META-ANALYSIS OF IDENTIFICATION TOOLS FOR AUTISM SPECTRUM DISORDER

Abstract

Focus on the early identification of autism spectrum disorder (ASD) has been made, although some scientists and policy-makers have questioned early authentic and credible identification for ASD. The aim of the present meta-analysis was to investigate the diagnostic accuracy of the various identification tools for ASD. A comprehensive literature search was organized across various databases, including PubMed, PsycINFO, and Scopus. Keywords included "Autism Spectrum Disorder," "identification tools," "screening instruments," "diagnostic tools," and "meta-analysis." Studies published from 2015 to 2024 were considered. The Bayesian model was employed to evaluate the accuracy of identification tools. Pooled sensitivity = 82%, specificity = 79% was effective in early screening but with variable predictive value in different age groups for the M-CHAT; pooled sensitivity = 81%, specificity = 77% demonstrated good performance in distinguishing ASD from other developmental disorders for the SRS; high sensitivity = 78, specificity = 76% considered a gold standard for comprehensive diagnostic assessment for the ADI-R; sensitivity = 79%, specificity = 75% found highly effective in a variety of settings, though more resource-intensive for the ADOS; sensitivity = 71%, specificity = 74% observed highly effective in a variety of settings, though more resource-intensive for the CARS; and sensitivity = 70%, specificity = 71% found moderately effective in a variety of settings, though more resource-intensive and needs specialize training for the AOSI. The researchers concluded that identification tools for ASD demonstrated consistent statistically significant results and are adequate thence to identify ASD at 12–36 months.