Mechanistic trials, therapy and developmental science—An exemplar from early autism care

Abstract Background Mechanistic design and analysis in clinical trials remains relatively rare in child mental health and autism, despite the considerable value that it could have in developing therapy practice and in illuminating basic science. Clinical trials themselves continue to have insufficient influence on actual clinical practice in child mental health. Methods Aspects of clinical trial design, including co‐design, are identified that could increase their face validity for clinicians. A number of recent developments in mechanistic analysis of complex intervention trials in mental health are described that can also increase the face validity and utility for clinicians, and provide causal inferences for clinical and developmental theory that are useful transdiagnostically. Results These possibilities are exemplified with findings from the Paediatric Autism Communication Therapy (PACT) trial programme in early autism care. This is one of the largest in the field and unique in achieving a long‐term trial follow‐up with preserved intention to treat analysis. Mechanistic analysis in this programme identifies a key cascade of mediation, with therapeutic changes in early child dyadic social initiation mediating generalised endpoint as well as follow up autism outcomes. These findings identified causal inferences regarding the responsiveness of sociability in autistic children to environmental adaptation in therapy, with implications for mechanism across developmental science. They also facilitated successful adaptation of this therapy into other non‐specialist environments. Conclusions Well‐designed mechanistic trials identify active processes underlying therapy effects. They have in consequence powerful potential to facilitate therapy adaptation and transdiagnostic refinement. They can also uniquely illuminate developmental theory, by allowing causal inferences in developmental science that are difficult in other ways.