Examining the Efficacy of Culturally Responsive Interventions for Autistic Children and Their Families: A Meta-Analysis.

Only 24 studies were pooled, 11 of them randomised trials, and the authors report that the analyses testing dose, setting, delivery mode and who delivered the intervention were probably underpowered to detect anything. Egger's test indicated publication bias for the child outcomes, so the pooled effect for children may be larger than the true effect. The studies adapted their interventions to very different degrees and described too little of that process for the authors to judge quality, so one pooled number mixes thorough cultural adaptation with token change. The authors state they cannot tell whether the gains came from the intervention itself or from it being more culturally matched to the families, because no study compared an adapted with an unadapted version of the same intervention. Fourteen of the studies ran in families' homes; whether the same effects hold in schools, clinics or community centres is unknown, and the authors call for that work. Single-case and group studies were combined using a conversion method the authors argue is appropriate, but they acknowledge the controversy about overestimating effects when both designs are pooled.

The article text supplied contains no funding statement and no declaration of competing interests, so there is nothing here to say about who paid for this work or what the authors declared.

Do not read the large pooled effect as proof that culturally adapting an intervention is what made it work. The review included no study comparing an adapted and a non-adapted version of the same intervention, and the authors state it remains unknown whether the gains came from the intervention alone or from it being more culturally relevant to the families.