Early detection of autism spectrum disorder (ASD) is critical yet disparities persist in screening accuracy and identification within underserved populations. This study sought to replicate the only other prior study examining universal ASD screening (Guthrie et al. in Pediatrics 144(4):e20183963, 2019, 10.1542/peds.2018-3963) and evaluate the diagnostic accuracy of the modified checklist for autism in toddlers (M-CHAT) in a diverse cohort of toddlers screened in a large rural healthcare system. A retrospective review was completed of 5300 primary care screening visits from 3956 children aged 16–30 months between 2015 and 2024. Screening completion rates at 18 months, 24 months, and both time points were examined, as well as diagnostic accuracy (sensitivity, specificity, positive predictive value, negative predictive value) both overall and within child/sociodemographic subgroups. Approximately 30% of children received screenings in alignment within AAP recommendations. Overall, the M-CHAT demonstrated moderate sensitivity (49.2%) and high specificity (94.2%), with a positive predictive value (PPV) of 28.9% and negative predictive value of 97.5% for detecting ASD. PPV increased substantially (91%) when identifying ASD or any other neurodevelopmental/behavioral condition. Diagnostic accuracy varied by child characteristics including sex, gestational age, and primary language. Children screening positive received an ASD diagnosis 20 months earlier on average. Findings provide evidence largely replicating and extending prior research, highlighting the importance of universal ASD screening. The M-CHAT demonstrates strong accuracy in ruling out ASD and identifying any neurodevelopmental/behavioral condition but shows poor to moderate predictive ability in detecting ASD specifically.
Harper et al. (2026) studied this question.