Dietary Effects on Autistic Behavior

An illustration of survey data for various diets parents have tried

It is not an uncommon strategy for parents to change an autistic child’s diet in the hope that it will result in a change in the autistic child’s behavior. Since 1967, the Autism Research Institute has been collecting survey data from parents of kids with autism on the efficacy of a range of interventions, including various diets. Below, I’ve graphed their most recent data that I could find relating to diets:

More than 27,000 surveys were collected for 11 different diets in this data set. This is impressive in terms of quantity, but limited in terms of quality for a couple major reasons.

First, any data set based on survey data is prone to errors that arise from a host of biases. In this case, confirmation bias is most relevant. Wikipedia will tell you that confirmation bias is the “tendency to search for, interpret, favor, and recall information in a way that confirms one’s preexisting beliefs or hypotheses.”

In other words, parents who have invested effort into revamping their autistic child’s diet are also the evaluators of how effective said diet is. These dietary changes are often arduous and motivated by the understandable hope for improvement, and that may effect how parents view the outcome.

And second, even in the absence of confirmation bias, this kind of survey can only demonstrate correlation — not causation. The surveys don’t account for factors such as whether the children were also receiving therapy or other interventions; or the cognitive development and behavioral changes that take place naturally with age in autistic and non-autistic children alike. Without the controls in place that you would expect in a clinical study, one can only assume which factors are and are not responsible for any signs of improvement.

It may be worth noting that the report also included data on 28 different supplements and 53 different pharmaceuticals. With supplements, parents reported data similar to that of diets, being effective 47% of the time, having no effect 46% of the time and having a negative effect 7% of the time, on average. Pharmaceuticals were less similar, being effective 32% of the time, having no effect 42% of the time and having a negative effect 26% of the time, on average.

For more information about this data, the diets, the theories and speculation behind their use, and survey data on other interventions besides diets, I suggest the full 53-page report by Arizona State University engineering professor James Adams, published in 2013. You can find it here.

Next, I’ll take a look at a rigorous controlled study published recently on the effectiveness of what is perhaps the most popular diet for autistic children: the gluten-free/casein-free (GFCF) diet [link here].