Autism: accuracy of primary care screening leaves room for improvement

  • Pediatrics

  • International Clinical Digest
Access to the full content of this site is available only to registered healthcare professionals. Access to the full content of this site is available only to registered healthcare professionals.

Takeaway

  • Universal primary care-based screening for autism spectrum disorder (ASD) using the Modified Checklist for Autism in Toddlers with Follow-Up (M-CHAT/F) had low accuracy, especially in certain groups.

Key results

  • 91% of all children were screened.
  • ASD prevalence: 2.2%.
  • M-CHAT/F performance:
    • Sensitivity: 38.8%.
    • Specificity: 94.9%.
    • Positive predictive value: 14.6%.
    • Negative predictive value: 98.6%.
  • Sensitivity was higher in toddlers who were:
    • Older (aged 21-26 months; 48.8%).
    • Screened twice (51.1%).
  • Positive predictive value was lower in girls (7.7%).
  • Specificity and positive predictive value lower in toddlers:
    • Who were black, Asian, other race, multiracial (90.4%-93.8% and 10.8%-13.4%).
    • From lower-income households (92.3% and 11.8%).

Study design

  • Prospective cohort study among 25,999 children aged 16-26 months having follow-up through age ≥4 years.
  • Universal, primary care-based screening conducted using M-CHAT/F, supported by electronic administration, integration into electronic health records.
  • Main outcome: M-CHAT/F performance against reference standard of diagnosis in electronic health record.
  • Funding: Allerton Foundation; Eagles Charitable Foundation; National Institute of Mental Health; National Institutes of Health.

Limitations

  • Unclear generalisability.
  • Diagnostic data available only through age 4-8 years.
  • Not all eligible children had follow-up interview.