Effects of Race, Ethnicity, and Maternal Education on Reported Regression in Children with Autism Spectrum Disorder (ASD)

Background:  According to parent report, about one-quarter to one-third of children with ASD experience developmental regression. Previous studies have noted that gender and socioeconomic status have no associations with reported regression among children with ASD. However, information about the effects of race, ethnicity, and maternal education on the prevalence of regression in children with ASD is limited.

Objectives:  To determine the effects of race, ethnicity, and maternal education on reported regression in children with ASD in the Simons Simplex Collection (SSC) and Autism Genetic Resource Exchange (AGRE) databases. 

Methods:  Data were analyzed for children with ASD in the SSC (N=2695; mean age at study participation=9.0 years (SD = 3.6); 87% male). Regression status was ascertained via the Autism Diagnostic Interview-Revised (ADI-R) and operationalized for this study as any loss of language and/or social engagement at or before 36 months of age. Logistic regression was used to explore possible associations with regression by race/ethnicity, maternal education, annual household income, and age at first concern.   

Results:  The overall prevalence of language and/or social regression in the SSC sample was 28%. Prevalence of regression by race/ethnicity was: non-Hispanic white 26%, non-Hispanic black 42%, Hispanic 37%, and non-Hispanic other 30% (p<0.001). In unadjusted analyses, regression was associated with race/ethnicity (p<0.001), age at first concern (p<0.001), and maternal education (p=0.013), but not annual household income (p=0.818). When controlling for age at first concern and maternal education, race/ethnicity was significantly associated with language and/or social regression (p=0.0002). Non-Hispanic black children were twice as likely to experience regression than non-Hispanic white children (95% CI 1.1, 3.5; p=0.011). Hispanic children were 1.6 times more likely to experience regression than non-Hispanic white children (95% CI 1.1, 2.2; p=0.004). When controlling for age at first concern and race/ethnicity, children with mothers with an associate’s degree or higher were 20% less likely to reportedly experience regression (p=0.019), although the difference in reported regression between maternal education groups was small (32% for mothers with no degree and 27% for mothers with a degree). Comparable analyses in the AGRE dataset are currently underway.

Conclusions:  In the SSC sample, non-Hispanic black and Hispanic children with ASD were at increased risk of parent-reported regression compared to non-Hispanic white children, when controlling for maternal education and age at first concern. Further studies are needed to evaluate whether these differences in race/ethnicity among children with ASD and regression are related to cultural and/or genetic differences possibly resulting in different phenotypes of ASD.