On average, diagnoses of autism spectrum disorder (ASD) occur after the fourth birthday, missing critical early intervention opportunities. Many children with ASD are referred to state early intervention agencies for possible developmental delay, but unless their ASD is identified, intervention services are generally low intensity and not specific to their core ASD symptoms. However, implementation of screening to detect ASD poses considerable logistical challenges. This can result in inaccurate or altered administration, which may hinder successful detection. While the importance of accurate and reliable early detection procedures in the ASD field is widely acknowledged, no study thus far has examined the issue of procedural fidelity (i.e., adherence to the prescribed protocol) in ASD screening. Another critical research gap concerns the proximal outcomes of ASD early detection procedures, such as how screening results relate to autism referrals and specific intervention services. As the ultimate goal of early screening is to facilitate timely access to appropriate intervention programs, it is critically important to examine the rates of referral and uptake of services for children who screen positive. The proposed project aims to address these critical gaps and increase scientific knowledge, focusing on the implementation of the Modified Checklist for Autism in Toddlers Revised with Follow-up (M-CHAT-R/F) in the context of a large urban high-risk preschool sample. Most research using the M-CHAT and the M-CHAT-R/F screened children during pediatric well-child check-ups (e.g., Chlebowski et al., 2013), although several studies have examined the performance of the tool with high-risk toddlers and preschoolers (e.g., Charman et al., 2015; Snow & Lecavalier, 2008). Overall, sensitivity is adequate when the M-CHAT is administered to samples of preschoolers at risk for developmental delay, although specificity ranges from poor to adequate. In order to test the study research questions we will extract intake, multidisciplinary evaluation (MDE), and service/placement data from Elwyn Special Education for Early Developmental Success (SEEDS), the agency providing intervention services to children age 3-5 in Philadelphia County, PA. SEEDS estimates that 5,000 children were screened January-December 2015. The first aim will examine the procedural fidelity of the M-CHAT-R/F implementation, by examining accuracy and completion data over time. Both inaccurate and missing data will be examined for barriers and biases. The second aim will examine the proximal outcomes of screening, namely the relationship between M-CHAT-R/F scores and MDE results, autism evaluation results, and services and placements offered to children in the SEEDS program. This study will be the first to examine fidelity of screening procedures in a high-risk setting. Identification of barriers or systematic bias may inform alternate approaches to more successful screening, improving care of children with ASD. This study also uses a novel approach to relate screening results to proximal outcomes, which will increase understanding of the role of screening in early intervention agencies.

Early detection of autism spectrum disorder (ASD) is essential to facilitate ASD-specific early intervention, which improves outcomes. Screening protocols can be embedded into community agencies that provide early intervention services, but it is important to study the accuracy with which these protocols are deployed, and the utility in identifying children in need of ASD-specific evaluation and treatment. The current study assesses these questions in a large, urban, community-based sample of preschool children referred for possible early intervention.