Parent-Mediated Interventions for Infants under 24 Months at Risk for Autism Spectrum Disorder: A Systematic Review of Randomized Controlled Trials.

The trials were small, from 16 to 103 infants each, and the smallest compared 11 coached infants with 5 controls. Results from numbers that small swing easily. Parents could not be blinded, and many outcomes were rated by those same parents, who knew whether they had been coached. Six of the seven trials were judged at high risk of bias for this. Overall, four trials were rated low risk of bias, two unclear and one high. Most control families were referred to, or already using, community services, and the review says it does not have enough information on the type or intensity of what they received. In two trials the control group got more service than the coached group, so no better than the control group here does not mean no better than nothing. There is no agreed definition of being at risk for autism. Some of these infants were simply younger siblings of an autistic child with no screening at all, while others had already met cut-offs on autism-specific assessments. They are not the same children. Five of the trials used between six and ten different instruments each, with multiple subscales inside each one. Run that many comparisons and some will come out significant by chance. The trials calculated effect sizes in different ways and nothing was pooled, so the review can describe the trials side by side but cannot weigh them against each other. Follow-up mostly ended early. The review notes that later skills such as language may not have emerged by the time the trials stopped measuring.

The text supplied carries no funding statement or conflict declaration for the review itself. The reviewers did assess the trials they included. Three of the seven had authors with a financial stake in what was being tested: royalties from the intervention materials, book royalties and positions in research and pharmacological companies, and intellectual property rights over instruments used to measure the outcomes. In one of those trials the authors held rights to three of the ten instruments used, which the reviewers say may have influenced its design. One trial gave no conflict statement at all, and three declared none.

Do not read this as proof that coaching parents of at-risk babies is a waste of time. Almost every control family in these trials was referred to or already using community services, and in some trials the control group received more therapy than the coached group. That makes this a comparison against other care, not against nothing. The review itself says the absence of treatment effects may not mean the interventions were ineffective. What it does rule out is prevention. No trial showed that coaching changed whether a child was later diagnosed.