Editor’s note: Earlier this month, research examining the link between the flu and fevers in pregnant women and the risk of autism raised eyebrows across the web and in the media. To get a better assessment, we asked University of Minnesota Physicians psychologist and autism spectrum disorders expert Amy Esler, Ph.D., for her take. Below is what she had to say about the study, outlined well by USA Today in this article.
There is a lot of information out there about the latest autism-related study from Denmark. As a result, I’m sure people have a lot of questions.
The strength of the study in question, published recently in Pediatrics, is that it uses a large population-based sample, followed prospectively from pregnancy, before any signs of autism-spectrum disorder (ASD) would be present. Thirty-one percent of all pregnant women in Denmark between 1996 and 2002 participated in the study.
The women were asked questions about illnesses during pregnancy at two time points during the actual pregnancy and then shortly after the child was born. This helps control recall bias, which occurs when people look back with hindsight and try to find potential causes of their child’s current problems. It also reduces problems with recall in general (i.e., prevents losing data due to parents forgetting what happened during pregnancy).
But what’s important for the general public to understand is that the study and its researchers actually argue the opposite of what is now appearing in headlines.
First and foremost, because they ran so many different tests and comparisons during their analyses, it increased the likelihood of finding something significant just due to random chance. This is a statistical problem called “multiple testing,” and it means that the more “tests” you run on a sample, the more likely it is you’ll get positive findings that aren’t really real. The authors state their results might have been due to random chance.
Second, assuming the findings are real, although it sounds scary, a two- or three-fold increase still would result in very low chances of having a child with autism.
The population rate of a child being born with autism spectrum disorder in Denmark (as well as in the United States) is one percent, so the results of the study, if accurate, would increase the chances to two or three percent. Also, the findings about having a three-fold increase if the woman had a fever lasting longer than seven days could have been an indicator of severity of illness rather than an actual link between fever and ASD.
Similarly, the finding on antibiotics, which actually were associated with a much smaller increase in risk, could also have been a marker of illness severity rather than a link between antibiotics and ASD.
As the CDC said in the USA Today article outlining , there are many reasons beyond ASD concerns to treat fevers during pregnancy and to vaccinate against influenza. The findings of this study do not change those recommendations.