Statistical analysis plan for the COMPARE trial: a 3-arm randomised controlled trial comparing the effectiveness of Constraint-induced Aphasia Therapy Plus and Multi-modality Aphasia Therapy to usual care in chronic post-stroke aphasia (COMPARE)

BACKGROUND: While high-quality meta-analyses have confirmed the effectiveness of aphasia therapy after stroke, there is limited evidence for the comparative effectiveness of different aphasia interventions. Two commonly used interventions, Constraint-induced Aphasia Therapy Plus (CIAT Plus) and Multi-modality Aphasia Therapy (M-MAT), are hypothesised to rely on diverse underlying neural mechanisms for recovery and may be differentially responsive to aphasia severity. COMPARE is a prospective randomised open-blinded end-point trial designed to determine whether, in people with chronic post-stroke aphasia living in the community, CIAT Plus and M-MAT provide greater therapeutic benefit compared to usual care, are differentially effective according to aphasia severity, and are cost-effective. This paper details the statistical analysis plan for the COMPARE trial developed prior to data analysis. METHODS: Participants (n = 216) are randomised to one of three arms, CIAT Plus, M-MAT or usual care, and undertake therapy with a study trained speech pathologist in groups of three participants stratified by aphasia severity. Therapy occurs for 3 h blocks per day for 10 days across 2 weeks. The primary clinical outcome is aphasia severity as measured by the Western Aphasia Battery-Revised Aphasia Quotient (WAB-R-AQ) immediately post intervention. Secondary outcomes include WAB-R-AQ at 12-week follow-up, and functional communication, discourse efficiency, multimodal communication, and health-related quality of life immediately post intervention and at 12-week follow-up. RESULTS: Linear mixed models (LMMs) will be used to analyse differences between M-MAT and UC, and CIAT-Plus and UC on each outcome measure immediately and at 12 weeks post-intervention. The LMM for WAB-R-AQ will assess the differences in efficacy between M-MAT and CIAT-Plus. All analyses will control for baseline aphasia severity (fixed effect) and for the clustering effect of treatment groups (random effect). DI