Music therapy and music-based support for autistic children: a neurodiversity-informed systematic review and meta-analysis

Abstract

Music-based interventions, including music therapy, are widely used to support autistic children, but reductions on autism-related scales do not fully represent communication, participation, quality of life, or autonomy. We retained clinically interpretable outcomes while using a neurodiversity-informed framework to organize intervention goals and interpretation. Web of Science, CNKI, Wanfang Data, VIP, PubMed, the Cochrane Library, Embase, Wiley Online Library, EBSCOhost, and Google Scholar were searched from inception to August 5, 2026. Randomized, quasi-randomized, crossover, and controlled studies involving autistic participants younger than 18 years were eligible. Outcomes were grouped as social communication/interaction, language, adaptive function/participation, quality of life/well-being, and autism-related characteristics and support needs. Positive Hedges’ g values favored the music-based condition. Random-effects models used restricted maximum likelihood estimation and modified Hartung-Knapp inference. Comparator intensity, technique, duration, and parent involvement were examined as exploratory moderators. Thirty-three independent studies, representing 1,880 participants in the selected quantitative comparisons, were included and contributed at least one outcome. Social communication/interaction showed a significant pooled effect (k = 10, g = 0.64, 95% CI 0.25 to 1.03, p = 0.0046; I² = 59.6%); the primary evidence layer also remained significant (k = 7, g = 0.57, 95% CI 0.02 to 1.12, p = 0.043). Language showed a significant small pooled effect (k = 11, g = 0.32, 95% CI 0.10 to 0.54, p = 0.009; I² = 0%), whereas the primary evidence layer yielded g = 0.19 (95% CI -0.17 to 0.55). Autism-related scale outcomes showed a significant pooled effect (k = 16, g = 1.23, 95% CI 0.93 to 1.53, p < 0.001; I² = 77.0%) and remained significant among studies reporting a reproducible randomization method (k = 6, g = 0.95, 95% CI 0.29 to 1.61). Adaptive function and quality-of-life estimates did not reach statistical significance. No between-group moderator test was significant. Duration was not associated with effect size (β = -0.034 per month, p = 0.552), and Egger’s regression did not detect funnel-plot asymmetry (p = 0.591). Music-based support produced favorable average effects on social communication/interaction, language, and autism-related scale outcomes.