Parent-defined target symptoms respond to risperidone in RUPP autism study: Customer approach to clinical trials.

L. Eugene Arnold et al. on behalf the Research Units on Pediatric Psychopharmacology Autism/PDD Network

Journal of the American Academy of Child and Adolescent Psychiatry 2003; 42(12): 1443-1450.7

Bottom Line

In real life situations does risperidone improve problem behaviour?

  • The behaviour of most of the children who remained on the drug was found to be “much improved” or “very much improved,” but the effect declined as time passed.
  • The types of behaviours most likely to be positively affected were hyperactivity, anger, and rapid changes in mood.
  • Children with PDD-NOS were more likely to respond to treatment than were children with a diagnosis of autistic disorder.
  • Risperidone provides relief of symptoms of aggression in children with PDD-NOS and autistic disorder, but likely should be used as a treatment of last resort when behavioural interventions have failed because of its side effects in some children.
  • The need for medication should be re-evaluated over time.

Question: Can clinical trials assess changes in symptoms that are of concern to parents/caregivers of children with autism?

Background: In most clinical trials it is the researchers who decide what symptoms they think will be helped by the therapy under study and it is these symptoms they look at. It is becoming increasingly clear that the things that researchers think need correcting and the things that family members and patients think need to be helped may not be the same. This study, while pursuing clinician/researcher-determined goals for treatment, also looked at the symptoms the parents or caregivers of children with autism. This is not a new research study, but a different analysis of the data from the same children taking part in the RUPP Risperidone study.

Participants: The participants were 101 children and adolescents with autistic disorder who took part in a study of risperidone as a treatment for aggressive behaviour, self-injurious behaviour, or irritability. They ranged in age from 5 to 17 years.

Design: This was a randomized, controlled trial.

Methods: 49 of the participants got active treatment, while 52 got an inactive treatment (placebo). A standardized test related to the behaviour problems, the clinical global impression of improvement test (CGI-I), was administered to the participants. The same clinician who carried out the CGI-I and who did not know if the participants were receiving active treatment or were on placebo interviewed the parents. They were asked about the one or two problems with their child that concerned them the most. These symptoms were looked for and described during the study, first at the beginning, again at 4 weeks, and then again at 8 weeks. The clinician recorded the number of times the symptoms were reported, how long they lasted, how severe they were, how much they interfered with family life or the individuals ability to function. After the study, five researchers then went over the data independently and rated whether the symptoms changed over the duration of the study. Thus, there were three raters who determined if improvement had or had not taken place.

Main Results: The symptoms reported by the parents were classified as aggression to others; self-injurious behavior; property destruction; tantrums; yelling/screaming; stereotypy; and hyperactivity/impulsivity/agitation (p.1445). The scores of the two groups were compared to each other. As time passed and exposure to the drug increased, the symptoms reported by the parents improved so that the particpants showed either definite improvement or marked improvement. The improvement seen in the symptoms reported by the parents and those detected by the CGI-I were virtually the same.

Conclusions: This study demonstrated that parent identified behavioural problems can be addressed within a clinical trial when other standardized tests of symptoms are used. The assessment tool used by the parents was easy to use, could be counted on to detect what it set out to detect (reliability), and was able to detect change over time.