The Intersection of Race and Place in the Time-Lag Between Diagnosis of Autism Spectrum Disorder and Onset of Early Intensive Behavioral Intervention

Background and Purpose:

Public funding of early intensive behavioral intervention (EIBI) for the 1 in 68 children who meet criteria for autism spectrum disorder (ASD) is rapidly expanding. Evidence indicates that children with ASD experience racial and geographic disparities in access to health care services, and the Interagency Autism Coordinating Committee cites disparities in access to early intervention among the most pressing yet understudied areas of research. Currently, ASD service research is dominated by inquiries into the age of diagnosis. We know little about disparities in an important indicator of service access, or the time-lag between ASD diagnosis and treatment onset.

To examine disparities in the time-lag between diagnosis and treatment onset, we partnered with the South Carolina Department of Disabilities and Special Needs (DDSN) to build a unique, comprehensive dataset of children who received EIBI through DDSN's Medicaid waiver. We examined: 1) the relationship between child race and time-lag; 2) the relationship between children's neighborhood racial composition, poverty, affluence, and urbanicity and time-lag; and 3) whether the relationship between race and time-lag varies by neighborhood characteristics.

Methods:

Data and samples: We integrated data from DDSN paper case records, spreadsheets, and electronic records and state Medicaid and Census data. The dataset includes children who enrolled in the waiver between its inception (January 2007) and March 31, 2015 (N=2,338). The current sample (N=473) includes children who were: 1) diagnosed after DDSN established a waitlist; 2) placed on a wait list; and 3) not missing dates of diagnosis, enrollment, assessment, and initial therapy session.

Measures: We measured child race as black, Hispanic, other non-Hispanic, and unknown (a DDSN and Medicaid category). We used the census tract ID of children's residential addresses to determine racial composition (percent of white residents), poverty (composite), and affluence (composite). We used USAA Rural-Urban Commuting Areas to determine urban, suburban, and rural tracts. Outcomes included the number of days between four time points: Date of Diagnosis to Date on Waitlist (Time 1), Date of Enrollment to Date of Assessment (Time 2), Date of Assessment to Initial Therapy Session (Time 3), and Date of Diagnosis to Date of Initial Therapy Session (Time 4). We estimated 20 contextual OLS regression models using PROC REG in SAS v9.4, or one main effects model and four interaction models for a total of five models per outcome. We applied a Bonferroni correction (α = .025) and compared changes in R².

Results:

Contrary to prior research, the number of days between diagnosis and treatment onset was largely unaffected by race and neighborhood characteristics. Average overall time-lag (Time 4) was 1040 days, and significant independent variables and covariates differed by outcome. Significant models explained 83% (Time 1), 6.8% (Time 2), and 88% (Time 4) of variability.

Conclusions and Implications:

Findings provide insight into the high number of days between diagnosis and treatment onset, underscore the need for future research that includes provider and organizational characteristics, and offer lessons on the collection and use of administrative data in research on EIBI.