Objectives/Goals: Telehealth introduces opportunities to improve access to Alzheimer’s disease and related dementias (ADRD) diagnostic services. We aim to investigate the effect of the 2020 Centers for Medicare and Medicaid Services (CMS) telehealth expansion (where virtual cognitive assessments were covered) on county-level prevalence of ADRD diagnoses. Methods/Study Population: A county-level difference-in-difference (DID) design was employed to investigate the change in ADRD prevalence among US 65+ fee-for-service beneficiaries from pre-expansion (2016-2019) to post-expansion (2021-2022) by comparing counties with and without broadband. 2016-2022 county-level ADRD prevalence estimates were defined by Chronic Conditions Data Warehouse methodology from the CMS Mapping Medicare Disparities Tool. Broadband presence was defined as counties with ≥1 high-speed internet provider and average download/upload speeds of 25/3 megabits-per-second. The parallel trends assumption was assessed using graphical and model-based approaches. Negative binomial regression models estimated the average effect of the CMS expansion on ADRD prevalence among counties with broadband. Results/Anticipated Results: Average county-level ADRD prevalence was 11.06% across the pre-expansion period and 7.15% post-expansion. Broadband was present in 70.35% of counties pre-expansion and 91.96% of counties post-expansion. Differences in ADRD prevalence by broadband were constant across the pre-expansion period. ADRD prevalence was 2.48% higher in counties with broadband versus counties without broadband post-expansion compared to pre-expansion (95% confidence interval (CI): 1.01, 1.04). Discussion/Significance of Impact: Improvements in ADRD detection may have occurred in counties with broadband following the CMS telehealth expansion. However, the effect is dwarfed by the substantial reduction in ADRD prevalence from pre- to post-expansion. This reduction is likely a consequence of the COVID-19 pandemic, which may have decreased overall diagnostic opportunities.
Rajczyk et al. (Wed,) studied this question.