Quality Matters: Application of Intervention Fidelity Benchmarks to Community Paraprofessional Delivered Social Communication Intervention for Toddlers with Autism

Background: Community Partnered Participatory Research can support the adoption of research-based practices in real world settings. The transfer of intervention protocols into community practice can be challenging, but deployment trials have demonstrated success sharing strategies with caregivers (Green et al., 2010; Kasari et al., 2014) and community educators (Chang et al., 2016). Although achieving 80% fidelity is a typical research standard, it is not as clear what constitutes “good enough” fidelity by which a minimum standard for program implementation may lead to empirical change in children’s target outcomes.

Objectives: To explore whether implementation benchmarks previously published (Shire, Shih, & Kasari, 2018) for caregiver-mediated Joint Attention, Symbolic Play, Engagement, and Regulation (JASPER: Kasari et al., 2008; 2014) intervention would also differentiate children’s outcomes when: (a) delivered by paraprofessional teaching assistants’ (TAs) and (b) when adapted for small group instruction (referred to as jasPEER).

Methods: Forty-eight TAs and 113 toddlers (mean age=32.28 months) across four classrooms were randomized to JASPER with a peer (jasPEER) or one-to-one JASPER waitlist control. Children had received outside diagnoses of autism (n=1 with Down Syndrome). All but two TAs, and two children were members of an ethnic minority group.

Intervention. Significant effects of JASPER over treatment as usual on joint engagement, joint attention gestures, play skills, and language were demonstrated with this center in a prior trial (Shire et al., 2018). Building upon TAs’ knowledge of one-to-one JASPER, they were provided with two weeks of in-vivo training and ongoing video feedback from the research team to learn the jasPEER adaptation. Children in waitlist classrooms received JASPER for 30 minutes a day for 11 weeks, while children in immediate jasPEER received the adapted peer version during this time.

Measures. Ten-minute video recorded TA-child interactions at entry, exit, and one month follow up were coded for children’s play level (number of 1-minute intervals in simple, combination, pre-symbolic, and symbolic play) and TAs’ implementation (total strategy implementation score).

Results: Examining TA’s implementation fidelity at exit, 34 children received sessions at 75% or greater fidelity (“high fidelity”). The number of children receiving high fidelity sessions did not significantly differ between JASPER and jasPEER. Significant growth in children’s pre-symbolic play (t(244)=2.39, p=0.017) was found for children in the high fidelity group, while gains were not significant for children in the lower fidelity group. Concurrent to this shift to higher level play was a significant decrease in combination play for the high fidelity group only (t(244)=-2.59, p=0.010). Simple and symbolic play occurred too infrequently to model.

Conclusions: Findings replicate differential outcomes for children when either intervention model was delivered with at least 75% implementation fidelity than below 75%. Considering the age and developmental heterogeneity of this diverse community sample of toddlers with ASD, gains in pre-symbolic play where children move beyond combining objects to extend acts to self and agents are clinically significant.