Objectives: A population of 16,235 children aged 18 months was screened using the Checklist for Autism in Toddlers (CHAT) to identify childhood autism (CA). Two further screening procedures were conducted at age 3 and 5 years. The population was followed up at age 7 years in order to establish the sensitivity, specificity, and positive predictive value of the instrument.
Method: A brief checklist assessing joint attention and pretend play behaviors was administered by primary health care practitioners when the children were 18 months old. Follow-up methods included screening through parents and health practitioners and checking medical and educational records.
Results: Nineteen cases of CA were successfully identified by the CHAT at 18 months. At follow-up a total of 50 cases of CA were identified via all surveillance methods. Thus, the CHAT has a sensitivity of 38% and a specificity of 98% for identifying CA. The positive predictive value of the instrument was maximized by concentration on the highest-risk group. Repeated screening 1 month later increased the positive predictive value to 75% for identification of CA but reduced the sensitivity to 20%, although the specificity was close to 100%. The screen also identified cases of pervasive developmental disorder as well as children with language and other developmental disorders.
Conclusions: The CHAT can be used to identify cases of autism and related pervasive developmental disorders at 18 months of age. It is emphasized that the CHAT is not a diagnostic instrument but can identify potential cases of autism spectrum disorders for a full diagnostic assessment.