Developmental Surveillance Versus Screening for the Identification of Autism Spectrum Disorders in Infants and Toddlers

Background: Despite the importance of early detection and intervention for children with Autism Spectrum Disorders (ASDs), there are currently no screening tools with sufficient psychometric properties that can be recommended for universal use. Barbaro and Dissanayake (2010) utilised a developmental surveillance framework in the Social Attention and Communication Study (SACS) to prospectively identify infants and toddlers with ASDs. The SACS was accurate and sensitive in the identification of ASDs from 12- to 24-months of age. They argued that this was a result of: 1) utilising repeated monitoring within a developmental surveillance framework, rather than ‘once-off’ screening at a particular age; and 2) the use of skilled observations, rather than sole reliance on parental report. The current study aimed to test these arguments.

Objectives: The first aim in the current study was to compare the psychometric properties of the SACS to several screening tools in identifying infants and toddlers ‘at risk’ for ASDs, in the same sample of children. The second aim was to examine the consistency of parental reporting of children’s behaviours across different ASD screening questionnaires. The third aim was to examine agreement between parental reporting on these questionnaires and skilled professional observations of the same behaviours.

Methods: Participants were drawn from the SACS sample (n = 110), and comprised children who were assessed at least once at 12-months (n = 10), or 18-months (n = 46). The screening tools utilised at these assessments included the First Year Inventory (FYI), the Communication and Symbolic Behavior Scales-Developmental Profile (CSBS-DP), the Checklist for Autism in Toddlers (CHAT), the Modified- CHAT (M-CHAT), and the Early Development Interview (EDI). The psychometric properties of these tools were calculated and compared to those of the SACS, reported in Barbaro and Dissanayake (2010). Furthermore, responses on items measuring the same behaviours across the different parental report tools were compared. Additionally, these same responses were also compared to skilled observations of the same behaviours in the SACS.

Results: Results indicated that the SACS demonstrated a better balance between good to excellent specificity, sensitivity, and positive predictive value compared to each of the screening tools, which consistently traded specificity for sensitivity, or vice versa. Furthermore, consistency of parental report on items measuring the same behavioural construct, across different screening tools, was poor, with only 9 out of 26 associations (35%) at 12- months, and 4 out of 12 associations at 18- months (33%) having acceptable Spearman rho values of .70 or more. Similarly, there were no acceptable correlations of .70 or more between parental report and skilled observations of children’s behaviours on the SACS. 

Conclusions: The current results confirmed that a development surveillance framework, utilising repeated monitoring of children’s behaviours via skilled observations, is more robust than the use of screening tools. This study revealed the variability in parental report of children’s behaviours both across different questionnaires and in comparison to skilled observation of the same behaviours. These results highlight possible reasons for the lack of screening tools for ASDs with sufficient psychometric properties to be recommended universally.