Utilising Technology for the Early Detection of Autism: Introducing Asdetect, an Early Detection Mobile Application for Caregivers

Saturday, May 13, 2017: 12:00 PM-1:40 PM
Golden Gate Ballroom (Marriott Marquis Hotel)
J. Barbaro and N. Kolivas, Olga Tennison Autism Research Centre, La Trobe University, Melbourne, Australia
Background: Early detection of autism is vital as it provides access to early intervention, facilitating children’s developmental outcomes and reducing family stress (Crane et al. 2016; Dawson et al., 2008; 2010; Howlin, 1997). Over the past 11 years, two community-based studies on the early detection of autism (Social Attention and Communication Study; SACS & SACS-Revised; Barbaro & Dissanayake, 2010; 2013) have been conducted within the Victorian Maternal and Child Health system, Australia. The SACS program is currently the most accurate and sensitive method for identifying autism in children under 24-months, but its use has been limited to universal services.

ASDetect (released February 2016) is a free mobile application that incorporates a modified version of the SACS training, and allows caregivers to monitor the signs of autism in children aged 11-30 months. Assessments at 12-, 18-, and 24-months contain 12-15 short videos demonstrating key social-communication behaviours, followed by ‘mostly’/‘rarely’ questions. Behaviours found to be most predictive of autism by 24-months (Barbaro & Dissanayake, 2013) are used to determine a child’s ‘likelihood’ for autism (‘high vs ‘low’), and caregivers are encouraged to share their child’s results with their doctor.

Objectives: This study’s objectives were to compare the responses for children at ‘high’ and ‘low’ likelihood for autism on each of the behavioural items, and qualitatively explore the experiences of, and actions taken by, caregivers following use of ASDetect.

Methods: 3452 assessments (with caregivers “opting in” for research) were undertaken at 12 (n=848), 18 (n=966), and 24-months (n=1638). Percentage of ‘mostly/rarely’ responses for each item was compared between children with a ‘high’ and ‘low’ likelihood of autism at each assessment using Fisher’s exact probability test. A brief email survey was also sent out to all caregivers who opted into research.

Results: 932 children retuned a ‘high-likelihood’ result (73% male, 27% female), with 71% of caregivers reporting they already had prior concerns. All items at each age significantly differed between ‘high’ and ‘low’ likelihood groups; the strongest associations across each assessment involved use of gestures, eye contact, pointing, and showing (phi coefficient range = .63-.75), with 81-94% of ‘high-likelihood’ children rarely engaging in these behaviours, compared to 7-11% of ‘low-likelihood’ children.

The survey (n= 122) indicated that 60% of parents whose children returned a ‘high-likelihood’ result arranged a follow-up appointment with their doctor, with 24% subsequently receiving a diagnosis (43% autism, 53% developmental/language delay, 14% “other”). Caregivers agreed/strongly agreed that ASDetect was “easy to use” (98%), the videos were helpful in illustrating the questions (97%), they knew more about social-communication milestones following its use (90%), and that they would recommend it to other parents (96%).

Conclusions: ASDetect has facilitated hundreds of families in seeking professional support following a ‘high-likelihood’ result for their child. Items that differentiated children at ‘high’ and ‘low’ likelihood for autism are strongly consistent with previous work identifying the predictors of autism in 12-24-month-old children. Projects are now underway to determine the psychometric properties of ASDetect in identifying children with autism, and translating the content into other major languages.