(This article has been updated with a correction.)
The government is paying an extra $2 billion a year to Medicare Advantage plans because of upcoding, according to a new working paper posted in the National Bureau of Economic Research.
The paper, which has not yet been peer reviewed, comes amid ongoing calls from members of Congress for federal officials to crack down on Medicare Advantage fraud, which has also spurred at least a few whistle-blower lawsuits in recent years.
The CMS pays private Medicare Advantage plans based partly on risk-scores, which reflect enrollee's health, as opposed to traditional Medicare which operates on a fee-for-service basis. Critics, however, have said that some are abusing the Medicare Advantage system, inflating risk scores, making patients appear sicker, to make more money.
The new paper, by University of Texas Professor Michael Geruso and Harvard Medical School research fellow Timothy Layton, estimates that enrollees in private Medicare Advantage plans have 6% to 16% higher diagnosis-based risk scores than the same enrollees would have under fee-for-service Medicare.
That difference not only costs taxpayers money; it also can steer patients toward choosing Medicare Advantage over traditional Medicare because it may allow those Advantage plans to offer better benefits or even rebates to enrollees, according to the paper's authors. The authors also found that coding seems to be more intense in integrated health plans.
"The integrated plans have more tools available to them to get the providers to code the way the insurers are incentivized to code,” Layton said in an interview.
Attempts to reach America's Health Insurance Plans, which represents the health insurance industry, for comment weren't immediately successful Monday afternoon.
But the paper's authors say their findings are important on a number of fronts.
“The manipulability of the risk adjustment system via diagnosis coding is an issue of significant practical importance, given the large and growing role of risk adjustment in regulated insurance markets for Medicare, Medicaid, and Exchange plans,” according to the paper.
The findings, for example, are relevant for risk adjustment within the state and federal insurance exchanges under the Affordable Care Act, according to the paper. Payments to the exchange plans are risk adjusted using a model similar to the one used in Medicare Advantage. Payments to insurers that claim sicker enrollees are funded by transfers from insurers that claim healthier enrollees, meaning plans that have more intense coding practices might get money from those with less intense coding. That transfer of money, in turn, will make consumers more often choose integrated plans, according to the paper.
The paper also notes that actions CMS has taken so far to address the issue, such as starting to decrease Medicare Advantage risk scores in 2010, haven't been enough.
“Our results suggest this adjustment is both too small and fails to account for large coding differences across insurance contract types,” according to the paper.
The paper, however, does not necessarily attribute all the differences in coding to intentional manipulation.
“There's a notion that differences in coding come up because insurers are gaming the payment system,” Geruso said. “While I think there's some of that going on, there are also probably some differences in models of managed care that result in different diagnosis codes being assigned.”
The paper, however, concludes that despite its problems, some type of risk adjustment is probably necessary to make sure insurers don't shy away from sick patients. The paper's authors suggest that risk scoring could be improved by making less easily manipulated factors such as age and gender more important in calculating risk scores than more easily-manipulated factors such certain diagnoses.
They also suggest lengthening the look-back period, or the length of time over which patients' past diagnoses are taken into account across both traditional Medicare and Medicare Advantage. That might help increase risk scores within traditional Medicare which would lead to a decrease in risk scores in Medicare Advantage because the calculation for Medicare Advantage payments takes those traditional Medicare risk scores into account.
Government agencies and organizations have raised similar concerns about risk-scoring over the years. According to a 2013 Government Accountability Office report, cumulative Medicare Advantage risk scores were 4.2% higher in 2010 than they likely would have been if the same beneficiaries had been enrolled in traditional Medicare.
HHS' Office of Inspector General has also said in its 2015 work plan that it plans to review medical record documentation “to ensure that it supports the diagnoses organizations submitted to CMS for use in CMS' risk-score calculations and determine whether the diagnoses submitted complied with federal requirements.”
(Note: This story has been updated to reflect a correction the paper's authors made to the paper Tuesday, a day after Modern Healthcare first published its article about the paper online. The paper's authors learned Tuesday that the CMS increased the amount it deflated risk scores in 2014, leading them to revise their original estimate that the government pays an additional $5 billion to Medicare Advantage each year because of upcoding. The authors revised that figure Tuesday down to about $2 billion.)