute risk.

This transcript was lightly edited for clarity.

AJMC: What prompted you and your colleagues to directly compare cardiac outcomes following SARS-CoV-2 infection with those following BNT162b2 vaccination in adolescents and young adults? What gap in the existing evidence were you hoping to address?

Toraih: The immediate prompt was the FDA memorandum of November 29, 2025, in which the director of the Center for Biologics Evaluation and Research stated that "at least 10 children had died after and because of COVID-19 vaccination." That conclusion rested on an unpublished review of 96 death reports submitted to the Vaccine Adverse Event Reporting System (VAERS) between 2021 and 2024, and it arrived with regulatory consequences attached, including proposed black-box warnings and stricter approval requirements.

Our concern was not the question. The question is fair, and parents ask it. Our concern was the method. The memorandum provided no ages, no pre-existing conditions, no autopsy findings, and no temporal detail, and it did not separate causation from temporal association. Former FDA commissioners, writing in the New England Journal of Medicine, described it as sweeping assertions about vaccine safety presented without supporting data, peer review, or transparent methodology. But the omission that mattered most to us as clinicians and data scientists was the absence of any comparator. There was no comparison with background mortality in this age group and no comparison with SARS-CoV-2 infection itself, which has caused more than 2000 documented pediatric deaths in the US.

That is the structural problem with passive surveillance. VAERS is a signal-detection system. It has no denominator and no control arm. Myocarditis occurs in adolescents regardless of exposure, at roughly 1.95 per 100,000 person-years, so a system that counts reports without counting the exposed population cannot distinguish an event the vaccine caused from an event that would have occurred that week anyway.

The research gap followed from that. The published evidence already pointed in a consistent direction: post-vaccination myocarditis at approximately 19.7 per million doses against 2.76 per 1000 infections, roughly a 140-fold difference. But those estimates come from separate studies, in different countries, using different definitions and different ways of finding cases. No large study had evaluated myocarditis, pericarditis, and mortality simultaneously across vaccinated, infected, and unexposed young people in the same network with the same codes, using methods that account for confounding and competing risks.

So, we built the study the regulatory debate required and did not have: 4 million Generation Z individuals aged 16 to 25 years, 4 mutually exclusive exposure groups, and a direct head-to-head comparison rather than 2 numbers compiled from different literature.

AJMC: Your study included more than 4 million individuals aged 16 to 25 years, making this a very large cohort. What did the size of the TriNetX population allow you to examine that previous studies may not have been able to capture?

Toraih: Size was what made the study possible. Design is what made it useful.

The main advantage of such a large population was that it gave us enough events to make a meaningful comparison. Myocarditis and pericarditis are rare in this age group. Myocarditis was recorded in about 1 per thousand individuals in the infection group over 6 months, and it was less common in every other group. Once we limited the population to ages 16 to 25, required sufficient prior medical history, applied our exclusions, and then matched the groups one to one, a much smaller dataset would not have provided enough events for reasonably precise estimates.

The large population also allowed us to separate individuals into 4 nonoverlapping groups: infection without recorded vaccination, BNT162b2 vaccination without recorded infection, infection followed by vaccination, and neither recorded exposure. This was important because we wanted to preserve the sequence of infection and vaccination rather than combine people with very different exposure histories.

For the main comparison between infection and vaccination, we were able to match 181,496 individuals in each group. This allowed us to compare the 2 exposures directly within the same US health care network, using the same eligibility criteria, outcome definitions, and follow-up period. Many previous studies focused on myocarditis during a short period after vaccination or examined infection and vaccination in separate populations. Our study compared the recorded risks after infection and vaccination in the same young population at several time points during the first 6 months, with additional exploratory follow-up extending to 60 months that we report as descriptive only.

The cohort size also allowed us to examine the results using several analytical approaches, including propensity-score matching, competing-risk analysis, and Cox regression. The overall pattern was consistent across these analyses, although that consistency does not eliminate residual confounding or establish causality.

I would not say this is the largest study of its kind. National registry studies in England and the Nordic countries were larger, and their data linkage is better than anything a hospital network can offer. Many earlier studies were also built for a different purpose, detecting safety signals soon after a dose, which is a different question from ours. What we added is a comparison built for 1 age group, with clearly separated exposure histories, followed for 6 months rather than a few weeks after vaccination.

Study timing was an advantage too. Our data run through September 2025, so the study covers several phases of the pandemic and not only the first rollout. But I want to be careful about what that means. We did not have variant or vaccine formulation at the individual level, and the calendar period entered the models only as a rough proxy. It gives the findings wider relevance in time. It is not a variant-specific or formulation-specific analysis.

Even so, a cohort of more than 4 million could not answer every question. We could not evaluate other vaccine products because outcome counts fell below the TriNetX reporting threshold. We could not separate the first dose from the second dose from the booster, or the monovalent from the bivalent formulation. After splitting by sex, myocarditis counts fell below the threshold as well. The hybrid-immunity group included only 10,677 individuals and produced unstable estimates. A large sample size improved precision, but it could not correct missing vaccination or infection records or the other limitations of observational electronic health record data.

AJMC: The infected/unvaccinated group had a substantially higher risk of myocarditis than the vaccinated/uninfected group, with the direct comparison showing an 85% lower relative risk among vaccinated individuals. How should clinicians and families interpret that relative difference alongside the very low absolute risk of myocarditis in both groups?

Toraih: Both numbers are true, and neither should travel alone.

In the matched analysis of 181,496 individuals per group, myocarditis was recorded in 69 people in the infected group. In the vaccinated group, the count was below the reporting threshold our data network permits us to disclose, which is itself informative about how uncommon the event was. The estimated relative risk was 0.15, an 85% lower risk, and the absolute difference was 0.0354%, or roughly 1 fewer recorded case per 2,800 individuals over 6 months.

Those 2 framings do different work, and clinicians need both. The absolute figure reassures a family that myocarditis was uncommon after either exposure or that most young people in either group had no cardiac event at all. The relative figure shows that among the rare events that did occur, most followed infection. Give only the relative reduction, and the individual risk sounds larger than it was. Give only the low absolute risk, and you erase a difference that was held in every analysis we ran.

I would add one caution about the number needed to vaccinate, because it is easy to misapply. It does not mean that vaccinating 2800 young people prevents one case of myocarditis. Our 2 groups were defined by what was recorded: vaccination with no recorded infection and infection with no recorded vaccination. Anyone vaccinated who was later infected was not in the vaccinated group by definition. So, this compares 2 observed populations after the fact. It is not a projection of what vaccinating an unselected group of teenagers would deliver, and the real-world figure would depend on how many still become infected.

2 further points about interpretation. The matched comparison balanced demographics, comorbidities, and medication use but not calendar period, and the 2 groups were indexed in somewhat different phases of the pandemic. We addressed that separately in the multivariable Cox models, which adjusted for calendar period and produced the same direction of effect. The size of any benefit is not a constant. It rises when transmission is high and when the circulating variant is more severe. Our study period ran from December 2020 through September 2025, so the counseling conversation should weigh local transmission, prior infection, and the individual's own health alongside the population figure.

In summary, our final practical message is that myocarditis was rare after both exposures, and, in our data, the recorded risk was lower after vaccination than after infection.