Antidepressants probably don't help kids with autism as much as has been believed, a new research review shows.

Kids with autism often repeat certain words or gestures. They might rock back and forth, or rapidly flick a hand over and over again. Or they may become fixated on a single thought, continually making the same point about the weather: "It's hot outside!"

Repetitive behaviors can be worrisome for parents, and they can keep autistic kids from being able to focus at school and participate in regular daily activities. Doctors frequently prescribe antidepressants called serotonin receptor inhibitors (SRIs) in an effort to help ease these kinds of symptoms. SRIs include the class of medications called selective serotonin reuptake inhibitors (SSRIs).

But a new research review shows the benefits of antidepressants in autism may have been overestimated.

That's because all the rigorous studies comparing antidepressants to have found that the drugs have small benefits. An equal number of studies, however, have been finished but not published. That means doctors and patients may not have all the facts when considering whether to use the drugs.

"When we realized there were as many unpublished studies with data as there were published studies providing data, it was definitely a little frustrating," says researcher Melisa Carrasco, PhD, a recent graduate of the neuroscience program at the University of Michigan in Ann Arbor. "It makes you wonder what data is not available and how it could help us better treat kids."
Evidence of 'Publication Bias'

The review, which is published in the journal Pediatrics, took a fresh look at five published studies that compared antidepressants such as SSRIs to placebo pills for treating repetitive behaviors in children with autism. Those five studies showed that the medications offered modest benefits, reducing repetitive behaviors by about 22%.

But in addition to the five published studies, researchers also found five others that had not been published, though they had been completed. They were able to get the results of one of the unpublished studies. It didn't measurably change the results of their analysis. Researchers say they can't be sure what the other studies found since they didn't have access to the data. But they believe the existence of so many studies with delayed or unpublished results -- at least as many as have been published -- is concerning.

That's because many medical journals are reluctant to publish studies that show a medication or treatment doesn't work. It's a problem called publication bias. And it's sometimes a good indication that those studies may have found that the medications were not effective.

When researchers adjusted their results to account for the unpublished studies and the risk of publication bias, the apparent benefit of the medications was cut in half. It dropped from reducing symptoms about 22% compared to a placebo, to reducing them just 12%, a difference that could have occurred just by chance and may not have been an actual benefit of taking the medication.