Skip to main content
off the charts
POLICY INSIGHT
BEYOND THE NUMBERS

Reducing Medicare Advantage Overpayments

Ahead of Friday’s announcement of Medicare’s preliminary payment rates and policies for Medicare Advantage insurers in 2016, insurers have launched an advertising campaign claiming that potential payment changes would undermine the program.  For numerous reasons, it’s hard to take these doom-and-gloom predictions seriously.

For starters, Medicare Advantage continues to thrive — enrollment has reached an all-time high and is expected to keep growing in 2016, according to the Congressional Budget Office (CBO) and Office of Management and Budget (OMB) — despite health reform’s ongoing, much-needed effort to curb overpayments to insurers.

In addition, Medicare Advantage payments still average about 105 percent of the cost of covering comparable beneficiaries in traditional Medicare, according to preliminary estimates by analysts from the Medicare Payment Advisory Commission (MedPAC), Congress’ official advisory body on Medicare payment policies.

MedPAC estimates that roughly 60 percent of that differential reflects the phenomenon known as “upcoding” — the first time MedPAC analysts have quantified how much upcoding inflates Medicare Advantage payments.  Medicare Advantage includes a risk adjustment system that raises or lowers payments to plans based on their enrollees’ relative health, measured by a “risk score” based on patient diagnoses; upcoding occurs when the risk scores that plans submit rise over time — making enrollees appear increasingly unhealthy — without actual changes in enrollees’ health.  This results in higher-than-warranted payments to Medicare Advantage plans.

Upcoding is a long-standing problem in Medicare Advantage, as CBO and the Government Accountability Office (GAO) have documented.  According to MedPAC, risk scores were 8 percent higher in Medicare Advantage, on average, than in traditional Medicare for comparable beneficiaries.  And MedPAC analysts noted that the amount of upcoding seems to be getting larger.

Policymakers could better address upcoding by raising Medicare’s annually required “coding intensity” adjustment.  Health reform requires the Centers for Medicare and Medicaid Services (CMS) to adjust Medicare Advantage’s risk adjustment system by at least a minimum amount each year to compensate for upcoding.  The President’s 2016 budget would raise that minimum annual adjustment very modestly starting in 2017, saving $36.2 billion over ten years, OMB estimates.

Moreover, while CMS has only applied the minimum required adjustment in recent years, it has the discretion to institute a larger adjustment than required, for example as part of the 2016 preliminary rate announcement.  MedPAC estimates that this year’s adjustment would have to have been more than 50 percent larger to fully offset the effects of upcoding.  (GAO similarly found that the annual adjustment likely needs to be substantially larger than the minimum level.)

Policymakers could also reduce upcoding by excluding in-home health assessments from Medicare Advantage risk score calculations unless the assessment diagnoses are later confirmed in treatment settings.  Medicare Advantage plans increasingly provide in-home health assessments of their enrollees; for example, a nurse may come to a patient’s home to do a physical exam.  CMS has found that insurers primarily use these assessments to “collect” diagnoses in order to increase enrollees’ risk scores for purposes of risk adjustment, rather than to improve follow-up care or identify illnesses requiring treatment.

CMS proposed last year to exclude any diagnoses identified during an in-home assessment that subsequent clinical encounters fail to confirm.  CMS, however, dropped the proposal in the face of industry opposition, opting to collect additional data about the impact of these assessments and revisit the issue later.  CMS could include this prior proposal in its 2016 preliminary rate announcement in order to limit the use of these assessments to promote upcoding.