What Are the Differences between Parent-Reported Autism Concerns on the M-CHAT-R/F and Clinical ADOS-2 Results?: Findings from a Sample of Ethnically Diverse, Low Income Families

Background: To detect autism spectrum disorder (ASD) in as many cases and as early as possible, the American Academy of Pediatrics recommends universal ASD-specific screening at 18- and 24-months. Pediatricians widely screen for ASD with the Modified Checklist for Autism in Toddlers Revised with Follow-Up (M-CHAT-R/F), a parent questionnaire with 20 binary items. Some research indicates sensitivity of the M-CHAT-R/F at rates up to 91%; however, in other studies the validity has been questioned for children who have: (a) families with lower socio-economic status, (b) varying symptom expression, such as females, and (c) ethnic minority group status. Understanding the areas of discrepancy and agreement on parent-report and gold-standard assessment (i.e., ADOS-2) is critical for informed recommendations on the M-CHAT-R/F's practical utility.

Objectives: To explore the relationship between parent-indicated concerns for ASD on the M-CHAT-R/F and corresponding gold standard assessment within a sample of ethnically diverse, low-income families.

Methods: Data were collected about 148 toddlers who had positive MCHAT-R/F screens in primary care and diagnostic assessment including ADOS-2 administration in specialty care. Participating parents, toddlers, and clinical sites were from a larger randomized controlled trial of Family Navigation across three U.S. cities. Two trained clinicians independently mapped M-CHAT-R/F items onto the ADOS-2 scoring system, based on measured ASD symptoms that corresponded across both assessments. Inter-rater agreement was 85%. The M-CHAT-R/F positive predictive value (PPV) in relation to ASD diagnostic resolution was calculated. Across toddlers with and without ASD, parent-indicated M-CHAT-R/F and clinician indicated ADOS-2 results were tabulated. Proportions of clinicians and parents with concordant present or absent concerns (i.e., positive predictive and negative predictive values) and with disagreement across corresponding items (i.e., false positive and false negative rates) were calculated.

Results: The participating toddlers were 73% male, 59.4% born in the U.S., and primarily from ethnic minority groups. The majority of parents had graduated high school, enrolled in social service programs, and received public insurance. Eighty-two of 148 toddlers met ASD diagnostic criteria, yielding an M-CHAT-R/F PPV of 0.55, which is within the range of its published performance characteristics. When a skill was indicated as a concern on the ADOS-2, the highest rate of M-CHAT-R/F-ADOS-2 agreement was on pretend play, pointing to ask for help, and imitating. In contrast, low agreement was identified on ratings of children's shared enjoyment and intonation. High false negative rates (i.e., absent M-CHAT-R/F item concern and present corresponding ADOS-2 item concern) occurred on pointing, imitation, response to name, and unusual finger movements. High false positive rates (i.e., present M-CHAT-R/F item concern and absent corresponding ADOS-2 item concern) occurred on responding to pointing and not requesting parents' attention.

Conclusions: In the context of research indicating poor M-CHAT-R validity for samples of ethnic minority and low SES children, this study identified specific areas that may contribute to a high level of inaccurate screens. Further exploration of parent-related factors (e.g., expectations of children, interpretation of items) and assessment-related factors (e.g., specificity of skill, wording of item, cultural bias) is recommended to explain the underlying causes of these findings.