Treatment fidelity of brief motivational interviewing and health education in a randomized clinical trial to promote dental attendance of low-income mothers and children: Community-Based Intergenerational Oral Health Study "Baby Smiles".

Background: Fidelity assessments are integral to intervention research but few published trials report these processes in detail. We included plans for fidelity monitoring in the design of a community-based intervention trial.

Methods: The study design was a randomized clinical trial of an intervention provided to low-income women to increase utilization of dental care during pregnancy (mother) or the postpartum (child) period. Group assignment followed a 2 × 2 factorial design in which participants were randomly assigned to receive either brief Motivational Interviewing (MI) or Health Education (HE) during pregnancy (prenatal) and then randomly reassigned to one of these groups for the postpartum intervention. The study setting was four county health departments in rural Oregon State, USA. Counseling was standardized using a step-by-step manual. Counselors were trained to criteria prior to delivering the intervention and fidelity monitoring continued throughout the implementation period based on audio recordings of counselor-participant sessions. The Yale Adherence and Competence Scale (YACS), modified for this study, was used to code the audio recordings of the counselors' delivery of both the MI and HE interventions. Using Interclass Correlation Coefficients totaling the occurrences of specific MI counseling behaviors, ICC for prenatal was .93, for postpartum the ICC was .75. Participants provided a second source of fidelity data. As a second source of fidelity data, the participants completed the Feedback Questionnaire that included ratings of their satisfaction with the counselors at the completion of the prenatal and post-partum interventions.

Results: Coding indicated counselor adherence to MI protocol and variation among counselors in the use of MI skills in the MI condition. Almost no MI behaviors were found in the HE condition. Differences in the length of time to deliver intervention were found; as expected, the HE intervention took less time. There were no differences between the overall participants' satisfaction ratings of the HE and MI sessions by individual counselor or overall (p > .05).

Conclusions: Trial design, protocol specification, training, and continuous supervision led to a high degree of treatment fidelity for the counseling interventions in this randomized clinical trial and will increase confidence in the interpretation of the trial findings.