The exchange centres on an original study by Slurink and her co-authors, published in the same journal, titled COVID-19 vaccination and short-term mortality risk: a nationwide self-controlled case series study in The Netherlands. That study used a self-controlled case series design, an epidemiological method that has gained prominence in vaccine safety research because of its unusual statistical properties. Rather than comparing vaccinated individuals with unvaccinated individuals—a comparison vulnerable to confounding by health status, age, and healthcare-seeking behaviour—the self-controlled case series compares each person against themselves, contrasting periods shortly after vaccination with periods further removed from it. Because every individual serves as their own control, stable characteristics such as underlying health, socioeconomic position, and genetics are, in principle, cancelled out of the analysis.

The method is not without its well-known vulnerabilities, however, and much of the debate between the Utrecht team and their critic turns on exactly these subtleties. The most important assumption of a self-controlled case series is that the event under study—here, death—must not change the probability of subsequent exposure. In vaccine research this assumption is complicated by the healthy-vaccinee effect and its mirror image, the frailty effect: people may delay vaccination while acutely ill, so that the days immediately before vaccination are artificially low-risk and the days immediately after appear artificially dangerous by comparison. Sophisticated implementations of the design therefore adjust for such time-varying confounding or restrict the observation window in ways designed to mitigate the bias. How faithfully a given analysis satisfies the method’s assumptions is precisely the kind of question that letters to the editor exist to probe.

Aarstad’s letter, published under the title Can we rule out that COVID-19 vaccination did increase all-cause mortality?, engages with the Utrecht study at the level of interpretation rather than raw computation. The question posed in its title reflects a broader argument that Aarstad has advanced in a series of papers elsewhere, including two 2026 articles in F1000Research. One of those articles examined mortality involving and not involving COVID-19 among vaccinated versus unvaccinated people in England between April 2021 and May 2023; the other analysed population-level data from more than 3,000 US counties covering over 320 million people, arguing that COVID-19 vaccination cannot be ruled out as an explanation for all-cause excess mortality in the aftermath of the pandemic. The county-level ecological approach and the individual-level self-controlled approach represent two very different methodological traditions, and their divergent conclusions illustrate why the choice of design matters as much as the data themselves.

Ecological studies of the kind Aarstad employed aggregate data across populations—in his US analysis, whole counties—and look for correlations between vaccination coverage and mortality outcomes. Such designs are useful for generating hypotheses and for examining population-level patterns, but they are notoriously susceptible to the ecological fallacy: an association observed at the group level need not hold at the individual level. Regions with high vaccination coverage may differ systematically from regions with low coverage in age structure, income, healthcare access, baseline COVID-19 mortality, and a host of other factors, any of which could drive an apparent correlation. Individual-level designs, by contrast, link exposure and outcome within the same person, which is why national registries in countries such as the Netherlands and England have become the preferred substrate for rigorous vaccine safety analyses.

In defending their original findings, the Utrecht authors were able to situate their work within a wider body of international evidence. Among the studies cited in the exchange is a 2023 analysis published in Nature Communications by Nafilyan and colleagues, which examined the risk of death following COVID-19 vaccination or a positive SARS-CoV-2 test in young people in England. That study, drawing on national linked data, found no increase in overall mortality in the period after vaccination, while identifying the expected small elevation in cardiac-related deaths among young women after a first dose of mRNA vaccines—an excess that was outweighed by the cardiac mortality risk associated with SARS-CoV-2 infection itself. Findings of this shape are typical of the modern vaccine safety literature: genuine, rare adverse effects are detectable and acknowledged, but they are small relative to the risks posed by the infection the vaccines prevent.

The myocarditis question looms large in any discussion of vaccine-associated mortality, and the response engages with it through a further cited study. Kim and colleagues, writing in PLoS ONE in 2023, reported ten-year survival outcomes after clinically suspected acute myocarditis in adults, using a nationwide study from the pre-COVID-19 era. The relevance of such work is methodological as much as clinical: to assess whether vaccine-associated myocarditis could plausibly produce a detectable signal in all-cause mortality statistics, one needs to know the long-term prognosis of the condition. If survival after clinically suspected myocarditis is generally favourable over a decade of follow-up, then even the modest excess incidence of myocarditis observed after mRNA vaccination would be expected to contribute very little to population-level death counts, particularly in the short term.

The structure of the exchange itself is worth understanding, because it reflects how scientific self-correction is supposed to operate. A letter to the editor is a formal critique of a published paper; a response is the original authors’ opportunity to answer that critique point by point, with the journal acting as arbiter. The Utrecht response was received on 3 September 2026, accepted on 23 September, and published on 12 October, indicating a rapid but peer-reviewed turnaround. The authors declare that no additional funding was received for the preparation of the response, and the competing-interest disclosures are unusually detailed for a correspondence item: co-author Marc Bonten notes regular interaction with pharmaceutical partners without personal fees, and the institution has received study funding from several vaccine manufacturers, while Miriam Sturkenboom heads a department that conducts studies for the European Medicines Agency and the European Commission under the ENCePP code of conduct. Such disclosures matter because transparency about funding and industry contact is central to credibility in vaccine research, whatever one’s conclusions.

For the general reader, the significance of this exchange lies less in any single statistical estimate than in what it reveals about the state of the evidence. A nationwide self-controlled case series in the Netherlands, a linked-data analysis of young people in England, a decade-long myocarditis prognosis study, and county-level ecological analyses of the United States all bear on the same question, and they do not all point in the same direction—or, more precisely, they weigh different biases and different populations. The Utrecht team’s position, maintained through the response, is that their individual-level analysis of Dutch national data supports the conclusion that COVID-19 vaccination was not associated with increased short-term mortality risk, and that the concerns raised in the letter do not overturn that finding. Aarstad’s position is that the possibility of a vaccination contribution to all-cause excess mortality cannot be excluded. Both positions can coexist in the literature because they rest on different designs, different assumptions, and different standards of proof.

What happens next is the quiet work of replication and synthesis. Correspondence exchanges rarely settle disputes on their own; their value is in forcing authors to clarify assumptions, release methods to scrutiny, and sharpen the interpretation of their estimates. The full text of the response, available through the European Journal of Epidemiology, provides the detailed technical argumentation for readers equipped to evaluate it. Meanwhile, the underlying question—whether the vaccination campaigns left any detectable footprint in all-cause mortality statistics—will continue to be examined as more registry data accumulate and as methods capable of handling the healthy-vaccinee effect and other subtleties mature. For now, the exchange stands as a compact demonstration of epidemiology doing what it is designed to do: subjecting consequential claims to adversarial scrutiny, in public, with the methods on display for anyone willing to read closely.

Subject of Research: COVID-19 vaccination and short-term all-cause mortality risk assessed with a nationwide self-controlled case series study

Article Title: Response to the letter to the editor by J. Aarstad

Article References: Slurink, I. A. L., Bonten, M. J. M., Sturkenboom, M. C. J. M., & Bruijning-Verhagen, P. C. J. L. (2026). Response to the letter to the editor by J. Aarstad. European Journal of Epidemiology. https://doi.org/10.1007/s10654-026-01478-z

Image Credits: AI Generated

DOI: 10.1007/s10654-026-01478-z

Keywords: COVID-19 vaccination, all-cause mortality, self-controlled case series, vaccine safety, excess mortality, epidemiology, myocarditis, letter to the editor, Netherlands, European Journal of Epidemiology, healthy-vaccinee effect, pharmacovigilance