CMS adjusts its payments to plans based on the expected health care costs of each plan’s enrollees. This process is known as risk adjustment. For each Medicare enrollee, the agency calculates a risk score that measures the expected healthcare costs for that enrollee. The higher the risk score, the greater the expected health care costs, and hence the greater the Medicare payments the government makes to the plan. The agency has proposed changes to its risk adjustment model in an effort to address underpayments for certain low-income Medicare beneficiaries (“dual eligibles”), who tend to have higher healthcare costs than other enrollees. As a result of differences in how states cover dual eligibles, the changes could increase payments for some plans, while dramatically reducing payments for others, depending on their location. Avalere experts note that the reduced payments could lead to reduced benefits for some beneficiaries.

“The proposed payment changes from CMS could introduce volatility into the Medicare Advantage market with unintended consequences on beneficiaries in some regions,” said Tom Kornfield, vice president at Avalere. “While payment accuracy is critical, CMS policy also needs to ensure stability for Medicare beneficiaries.”

On October 28, CMS announced proposed future changes to the risk model that is used to adjust payment for MA plans. These changes are intended to improve the adequacy of payments to plans for the higher costs associated with dual eligible beneficiaries. The changes introduced by CMS would increase risk scores for certain populations while reducing them for others.

Risk adjustment affects MA payment in several ways. It adjusts plan payments on an individual basis, meaning plans are paid less for a lower risk score and more for a higher risk score. In addition, CMS uses the risk-adjustment model to create geographic adjustments in payment-called county benchmarks-to account for regional variations in the cost of healthcare. Finally, if plans bid to cover enrollees in a given region at less than the cost of the regional benchmark, the extra payment is granted as a “rebate” that can be used to provide enhanced benefits to enrollees. Reductions in county benchmarks can lead to less additional benefits because the differences between the bids and the benchmarks will be reduced creating fewer opportunities for rebate-funded benefits. The proposed risk model changes have the potential to affect each of these payment streams.

As shown in the figure below, large payment decreases are possible in nearly all counties in California, while portions of Florida (except for Miami-Dade) could see increases.

{embed=”embeds/_article_slideshow” group_name=”MARateBook1″}

Avalere estimates that 7 of the 10 counties with the largest MA enrollment as of September 2015 would experience a lower payment rate from the risk model changes. The largest county in terms of enrollment, Los Angeles, would see a decrease of 3.4 percent. This decrease could mean substantial reductions in additional benefits for beneficiaries in this county. Other large counties that would see reductions include Orange, CA and Queens, NY.

{embed=”embeds/_article_slideshow” group_name=”MARateBook2″}

“Because CMS does not plan to publish the 2017 county benchmarks until April 2016, Medicare Advantage plans may want to ask CMS for this information earlier in order to determine the potential impact of these changes and whether or not they would have an adverse impact on beneficiaries,” says Caroline Pearson, senior vice president at Avalere.

Methodology

Avalere estimated relative risk factors for each county under the new model proposed by CMS, based on: 1) the data published by CMS on October 28, 2015, 2) the calculation data for the Medicare Advantage county ratebook for 2016, and 3) the distribution of Medicare/Medicaid eligibles by county and eligibility type for 2014 and 4) the 100 percent denominator file from CMS. Sources 2) and 3) are data files available on CMS website.

For each county, Avalere first used the county’s five year average risk score from 2009 to 2013 from the calculation file. Then, we derived the risk score for each of the six categories (aged non-dual, aged full-dual, aged partial-dual, disabled non-dual, disabled full-dual, disabled partial-dual) listed in the CMS website by applying an estimate of the number of dual and non-dual eligibles not enrolled in a managed care plan as of July of 2014 from the denominator file with the distribution of full- and partial-duals for each county from the CMS website. The relative factors from the October 28, 2015 file were used to determine the risk scores for each of these categories. Then, we used the predictive ratio estimates from CMS to develop the new risk scores. Finally, we adjusted the payment rates with the new risk scores and compared those rates with the old risk scores to determine the estimated change in the county benchmark.