Interim treatment fidelity of SFA RCT (Babiak et al., 2025)

Purpose: This study evaluated treatment fidelity for an ongoing randomized controlled trial comparing two variants of semantic feature analysis (SFA) treatment for naming impairment in aphasia. In this trial, participants are being randomized to treatment conditions (many- or few-features) in which they are asked to generate either 11 or five features per SFA trial.

Method: Treatment fidelity was addressed via study design, manualization of the treatment, provider training, and maintenance of daily treatment logs, among other approaches. Fidelity was assessed through protocol adherence checks, interrater reliability of feature counting, and analysis of treatment dosage. Protocol adherence was measured using a 20-item checklist. Interrater reliability was evaluated for feature counts on 10% of treatment sessions. Treatment dosage was analyzed to confirm differentiation between many-features and few-features groups in terms of features generated and exposed and equivalence in terms of total treatment time.

Results: Overall protocol adherence was 99%. Interrater reliability for feature counting showed good-to-excellent agreement. Analysis of treatment dosage in terms of features per trial confirmed clear differentiation between groups, with the many-features group generating or being exposed to approximately 10–11 features per trial compared to four to five features for the few-features group. Analysis of other aspects of dosage also suggested good fidelity to the design.

Conclusions: This ongoing trial of SFA is being implemented with high levels of treatment fidelity. The study demonstrates the importance and feasibility of comprehensive fidelity monitoring in aphasia treatment research. Detailed fidelity procedures and outcomes support interpretation, replication, and clinical implementation of study findings.

Supplemental Material S1. 16-page protocol that provides details regarding treatment execution, as well as examples of general and second-level cues, determining strong versus weak features, response coding, and the sentence generation task.

Supplemental Material S2. Treatment dosage by participant. Specific data includes blinded participant ID, anomia severity, group (many vs. few), number of treatment hours, total number of trials, average features generated per trial, average features exposed per trial, total features generated, and total features exposed.

Supplemental Material S3. HTML R markdown document providing the code for running all the statistical models reported in the manuscript, as well as three data frames including anonymized subject-level data.

Babiak, M. C., Hula, W. D., Autenreith, A., Nader, M. M., Hula, S. A., Swiderski, A., Cavanaugh, R., Nunn, K., Johnson, J. P., & Dickey, M. W. (2025). Interim treatment fidelity for a randomized controlled comparative effectiveness trial of two variants of semantic feature analysis treatment for aphasia. American Journal of Speech-Language Pathology, 34(4), 2081–2097. https://doi.org/10.1044/2025_AJSLP-24-00331