Insight

Proposed Changes to Medicare Advantage Payment Policies Threaten Care for Seniors

Each February, the Centers for Medicare and Medicaid Services (CMS) releases its Advance Notice for the following year, proposing the rates and parameters for the Medicare Advantage (MA) and Part D Prescription Drug programs, similar to a proposed rule. This year’s notice, released on February 19, 2016, contained several provisions that could have a negative impact on the nearly 18 million seniors served by these plans. The final notice is due to be released on April 4.

The Growth Rate
The payment formula for MA plans is complicated—as explained in this AAF Primer—and while the growth rate is just one component of that formula, it is an important one. The Affordable Care Act (ACA) included provisions requiring an adjustment to the way the benchmarks (the basis of MA payments) are set and placing a cap on future benchmarks in order to constrain spending. While the transition period is now over, benchmarks under the new formula must not be more than what they would have been under the pre-ACA formula. Would-be pre-ACA benchmarks are calculated using the prior year’s benchmark increased by the estimated MA growth rate. Thus, insurance companies use the growth rate to approximate the maximum base amount that they will be paid for each enrollee (before any adjustment is made based on other factors).

While the growth rate is supposed to be an approximation of the rate of increase in costs per beneficiary from one year to the next, this rate has often changed (sometimes as dramatically as 4 percentage points or more, and sometimes from negative to positive and vice versa) from the publishing of the Advance Notice in February to the publishing of the Final Notice in April, just two months later. For insurance companies trying to manage their business and plan their expenses, it is difficult to make necessary business projections when a company’s expected revenues change so drastically in such a short period of time, without explanation. There is no other known metric or index used by the federal government to calculate payments that operates with such opaqueness. CMS owes it to those offering MA plans to provide some transparency and explanation regarding the methodology used to calculate the growth rate.

Further, because the cap on benchmarks applies to the bonus-adjusted benchmark, many plans are not receiving the full bonus for which they qualify, resulting in an estimated loss of tens of millions of dollars over the last few years.[1] Insurers and care providers have made great efforts and invested significant resources to improve the quality of care provided to their beneficiaries; not providing plans the full payment for which they qualify not only reduces revenues to companies who earned them, but more importantly, it reduces the level of benefits that can be provided to the seniors who enroll in such plans.

Risk Adjustment
Another component of the MA payment formula is the beneficiary’s risk score. Payments to plans are adjusted based on the average risk of the individuals that enroll in that plan; healthier individuals have a lower risk and thus payments on their behalf are lower, while less healthy patients have a higher risk of needing care and thus payments for these individuals are increased. In order to determine a beneficiary’s risk score, CMS uses both demographic factors as well as reported diagnoses from the previous year.

There is belief among some in the industry that the health risk of certain beneficiaries, particularly full dual-eligibles (individuals fully eligible for both Medicare and Medicaid, i.e. in the lowest income bracket), is not appropriately being measured—and thus not appropriately paid for. In response, CMS plans to make adjustments to the risk score model used in 2017. Primarily, CMS plans to replace the single community segment of the current model with six separate segments, representing each category of dual (full, partial, or non) and aged or disabled status, with a patient’s Medicaid status being determined on a monthly basis.  Few are likely to argue that the model is not flawed in its current form, but there is disagreement over the soundness and validity of these proposed changes. Whether or not an individual is dually-eligible for Medicare and Medicaid is based on that individual’s financial status. While studies have shown that there is some correlation between one’s socioeconomic status and their health, such a factor has not been proven to be either causal or the primary determinant of one’s health. As such, it may not be appropriate to use an individual’s financial status as the primary basis for determining one’s risk score.

Adjusting risk scores based on whether or not an individual suffers from multiple chronic conditions may be more sensible, and CMS is making adjustments to the disease interactions used in the model. CMS analyzed the interaction of diseases that occurred with a high frequency for each of the six subgroups described above and identified diseases that, when occurring together, significantly increase a patient’s health care costs. However, some are concerned that these adjustments are still not correctly reflecting the risk and costs for certain populations, particularly given that, as CMS admits, a majority of beneficiaries in the lowest risk groups do not have any of the conditions used to determine the risk score, but they do have other conditions not being accounted for. Further, one study found that partial-dual beneficiaries were more likely than full-duals or non-duals to have cancer, diabetes, depression, chronic obstructive pulmonary disorder (COPD), and ischemic heart disease. Given that partial-dual beneficiaries are still responsible for some cost-sharing, which for some will be unaffordable, these individuals may be more vulnerable to payment and benefit reductions than full-duals.

In addition to these various policy changes to the model, the underlying data supporting the model has also been updated. Simply changing the underlying data that the model relies on would alone yield different results; it is not clear that CMS sufficiently controlled for these changes to adequately assess the impact of the policy changes it has proposed to make in addition to the data update. Given such uncertainty, plans must be prepared for the possibility that additional changes will have to be made again next year; changes were also made in 2013 and between 2014 and 2016. Constantly changing the model creates real uncertainty and imposes instability in the market.

Ultimately, if a plan’s patient mix is not representative of the larger Medicare population, the revenue impact to that plan will likely be significantly different than what CMS has projected the average impact will be. Plans enrolling large numbers or percentages of partial-duals and non-dual beneficiaries will particularly be harmed, given that the focus of the recalibration was on full-dual beneficiaries. CMS should instead work to correct the underlying risk adjustment model to focus on the actual determinants of health rather than continuously making patch-work adjustments for certain subpopulations; such a piecemeal approach undermines the theoretical basis of a risk adjustment model.

Encounter Data
As mentioned, one factor used to determine a patient’s risk score is the diseases and conditions that patient was diagnosed with the previous year. CMS began requiring plans to collect and report encounter data in 2012, and began factoring such data into the calculation of patients’ risk score for the first time in 2015.[2] CMS will continue to use a blended risk score in 2017, relying on both the Risk Adjustment Processing System (RAPS) and the Encounter Data System (EDS), but will increase the weight given to encounter data from the current 10 percent to 50 percent. While encounter data is a valid source for determining a patient’s diagnoses—or at least the ones a patient is being treated for—and subsequently as a metric for adjusting risk scores, the system does not appear to be ready for this transition. The non-partisan Government Accountability Office (GAO) has found significant operational challenges with the EDS. Placing so much weight on data from a flawed system will likely yield inaccurate results and, in turn, improper payments.

CMS did not release the filtering logic to be used for payment year 2015 until December 22, 2015; in other words, plans were not told how or what data would be used to determine their payment amounts for the year until nine days before that year ended. Further, the data used to adjust payments for a year is from encounters occurring in the prior year (ie, 2014 encounter data is used for payments made in 2015); given that patients often change plans from one year to the next, this can be problematic.

Star Rating System
For seemingly the same reason that CMS has proposed adjusting the risk model, CMS is also proposing to adjust the methodology for determining the Star Ratings by which plans’ bonuses and rebates are determined. Again, CMS points to the impact of socioeconomic status on one’s health, and the impact on a plan’s Star Rating if a significant percentage of its enrollees are disabled and/or low-income subsidy (LIS) eligible. For example, being either disabled or low-income may inhibit one’s ability to adhere to a treatment plan, because either they forgot or were unable to make it to an appointment or take their medication, which factors into plans’ ratings. Plans’ ratings determine whether or not they qualify for a bonus and/or rebate and how big each will be. Accordingly, CMS has proposed applying an adjustment to plans’ Overall and/or Summary Star Ratings based on the percentage of dual/LIS and/or disabled beneficiaries enrolled in a given plan. The current “fix” is only temporary though as CMS works towards a more permanent solution. Again, continued changes to the model creates instability for plans and beneficiaries.

EGWP Bidding Waiver
CMS also proposed changing the way MA-Employer Group Waiver Plans (EGWPs) are paid. The proposal would eliminate the bidding process for these plans, instead paying them based on a weighted average of bids submitted by non-EGWP plans. The reasoning behind CMS’s decision is their belief that EGWP plans are overpaid based on the fact that their plan bids tend to be higher than those of non-EGWP plans while their enrollees’ risk scores tend to be significantly lower. However, it is important to understand the differences in incentives and plan design that result in differences in cost and benefits.

In the individual market (non-EGWPs), plans have an incentive to bid low in order to gain market share among enrollees. Since the cost of any bid in excess of the benchmark must be paid by the individual, plans with high bids are less likely to gain enrollees. Further, the lower a plan bid and the higher the quality rating, the more money a plan is paid by CMS (in the form of a bonus) to provide additional benefits, which can of course be instrumental in attracting more enrollees.

Employers offering EGWP plans, on the other hand, are incentivized to provide substantial benefits to their retirees in order to recruit the best employees. One aspect of such benefits pertains to the size and coverage in and out of the provider network. EGWPs typically provide broader (sometimes even national) coverage than non-EGWP plans since an employer’s retirees may live anywhere in the country. As shown in the chart below, most EGWP enrollees (64 percent) are in local preferred provider organizations (PPOs) which tend to have much broader provider networks and typically offer at least some coverage for care received from non-network providers. Conversely, 69 percent of enrollees of non-EGWP plans are in health maintenance organizations (HMOs), which tend to have more limited provider networks and tend to not offer coverage for services received from non-network providers.

Picture15

The cost and value of maintaining a national network and covering some of the cost for care received outside the network is certainly greater than that of a county-wide network where the cost of non-network care is borne fully by the patient. Under the proposal, EGWPs will be paid based on the bids submitted by non-EGWP plans. Paying PPO plans based on the bids of HMO plans does not reflect the differences in benefit design and value between these types of plans.

Conclusion
Enrollment in Medicare Advantage has continued to increase, covering more than 30 percent of Medicare beneficiaries in 2016, despite projections that enrollment would shrink, particularly given the payment reductions scheduled under the Affordable Care Act. The fact that enrollment has continued to increase despite these cuts is a testament to private insurers’ ability to provide high value care and satisfy their beneficiaries, unlike the traditional fee-for-service Medicare model. But these plans can only sustain so much; continued cuts and flawed policy changes will eventually result in benefit reductions, harming the individuals most at risk.


 

[1] Calculated by the Alliance of Community Health Plans using CMS data:

https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/MCRAdvPartDEnrolData/Monthly-Contract-and-Enrollment-Summary-Report.html

https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Ratebooks-and-Supporting-Data.html

https://www.cms.gov/Medicare/Prescription-Drug-Coverage/PrescriptionDrugCovGenIn/PerformanceData.html

[2] https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Downloads/Advance2015.pdf

Disclaimer