posted by Mark Hamelburg, Greg Berger, Tom Kornfield*
on October 30, 2019
In the past decade, enrollment in Medicare Advantage (MA) has increased dramatically. Today, more than 22 million Americans – over one-third of all Medicare beneficiaries – are enrolled in MA plans. These plans provide financial security by limiting out-of-pocket costs, offering integrated drug coverage, and providing a rich array of benefits not available in traditional Medicare, including dental, vision, hearing, and other supplemental benefits. Enrollees in MA plans are highly satisfied with the MA program.
MA plans provide these benefits at the same cost as traditional Medicare costs, while achieving better clinical outcomes. And in areas of the country where MA is popular, additional enrollment leads to slower traditional Medicare spending growth as providers employ MA practice patterns and care guidelines for their remaining traditional Medicare patients.
Beneficiaries can choose to have their Medicare benefits covered through a MA plan. These plans must cover all Medicare Part A (hospital) and Part B (physician) services and may offer additional supplemental benefits, as well as prescription drugs under the Part D program. The Centers for Medicare & Medicaid Services (CMS) calculates “benchmarks” for each county and MA plans submit bids against these county benchmarks to determine their payment rate. By law, the county benchmark is based on what the equivalent costs would be for Medicare if that individual were enrolled in traditional, or fee-for-service, Medicare.
CMS calculates benchmarks based on claims experience for traditional Medicare enrollees. Specifically, CMS calculates a per capita rate separately for Part A and Part B services and adds them together to arrive at the benchmark. As a result, CMS includes beneficiaries with both Part A and Part B, as well as beneficiaries with only Part A or only Part B, in the calculation. However, only Medicare enrollees with both Part A and Part B are permitted to enroll in a MA plan.
CMS’ methodology is valid only if the rate obtained by calculating Part A and B per capita spend separately is the same as the rate calculated when only including individuals with both Part A and B. However, the Medicare Payment Advisory Commission (MedPAC) has noted that the per capita costs for an enrollee with Part A or B coverage is less than the per capita cost for someone with Part A and B coverage. In particular, MedPAC found that “Part A spending for beneficiaries enrolled in Part A and Part B all year averaged 8% more than average Part A spending for beneficiaries enrolled in Part A (with or without Part B).”
As such, MedPAC recommended that CMS calculate the benchmarks based on spending for those who are enrolled in both Part A and Part B. In fact, CMS already calculates rates for Puerto Rico using claims for individuals with both Part A and B. CMS’ current methodology results in an inaccurate estimate of the county payment rates.
Using summary claims data from CMS, we have estimated what the 2019 county benchmarks would have been had CMS used individuals with Part A and B coverage and compared those benchmarks with the published 2019 county benchmarks. As shown here, CMS’ methodology significantly understates the county benchmarks, and by a large amount in certain geographies. But first, we discuss the relationship between the percentage of enrollees who have Part A only and the impact on the level of the county benchmarks.
Upon reaching Medicare eligibility age, enrolling in Part A is free, but enrolling in Part B requires paying a monthly premium. A beneficiary may choose not to enroll in Part B for multiple reasons upon reaching age 65. For instance, a beneficiary may decline Part B if they have retiree coverage. In addition, a beneficiary who is actively working may choose to decline Part B. In recent years, the number of Americans working past the age of 65 has increased. In 2016, over one-third of those with Part A only were active workers.
Based on an analysis of Medicare enrollment files, we found that the percentage of enrollees with Part A only has increased in the past 10 years. In 2008, 8.6% of enrollees in traditional Medicare had Part A only, while in 2017, this percentage increased to 12.1%. This increase occurred at a time when Medicare managed care enrollment doubled from 9.2 million in 2008 to 18.4 million in 2017. As MA enrollment increases, the percentage of enrollees left in traditional Medicare without Part B also increases, because an enrollee must have Part A and B to enroll in MA.
Exhibit 1 shows the relationship between Medicare managed care enrollment and the Part A only rate from 2008 to 2017. Of note, more and more managed care enrollees are living in counties with higher Part A only rates. By 2017, fewer than one-quarter of Medicare managed care enrollees lived in counties with a Part A only enrollment rate below 10%, while 38% of managed care enrollees lived in counties where at least 15% of enrollees had Part A only.
Source: Analysis of 2008-2017 CMS 100% Denominator Files
Exhibit 2 shows the variation in the Part A only rates by state, as well as the relationship between the Part A only rate and the Medicare managed care penetration rate. The rate of Part A only enrollment varies significantly, from a low of 7% in Mississippi to a high of 23.7% in Hawaii. Five other states – California, Minnesota, Nevada, New York, and Rhode Island – along with the District of Columbia have Part A only enrollment of above 15%.
Source: Analysis of 2017 CMS 100% Denominator File
The 2019 FFS rate for each county is the United States Per Capita Cost, or USPCC, multiplied by the county’s cost relative to the national average, also known as the Average Geographic Adjustment, or AGA. The county benchmark is based on the product of the USPCC and the AGA, where the AGA is standardized by the average risk score for the county.
To determine the USPCC, CMS estimates a Part A per capita rate and a Part B per capita rate for the payment year. CMS sums these two rates together to arrive at the national USPCC. To calculate the AGA, CMS calculates the Part A enrollment and total spending, and the Part B enrollment and spending, for a five-year period. For the 2019 rates, CMS used claims from 2012 to 2016. For each county, CMS calculates a Part A and Part B per capita monthly payment for each year from 2012 to 2016. These data are then averaged and risk adjusted to arrive at the AGA – which represents that county’s cost compared to the national average. As an example, an AGA of 1.02 would mean that the county costs are 2% higher than the average. CMS’ county benchmark methodology has been largely unchanged since it first began determining these payment rates in 1985. A more detailed methodology of the approach used by CMS, including the underlying data for each county from 2012 to 2016, is published here.
If CMS were to calculate the benchmarks for MA by excluding Part A only enrollees, benchmarks would increase considerably for the vast majority of counties. In a small number of counties that serve less than 2% of MA enrollees, benchmarks would be lower. However, it should be noted that the change in methodology could be phased in, or it could be done in such a way as to hold counties harmless.
The impacts would also vary considerably by geography, which in many ways is expected given the differences cited earlier about the rate of enrollees with Part A only. Exhibit 4 shows the impacts by State. The impact would be largest in the District of Columbia, followed by Hawaii. As shown in Exhibit 2, the managed care enrollment rate is much higher in Hawaii than in the District of Columbia, and both have high levels of Part A only enrollment.
Exhibit 5 shows the results in Metropolitan Areas which had over 100,000 enrollees in MA as of July 2019 – these areas represent about half of all MA enrollment. The impacts range from a low of 3% in Tampa to a high of 11.5% in New Orleans.
Correcting the benchmark calculation would have significant impacts on benchmarks, and these impacts would vary considerably by area. By CMS’ own admission, the benchmarks would be different if the calculation were fixed. In the 2020 Final Rate Notice, CMS stated, “While we recognize that calculating rates based on data that excludes beneficiaries entitled only to Part A would yield different results than calculating rates based on our current methodology, that fact alone does not determine which methodology should be employed.”
However, a more accurate representation of expected costs for an enrollee in MA would exclude Part A only enrollees, since these enrollees are prohibited from choosing MA. The disparity in county benchmark rates resulting from the current CMS methodology will only continue to grow as Medicare managed care penetration increases – the Congressional Budget Office projects that there will be 36 million beneficiaries (over 40%) in Medicare managed care by 2029.
*Tom Kornfield was a Vice President at AHIP when this blog was authored. He is currently a Senior Consultant at Avalere Health.