Jay Wile
[Originally published in 2014 as Not a CDC Coverup…A Case of Using the Wrong Analytical Method]
The news story headline is astounding: “Fraud at the CDC uncovered, 340% risk of autism hidden from public.” The article says that data from a 2004 CDC study on the relationship between the Measles, Mumps, and Rubella (MMR) vaccine and autism were purposefully hidden so it could deny a relationship between the two. Those hidden data supposedly show that some who got the MMR vaccine were significantly more likely to become autistic than those who didn’t.
According to the article, this shows that the original study is “fraudulent,” and there is now a petition to get the study retracted. It also calls into question the other studies that the CDC often cites to show that there is no relationship between vaccination and autism. And this article has to be reliable. After all, it is on CNN’s website, right?
Well, not exactly.
If you go to the article, you will see “NOT VERIFIED BY CNN” at the top, and you will find a CNN producer note that this website is the network’s “user-generated news community.” So, the article wasn’t written by someone at CNN. It was written by a blogger.
Does that mean it’s not reliable? Of course not. I read several blogs regularly, and I find most of the articles written on them to be very reliable. In fact, I would say that some blogs are more reliable than some standard media outlets! The question, of course, is whether or not this particular blog article is reliable.
When you look into the details, you find that it’s not.
The article’s big claim is that by including data that were supposedly covered up by the CDC, you can find that African American boys have a 340% increased risk of autism if they get the MMR vaccine. This conclusion, however, was “hidden due to pressure from senior officials.” Of course, to make such a claim, someone must have done some sort of study. The article itself tells you nothing about that study, but the CNN Producer Note at the top indicates that it was a study done by Dr. Brian Hooker (a bioengineer) and was published in a journal called Translational Neurodegeneration. An update to that note indicates that the journal has pulled the study, and the journal says this is “because of serious concerns about the validity of its conclusions.”
That doesn’t sound very good, but then again, maybe the journal has been pressured by the CDC to participate in their elaborate coverup. Fortunately, I was able to read the study before it was pulled, and I have to agree with the journal’s decision. The study’s conclusions are obviously wrong because it used the wrong kind of tool to evaluate the data.
How the Research was Conducted
Let’s start with the original study, which is supposedly now “fraudulent” because these new data were uncovered. It looked at 624 children with autism and matched them to 1,824 children who didn’t have autism.
Right away, then, we see that this is a case-controlled study. Such studies are useful because there are probably a lot of factors that determine whether or not a child has autism. In order to compare autistic children with non-autistic children, then, researchers must try to compare “equivalent children.” A case-controlled study tries to do that by specifically looking for non-autistic children that have the same relevant characteristics as the autistic children being studied (school district, the age of the mother at birth, the kind of birth that was experienced, etc.).
Here is what they found:1
Similar proportions of case and control children were vaccinated by the recommended age or shortly after (ie, before 18 months) and before the age by which atypical development is usually recognized in children with autism (ie, 24 months). 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.
So the percentage of children who got the MMR vaccine before the age when most autistic children show signs of autism was the same in both groups. This indicates the MMR vaccine has no effect on autism. However, the authors do note that if you expand the age of vaccination to 36 months, a slightly larger percentage of autistic children (93.4%) had gotten the MMR than non-autistic children (90.6%).
Since this is past the age when most autistic children show signs of autism, the authors suggest that the small difference was probably related to the vaccination requirement for early intervention programs. Non-autistic children wouldn’t have a vaccination requirement until kindergarten, so the parents of the unvaccinated, non-autistic children wouldn’t be forced to deal with vaccination until then. If a parent had an unvaccinated autistic child and wanted early intervention, he or she would have to deal with the issue sooner.
What about the new study that supposedly includes the data that weren’t included in the original study? Why does it come to a different conclusion?
Well, let’s first look at why these data weren’t included in the original study. They were from children who didn’t have a Georgia birth certificate! Remember, the original study was a case-controlled study. It needed to match autistic children to non-autistic children. One of the things you want to match is the conditions that occur at birth. If the child was born in another state, the conditions would likely be different. Thus, you don’t want those children in a study that is trying to match children as closely as possible!
Rather than being the result of some sort of “coverup,” then, the data were excluded because they didn’t meet the criteria of the study. Another kind of study might have been able to use those data, but this kind of study could not.
That brings me to the second problem associated with this new study.
The author used the wrong data analysis method. He treated the data like they were gathered for a cohort study and analyzed them that way. This doesn’t work.
In a cohort study, you need large sample sizes. Since you aren’t trying to match autistic and non-autistic children, you need to hope that all the other factors that contribute to autism average out between the groups you are comparing. The only way to do that is to have a very large sample size. For example, a good cohort study on the MMR vaccine and autism was published in 2002.2 It studied 537,303 children. That’s a good sample size for a cohort study, and it found no relationship between the MMR vaccine and autism.
The number of children in this new study is great for a case-controlled study, but it is very poor for a cohort study. So, using cohort study techniques in a case-controlled study is like using a knife as a screwdriver. You’re just using the wrong tool for the job.
Of course, you don’t have to know this to see that there is a problem with the study. All you have to do is look at the conclusions.
The author saw an increased risk of autism only for African American boys who received the MMR. He didn’t see an increased risk among the other groups that were analyzed. Why? Other studies show that the rate of autism diagnosis is the same among African Americans as whites; it just tends to occur later.3 How can something cause autism in African American boys and not in other children, despite the fact that the condition is roughly equivalent between the two groups?
The reason the new study sees the effect among African American boys is that they represent an even smaller sample. As a result, the analysis technique the author is using is even more unfit for this group. When you use the wrong tool for the job, you often end up with bad results. That’s what has happened here.
I am not surprised that Translational Neurodegeneration pulled the paper. I am surprised they published it in the first place.
References
- Frank DeStefano, Tanya Karapurkar Bhasin, William W. Thompson, Marshalyn Yeargin-Allsopp, Coleen Boyle, “Age at First Measles-Mumps-Rubella Vaccination in Children With Autism and School-Matched Control Subjects: A Population-Based Study in Metropolitan Atlanta ,” Pediatrics 113(2):259 -266, 2004
- Madsen KM, Hviid A, Vestergaard M, Schendel D, Wohlfahrt J, Thorsen P, Olsen J, Melbye M, “A population-based study of measles, mumps, and rubella vaccination and autism,” New England Journal of Medicine 347(19):1477-82, 2002
- Gourdine RM, Baffour TD, and Teasley M, “Autism and the African American community,” Social Work in Public Health 26(4):454-70, 2011
