Prevalence, Correlates and Predictors of Good Outcomes in Autism Spectrum Disorder in Late Childhood

Background: The outcome literature in ASD has focussed largely on changes in symptoms and cognitive skills over relatively short time frames with few data points. There are very few studies that investigate what constitutes a ‘good’ outcome in late childhood in ASD. This is an important developmental stage as health status at this stage is often more stable compared to earlier periods and predicts health status at later ages.

Objectives: To describe the process whereby parents defined domains to be assessed in measuring good outcomes and to describe the prevalence, agreement, correlates and predictors of good outcomes in children with ASD at 8-11 years of age followed longitudinally from 2-4 years of age.

Methods: Data are drawn from the Pathways in ASD study which assembled an inception cohort of children with ASD (N=421) soon after the diagnosis was given at between 2- 4 years of age. The children were seen at several time points and this presentation will focus on outcomes investigated at 8-11 years of age. Multiple domains of a good outcome were defined by parents who identified socialization, communication, independence, and emotional challenges as key domains. These domains were measured using the Vineland Adaptive Behaviour Scales and the CBCL. A ‘good outcome’ in these domains was defined by scoring in the ‘typical’ range on any instrument or by an improvement of one standard deviation in standardized scores between the two time points. The prevalence of a good outcome in each domain by either metric was calculated, the agreement between good outcomes definitions was calculated, and the predictors of these outcomes were assessed

Results: The prevalence of a good outcome on any of the 5 domains ranged between 4 and 26% depending on the domain and the metric. There was little agreement between the different ways of defining a good outcome or between different outcome domains, even within a measurement tool. In particular there was no agreement (beyond chance) between the good outcomes defined by scores in the typical range and good outcomes defined by improvement in key domains. A good outcome defined by either metric was associated with lower ASD symptom scores, higher IQ scores and more friendships in late childhood. Important contextual factors such as family functioning were important predictors of a good outcome, controlling for baseline scores.

Conclusions:

The results suggest that there are multiple ways of defining a good outcome that individualy are not uncommon in this cohort during late childhood. Given the heterogeneity seen in ASD, it is possible to arrive at a ‘good’ outcome by multiple pathways, either by achieveing a typical score or by showing clinically significant improvement. Although these ratings of outcome were based on parental input, we do not have the perspective of children with ASD which will be crucial in future studies. In addition, the stability of a good outcome in unknown. The different domains of outcome appear to be relatively independent suggesting that interventions must be broadly based to ensure improvement across multiple domains.