A paper whose message outran its evidence

BMJ Public Health has retracted a controversial 2024 paper that was widely used in vaccine-sceptical narratives about COVID-19. The article examined excess mortality estimates in 47 Western countries from 2020 to 2022 and reported roughly 3.1 million deaths above a pre-pandemic expected baseline.

That figure was real within the study’s chosen dataset and method, but it did not establish a single cause of those deaths. The central problem was not that excess mortality did not occur. It did. The problem was the paper’s framing of possible explanations, which placed substantial emphasis on COVID-19 vaccination and pandemic-control measures without conducting an analysis capable of determining their contribution to mortality.

In its retraction notice, the journal said the discussion of causes was imbalanced, insufficiently rigorous and included misinformation. That is a consequential editorial finding because the article circulated far beyond specialist research audiences, often with a much stronger claim attached to it: that it had demonstrated vaccines were responsible for millions of deaths. It did not make that causal demonstration.

What the study measured

Excess mortality compares the number of deaths observed during a period with the number expected from prior mortality patterns. It is a valuable public-health measure, particularly during a crisis when deaths directly attributed to an infection may be undercounted or classified differently between places.

The retracted paper used estimates published through the Our World in Data project and aggregated results across 47 countries. It reported excess deaths in each of the three years covered, including 2022, when many jurisdictions had relaxed the most stringent pandemic restrictions and vaccination was widespread.

Those descriptive results are not, by themselves, evidence that vaccination or containment policies caused excess deaths. Population mortality during the pandemic was shaped by overlapping factors, including waves of SARS-CoV-2 infection, disruption to health services, delayed diagnosis and treatment of other diseases, demographic change, influenza and other respiratory illnesses, heat events, and differences in death registration and baseline modelling.

A country-level comparison is especially poorly suited to assigning cause to an individual medical intervention. Vaccination uptake, infection exposure, age structure, access to care, timing of policy changes and the reliability of mortality data differ greatly across countries. When such factors move together, a temporal association can look persuasive while still failing to identify a causal relationship.

Why the discussion drew criticism

The authors did not state that they had conclusively proven vaccines caused the measured excess deaths. However, the paper’s introduction and discussion gave unusual prominence to suspected vaccine adverse events and the possible harms of non-pharmaceutical interventions. That framing made the paper highly vulnerable to a causal interpretation that its methods could not support.

The Princess Máxima Center, where three of the four authors were employed at the time, distanced itself from the publication shortly after it appeared in June 2024. Its subsequent scientific-integrity process found no evidence of plagiarism, data falsification or deliberate misconduct by the two researchers it investigated. But it concluded that the publication process had not met expected scientific standards and that principles of the Dutch research-integrity code had been breached.

The centre also said that authors involved had asked the journal to retract the article because of the risk of misinterpretation and a possible decline in vaccine willingness. That distinction matters. A retraction for misleading reasoning or inadequate scholarly standards is not equivalent to a finding that every underlying mortality number was fabricated. Nor does the absence of fabrication make an article’s causal discussion reliable.

The broader evidence on vaccine safety

COVID-19 vaccines, like all medical interventions, can have adverse effects. Rare serious events have been identified and are reflected in public-health guidance. For example, myocarditis and pericarditis have been observed rarely after vaccination, with risk patterns varying by product, age and sex. Safety surveillance is intended to identify precisely such risks, investigate them and adjust recommendations when necessary.

That is different from a claim that vaccination explains broad national or international mortality patterns. Assessing that proposition requires linked individual-level data, careful adjustment for age and pre-existing illness, comparison groups, information on infection history and timing, and transparent tests of alternative explanations. It cannot be resolved by observing that excess deaths and vaccine programmes occurred during overlapping years.

Large safety-monitoring systems combine reports from clinicians and patients with electronic health records, insurance claims, clinical research and population data. A report submitted to a passive reporting system is an alert for investigation, not proof that a vaccine caused the outcome. Similarly, an excess-death total is an important signal requiring investigation, not an automatic attribution to a particular policy or product.

A test of scientific communication

The episode also illustrates why journal peer review is not the end of scientific scrutiny. Papers can be questioned after publication, corrected, subjected to expressions of concern or retracted. Those mechanisms are imperfect and sometimes slow, but they are part of how science revises the public record when a study’s claims, methods or presentation do not withstand review.

For readers, the practical lesson is to separate three questions that are often merged in online debate:

  • Did excess mortality occur? Yes, it was documented across many countries during the pandemic period.
  • Is it legitimate to investigate all potential contributors, including the unintended effects of pandemic policy and rare vaccine harms? Yes.
  • Did this particular ecological analysis prove that COVID-19 vaccines caused millions of deaths? No.

The retraction does not close research into pandemic-era mortality. It reinforces the need for better-designed studies that distinguish correlation from causation and communicate uncertainty without inviting unsupported conclusions. In an area as politically charged as COVID-19, precision is not a cosmetic feature of research. It is essential to public understanding and public health.

Sources