Eli Lilly recently commissioned Milliman to analyze the risk scores and claim costs for Medicare beneficiaries with mild cognitive impairment (MCI) or Alzheimer’s disease (AD). For the risk scores, we analyzed the impact of reclassifying MCI into a distinct hierarchical condition category (HCC) within a risk score model introduced by the Centers for Medicare and Medicaid Services (CMS) in 2024 as the next evolution of Medicare Advantage (MA) risk adjustment. Our analysis involved performing a ground-up rebuild of the risk score model using the same data period (2018 diagnoses and 2019 expenditures) and methods used by CMS to develop the 2024 CMS-HCC Version 28 risk score model with the 100% fee-for-service Medicare claims data. With regard to healthcare costs, we examined relative differences in costs and key quality measures for Medicare beneficiaries diagnosed with MCI or AD compared with matched controls. We also looked at medical (Part A/B) and pharmacy (Part D) claim costs and key Healthcare Effectiveness Data and Information Set (HEDIS) quality measures associated with beneficiaries diagnosed with MCI and AD, comparing results with undiagnosed matched control cohorts.

Key findings:

• CMS-HCC risk scores are under-projected by 5% to 10% for patients with MCI under the current risk-adjustment model.
• MCI would be assigned a risk score coefficient of 0.15 to 0.20 if included in the model.
• All statistical measures tested indicated model accuracy would improve by including MCI.
• Healthcare costs were higher for the cohorts diagnosed with MCI and AD.
• Comorbidities amplify cost differentials.
• Inpatient services accounted for the largest share of costs in the month of incident diagnosis for those diagnosed with MCI and AD.
• Healthcare costs were higher than controls for those treated with amyloid-targeting therapies.
• The performance of proxy HEDIS quality metrics for MCI and AD cohorts was generally similar to matched controls.

This report was commissioned by Eli Lilly.