Wednesday, April 09, 2014

Autism And Obese Fathers. Today's Research Popularization Critique.

A Norwegian study has found a statistically significant correlation between autism in children and their  fathers'  obesity:

A Pediatrics study finds the risk for autism spectrum disorder strongly associated with paternal obesity (that is, with a BMI of 30 or more).
Investigators examined the relationship in a Norwegian cohort of some 93,000 children, whose health was tracked in national databases. By the end of follow-up, at a mean age of 7 years, the odds ratio for autism spectrum disorder among children of obese fathers was 1.53, compared with those of normal-weight fathers. Maternal obesity was not significantly associated.
The authors express surprise at the findings, which suggest an unidentified genetic or epigenetic mechanism.
I would be a bit hesitant about interpreting the results at face value until they have been replicated in other studies.  The reason, oddly, is that very large sample of data (93,000) that they work with*.  Even though a larger sample is usually  better than a smaller sample, analyses which use very large samples can sometimes be too good a thing.  For instance, they can find differences which are tiny or unimportant in practical terms statistically significant, because (from the comments to the linked post by its author):

In short, it happens because a large sample size greatly increases the power of an analysis to detect a difference. And THAT occurs because the test statistic that determines statistical significance is based on a formula that, in effect, gets multiplied by a value based on the sample size (such as the square root of the sample size). So, assuming a given difference and a given standard deviation for the data, increasing the value of sample size (n) increases the test statistic and results in an increased likelihood of statistical significance. Therefore, a humongous sample size can "override" a difference of very small size, inflating the test statistic, and result in statistical significance, even though the difference is so small that it has no practical consequence.
Indeed, the head researcher warns us about taking the results as definitive, though he also seems to believe in a genetic or epigenetic explanation.

But what I really wanted to talk about is the tentative interpretation of what to do IF the results are meaningful in practical terms.  An example from Fox News site:

The researchers noted that more research needs to be done to confirm the link between paternal obesity and ASD in children.
"The potential effects should be further investigated through attempts at replication of our analysis, and, if these are positive, through genetic and epigenetic studies," the researchers said. "It should also be explored whether paternal overweight and obesity are associated with an increased risk of other neurodevelopmental disorders in children."
Dr. Andrew Adesman,  chief of developmental and behavioral pediatrics at Steven & Alexandra Cohen Children's Medical Center of New York in Lake Success, N.Y., noted that the impact of paternal obesity on a child’s risk of developing ASD is relatively small.
"Most of the children with autism in this study were not born to obese fathers and most of the children born to obese fathers did not develop autism," Adesman, who did not work on the study, told MedPage Today. "The risk (for autistic disorder) increased from 15 per 10,000 cases (children with normal-weight dads) to 25 in 10,000 cases (children with obese dads), which is still very, very low."

In short, we need more research, we need to note that the effect is quite small overall and so on.  All good and careful.

Now have a look at this 2012 Science Daily summary of somewhat similar research but connecting obesity in mothers with their children's autism**. Then compare it to the 2014 summary of the Norwegian study, also at the Science Daily site. Granted, there are all sorts of reason for the different tone in the two summaries, but I tend to think that one reason has to do with the responsibility mothers are assumed to have for their children's well-being.

Thus, several sites in 2012 discussed the importance for mothers to affect their own obesity.  Examples:

"The fetus depends on the mother for nutrients," noted Dr. Hertz-Picciotto. "So at certain times in the fetal development if sugar levels or other nutrients are too high or too low, the imbalance can affect the fetus, especially when it comes to the brain."
But Hertz-Picciotto said there was good news from this study.
"The best thing about this is a lot of this can be modified," she said. "If you are thinking about getting pregnant, watch your weight and if you have diabetes have your doctor keep a close eye on you so you keep your glucose under control while you're carrying your baby."


Although the study cannot prove that one condition causes the other, its authors caution that even the possibility is worrisome in the light of rising U.S. obesity rates.
"If there is anything you can do to make yourself healthier, this is yet another reason for moms to consider," said Paula Krakowiak, a researcher at the University of California, Davis, who led the study.

After about an hour's worth of search*** I found one somewhat similar warning after the dads-and-autism study:

In short, we have a long way to go before we figure out what causes autism spectrum disorders. Until that time comes, it's probably a good idea no matter what to try to maintain a healthy weight!

The tones of popularizations can vary in stark ways and they can vary in more subtle ways.  It's unlikely that the writers create the tone differences on purpose.  The reasons are probably in our upbringing, the assumption that mothers are responsible for children's health and the very strong focus on mothers, not on fathers, in medical research about children.  For example, the 2012 study about maternal morbidity didn't use data on the fathers' obesity levels, probably because it wasn't available.  But that omission meant that any paternal obesity effects might be combined with the maternal obesity levels if partners are both more likely to be obese or not to be obese.   And all of them would be attributed to the mother.

*I'm assuming that they use the whole data set, based on the study abstract.  I haven't read the whole study (too many $$ needed for access).
**My post on that in 2012 is here.
***I'm prepared to be corrected if a longer search unearths more health sermons for fathers.