The Medicare Payment Advisory Commission (MedPAC) is supposed to give Congress good advice about Medicare. MedPAC is good at telling Congress that the managed care experiments Congress keeps foisting on Medicare are not saving money. But MedPAC is terrible at telling Congress why these experiments are failing and whether they can be salvaged. The Medicare Advantage program is the longest-running example of MedPAC’s chronic inability to explain to Congress why a managed care program isn’t working and what should be done about it. The ACO and MACRA programs are two other examples.
MedPAC’s meandering discussion about converting Medicare into a voucher program, which began three years ago, illustrates the problem. I’ll focus on that discussion in this post.
MedPAC’s willingness to identify and discuss shortcomings of managed care programs creates the impression that it is offering Congress good advice on what to do about them. But that impression is an illusion. What MedPAC does after identifying a problem may accurately be called dithering – refusing to state clearly that a problem MedPAC has identified cannot be fixed and, therefore, the managed care program in question will never function well or at all. By dithering, MedPAC avoids the discomfort of being the bearer of bad news to Congress and the managed care movement.
MedPAC concedes risk adjustment is essential … and impossible
To illustrate how MedPAC dithers on fundamental issues, I call your attention to chapters 1 and 2 of the commission’s June 2014 report to Congress. Chapter 1, entitled “Synchronizing Medicare policy across payment models,” presents what MedPAC called its “initial exploration” of a proposal to convert Medicare into a “premium support” or voucher program. (p. 4) The “payment models” MedPAC seeks to “synchronize” with a voucher program are Medicare’s traditional fee-for-service (FFS) program, the Medicare Advantage (MA) program, and the new ACO program.
MedPAC began discussing “synchronization” via vouchers in the fall of 2013. The underlying idea is that giving all Medicare recipients vouchers and instructing them to use those vouchers to shop among the three Medicare programs will somehow “synchronize” – make more fair and efficient – both the “quality measures” applied to the three programs and the payments made to the insurers in the MA program and to the clinics and hospitals in the FFS and ACO programs. Once this “synchronization” occurs, the “value” of the insurance offered by the three “payment models” will somehow be exposed for all to see, this in turn will allow Medicare recipients to bestow their vouchers on the program with the highest value, and, presto, competition will then sort out which of the three programs is the most efficient. No need for MedPAC commissioners and staff to tie their brains in knots trying to determine from high above the earth which of the programs is more efficient. No need, in other words, for MedPAC to tell Congress the MA and ACO programs are more expensive than the FFS program, are causing other side effects such as consolidation and physician burnout, and should be terminated. MedPAC’s staff and commissioners can sit back in their hovercraft at 80,000 feet and watch the Earthlings, goosed by the “invisible hand” of competition, choose winners and losers.
MedPAC offers its initial thoughts on this wondrous vision in Chapter 1 of its June 2014 report. In this chapter MedPAC states that the vouchers (the subsidies Medicare recipients will receive to help them pay their premiums) must be accurately risk adjusted for any voucher program to work. The justification for this statement is obvious: (1) If the vouchers assigned to sicker Medicare recipients are not adjusted upward to reflect the higher cost of insuring sicker people, insurers and providers will have an incentive to avoid sicker people and to ration care to the sicker people they cannot avoid; and (2) the “competition” unleashed by the voucher program will not reveal which of the FFS, MA, and ACO programs are more efficient, but will reveal instead how well insurers and providers avoid and screw the sick.
But in the very next chapter, Chapter 2 (which is adorned with the happy title, “Improving risk adjustment in the Medicare program”), MedPAC admits CMS’s current risk adjustment method – a method developed over the last 15 years to adjust payments to MA plans – is terribly inaccurate and, to make matters worse, there is no hope of improving it. (So why was the chapter entitled, “Improving risk adjustment….”?)
I’ll discuss Chapter 2’s dismal findings in more detail in a moment. But let me stop here and ask an obvious question: Shouldn’t MedPAC have closed Chapter 2 by warning the reader that the voucher program discussed in Chapter 1 must be shelved because accurate risk adjustment is not possible? Or better yet, shouldn’t MedPAC have held off even discussing a voucher program until that fine day when CMS has developed an accurate risk adjuster? MedPAC did neither. They outlined a voucher program, noted that accurate risk adjustment is essential to such a program, then stated that accurate risk adjustment is not possible … and then just walked away.
How bad is CMS’s risk adjuster?
Now let’s take a closer look at Chapter 2 in MedPAC’s June 2014 report. It illustrates the two behaviors I’m talking about – identifying problems honestly and accurately, and dithering about those problems. Early in that chapter MedPAC repeats the warning it offered in Chapter 1 – that accurate risk adjustment is “vital” if competition via a voucher system is going to reward the most efficient of Medicare’s three programs.  MedPAC then tells us CMS’s adjuster, known as a “hierarchical condition category” (HCC) adjuster, has an R-squared value of only .12, which means the HCC method explains only 12 percent of the variation in spending among Medicare recipients. 
To drive home how bad an R2 of .12 is, MedPAC presents Table 2-1. This table shows how much CMS’s HCC adjuster overestimates the cost of healthy people and underestimates the cost of sick people. The table shows, for example, that the adjuster pays MA insurers 62 percent too much for the average person in the healthiest quintile and 30 percent too much for the second healthiest quintile. It’s not till you get into the fourth quintile – people between the 60th and 80th percentile – that the adjuster pays too little (5 percent too little). For those between the 95th and 99th percentile, it underpays by 18 percent, and for the sickest one percent it underpays by 29 percent.
MedPAC spends the rest of Chapter 2 examining three ways to improve CMS’s very crude adjuster and concludes “the alternative approaches we evaluated either do not improve the performance of the CMS–HCC model or could create other problems, including less incentive for plans to manage care and hold down costs, penalizing plans that do so, and increasing incentives to upcode.” (p. 33)
So let’s stop here and give MedPAC credit for clearly describing a serious and intractable problem – it’s not possible to risk-adjust CMS’s lump sum payments to MA plans or anyone else accurately. But then let’s look at what doesn’t happen. MedPAC refuses to retract the happy voucher vision it presented in Chapter 1. The end of Chapter 2 would have been an obvious place for MedPAC to do that – to state that the voucher proposal discussed in Chapter 1 must be shelved because accurate risk adjustment is neither technologically nor financially feasible. A simple “never mind Chapter 1” would have sufficed. But MedPAC doesn’t do that. MedPAC just walks away from their own absurd logic: “Premium support” is a great way to “synchronize” payment to Medicare’s three programs; accurate risk adjustment will be necessary for “premium support”; but accurate risk adjustment is not possible. 
What word should we use to describe such irresponsible behavior, such illogical thinking? I have chosen “dithering.” There may be a better label for such baffling behavior, but right now I can’t think of it. I invite readers to submit suggestions.
Dithering over “Quality Measurement”
Risk adjustment of payment is not the only issue MedPAC must resolve to make vouchers work. Other issues include:
- Accurate risk adjustment of the “quality measures” that MedPAC intends to inflict on the three Medicare programs,
- accurate “attribution” of Medicare recipients to ACOs and to individual doctors in the FFS program,
- rational definition of the geographical areas that will constitute the “markets” within which the FFS, ACO and MA programs will compete for vouchers, and
- making all this complexity mesh with the insanely complex rules MACRA imposes on the FFS and ACO programs.
As was the case with the risk-adjustment-of-payment issue, the commission identifies these other issues as problems, then just floats away.
A conversation commissioners had at their October 7, 2016 meeting illustrates this pattern. The agenda item for this meeting was “Quality measurement and premium support.” The question for the day was how to subject the FFS, MA and ACO programs to uniform measurement of quality. The staff presented five “quality measures” for discussion, including emergency department visits, readmission rates, and a novel and grandiose “measure” called “healthy days at home,” which the staff said represented “the number of days within a year that a local area’s beneficiaries are alive and did not have interactions with the health care system that imply less than optimal health.” (p. 6 of the transcript of the morning session.)
Midway through this session commissioner David Nerenz pleaded with the other commissioners to start demanding accuracy in risk adjustment. I quote a portion of his remarks to give you an idea of how cavalierly the commission has treated this issue to date and the level of Nerenz’s concern.
[W]e don’t think sharply enough about exactly what it is that we think we’re measuring…. (p. 39)
[T]he question that I want us to be looking at in all these measures [is], What is the signal-to-noise ratio? …. We almost never ask that about quality measures that we use. But if you just think about it in a signal-to-noise framework, when we measure the outcome, only maybe 5 percent of the variance in the outcome [is] influenced by the signal that we’re after. It really can be that weak. And all this other stuff is floating around in there…. [W]e have to take the issue of risk adjustment so seriously when we’re focusing on outcomes….
And even our language – you know, we talk about these things as measures of quality. Well, that … makes sense if … something like 70 percent or so of the variance in the outcome is explained…. Then I can say, okay, that’s a measure of quality. But as soon as that R-squared starts to drop down, then I think the word “measure” begins not to be quite right. But we keep using it, and we get trapped in our language…. (pp. 41-42)
Five or six other commissioners agreed with Nerenz’s criticism. Nevertheless, I predict that Nerenz’s tutorial will turn out to have been a waste of breath. The brutal fact is the staff cannot improve the accuracy of quality risk adjustment at the “payment model” level any more than they can improve cost risk adjustment. But I predict the staff, with Chairman Crosson’s blessing, will keep asking the commission to endorse “healthy days at home” and other trendy “measures” of quality; Nerenz and one or two other commissioners will object once or twice more, then they’ll go quiet; and sometime early next year the commission will vote for a voucher proposal to Congress that assumes accurate adjustment of quality “measures” is possible when everyone knows it isn’t. The commission will keep their voucher proposal vague and they’ll avoid explicitly endorsing it, but it will be clear they think a voucher program could work.
Dithering at 80,000 feet: A cure for cognitive dissonance
I’ll close with one more example of dithering over an issue that must be solved if a voucher program is to work. In response to a question posed by Commissioner Alice Coombs at the October 7 meeting about how MACRA’s Merit-based Incentive Payment System (MIPS) would work within a voucher program, staff director Mark Miller said: “You know, a lot of what we’re headed towards here is a chapter in June [meaning a chapter in the June 2017 report to Congress] that kind of goes through high-level design issues, and so exactly all the MIPS interactions and everything. [The preceding dead end is in the original]. Just to be very direct, for the purposes of this paper, we haven’t thought that through yet….[T]his is just a conceptual conversation.” (p. 22) 
Notice the words “high-level design” and “conceptual.” Any bets on whether MedPAC will ever bring their hovercraft down to Earth and actually “think through” how MIPS is supposed to work in a voucher system?
Congress and the public do not need an advisory commission that merely goes through the motions of advising. We need a MedPAC that is willing to state clearly that Medicare’s managed care programs are not working and should be replaced by an entirely different approach, an approach that focuses on specific evidence-based services, not faith-based, aspirationally defined organizations such as HMOs, ACOs and “medical homes.”
 Here is how MedPAC articulated, in Chapter 2 of the June 2014 report, why accurate risk adjustment of expenditures is essential to a voucher program: “A final issue to consider is how payment inaccuracies related to level of health care costs affect equity among MA plans, FFS Medicare, and accountable care organizations (ACOs). If payment equity among these three sectors is a goal, risk adjustment that results in more accurate payments for high-cost and low-cost beneficiaries is vital. For example, if the MA sector can attract low-cost beneficiaries and avoid high-cost beneficiaries, the risk-adjusted payments in the MA sector would exceed what their enrollees would cost in ACOs or FFS Medicare. The result would be program spending that is higher than if all beneficiaries were in FFS Medicare.” (p. 23)
Three pages later MedPAC repeats this warning: “If we desire financial neutrality among FFS Medicare, MA plans, and … ACOs, over-prediction for low-cost beneficiaries and under-prediction for high-cost beneficiaries could present a problem. If MA plans have high shares of low-cost beneficiaries, payments in the MA sector that are risk adjusted with the existing CMS–HCC model would exceed what Medicare would pay for their enrollees in ACOs or FFS Medicare.” (p. 26)
 CMS adopted the HCC risk adjuster in 2004. Prior to that, its risk adjuster predicted a measly 1 percent of the variation in expenditures (which means the taxpayer has vastly overpaid MA insurers for a long time). The HCC method implemented in 2004 had an R2 of .11 (see Pope et al. here https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4194896/). According to MedPAC’s June 2014 report, the version of the HCC method CMS began to use in 2014 has an R2 of .12. (p. 30)
 Here is how Chapter 2 of the June 2014 report actually ends: “Therefore, we may need to consider administrative measures to address the imprecision of the CMS–HCC model and incentives for plans to engage in selection. One possibility is penalties for disenrollment of high-cost beneficiaries. Also, CMS may be able to obtain helpful information about factors that contribute to disenrollment through surveys of disenrollees and evaluating disenrollees for changes in their risk factors over time.” (p. 33) In other words, CMS is saying the only idea they can come up with that might reduce the favorable selection into MA plans encouraged by CMS’s crude risk adjuster would be to punish MA insurers for driving sicker people back to Medicare’s FFS (and therefore ACO) programs.
For two reasons, this is a pathetic response. First, MedPAC offers not a shred of evidence for this suggestion and does not elaborate on it. Is it possible to detect and punish “excess” disenrollment of sicker people? What would it cost to do this well? If it’s possible, would it have any effect on the effective R2? We get no answers. The chapter just peters out here. Second, even if CMS could devise a way to punish MA plans for “excess” disenrollment of sicker people, that would do nothing to alter favorable selection into MA plans in the first place. Nor would it reduce the incentive MA plans have to upcode, which is another problem MedPAC recognizes and has become very good at dithering over.
 An article published in the November 7, 2016 CQ Healthbeat http://www.commonwealthfund.org/publications/newsletters/washington-health-policy-in-review/2016/nov/november-7-2016/medpac-examines-how-medicare-premium-support-might-work gives you some idea of how MedPAC will articulate its support for “synchronization” via vouchers. They will claim they are neutral, but they’ll make statements indicating the issues MedPAC raises are solvable and a voucher program can work.