Integrating the Early Start Denver Model into Israeli community autism spectrum disorder preschools: Effectiveness and treatment response predictors.

The study was not randomised; preschools showed interest in implementing the Early Start Denver Model and comparison preschools were selected by child age, so group differences may reflect selection. The sample was small and nested in eight preschools, with 4 to 9 assessed children per setting, so the authors could not use multilevel modelling or examine many covariates without losing statistical power. The Early Start Denver Model and comparison programmes differed in manualisation, fidelity requirements, parent involvement, and progress monitoring, so the active ingredients cannot be separated. Parents were encouraged to practise the model at home, but no fidelity data were taken on what parents actually delivered at home. The outcome measures were broad developmental and adaptive tests, so children who improved in more specific skills may not have been detected. Children whose parents did not consent received the preschool intervention but were not assessed, so the results may not represent all children in the preschools. The study did not examine implementation feasibility or social validity of the preschool-based Early Start Denver Model.

Do not conclude that the Early Start Denver Model alone caused the gains. The groups differed in manualisation, fidelity, parent involvement, and progress monitoring, and the study was not randomised.