Medicare Advantage Plan Double Bonuses Drive Racial Disparity In Payments, Yield No Quality Or Enrollment Improvements

Abstract

Under the Medicare Advantage Quality Bonus Payment Program, initiated in 2012, MA plans in eligible counties, with relatively high quality performance, and located in counties deemed eligible, receive quality bonuses twice as large as counterparts in doubling-ineligible counties. Relative to ineligible counties, eligible counties have larger populations and higher MA enrollment. Using national data for 2008 through 2018, we found double bonuses were not associated with either higher plan quality or increased enrollment. Additionally, because Black beneficiaries were less likely to reside in eligible counties, double bonus payments increased spending by $60 per year for Black enrollees $60, compared to $91 for White enrollees. Our findings suggest double bonuses are not only ineffective in achieving desired improvements, but also foster racially inequitable distribution of Medicare funds that disfavors Black MA enrollees. Our study supports eliminating double bonuses, saving an estimated $1.7 billion per year.

Introduction

Enrollment in Medicare Advantage – private plans for Medicare beneficiaries – has grown remarkably, increasing from 11.1 million in 2010 (24% of beneficiaries) to 24.1 million in 2020 (36%).(1) Fueled by generous payments from the Centers for Medicare and Medicaid Services (CMS), Medicare Advantage offers an attractive set of benefits coupled with typically modest premiums and cost-sharing. However, some policymakers have argued that Medicare Advantage plans are overpaid(2) and questioned the program’s value. In response, the Affordable Care Act (ACA) cut payments to Medicare Advantage plans while simultaneously creating the Quality Bonus Payment (Bonus Program).(3) Originally implemented as a demonstration in 2012, the Bonus Program is now a fixture of Medicare Advantage. It provides financial bonuses to plans that achieve higher star ratings (out of a possible five),_ based on quality of preventive care, chronic disease management, patient experience, and plan performance. The Bonus Program is intended to ensure the quality of care in Medicare Advantage (MA) plans, which receive risk-adjusted capitated payments and thus may otherwise face an incentive to ration needed care.

An unusual feature of the MA Bonus Program is the delineation of “double-bonus” counties. In these counties, higher quality plans receive certain MA bonuses at double the dollar level paid to comparably performing plans in counties ineligible for double bonuses. Through the ACA, Congress created three criteria a county must meet, to be eligible for double bonuses: historically high MA enrollment (at least 25% in 2009); low Medicare fee-for-service (FFS) spending (below the national average in a given year); and a 2004 “urban floor” designation, given to metropolitan statistical areas [MSA] with at least 250,000 residents that qualify for the minimum MA benchmark rate, granted to low FFS spending areas). While the number of counties qualifying for double bonus status is small – around 7% of counties nationally – their impact is large because 27% of MA beneficiaries live in these counties.

The intent of double bonuses is to generate savings for Medicare by increasing MA enrollment in areas with high Medicare Advantage penetration and low FFS spending. Applying this policy at the county-level takes advantage of the fact that insurers are allowed to selectively introduce competitive MA plans in certain counties and not others. Historically, insurers have sought to increase enrollment by offering more Medicare Advantage plans in counties that are eligible for double bonuses.(4) Given the opportunity for larger bonuses, MA insurers may also implement strategies to increase quality of care via enhanced care management programs or passing incentives for quality through to provider organizations in their networks.

At the same time, double bonuses might drive geographic or racial differences in MA payment rates. The capacity to produce both unwanted outcomes becomes apparent when one considers the case of Wayne and Lapeer Counties, in Michigan. Wayne County, which includes Detroit, has a population of 1.8 million, and 39% of its Medicare beneficiaries are Black. Despite the county’s urban floor designation, high-quality MA plans serving local beneficiaries were not eligible for double bonuses in 2018; Wayne County’s average annual FFS spending was just above the national average (too high) and its Medicare Advantage enrollment was 1 percentage point (pp) below the minimum 25% cutoff in 2009.

Lapeer County is 55 miles to the north, but belongs to the same MSA as Detroit. It has a population of only 88,000, and only 0.4% of its Medicare beneficiaries are Black. Unlike Wayne County, Lapeer meets all three requirements for MA bonus doubling (i.e., also meets the FFS spending and enrollment criteria). A high-quality plan serving Lapeer beneficiaries, therefore, would receive double bonuses. Put simply, if MA plans of equally high quality served Wayne and Lapeer residents, the one serving more Black enrollees would receive only half as many quality bonus dollars.

Because CMS expects quality bonus payments to be partially passed on by MA plans to beneficiaries, in the form of enhanced benefits or premium support, differences in allocation of MA bonus payments to counties eligible and not eligible for doubled bonuses could drive geographic or racial disparities. We define such disparities as differences in treatment, access, or costs not justified by differences in health status or preferences.(5,6) For instance, differences in double-bonus eligibility could lead to fewer benefits—such as coverage for dental care or hearing aids—being offered to MA enrollees in one geographical area but not another. It could also translate to higher premiums for the same benefits when offered to primarily Black versus primarily White populations, which could harm the financial well-being of Black beneficiaries.

Despite the substantial investments by CMS in double bonuses for MA plan quality, it is unclear whether those double bonuses are allocated inequitably, and if so, across which groups. Moreover, there is limited evidence of whether the policy has achieved its intended goals of improving quality and enrollment since its introduction to Medicare Advantage in 2012.

In this study, we used national Medicare data from 2008 and through 2018 and difference-in-differences methods to address three questions. First, are double bonuses associated with changes in quality of care in Medicare Advantage? Second, are double bonuses associated with changes in enrollment in Medicare Advantage? Because counties entered double-bonus eligibility in a staggered manner from 2012 onward, we used event-study methods to compare changes in MA quality and enrollment in double-bonus counties and non-double-bonus counties, before and after the start of double-bonus eligibility. Finally, do differences in double-bonus eligibility drive disparities in payments for plans to care for Black versus White beneficiaries?

Background

Quality-based Payment in Medicare Advantage

Payments to Medicare Advantage plans are based on average risk-adjusted Medicare FFS spending in the plan’s county (the “benchmark”).(4) For each plan, private insurers bid to offer coverage to MA beneficiaries. If an insurer bids above the benchmark, the beneficiary pays the difference via a premium. If an insurer bids below the benchmark, the insurer shares between 50% and 70% of the savings (the “rebate”), all which must be returned to beneficiaries via premium support, enhanced benefits, or lowered copayments or deductibles. With the introduction of the Bonus Program in 2012, CMS began linking these payments to plan performance in CMS’ star rating system (Appendix exhibit A1).(7)

Under the Bonus Program, higher quality plans receive financial incentive payments via two mechanisms: a “benchmark bonus” (an additional 5% of the benchmark) and a “rebate bonus” (with shared savings rising from 50% for 3-star or lower plans to 70% for 4.5-star or higher plans). In double-bonus counties, higher quality plans receive benchmark bonuses that are twice as large (10% versus 5%). The Bonus Program has evolved since its inception. Plans with at least 3 stars received bonuses in 2012–2014, while only those with at least 4 stars have remained eligible since 2015. The Bonus Program has also discontinued some quality measures (e.g., glaucoma testing) and introduced others (e.g., hospital readmissions). Other quality measures have been a consistent focus of the program since 2012, including those assessed in this study. The CMS double-bonus policy has also remained a central feature throughout the Bonus Program.

Methods and Data

Exposure to Double Bonuses.

We used the Medicare Advantage Ratebook to determine counties’ staggered entry into double-bonus eligibility from 2012 through 2018 (Appendix exhibit A2).(7) As counties could gain or lose double-bonus eligibility during the study, we defined double-bonus exposure for difference-in-difference analyses by whether the beneficiary resided in a double-bonus county with current or prior double-bonus eligibility.

Medicare Advantage Quality.

To evaluate changes in Medicare Advantage quality, we used Optum’s de-identified Clinformatics® Data Mart Database. Because MA plans are administered by private insurers, there is no single database that includes claims for MA beneficiaries nationwide. However, the Optum database represents the largest commercial MA claims database in the U.S. We defined performance on 9 quality performance measures that were consistently included in the Bonus Program and that we could estimate using insurance claims from the concurrent year of analysis (Appendix exhibit A2).(7) This included 6 Medicare Advantage Part C measures: breast cancer screening (mammogram in year of analysis); four diabetes measures (annual hemoglobin A1c testing, low-density lipoprotein [LDL] screening, retinopathy screening, nephropathy management); and any use of a disease-modifying anti-rheumatic drug (DMARD) for rheumatoid arthritis. We used A1c testing instead of A1c control, as the latter cannot be measured using claims data. We also included 3 measures related to Medicare Part D prescription drug coverage: adherence to statins; adherence to diabetes medications; and adherence to renin-angiotensin system [RAS] antagonists, as defined by at least 80% of proportion days covered. As specified by the star rating system, Part C and D measures were defined according to the Healthcare Effectiveness Data and Information Set and Pharmacy Quality Alliance specifications, respectively.(8,9) To estimate annual quality changes more precisely and avoid bias from varying lengths of historical data among beneficiaries in double-bonus versus non-double bonus counties, quality measures did not include look-back periods from prior years. (For more details on measure selection and specification, see the Appendix Methods and Appendix exhibit A3).(7) For quality analyses, we used Optum data to define beneficiary age, sex, and the number of chronic conditions (as defined by CMS Chronic Conditions Data Warehouse specifications). We also linked beneficiaries to U.S. Census data to define area-level characteristics.

We used an event-study framework to test the association of double-bonus eligibility with changes in Medicare Advantage quality, comparing differences in quality performance among plans serving MA beneficiaries in double-bonus versus non-double-bonus counties, before and after the start of double-bonus eligibility. We then estimated the association of double bonuses with quality performance using a difference-in-differences approach performed with two-way fixed effects. Specifically, we estimated a linear regression model at the measure-beneficiary-year level that included an indicator for double-bonus eligibility, county fixed effects, year fixed effects, beneficiary age, sex, number of chronic conditions, and ZIP code tabulation area percent Black, percent with at least high school education, percent living below the federal poverty level, percent unemployed, and median household income.

To ensure comparability of the analytic sample over time, we included beneficiaries ages 50 to 74 who were continuously enrolled in Medicare Advantage during the year. Because the 2016 double-bonus cohort demonstrated differential pre-period trends for several diabetes quality measures (Appendix exhibits A4-A5),(7,9,10) we excluded beneficiaries in counties entering double-bonus eligibility from 2016 through 2018 from event studies of changes in quality and enrollment (1.5% and 4.8% of the Medicare Advantage quality and enrollment samples, respectively). For all quality, enrollment, and equity analyses, we also excluded beneficiaries residing in U.S. territories and those with missing covariate or double-bonus data.

Medicare Advantage Enrollment.

To evaluate changes in Medicare Advantage enrollment, we used the Medicare Beneficiary Summary File to define beneficiary enrollment in Medicare Advantage versus Traditional Medicare (based on January enrollment each year), age, sex, race (non-Hispanic Black, non-Hispanic White, Hispanic, Asian, Other/Unknown), original reason for Medicare entitlement, and Medicaid dual-eligibility (at least one month in year). We used an event-study framework to test the association of double-bonus eligibility with changes in MA enrollment, comparing differences in MA enrollment (versus Traditional Medicare) for beneficiaries residing in double-bonus versus non-double counties, before and after the start of double-bonus eligibility. We then estimated the association of double-bonus eligibility with MA enrollment using a difference-in-differences approach performed with two-way fixed effects and adjusting for the covariates above. We also tested whether changes in Medicare enrollment varied across beneficiary characteristics using analogous difference-in-difference-in-differences models, interacting the covariate of interest (age, sex, race, reason for Medicare entitlement, dual-eligibility) with the double-bonus indicator and year fixed effects. (For more details on model specification, see the Appendix Methods).(7)

Racial Disparities in Double-Bonus Allocation.

To evaluate whether double bonuses were allocated equitably across Medicare Advantage beneficiaries, we again used the Medicare Beneficiary Summary File as above but now restricted to only MA beneficiaries in the double-bonus period (2012–2018). We then compared the probability of residing in a double-bonus county among Black and White MA beneficiaries. We estimated a logistic regression model at the beneficiary-year level that controlled for beneficiary age, sex, race, original reason for Medicare entitlement, Medicaid dual-eligibility, and year fixed effects.

Impact of Double Bonuses on Payment.

Finally, we estimated the effect of double-bonus eligibility on payment rates to Medicare Advantage plans overall. We also tested whether potential racial disparities in double-bonus eligibility drove disparities in payment rates across Black versus White beneficiaries. To calculate payments to MA plans, we linked publicly-available CMS data on plan enrollment, star ratings, and county-specific payment rates from 2012 through 2018. We estimated the effect of double-bonus eligibility by comparing average payments for plans in double-bonus and non-double bonus counties. To do so, we estimated a linear regression model where the outcome was plan payment and the main exposure was whether the plan was in a double-bonus county (versus non-double-bonus county). Models were estimated at the plan-year level, weighted by plan enrollment, and controlled for the plan’s star rating, county-specific payment benchmarks, and year fixed effects (For more details on model specification, see the Appendix Methods).(7)

We performed an array of robustness tests of our event-study results. These analyses included using an alternate definition of double-bonus eligibility that could change in each study year; restricting to counties that do not lose double-bonus eligibility; and restricting the sample to counties with baseline Medicare Advantage enrollment between 20% and 30% (i.e., within 5pp of the double bonus eligibility cutoff of 25% enrollment). For our enrollment analyses, we also tested alternate MAdefinitions (enrollment in December or for the entire year). For our quality analyses, we used a difference-in-differences framework to test for compositional changes among MA beneficiaries that might bias estimated changes in quality. Likewise, for our enrollment analyses, we used a difference-in-difference-in-differences framework to test for differential changes in the composition of MA versus Traditional Medicare beneficiaries.

Analyses used Stata software, version 16 (StataCorp, College Station, TX). Standard errors were clustered at the county level. The source code for the statistical analyses is available at github.com/[redacted]/[redacted].

Limitations

First, differences between double-bonus and non-double-bonus counties could lead to differential changes in Medicare Advantage enrollment or quality in absence of double bonuses. However, our event studies showed parallel pre-period trends and minimal compositional changes after the start of double-bonus eligibility. Estimates were also robust across a range of model specifications. Second, given higher baseline levels of MA enrollment in double-bonus counties, regression to the mean could negatively bias estimated enrollment changes in double-bonus versus non-double-bonus counties.(12) However, estimated changes in enrollment were comparable among counties with baseline MA enrollment between 20% and 30%. Third, our quality analyses focused on process measures that could be assessed with claims data, were consistently included in the Bonus Program, and were applicable to beneficiaries ages 50–74. Thus, our study did not capture potential changes in measures that were based on intermediate health outcomes (e.g., hemoglobin A1c control), applicable to elderly beneficiaries only (e.g., fall reduction), or introduced to the Bonus Program later in our study period (e.g., hospital readmissions). Yet because the process measures included in our study were easier to achieve than excluded measures pertaining to patient outcomes, our estimates likely capture an upper bound of the effects of double bonuses on quality. Fourth, estimated changes in enrollment and quality may not generalize to counties entering double-bonus eligibility after 2015 (which were excluded, as noted above), although estimated changes in MA enrollment and quality were similar when including later double-bonus cohorts. Fifth, estimated changes in quality may not generalize to MA plans not included in the Optum Clinformatics® Data Mart Database. Nonetheless, this national database includes the insurer with the largest share of MA beneficiaries – encompassing more than 25% of this group with approximately 5.7 million beneficiaries in 2019(13) – and these MA contracts have received the largest share of bonus payments under the Bonus Program.(14) Finally, because publicly-available CMS data on individual MA plans does not include beneficiary race, estimates of the impact of double-bonus eligibility on racial disparities in payments assumed that Black and White beneficiaries are similarly distributed across higher and lower quality plans. However, because quality performance is lower for Black beneficiaries than White beneficiaries in Medicare Advantage,(15) the true impact on racial disparities in payment is likely larger than our estimate.

Results

Double-Bonus Exposure.

Between 2012 and 2018, 7% of counties and 19% of Medicare beneficiaries resided in counties exposed to double bonuses in a given year, on average. Although counties could gain or lose double-bonus eligibility, 85% of those counties that were ever eligible for double bonuses remained consistently eligible between 2012 and 2018 (Appendix exhibit A6).(7)

Medicare Advantage Quality.

For the analysis of changes in Medicare Advantage quality, MA beneficiaries contributed 25,938,455 measure-beneficiary-years (see Appendix exhibits A7 and A8 for CONSORT diagram and beneficiary characteristics).(7) In the pre-double bonus period, quality measures were achieved for 69.5% and 68.3% of beneficiaries residing in double-bonus versus non-double counties, respectively (Appendix exhibit A3).(7) Pre-period trends in quality were similar between beneficiaries residing in double-bonus versus non-double-bonus counties (trend, −0.2pp, 95% CI, −0.5, 0.1, Appendix exhibits A9-A10).(7)

Exhibit 1 shows that double bonuses were not associated with changes in quality for Medicare Advantage beneficiaries (difference-in-differences, −0.1 pp; 95% CI, −0.8, 0.6, Appendix exhibit A10).(7) This finding was robust across beneficiary characteristics, individual performance measures, and a range of sample definitions and model specifications (exhibit 1 and Appendix exhibits A11-A12).(7) Additionally, double bonuses were not associated with changes in beneficiary characteristics, suggesting that double bonuses did not cause compositional changes that might bias estimated changes in quality (Appendix exhibit A13).(7)

Exhibit 1. Event study of the association between Double Bonus Exposure and Quality Performance for Medicare Advantage Beneficiaries.

Source: Authors’ analysis of 2008–2018 Optum Clinformatics® Data Mart Database, Medicare Advantage Ratebook, and U.S. Census Data.

Notes: Pp denotes percentage point. Shaded region represents 95% confidence intervals. The event study shows the difference in the estimated probability of a given quality measure among Medicare Advantage beneficiaries in double-bonus versus non-double counties, before versus after the start of double-bonus eligibility. A single linear regression model was estimated at the measure-beneficiary-year level that included indicators for years to the start of double-bonus eligibility, as well as adjusting for county fixed effects, year fixed effects, performance measure fixed effects, and the covariates listed in the Methods. The difference-in-differences estimate of association between double bonuses and quality performance was −0.1pp (95% CI, −0.8, 0.6) (Appendix exhibit A11).

Medicare Advantage Enrollment.

For our study of changes in Medicare Advantage enrollment, Medicare beneficiaries contributed 583,038,891 beneficiary-years from 2008 through 2018 (see Appendix exhibits A14-A15 for CONSORT diagram and beneficiary characteristics).(7) In the pre-double bonus period, MA enrollment was 36% and 20%, respectively, in double-bonus versus non-double-bonus counties (Appendix exhibit A9).(7) Pre-period trends in enrollment were similar between beneficiaries residing in double-bonus versus non-double-bonus counties (trend, 0.3pp per year; 95% confidence interval [CI], −0.1, 0.6, Appendix exhibits A10 and A16).(7)

Exhibit 2 shows that double bonuses were not associated with changes in Medicare Advantage enrollment (difference-in-differences, −0.6pp; 95% CI, −1.6, 0.3, Appendix exhibit A17).(7) This finding was robust across a range of model specifications, sample definitions, and beneficiary characteristics (Appendix exhibits A17-A18),(7) including when restricted to counties with baseline MA enrollment between 20% and30%. In difference-in-difference-in-difference models, double bonuses were not associated with differential changes in the composition of Medicare Advantage versus Traditional Medicare beneficiaries (Appendix exhibit A19),(7) suggesting that double bonuses did not affect beneficiary selection into Medicare Advantage.

Exhibit 2. Event study of association between Double Bonus Exposure and Change in Enrollment in Medicare Advantage versus Traditional Medicare (2008–2018).

Source: Authors’ analysis of 2008–2018 Medicare Beneficiary Summary File and Medicare Advantage Ratebook.

Notes: Pp denotes percentage point. Shaded region represents 95% confidence intervals. The event study shows the difference in the estimated probability of beneficiary enrollment in Medicare Advantage (versus Traditional Medicare) among beneficiaries in double-bonus versus non-double counties, before versus after the start of double-bonus eligibility. A single linear regression model was estimated at the beneficiary-year level that included indicators for years relative to the start of double-bonus eligibility, as well as adjusting for county and year fixed effects and the covariates listed in the Methods. The difference-in-differences estimate of association between double bonuses and enrollment was −0.6 pp (95% CI, −1.6, 0.3) (Appendix exhibit A17).

Racial Disparities in Double-Bonus Allocation.

Exhibit 3 shows county-level proportions of Black Medicare Advantage beneficiaries across double-bonus and non-double counties, indicating that Black beneficiaries were less likely than White beneficiaries to reside in counties offered double bonuses. Exhibit 4 shows that Black MA beneficiaries were 9.9 pp (95% CI, −14.2, −5.6) less likely than White beneficiaries to reside in double-bonus counties, a relative difference of 35% (Appendix exhibit A20).(7) This racial disparity is driven by the finding that Black beneficiaries were 22% less likely to reside in counties with below-average FFS spending, one of three criteria for double-bonus eligibility (Appendix exhibit A20).(7)

Exhibit 3.

Impact of double-eligibility on payments to Medicare Advantage plans, overall and by race, 2012–18

Statistic/Measure	Overall	Black beneficiaries	White beneficiaries	Black-White disparity
Share of Medicare Advantage (MA) beneficiaries residing in double-bonus counties
(percent) _	26.9%	18.8%	28.7%	--
(percentage points)	--	--	--	-.9.9
Number of MA beneficiaries residing in double-bonus counties, 2012–2018 (beneficiary-years)	119,866,713	14,162,321	89,315,036	--
Incremental effect of double-bonus eligibility, on average payment to MA plans
(per beneficiary per year, dollars)	$85	$60	$91	-$31
(per beneficiary per year, percent)	0.8	0.6	0.9	-0.3
Aggregate effect of double-bonus eligibility on payment to MA plans, 2012–2018 (billions of dollars)	$10.2	--	--	-$0.446

Sources: Authors’ analysis of 2012–2018 Medicare Beneficiary Summary File, Medicare Advantage Ratebook, Medicare Advantage Plan Enrollment Data, Medicare Advantage Plan Star Rating Data

Notes: Pp denotes percentage point, CI denotes confidence interval. Each statistic uses a different measure, reported in the row labels. Our analysis estimates the Black-White disparity in the share of MA beneficiaries living in double-bonus counties nationwide at −9.9 percentage points (95% CI, −14.2, −5.6). The analysis of the incremental effect on average payment rates included 243,760 plan-years (representing 101,627,486 beneficiary-years). The incremental effect on average payment was calculated by multiplying the average difference in payments between double-bonus and non-double bonus counties by the percent of beneficiaries residing in double-bonus counties ($317.86 per beneficiary-year x 26.9%=$85.36). The aggregate effect on payment was calculated by multiplying the incremental effect by the number of beneficiaries in double-bonus counties (2012–2018) ($85.36 per beneficiary-year x 119,866,713 beneficiary-years=$10.2b). The aggregate payment disparity was calculated by multiplying the disparity in the incremental effect by the number of Black beneficiaries in Medicare Advantage (2012–2018) (-$31.48 per beneficiary-year x 14,162,321 beneficiary-years=$446m). Race-specific aggregate effects are not presented because they are not meaningful without normalizing to population size.

Impact of Double Bonuses on Payment per Beneficiary.

After adjustment for differences in star ratings, benchmark payments, and county-level differences in Medicare Advantage beneficiary characteristics, MA plans in double-bonus counties received $85 per beneficiary per year (0.8%) more than plans in non-double-bonus counties (exhibit 4). This translates to $1.7 billion in additional MA payments in 2018, or $10.2 billion from 2012 through 2018. However, because Black beneficiaries were less likely to reside in double-bonus counties than White beneficiaries, the increase in payments due to double bonuses was smaller for Black beneficiaries, on average. Assuming that Black and White beneficiaries enroll in high- versus low-quality plans in a similar manner, double-bonus eligibility would increase per-beneficiary per-year payments by $91 for plans to care for a White beneficiary and by only $60 for plans to care for a Black beneficiary, a disparity of -$31 per beneficiary per year, on average. In the aggregate, White Medicare beneficiaries gained $446 million than Black beneficiaries from 2012 through 2018 as a result of double bonuses.

Discussion

In this national study of the impact of double bonuses in the Medicare Advantage Quality Bonus Program, we report three main findings. First, double bonuses were not significantly associated with changes in quality performance among health plans in the largest national commercial Medicare Advantage claims database. Second, double bonuses were not significantly associated with changes in Medicare Advantage enrollment. Third, Black beneficiaries were substantially less likely to reside in counties offered double bonuses than White beneficiaries, thus contributing to racial disparities. These findings suggest that double bonuses are not an efficient or equitable mechanism for improving the Medicare Advantage program.

Our findings support longstanding concerns from the Medicare Payment Advisory Commission (MedPAC) and the Congressional Budget Office (CBO) that the double-bonus policy is inequitable on its face by granting bonuses to highly-rated plans in double-bonus counties that are twice as large as those given to plans with identical quality in non-double-bonus counties.(16) Our results are also consistent with prior ecological findings that double bonuses were not associated with county-level changes in Medicare Advantage enrollment or composite star ratings.(4,18) Our study does not shed light on where these additional payments flowed within the county – benchmark bonuses may be passed on to health care providers, passed through to beneficiaries as enhanced benefits or premium support, or remain with the plan as profit or administrative costs. However, recent evidence suggests that approximately 60% of benchmark changes are passed through to beneficiaries,(17) which would suggest that White beneficiaries gain an average of $19 more per year from double bonuses relative to Black beneficiaries. These payment disparities may exacerbate racial health care disparities in Medicare Advantage, where Black beneficiaries receive lower quality care than White beneficiaries, on average.(15) Thus, the case of the two Michigan counties in this paper’s introduction is not an isolated example. Because these bonuses are allocated along geographic lines that embody a legacy of racial discrimination and segregation, double bonuses in the MA Bonus Program represent are a form of structural racism, contributing to an unequal and unjust distribution of wealth and health care resources across Black and White communities (27, 28).

We estimate that the double-bonus policy increased payments to Medicare Advantage plans by $10.2 billion from 2012 through 2018. There are several explanation was this substantial influx of federal monies did not improve quality or increase enrollment. First, unlike most value-based reforms that directly compensate provider organizations for meeting quality performance incentives, the Bonus Program relies on MA insurers passing quality incentives through to providers, an added layer of complexity that may have blunted the program’s impact.(4) Second, the Bonus Program encompasses a broad mix of population-based quality measures, many of which have proven difficult to improve even within interventions specifically designed for those measures.(23) Finally, the incremental revenue from double bonuses may be outweighed by the marginal cost of increasing quality or enrollment in double-bonus counties, where MA enrollment is by definition already high. However, we observed no difference in enrollment changes among counties with baseline MA enrollment between 20% and 30%. Similarly, it is not clear why improving quality would be costlier in double-bonus versus non-double-bonus counties, in particular given our finding that baseline quality was comparable across the two groups.

Policy Recommendations

Both the MedPAC and CBO have called for elimination of double bonuses, citing the policy’s growing expense and the lack of evidence for its effectiveness.(16,24) Our study supports these concerns, demonstrating that double bonuses are not an efficient or equitable use of taxpayer dollars and disproportionally benefit White beneficiaries relative to Black beneficiaries. Moreover, because double bonuses do not improve quality or enrollment, eliminating them should not negatively impact the Medicare Advantage program. However, because double bonuses were enacted by Congress as part of the ACA, their elimination cannot occur via CMS rulemaking and will instead require Congressional action.

Our study’s findings provide strong support for congressional action to remove double bonus provisions from the MA Bonus Program. In lieu of double bonuses, Congress could also allocate funds to support evidence-based approaches to improving equity in Medicare Advantage. Certain plans such as Kaiser Permanente’s have made substantial strides toward eliminating racial disparities among MA beneficiaries.(25)

Conclusions

This national study contributes important evidence to the ongoing policy debate on the merits of the double-bonus provisions in the Medicare Advantage Quality Bonus Payment Program. The findings indicate that double bonuses have not only failed to improve quality of care or enrollment in Medicare Advantage, but also directly contribute to racial disparities in wealth and health care resources. As Black Medicare beneficiaries increasingly select into MA plans (26) and efforts to expand Medicare Advantage continue, CMS may wish to consider alternative strategies to ensure efficiency, quality, and racial equity in the Medicare Advantage program.