Over the next several weeks, we’ll be posting highlights of new data, analysis and recommendations from MedPAC’s March 2015 Report to Congress, which was released on March 13.
MedPAC’s March 2015 report contains a new analysis of differences in risk scores between beneficiaries in Medicare Advantage (MA) plans and fee-for-service Medicare (FFS).
Medicare calculates its payment to plans separately for each beneficiary, multiplying the plan’s payment rate by the beneficiary’s risk score. The risk models are based on diagnoses that FFS providers coded during the year before the payment year. MA plans report diagnoses to Medicare for their enrollees. Because risk scores influence payment, MA plans have a strong incentive to maximize coding. In contrast, FFS providers don’t have as strong an incentive to maximize coding. For a given beneficiary, all else equal, this would result in Medicare paying more when the beneficiary enrolls in MA than if they had remained in FFS.
Recent research has found that risk scores for MA plan members have been growing more rapidly than risk scores for FFS beneficiaries (Kronick and Welch 2014). Since 2010, as mandated by law, CMS has been making an across-the-board reduction in MA risk scores to account for these differences, which do not appear justified by enrollees’ actual health risk. In 2012, the Government Accountability Office found that CMS should make larger reductions to fully account for coding differences. For 2015, CMS has chosen to reduce risk scores by 5.16 percent, the minimum reduction under current law.
This year, MedPAC examined coding differences between beneficiaries in FFS Medicare and those enrolled in MA plans. We found that beneficiaries in MA had more growth in risk scores than beneficiaries who had remained in FFS. And those differences grew the longer enrollees stayed in MA.
We began our analysis by comparing two groups of beneficiaries who joined Medicare in the same year: one group who enrolled in FFS for their first year in Medicare and then switched to MA for their second and subsequent years, and one group who enrolled in FFS for their first year in Medicare, and remained in FFS for their second and subsequent years. We first looked at a cohort of beneficiaries who joined FFS Medicare in 2006 and switched to MA in 2007, and found that in their first year spent in MA, beneficiaries’ risk scores grew 6 percent faster than the comparison population that remained in FFS. This first-year MA effect was even larger for more recent enrollees: among beneficiaries who joined Medicare in 2010 and switched to MA in 2011, their risk scores grew 7.5 percent faster in 2011 than the comparison FFS population. After the first year, average MA risk scores for these cohorts grew about 2 percent faster annually than FFS risk scores.
Next, we examined the cumulative effect of these differences in growth rates over multiple years. We built cohorts of 2013 MA and FFS enrollees based on how long they had been continuously enrolled in MA and FFS, respectively, and compared them. Our analysis showed that the longer cohorts remained in MA, the greater difference they experienced in risk score growth relative to FFS. Depending on the time period, the difference in risk score growth ranged from 4 percent to 13 percent. When these growth rates are weighted for length of enrollment, on average, the MA enrollees’ risk scores grew about 8 percent faster than scores in the FFS population.
Those differences in coding suggest that if CMS raised the coding adjustment by about 3 percentage points, the aggregate level of coding in the FFS and MA sectors would be roughly equal. In other words, due to coding differences that are not adjusted for, the Medicare program spends 3% more on average for a beneficiary who enrolls in MA compared to a similar beneficiary who stays in FFS.
Given the presence of uncorrected coding differences in MA, the Commission finds that payments to MA plans are 105 percent of FFS for 2015. The 105 percent includes the 102 percent that we estimated using the methodology we have traditionally used, which assumes that CMS’s risk and coding adjustments properly adjust for differences in the MA and FFS populations, plus the additional 3 percent that we found should be added to the coding adjustment.
To read more about this analysis, check out The Medicare Advantage Program: Status report in MedPAC’s March 2015 Report to Congress.