Adaptive Interventions and SMART Designs: Three Case Studies in Autism

Background:   Due to the great heterogeneity of children with autism spectrum disorders (ASDs), effective treatment often requires individualized, sequential decision-making. To do this, each child’s treatment is dynamically tailored over time based on the child’s changing state, including response to prior treatment. Adaptive interventions (also known as dynamic treatment regimens) operationalize such individualized decision making using a sequence of decision rules that specify which intervention option to offer, for whom, and when. Intervention options correspond to varying doses, types or delivery modes of pharmacological or behavioral treatments. The sequential, multiple-assignment, randomized trial (SMART) is a type of multi-stage randomized trial research design used to build high-quality adaptive interventions. 

Objectives:   The objectives of this talk are to (i) provide a brief introduction to adaptive interventions, and (ii) present case studies of three SMART designs in autism.  

Methods:   The first SMART study was designed to develop an adaptive intervention for children with ASD who are minimally verbal using a speech-generating device in the context of a naturalistic intervention involving joint attention, symbolic play, engagement and regulation (JASPER) plus enhanced milieu training (EMT).  The second study, again among children with ASD who are minimally verbal, was designed to develop an adaptive intervention involving JASPER+EMT, discrete trials training (DTT), parent training, combined JASPER+EMT+DTT and a clinician-rated clinical global impressions measure to monitor child’s progress.  The third is a SMART pilot study, which aims to develop an adaptive intervention to improve social engagement outcomes in children with ASD in inclusive school settings. This novel study examines the feasibility and acceptability of adaptive interventions involving Remaking Recess (a school-level intervention), Classroom Supports (a classroom-level intervention), Parent- and Peer-mediated social skills interventions, and the use of paraprofessionals to monitor child progress on the playground and inform subsequent treatment decision-making. For each study, we present the SMART design, its rationale, and the scientific questions. 

Results:   Three SMART studies have been designed to develop adaptive interventions in children with autism. Each is designed to answer novel questions concerning how best to individualize treatment in children with ASD. The first study is completed; the second and third studies are in the field.  

Conclusions:  Adaptive interventions are intended to serve as a guide for sequential treatment decision-making in actual clinical practice. They hold the promise of improving outcomes for greater numbers of children with autism by providing the right treatment to children who need it, when they need it. Currently, there is great interest in the use of SMART designs to build high-quality adaptive interventions in autism.