COVID Deaths: When Scientific Questions Become Too Dangerous to Ask
Academic freedom is easiest to defend when nobody cares about the answer. The real test comes when researchers investigate questions that governments, universities, medical institutions or scientific establishments would rather regard as settled. COVID-19 provided perhaps the greatest such test of recent times, and the controversy surrounding a now-retracted paper in BMJ Public Health raises uncomfortable questions about how contemporary science deals with research that challenges an established consensus.
The paper, by Saskia Mostert and colleagues, examined excess mortality across 47 Western countries between 2020 and 2022. Its basic finding was striking but not especially controversial: more than three million excess deaths occurred during those years, and elevated excess mortality continued after the first pandemic year. The authors argued that the causes deserved investigation.
That is where things became controversial. The researchers did not demonstrate that COVID vaccines caused those excess deaths. Their data could not establish such a causal relationship, and claims that the paper proved millions had died because of vaccination went far beyond what the study showed. The authors instead identified several possible contributors requiring further investigation, including COVID infection itself, indirect consequences of containment measures and vaccination programs. That distinction matters enormously. Science depends upon distinguishing a hypothesis from a demonstrated causal conclusion.
But it creates another question. If researchers observe persistent excess mortality after an enormous medical and social intervention, why should examining all plausible contributing factors be regarded as scientifically improper? A scientific response should surely be straightforward: formulate hypotheses, obtain better data and test them.
If vaccination made no significant contribution to excess mortality, sufficiently rigorous research should demonstrate that. If lockdowns contributed through delayed diagnoses, interrupted medical treatment, psychological deterioration or other mechanisms, investigate those possibilities. If COVID infection itself explains most of the excess, demonstrate that. If demographic changes, ageing populations, cardiovascular disease, cancer, drug deaths or some combination of factors accounts for it, establish that. That is what science is supposed to do.
Instead, the paper rapidly became embroiled in controversy. BMJ Public Health issued an expression of concern in June 2024, warning that the research had been misrepresented as establishing a direct causal relationship between COVID vaccination and mortality. The journal also investigated concerns about the quality and messaging of the work.
There were substantive criticisms. Critics argued that the paper's discussion gave disproportionate attention to vaccination as a possible explanation despite the ecological mortality data being incapable of establishing individual-level vaccine causation. An institutional investigation also raised concerns about selection and interpretation of evidence.
Those criticisms deserve consideration. Academic freedom cannot mean immunity from criticism, peer review, correction or even retraction. Bad research does not become good research merely because its conclusions are unpopular, just as establishment research does not become sound merely because prestigious institutions endorse it. But the affair raises a broader problem that cannot be dismissed merely by pointing to methodological criticism.
Why do some scientific questions appear to generate an institutional reaction extending far beyond ordinary scientific disagreement? Scientists publish weak papers constantly. Journals contain questionable assumptions, inadequate datasets, exaggerated conclusions and analyses that subsequently fail replication. That is one reason science is supposed to be self-correcting. Researchers challenge previous work, obtain better evidence and gradually determine which explanations survive.
Retraction traditionally serves a more serious purpose. Fabricated data, plagiarism, fundamental errors, unethical research or findings rendered unreliable by serious methodological defects are obvious candidates. Retraction should not become an instrument for removing papers merely because their interpretation becomes politically inconvenient or because journalists and activists misuse their conclusions.
That distinction is crucial in the Mostert controversy. A journal cannot reasonably hold researchers entirely responsible for every claim subsequently made about their work on social media or in political debate. If a paper says that several possible causes of excess mortality require investigation, someone else claiming that the paper "proves vaccines killed millions" does not magically transform the original statistical analysis into fraud.
Correct the misrepresentation. Criticise the methodology. Publish rebuttals. Produce superior research. That is how science advances. The danger arises when scientific institutions begin treating certain hypotheses differently because of their political implications. Once researchers learn that asking one question advances a career while asking another may trigger investigations, reputational attacks or institutional difficulties, formal censorship becomes unnecessary. Researchers censor themselves.
This is one of the oldest mechanisms operating within intellectual institutions. Nobody needs to circulate a memorandum announcing that a particular conclusion is forbidden. Young academics observe what happens to senior researchers who challenge prevailing assumptions. Grant committees establish priorities. Editors decide which subjects appear respectable. Universities worry about reputation. Researchers quickly learn which questions produce funding and promotion and which produce trouble.
The resulting conformity can be perfectly sincere. Human beings are extraordinarily good at convincing themselves that conclusions beneficial to their careers also happen to be intellectually correct.
COVID intensified this problem because science and government policy became unusually intertwined. Governments were spending enormous sums, imposing unprecedented restrictions and encouraging or requiring populations to accept newly developed vaccines. Public-health agencies were simultaneously scientific authorities and participants in policies whose success scientists were being asked to evaluate. That created an obvious institutional problem. Organisations rarely enjoy investigating themselves.
Even if for the sake of argument we grant the effectiveness of the COVID vaccination, it does not establish that every vaccine policy was optimal, that every adverse event was adequately investigated, that every age group had an identical risk-benefit calculation or that vaccination could have made no contribution whatsoever to mortality in particular circumstances. Those are empirical questions.
The phrase "the science is settled" is therefore particularly dangerous when applied to complicated and evolving subjects. Science is never strengthened by declaring questions illegitimate. Newtonian mechanics survived centuries of investigation not because critics were silenced but because it repeatedly worked within its domain. Einstein displaced part of the Newtonian picture because observations eventually demanded something more. Scientific authority derives from surviving criticism, not preventing it. The same principle should govern medical science.
If we vaccine sceptics are wrong about excess mortality, demonstrate it. Break mortality figures down by age, sex, cause of death, vaccination status, number of doses, time since vaccination, previous infection and relevant medical conditions. Make sufficiently detailed anonymised data available to independent researchers. Encourage competing analyses. Let researchers who believe vaccines contributed significantly to mortality attempt to prove their case and let other researchers attempt to demolish it. That would strengthen confidence in vaccination far more effectively than institutional defensiveness ever could. We dare the establishment!
The deeper problem is that retraction has acquired an unavoidable political meaning in intensely contested fields. Whatever the editors' intentions, removing a paper from the scientific literature can be interpreted publicly as declaring not merely that its methodology was inadequate but that its questions were illegitimate. That perception becomes especially powerful when the paper challenges policies defended by governments and major scientific institutions.
If journals aggressively police speculative claims that challenge establishment positions, they must apply the same standards to papers supporting those positions. Papers exaggerating the benefits of lockdowns, masks, vaccines or other interventions should receive precisely the same scrutiny as papers exaggerating their harms. Scientific standards must operate symmetrically. Otherwise peer review risks becoming worldview review.
Academic freedom does not guarantee academics the right to be correct. It guarantees something much more important: the freedom to be wrong while asking legitimate questions, subject to evidence, criticism and the normal processes of scholarly correction. That freedom is especially important when the subject concerns powerful institutions, like Big Pharma.
There is nothing inherently scientific about believing governments. There is nothing inherently scientific about distrusting them either. Science begins when both impulses are put aside and evidence is allowed to decide.
More than three million excess deaths across the Western countries examined by Mostert and her colleagues represented an enormous human event. Determining what caused those deaths should not belong to the political Left, Right, vaccine advocates or vaccine sceptics. It should be an empirical problem pursued without fear or favour. But, it is not.
