Concordance between Community Supervisor and Provider Ratings of Fidelity: Examination of Multi-Level Predictors and Outcomes

Fidelity is conceptualized as a key mechanism impacting treatment outcomes, yet community examination of fidelity is limited, partly due to limited feasible training methods and tools. Provider-report tools represent a feasible method for improving routine fidelity monitoring. Additional work examining the effectiveness of these methods and concordance with other fidelity sources is needed. Further, provider and client factors impact both intervention implementation and concordance, but their impact on fidelity measurement is poorly understood. Using a train-the-trainer methodology, we examine concordance between three methods of assessing fidelity (trained independent coders, supervisor evaluation and provider self-report) using a fidelity assessment tool adapted for community use. Multilevel models examined predictors (provider- and child-level factors) and outcomes (provider average fidelity) of concordance. Results suggest supervisors and providers are able to use the fidelity tool, but demonstrated variable concordance, with higher concordance with trained coders for supervisors than providers. Provider credentials and child language were marginally predictive of rater concordance. Higher supervisor accuracy was associated with higher provider fidelity. Findings inform broader efforts to improve the development and implementation of feasible fidelity tools for routine use in community settings and highlight multi-level factors impacting the adoption and implementation of such tools.