Up-coding in MA greater than in trad Mcare. http://t.co/HBtRA9UUPs
— Ari Friedman (@AriBFriedman) August 14, 2014
This is massive abuse of the system. I’ll go into details shortly after the CMS study’s results:
The Results: Each year the average MA risk score increased faster than the average FFS score. Using the risk adjustment model in place in 2004, the average MA score as a ratio of the average FFS score would have increased from 90% in 2004 to 109% in 2013. Using the model partially implemented in 2014, the ratio would have increased from 88% to 102%. The increase in relative MA scores appears to largely reflect changes in diagnostic coding, not real increases in the morbidity of MA enrollees. In survey-based data for 2006–2011, the MA-FFS ratio of risk scores remained roughly constant at 96%. Intensity of coding varies widely by contract, with some contracts coding very similarly to FFS and others coding much more intensely than the MA average. Underpinning this relative growth in scores is particularly rapid relative growth in a subset of HCCs.
Now what does that mean?
One of the great problems with Medicare Advantage (besides that we as a society pay too much incremental costs for minimal extra benefits to the elderly … most of the extra money goes to marketing, cherry picking, hookers and blow) is that the basic structure of the privatized market encourages insurance companies to do their best to cherry pick only the healthier segments of the Medicare population. This is because they get paid a risk adjusted flat fee based on average Medicare costs per enrollee plus a kicker.
The MA payment system uses diagnostic information to assign a risk score to each beneficiary, where the average beneficiary in fee-for-service has a risk score of 1.0. MA plans are paid the product of their bid multiplied by the enrollee’s risk score—that is, if an MA plan bids $1,000/month for an enrollee with a risk score of 1.0, and then enrolls a beneficiary with a risk score of 1.2, the plan gets paid $1,200/month for that enrollee (1.2 * $1,000/month). This payment system creates incentives for MA plans to find and report as many diagnoses as can be supported by the medical record…
payment to MA plans is calibrated based on coding patterns in FFS. If, for example, Jane Doe would have a risk score of 1.0 if she were in FFS, the implicit assumption in the design of the MA payment system is that she would have the same risk score of 1.0 if she were enrolled in MA. Alternatively, if the same Jane Doe had a risk score of 1.1 if enrolled in MA, then her plan would be overpaid.
One of the policies in place that is designed to blunt that incentive is back end risk adjustment reconciliation. At the end of a claims year, the Center for Medicare and Medicaid Services (CMS) looks at all the claims submitted by Medicare Advantage insurers and attempts to identify which plans have really healthy populations based on claims data, and which plans have really sick populations based on claims data. This information is used to create “risk adjustment” reweights of what the base fee plus a kicker is for each plan. Healthier plans see a smaller kicker, sicker plans see more money.
On one set of scales, the Medicare Advantage population as a whole has been slightly healthier than the traditional Medicare population, and that has been consistent on a population basis for a while. On another scale, the Medicare Advantage population went from being significantly healthier than the Medicare Fee for Service (FFS) population in 2004 to way sicker than the FFS population in 2013. Being seen as “way sicker” means the Medicare Advantage kicker is bigger.
So the two things disagree. What is going on here? The study argues that the increased “sickness” of the Medicare Advantage population is an artifact of how claims are submitted and diagnoises are reported. Medicare Advantage billers have every incentive in the world to throw any medically supportable or vaguely supported diagnosis on a claim plus go through records. Fee for Service is incentivized to provide a minimally relevant claim. There are teams at almost all insurers whose job it is to look at high cost individuals and make sure their doctors are noting every diagnosis from acne to congestive heart failure on a claim for a strep throat test. Fee for Service would normally bill a procedure code of a strep throat test, a sick office visit and a diagnosis of strep throat while Medicare Advantage would have all of that plus acne, congestive heart failure, diabetes and restless leg syndrome as ongoing diagnosises. The last four items are of minimal clinical value but allow the Medicare Advantage plan to argue that this individual is far sicker than that person would be categorized in the Fee for Service world.
CMS has been trying to deal with this problem by rejiggering their risk adjustment model every couple of years as well as having Congressionally mandated chops off the top but insurance companies are engaged in a continuous arms race to beat the risk adjustment system, or at least do no worse than any other Medicare Advantage insurer from beating the sytsem. This is one of the weaknesses of long term risk adjustment and risk weighing schemas and it is one that will show up in the Exchanges. The insurers have every incentive in the world to create a paper trail of very sick individuals that the claims data does not support nor does the true clinical assessment data support. They need to do this to compete with other insurers who are looking to maximize their surface risk factor at the very least. Insurance companies have devoted significant resources to getting very good at generating big gaps between the reported risk of their population and the actual risk of their population. This is an area where further ‘expertise’ will develop under current incentives. And it will be profitable for insurers while continuing to make American claims processing and office billing procedures an expensive mess.