Treatment fidelity in a pragmatic clinical trial of music therapy for premature infants and their parents: the LongSTEP study

Background Treatment fidelity (TF) refers to methodological strategies used to monitor and enhance the reliability and validity of interventions. We evaluated TF in a pragmatic RCT of music therapy (MT) for premature infants and their parents. Methods Two hundred thirteen families from seven neonatal intensive care units (NICUs) were randomized to receive standard care, or standard care plus MT during hospitalization, and/or during a 6-month period post-discharge. Eleven music therapists delivered the intervention. Audio and video recordings from sessions representing approximately 10% of each therapists’ participants were evaluated by two external raters and the corresponding therapist using TF questionnaires designed for the study (treatment delivery (TD)). Parents evaluated their experience with MT at the 6-month assessment with a corresponding questionnaire (treatment receipt (TR)). All items as well as composite scores (mean scores across items) were Likert scales from 0 (completely disagree) to 6 (completely agree). A threshold for satisfactory TF scores (≥4) was used in the additional analysis of dichotomized items. Results Internal consistency evaluated with Cronbach’s alpha was good for all TF questionnaires (α ≥ 0.70), except the external rater NICU questionnaire where it was slightly lower (α 0.66). Interrater reliability measured by intraclass correlation coefficient (ICC) was moderate (NICU 0.43 (CI 0.27, 0.58), post-discharge 0.57 (CI 0.39, 0.73)). Gwet’s AC for the dichotomized items varied between 0.32 (CI 0.10, 0.54) and 0.72 (CI 0.55, 0.89). Seventy-two NICU and 40 follow-up sessions with 39 participants were evaluated. Therapists’ mean (SD) TD composite score was 4.88 (0.92) in the NICU phase and 4.95 (1.05) in the post-discharge phase. TR was evaluated by 138 parents. The mean (SD) score across intervention conditions was 5.66 (0.50). Conclusions TF questionnaires developed to assess MT in neonatal care showed good internal consistency and moderate interrater reliability. TF scores indicated that therapists across countries successfully implemented MT in accordance with the protocol. The high treatment receipt scores indicate that parents received the intervention as intended. Future research in this area should aim to improve the interrater reliability of TF measures by additional training of raters and improved operational definitions of items.