Comparative effects of different types of exercise on improving motor impairments in children and adolescents with Autism: A network meta-analysis

Abstract

Background Motor impairments are highly prevalent and clinically relevant in children and adolescents with ASD, yet the relative effects of different exercise modalities on motor function remain unclear. Methods We conducted an NMA of RCTs evaluating structured exercise interventions versus control in children and adolescents with ASD aged 3–18 years. Eligible trials included interventions lasting ≥ 4 weeks and reported standardized motor outcomes (e.g. TGMD/TGMD-2, BOT/BOT-2, MABC/MABC-2). Exercise programs were grouped into six modalities: CT, AS, BG, MSG, MBE and ES. A random-effects NMA was performed to estimate standardized mean differences (SMDs) and derive P-scores. Network heterogeneity and inconsistency were assessed using Q statistics and node-splitting. Small-study effects and potential publication bias were examined with funnel plots, Egger tests, trim-and-fill and selection models. Certainty of evidence was evaluated using CINeMA. Results Sixteen RCTs conducted in ten countries met the inclusion criteria. Exercise overall was associated with a moderate-to-large improvement in motor outcomes versus control (study-level pooled SMD 0.81, 95% CI 0.51–1.11). At the modality level, CT showed the largest benefit (SMD 1.10, 95% CI 0.61–1.60; P-score 0.81), followed by AS (SMD 0.69, 95% CI 0.34–1.10; P-score 0.78). BG (SMD 0.65, 95% CI 0.18–1.10; P-score 0.62; one trial) and MSG, MBE and ES (SMDs 0.23–0.47) showed small-to-moderate, statistically uncertain effects. Network heterogeneity and inconsistency were low, but funnel plots, Egger tests and selection models indicated small-study effects and potential publication bias, with bias-adjusted pooled SMDs decreasing to small-to-moderate magnitudes. CINeMA rated most comparisons as low or very low certainty, mainly due to within-study bias, reporting/publication bias and imprecision. Conclusions Structured exercise appears to yield group-level improvements in motor outcomes in ASD, with CT and AS emerging as promising modalities in this NMA. However, the certainty of the underlying evidence is low, and small-study effects may inflate some estimates. These findings should be interpreted cautiously and considered hypothesis-generating; larger, rigorously reported RCTs with harmonised motor outcomes are needed to confirm or revise the current modality rankings.