Mixed Verdict Delivered on Autism Treatments

There were problems with studies of other medical treatments, too. In fact, insufficient evidence existed to even evaluate the use in autism of selective serotonin-reuptake inhibitors or stimulants, said the researchers.

They did find sufficient evidence to support the use of the antipsychotic drugs risperidone and aripiprazole for "challenging and repetitive behaviors," such as irritability, aggression, and self-injurious behavior.

But the benefits of those drugs came with a high price—increased risk for weight gain, sedation, and extrapyramidal symptoms—that may confine the medications' use to patients whose severe impairment or injury risk outweighs the potential adverse effects.

Poorly designed and inadequately powered trials generally limited the value of studies on other medical interventions for ASDs, and the two strongly positive trials the authors identified were funded by drug companies. "[M]ore publicly funded studies of medications for ASDs are warranted," they said.

Not all the news was gloomy, Veenstra-Vander Weele told Psychiatric News.

"The early intensive behavioral intervention data are actually somewhat encouraging, just not as clear or as specific as we might hope," he said. "Despite the absence of certainty, this is the area where interventions seem most likely to have a large impact on educational outcomes."

The researchers looked at studies of three categories of behavioral therapy: the early intensive behavioral intervention (EIBI) pioneered by the late University of California at Los Angeles clinical psychologist Ole Lovaas, Ph.D.; comprehensive approaches for children under age 2, including the Early Start Denver Model (ESDM); and parent training.

Both the Lovaas model and ESDM use teaching strategies that are based on learning strategies called applied behavior analysis. Of the 23 EIBI studies assessed, 15 were rated as having poor quality and eight as fair, including one randomized controlled trial. Children in the treatment group in that trial improved their IQ test scores, but selection of the ASD patients was not uniform. Most study participants who showed considerable improvement were diagnosed with pervasive developmental disorder not otherwise specified, but most of those who did poorly in the trial were diagnosed with "classically defined autism disorder," said the authors.

In the second category, involving comprehensive approaches, only four studies of treatments for very young patients (younger than age 2) met the reviewers' standards for inclusion, and just one of those, the ESDM, was a randomized controlled trial.

"In the ESDM model, the teaching strategies are delivered as part of interactive play activity that is child-directed and provides an opportunity for the child's choice of materials and activities," said Geraldine Dawson, Ph.D., chief science officer of the advocacy group Autism Speaks and principal investigator of the ESDM randomized controlled trial, in an interview. "Skills such as making eye contact, gesturing, and responding to a request, are not taught in isolation … [but] within shared playful activity."

The trial indicated that children in the intervention arm showed greater gains over a two-year period in IQ (17.6 points) than did a comparison group (7.0 points) that received a less-intensive intervention.

Other studies of potentially promising comprehensive interventions had methodological flaws such as a lack of baseline comparisons, poor case definitions of included patients, or inadequate documentation, the authors pointed out.

The third category of programs reviewed covered seven home-based interventions delivered by parents. Results were mixed, the authors determined. Criteria assessed in some studies seemed to improve, but others did not. Again, methodological problems (sample size, lack of randomization, and lack of standardized measures of performance) limited the studies' utility.

Other issues detract from the utility of many of these clinical trials as well.

For one thing, definitions of autistic disorder and autism spectrum disorders have changed significantly over the past 30 years and may continue to change with the development of DSM-5, making it hard to compare studies over time, he said. In addition, individual children with ASDs may differ substantially from one another in their symptoms, diagnoses, and response to treatment.

"It is therefore extremely important that studies fully characterize the participant population so that clinicians and families can evaluate whether a particular child is likely to benefit," he said.

Outcomes must be standardized as well. Autism Speaks has already adopted that approach, said Dawson.

"We recently funded seven clinical trials of early interventions for infants and toddlers with autism, and required that all investigators use the same outcome measures so that the data could be compared and combined across trials," she said. "Similarly, NIH now requires that investigators use common diagnostic, IQ, adaptive behavior, and language measures so that outcomes can be compared across different studies."

Another problem with autism studies is that the research sometimes stops with an initial clinical trial of a treatment performed by its developers.

"That is perfectly appropriate, so long as the study is appropriately controlled, but replication of research findings is critical," said Veenstra-Vander Weele. "It is also difficult to know how a treatment will translate into the real world without knowing that it has succeeded in multiple places."

Finally, how well improvements are sustained over time is poorly understood, he said.

"As far as I know, there are no long-term follow-up studies of intervention effects, but clinical experience suggests that continuing (but possibly occasional) treatment activity is needed," Rutter agreed.

Participants in the recently published trial of the ESDM model are being followed with the help of NIH funding, said Dawson. She expects publication within the next few years.