Early Identification in Autism: Subtypes Based on Child, Family, and Community Characteristics

Background: The American Academy of Pediatrics (AAP) screening guidelines highlight the importance of listening closely to parent concerns as an effective strategy to identify children earlier (Johnson & Meyers, 2007). While the AAP guidelines have led to increased rates of developmental screenings and earlier identification of ASD (e.g., Radecki, Sand-Loud, O’Connor, Sharp, & Olson, 2011), many children, especially those considered underserved (i.e., low-income, rural, or racial/ethnic minorities), still receive later diagnoses. Given the immense heterogeneity across autism spectrum disorders (ASD), we investigated subtypes of children with ASD based on parent concerns and socio-demographics.

Objectives: Research questions included: 1) among children with ASD, how do early parent concerns, child (i.e., age and gender), family (i.e., race/ethnicity, socioeconomic status [SES]), and community (i.e., provider access) characteristics group by subtypes, and 2) to what extent do subtypes of children with ASD differ by chronological age at the diagnostic evaluation?

Methods: We performed a secondary analysis with data drawn from a medical university child diagnostic center. The sample included children (n=712) 12 months- 12 years (M=66.68 months; SD= 34.28) who eventually received a diagnosis of ASD. We examined intake information completed by parents prior to the diagnostic evaluation including: 1) parent concerns, 2) child’s age, 3) child gender, 4) family SES, 5) race/ethnicity, and 6) access to service providers. For parent concerns, we coded these statements into six categories (adapted from Ozonoff et al., 2009), including: 1) behavior/temperament, 2) cognitive development, 3) speech/communication, 4) social interactions, 5) stereotyped behaviors, and 6) medical. Coders examined percent agreement with 20% of the dataset and achieved 89%. For the first research question, we used latent class analysis (LCA). For research question two, we performed a Kruskal-Wallis H-Test.

Results: After comparing the LCA fit statistics (i.e., Akaike Information Criterion, Bayesian Information Criterion, Adjusted Bayesian Information Criterion, Vuong-Lo-Mendell Rubin, Lo-Mendell-Rubin, entropy) of a two to seven class solution, results revealed a five-class solution fit best. Two subtypes were identified younger (i.e., approximately 3.5 years of age) and were differentiated by communication and medical concerns. One of these younger subtypes included non-white, Hispanic children utilizing Medicaid. Another subtype was identified around 5.5 years and was differentiated by stereotyped and by developmental parent concerns. Lastly, two subtypes were identified at an older age (i.e., approximately 9 years of age) with either cognitive concerns, or social and behavior concerns. One of the oldest subtypes was characterized by females with ASD.

Conclusions: By understanding ASD subtypes based on parent concerns from diverse socio-demographics, we may inform universal screening procedures. Our study suggests that children with speech/communication parent concerns are most likely identified earlier regardless of race, ethnicity, or SES. Whereas, children with social and behavior parent concerns, as well as females were diagnosed later. Future research should examine the distinct subtypes of females with ASD. While certain social and behavior concerns are acceptable during early years of childhood, more research is needed to determine methods to distinguish these types of parent concerns earlier.