Timing of Autism Spectrum Disorder Identification in the U.S. from the Autism and Developmental Disabilities Monitoring Network 2006-2012

Poster Presentation
Saturday, May 4, 2019: 11:30 AM-1:30 PM
Room: 710 (Palais des congres de Montreal)
L. Stewart1, J. Baio2, S. Rosenberg3, C. Robinson Rosenberg4, M. S. Durkin5, R. S. S. Kirby6, J. A. Hall-Lande7, B. Harris8, L. Nikolaou9, D. Christensen2 and L. C. Lee10, (1)Mental Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, (2)Centers for Disease Control and Prevention, Atlanta, GA, (3)University of Colorado Anschutz Medical Campus, Aurora, CO, (4)University of Colorado / JFK Partners, Aurora, CO, (5)Population Health Sciences, University of Wisconsin School of Medicine and Public Health, Madison, WI, (6)Community and Family Health, University of South Florida, Tampa, FL, (7)UCEDD, University of MN, Minneapolis, MN, (8)Department of Pediatrics, University of Colorado, Denver, CO, (9)Oak Ridge Institute for Science and Education, Oak Ridge, TN, (10)Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD
Background: Timely identification of autism spectrum disorder (ASD) is one of the critical factors determining whether children have access to early intervention services. The most recent estimate of median age at ASD diagnosis among those with a diagnosis by the age of 8 years from the 2016 Autism and Developmental Disabilities Monitoring Network (ADDM) cohort in the U.S is 4.3 years. Prior work in the ADDM network, as well as in smaller community-based studies suggests that age at ASD identification varies across subgroups by factors such as sex, race, ethnicity, geographic location, and socioeconomic status.

Objectives: Examine influence of socio-demographic characteristics on timing of ASD identification during 2006-2012 within the ADDM network.

Methods: The ADDM Network is an active surveillance system that provides ASD prevalence estimates among children age 8 years in parts of the U.S. Data for this analysis come from ADDM surveillance years (SY) 2006, 2008, 2010, and 2012 from communities in Alabama, Arkansas, Arizona, Colorado, Georgia, Maryland, Missouri, North Carolina, New Jersey, and Utah. Children were included in this analysis if they had an ADDM determined ASD case status, information from birth certificates and neighborhood level census data. Median age at ASD identification (AAI) was assessed for the full sample with ASD status as well as across socio-demographic, developmental, site, and surveillance subgroups. Nested, multivariate survival analyses examined how AAI associated with covariates of interest. A series of increasingly complex mixed-effects models were fit nesting child, family, surveillance characteristics clustered on census-tract level poverty and study site.

Results: 13,731 children with ASD case status were analyzed, 30% of whom were not identified as having ASD prior to ADDM abstraction and review at age 8. Compared to those with a documented AAI, children without a documented AAI were more likely to belong to a non-white racial/ethnic group and live in areas with >= 20% of the households below the federal poverty line, or “poverty areas”. Incorporating the traditionally censored not-yet-identified children back into the sample with an ADDM-identification age of 8 years yielded a median AAI of 5.8. From SYs 2006 to 2012, median AAI decreased from 6.3 to 5.3 and there was a statistically significant trend towards earlier AAI across SYs. This trend towards earlier AAI was more pronounced in non-white racial/ethnic groups, as the gap in AAI was reduced from 2006-2012. The percentage of children not-yet-identified decreased across SYs from 35% to 25% between 2006 and 2012. The nested multivariate survival models showed that a later AAI was significantly associated with African-American or Hispanic race/ethnicity, lower maternal education, earlier cohort year, and single-surveillance record source. Earlier AAI was significantly associated with having below average or missing data on IQ, higher maternal education, or being a member of a later birth cohort. Non-substantial variation was found across poverty level and study site variables.

Conclusions: Further examination of the role socio-demographic factors play in the ASD identification timing will help to inform and strengthen current identification infrastructure.