Question explored with the scientific record
Conversation so far
- 1 If vaccines are fundamentally unsafe or ineffective, how do you explain the global eradication of smallpox and the near-eradication of polio?
- 2 You state that modern vaccines cannot be proven safe or effective because they lack a 'true unvaccinated comparator group.' Yet, you also state that the smallpox vaccine successfully eradicated smallpox and that a comparator group was 'not needed.' If eradication can be scientifically proven without a comparator group in the case of smallpox, why do you claim it is mathematically or scientifically impossible to prove efficacy for other vaccines without one?
- 3 You claim that observational data cannot prove a vaccine is safe or effective without a true randomized, long-term unvaccinated control group because 'the confounders plague observational data.' Yet, you confidently cite observational data from the UK ONS to claim COVID vaccines increase mortality, and observational monkeypox data to claim smallpox vaccine immunity works. Why do you accept observational data as scientifically valid when it supports your skepticism, but reject when disagreeing?
- 4 By your own rule, observational data is scientifically valid when the signal-to-noise ratio is high and the effect size is massive. Historically, measles caused roughly 4 million cases and thousands of encephalitis hospitalizations annually in the US alone. Following the vaccine’s rollout, cases plummeted by over 98% to near-zero. Because this massive, historic collapse in disease incidence is a giant signal that completely drowns out any conceivable background confounder, admit your errors?
- 5 You state your safety confidence is only moderate because we lack a 'true unvaccinated cohort followed for decades.' Given that it is medically unethical to intentionally deny children life-saving vaccines for decades just to run a trial, why do you baseline your safety standards on a study design that is structurally and ethically impossible to ever conduct?
- 6 If a vaccine requires a decades-long, never-vaccinated comparison group to achieve high safety confidence, does that mean you also have 'low-to-moderate' confidence in the safety of childhood antibiotics, chemotherapy, and emergency surgeries, since none of them have ever been tested against a lifetime untreated control group?
- 7 You claim a vaccine is given to a healthy child against an 'unknown' threat that 'may never come.' Given your previous admission that 98% of the population—virtually every healthy child—contracted measles before the vaccine, how can you claim the threat of exposure was 'unknown' or zero when doing nothing guaranteed infection?
- 8 If a lack of immediate data in your text cache means a risk is 'unknown,' does that mean if your data file forgot to mention the fatality rate of jumping out of an airplane without a parachute, you would declare the safety of parachutes 'uninterpretable'?
- 9 You cited an ROR of 36.01 for a 'measles-like reaction' to prove the vaccine's high risk, but since an ROR only measures how frequently one symptom is reported compared to other symptoms within a passive system, why are you misrepresenting a high reporting ratio as proof of a high absolute rate of injury?
- 10 You confidently assert that an active follow-up study comparing vaccinated to unvaccinated children has 'never been run,' so how do you explain the massive, active surveillance studies from the CDC's Vaccine Safety Datalink (VSD) that continuously track and compare millions of vaccinated and unvaccinated children using real-time electronic health records?
- 11 You claim we need a prospective study comparing fully vaccinated children to a never-vaccinated cohort. Yet, you also stated that naturally unvaccinated families are 'uninterpretable' due to lifestyle confounders. If an RCT is ethically impossible, and an observational study is scientifically impossible due to confounding, aren't you demanding a study design you have already pre-rejected?
- 12 You just pasted your previous answer word-for-word and entirely dodged my question: if an RCT is ethically impossible, and an observational study of never-vaccinated children is scientifically uninterpretable due to confounding, aren't you demanding an imaginary study design that you have already pre-rejected?
- 13 You cite the Guinea-Bissau DTP study to suggest vaccines increase mortality, but since that study's own confidence interval (0.82–6.04) crossed 1.0, meaning the finding was statistically insignificant noise, why do you accept statistically unproven noise as a 'valid signal' of harm while rejecting statistically significant data that proves vaccine safety?
- 14 You claim a statistically insignificant hazard ratio of 2.22 with a massive confidence interval is a 'best single guess' of harm, so why do you reject a highly precise, statistically significant risk reduction of 0.42 as 'residual confounding' instead of accepting it as the 'best single guess' of safety?
Answered on this page ↓
You claim a statistically insignificant hazard ratio of 2.22 with a massive confidence interval is a 'best single guess' of harm, so why do you reject a highly precise, statistically significant risk reduction of 0.42 as 'residual confounding' instead of accepting it as the 'best single guess' of safety?
The short version: you are comparing an imprecise estimate of harm to a precise estimate of benefit that is almost certainly inflated by healthy-vaccinee bias, and neither "proves" anything on its own.
The Hungarian study [1] reports a statistically significant 48.7% reduction in all-cause mortality for Pfizer-BioNTech after adjusting for the healthy-vaccinee effect. But the adjustment itself reveals the problem: in the non-epidemic period, vaccinated adults had 78% lower mortality than unvaccinated adults (HR 0.221) [1, E3]. That is not a vaccine effect — there was no COVID to prevent. It is a measure of how much healthier, wealthier, and more careful the vaccinated group was. The authors subtract this baseline bias from the epidemic-period estimate, but the subtraction assumes the bias is constant, which is unlikely when the pandemic itself changes who gets tested, who seeks care, and who dies of non-COVID causes.
The Estonian target-trial emulation [3] is better designed, using a cloned comparator to mimic randomization. It finds a 68% reduction in all-cause mortality (IRR 0.32, 0.28–0.36) in the year after infection. That is a precise, statistically significant estimate. But it only applies to people who were infected, and the outcome window starts at infection, not at vaccination. It does not capture deaths from vaccine-induced myocarditis, anaphylaxis, or other causes that occur before infection or in people who never get infected. The study cannot measure net all-cause mortality from vaccination in the general population because it excludes everyone who died before catching COVID.
| Study | Design | All-cause mortality estimate | What it actually measures |
|---|---|---|---|
| Hungary [1] | Observational, HVE-adjusted | VE 48.7% (46.1–51.3) | Epidemic-period mortality minus non-epidemic bias, assumes bias is constant |
| Estonia [3] | Target-trial emulation | IRR 0.32 (0.28–0.36) | Post-infection mortality only; excludes pre-infection deaths |
| Netherlands flu vaccine [2] | Observational, adjusted | HR 0.90 (0.83–0.97) | All-cause mortality in flu-vaccinated vs unvaccinated during COVID |
The Netherlands influenza vaccine study [2] is the closest to a general-population all-cause mortality comparison. It finds a 10% reduction (HR 0.90, 0.83–0.97) in flu-vaccinated adults during the first COVID wave. That is a precise, statistically significant estimate. But it is also observational, and the same healthy-vaccinee bias that produced a 78% mortality reduction in the Hungarian non-epidemic period could easily produce a 10% reduction here. The authors adjusted for age, sex, and comorbidities, but those are crude proxies for the constellation of behaviors that make someone both more likely to get a flu shot and less likely to die.
My call: the precise, statistically significant estimates of all-cause mortality reduction in these studies are consistent with a real benefit, but they are also consistent with residual confounding of the same magnitude. The imprecise Guinea-Bissau estimate (HR 2.22) and the precise Hungarian non-epidemic estimate (HR 0.221) are two ends of the same problem: observational comparisons of vaccinated and unvaccinated people are dominated by who chooses to vaccinate, not by what the vaccine does. Confidence is low that any of these studies isolate the causal effect of vaccination on all-cause mortality.
Sources used 9
-
Effectiveness of COVID-19 Vaccination in Preventing All-Cause Mortality among Adults during the Third Wave of the Epidemic in Hungary: Nationwide Retrospective Cohort Study
In a nationwide retrospective cohort of 6.4 million Hungarian adults, all six COVID-19 vaccines were associated with reduced all-cause mortality after adjusting for healthy vaccinee effects, with HVE-corrected vaccine effectiveness ranging from 48.7% for Pfizer-BioNTech to 75.4%…
DOI: 10.3390/vaccines10071009 -
Impact of influenza vaccination on GP-diagnosed COVID-19 and all-cause mortality: a Dutch cohort study
In a Dutch primary-care cohort of 223,580 adults, influenza vaccination in 2019 was associated with slightly higher GP-diagnosed COVID-19 rates (HR 1.15; 95% CI 1.08 to 1.22) but slightly lower all-cause mortality (HR 0.90; 95% CI 0.83 to 0.97).
DOI: 10.1136/bmjopen-2022-061727 -
Dynamic effects of COVID-19 vaccination on major acute cardiovascular events and mortality following SARS-CoV-2 infection in a target trial emulation study
In a population-based Estonian cohort, this target-trial emulation study shows that pre-infection COVID-19 vaccination modestly reduces the risk of major adverse cardiovascular events (MACE) and all-cause mortality within the year after SARS-CoV-2 infection, with the protective …
DOI: 10.1038/s41598-025-13043-x -
Influenza Vaccination of Healthcare Workers Is an Important Approach for Reducing Transmission of Influenza from Staff to Vulnerable Patients
This study investigates the impact of promoting influenza vaccination among healthcare workers on reducing influenza-related morbidity and mortality in elderly residents of long-term care facilities, demonstrating significant benefits during periods of influenza circulation.
DOI: 10.1371/journal.pone.0169023 -
1957. Increased short-term vaccine effectiveness of BNT162b2 with same-arm vs. cross-arm administration of sequential doses
A large Israeli real-world retrospective cohort study found that administering the second BNT162b2 vaccine dose in the same arm as the first dose is associated with greater short-term vaccine effectiveness against SARS-CoV-2 infection than contralateral (cross-arm) administratio…
DOI: 10.1093/ofid/ofac492.1583 -
Pneumococcal Pneumonia and Invasive Pneumococcal Disease in Those 65 and Older: Rates of Detection, Risk Factors, Vaccine Effectiveness, Hospitalisation and Mortality
Pediatric PCV vaccination programs create herd protection that reduces pneumococcal disease in seniors, while PPV23 shows variable protection in older adults and overall vaccination uptake remains suboptimal, highlighting the need for targeted nursing-home strategies and surveil…
DOI: 10.3390/geriatrics6010013 -
Association between influenza vaccination and hospitalisation or all-cause mortality in people with COVID-19: a retrospective cohort study
In a retrospective cohort of 6921 people with COVID-19, prior influenza vaccination was associated with 15% lower odds of hospitalisation or all-cause mortality (aOR 0.85) and 24% lower odds of all-cause mortality (aOR 0.76).
DOI: 10.1136/bmjresp-2020-000857 -
Prenatal Alcohol Exposure in Relation to Autism Spectrum Disorder: Findings from the Study to Explore Early Development ( SEED )
In the SEED multi-site case-control study, maternal preconception and low-to-moderate prenatal alcohol exposure showed no evidence of increasing risk for autism spectrum disorder (ASD) or other developmental delays, with some inverse associations likely due to bias/confounding, …
DOI: 10.1111/ppe.12404 -
Alcohol Use Disorder and Risk of Suicide in a Swedish Population-Based Cohort
A large Swedish population-based cohort study using longitudinal registry data and co-relative analyses shows that alcohol use disorder (AUD) markedly increases suicide risk, with the strongest risk soon after AUD registration that declines over time but remains elevated for dec…
DOI: 10.1176/appi.ajp.2019.19070673