Tuesday, April 10, 2012

On Autistic Children and Obese Mothers

A study in the Pediatrics journal has found a statistically significant correlation between mother's obesity and autism-related disorders in the child:
Krakowiak and her colleagues looked at 1004 children who were between two and five years old, born in California and already participating in a study underway at UC Davis.
Of those kids, 517 had an autism spectrum disorder and 172 had developmental delays. For Krakowiak's study, the children's diagnoses were confirmed by a reevaluation at the UC Davis MIND Institute.
Autism is attributed to atypical brain development and characterized by a group of symptoms that include problems with socialization, communication and behavior.
Milder versions of autism, such as Asperger's syndrome and related conditions, form a "spectrum" of autism-related disorders. In addition, impairments in any one of the autism-related cognitive skill areas are considered developmental delays.
Among the kids in the study with an autism spectrum disorder, 48 were born to mothers with Type 2 or gestational diabetes, 111 to mothers who were obese and 148 to moms with any sort of metabolic condition, like high blood pressure.
For children with a developmental delay, 20 were born to mothers with Type 2 or gestational diabetes, 41 to mothers who were obese and 60 to moms with any sort of metabolic condition.
Overall, the connection between diabetes in a mother and her child being diagnosed with autism was not significant, but the researchers did find links between a mother being obese or having any other metabolic condition and her child having autism.
Developmental delays were associated with both obesity and diabetes, along with having any other metabolic condition.
"There is definitely an association present and it adds to the reasons for finding ways to lower obesity rates or diabetes rates and make greater efforts to change lifestyle factors," said Krakowiak.

And what are the policy implications? Krakowiak and her colleagues:
Meanwhile, Krakowiak and her colleagues note that nearly 60 percent of U.S. women of childbearing age (20-39 years) are overweight, one-third are obese and 16 percent have so-called metabolic syndrome -- a constellation of symptoms, including high blood pressure and insulin resistance, that raise heart disease risk.
Although no one can say the nation's rising obesity rate is to blame for the prevalence of autism, Krakowiak said the parallel increases did catch her attention.
The study itself is available for reading. The initial study design matches the mother-child pairs with control mother-child pairs where the children have typical development, without any autism or developmental delays. Here are the frequencies of the mothers' Body Mass Index in the autistic-children group and the typical-development-children group:
Percent underweight (BMI less than 18.5)
Autistic group 3.7%
Control group 3.2%

Percent normal weight (BMI between 18.5 and 24.9, inclusive)
Autistic group 52.0%
Control group 54.6%

Percent overweight (BMI between 25.0 and 29.9, inclusive)
Autistic group 22.8%
Control group 27.9%

Percent obese (BMI 30.00 or more)
Autistic group 21.5%
Control group 14.3%

Those frequencies are worth a closer look. The percent of obese mothers is clearly higher in the autistic group. But the percent of overweight-but-not-obese mothers is actually higher in the control group.

I'm nitpicking here, sure. But given how studies like these are interpreted, it's worth being careful with the numbers and also worth noting that the majority of mothers in the autistic group (55.7%) were not overweight or obese. In short, autism would not be cured if every single woman belonged to those weight categories.

As the researchers point out, it's difficult to know what this correlation means, and further research is necessary.

Speaking of further necessary research, I'm always astonished by the way fathers are eradicated in these kinds of studies. This study, for instance, uses good controls on all sorts of demographic and sociological variables which might intervene with the identification of autism in a child, say, including data on the mother's education level. But there is no control data on the fathers. None whatsoever.

That rules out some theories about altruism at the starting line (anything having to do with the sperm). If we rarely look at data on fathers we won't find much, either. I sometimes imagine future people reading the studies of this era and finding them a bit shocking in the hidden assumption that only mothers matter when it comes to children's health and development.