COVID vaccination and heart events: read the VA study’s absolute risk

A study of U.S. veterans associated receipt of a 2024–2025 COVID-19 vaccine with fewer COVID-associated major adverse cardiovascular events over eight months. The result supports a specific observational comparison, not the earlier article's broad instructions for choosing a brand, spacing doses or restricting exercise.

The paper by Miao Cai, Yan Xie and Ziyad Al-Aly used Veterans Affairs electronic health records in a target-trial emulation. Calling the design an emulation matters: it is not the same as randomly assigning participants to receive a vaccine or a particular clinical schedule.

Identify the actual comparison

The study compared people receiving COVID and influenza vaccination on the same day with those receiving influenza vaccination alone. It included 1,039,659 participants; 349,085 received both vaccines and 690,574 received influenza vaccine alone.

That control group is not an entirely unvaccinated population. Renaming it that way would change the study's exposure comparison. The analysis also concerns a specified vaccine season and follow-up period, rather than every vaccine formulation and every subsequent year.

The authors estimated COVID-associated major adverse cardiovascular events, a composite including cardiovascular death, myocardial infarction, stroke and hospitalization for heart failure. It should not be expanded to every possible heart condition simply because those outcomes share an organ system.

Relative and absolute effects tell different parts of the story

The primary abstract reports estimated effectiveness of 37.7%, with a 95% confidence interval of 18.2% to 54.9%. It also reports a risk difference of about two events per 10,000 people over eight months, with an interval of 0.9 to 3.7.

Two per 10,000 is 0.02 percentage points. That is an arithmetic conversion of the reported absolute difference, not an additional clinical estimate. It is not a two-percentage-point reduction, and it is not interchangeable with the 37.7% relative figure.

Reporting both scales helps avoid two opposite mistakes: presenting a large relative percentage without the baseline scale, or dismissing a population-level association because an individual's absolute difference appears small. The decision-relevant interpretation still depends on the population and outcome actually studied.

The age profile limits a universal conclusion

Participants had a mean age of 70.1 years and 91.8% were men. The abstract reports statistically significant COVID-associated cardiovascular effectiveness only in those older than 75; the younger age strata did not have statistically significant estimates for that endpoint.

This does not prove an absence of benefit in every younger person. It does mean that this study cannot be cited as a precise demonstration of the same benefit in all ages. A non-significant subgroup result and a finding of zero effect are different statements.

The older article skipped that boundary while giving recommendations to young men, pregnant people, athletes and people with prior vaccine-related myocarditis. Those recommendations require additional, current clinical evidence and guidance. They have been removed from this study report.

A composite result is not a complete clinical policy

The finding does not directly compare vaccine brands, determine an ideal interval after infection or test an eight-week dose gap. It also does not evaluate a twenty-four- to forty-eight-hour exercise restriction, pain-medication advice or a strategy for people taking anticoagulants.

Those prescriptions appeared in the earlier text despite not being the intervention tested here. Removing them preserves the study's contribution without allowing a news article to become an unsupported treatment schedule.

The same applies to causal mechanisms. A lower observed event risk does not, by itself, identify how much came from fewer infections, less severe illness or another pathway. A plausible explanation should not be relabelled as a mechanism measured by this comparison.

Observational evidence remains useful with its boundaries intact

Target-trial emulation can organize an observational comparison around a clear question, but it does not make every potential difference between groups disappear by definition. The reported estimate and uncertainty should travel with the study design, rather than being converted into a guarantee for a particular individual.

This revision relies on the primary abstract available through the authors' institutional publication record and the contemporaneous reporting. It does not claim a new analysis of the underlying health records or an independent assessment of every modelling assumption.

The supported result is meaningful and bounded: an association between the specified vaccination exposure and fewer specified cardiovascular events in this predominantly older male VA cohort. Presenting the absolute difference, age limits and study design makes that result more informative than a generic instruction that everyone should choose and time a shot in the same way.

Sources

Primary abstract and publication record: https://profiles.wustl.edu/en/publications/2024-2025-covid-19-vaccine-and-major-adverse-cardiovascular-event/

Paper DOI: https://doi.org/10.1001/jamainternmed.2026.1929

Original reporting: https://arstechnica.com/health/2026/06/covid-vaccines-still-protect-against-heart-problems-large-study-finds/