Algorithms to Improve Fairness in Medicare Risk Adjustment.

Area of Science:

• Health economics
• Health equity
• Algorithmic fairness

Background:

• Payment system design significantly influences healthcare spending, access, and outcomes.
• Medicare Advantage represents over half of Medicare spending, making its risk adjustment algorithm crucial for broad impact.

Purpose of the Study:

• To evaluate algorithmic tools for equitable Medicare risk adjustment payment.
• To maintain performance, flexibility, feasibility, transparency, and interpretability.

Main Methods:

• Retrospective analysis of Medicare enrollment and claims data (2017-2020).
• Utilized demographic indicators and hierarchical condition categories to predict subsequent year Medicare spending.
• Employed constrained regression and postprocessing to assess fairness and performance.

Main Results:

• Constrained regression and postprocessing achieved fair spending targets with minimal reduction in payment system fit (12.6%-12.7%).
• Postprocessing increased payments for minoritized racial/ethnic groups.
• Constrained regression benefited minoritized groups and those in socioeconomically disadvantaged areas.

Conclusions:

• Constrained regression and postprocessing offer feasible methods to integrate fairness into Medicare risk adjustment.
• These algorithmic adjustments can help policymakers address health care disparities via payment reform.