Characterizing mechanisms of caregiver-mediated naturalistic developmental behavioral interventions for autistic toddlers: A randomized clinical trial.

There was no untreated control group. Every family received an active eight-week programme, so the study can compare the two coaching approaches but cannot show that either one works. The trial was powered for its original primary aim, not for these outcomes. Power to detect the reported effect for spontaneous communication during the play sample was 63% straight after intervention and 73% at follow-up, and only 31% and 55% for the mother-child interaction context, so some comparisons here may be underpowered. The joint engagement analyses were exploratory and done after the fact, and power for them ranged from 50% to 98%. Only mothers took part, so nothing here speaks to fathers or other caregivers. Children had to be 18 to 48 months old and not yet using flexible phrase speech, so the findings say nothing about autistic children who already talk in phrases, or about older children. 23 of the 111 children did not have a formal medical diagnosis of autism when they joined; their autism characteristics were identified using the ADOS-2 together with clinician judgement. 84 of the 111 families completed the three-month follow-up, and missing data were handled by carrying the last recorded score forward, which assumes children did not change after they stopped being assessed. The two groups differed at the start in how often they showed supported joint engagement, which the analysis adjusted for. The paper does not test whether how well each mother actually used the strategies was related to how much her child improved.

The article text does not include a funding or competing-interests statement. The authors state that there was no community involvement in the study.

Do not read this as showing that directive, prompting-heavy strategies are the better way to teach autistic children to talk, or that responsive approaches should be dropped. Every family received an active eight-week parent programme and there was no untreated control, so nothing here shows either approach beats no intervention, and the responsive group showed longer stretches of shared attention. Several of these comparisons were underpowered, and the joint engagement and mediation analyses were done after the fact.