Identifying Coding Intensity in Medicare Advantage Through Switchers.

Abstract

Objective: To estimate the extent of differential coding of health risk in traditional Medicare (TM) compared with Medicare Advantage (MA).

Study setting and design: Payments to MA plans are based on reported medical conditions, and research has shown the number and severity of diagnoses are larger when beneficiaries are enrolled in MA plans rather than TM. We compare the risk scores of Medicare beneficiaries who switch from TM into MA over the 2013-2021 period to the scores of beneficiaries who stay in TM, incorporating heterogeneous treatment effects across switching cohorts and over time.

Data sources and analytic sample: We use a 10% sample of Centers for Medicare and Medicaid Services data containing individual risk scores and enrollment status for 2012-21. After applying exclusion criteria, our sample consists of 1,911,968 beneficiaries with data available for each year. We also link administrative data to the Medicare Current Beneficiary Survey to assess measures of health status.

Principal findings: We find the risk scores of switchers to MA were 0.120 points (12.4%; 95% confidence interval [CI]: 12.0%-12.8%) higher than stayers in the second year, 0.166 points (17.2%; 95% CI: 16.7%-17.6%) higher in the third year, and 0.216 points (22.3%; 95% CI: 21.7%-22.9%) higher by the sixth year after switching. Averaged over all MA enrollees in 2021, our estimates suggest coding intensity in MA led to risk scores that were 18.6% higher than for comparable enrollees in TM.

Conclusions: Our estimates of coding intensity are at the higher end of the range in the prior literature while addressing concerns of endogenous switching. Our estimates of increasing coding over time and across enrollment cohorts can help inform decisions regarding adjustments to MA payments for coding intensity.