Informal surveys are fraught with problems. There can be all sorts of biases. Many ways for the data to be skewed one way or another. A great example of this was the Generation Rescue phone survey. The results were all over the map, and clearly flawed. Many of the results pointed to higher autism rates in partially vaccinated over fully vaccinated children.
That was four years ago. Recently an internet survey was undertaken to explore vaccine questions. The survey results can be found on the site vaccineinjury.info. People with agendas certainly can do surveys, but they shouldn’t be surprised when their agendas are pointed out. I won’t go into the survey or results in detail. I have a suspicion the results will be analyzed elsewhere. Rather, let’s just look at the autism results.
Here is the age distribution of the (mostly) children reported on in the study. Very skewed towards the very young.
Here is the distribution of an age distribution of the fraction of kids in each group reported to be autistic (click to make bigger)
If you are wondering, “did they just publish a graph showing high autism ‘prevalences’ in unvaccinated kids?”, you are correct. They are showing that in some age groups the fraction of unvaccinated autistic kids is about 2%. The current prevalence estimate for the U.S. and the U.K. is about 1%.
Do we really want to put any weight on this result? No, not really. It is a nicely packaged internet survey, but it is an internet survey after all. Can we speculate a little as to what this means? Sure. It could mean that the groups that were recruited for this survey included more families with autistic children. Given the high recurrence risk of autism (i.e. the high chances that a younger sibling of an autistic kid would be autistic) it is not only reasonable, but predicted, that such a survey might show a high fraction of autistic kids.
Keep something in mind–the number of autistic kids is small. If I did the math correctly, there are only 37 autistic kids total. [edit to add–this is incorrect. There are 44. I left out two age groups in my total] That means big uncertainties (error bars) in the “results”. Results which, as we’ve already discussed, are pretty skewed just by the design and limitations of the survey.
Must be unexpected for the people doing the survey. Rather than show a low autism prevalence, they show a high one. For those who claim (sometimes over and over) that there are no unvaccinated autistic kids, here is another piece of evidence that they are wrong.
It’s clear enough that even commenters at the Age of Autism blog have noticed it. From “Sarah”:
Help! What am I doing wrong trying to read these graphs? The one titled autism in UNVACCINATED children shows autism percentages above the commonly accepted level in vaccinated children, but in the full study, which also shows a similar graphic, they say only 4 children in the study had severe autism. Are the graphs supposed to be the levels in vaccinated children? What am I missing here? Does this study show higher rates of autism among the unvaccinated?
Yes, Sarah, you are seeing correctly. The study shows higher rates of autism amongst the unvaccinated. That’s what the >1% values for ages 3-12 means when compared to the 1% value currently reported for the United States.
And, you can see people trying to “interpret” these results in interesting ways:
Sarah, I see 0.57% for autism roughly half of the rate seen in vaccinated children.
Note also that many who are not vaccinating are doing it because they have a child with autism already and these families have an increased likelihood of another child having autism as well. These numbers could be much lower in a larger survey and the cases are less severe as noted already. I hope this helps.
Yes, if you want to water down the results by averaging, including infants too young to be reliably diagnosed, you can get a lower prevalence. Of course, to do that one has to ask: what is the prevalence of diagnosed autism amongst infants? We don’t know.
The website notes:
Due to the fact that the majority of children in the survey are between 0 and 2 years of age and some diseases generally do not appear in this age group, the results are subdivided into different age groups (click on the graphic). Information about country, gender, age, age distribution, breastfeeding, preferred tretment[sic] can be found here.
And they are correct. There are no autistic kids in the age 0-2 group reported. Looking at just the ages 3-18 kids, there are 4326 kids, making a “prevalence” of 0.85, if you want to average. That would be within the error bars of the current autism prevalence estimate in the U.S..
As I’ve noted, the Age of Autism blog has already discussed this (thanks to those who sent me the link). I doubt this will get much play from these groups as the results are really not good from their point of view.


Recent Comments