Attributing Racial Differences in Care to Health Plan Performance or Selection.

Area of Science:

• Health Services Research
• Health Equity
• Healthcare Policy

Background:

• Public reporting and financial incentives increasingly link to organizational performance on health equity metrics.
• Variation in organizational performance may stem from actual differences or selection bias.

Purpose of the Study:

• To determine if observed racial disparities across health plans are due to performance or selection bias.
• To assess the impact of random assignment versus self-selection on health equity metrics.

Main Methods:

• A natural experiment using a southern US state's random assignment of Medicaid enrollees to 5 managed care plans.
• Analysis of enrollee-level administrative claims and enrollment data (2011-2015) for Black and White enrollees.
• Comparison of outcomes (primary care visits, ED visits, prescription fills, spending) between self-selected (observational) and randomly assigned (randomized) populations.

Main Results:

• In the randomized population, racial differences in healthcare utilization and spending did not vary significantly across plans.
• In the observational population, significant variation in racial differences was observed across plans, exceeding that in the randomized group.
• Selection bias, not performance differences, likely explains the observed variation in racial disparities in the observational group.

Conclusions:

• Selection bias can misrepresent health plans' performance on health equity measures.
• Relying on disparity measures for payment adjustments in Medicaid may have unintended consequences.
• Careful consideration of enrollment mechanisms is crucial when evaluating health plan performance on equity.