Vaccinst studies Restore trust in science

In 1942 the German physicist Max Planck wrote in his autobiography:

“A new scientific truth does not triumph because opponents become convinced and admit they are wrong, but because the opponents gradually die out and a new generation grows up that is familiar with the new insights.” ¹Quote Max Planck: https://citaten.net/quotes/max_planck/30925/citaat-een-nieuwe-wetenschappelijke-waarheid-zegeviert-niet-doordat.html

Twenty years later, the American philosopher of science Thomas Kuhn developed this idea further The Structure of Scientific Revolutions (1962). According to him, science usually proceeds within a widely shared paradigm, in which researchers rely on the same assumptions and methods. Anomalous findings or anomalies have long been ignored or incorporated into the existing framework. Only when tensions mount and anomalies pile up can a crisis arise that ultimately leads to a paradigm shift: a fundamental revision of the way reality is understood.²Wikipedia about these Phases: https://nl.wikipedia.org/wiki/De_structuur_van_wetenschappelijke_revoluties#:~:text=37%3A144-,Fases,-%5Bbewerken%20%7C

Kuhn emphasized that such a revolution is not only a rational process, but also a social and psychological process. Positions of power, reputations and vested interests ensure that dissenting voices are downplayed or even disqualified. Here Kuhn directly echoes Planck: new paradigms rarely win because opponents are convinced, but because they eventually disappear and a new generation grows up with different assumptions.

We also see this mechanism in the current discussions surrounding COVID-19 and vaccine studies. Critical analyzes that point out methodological biases — such as the healthy vaccinee effect and immortal time bias — should be regarded as valuable contributions to correction and refinement in a self-critical scientific climate. Instead, such criticisms are often dismissed as anti-scientific and anti-institutional (what if extreme right is seen).

There are plenty of examples. Top virologist Marion Koopmans warned about this on X yesterday “the dangers of antivax rhetoric” , in response to Robert F. Kennedy Jr.'s plans)³Tweet from Koopmans: https://x.com/MarionKoopmans/status/1964024089892278662.

Science in motion

Since the beginning of the vaccination campaigns, studies have been published worldwide reporting extremely high effectiveness and safety of the COVID-19 vaccines. These figures were often convincing: 90 to 98% protection against infection, hospitalization and death. But upon closer inspection, these analyzes show a systematic pattern: bias in favor of the vaccines.

Two mechanisms play a key role in this:

• Healthy Vaccinee Effect (HVE): Healthier people are more likely to be vaccinated, terminal people are not vaccinated at all. As a result, vaccinated groups artificially show significantly lower mortality, also due to causes that have nothing to do with COVID.
• Overige Biases, o.m. Time Related Bias (TRB): Time Related Bias is a type of selection bias that causes events (or deaths) to be missed (starting measurements too late and/or stopping too early) or people end up in the wrong research group (misclassification bias) because of the period chosen for the study. It may even have to do with administrative problems, especially in the Netherlands. More about this from Herman Steigstra⁴First week after vaccination = unvaccinated https://steig.nl/2024/11/99-gevaccineerde-overlijdens-in-eerste-week-ongevaccineerd/. However, selectively choosing and shifting periods and time zones is a recurring and recognizable phenomenon in research worldwide.

A well-known - simple - example is only granting the 'vaccinated' status two weeks after the second injection. The result is that in the first weeks after injection, vaccinated people often appear to not only have a lower risk of COVID mortality, but even of all-cause mortality (ACM) because the deaths are still counted as 'unvaccinated' (misclassification) and the vaccinated do relatively much better. It also cannot be due to COVID immunity: the disease was not fatal enough to explain such sharp declines in overall mortality (including from non-Covid causes). These observed effects are therefore not proof of safety or additional health benefits, but signals of bias artifacts.

If vaccinated people in a study show lower mortality from all causes (including non-Covid), this indicates a statistical artifact.

In initial publications (NEJM, UKHSA, RIVM/CBS, NIVEL, ISS, CDC) these biases were largely ignored, at most mentioned in the small print. Only in reanalyses and critical reflections (Chemaitelly, Shahar, Pescara, Doshi) did it become clear how strongly HVE and TRB actually distort the results, and that both effectiveness and safety were often much lower than initially suggested.

Below are a number of flagship studies in a heat map (click on the cells for a brief characterization). We start with the studies on which WHO and EMA based their assessment, and then follow the line through national reports and reanalyses.

Textual overview:

🇺🇸 VS – Polack et al., NEJM 2020 (Pfizer phase 3 trial)

• Claim: 95% effectiveness against symptomatic COVID-19 in the trial population.
• Signal: Trial population was selected (few elderly, no fragile groups, short follow-up). Mortality from other causes was hardly measured → no insight into all-cause mortality.
• Bias: Healthy volunteer effect through strict inclusion criteria; possible TRB by definitions of “fully vaccinated” (≥7 days after 2nd dose).
• Quote: “A two-dose regimen … conferred 95% protection against Covid-19 (95% CI, 90.3–97.6).”

🇬🇧 Voysey et al., Lancet 2021 (AstraZeneca trial)

• Claim: ~70% effectiveness from 14 days after the second dose.
• Signal: Only events after day 14 were counted → early infections automatically attributed to the control group.
• Bias: Time Related Bias built-in; selection of relatively healthier subjects → healthy volunteer bias.
• Quote: “Vaccine efficacy was 70.4% … from 14 days after the second dose.”

🇺🇸 Tenforde et al., MMWR 2021 (VISION network)

• Claim: 94% effectiveness against hospitalization in ≥65 year olds.
• Signal: Vaccinees had fewer comorbidities than controls; early events fell into the 'unvaccinated' group.
• Bias: Healthy vaccinee effect + Time Related Bias (vaccine status pas na ≥14 days).
• Quote: “Effectiveness of full vaccination against hospitalization among adults ≥65 years was 94%.”

🇬🇧 Andrews et al., NEJM 2022 (UKHSA boosters)

• Claim: Booster gave VE of ~90% against hospitalization with Delta, ~75% against Omicron.
• Signal: A booster would increase the protection against hospitalization with Omicron to 88%. However, negative VEs also occurred in the same analyses, meaning a higher risk of hospitalization among vaccinated people. Only the negative effects were attributed to bias by the authors.
• Bias: Healthy vaccinee effect mentioned for negative VE; possible TRB by onset definition (≥14 days). No correction applied for positive VE estimates.
• Quote: “Vaccine effectiveness against hospitalization with Omicron was 88% … after a booster dose.”

🇮🇱 Dagan et al., NEJM 2021

• Claim: VE >90% against infection, hospitalization and death.
• Signal: Also non-COVID death lower in vaccinated people, biologically impossible.
• Bias: HVE + TRB → healthy people were vaccinated earlier, and mortality shortly after vaccination was assigned to the “unvaccinated” group.
• Quote: “Estimated vaccine effectiveness for documented infection was 92%, for hospitalization 87%, and for severe disease 92%.”

🇬🇧 UKHSA Vaccine Surveillance Reports

• Claim: VE against mortality often >95% in 2021.
• Signal: It was only in later reports that called HVE as an explanation for negative VE in older age groups.
• Bias: HVE + TRB remained largely unnamed or uncorrected.
• Quote: “Vaccine effectiveness against death following the Delta variant was 95% after 2 doses of Pfizer and 96% after 2 doses of AstraZeneca.” (VSR Week 42, 2021)

🇳🇱 NIVEL (2024: Birds during the COVID-19 pandemic)

• Quote: “Vaccinated people have up to 50% less mortality than expected, while unvaccinated people show significant excess mortality.” (NIVEL news item
• Claim: This was the case among vaccinated people under-mortality (less mortality than expected), while unvaccinated people showed clear excess mortality, rising to 15 to 20 percent. This was true even after adjusting for demographic, socioeconomic, and medical factors.
• Signal: The observed under-mortality in vaccinated persons ranged from −3% to even −50%, depending on age group and period. Unvaccinated people actually showed significant excess mortality, sometimes hundreds of percent higher than expected. This pattern exactly matches the classic signature of it healthy vaccinee effect.
• Bias: NIVEL suggests explanations such as corona measures, lifestyle and medical history, but does not explicitly mention HVE. However, the finding that vaccinated people die less from all causes cannot be explained by vaccination alone.

🇳🇱 RIVM/CBS (2022: Mortality and excess mortality in 2020 and 2021)

• Claim: Vaccine effectiveness against COVID-19 mortality was very high: > 90% in the first months after the baseline series, decreasing to ~80% after 7–8 months, and again > 85% after booster vaccination.
• Signal: The joint report by RIVM and CBS showed that vaccinated people had a hazard ratio of only 0.27 for mortality from causes other than COVID in the weeks after their injection, which means a 73% lower than normal risk of dying.
• Bias: This pattern directly points to the healthy vaccine effect. The report cautiously mentions this as a “possible explanation”, but indicates that it was not quantifiable. TRB is not mentioned, nor are the effects of the administrative delay.
• Quote: "The results show... a reduced risk of death from causes other than COVID-19... compared to status without this vaccine dose. This may indicate a strong selection of relatively healthy individuals."
  (RIVM/CBS final report 2022

🇮🇹 ISS reports Italy

• Claim: VE against mortality up to 96%.
• Signal: Graphs show structurally lower total mortality among vaccinated people.
• Bias: HVE ignored; TRB not discussed.
• Quote: “The efficacy in preventing death is estimated at 96% in vaccinated with a complete cycle.” (Rapporto ISS, 15 sept 2021)

🇺🇸 Tenforde et al., MMWR 2021 (VISION network)

• Claim: VE against hospitalization 94–95%.
• Signal: Figures extremely high in the first months, then decline; bias not mentioned in the paper itself.
• Bias: HVE and TRB were only later identified by external critics.
• Quote: “Among adults ≥65 years, effectiveness … against COVID-19–associated hospitalization was 94%.”

🇶🇦 Chemaitelly et al., eLife 2025, Qatar data (kritiek)

• Claim: Researched specifically non-COVID death in vaccinated vs. unvaccinated people.
• Signal: From nun-COVID mortality was much lower in vaccinated people in the first 6 months, then back to normal → typical HVE pattern.
• Bias: HVE explicitly named and quantified; TRB ignored.
• Quote: “…risk of non-COVID-19 death was substantially lower among vaccinated persons for the first 6 months… This protection waned thereafter.”

🇶🇦 Shahar, Cureus 2025 (reanalysis Qatar data) (critical)

• Claim: Reanalysis of early Qatar data.
• Signal: Adjustment for ITB halved VE against severe disease (100% → ~50%). Rudimentary correction for HVE even made the effect negative in the first month.
• Bias: Both HVE and TRB are explicitly identified and addressed.
• Quote: “Accounting for immortal time bias alone … from 100% to about 50%. Then … correction for healthy vaccinee bias … eliminated meaningful benefit…”

🇮🇹 Italy – Pescara cohort 2025 (critical)

• Claim: Vaccination reduced all-cause mortality, but increased risk of cancer admission (HR 1.23).
• Signal: Authors themselves acknowledge this HVE and unmeasured confounders are not quantifiable.
• Bias: HVE recognized but not corrected; TRB not named.
• Quote: “Given that it was not possible to quantify the potential impact of the healthy vaccinee bias and unmeasured confounders, these findings are inevitably preliminary.”

🌍 Doshi et al., Vaccine 2022 (trialkritiek)

• Claim: Reanalysis of Pfizer/Moderna trial data.
• Signal: Serious adverse events were more common than reported; ITB and selective reporting distort outcomes.
• Bias: TRB explicitly mentioned; HVE pointed out as trial bias; no formal quantification but strong methodological criticism.
• Quote: “…serious adverse events of special interest were higher in the vaccine group than the placebo group…”

Conclusion

From the first trials to national reports and reanalyses, we see the same pattern: the belief of “safe & effective” was based on studies that systematically gave an overly favorable picture due to selection bias (HVE: healthy vaccine effect) and timing bias (ITB: immortal time bias).

• Fase 3-trials (Pfizer, AstraZeneca) excluded the vulnerable and used definitions that automatically increased effectiveness.
• Observational flagships (CDC, UKHSA) reported high VE, but at the same time showed signals that can only be explained by bias.
• Nationale analyses (RIVM/CBS, NIVEL, ISS) confirmed the narrative, but also showed biologically impossible patterns, such as lower non-COVID mortality among vaccinated people.
• Reanalyses (Chemaitelly, Shahar, Doshi, Pescara) showed that correction for these distortions caused the VE to drop sharply, sometimes to zero or negative.

Even in the field of safety, the dynamic was repeated: early publications (such as the NEJM study in pregnant women) were presented as evidence of safety, while the underlying data were too limited.

The picture is consistent: the “safe & effective”studies almost automatically produced the desired result by design. But the scientific method is unwavering, despite vested interests and researchers digging in their heels.

Science corrects itself, but as Planck and Kuhn already described, this can take a long time: old paradigms do not disappear because opponents become convinced, but because a new generation grows up with the facts. Exactly that process is now taking place before our eyes.

The lesson of COVID is therefore not anti-scientific: science corrects itself - but certainly not in warp speed. Whether political leaders will succeed in accelerating that internal process from outside will remain to be seen in the near future, especially now that new booster campaigns are already in the pipeline.

References

• 1
  Quote Max Planck: https://citaten.net/quotes/max_planck/30925/citaat-een-nieuwe-wetenschappelijke-waarheid-zegeviert-niet-doordat.html
• 2
  Wikipedia about these Phases: https://nl.wikipedia.org/wiki/De_structuur_van_wetenschappelijke_revoluties#:~:text=37%3A144-,Fases,-%5Bbewerken%20%7C
• 3
  Tweet from Koopmans: https://x.com/MarionKoopmans/status/1964024089892278662
• 4
  First week after vaccination = unvaccinated https://steig.nl/2024/11/99-gevaccineerde-overlijdens-in-eerste-week-ongevaccineerd/