Development, Testing and Use of a Targeted, Efficient Fidelity Measure in the Attachment and Biobehavioral Catch-up Intervention

Introduction: Measurement of fidelity is critical to ensure that an intervention is being implemented as intended, both in the lab and in the field. However, for a fidelity measurement to be adopted for use in the field, it must be feasible and efficient as well as reliable and valid (Schoenwald et al., 2011). We designed and tested a fidelity measure for the Attachment and Biobehavioral Catch-up (ABC) intervention, and adapted this measure for use with trainees at dissemination sites.

Two features distinguish our measure of fidelity. First, the measure is focused on one critical component of intervention, the use of in the moment comments, or live feedback, to respond to parents' intervention-relevant behavior with their children. Second, because the measure is so targeted, we are able to code fidelity from 5-minute video clips of intervention sessions. 

Study 1: We designed a research measure of fidelity. In a sample of 50 cases from our lab, we found that the frequency of in the moment comments in an early session of intervention predicted parental sensitivity in a later session, even when controlling for earlier parental sensitivity. 

Study 2: We simplified the fidelity measure for use with dissemination sites. We examined associations between fidelity and parent behavior change in a sample of 56 cases seen by 7 interventionists at 7 different agencies in Hawaii.  We found that the frequency and on-target hit rate of in the moment comments in the first half of intervention predicted parental change in sensitivity and intrusiveness. In this study, parent behavior was measured with a semi-structured play activity, independent of the intervention sessions.

Study 3: When learning to implement ABC, interventionists are trained to code their own fidelity. Undergraduate-level expert coders provided feedback to interventionists about their coding. In this study, we found that interventionists became more reliable coders over time, as measured by increased inter-rater reliability over time with expert coders. 

Conclusions: Results suggest that our adapted measure of ABC fidelity is reliable across expert coders, and is valid as a predictor of parent behavior change during the intervention, both in the lab and in the field. We are currently using the fidelity measure as a tool to train interventionists to reflect on their performance, and examining the ability of parent coaches to code reliably.

[2427 characters/2800 character limit]