Diagnostic Accuracy of the Autism-Spectrum Quotient: A Brief Systematic Review and Meta-Analysis

The Autism-Spectrum Quotient (AQ) is widely used to assess autistic traits and is recommended in clinical guidelines to screen for and assess autism in adults. However, evidence for its accuracy largely comes from case–control studies, which tend to overestimate diagnostic performance. This systematic review and meta-analysis (1) evaluated how well the AQ predicts autism diagnostic status in adults, (2) calculated pooled sensitivity and specificity of the AQ, and (3) considered number of false positives and false negatives across different prevalence levels. Eligible studies needed to administer the AQ prior to a diagnostic assessment to confirm diagnostic status. Twelve studies were synthesised, with up to 2536 people included in each analysis. Random effects meta-analysis estimated pooled area under the curve (AUC) values, and bivariate random effects models pooled sensitivity and specificity. Predictive values were calculated across prevalence rates from 10% to 70%. The pooled AUC was 0.65 (95% CI 0.58–0.71), indicating limited discrimination. Sensitivity was moderate to high, but specificity was consistently low. False positives outnumbered true positives unless autism prevalence was over 40%. In clinically relevant samples, the AQ shows limited diagnostic accuracy and high false positive rates. Its use in adult clinical pathways should be approached with caution, and guideline recommendations may require re-evaluation.