Comparative effects of non-pharmacological intervention categories on cognitive domains in children with autism spectrum disorder: A systematic review and network meta-analysis of randomized controlled trials.

Abstract

This systematic review and network meta-analysis aimed to compare the effects of non-pharmacological intervention categories on attention, working memory, comprehensive executive function, and social-emotional cognition in children with autism spectrum disorder. Five databases were searched for randomized controlled trials published from January 2005-12 April 2026. Random-effects pairwise meta-analyses and frequentist network meta-analyses were conducted using standardized mean differences with 95% confidence intervals. Risk of bias and confidence in network estimates were assessed using the Cochrane Risk of Bias 2 tool and CINeMA, respectively. Thirty-six reports met the eligibility criteria, and 35 reports involving 1694 children contributed to the quantitative synthesis. Non-pharmacological interventions produced a statistically significant moderate overall improvement in cognitive outcomes (standardized mean difference = 0.54, 95% confidence interval 0.36-0.71), with significant benefits for working memory (0.61, 0.09-1.13) and social-emotional cognition (0.47, 0.30-0.65). Social-emotional cognition was the only domain with a sufficiently connected network for comparisons across intervention categories; social-behavioral and technology- and robot-assisted interventions showed the clearest benefits relative to active controls. Evidence networks for attention, working memory, and comprehensive executive function were sparse or disconnected. Confidence was moderate for selected social-emotional cognition comparisons but low or very low for most estimates. Non-pharmacological interventions may improve selected cognitive outcomes in children with autism spectrum disorder, but effects appear domain specific. More adequately powered head-to-head trials using standardized outcomes are needed.