Validating a New Visual Screener for Autism in an Unselected Sample of Low Income and Ethnic Minority Children

Background:

Significant disparities exist in early access to care among low income and minority children at risk for ASD. These disparities may be due in part to the lack of ASD screening tools designed for this population. Existing tools typically require parents to have sophisticated reading comprehension and understanding of developmental milestones. The Developmental Check-In (DCI), a primarily visual tool with minimal literacy demands, was developed specifically for this population. Previous research with the DCI found that it outperformed the Modified Checklist for Autism in Toddlers-Revised (MCHAT-R) and Social Communication Questionnaire (SCQ) at identifying ASD among young underserved children at risk for ASD.

As a next step in validating the DCI, we examined its psychometric properties in an unselected sample of young underserved children.

Objectives:

To compare the performance of the DCI with that of other, validated ASD screening tools (M-CHAT-R & SCQ) in detecting ASD among an unselected sample of young underserved children.

Methods:

Subjects were 24 to 60 month old children enrolled at Head Start programs, which are located in underserved communities and serve low income families. Parents completed the DCI, MCHAT-R, SCQ, medical history form, adaptive behavior measure (ABAS) and demographic information. Parents provided consent for their child’s Head Start teacher to complete the same 3 screening tools. Evaluations were offered to children at risk on any parent or teacher-completed screening tool. The children were evaluated using the Mullen Scales of Early Learning and the ADOS-2. Clinical diagnosis was determined by a licensed, experienced clinician (developmental pediatrician, psychologist or advanced practice nurse) based on patient history, clinical observation and ADOS-2 classification. Differences in performance were estimated using receiver operating characteristic curves.

Results:

Completed parent-report measures were collected for 439 children. Child care providers also completed screening for 68% of the sample. Forty-one children (9%) were subsequently diagnosed with ASD. The parent-completed DCI better identified children at risk for ASD than the MCHAT-R (p<0.0001) or the SCQ (p= 0.0139). Area under the curve in the ROC analysis was 0.84, 0.67, and 0.76, respectively. For teacher-completed tools, there were no significant differences between the DCI and the MCHAT-R (p = 0.06) or SCQ (p = 0.48).

Conclusions:

The parent-report version of the DCI performed better than two validated screening tools in identifying young children with ASD who were from families with low income, and/or who were non-white, Hispanic, and/or for whom English was not the primary language spoken in the home. When considered with previous findings that the DCI also performed better among young, underserved children at-risk for ASD, the DCI is a promising tool for reducing disparities in diagnosing ASD. Based on the rate of ASD (9%) among children with completed parent screens, this study also suggests that ASD screening in underserved childcare settings is a useful method for identifying children at risk.