The correspondence engages directly with a recent nationwide self-controlled case series study from the Netherlands, published in the same journal by Slurink and colleagues, which examined COVID-19 vaccination and short-term mortality risk. Self-controlled case series methodology is an epidemiological technique in which each individual serves as their own control, comparing mortality risk in defined windows immediately after vaccination with risk in later periods. Because the design compares a person with themselves, it automatically controls for all fixed characteristics of that person, such as genetics, socioeconomic background, and stable health conditions. This makes it a powerful tool for detecting short-term hazards that follow vaccination, and the Dutch study is one of the largest applications of the method to COVID-19 vaccines.

Yet Aarstad’s correspondence highlights why this strength does not translate into a definitive answer about long-term or population-level mortality effects. The self-controlled case series approach is designed to detect hazards concentrated in a short window after vaccination, such as the rare myocarditis signal observed in young men within days of mRNA vaccination. If a mortality effect were distributed diffusely over months or years, or if it interacted with subsequent infections, seasonal factors, or age-related decline, the method would have limited power to detect it. Moreover, the design conditions on the timing of vaccination itself, which means it cannot address the question of whether vaccinated and unvaccinated populations differ in their overall mortality trajectories over the years following the campaign.

Central to the correspondence is the problem of healthy vaccine bias, sometimes called the healthy vaccinee effect. People who receive vaccines are, on average, healthier than those who do not, either because illness prevents them from getting vaccinated or because they and their physicians are more health-conscious. In observational comparisons of vaccinated versus unvaccinated individuals, this bias tends to make vaccines appear protective against deaths from causes that have nothing to do with the infection the vaccine targets. When studies report that vaccinated people die less often from cardiovascular disease, cancer, or accidents, critics argue that this pattern is difficult to interpret as a genuine biological effect of vaccination and more plausibly reflects differences in underlying health status between the groups.

Aarstad also invokes the mortality displacement effect, a phenomenon long recognized in demography. Mortality displacement, or harvesting, occurs when a stressful event, such as a heat wave or an epidemic wave, advances the deaths of frail individuals who would otherwise have died weeks or months later. In the pandemic context, this means that mortality patterns in the aftermath of COVID-19 waves are shaped by the timing of earlier deaths, complicating any attempt to attribute excess mortality in a given period to a single exposure. If vaccination shifted mortality risks in complex ways across time, simple before-and-after or vaccinated-versus-unvaccinated comparisons could obscure rather than reveal the underlying dynamics.

The correspondence draws on Aarstad’s own prior work, including a study published in F1000Research in 2026 analyzing mortality involving and not involving COVID-19 among vaccinated and unvaccinated people in England between April 2021 and May 2023, and a second F1000Research study examining more than 3,000 US counties with a combined population of over 320 million people. The county-level analysis, according to its title, addresses why COVID-19 vaccination cannot be ruled out as an explanation for all-cause excess mortality in the pandemic’s aftermath. These population-level analyses form the empirical backdrop for the correspondence’s central claim: that the aggregate data patterns observed across large jurisdictions are not inconsistent with a vaccination contribution to excess mortality, even though they do not prove one.

It is important to situate this argument against the broader evidence base that the correspondence acknowledges implicitly through its cited references. A study by Nafilyan and colleagues published in Nature Communications in 2023 examined the risk of death following COVID-19 vaccination or a positive SARS-CoV-2 test in young people in England and found no increase in all-cause mortality in the weeks after vaccination, while deaths after infection were markedly elevated. Research on long-term outcomes after clinically suspected acute myocarditis, such as the nationwide pre-COVID-19 era study by Kim and colleagues published in PLoS One in 2023, provides context for evaluating the prognosis of the rare cardiac side effects associated with mRNA vaccines. These studies collectively support the conclusion that any direct vaccine-induced mortality risk, if present, is small and largely confined to identifiable short-term windows.

The correspondence’s contribution is therefore not a new data analysis but a challenge to how the existing evidence is interpreted. Aarstad’s framing, captured in the deliberately convoluted title, asks whether the community can rule out that vaccination did not increase all-cause mortality, a double negative that underscores the asymmetry between proving safety and failing to detect harm. In epidemiology, absence of evidence is not evidence of absence, and the correspondence argues that the tools used to date, however sophisticated, leave open a logical space in which a modest population-level mortality effect could exist without having been detected. This is a standard epistemological position in drug and vaccine safety surveillance, where pharmacovigilance systems are built precisely on the assumption that rare or delayed harms may emerge only after very large populations and long follow-up periods are studied.

The significance of the debate extends beyond academic argument. All-cause excess mortality has remained elevated in many countries in the years following the acute pandemic phase, and competing explanations include delayed effects of SARS-CoV-2 infections, disruptions to healthcare and screening programs, economic and psychological stressors, and, in the view of some commentators, vaccination itself. Because vaccination was administered to billions of people under emergency and mass-campaign conditions, even a very small per-person effect could translate into a non-trivial population burden. Establishing whether such an effect exists requires study designs capable of comparing vaccinated and unvaccinated populations over multi-year horizons while adequately handling the healthy vaccinee bias, mortality displacement, and the massive confounding introduced by differential infection rates.

Aarstad declares no relevant financial or non-financial competing interests and reports no funding for the manuscript, and the correspondence is single-authored. As a correspondence item, it did not analyze new data, and the journal’s ethics declarations note that no ethical approval, consent, or data analysis was applicable. The piece should be read as an invitation to methodological scrutiny rather than a conclusion: it argues that the burden of proof for ruling out a vaccination contribution to all-cause excess mortality has not yet been met, and that researchers should continue developing designs, such as long-horizon individual-level cohort studies with careful bias correction, capable of closing the remaining evidentiary gap. Whether future analyses will confirm the reassurance offered by short-term safety studies, or reveal patterns they could not detect, remains an empirical question that the scientific community is still working to answer.

Subject of Research: Whether COVID-19 vaccination can be ruled out as a contributor to all-cause excess mortality after the pandemic

Article Title: Can we rule out that COVID-19-vaccination did not increase all-cause mortality?

Article References: Aarstad, J. (2026). Can we rule out that COVID-19-vaccination did not increase all-cause mortality?. European Journal of Epidemiology. https://doi.org/10.1007/s10654-026-01443-w

Image Credits: AI Generated

DOI: 10.1007/s10654-026-01443-w

Keywords: COVID-19 vaccination, all-cause mortality, excess mortality, self-controlled case series, healthy vaccine bias, mortality displacement, epidemiology, vaccine safety, non-COVID-19 mortality, population-level study, European Journal of Epidemiology, Jarle Aarstad