The “CDC whistleblower scandal” which erupted in 2014 is about a supposed cover-up of data linking MMR vaccines to autism. However, leaked documents tell a different story. In this story, actions almost certainly done in good faith are misconstrued as fraud. This distorted interpretation then gets repeated, is never verified, and soon becomes taken as fact. Productive arguments as to why vaccines are still plausible causes of autism are then drowned out by talk of conspiracy. Instead of making progress, vaccine skepticism is set back from rallying behind a poor argument.
The real reason the research community is wrong about vaccines being “proven” not to cause autism has nothing to do with a conspiracy, but rather systematic methodological flaws. The most notable of these is called “confounding by contraindication”, which has the potential to dramatically alter study results. It’s time for vaccine skeptics to ditch harmful conspiracy theories and to take up discussion of actual methodological flaws.
- Timeline of Events In the CDC Whistleblower Scandal
- The 2004 MMR-Autism Study
- Thompson Talks With Hooker
- Hooker Publishes A Reanalysis
- Hooker and Wakefield Go Public
- Thompson Gets a Lawyer, Makes a Statement, Communicates with Posey, and Goes Silent
- Hooker and Wakefield Make Formal Accusations
- Vaxxed Documentary Popularizes the Controversy
- The Documents Are Released Publicly
- Fact Checking the Accusations
- Claim #1: Immunization requirements do not explain the reported vaccine-autism associations.
- Claim #2: That the above immunization requirements explanation is so poor that it suggests deliberate misdirection.
- Claim #3: Race analysis was planned in the overall sample.
- Claim #4: That the above race analysis found an association between MMR and autism in African American boys at 36 months.
- Claim #5: The researchers deviated from the study protocol by not publishing this association.
- Claim #6: The researchers introduced a birth certificate requirement after-the-fact in order to conceal the association.
- Claim # 7: The researchers held a meeting to destroy documents relating to the study in order to cover their tracks.
- Timeline of Events In the CDC Whistleblower Scandal
- Conspiracy Theories Versus Systematic Confounding
Timeline of Events In the CDC Whistleblower Scandal
Before we analyze whether accusations of a cover-up are true, let’s review some notable events in timeline that led to this controversy.
The 2004 MMR-Autism Study
The story starts with a CDC researcher named William Thompson who was involved in a study published in 2004 (DeStefano et al.) that examined whether or not the age at which a child gets MMR vaccination has any association with autism.1 The authors did some unadjusted analyes on an overall sample, and then more adjusted analyses on a smaller sample of just children for whom they had more detailed birth certificate information. They found that autistic and nonautistic children were about equaly likely to be vaccinated before 18 or before 24 months of age. However, in the unadjusted analyses autistic children (specifically boys) were more likely to be vaccinated before 36 months of age compared to controls, especially among children who were 3 to 5 years old. The authors concluded this was likely due to immunization requirements that were enacted around that time. Autistic children in the study needed to be vaccinated in order to attend early intervention or special education programs.
Thompson Talks With Hooker
In 2013 and 2014, Thompson got in touch with a notable advocate for a vaccine-autism connection named Brian Hooker. Thompson shared his misgivings about the CDC and his colleagues with Hooker. He also disclosed that he believed he and his coauthors of the 2004 study buried a statistically significant association (in African American boys) that should have been reported, according to the researchers’ predefined study protocol. Thompson shared internal CDC documents with Hooker pertaining to the 2004 study.
Hooker Publishes A Reanalysis
In August 2014 Hooker went on to publish a reanalysis of the same dataset used in the original 2004 study2, apparently with some encouragement from Thompson. Hooker reported an autism risk only in African American males who were vaccinated not on schedule, but before the age of 3. The paper was soon retracted.
Hooker and Wakefield Go Public
Unknown to Thompson, Hooker had been recording their phone calls. Hooker had shared Thompson’s disclosures with Andrew Wakefield, the first author of the infamous retracted 1998 paper in the Lancet which suggested MMR vaccines might be related to autism. The two, apparently in violation of Thompson’s confidence, made public a narrative that said Thompson was disclosing fraud at the CDC. A short video espousing this was published also in August 2014.3
Thompson Gets a Lawyer, Makes a Statement, Communicates with Posey, and Goes Silent
Thompson, presumably concerned that his anonymity was lost, hired legal counsel. He has been more or less silent ever since. He posted a statement on his lawyers’ website.4 He apparently then also shared internal documents pertaining to the 2004 study with Congressman Posey, who had previously been noted for his receptivity to criticisms of vaccines. Posey read parts of a statement sent to him by Thompson into the public record.5
Hooker and Wakefield Make Formal Accusations
In late 2014 and early 2015, Hooker and Wakefield jointly sent two complaint letters to the HHS Office of Research Integrity, alleging fraud pertaining to the 2004 study.6,7 Hooker also published his own statement much later in 2016.8 Though the two made many claims, it will suffice to review what we identify as seven of the most important ones:
- That the researchers’ explanation that the vaccine-autism associations they reported at 36 months are likely explained by immunization requirements is not founded.
- That the above explanation is so poor that it suggests deliberate misdirection.
- That the researchers planned to do analysis by race in the overall sample.
- That the above race analysis found an association in the overall sample between MMR and autism in African Americans boys at 36 months.
- That the researchers then chose to deviate from their predefined study protocol and omit the above association from publication.
- That the researchers then changed the protocol to introduce a birth certificate requirement, which they knew would comprise a sample less likely retain the effect.
- That the researchers held a secret meeting to destroy documents relating to the study in order to cover their tracks.
Vaxxed Documentary Popularizes the Controversy
Wakefield directed a 2016 documentary called Vaxxed which presented the narrative of a cover-up at the CDC pertaining to the 2004 DeStefano et al. study.9 This resulted in widespread awareness of what has been dubbed the “CDC whistleblower” scandal.
The Documents Are Released Publicly
Congressman Posey’s office has shared Thompson’s documents with at least a couple of other people or groups. One such group is Matt Carey and Dorit Reiss, who are both autism activists (who don’t believe vaccines cause autism). Carey is the present author of the website Left Brain Right Brain. He has the entire cache of documents available as a zip file on the site.10 The most salient screenshots from the documents are also worked into his articles. Wakefield has also provided a copy of the documents on the Vaxxed documentary website.11 In both cases, these websites seem to have made the documents available in early 2016.
Fact Checking the Accusations
That pretty much finishes the timeline of the most pertinent events. Now the question is whether or not any of Hooker and Wakefield’s claims are true? Answering these questions obviously requires knowing what Thompson’s documents contain. Unfortunately, the entire internet has been content to just assume they pointed to a conspiracy, without actually checking them.
If we were to look at any epidemiological study whatsoever with the kind of lens that Hooker and Wakefield are applying, it is likely we would be able to find actions that can be considered suspicious, or at least biased. While some actions taken by DeStefano et al. could be interpreted as being questionable, they are nonetheless still best explained as being actions done in good faith. Saying something is questionable is a far cry from intentional misconduct. With this in mind, and with the help of the documents, let’s go through the accusations:
Claim #1: Immunization requirements do not explain the reported vaccine-autism associations.
Hooker and Wakefield make various arguments as to why the reported significant results in the unadjusted analyses are not spurious, as DeStefano et al. suggest.7 Some of their arguments may be reasonable. For example, Hooker and Wakefield argue that if immunization requirements are the explanation of the reported association, why are boys and girls not affected equally? It’s a valid question. So for this claim, we conclude that Hooker and Wakefield might actually have a point.
This is really reaching on the part of Wakefield and Hooker. The researchers interpretation of the data regarding immunization requirements is actually totally reasonable. For example, it would be hard to come up with a biological explanation as to why it is more dangerous for boys to receive MMR at a later age than at an earlier age (where no significant association was reported). Also, the researchers did explicitly report that there was no statistically significant difference between the reported odds ratios for boys and girls at 36 months, even though the odds ratio for boys was higher. And lastly, the researchers only state that the immunization requirements “likely” explain the association. That is not a definitive dismissal of other possible interpretations.
Claim #3: Race analysis was planned in the overall sample.
The researchers reported on race in the birth certificate sample, but not in the overall sample. Was this in violation of the analysis plan (a.k.a the study protocol)? The released documents contain drafts of the researchers’ analysis plans . There is nothing in them that indicates that any race analysis was ever intended in the overall sample. The closest thing that can be construed as such is a statement in the finalized plan (dated September 5, 2001) saying:
“The only variable available to be assessed as a potential confounder using the entire sample is child’s race.”12
Though that does arguably sound like a race analysis was planned, from the larger context of where that quotation appears, it is clear that this statement is merely describing the fact that the data they have available to analyze was limited, and hence separate analysis of just those with birth certificate data (which provides more information, including more complete information about race, which was missing in about 5% of the overall sample) was needed. But Wakefield and Hooker have misquoted this line in the complaint they sent to the CDC office of research integrity, writing:
“The only variable that will be assessed as a potential confounder using the entire sample will be the child’s race.”6 [Emphasis added]
It is possible that Hooker and Wakefield just made a mistake here, as opposed to intentionally misrepresenting what was stated in the analysis plan. Though it is also arguable that it was intentional…
Thompson posted a statement on his lawyers’ website in 2014. It in part read:
“I regret that my coauthors and I omitted statistically significant information in our 2004 article published in the journal Pediatrics. The omitted data suggested that African American males who received the MMR vaccine before age 36 months were at increased risk for autism. Decisions were made regarding which findings to report after the data were collected, and I believe that the final study protocol was not followed.”4
There is no need to doubt the truthfulness that such a statistically significant association was indeed found and not reported. But that does not mean the association was correct, or that it was misconduct to omit it. Thompson himself wrote in his statement to Congressman Posey:
“The fact that we found a strong statistically significant finding among black males does not mean that there was a true association between the MMR vaccine and autism-like features in this subpopulation.”13
Claim #5: The researchers deviated from the study protocol by not publishing this association.
As already discussed, the analysis plan never stated that a reporting on race was planned in the overall sample. If Thompson believed that it was, he was mistaken. Sometimes gaps in communications happen when working in teams. The written analysis plan does not support Thompson’s claim that the study protocol was not followed. There are valid reasons why epidemiologists would omit a finding in a variety of contexts if they believe the data is unreliable, and if no predefined protocol called for it to be reported. This is commonplace in epidemiology.
In retrospect, it would have been wiser had the researchers reported the significant finding, as that would have given the public the opportunity to scrutinize the association, and determine for themselves that it was not reliable. Having failed to do this, room for doubt was created, which has allowed a conspiracy theory to develop. The researchers decision to omit this statistically significant finding, albeit possibly foolish, was clearly done in good faith and in accordance with their predefined protocol.
A copy of even the earliest draft of the analysis plan (dated April 3, 2001) show that the plan stated:
“For the children born in Georgia for whom we have a birth certificate, we will perform several sub-analyses similar to the main analyses, and will include several additional potentially confounding variables.”10
Seeing as the birth certificate analysis was always planned from the beginning, this accusation is certainly false. So Hooker and Wakefield were wrong. Hopefully this error was not intentional, though it does look suspect…
Thompson says that the researchers scheduled a meeting in 2002 and got together to destroy documents relating to the study. He stated:
“At the bottom of Table 7 it also shows that for the non-birth certificate sample, the adjusted [sic?] race effect statistical significance was huge. All the authors and I met and decided sometime between August and September ’02 not to report any race effects for the paper. Sometime soon after the meeting, we decided to exclude reporting any race effects, the co-authors scheduled a meeting to destroy documents related to the study. The remaining four co-authors all met and brought a big garbage can into the meeting room and reviewed and went through all the hard copy documents that we had thought we should discard and put them in a huge garbage can. However, because I assumed it was illegal and would violate both FOIA and DOJ requests, I kept hard copies of all documents in my office and I retained all associated computer files. This included all the Word files (agendas and manuscript drafts), Excel files with analysis and results, and SAS files that I used to generate the statistical findings. I also kept all my written notes from meetings. All the associated MMR-Autism Study computer files been retained on the Immunization Safety Office computer servers since the inception of the study and they continue to reside there today.”14
There are some problems with the claim that trashing documents was part of a cover-up.15 Firstly, the year is 2002. The CDC has digital copies of everything. And secondly, it is only a requirement to keep enough records so that the study can be recreated from scratch. And we know the CDC met these requirements because Hooker himself performed his reanalysis based on data he got from the CDC directly. The data is potentially available to any researcher who requests it from the CDC.16 Thompson himself said as much. In other words, throwing out unneeded documents could have easily just been housekeeping. Carey points out that most of the 1,000 pages provided to him by Congressman Posey are uninteresting and could easily stand to be discarded, and that such housekeeping is not an unheard of thing in the research world.13
So in short, none of Hooker and Wakefield’s claims are convincing.
Everyone’s Missing the Bigger Picture
There are several other angles to this controversy. For example, why was Hooker’s reanalysis retracted? The journal’s official retraction notice states undisclosed conflicts of interest and concerns over the validity of the methodology.2 Whether or not those concerns are correct, and whether or not a subsequent retraction was in keeping with the journal’s stated retraction policies have been debated.17,18 Endless detail could go into discussing this and a number of other issues.
But the thing is, it doesn’t really matter. Everyone is missing the bigger picture. Something like 1 in 36 children in the USA may have autism.19,20 However much one may debate the accuracy of that estimate, the true number is still going to be a lot. And you can bet most of them are not black. Even if a cover-up had happened, this would not be able to explain why numerous studies done all around the world have not found an association between MMR and autism. Yes, some MMR-autism studies have concerning reported conflicts of interest21 (as well as sometimes dubious methodologies), but some do not. Smeeth et al. for example report no conflicts of interest, and yet they did not find an association between MMR and autism.22 An isolated alleged cover-up simply cannot explain results like this.
Conspiracy Theories Versus Systematic Confounding
So vaccine skeptics need to ask themselves: Are vaccines a primary culprit in the autism epidemic? Or are they not? If they are, then positing a cover-up of an autism risk in just blacks would in no way explain why vaccines are causing autism in everyone else. That leaves a person with three choices:
- Accept the many MMR-autism studies as being correct, and abandon the MMR-autism hypothesis.
- Posit that corruption is pervasive and that every single MMR-autism study is tainted.
- Posit that there is systematic bias/confounding that similarly affects all MMR-autism studies, causing them to give wrong (or at least worthless) results.
Spoiler alert: The correct answer is number three. It doesn’t require widespread corruption for the research community to be chronically wrong. It merely requires just one mistake made repeatedly. Suppose ten people are looking through a glass window that is tinted green. They see the white wall of a building across the street and are asked to report what color the wall is. Every one of them may say that it is green. There would be a consensus and so they would likely conclude that the wall is green. Of course, they are all wrong because none of them realize that the window is tinted, distorting their perceptions.
Confounding by Contraindication – The #1 Reason All MMR-Autism Studies Are Worthless
Confounding by contraindication (CBC) is the foremost systematic source of confounding that invalidates literally all MMR-autism studies. I will be publishing an in-depth peer-reviewed paper on the subject in the future. Read this preview if you want to know more. The rabbit hole runs very deep. But here is the abridged version:
- Parents notice traits in their children that are associated with increased risk of ultimately being diagnosed with autism.
- Such parents are particularly concerned that vaccines may be unsafe for their children.
- Such parents are therefore more likely to not allow their children to receive the MMR vaccine.
- Autism outcomes are therefore concentrated in the control groups of MMR-autism studies.
This is not an unsubstantiated theory. I have documented in great detail that this is happening in several different ways. In other words, for all we know, the real reason MMR is not positively associated with autism in any study is not because MMR doesn’t cause autism, but because autism-associated traits cause avoidance of MMR! Researchers have the whole thing backwards.
So what kind of results would MMR-autism studies have if CBC was properly accounted for? I am still working on a sensitivity analysis going over a variety of possible scenarios. In short, the answer is terrifying…
Calling People Shills Makes the Rest of Us Look Stupid
True, vaccine promoters can be cruel and ignorant. But so can vaccine skeptics. The fact of the matter is, there is no convincing evidence of a cover-up. Yet unfortunately, anywhere this CDC whistleblower scandal is doubted, the comments are sure to be flooded with accusations of people being shills for Big Pharma. Aside from being obnoxious, it makes the rest of us look stupid. Sincere physicians are literally being sued and terrorized online through vile comments and even fake practice reviews, simply for stating their favorable opinions of vaccines.23,24 Worse than that, online groups of vaccine critics have been sometimes found to coordinate attacks on others.25 The result is that important issues like CBC get drowned out by the noise. Just ask yourself, how many people have even thought for a moment about the possibility that some kind of selection bias may exist in MMR-autism studies? The answer is very few.
On one hand, it is understandable and even encouragable to doubt the CDC’s narrative on vaccine safety. For example, the CDC’s website brazenly states that “vaccines do not cause autism”, citing a 2011 report by the Institute of Medicine (IOM).26 Yet such a claim is absurd, considering the IOM report contains grossly incompatible statements like:
“The evidence is inadequate to accept or reject a causal relationship between diphtheria toxoid–, tetanus toxoid–, or acellular pertussis–containing vaccine and autism.”27
So while it is true that the CDC has its obvious bias, be mindful that bias is not the same thing as fraud, especially in any and every particular instance in question. Vaccine skeptics, I am definitely on your side. But it’s time to wake up. Constantly playing the corruption card doesn’t help move things forward. It’s time to let go of the alleged CDC scandal, and start talking about reasonable criticisms like confounding by contraindication. That actually has the potential to turn the tide. Bad arguments and online harassment can never be effective in changing vaccine policy or the direction of research. Only good arguments can.
Conclusion – We’ve Been Duped… By Ourselves
Hooker and Wakefield have construed the 2004 DeStefano et al. study in ways that are demonstrably false. How this happened is debatable. Did they just not read the documents well? Did they intentionally misrepresent them? Have their own personal histories affected their judgement? (In my opinion, Wakefield was excessively maligned following his original retracted 1998 paper. Though not free of fault, the controversy is not nearly so black and white as has been uncritically popularized, and there is much to be said in his defense.28,29) Whatever the case may be, it doesn’t really matter. The bigger issue is that we all allowed ourselves to believe a story without scrutiny, and without realizing that – even if true – it would not at all change the bigger picture. On the other hand, confounding by contraindication does change the bigger picture.