Published April 23, 2012
About a quarter of children with autism may be taking antidepressants that won't help them, a new study suggests.
Researchers at the University of Michigan analyzed results from both published and unpublished studies (unpublished studies have been completed, but have not appeared in a scientific journal). In all studies, children with autism were randomly assigned to receive serotonin receptor inhibitors (SRIs) or a placebo for at least four weeks.
When the researchers looked at results from the published studies alone, they saw a small but real benefit to the drugs. But when the results were adjusted to take into account the role of unpublished studies, the benefit of the drugs disappeared.
The researchers said the purported benefit of taking SRIs for kids with autism could be due to publication bias, a term used to refer to the selective publication of studies with positive results — or those that show a drug works.
"This research made it clear that the effects of SRI treatment in [autism spectrum disorder] are considerably overrated because of publication bias," the researchers write in an article published today (April 23) in the journal Pediatrics.
Finding effective treatments for autistic children will be difficult if publication bias persists, the researchers said, because it will remain unclear whether drugs are actually effective.
It's also unclear why five of the studies went unpublished. It could be that journals would rather publish studies with positive results, or that studies with negative results tend to take a longer time to get published. The Michigan researchers were able to obtain results from only one of the unpublished studies — the researchers who worked on the other four would not provide this information when it was requested.
It is not uncommon for studies involving children to go unpublished. Another study, also published today in Pediatrics, found that, of the 3,400 clinical trials involving children that have taken place since 2000 and have been registered with the government, about 70 percent have been completed, but less than one-third have been published.
When information from studies is not made available, "trials may be unnecessarily repeated, and the information cannot be used to guide therapy," Dr. Scott Denne, of Indiana University School of Medicine, wrote in a comment about the new Pediatrics studies.
This problem could be solved by mandating that researchers post information from their trials to the same government website that is used to document trial registration (ClinicalTrials.gov), according to the researchers of the study on child clinical trials. Currently, registration of studies involving children and posting of results on the site is voluntary.