• home
• about
• new
• essays
• AI
• book

Frank Visser, graduated as a psychologist of culture and religion, founded IntegralWorld in 1997. He worked as production manager for various publishing houses and as service manager for various internet companies and lives in Amsterdam. Books: Ken Wilber: Thought as Passion (SUNY, 2003), and The Corona Conspiracy: Combatting Disinformation about the Coronavirus (Kindle, 2020).

Debunking Vaccine Alarmism

A Closer Look at Steve Kirsch' 'Most Important Article'

Frank Visser / Grok

I've written over 1,800 articles on the COVID vaccines and related topics. This article is the most important because it reveals the truth about the COVID vaccines using a new methodology applied to a publicly available government dataset. — Steve Kirsch[1]

Steve Kirsch

The essay by Steve Kirsch on the Czech COVID vaccine data claims that the mRNA vaccines, particularly Moderna, are associated with significantly increased all-cause mortality (ACM) compared to Pfizer, based on his analysis of record-level data from the Czech Republic. Kirsch argues that this data, analyzed using his Kirsch Cumulative Outcomes Ratio (KCOR) method, shows a clear harm signal, with Moderna linked to a 30-50% higher mortality rate than Pfizer, and that the vaccines overall provide no mortality benefit. Below, I critically examine his claims, methodology, and conclusions to provide a balanced debunking, while acknowledging potential issues in the data and its interpretation.

Key Claims and Analysis

1. Claim: Moderna Vaccines Increased All-Cause Mortality by 30-50% Compared to Pfizer

• Critique: Kirsch's analysis hinges on comparing mortality rates between Moderna and Pfizer recipients, asserting that Moderna's higher ACM indicates harm. However, several issues undermine this conclusion:

• Confounding Factors: Kirsch claims vaccine distribution was “ quasi-random,” but a commenter on his Substack notes that an online reservation system allowed vaccine choice two months after the campaign began, potentially introducing selection bias. Moderna recipients were older, had higher age-normalized comorbidity indices, and were more likely to be in senior homes, which could explain higher mortality rates without implicating the vaccine itself. Kirsch dismisses these confounders, but his analysis doesn't adequately adjust for them.

• Comorbidity Index: The Czech data includes a comorbidity index (DCCI), which Kirsch claims doesn't affect his results. However, external analyses show that normalizing for comorbidities reduces the Moderna-Pfizer mortality ratio, suggesting comorbidities play a significant role. Kirsch's failure to fully account for this weakens his causal inference.

• Statistical Significance: Kirsch cites p-values (e.g., 8e-14) to argue statistical significance, but external critiques suggest his approach to determining significance is flawed, as it doesn't account for the number of deaths and baseline mortality rates properly. Small sample sizes in certain age groups or time periods also introduce noise, especially for younger or older cohorts.

• Non-COVID Periods: Kirsch emphasizes higher Moderna mortality in non-COVID periods to argue the vaccine itself causes harm. However, external analyses show unvaccinated individuals had significantly higher mortality during COVID waves (e.g., 84% increase vs. 8% for Moderna), suggesting vaccines mitigated COVID-related deaths. This contradicts Kirsch's claim of no benefit.

2. Claim: The Czech Data Shows No Mortality Benefit from Vaccines

• Critique: Kirsch argues that the case fatality rate ratio (CFRR) for vaccinated vs. unvaccinated matches the non-COVID all-cause mortality ratio (NCACMR), implying no vaccine benefit. This overlooks several points:

• Healthy User Bias: Kirsch acknowledges that unvaccinated individuals were 3.3x more frail, which could explain higher mortality rates independent of vaccination status. This selection bias inflates the apparent mortality difference, making vaccines appear less effective.

• Hospitalization Data: External analyses of the Czech data show significantly lower age-standardized COVID hospitalization rates for vaccinated individuals (e.g., 750 per 100,000 person-years for Pfizer vs. 5,274 for unvaccinated in December 2021), indicating a protective effect against severe outcomes.

• Excess Mortality Context: Official Czech data shows excess mortality peaked in 2021 when vaccine coverage was low (14.5% by March 2021) and dropped near baseline during the Omicron wave in 2022, despite high case numbers, suggesting vaccines reduced mortality. Kirsch's focus on all-cause mortality ignores this context.

3. Claim: KCOR is a Robust, Objective Method

• Critique: Kirsch's KCOR method, which compares cumulative mortality slopes across cohorts, is presented as simple and objective, requiring only date of birth, vaccination, and death. However:

• Oversimplification: Critics, including Stanford Professor Konstantina Stankovic, argue that KCOR oversimplifies mortality patterns by assuming linear death rates, ignoring comorbidities, lifestyle, and seasonality. Mortality is complex and influenced by multiple factors, which KCOR doesn't fully address.

• Validation Issues: While Kirsch claims KCOR is validated (e.g., by Norman Fenton), external epidemiologists question its reliability for detecting small signals or adjusting for non-proportional hazards like COVID. Standard epidemiological methods, such as Cox proportional hazards models, are more robust for such analyses but are absent in Kirsch's work.

• Mirage Effects: Kirsch himself acknowledges that certain analyses (e.g., risk ratio comparisons for booster doses) can produce statistical mirages, yet he doesn't fully address how KCOR might be similarly affected by cohort mismatches or unmeasured confounders.

4. Claim: No Epidemiologist Analyzes the Czech Data Due to Narrative Bias

• Critique: Kirsch suggests the lack of peer-reviewed studies on the Czech data indicates institutional censorship. However:

• Data Accessibility: The Czech data, while publicly available, is complex (1.3GB, 10M+ records) and requires specialized tools, which may deter researchers without adequate resources. This doesn't necessarily imply censorship.

• Published Studies: Contrary to Kirsch's claim, some studies have analyzed Czech data. For example, a 2024 paper in Scientific Reports used Czech molecular surveillance data to assess pandemic trends, showing vaccines reduced excess mortality during Omicron. Another study on batch-dependent safety exists, though it's less comprehensive.

• Reputational Risk: While institutional bias may exist, researchers may avoid controversial analyses due to the need for rigorous validation, not just fear of backlash. Kirsch's reliance on his own method without peer-reviewed validation undermines his credibility.

5. Claim: Vaccines Caused Widespread Harm (e.g., 500,000 US Deaths)

• Critique: Kirsch extrapolates from the Czech data to claim massive vaccine-related deaths globally. This is unsupported:

• Lack of Causal Evidence: Higher mortality in Moderna recipients doesn't prove causation. Alternative explanations (e.g., socioeconomic status, healthcare access) are plausible but underexplored by Kirsch.

• Anecdotal Evidence: Kirsch cites anecdotes (e.g., a stepson's death post-Moderna) to support his claims, but these are not statistically robust and are prone to confirmation bias.

• Contradictory Data: Global studies, such as WHO estimates and European excess mortality analyses, show correlations between vaccination and reduced mortality, not increased harm. Kirsch's claim of 500,000 US deaths lacks direct evidence and contradicts broader epidemiological trends.

Broader Context and Counterpoints

• Vaccine Efficacy: Multiple studies, including a 2022 meta-analysis, show COVID vaccines reduced severe outcomes and mortality, particularly in high-risk groups. The Czech data itself shows lower hospitalization rates for vaccinated individuals, contradicting Kirsch's no-benefit claim.

• Safety Signals: While adverse events exist, large-scale studies (e.g., VAERS analyses) show serious events are rare, and batch-dependent variations don't necessarily indicate widespread harm. Kirsch's focus on Moderna overlooks that Pfizer also had safety concerns, yet he uses it as a baseline, assuming it's “ safer” without sufficient evidence.

• Transparency and Bias: Kirsch's call for data transparency is valid, but his dismissal of mainstream science as corrupt ignores the complexity of epidemiological research. His reliance on unpeer-reviewed Substack posts and limited external validation (e.g., from Denis Rancourt) weakens his argument compared to rigorous, peer-reviewed studies.

Conclusion

Kirsch's essay overstates the Czech data's implications by relying on a simplistic methodology (KCOR) that doesn't adequately account for confounders like comorbidities, age, or selection bias. His claims of widespread vaccine harm, particularly for Moderna, are not supported by robust causal evidence and contradict broader epidemiological data showing vaccines reduced COVID-related mortality and hospitalizations. While the Czech dataset is valuable, Kirsch's analysis lacks the rigor of standard epidemiological methods and ignores contradictory findings, such as lower hospitalization rates among vaccinated individuals. His narrative of institutional censorship is plausible but overstated, as data complexity and validation requirements likely contribute to limited analyses. Critical examination of the data suggests vaccines provided measurable benefits, particularly during COVID waves, and Kirsch's conclusions should be approached with skepticism until validated by peer-reviewed research.

Appendix: Methodological Concerns

Steve Kirsch's Kirsch Cumulative Outcomes Ratio (KCOR) is presented as a simple, objective method to compare mortality outcomes across cohorts (e.g., vaccinated vs. unvaccinated or Moderna vs. Pfizer) using the Czech COVID vaccine data. Below is a detailed critique of KCOR, focusing on its methodological flaws, limitations, and potential misinterpretations, grounded in epidemiological principles and the specific context of Kirsch's analysis.

Overview of KCOR

KCOR is described as a ratio of cumulative mortality slopes derived from plotting deaths over time for different groups (e.g., vaccinated vs. unvaccinated). It uses basic data— date of birth, vaccination status, and date of death— to calculate mortality rates and compare them across cohorts. Kirsch claims KCOR is robust because it avoids complex adjustments and directly reflects real-world outcomes, making it resistant to manipulation.Critique of KCOR

1. Oversimplification of Mortality Dynamics

• Issue: KCOR assumes mortality rates are linear and comparable across cohorts, ignoring the complex, non-linear nature of mortality influenced by time-varying factors like COVID waves, seasonality, and healthcare access.

• Impact: By focusing on cumulative slopes, KCOR fails to account for time-dependent hazards (e.g., higher mortality during COVID peaks). For example, external analyses of the Czech data show unvaccinated individuals had significantly higher mortality during COVID waves (e.g., 84% increase vs. 8% for Moderna in early 2021), which KCOR flattens into a single ratio, obscuring vaccine benefits.

• Alternative: Standard epidemiological methods like Cox proportional hazards models or time-stratified analyses adjust for time-varying risks and are better suited for capturing dynamic mortality patterns.

2. Inadequate Adjustment for Confounders

• Issue: KCOR relies on minimal data inputs, which Kirsch touts as a strength, but this ignores critical confounders like comorbidities, socioeconomic status, or healthcare access. In the Czech data, Moderna recipients were older (mean age difference noted in external critiques) and had higher comorbidity indices (DCCI), yet KCOR doesn't robustly adjust for these.

• Impact: Higher mortality in Moderna cohorts could reflect these confounders rather than vaccine harm. A commenter on Kirsch's Substack noted that normalizing for comorbidities reduces the Moderna-Pfizer mortality ratio, undermining Kirsch's 30-50% harm claim.

• Alternative: Multivariable regression models or propensity score matching can better control for confounders, ensuring fair comparisons across cohorts.

3. Susceptibility to Selection Bias

• Issue: KCOR assumes quasi-random vaccine allocation, but Czech data suggests non-random distribution after an online reservation system allowed vaccine choice two months into the campaign. Moderna recipients were more likely to be in senior homes or higher-risk groups, skewing mortality outcomes.

• Impact: This selection bias inflates Moderna's mortality signal in KCOR analyses, as higher-risk individuals were disproportionately represented. Kirsch's dismissal of this as “ random enough” lacks statistical rigor.

• Alternative: Stratified analyses by risk group (e.g., nursing home vs. community-dwelling) or inverse probability weighting could mitigate selection bias.

4. Statistical Noise in Small Subgroups

• Issue: KCOR's reliance on cumulative mortality slopes is sensitive to small sample sizes, especially in younger or older age groups with fewer deaths. Kirsch's analysis includes age-stratified KCORs, but small death counts (e.g., in younger cohorts) produce noisy estimates.

• Impact: This noise can exaggerate differences in mortality ratios, leading to misleading conclusions. For instance, Kirsch's claim of a 50% higher Moderna mortality rate in certain groups may reflect statistical artifacts rather than a true signal.

• Alternative: Bayesian methods or larger sample sizes could stabilize estimates, while confidence intervals would better convey uncertainty.

5. Failure to Account for Competing Risks

• Issue: KCOR treats all-cause mortality as a single outcome, ignoring competing risks (e.g., COVID vs. non-COVID deaths). Vaccines reduce COVID mortality but may not affect non-COVID causes, which dominate in non-COVID periods.

• Impact: Kirsch's focus on non-COVID periods to argue vaccine harm overlooks the vaccine's protective effect during COVID waves. External analyses show vaccinated individuals had lower COVID hospitalization rates (e.g., 750 vs. 5,274 per 100,000 person-years for unvaccinated in December 2021), which KCOR doesn't capture.

• Alternative: Competing risk models (e.g., Fine-Gray models) could distinguish between COVID and non-COVID mortality, providing a clearer picture of vaccine impact.

6. Lack of Peer-Reviewed Validation

• Issue: Kirsch claims KCOR is validated by figures like Norman Fenton, but it lacks peer-reviewed scrutiny in epidemiological literature. Its simplicity, while appealing, hasn't been tested against established methods like standardized mortality ratios (SMRs) or hazard ratios.

• Impact: Without rigorous validation, KCOR's reliability for detecting small mortality signals is questionable. Critics, including Stanford's Konstantina Stankovic, argue it oversimplifies complex data, risking misinterpretation.

• Alternative: Peer-reviewed methods like SMRs or Kaplan-Meier survival curves, widely used in vaccine studies, offer validated frameworks for mortality comparisons.

7. Potential for Statistical Mirages

• Issue: Kirsch acknowledges that certain analyses (e.g., risk ratios for booster doses) can produce “ mirages” due to cohort mismatches or timing effects. KCOR is similarly vulnerable, as cumulative slopes can be distorted by differences in observation periods or cohort composition.

• Impact: For example, if Moderna recipients were vaccinated later (as some Czech data suggests), their exposure to later COVID variants or different healthcare conditions could skew KCOR results, mimicking a harm signal.

• Alternative: Time-to-event analyses with clear cohort alignment (e.g., matching vaccination dates) could reduce such artifacts.

8. Misleading Interpretation of Ratios

• Issue: KCOR's ratio-based approach emphasizes relative differences (e.g., Moderna vs. Pfizer) without contextualizing absolute risk. A 30-50% higher mortality ratio sounds alarming, but if baseline mortality is low (e.g., in younger groups), the absolute impact is minimal.

• Impact: Kirsch's presentation of KCOR results amplifies perceived harm without providing absolute risk estimates, which are critical for public health decisions. For instance, global data shows vaccines reduced mortality by millions (WHO estimates), a context KCOR ignores.

• Alternative: Reporting absolute risk differences alongside ratios, as done in vaccine trials, would provide a balanced perspective.

Broader Implications

KCOR's simplicity, while appealing for public communication, sacrifices the nuance required for robust epidemiological analysis. Its failure to adjust for confounders, time-varying risks, and selection biases makes it prone to misinterpreting mortality signals, especially in complex datasets like the Czech records. Established methods, such as Cox models or propensity score analyses, are more rigorous for causal inference and have been used in peer-reviewed studies to demonstrate vaccine benefits (e.g., 2022 meta-analyses showing reduced mortality). Kirsch's reliance on KCOR, without addressing these limitations, undermines his claims of vaccine harm and overstates the method's objectivity.

Conclusion

KCOR is a flawed tool for analyzing vaccine-related mortality due to its oversimplification, inadequate confounder adjustment, susceptibility to bias, and lack of peer-reviewed validation. It risks producing misleading conclusions, as seen in Kirsch's interpretation of the Czech data, where Moderna's higher mortality likely reflects cohort differences rather than vaccine harm. For reliable insights, epidemiologists should prioritize established methods that account for the complexity of mortality data.

NOTES

[1] Steve Kirsch, "New analysis of the Czech COVID vaccine data reveals that the mRNA shots were deadly for all ages. They should be pulled from the market", kirschsubstack.com, August 28, 2025.