Exclusive: AI analysis of internal CDC documents from the 2004 DeStefano study on MMR and autism shows they violated protocol and unethically omitted a 2.25 - 2.4 odds ratio associating MMR and autism

Hooker's analysis had OR=3.86 similar to values found in other studies; it is statistically significant. The CDC knew in 2004 that the data showed the association, so they destroyed the data.

Executive summary

Eighteen months ago, I wrote a substack article, “Over 300 pages of evidence from the CDC show that vaccines cause autism.”

The documents are all related to the 2004 CDC DeStafano study published in Pediatrics: Age at first measles-mumps-rubella vaccination in children with autism and school-matched control subjects: a population-based study in metropolitan atlanta. The paper claimed there was no association between vaccines and autism while at the same time admitting in the abstract a statistically significant association.

I offered the CDC documents in my possession to the press for verification. As you can imagine, no journalist was interested in looking at the documents.

I recently decided to have AI analyze the document set.

AI verified the result that:

1. The researchers found a statistically significant association between the MMR vaccine and autism (Black Model, MMR <36 months: OR 2.4, 95% CI 1.4-4.4). This is similar to values found in other papers that used different methodologies as noted here and here. These OR findings from multiple studies imply that around 75% of autism cases are likely caused by vaccines as Toby Rogers recently explained in his substack article. He did this independently of my earlier work and found the same numbers that I found. This follows from the large OR numbers from the studies (we both found the same studies) and the fact that nearly all kids in the US are vaccinated. It’s inescapable. It’s simple math.

2. The researchers violated their own research protocol by seeing the subgroup analysis and then deciding that it didn’t fit their narrative so they omitted mentioning it in the paper which is a very serious breach of scientific ethics.

3. The researchers engaged in an illegal destruction of government property in order to suppress the signal so that they could write a clean paper showing no association between the MMR vaccine and autism.

4. CDC executive Coleen Boyle, who allegedly ordered the document destruction, admitted that she would “never go and testify again” in Congress.

If the researchers truly thought that the association between MMR and autism was insignificant, why did they willfully engage in what was clearly an illegal purging of US government data and refuse to testify about it? There is only one reason I can think of and that is because the signal was serious so they felt they had to bury it. If they didn’t delete the data, they’d be showing that Andy Wakefield was right and we can’t have THAT happen can we?

In this article, I’ll share with you the excerpts of the AI analysis.

Also, the study itself found a OR=1.49 for the association between MMR and autism. The population attributable risk (PAR) is 33% meaning 33% of autism cases are likely caused by the MMR vaccines. But that’s based on just one vaccine type and there are 18 different vaccine types. For black boys, the study found a OR=2.4 so they destroyed that day. That’s a PAR of 58%. So you’re likely looking at the #1 cause of autism based on this one CDC study alone. So when scientists claim there is no association between vaccines and autism, they are simply not telling the truth.

From the DeStefano paper, here’s the 1.49 right in the abstract showing it is statistically significant:

From the December 2001 spreadsheet (race3.xlsx)

They’ve known about the association for over 25 years! This is a screenshot from one of the spreadsheets created by the lead scientist in the study.

AI analysis highlights

I used ChatGPT Codex for the analysis since there were around 400 documents in the dataset. Here are some of the conclusions.

Note that the published DeStefano paper itself admitted to a statistically significant association (OR 1.49), then tried to downplay it:

Summary

The CDC knew the MMR vaccine was linked to higher rates of autism. Their paper even admitted to a weaker signal (OR=1.49) than what they found (OR=2.4). They decided it was best to cover it up.

The mainstream media plays along by refusing to investigate. Congress isn’t going to hold hearings to expose the truth and the CDC officials will refuse to testify even if they did.

But we have the evidence in plain sight of the association.

More troubling of course is if “science” really wanted to answer the question, a very simple study creating a histogram of date from last vaccination to date parents first noticed onset of ASD symptoms (where there is a specific date that can be identified where a child goes from normal to severely autistic within <24 hours) would reveal the association. Pediatrician Doug Hulstedt is the only pediatrician I know of who actually tracked this number and he found that all 44 cases of rapid onset autism in his practice happened within 2 weeks after a vaccine. That is simple not possible if vaccines don’t cause autism. Even if kids are vaccinated every 4 months, the chance of observing 44 events within 2 weeks after vaccination is vanishingly small (about 1.6e-40). Other doctors have similar statistics which I wrote about in my substack article No plausible explanation given for the lack of symmetry in “rapid onset” autism cases.

Instead, they all use study designs which are virtually guaranteed to find no signal. Denmark is the perfect place to do such studies because of how autism is diagnosed in Denmark. And when you call the Danish researchers out on inappropriate methods, they block you. That’s how scientific disagreements are resolved: through censorship.

Finally, nobody can explain the McDowell triplets. The only time in human history that triplets get autism within hours of each other and it happened within 2 hours after their vaccine appointment.