Brian Hooker proves Andrew Wakefield wrong about vaccines and autism

Truth

Here we go again.

If there’s anything that ignites the fevered brains (such as they are) of antivaccine activists, it’s a good seeming conspiracy. Indeed, as we’ve seen before, if they can’t find a legitimate one, they’ll either exaggerate one or make one up out of whole cloth. This week, an “alleged” conspiracy has been brewing. It’s really the damnedest thing in that it’s hard to figure out exactly what’s going on. Whatever is going on, though, I would recommend extreme skepticism because two people are involved whose word you would be very foolish to trust on any scientific matter relating to vaccines: Andrew Wakefield and Brian Hooker. It began with a paper published in yet another journal I’ve never heard of, Translational Neurodegeneration, and accelerated last night with the release of a video that claims to name a former high ranking CDC official as a “whistleblower” for the finding that the CDC has been “covering up” (of course!) the “truth” that the MMR vaccine causes autism. This is one that might require multiple posts as information dribbles out and people figure out exactly what is really going on. In the meantime, let’s start with the video, released by Andrew Wakefield’s Autism Media Channel, the same “channel” that tried to make antivaccine hay over the murder of an autistic boy. Here’s the video, which is now also on YouTube:

(Note that I’ve downloaded a copy in case it changes, as apparently it has before.)

A partial transcript can be found at—where else?—that wretched hive of antivaccine scum and quackery, Age of Autism.

The first thing one notices about the video is how intentionally inflammatory it is. The “malfeasance” (if such it is) being discovered is compared unfavorably—yes, unfavorably—to the Tuskegee syphilis experiment, complete with lurid pictures patients suffering from advanced syphilis and the introduction of Peter Buxton, the Public Health Service investigator who blew the whistle on the experiment, because, apparently, to Wakefield and Hooker autism is just like end stage syphilis. (Yes, there is another “whistleblower” in this video.) The “malfeasance” being claimed is that the CDC supposedly covered up the link between MMR and autism in African American boys, hence the puffed up rhetoric about the Tuskegee syphilis experiment. Before I go into that more, there is one thing that bears mentioning here. The video even concludes with the sheer Godwin-y goodness (from an entertainment standpoint) of references to the crimes of Hitler, Stalin, and Pol Pot, because, I guess, autism is just like the mass murder of millions, at least in the minds of Wakefield supporters.

Of course, the key finding in Brian Hooker’s paper is that Wakefield was wrong. Indeed, in this video, Wakefield even admits that he was mostly wrong about MMR and autism. Let that sink in again. He admits that he was mostly wrong about MMR and autism. OK, he says we were “partially right,” but the flip side of that is that he must have been mostly wrong. What do I mean? I’ll explain.

The claims being put forward in the video is that earlier MMR vaccination is associated with an increased risk of autism in African-American boys and that the CDC has spent the last 13 years covering this linkage up. These charges are based the result of a “reanalysis” by Brian Hooker in Translational Neurodegeneration entitled Measles-mumps-rubella vaccination timing and autism among young african american boys: a reanalysis of CDC data. The study which has been “reanalyzed” is from a study by DeStefano et al in 2004 published in Pediatrics entitled Age at first measles-mumps-rubella vaccination in children with autism and school-matched control subjects: a population-based study in metropolitan Atlanta. That study was a case-control study in which age at first MMR vaccination was compared between autistic “cases” and neurotypical controls. Vaccination data were abstracted from immunization forms required for school entry, and records of children who were born in Georgia were linked to Georgia birth certificates for information on maternal and birth factors. Basically, no significant associations were found between the age cutoffs examined and the risk of autism. I note that, even in this “reanalysis” by Brian Hooker, there still isn’t any such correlation for children who are not African American boys.

I’ll get into the issues with this study a bit more later, but I’ll admit up front that it’s hard to go too deeply into this study without a statistician and access to the actual restricted data set from the CDC used. Not having access to the data set, I have no way of knowing if the analyses Hooker used were appropriate. However, for purposes of this post, let’s just, for a thought experiment, assume Hooker’s study comes to a valid conclusion (which is, given that it’s Hooker, highly unlikely, but stay with me for a moment). If that were the case, these results are no reassurance whatsoever to the vast majority of antivaccinationists supporting Wakefield. This study says nothing whatsoever about, for instance, Jenny McCarthy and her son’s autism, other than that there is no link between MMR and autism for children like him. Remember, the most vocal antivaccinationists jumping all over this are not African-American but instead tend to be UMC or even highly affluent Caucasians. There’s absolutely nothing in even Hooker’s ham-fisted “reanalysis” of this data to tell them that the MMR vaccine caused their children’s autism.

All there is left is a chance to hype up the conspiracy mongering machine against the hated CDC, because even Hooker’s reanalysis doesn’t support an increased risk of autism with earlier MMR vaccination in white babies. Zero. Nada. Zip. This leads to a bunch of “Tuskegee” handwaving to hide that finding, that even taking his best shot at it the most Hooker could come up with after he tortured the data was a correlation between age of MMR vaccination and autism in African Americans babies—and not just African-American babies, but African-American male babies. Even taken at face value, Hooker et al is a disaster for the vast majority of antivaccine activists. This can’t be repeated often enough. But does the study support an increased risk for African American males, as claimed?

There are a couple of things you have to remember whenever looking at a study that is billed as a “reanalysis” of an existing data set that’s already been published. The first is that no one—I mean no one—”reanalyzes” such a dataset unless he has an ax to grind and disagrees with the results of the original analysis so strongly that he is willing to go through the trouble of getting institutional review board (IRB) approval, as Hooker did from Simpson University, going to the CDC to get this dataset, and then analyzing it. Think, for instance, the infamous “reanalysis” by homeopaths of the meta-analysis of Shang et al that concluded that the effects of homeopathy are placebo effects. The reanalysis did not refute the original meta-analysis. The second thing you have to remember is that it’s pretty uncommon for such a “reanalysis” to refute the original analysis. Certainly, antivaccine “researchers” like Hooker try to do this all the time. Occasionally they get their results published in a bottom-feeding peer-reviewed journal (Translational Neurodegeneration doesn’t even appear to have an impact factor yet), as Hooker has. It means little.

So what about the paper itself? First, one has to go back to Destefano et al 2004. Basically, this was a case-control study in which 624 case children were identified from multiple sources and matched to 1,824 control children on age, gender, and school. In case control studies, researchers look at (usually) two groups of people, cases (those with the condition under study) and controls (those without) and tries to match them as closely as possible to every other confounding factor except for the risk factor being studied, in this case, the age of receiving first MMR vaccination. Porta’s Dictionary of Epidemiology defines the case-control study as: “an observational epidemiological study of persons with the disease (or another outcome variable) of interest and a suitable control group of persons without the disease (comparison group, reference group). The potential relationship of a suspected risk factor or an attribute to the disease is examined by comparing the diseased and nondiseased subjects with regard to how frequently the factor or attribute is present (or, if quantitative, the levels of the attribute) in each of the groups (diseased and nondiseased).” They then see if that risk factor is higher in the case population than it is in the control population. This is in comparison to cohort studies, in which researchers look at groups of people who vary in exposure to a given putative risk factor (for instance, vaccines), each controlled for every other potential risk factor that the authors can control for, and then determine if the condition for which that putative risk factor is suspected to be a risk factor for. A cohort study can be retrospective (looking at existing data) or prospective (the cohorts determined in advance and then followed over time), while case control studies are retrospective.

It’s also not uncommon for epidemiologists to choose more controls than cases in case-control studies. In any case, one thing Destefano et al did was to perform a case control study of children in metropolitan Atlanta looking at age at first MMR vaccination (0-11 months; 12-17 months; 18-23 months; 24-29 months; 30-35 months; and 36+ months). They found no statistically significant correlations. They also looked at a subgroup of the groups, children for whom a Georgia birth certificate could be located, in order to test correlations for other traits:

We matched 355 (56%) case and 1020 (56%) control children to Georgia state birth certificate records, which allowed us to obtain additional information, such as each child’s birth weight and gestational age and the mother’s parity, age, race, and education.

There was no significant correlation noted in various groups based on race, maternal age, maternal education, and birth weight. It’s all pretty straightforward, at least a straightforward as a epidemiological study can be. The only hint of a whiff of anything in it helpful to antivaccinationists was this:

Vaccination before 36 months was more common among case children than control children, especially among children 3 to 5 years of age, likely reflecting immunization requirements for enrollment in early intervention programs.

In other words, it’s a result that is likely not due to an actual effect.

Fast forward to Brian Hooker’s study. The first thing I noticed reading it was that it contains a lot of the usual red flags of antivaccine papers. Hooker cites several Mark Geier papers as evidence of a correlation between vaccines and autism, to try to make it seem as though there is an actual scientific controversy. He even cites a Wakefield paper.

Then there is the methods section. It’s really not very clear exactly what Hooker did with this dataset, other than muck around with it using SAS® software. He keeps referring to “cohorts,” which made me wonder right away whether he was not doing the same sort of analysis as Destefano. Instead of doing a case control study, it looks as though he did a cohort study:

The Pearson’s chi -squared test contained in the SAS® software was utilized for current statistical analyses, and a two-sided p-value <  0.05 was considered statistically significant. This is in contrast to the original Destefano et al. [14] (CDC) study, where a case–control study design was used, where 3 control children were matched to each case child, and analyzed using conditional logistic regression dichotomized for the three age cut-offs at 18, 24 and 36 months...In the present study, frequencies of cases were determined for first MMR ages of less than versus greater than 18 months, 24 months and 36 months in each separate analysis.

Yep, he did a cohort study. Basically, he looked at the risk of an autism diagnosis in the groups first exposed to MMR at different age ranges. Remember, case control = comparing risk factor frequency in people with a condition compared to controls; cohort = examining risk of condition in people with different exposures.

There’s an old saying in epidemiology (and in science in general) that says that if you torture data enough, eventually they will confess. With this in mind, it’s hard not to think of Brian Hooker as the Spanish Inquisition. I find it very telling that Hooker couldn’t find (or didn’t bother to look for) coauthor who is an actual epidemiologist or statistician. Heck, Jake Crosby wasn’t available? He might not be an epidemiologist, but he does have a degree in epidemiology and is in graduate school. What training in epidemiology or statistics does Brian Hooker have that qualifies him to do a retrospective cohort study like this? None that I can see. My first rule of thumb doing anything involving anything more complicated than the rudimentary statistics that I use to analyze laboratory experiments (such as even a “simple” clinical trial) is to find a statistician. While it’s true that Hooker used to lead a high throughput biology team, which likely required some statistical expertise, that’s a different sort of statistics and experimental design than epidemiology. Basically, if you’re going to do epidemiology, you should find an epidemiologist to collaborate with, and if you’re going to do something that requires some heavy statistical lifting you really need to get a statistician on board as well before you start the study.

So is Hooker’s result valid? Was there really a 3.36-fold increased risk for autism in African-American males who received MMR vaccination before the age of 36 months in this dataset? Who knows? Hooker analyzed a dataset collected to be analyzed by a case-control method using a cohort design. Then he did multiple subset analyses, which, of course, are prone to false positives. As we also say, if you slice and dice the evidence more and more finely, eventually you will find apparent correlations that might or might not be real. In this case, I doubt Hooker’s correlation is real. More importantly, even if his statistics were correctly done, his changing the design is highly suspect, particularly when coupled with claims being promulgated by our good buddy Jake Crosby that the CDC intentionally left out subjects in order to hide this correlation:

According to Dr. Hooker, the CDC whistleblower informant — who wishes to remain anonymous — guided him to evidence that a statistically significant relationship between the age the MMR vaccine was first given and autism incidence in African-American boys was hidden by CDC researchers. After data were gathered on 2,583 children living in Atlanta, Georgia who were born between 1986 and 1993, CDC researchers excluded children that did not have a valid State of Georgia birth certificate — reducing the sample size being studied by 41%. Hooker explains that by introducing this arbitrary criteria into the analysis, the cohort size was sharply reduced, eliminating the statistical power of the findings and negating the strong MMR-autism link in African American boys.

This is an accusation neither Hooker’s study nor anything any antivaccinationist has published thus far provides any tangible evidence for. Requiring the birth certificate was not an “arbitrary” criterion, either. It allowed investigators to account for known confounders related to autism risk, such as birth weight, at least in this subset of the case and control groups. I also can’t help but think there is likely to be a confounder that is unaccounted for in this study, particularly given how there increase in risk is found in only one group. In fact, as Reuben at The Poxes Blog explains, there almost certainly was just such a confounder:

Next come the statistics. Hooker uses Pearson’s chi squared test to see if there is a significant association between MMR and autism in children at different ages. DeStefano et al used conditional logistic regression. For the non-biostatisticians out there, the technique that DeStefano et al used accounts for confounders and effect modifiers, different traits in their population that could skew the results. Hooker’s technique doesn’t really do that, unless you stratify results and use very, very large datasets. Hooker’s approach is more “conservative,” meaning that it will detect small effects and amplify them, and those effects can come from anything.

In other words, Hooker used a method prone to false positives. Then:

The nail in the coffin for the Hooker paper is that autism is usually diagnosed by the time a child is three years old. There was no increased risk at 18 months, higher but not by a whole lot at 24, and then the three-fold increase at 36 months. Gee, was it the MMR vaccine, mister? No, the effect is being modified by age. It’s as if I asked you if your shoe size was bigger at 36 months because you drank milk vs because you were 36 months. It’s age. It’s the way that autism is diagnosed. You’re going to have more children diagnosed as autistic at 36 months than you will at 18 months or at 24 months. Using the chi square test doesn’t tease this out, Dr. Hooker! That’s more than likely why DeStefano et al used conditional logistic regression, to take age into account in the analysis.

So why did we not see this in the other ethnic groups or in girls? The answer here is simple, again. Hooker had a limited dataset to work with when he boiled it down to African-American baby boys. In this table, for example, he tells us that he had to modify the analysis to 31 months instead of 36 because he had less than 5 children in that group. It’s the same goddamned mistake that Andrew Jeremy Wakefield wanted to pass off as legitimate science. You cannot, and must not use small numbers to make big assertions…

Quite right. I should have seen that right off the bat.

And, of course, there’s no biologically plausible reason why there would be an effect observed in African-Americans but no other race and, more specifically than that, in African-American males. In the discussion, Hooker does a bunch of handwaving about lower vitamin D levels and the like in African American boys, but there really isn’t a biologically plausible mechanism to account for his observation, suggesting that it’s probably spurious. Finally, even if Destefano et al is thrown out, it’s just one study. There are multiple other studies, many much larger than this one, that failed to find a correlation between MMR and autism. Even if Hooker succeeded in “knocking out” Destefano et al, it doesn’t invalidate all that other evidence.

Hooker’s “reanalysis” aside, somehow, some way, a senior CDC scientist has made the massive mistake of speaking with Brian Hooker. That CDC scientist is William Thompson, well-respected (until, possibly, now) scientist and co-author of Destefano et al, as well as first author on a widely cited NEJM study showing no correlation between thimerosal in vaccines and neurodevelopmental disorders, among other studies. The first thing I noticed listening to Thompson in Wakefield’s video is just how little he is quoted. Instead he’s paraphrased by Hooker, who portrays himself as Thompson’s “confessor” to whom Thompson is “confessing.” The parts with Thompson’s voice appear highly edited, brief sound bites. They sound, at least the way they are presented, highly damning on first listen. It seems very odd on first listen. Heck, it sounds very odd on second listen. So what really happened? Again, who knows? You’ll excuse me if I reserve judgment until more information comes in from sources other than Andrew Wakefield and Brian Hooker because I suspect that what we’re seeing is a highly one-sided presentation of cherry picked information. It is, after all, Wakefield and Hooker we’re talking about.

In the meantime, remember this. Even if Hooker is “right,” he has just undermined the MMR-autism hypothesis and proven Wakefield wrong, with the possible (and unlikely) exception of a single group, African American males. Given the dubiousness of his analysis and background, he hasn’t even demonstrated it for them, either, particularly given the copious other studies that have failed to find a correlation between MMR and autism. What he has done, apparently, is found grist for a perfect conspiracy theory to demonize the CDC, play the race card in a truly despicable fashion, and cast fear, uncertainty, and doubt about the CDC vaccination program, knowing that most of the white antivaccine activists who support hate the CDC so much that they won’t notice that even Hooker’s reanalysis doesn’t support their belief that vaccines caused the autism in their children.

ADDENDUM:

Whoa. The beat goes on and on and on. Our “buddy” Jake Crosby is now claiming that Andrew Wakefield betrayed William Thompson by identifying him in the video without Thompson’s permission:

Andrew Wakefield has betrayed the CDC whistleblower by releasing his name without his consent. On the Autism Media Channel website, a video hosted by Wakefield is up announcing the whistleblower’s name and playing recordings of his voice. In the video is scientist and parent Dr. Brian Hooker, who had been in discussions with the whistleblower and made the catastrophic mistake of sharing his identity with Wakefield. Complicit in the betrayal is Age of Autism, which is promoting Wakefield’s video while repeating the whistleblower’s name.

In commentary to a small group of people later relayed to Autism Investigated, attorney Robert Krakow commented:

“I am very familiar with the information [whistleblower] offered. Disclosure of [whistleblower]’s existence and identity at this point in time is a colossal blunder and an inexplicable error in judgment that damages irreparably the opportunity to use [whistleblower]’s very valuable information and testimony effectively. I know that Brian Hooker did not make the disclosure.”

It remains truly ironic that Andrew Wakefield – a man betrayed by the Lancet editor a decade ago – would turn around and betray the trust of someone who has come forward with valuable information about the fraud committed in a federal agency. Also ironic is that Wakefield similarly betrayed the trust of Dr. Brian Hooker, whose congressional activities have been repeatedly undermined by groups associated with Age of Autism.

Given Jake’s past reliability, take this with a grain of salt. It does have plausibility in that, given Wakefield’s history I could totally see him betraying a “whistleblower” like that. However, I don’t for a moment believe that, if it’s true that Wakefield “outed” Thompson without his permission, Hooker didn’t know about it and at least acquiesce, if not actually approve. I mean, seriously. Hooker spent a lot of time in front of the camera bragging about how he had become William Thompson’s “confessor” and relating what Thompson had said to him, and he didn’t think Wakefield would use that footage? He’s either complicit or a irredeemably stupid and gullible. Take your pick. Of course, Jake is Brian Hooker’s best bud these days and admires him far out of proportion to any reason there might be to admire Hooker; my guess is that he wants to protect Hooker from this allegation, and that’s why he made the claim that Hooker knew nothing.

ADDENDUM #2:

Well, well, well. Andy Wakefield jumps in:

Jake, in light of your serious allegation that I “betrayed” the whistleblower by disclosing his name without permission, I asked you how you knew this. You replied “my sources”. There are no sources other than the whistleblower himself that would know whether or not this was the case since this matter was discussed between the two of us, in several of our conversations. I did, as a matter of fact, have his full knowledge and his permission to do what “I felt best” with the uncensored video.

Although it is a moot point, I remain perplexed as to quite why a man who participated in a prolonged scientific fraud – one that likely resulted in harm to many children – should merit anonymity on his terms. The misinformation you have put out has spread through an already divided community. Your reporting on this matter does you no credit.

So that’s Wakefield’s story, is it? That Thompson gave him permission to “do what he felt best,” but that Thompson doesn’t deserve anonymity on his terms. Why is it that I don’t believe Wakefield’s first claim in light of his attitude expressed in the second part of his comment?