Sunday, March 25, 2012

Autism Risk In Half Siblings

If you have one child with autism, what is the chance that you will have a second child with autism?  It turns out the answer to that question (like everything to do with autism) is it depends.

It depends on whether the children are identical or fraternal twins.  It depends on how many of your older children have autism.  It might depend on whether the older children with autism are boys or girls.  It might depend on far apart they are in years.

Now, it might depend on whether they are full or half siblings.

According to a new paper, the risk of developing autism if you have a half sibling with autism is roughly half that of a full sibling.  The paper is sparse and light on details because it is a letter to the editor instead of a full-fledged paper, but the results are still interesting nonetheless.

In this paper, researchers looked at two different groups of children - one from the Interactive Autism Network (IAN) and another from the U.S. Autism Centers of Excellence (ACE).  The IAN data is based on parental reports but has been shown to be reliable and has been used in quite a few recent papers.  The ACE data set is (presumably) based on children evaluated at one of the the Autism Centers from around the US and was designed to include more minorities that you would get in a typical group of children with autism (oversampled for minorities).

Both sets were filtered to include only families that had at least one child with autism and at least one additional sibling.  The siblings were separated out into full, half-maternal, and half-paternal and at most one sibling of was randomly selected from each family to be included in the paper.  Although it isn't clear from the paper's text whether a full sibling and half sibling could both be included or whether precedence was given to half siblings.

The breakdown of the number and type of each sibling pair types is in the table below.

As you can (hopefully) see, the risk for half siblings is less than full siblings.

But there is another interesting bit that this data shows - the risk is very different depending on whether you have a mother or father in common.  Having a mother in common means that you have roughly half the risk of a typical sibling while having a father in common gives you roughly the same risk as the general population.

Now, this difference in risk might not be as large as the data suggests because there aren't that many paternal half-siblings in either the IAN or ACE data set compared to maternal half-siblings.  But the maternal half-siblings in the ACE data set, which is much smaller than the IAN data and comparable to the number of paternal half-siblings in both data sets, still shows the elevated risk for the mother.   If the paternal risk were equivalent to or higher than the maternal risk, one or both data sets should have shown an increased risk.

So I think it is safe to assume from this data that the half sibling risk is different depending on whether you have a mother or father in common and that the risk is less than it would be for a full sibling.

The authors of this paper are saying that the reduced but elevated maternal risk shows that autism is strongly genetic.  But I see a different picture in this data.  To me, this data seems to show that not only the mother and father's genetics contribute to the risk but also that there are other major risk factors that combine to create the greater risk inside of a given family.

If there were few other risk factor and it was the genetics of the parents was the major risk factor (i.e. strongly genetic) then you would expect the risk to be elevated and roughly equal for both parents.  And failing that, you would expect the risk for the higher parent to be much closer to the combined risk.  Yet that is not the case here as both the mother and father have a substantially reduced risk but, taking their risks together, you don't get back to the full sibling risk.

So what could explain these results?  Well, I'm not really sure but I can think of three possible explanations.

First, if you look at the ages of the full and half siblings you would see that the full siblings are closer together in age and younger.  A shorter interval between pregnancies has been shown to be associated with an increased risk of autism and autism is more common in younger children than older ones.

Second, the presence of half-siblings suggests a major shift in the family dynamics such as a divorce, marriage, death of spouse, or other event that leads to the parents no longer being together.

Since the original parents are no longer together, it is very likely that at least one of the parents has moved to a different geographic area.  In my experience, the children are much more likely to stay with the mother and the mother is less likely to move to a new area.  So perhaps staying in the same area for later children means she is exposed the same geographic risk factors that gave her an increased risk in the first place.

Other possible changes are a reduced level of parental stress (constant stress, as is seen in failing marriages, can lead to all sorts of health problems), a change in social-economic status (higher social-economic status has been linked to higher rates of autism), parental diet, or any one of a number of lifestyle changes.

Third, this difference in risk could just be an artifact of the data and might not be real at all.  After all, the full sibling rate found in this paper is much less than the most recent (and best) estimate to date (10% vs 18% or higher for families with multiple children with autism).  The current data also doesn't adjust for birth order, an increasing risk of autism with birth years, premature birth, parental age, or any one of the other factors that have been linked to an increased risk of autism.

Regardless, assuming the data has some validity, I think that it suggests that something other than parental genetics increases the risk of autism inside of a family.


Constantino, J N et al. 2012. “Autism recurrence in half siblings: strong support for genetic mechanisms of transmission in ASD.” Molecular psychiatry.

Hallmayer, Joachim et al. 2011. “Genetic Heritability and Shared Environmental Factors Among Twin Pairs With Autism.” Archives of general psychiatry 1-8.

Ozonoff, Sally et al. 2011. “Recurrence risk for autism spectrum disorders: a baby siblings research consortium study.” Pediatrics 128(3):e488-95.

Cheslack-Postava, Keely, Kayuet Liu, and Peter S Bearman. 2011. “Closely spaced pregnancies are associated with increased odds of autism in California sibling births.” Pediatrics 127(2):246-53.


  1. Hi MJ -

    Interesting finding.

    What if, instead, part of this can be explained not by genetics per se, but rather, the impact of what happens in the womb being a critical determinant of autism diagnosis? If the brain differences found in autism are the result of developmental patterns, and those patterns have high concordance with properties of the mother herself; i.e., chemical exposures, stress responses, immune function, metabolic profiles, then in these instances, switching up fathers wouldn't change too much.

    - pD

    1. Hi pD,

      I think you probably correct that the prenatal environment plays a large part in the increased risk. I was even going to list that in the reasons, but I took it out because the current study doesn't really speak directly to it and I couldn't think of a way to work it in that didn't sound like I was "blaming" the mother.

      Although I think that you said it well and the properties you mentioned would vary with time and lifestyle choices.