Interventions for children with autism: identifying what works for whom

Abstract

Background: Despite the heterogeneity of autism, most intervention research focuses on group outcomes; little is known about the individual characteristics of children who do, or do not, respond to treatment. Methods: We explored factors predictive of treatment response, as opposed to variables related to prognosis in autism more generally, in two large‐scale data sets: (1) 152 pre‐school children involved in a parent‐mediated, social‐communication randomised control trial and (2) 240 school‐aged children receiving social‐cognition interventions. The Reliable Change Index was used to identify children as responders or non‐responders to intervention. Logistic regression was used to identify baseline variables predictive of treatment response. Results: Group analyses indicated moderate‐large improvements in each intervention group compared with treatment as usual (TAU). However, in both treatment and TAU conditions, some children failed to improve while others made considerable gains. None of the variables typically associated with a good prognosis in autism, such as cognitive, linguistic and social skills were predictive of treatment response. However, in the parent‐mediated trial, TAU children in the site with greater social deprivation had a poorer outcome (P < .01) than TAU children in other sites. In the social‐cognition cohort, children with poorer baseline social skills were more likely to respond to intervention (P = .01) than those with better‐developed social skills. Conclusion: The findings illustrate the complexity and challenges of identifying factors related to individual responses to intervention; they also highlight the importance of distinguishing between prognostic indicators of natural improvement over time and variables that predict treatment outcome. To date, little is known about which children may benefit from a particular intervention or for whom it is contraindicated. Larger, multicentre trials are essential to understand the variables related to treatment response and to enable educators and clinicians to choose appropriately individualised interventions for children with autism.