Neurodevelopmental Outcomes in COVID-19-Vaccinated Versus Unvaccinated U.S. Children: Analysis of the National Health Interview Survey, 2022–2024
Verdict
UNSCIENTIFIC
Neither the paradigm nor the study inside it passes. The verdict is unscientific.
1·Study methodology · inside the paradigm
UNSCIENTIFIC
Does not follow the scientific method within Psychiatric Model.
2·Paradigm · Psychiatric Model
UNSCIENTIFIC
Fails the tests. The worse layer decides.
How the verdict is decided
The verdict grades both layers: the paradigm this study assumes, and how the study was carried out inside it. Both must be scientific for the whole thing to be scientific. A clean method inside an unscientific paradigm is unscientific. An unscientific method inside a better paradigm is unscientific. A split on either layer is mixed. The worse layer decides.
1·Study methodology
Did this study test its claim with methods that are independent, falsifiable, and non-circular?
UNSCIENTIFIC
This study claims that children who got a COVID-19 vaccine have higher odds of autism, ADHD, anxiety and other diagnoses than unvaccinated children, and that the odds rise with more doses. All three tests fail.
Independently Verifiable
Fail
Parents said whether a doctor had ever diagnosed autism, and that answer became the autism variable; nobody in the study watched a child.
Falsifiable
Fail
The raw autism rates were flat, 4.19% against 4.28%, and the authors wrote that off as negative confounding, so only the adjusted number could count.
Non-Circular
Fail
The same adjustment that pushed ER visits to zero effect was called proof it works, then used to lift autism from no difference to 1.26.
Why
The authors started from the belief that the vaccine harms brain development. They used a parent survey asking whether a child ever got the shot and whether a doctor ever diagnosed autism, ADHD, anxiety and similar labels. The raw counts showed no autism difference. They then added income, education and insurance, and the odds rose to 1.26. They called the flat raw result negative confounding. ER visits moved from 0.72 to 1.02 under the same adjustment, and the authors called that proof the adjustment was sound. Flu shots, their control, also showed higher ADHD and hospitalization. Early-exposure groups showed no difference, 1.11 and 1.02, and were called undetermined. They ran 27 specifications with no correction for multiple comparisons. Nobody observed a child. They counted what parents said clinicians had recorded.
2·Paradigm · Psychiatric Model
Does the framework this study assumes pass the three tests?
UNSCIENTIFIC
Independently Verifiable
Fail
A child's behavior can be watched, but which behaviors count, how many, and by what age comes from criteria, so the diagnosis count measures the criteria.
Falsifiable
Fail
Rising diagnosis counts get explained by awareness, wider criteria and a spectrum, so no count could show the labels are not distinct disorders.
Non-Circular
Fail
The criteria were written from clinicians' own cases, and diagnoses made with those criteria are then offered as proof the criteria find real conditions.
Why
This study counted diagnoses, not behavior. Whether a child has autism, ADHD or anxiety depends on a checklist that a committee wrote, with its own cutoffs, ages and subtypes. Change the checklist and the count changes while every child acts exactly as before. When counts rise, the field credits awareness, broader criteria, or the spectrum. When subtypes fail to separate, it adds comorbidity. None of that can show the labels are wrong. The criteria come from clinicians' judgments, and clinicians' diagnoses made with those criteria are used to show the criteria work. The model treats a recorded label as the disease and never asks what is happening to the child.
From the study
Lifetime parent-reported diagnoses were autism spectrum disorder (ASDEV_C; ages 2–17), ADHD (ADHDEV_C; 2–17), learning disability (LDEV_C; 2–17)
Crude ASD prevalence did not differ by vaccination status (4.19% vs 4.28%; OR 0.98, 95% CI 0.83–1.16; Table 2, Figure 1B).
Autism estimates are null in crude and M1 and emerge at M2, when socioeconomic position enters the model; the reference outcome moves from 0.72 to 1.02 under the same adjustment.
Two-sided p < 0.05 was considered statistically significant, without correction for multiple comparisons, consistent with an exploratory signal-detection design.