Real-World Outcomes of a Telephone-Based Virtual Cognitive Rehabilitation Therapy Program: A Retrospective Cohort Analysis

Background: Cognitive rehabilitation (CR) is an established behavioral intervention that improves daily functioning for individuals with mild cognitive impairment (MCI) and early-stage dementia. Traditional models of in-person delivery limit access, particularly for individuals living in rural areas. This study evaluated the efficacy of a novel telephone-based virtual CR model combining speech-language pathologist (SLP)-led sessions with cognitive exercises delivered by an automated voice agent between visits. Methods: We conducted a retrospective observational analysis of 141 older adults who completed treatment to discharge (58% female; mean age 71.2, standard deviation 10.8 years; MCI diagnosis rate 61.7%, dementia diagnosis rate 29.1%; Montreal Cognitive Assessment mean score 20.8, standard deviation 4.3). Changes in four outcome measures from initiation of treatment to discharge were evaluated for statistical significance. The four outcomes studied were patient-reported quality of life and three therapist-rated Functional Communication Measures (FCMs): overall cognition, spoken language, and language comprehension. Changes were compared to FCM averages from the American Speech-Language-Hearing Association National Outcomes Measurement System (ASHA NOMS). Models were developed to predict changes in outcome measures based on patient demographics, clinical status, program engagement and treating therapist. Results: All four outcomes improved significantly over the course of treatment (p<0.05), with medium to very large effect sizes. Mean changes in the three FCM outcomes exceeded ASHA NOMS benchmarks for in-person outpatient care. A majority of patients saw an improvement in each clinical outcome measure. Models with meaningful predictive power were identified for changes in all outcome measures except the FCM for language comprehension. Baseline cognitive function was the most influential and negatively correlated predictor of an improvement in overall cognitive abilities and language expression. Baseline quality of life was the dominant and negatively correlated predictor of improvement in quality of life. Conclusions: Telephone-based virtual CR led by SLPs with automated exercises delivered by a voice agent produced clinically meaningful functional and quality of life gains relative to external benchmarks for in-person clinical practice. These results support the use of virtual CR within post-diagnostic care for older adults experiencing cognitive impairment, particularly for rural and underserved communities.