A Randomized Trial to Identify Accurate Measurement Methods for Adherence to Cognitive-Behavioral Therapy

•We tested the relative performance of three CBT adherence measurement approaches.•Each approach was compared to direct observation (DO)•Semistructured role-plays yielded adherence scores comparable to DO.•Self-report and interviews yielded inflated adherence estimates. Clinician fidelity to cognitive behavioral therapy (CBT) is an important mechanism by which desired clinical outcomes are achieved and is an indicator of care quality. Despite its importance, there are few fidelity measurement methods that are efficient and have demonstrated reliability and validity. Using a randomized trial design, we compared three methods of assessing CBT adherence—a core component of fidelity—to direct observation, the gold standard. Clinicians recruited from 27 community mental health agencies (n = 126; M age = 37.69 years, SD = 12.84; 75.7% female) were randomized 1:1:1 to one of three fidelity conditions: self-report (n = 41), chart-stimulated recall (semistructured interviews with the chart available; n = 42), or behavioral rehearsal (simulated role-plays; n = 43). All participating clinicians completed fidelity assessments for up to three sessions with three different clients that were recruited from clinicians’ caseloads (n = 288; M age = 13.39 years SD = 3.89; 41.7% female); sessions were also audio-recorded and coded for comparison to determine the most accurate method. All fidelity measures had parallel scales that yielded an adherence maximum score (i.e., the highest-rated intervention in a session), a mean of techniques observed, and a count total of observed techniques. Results of three-level mixed effects regression models indicated that behavioral rehearsal produced comparable scores to observation for all adherence scores (all ps > .01), indicating no difference between behavioral rehearsal and observation. Self-report and chart-stimulated recall overestimated adherence compared to observation (ps