Screening for autism in pre-school children in primary care: systematic review of English Language tools.

Objective: To review the accuracy of brief screening tools for autism in pre-school children.

Methods: Systematic review of diagnostic accuracy studies. Methods: Medline, Embase, Cinahl and Psychlit plus references of identified papers and contact with authors. Methods: Children and infants aged 5 years or less without a prior diagnosis of autism or pervasive development delay. Methods: Tools/checklists appropriate for use in screening for autism in primary care settings. Methods: Sensitivity, specificity, positive and negative predictive value of screening tools and likelihood ratios relative to a diagnostic assessment made using either DSM-III/IV or ICD 10 diagnosis.

Results: Three studies considering two tools were identified. The CHecklist for Autism in Toddlers (CHAT) was tested on an appropriate population sample with moderate long-term follow-up but demonstrated poor sensitivity and positive predictive value. Weaker evidence suggested that the Modified CHecklist for Autism in Toddlers (M-CHAT) had high sensitivity but follow-up was of shorter term and less comprehensive.

Conclusions: The CHAT demonstrated a level of sensitivity unlikely to be useful for population screening purposes, however, its high specificity suggests it has utility in secondary screening. The M-CHAT is a parent only report and might be more sensitive, and therefore appropriate for population screening. However, full conclusions regarding its accuracy cannot be drawn until follow-up data has been collected.