Impact of the Medicare Annual Wellness Visit on Geographic Variation in Dementia Diagnosis

Abstract

Background: Alzheimer's-disease-and-related-dementia (ADRD) is often misdiagnosed or diagnosis is late in disease progression. Diagnosis variations can be driven by access variations related to geographic location. To improve timely and accurate ADRD diagnosis, the Medicare Annual Wellness Visit (AWV) required a cognitive assessment, starting in 2011, which may reduce diagnosis variations.

Objective: Assess impact of the AWV on geographic variation in ADRD diagnosis and outcomes.

Methods: Aim 1: ADRD patients and caregivers from western, central, and eastern Maryland were interviewed to understand diagnosis-pathway geographic differences. Data were analyzed using interpretative phenomenological analysis to identify themes. Aim 2: Using the CMS Chronic Condition Data Warehouse (CCW) and HRSA Area Health Resource Files, 5-year, county-level ADRD cumulative incidence were compared pre- (2006-2010) and post- (2011-2015) AWV implementation in Mid-Atlantic states. Geographically-weighted, generalized linear models assessed the association between the AWV and ADRD cumulative incidence, controlling for demographic and access measures. Aim 3: A retrospective cohort study using CCW was conducted in newly diagnosed ADRD individuals. Health care utilization (HCU) was compared for individuals with an AWV pre-diagnosis versus no AWV. Difference-in-difference models assessed 10-month outcomes between exposure groups.

Results: Aim 1: Average time from first doctor visit (concerning signs/symptoms) to ADRD diagnosis was 3.3, 2, and 5.3 years for western, central, and eastern regions, respectively. Aim 2: AWV participation was not significantly associated with increased 5-year ADRD cumulative incidence. The association between AWV and 5-year ADRD cumulative incidence varied by county with stronger associations clustered in eastern Virginia, Maryland, and Delaware. Aim 3: Receiving an AWV pre-diagnosis was associated with increased HCU [ED (rate ratio [RR]: 1.20, 95% confidence interval [CI]: 1.11, 1.30), hospitalizations (RR: 1.26, 95% CI: 1.13, 1.40), and outpatient (RR: 1.08, 95% CI: 1.04, 1.12)].

Conclusions: Longer times to ADRD diagnosis were observed in rural versus urban regions. The AWV demonstrates the potential to minimize geographic differences by increasing diagnosis rates and HCU. Due to low and variable participation during early implementation, the true impact of the AWV is yet to be established. It may take longer to see impacts on longer-term outcomes.