An antidepressant that is among the most popular kinds of medicine used for treating autism didn't work for most kids and caused nightmares and other side-effects, new research found.

Results showed risks with citalopram, sold in Canada and the U.S. as Celexa, outweighed any benefits in the largest published study of medication versus dummy pills for autism.

'We are still so challenged to come up with medications that can address core symptoms.'— Geraldine Dawson

That's according to the lead author, Dr. Bryan King, director of child and adolescent psychiatry at Seattle Children's Hospital and the University of Washington medical school.

The drug is not approved for treating autism. However, many doctors have prescribed it, thinking it might help prevent repetitive behaviours such as spinning, twirling and head-banging that are hallmark autism symptoms.

Similar antidepressants have been shown to help treat repetitive actions in people with obsessive-compulsive disorder.

In the autism study, however, citalopram worked no better than dummy pills. In fact, compared with kids on placebo, those on citalopram were more than twice as likely to develop repetitive behaviours, as well as other side-effects including sleep problems and hyperactivity.

Citalopram is in a class of antidepressants known as selective serotonin reuptake inhibitors, or SSRIs, which are among the most widely used medicines given for autism.

The new research could "change this practice," said prominent Yale University autism researcher Dr. Fred Volkmar. He commented in an editorial released with the study Monday in the June issue of Archives of General Psychiatry.

The results echo a separate study reported in February that showed a low-dose form of Prozac, another SSRI, also did not reduce repetitive behaviors in autism.

The overall global market for drug treatment in autism is at least $2 billion US and SSRI antidepressants account for nearly 60 per cent of that, the study authors said.

Celexa's maker, Forest Laboratories Inc., issued a statement saying the company "was not involved in this study and therefore cannot provide comment."

Little improvement found in most children

Geraldine Dawson, chief science officer of the advocacy group Autism Speaks, said the new results underscore the difficulty in treating a condition with an uncertain cause and symptoms that range from mild to severe.

"We are still so challenged to come up with medications that can address core symptoms," she said, "largely because we still don't understand the biology of autism."

The study involved 149 autistic children aged five to 17 who were randomly given either up to 20 milligrams daily of for 12 weeks or dummy pills.

Doctors rated children's symptoms during treatment on a scale of 1 to 7, with high scores reflecting worsening symptoms. The rating method allowed doctors to evaluate improvements in repetitive actions and also other behaviours.

Only about one-third of children on citalopram showed substantial improvement; most showed little or no improvement or got worse.

Celexa is among antidepressants labelled with a warning about the potential for increasing risks for suicidal thoughts and behaviour in children, but these symptoms didn't occur in the study.

About one-third of children on dummy pills also improved. King said reasons for that are uncertain. It could be they expected to get better with any kind of pill — the well-known "placebo effect." Or it could just have been a coincidence since autism symptoms tend to fluctuate over time.

That tendency might also explain why many kids on placebo also developed new or worse symptoms, he said.

The U.S. National Institutes of Health paid for the research.