Acting CDC Director Dr. Jay Bhattacharya tried very hard to prevent the findings from comimg out. Bhattacharya even exerted his power as CDC head to block publication of the paper. Undaunted, CDC scientists went around him and published the data in the Journal of the American Medical Association today.
What was al the fuss about? The study by CDC showed that the COVID booster for 2025-2026 reduced ER visits and hospitalization by 55%. What was the objection? Why did Bhattacarya block the release of these positive results?
Because he is a committed antivaxxer and his mind is made up. It has been from the beginning. He doesn’t believe the COVID mRNA vaccine works and he doesn’t want you to know that it does.
What specifically does Bhattacharya object to? He doesn’t like the study design - the “test negative” methodology. What is the “test negative” design?
The test-negative design (TND) is an observational epidemiologic method that the CDC and many other researchers have used for years to estimate vaccine effectiveness against influenza, COVID-19, RSV, and other respiratory diseases. It became a major point of controversy in 2025–2026 when some CDC COVID vaccine studies using this design were challenged by critics such as Jay Bhattacharya and Martin Kulldorff.
How the test-negative design works
Instead of comparing vaccinated and unvaccinated people in the general population, researchers:
Enroll people who seek medical care for similar symptoms (fever, cough, respiratory illness).
Test everyone for COVID.
Classify:
Cases = test positive for COVID
Controls = test negative for COVID
Compare vaccination rates between the two groups.
A simplified example:
GroupVaccinatedUnvaccinatedCOVID-positive200400COVID-negative800600
Vaccinated people are proportionally less common among the COVID-positive group, suggesting protection.
Vaccine effectiveness is estimated as:
VE = (1 − Odds Ratio) × 100%
This is the formula CDC’s COVID vaccine effectiveness studies commonly use.
Why researchers like it
The biggest problem in vaccine studies is that vaccinated and unvaccinated people often behave differently.
For example, vaccinated people may:
Seek medical care more often
Be older
Have more chronic illness
Be more health-conscious
The test-negative design partially controls for this by only comparing people who were sick enough to seek care and get tested. This reduces “health-care-seeking bias.”
The main criticism
Critics argue that the design depends on assumptions that may not always be true.
Key concerns include:
1. Prior infection bias
By 2024–2026, many unvaccinated people had already had COVID.
Natural infection provides some immunity.
If prior infections are more common among unvaccinated individuals and are not fully measured, vaccine effectiveness can appear lower or higher than reality.
2. Misclassification
Some people who test negative for COVID may actually have:
Influenza
RSV
Another respiratory virus
The design assumes these illnesses are not affected by COVID vaccination. If that assumption fails, bias can occur.
3. Residual confounding
Vaccinated and unvaccinated people may differ in ways researchers cannot completely adjust for:
Risk-taking behavior
Occupation
Health status
Prior exposure history
These differences can influence the estimate.
4. Testing behavior
If vaccinated people are more or less likely to seek testing, results can be distorted. CDC acknowledges changing testing patterns as a challenge in vaccine effectiveness studies.
What supporters say
Supporters argue:
Randomized placebo-controlled trials are generally not feasible once a vaccine is widely available.
The test-negative design has been used successfully for influenza vaccine studies for many years.
It is not perfect, but it often provides the best available real-world estimate of effectiveness.
Multiple studies comparing TND results with randomized trial data have found reasonably similar estimates when the method is applied correctly.
What the CDC COVID study found
The CDC study that became controversial in 2026 used a test-negative design and found that updated COVID vaccination reduced:
Hospitalizations by about 55%
Emergency and urgent care visits by about 50%
among adults during the studied winter season.
My assessment
As a physician familiar with epidemiology, I recognize that the test-negative design is essentially a specialized case-control study. It is not as strong as a randomized controlled trial, but it is also not some novel CDC invention for COVID. It has been a mainstream vaccine-effectiveness method for influenza and other respiratory infections for over a decade. The real scientific debate is not whether TND is valid at all, but rather how much bias remains in the estimates after accounting for prior infection, testing behavior, and other confounders.
The criticisms raised by Jay Bhattacharya, Martin Kulldorff, and others regarding CDC COVID vaccine studies generally fall into a few major categories. Some are legitimate methodological concerns; others are arguments about the interpretation of results rather than proof that the studies are invalid.
1. Prior Infection (”Hybrid Immunity”) Bias
This is probably the strongest criticism.
By 2023–2026, most Americans had experienced at least one SARS-CoV-2 infection. Prior infection provides some degree of protection against reinfection and severe disease.
The concern is:
Unvaccinated people were more likely to have been previously infected.
Many prior infections were undocumented.
If prior infection is not adequately measured, the unvaccinated group may appear more protected than expected.
Example:
If a person avoids hospitalization because of natural immunity, the study may underestimate vaccine effectiveness.
This criticism is widely accepted across the epidemiology community. CDC investigators themselves have increasingly tried to account for prior infection in later studies.
Counterargument
The bias usually drives vaccine effectiveness estimates downward rather than upward. In other words, failure to fully account for natural immunity often makes vaccines look less effective than they really are.
2. Healthy Vaccinee Bias
Another legitimate concern.
People who receive boosters may differ systematically from those who do not.
Boosted individuals may:
Have better access to healthcare.
Follow medical advice more closely.
Seek treatment earlier.
Have different socioeconomic status.
These factors can independently reduce hospitalization risk.
Counterargument
The test-negative design was specifically developed to reduce some of this bias by restricting analysis to people who sought care for similar symptoms.
It does not eliminate the problem entirely.
3. Differential Testing Behavior
Critics argue that vaccinated and unvaccinated people may not seek testing at the same rate.
For example:
Vaccinated individuals may be more likely to seek testing when mildly ill.
Unvaccinated individuals may avoid healthcare settings.
If true, this can distort effectiveness estimates.
Counterargument
This is a recognized limitation, but most severe-outcome studies focus on emergency department visits and hospitalizations, where testing rates tend to be high regardless of vaccination status.
4. The “Depletion of Susceptibles” Argument
This is more technical.
Early in a pandemic:
Higher-risk unvaccinated people become infected first.
Many subsequently acquire immunity.
The remaining unvaccinated population may become increasingly enriched with survivors who already possess natural immunity.
This can make vaccine effectiveness appear to decline over time even if biological protection remains relatively stable.
Many epidemiologists agree this is a real phenomenon.
5. Selection Bias Within Test-Negative Designs
Kulldorff has argued that selecting only symptomatic people who present for testing can introduce bias if vaccination affects the likelihood of seeking care.
This criticism is methodologically valid.
The question is magnitude.
Most experts would say:
Bias exists.
The bias is usually modest.
The design remains useful when carefully implemented.
6. Were the CDC Estimates “Made Up”?
No.
The criticisms generally do not demonstrate fraud or statistical misconduct.
The CDC studies:
Used standard epidemiologic methods.
Published protocols.
Reported confidence intervals.
Underwent peer review.
The debate is about how much residual bias remains, not whether the calculations were fabricated.
What Most Epidemiologists Agree On
There is broad agreement that:
Strong evidence exists that COVID vaccines reduced:
Hospitalization
ICU admission
Death
particularly during the original, Alpha, Delta, and early Omicron periods.
Evidence comes from:
Randomized trials
Test-negative studies
Cohort studies
National registry studies from multiple countries
The magnitude of protection and durability of protection remain debated.
Where the Real Scientific Disagreement Lies
The disagreement is usually not:
“Did vaccines work at all?”
Instead it is:
“How much protection remained after repeated infections and repeated vaccine doses in highly immune populations?”
By 2025–2026, when most people had either:
vaccination,
infection,
or both,
estimating the incremental benefit of an additional booster became much harder than estimating the benefit of the original vaccine series in 2020–2021.
That is where Bhattacharya, Kulldorff, and others focus many of their methodological criticisms. The TND remains a standard and generally accepted tool, but its assumptions became increasingly strained as population immunity became nearly universal.
Neither Bhattacharya nor Kulldorf have the appropriate academic background. One is an economist, the other has a PhD in Operations Managament. Although Bhattacharya has an MD, he never did any post graduate medical education. Kulldorff is not an MD. Neither are epidemiologists. Their lack of training shows.
The fact that Bhattacharya tried to use his authority to block release of these data is reprehensible. It is an aritrary and capricious abus of power. He should be fired
