CMS will likely overpay Medicare Advantage plans by hundreds of billions of dollars over the next 10 years if the agency doesn’t account for "coding intensity" in its risk scores.

As University of California, San Diego family medicine professor Richard Kronick wrote in a Health Affairs study, the risk score for MA enrollees has risen by around 1.5 percent per year over the past decade. He credited this rise not to a change in enrollees’ health, but rather to MA plans making a concerted effort to identify additional diagnoses for beneficiaries—making them appear less healthy than the average Medicare fee-for-service (FFS) beneficiary and thus increasing their risk score.

This has been an issue since 2000, when CMS began phasing in use of diagnostic as well as demographic information as risk score adjusters. The idea was the old system penalized plans which covered potentially higher-cost enrollees. After the switch, however, Congress became concerned this would lead to overpayment of MA plans after giving them strong incentives to report diagnostic information more completely than in FFS Medicare. With each plan paid its bid multiplied by the risk score,

According to Kronick, lawmakers were right to be concerned. He determined there was a 8 percent difference between MA coding and the actual risk. MA enrollees had 97.1 percent of the average FFS Medicare beneficiary’s risk score using an adjusted average per capita cost method, but when using diagnostic risk adjustment, the average MA risk score swung the other way, to 104.9 percent of FFS. After the 2014 minimum coding intensity adjustment of 4.91 percent required by Congress is subtracted, the net effect of Medicare Advantage coding intensity in 2014 on risk scores is approximately 3 percent.

A coding intensity adjustment was implemented in 2010, which reduced risk scores by a minimum of 3.41 percent in 2010, ramping up to a 5.91 percent reduction in 2015—still not enough to cover the 8 percent gap Kronick identified.  

“These responses have not yet established a payment system that will protect taxpayers and beneficiaries from the effects of differential coding over the coming decade, with a likely liability of $200 to $250 billion,” he wrote. “Both political and technical problems have impeded the development of a robust solution.”

The effects could worsen if MA enrollment grows beyond what Kronick considered to be conservative estimates.

Part of the solution, he argued, can be found, is to revive a 2004 proposal. This would’ve required the CMS adjustment to be budget-neutral and adjust for coding intensity “using the principle that MA beneficiaries are no healthier and no sicker than demographically similar fee-for-service Medicare beneficiaries.”

Kronick said this may still be “slightly too generous” to MA plans, as their enrollees be somewhat healthier than comparable FFS beneficiaries. With the change, CMS may have to look out for MA plans trying to avoid covering higher-risk patients.

“If CMS adopts the method of calculating the coding intensity adjustment proposed here, it would be important for the agency to conduct annual analyses of mortality data, Medicare Current Beneficiary Survey data, and other data such as prescription drug data to monitor selection into Medicare Advantage. If these analyses were to find adverse selection against MA plans, then the coding intensity adjustment would need to be modified to account for this,” Kronick wrote.