Ancestry of Pink Disease and Autism

A study was published a few weeks ago that took a look at whether having a family history of pink disease (Infantile Acrondynia) leads to an increased risk of autism.  Pink disease is, as most of you probably know, a form of mercury poising linked to use of teething powders that used to be commonly seen in infants and toddlers.  And so, as would be expected when someone mentions the words "autism" and "mercury" together in the same sentence, this study has produced some rather strong opinions from all corners of the autism world.

I didn't want to write anything about this one until I had the chance to read what the study did and did not say (unlike some other people).  But after reading it, I don't really understand what the big deal is.  The data contained in the paper is interesting, if very limited, but outside of the typical theories that fill the introduction and conclusion sections of almost every paper, there is no direct link made between mercury and autism.

Or to put it another way, even if every bit of actual data in the study is 100% true and having a parent or grandparent who survived pink disease does increase the risk of having autism, it does not speak to why the relationship is there.  So this is just a simple association study saying that A might be associated with B, and a very tenuous one at that.

The core of the data is quite simple.  The researchers sent out surveys via mail and e-mail to approximately 2,300 survivors of pink disease.  The surveys indicated that they were collecting data on the general health outcomes in the descendants of survivors of pink disease instead of identifying the interest in autism specifically.

The surveys asked questions about the number of biological children and grandchildren that each survivor had that lived until at least their fifth birthday.  It asked how many of the children and grandchildren had one or more of the following conditions : autism, Asperger's, ADHD, epilepsy, fragile X, intellectual disability, or Down syndrome.

Out of all of the surveys sent out, only 522 completed, non-duplicate surveys were received back.  If you are looking for problems with the study, this is one of the first ones.  Only a small fraction of the surveys were returned leading to the very real possibility that people who chose to respond were biased in some way.

The breakdown of the responses is as follows.  I am only going to report the autism and ADHD numbers as those are the interesting ones.

There were a total of 1,103 children who had a mean age of 37.1 years old.  In this group, there were 11 children with any form of autism (3 autism, 8 Asperger) and 38 children with ADHD.

There were a total of 1,366 grandchildren who had a mean age of 11.3 years old.  In this group, there were 34 grandchildren with autism (12 autism, 22 Asperger) and 29 grandchildren with ADHD.

The researchers then compared the rates of autism in the grandchildren against other published estimates of autism prevalence from Australia.  There are some other sites out there (some of which didn't read the paper) that are claiming that this is one of the problems with this paper.  The claim is that there is no accepted autism prevalence number so there is nothing to compare against.

These claims are largely baseless as the prevalence estimates used in the paper are from other published research that specifically looked at autism prevalence in Australia in and around the age of the grandchildren.  Furthermore, the estimates used are in line with other published estimates in other countries during the same time period.

Personally, I don't find these comparisons to other prevalence estimates all that useful or meaningful.  I think the raw numbers speaks for themselves even in terms of the most recent autism prevalence estimates.

The bottom line is that there were far more cases of autism in grandchildren than would be expected in a group of this size.  The total rate in this population would be almost 2.5%, or more than double what would be expected even in the youngest children today.  When the grandchildren are grouped according to age, the autism prevalence are -

1 in 25 for 6-12 year olds (398 kids)
1 in 35 for 13-16 year olds (141 kids)
1 in 60 for 16+ year olds (827 kids)

That last point is my own extrapolation from the data included in the study.  As a point of comparison, the accepted expected rate of autism today is about 1 in 110, or at least it is in the US.

The study doesn't go into details about the children, but simple math would show you that about 1 in 100 of the direct children had a form of autism.  That number is far larger than would be expected considering that the majority of these people were born before 1980.  Back then the expected rate would have been something like 4 in 10,000 - not 100 in 10,000.

But as shocking as these numbers are, you can't read too much into them because they are based on a survey.  No one went out and actually verified that the people had the condition that they said they did.  No one checked for any sort of response bias in the data.  No one even checked that the people who responded did actually survive pink disease.

So the results are interesting but hardly earth-shattering.

There are also two interesting artifacts in the data that give me pause.

First, the proportion of people with autism compared to Asperger's is out of balance.  Normally you find one person with Asperger's for each 9 with classic autism or pdd-nos.  Yet in both the children and grandchildren Asperger's made up almost two thirds of the autism group.

Perhaps the survivors whose children had autism were less likely to have grandchildren or were less likely to be included in the sample.  This could especially be true since the response to autism back them was to stick people in an institution for the rest of their lives.

Second, there is something interesting going on with the total number of autism and ADHD cases.  In the children, the percent with either form of autism or ADHD was about 4.5%.  In the grandchildren, this percent was about almost identical at 4.6%.  The two populations aren't the same size but it looks like about half of the increase in the autism group could have come from the decrease in the ADHD group.

But again, this is just survey data, so we can't read too much into these quirks just like we can't read too much into the autism rates.

I am not going to go into the mercury/autism theories as there isn't too much point.

Any possible relationship between pink disease and autism is purely hypothetical at this point.  If you wanted to prove some sort of relationship you would have to do a much better job with data collection and analysis.

And on the flip side, we know almost nothing about the relationship between the type of mercury associated with pink disease and autism.  It is a different type of mercury delivered in different doses in a different way.  You can't look at the existing data on  intramuscular ethyl mercury exposure and draw conclusions about ingested mercury chloride.

References

Shandley, Kerrie, and David W Austin. 2011. “Ancestry of pink disease (infantile acrodynia) identified as a risk factor for autism spectrum disorders.” Journal of toxicology and environmental health. Part A 74(18):1185-94.
PMID : 21797771