Evaluating BOSCC and ELSA as Outcome Measures in the Context of a JASPER Intervention Trial

Overview

Research teams in Boston University, UCLA, and Weill Cornell will recruit 90 participants in 3 years (10 participants per site per year) and evaluate JASPER, play-based intervention, using the BOSCC and ELSA-T. Participants will be randomized to receive JASPER facilitated either by a clinician or the caregiver. After 10 weeks, the participants will be evaluated using the CGI to determine if they are "responders" or "non-responders". Nonresponders will be given a mix of clinician and caregiver-facilitated JASPER and responders will remain the course for the following 10 weeks. Coding of the BOSCC and ELSA-T will be the outcome measures and change will be evaluated throughout the study.

Participants: 90 children with confirmed diagnoses of ASD (based on ADOS and clinical judgement) between 18 months and 5 years of age will be randomized into clinician-mediated vs. parent-mediated JASPER intervention at Weill Cornell (n=30; PI Dr. Kim), BU (n=30, PI Dr. Tager-Flusberg), and UCLA (n=30, PIs Drs. Lord \& Kasari). Minimally verbal children eligible for the ADOS Toddler Module or Module 1 (pre or minimally verbal) or those who have emerging phrases who may receive Module 2 (phrase speech) will be enrolled. The sample will be limited to children with nonverbal mental ages greater than or equal to 12 months. Pre-Intervention Behavioral Testing: We will administer the ADOS and Mullen Scales of Early Learning (MSEL), ESCS, SPA, and CCX. Parents will complete the interview version of the Vineland Adaptive Behavior Scales -3rd edition. In addition, the investigators will administer ELSA-T (following the same procedures that we developed in the first phase of this project described above) as well as the BOSCC with clinicians and caregivers (independent, standardized administration by caregivers). A subset of children (n=20) will be invited within a week of the first assessment for another ELSA/BOSCC sessions for test-retest reliability. Responder vs. non-responder criteria: Children who show improvements in core ASD symptoms based on the treating therapists' CGI-Improvement ratings will be considered as "responders." Children who show stability or worsening on the CGI-I will be considered as "non-responders." Children will be stratified based on age and language levels across sites, and randomized into 10-week clinician- vs. caregiver- mediated JASPER interventions initially by the statistician at BU who will not be directly involved in the data collection. Both "responders" and "non-responders" will have monthly check-ins (2 total) with the research staff for an additional 10 weeks. Both groups will proceed to T3 assessments following the 20 weeks of treatment and check-ins. At T2 (after first 10 weeks) and T3 (after 20 weeks, end of study), the investigators will again administer ELSA-T, BOSCC, as well as CGI, ESCS, SPA, and CCX. Coding ELSA: Naïve coders will be trained to use ELAN to mark the beginning and end of each child speech vocalization/utterance and each examiner/parent utterance, which are coded on separate tiers. Coders will be blind about which kind of intervention the child received and when the ELSA recording was collected pre- or post-intervention. ELAN provides as output frequency of child utterances and frequency of adult utterances (corrected per unit time). The investigators have developed an algorithm that takes ELAN exported files containing the start and end time of each utterance from each speaker to derive a measure of conversation turn-taking (number of consecutive utterances produced by same speaker). Note that these two measures are not independent, but clinical trials could define either one as a primary outcome measure depending on whether the targeted outcome focuses on the child's communicative abilities or social engagement. ELSA-T recordings will be collected at three sites, but all ELSA coding will be completed at BU. Coding BOSCC: PI Dr. Kim's research team has trained, blind coders who will score the BOSCC without previous knowledge of the child and the time points of interactions. Different coders will be assigned for the clinician vs. caregiver BOSCC videos and for the parent behavioral coding. BOSCC videos will be collected at three sites, but all BOSCC coding will be done at Weill Cornell. The same research activities will be conducted at UCLA, Boston University, and Weill Cornell.