How CMS Undermines ACOs and What to do About It

In my first post  in this three-part series, I documented three problems with Pioneer ACOs: High churn rates among patients and doctors; assignment to ACOs of healthy patients; and assignment of so few ACO patients to each ACO doctor that ACO “attributees” constitute just 5 percent of each doctor’s panel. I noted that these problems could explain why Medicare ACOs have been so ineffective.

These problems are the direct result of CMS’s strange method of assigning patients to ACOs. Patients do not decide to enroll in ACOs. CMS assigns patients to ACOs based on a two-step process: (1) CMS first determines whether a doctor has a contract with an ACO; (2) CMS then determines which patients “belong” to that doctor, and assigns all patients “belonging” to that doctor to that doctor’s ACO. This method is invisible to patients; they don’t know they have been assigned to an ACO unless an ACO doctor tells them, which happens rarely, and when it does patients have no idea what the doctor is talking about. [1]

This raises an obvious question: If CMS’s method of assigning patients to ACOs is a significant reason why ACOs are not succeeding, why do it? There is no easy way to explain CMS’s answer to this question because it isn’t rational. The best way to explain why CMS adopted the two-step attribution method is to explain the method’s history.

I will do that in this essay. We will see that Congress, on the basis of folklore, decided that doctors in the traditional fee-for-service (FFS) Medicare program were ordering too many services and needed to be herded into “physician groups” that would resemble HMOs (see my comment here  on the obsession with overuse). But Congress also decided they didn’t want to force Medicare beneficiaries to enroll with the quasi-HMOs. That was a critical decision because it required that Congress figure out some method other than enrollment to determine which patients “belonged” to which physician groups. Congress, in its Infinite Wisdom, decided they would let someone else figure that out. They assigned that task to CMS.

The assignment was impossible. CMS should have told Congress they were nuts but, understandably, CMS eschewed that option. So CMS did the best they could. Based on some arbitrary assumptions, CMS devised the two-step assignment method that is now causing so much trouble for ACOs.

The first mistake: Demonizing fee-for-service

The label “accountable care organization” was invented by Elliott Fisher and members of the Medicare Payment Advisory Commission at MedPAC’s November 9, 2006 meeting. [2] At that meeting Fisher presented to MedPAC a version of the two-step process for assigning Medicare beneficiaries that CMS was already using for the Physician Group Practice (PGP) demonstration (which began in 2005) and that CMS would go on to use for its first two ACO programs – the Pioneer ACO program and the Medicare Shared Savings Program (MSSP) (both inaugurated in 2012). That fact, and Fisher’s aggressive promotion of ACOs after that meeting, earned Fisher the title of “father of the ACO.”

But the ACO concept was being discussed by CMS by the early 1990s (at that time CMS was known as the Health Care Financing Administration), and the two-step method of assigning patients to groups of doctors was being discussed within CMS by the early 2000s. The impetus for these discussions was a series of laws enacted by Congress in1989, 2000, 2005, and 2010, all aimed at reducing inflation in the cost of Medicare’s traditional FFS program.

In the Omnibus Budget Reconciliation Act (OBRA) of 1989, Congress authorized a “volume performance standard” (VPS) for Part B, the first version of what would soon become the Sustainable Growth Rate formula (which would in turn be replaced by MACRA in 2015). The VPS was a limit on total Part B spending. Because Congress had some doubts about how well the VPS would work, and perhaps more importantly, because Congress had bought the conventional wisdom that FFS causes overuse and overuse was causing health care inflation, Congress included a provision within OBRA authorizing CMS/HCFA to develop an alternative to Part B’s FFS method. This alternative was supposed to employ managed care tactics, including shifting insurance risk to doctors.

This was the first mistake Congress would make on its way to endorsing ACOs and ultimately MACRA. By demonizing FFS and lionizing managed care, Congress got it totally backwards. Congress should have investigated how to make both the insurance industry and the privatized portion of Medicare (what we now call Medicare Advantage) look more like traditional Medicare, not the other way around. By the early 1990s Congress had been warned numerous times that there was little evidence for the claims being made on behalf of HMOs and the HMO wannabees that were rapidly taking over the insurance industry, and much evidence indicating that the portion of Medicare run by HMOs (today’s Medicare Advantage) was costing much more per insured beneficiary than the traditional FFS program. [3]

But Congress, under the spell of the managed care movement, didn’t grasp that it had it backwards. So rather than instruct CMS to look for ways to induce private-sector insurers to act more like traditional FFS Medicare, Congress endorsed the opposite policy. Provisions in OBRA instructed CMS to start looking for ways to make the traditional Medicare program look more like the managed care insurance companies that were taking over the private sector.

The sound of one hand clapping

Predictably enough, the VPS system authorized by OBRA didn’t work and Congress replaced it with the doomed Sustainable Growth Rate (SGR) formula in 1997. The SGR soon proved it wasn’t going to work either.

The problem with both the VPS and SGR (aside from the fact they were designed to address overuse, a problem that was minor compared to underuse and excessive prices and administrative costs) was that they both applied expenditure growth limits to the entire pool of 700,000 American doctors who treated Medicare patients. That pool was too large; there was no way individual doctors could perceive that it was in their self-interest to reduce their own contribution to the alleged overuse problem by cutting back services to their own patients.

By the late 1990s, Congress and the managed care movement were even more obsessed with overuse and, given the failure of the VPS and SGR mechanisms, even more determined to find a way to break the ocean of Medicare doctors into smaller pools to which mini-SGRs and managed-care tactics could be applied. The thinking was that if doctors were no longer in a pool of 700,000 doctors but were instead in much smaller pools (say 200 to1,000 doctors), doctors would find it in their financial interest to stop ordering all those unnecessary services and, if they didn’t, they could be micromanaged by a third party.

The failure of the VPS and then the SGR led Congress to enact two laws that contained provisions that accelerated the search for the Holy Grail – quasi-HMOs that could serve as the holding pens for pools of doctors much smaller than the national pool. The first of these laws, the Medicare, Medicaid, and State Child Health Insurance Program Benefits Improvement and Protection Act (BIPA) of 2000, authorized CMS/HCFA (hereafter just CMS) to create the Physician Group Practice demo, and the other, the Deficit Reduction Act of 2005, instructed the Medicare Payment Advisory Commission (MedPAC) to dream up some other small-pool ideas. It was the 2005 instructions to MedPAC that caused MedPAC to hold that November 9, 2006 meeting with Elliot Fisher at which the “ACO” label was invented and endorsed. MedPAC’s endorsement in turn contributed significantly to the groupthink that induced Congress to include provisions in the Affordable Care Act authorizing CMS to start the Pioneer and MSSP ACO pilots.

The BIPA law of 2000 was the second in which Congress asked CMS to solve the Zen riddle they refused to solve – how to determine “belongingness” of patients to doctors without making patients enroll with a doctor or clinic. But this time CMS would have to do more than produce a study on how the impossible question might be answered. This time they would have to choose a method and use it in an actual demonstration – the PGP demo. There could be no more delay. CMS had to solve the Zen riddle presented to them by Congress – it had to devise a way to assign patients to groups of doctors so that those doctors could be punished if “their” patients got too many services even if many of those patients, um, weren’t “theirs.”

Solving the unsolvable

After Congress passed OBRA, CMS contracted with scholars at Brandeis to make recommendations on how to expose groups of doctors to financial incentives to reduce medical services. The Brandeis scholars delivered their first paper on “group specific volume performance standards” to CMS in 1991, and subsequent installments in 1992 and 1995 (see their 1995 paper here and a 2003 version of it here ).

These papers proposed the basic elements of what would later be called the ACO. They proposed “shared savings” programs under which groups of doctors allegedly large enough to bear some insurance risk would somehow cut medical costs and share the savings with CMS. CMS would measure savings (or, perish the thought, increased costs) by calculating total spending on all the patients seen by a “physician group” in a baseline year, and then compare that with total spending on the patients seen by that group in a subsequent (“performance”) year. The Brandeis papers regurgitated the folklore peddled by the dominant managed care movement, to wit:

  • FFS was responsible for “runaway” growth in Part B spending (whether growth was even more “runaway” in the private sector didn’t matter);
  • “managed care” was the solution;
  • under the lash of exposure to insurance risk, doctor groups would adopt managed care tactics;
  • those tactics would generate savings and improve quality, not the other way around;
  • CMS would find a way to measure physician cost and quality accurately; and
  • the savings would be shared between Medicare and the doctors.

The Brandeis papers offered no evidence for these claims. [4]

Although the Brandeis scholars acknowledged in their reports that Congress didn’t want Medicare recipients to be forced to enroll with physician groups, they dodged the question of how, short of forced enrollment, CMS would know which recipients belonged to which doctors. [5] It was not until the early 2000s, when CMS began planning the PGP demo, that CMS “solved” that problem. Sometime shortly before the 2005 inauguration of the PGP demo, CMS adopted the peculiar two-step method of assigning patients that they would use in the PGP, Pioneer, and MSSP programs.

When CMS began designing the PGP demo, they contracted with RTI International, not the Brandeis scholars. In a paper published in a 2007 edition of Medicare and Medicaid Research Review, RTI’s John Kautter and colleagues laid out the design of the PGP demo they had recommended and that, by then, CMS had adopted. Kautter et al. stated their recommendations “build on” the Brandeis papers, then went beyond the Brandeis studies and proposed an algorithm by which CMS could assign Medicare beneficiaries to the ten groups participating in the PGP demo.

Kautter et al. recommended that CMS first determine to which PGPs doctors belonged, and then assign patients to those doctors based on the plurality-of-primary-care-visits method. Under this method, patients would be assigned to the primary doctor they saw most often. Thus, if I see two primary care doctors during a baseline year (say 2017) a total of five times, and three of those visits were to Dr. Inside ACO and two were to Dr. Outside, I will be assigned to Dr. Inside during the performance year (say 2018). Even though I’m free to see Dr. Outside and other doctors in 2018, and even though I may never again visit Dr. Inside after 2017, Dr. Inside is still “accountable” for me in 2018.

Kautter et al. did not comment on the irrationality of the riddle posed to CMS by Congress. They merely declared, “Because the PGP demonstration is a Medicare FFS innovation, there is no enrollment process whereby beneficiaries accept or reject involvement. Therefore, we developed a methodology to assign beneficiaries to participating PGPs based on utilization of Medicare-covered services.” Having thus delicately skirted the issue of congressional sanity (don’t you just love the unctuous phrase “FFS innovation”?), they went on to say they used two criteria to determine the best assignment method:

We evaluated the alternative assignment methodologies on two criteria: Provider responsibility and sample size. First, providers must believe that the numbers and types of services they provide mean that they have primary responsibility for the health care of beneficiaries assigned to them. Otherwise, PGPs may have difficulty responding effectively to the demonstration incentives…. Second, sample size is critically important for the statistical reliability of performance measurement. If the number of beneficiaries assigned to a participating PGP is too low, then cost and quality performance measurement may be unstable.

Note the phrase “providers must believe.” How did Kautter et al. determine what doctors “must believe” about how turbulent the pool of patients assigned to them should be? Answer: They interviewed a few doctors. What did the doctors tell them? Kautter et al. didn’t say. How did Kautter et al. determine what constitutes accurate “performance measurement” and how big the pool of patients must be to achieve that? They didn’t say. They simply concluded that when they balanced the two criteria in their own minds – “responsibility” and sample size – they came up with the plurality-of-visits method.

Kautter et al. simulated their plurality-of-visits method and discovered it would cause great churn among patients. “PGPs generally retained approximately two-thirds of their assigned beneficiaries from one year to the next,” they reported. How did Kautter et al. justify such a high churn rate? Other than to say they interviewed some doctors, they didn’t. They also had no comment on the possibility that the two-step process would assign few patients to doctors and that those patients might be healthier than average.

Sometime shortly before CMS implemented the PGP demo in 2005, CMS adopted Kautter et al.’s two-step assignment method for that demo. In November 2006, the ineffable phrase “accountable care organization” was concocted by Fisher and MedPAC. And sometime between the enactment of the Affordable Care Act in March 2010 and the 2012 start date of the Pioneer and MSSP ACO programs (probably 2011), CMS decided to use the same two-step method they had adopted for the PGP demo for those ACO programs.

We have seen the consequences. PGPs/ACOs can’t cut costs and, at best, make modest improvements on a tiny handful of quality measures, an improvement which may have been accompanied by a decline in the quality of unmeasured care.

No exit

If you followed my discussion of how Kautter et al. struck a balance (at least in their own minds) between their two criteria – “belongingness” and adequate sample size – then you already know that the problems created by CMS’s two-step assignment method are not fixable.

Consider CMS’s only option to reduce patient churn. If CMS abandons the plurality-of-visits rule in favor of, for example, an 80-percent-of-visits rule, that would greatly increase the odds that the patients assigned to a doctor really do “belong” to that doctor and will continue to see that doctor in the performance year. But that would also assign even healthier patients to ACOs, and it would greatly reduce the number of patients that could be assigned to ACOs. The reduction in the number of patients assigned to ACOs would in turn make CMS’s measurements of cost and quality even cruder, and it would push the percent of ACO patients in a doctor’s panel even lower than the 5 percent level I discussed in my previous post.

ACO proponents have only two options: Explicitly prohibit patients from visiting doctors outside their ACO, which would be tantamount to admitting ACOs really were HMOs in drag all along; or redefine ACOs so that they are no longer responsible for entire “populations” but instead focus on the chronically ill. I will discuss these options, the impact these options would have on MACRA, and the final Pioneer ACO evaluation in my next post.

[1] Evidence on the near-total lack of awareness among patients of their assignment to an ACO appears in the final evaluation of the Pioneer ACO program, released last December. The author of that evaluation, L&M Policy Research, held focus groups with Medicare recipients who had been assigned to a Pioneer ACO. “[W]e learned that beneficiaries were generally unaware of the ACO organization and the term ‘ACO,’” L&M reported. “In the few cases where the beneficiaries reported hearing the term ACO, they were not able to describe what an ACO is and its relationship to them as recipients of health care services. Since beneficiaries were not even aware of the term ‘ACO,’ they also were unaware that their care was being provided or coordinated by an ACO.” (p. 51)

Patient ignorance of their status as an ACO member may have been aggravated by physician ignorance. According to L&M, “In several respects, physicians were not particularly knowledgeable about the ACO. When asked if they knew which of their patients were aligned with the Medicare ACO, just over a third of Pioneer physicians reported knowing which beneficiaries were aligned and a similar proportion reported not knowing their aligned beneficiaries at all. When asked about the elements of their compensation, almost half of physicians participating in the Pioneer model reported not knowing whether they were eligible to receive shared savings from the ACO if the ACO achieved shared savings.” (p. 43)

[2] As Kelly Deverson and Robert Berenson  put it, “Together, the Medicare Payment Advisory Commission … and [Elliot] Fisher provided the impetus for the current concept and interest in ACOs.” (p. 2)

[3] It should have been obvious to Congress why traditional FFS Medicare was beating the pants off the insurance industry. First, the traditional Medicare program paid doctors and hospitals substantially less than private-sector insurers did. Second, the traditional program devoted a much smaller percent of its expenditures to overhead (2 percent since the early 1990s) compared with the 20-percent overhead of the insurance industry. (For evidence that the insurance industry’s overhead is 20 percent, see this graphic  published by America’s Health Insurance Plans. For evidence that traditional Medicare’s overhead is 2 percent, see citations to reports by the Medicare trustees, the Congressional Budget Office and others in my paper on this subject in the Journal of Health Politics, Policy and Law.) The insurance industry has never figured out how to overcome those two advantages – lower payment to providers and lower overhead – and it never will.

Moreover, as of the early 1990s the traditional Medicare FFS program did not micromanage Part B doctors as the insurance industry did and as MACRA is forcing CMS to do today. That in turn meant traditional Medicare was not driving up physician overhead costs, and was not burning doctors out, anywhere near as much as the insurance industry did and does now.

[4] Here is just one example of numerous evidence-free paeans to HMOs and managed care strewn throughout the Brandeis papers produced under contract with CMS, then HCFA: “The efficiencies … should be achieved through effectively managed care…. Presence of utilization review and quality assurance programs and other features associated with managed care may also be prerequisite.” (“Models for Medicare payment system reform based on group-specific volume performance standards (GVPS),” unnumbered page, Appendix B )

[5] The authors of the Brandeis papers might object to my statement that they “dodged” the issue of how to assign patients to physician groups. They could argue, correctly, that they did mention the issue and decided they didn’t need to assign patients to each doctor. I won’t try to explain here the strange logic they used to justify that position. Suffice it to say their primary argument, presented without a shred of evidence, was that risk adjustment could accurately detect changes in the average health status of an ever-changing pool of patients.

Categories: Uncategorized

Tagged as: , , ,

12 replies »

  1. Kip, thanks for responding and thanks for trying to clarify the situation. There is not just a small kernel of truth in what I said but a Costco sized bag of kernels. Firstly, comparing MA to FFS Medicare is comparing two different systems to one another. That is, I think, where you were going with your citation and I think that was a bit of deflection.

    I am comparing the 2% Medicare overhead with the overhead of those insurers in the for profit sector insuring those below 65. When that is carefully observed there is no doubt that the methodology used to compare these percentages is intentionally skewed to pretend Medicare’s overhead is very low. Based upon this calculation, that you seem to favor, if Medicare were to to increase its expenditures per patient infinitely its overhead would trend to 0%. That demonstrates the foolishness in using this type of calculation in this type of comparison. MA overhead has to be higher than Medicare FFS overhead because MA has to deal with all the other costs left out in the 2% figure. Some of those costs were mentioned in my earlier reply below.

    “Can you post a link to any research that supports your assumption that administrative costs do not rise as per-enrollee bills rise?”

    They do rise, but generally not as fast. …And you are wrong, I have had ownership experience of a private insurer. Are you going to stand there and state that the 2% overhead is linearly based upon the size of the bill? That would mean a hospital surgical bill of $10,000 would have overhead costs of $200 while a $100 bill would only incur a $2 overhead charge? Of course not, no matter what assumptions you make.

    “ Please read my paper.” You also ask “Let’s treat this issue as an empirical issue, not one to be determined by ideology. Unfortunately, we have very little research to turn to.”

    Right off the bat we have Mark Litow’s paper, common sense, and the experience of huge numbers of businesses that recognize the inappropriate way your 2% calculation was performed.

    Why don’t you simply provide a complete list of documented expenditures included in the 2% figure. Mark Litow a statistician (from Milliman) carefully demonstrated that your emphasis on the 2% overhead is wrong.

    “The study estimated that Medicare administra-tive costs during 2003 were lower as a percentage of claim costs than private health insurance admin-istrative costs for functions that were readily comparable. Medicare administrative costs were estimated at 5.2 percent of total costs (benefit payments plus administrative costs) and private insurance administrative costs were estimated at 8.9 percent of total costs (premiums). But this comparison does exclude some significant differ- ences between Medicare and private health insurance.”

    Take note of the 5.2% instead of the 2%

    “Medicare’s method of calculation makes administrative costs higher as a percentage for commercial insurance, but lower when calcu- lated as a PMPM. The right answer is somewhere between a percentage of premium and a PMPM, but the point is an important one when trying to make comparisons.”

    Some comments on the comparisons above are as follows: From the Society of Actuaries Health Watch Newsletter May 2006 – Issue 52

    • ♦ Administrative costs are higher than reported in the Federal budget (about 2 percent).

    • ♦ Medicare administrative costs are expected to decrease over time because Medicare benefit costs increase at a higher rate than administra- tive costs.

    • ♦ The private market administrative costs are expected to remain at about 9 percent of total private insurance cost, excluding premium taxes, commissions and profit.

    • ♦ Other significant differences exist between Medicare and private health insurance, which could significantly alter the comparison if recognized as discussed below.

    The graph presented in Litow’s article shows how Medicare’s costs decline as their payments rise making this type of comparison look a bit ridiculous.

    Actually I believe Litow was being overly generous to your side of the story because the unintended consequences of Medicare also have costs, but that is not something that can easily be discussed in short form in a blog such as this. I would, however, like to know how 60 minutes $60 Billion in fraud or ~10% of Medicare’s ~$600Billion fit into your calculations. Your response to almost every question is for another to prove a negative or search the literature which has been done (See Litow’s original paper). Why not first simply respond to the question asked?

  2. Ok, I think I understand your question. You’re asking whether I think CMS’s failure to assign a random sample of Medicare beneficiaries to ACOs is a major factor in ACOs’ inability to cut Medicare’s costs.

    I’m going to address that question in my third and last installment in this series (hope to get that done in the next few days) so I’ll try to keep this short. I don’t think it is a major factor. I think it’s only a minor, secondary factor. Even if ACOs were to get their share of the really sick, they still wouldn’t be able to cut Medicare’s costs or make enough money for themselves to pay for the cost of whatever interventions they use to improve quality and lower cost.

    The far more important obstacles to the ACO proposal are:

    * The impotence of every managed care tool deployed by the first wave of HMOs (closed networks, utilization review, formularies, profiling, public report cards, more preventive services) and every tool promulgated since the HMO backlash of the 1990s (disease management, P4P, EMRs); and

    * the high turnover rates among ACO docs and patients.

    It is conceivable that some of the tactics we lump under the label “managed care” would not only improve quality but lower costs if they were applied ONLY to a select group of chronically ill patients as opposed to entire populations. The HMO and ACO experiments have failed not just because the tools available to them were so crude, but because HMO and ACO proponents insisted that HMOs and ACOs apply their magic to the healthy and the sick. They insisted, in other words, that the crude managed care tools be applied not just to the 20 percent of us who account for 80 percent of total spending, but also to the 80 percent of us who account for only 20 percent of total spending. Thus, the net savings that MIGHT be achievable if a few carefully selected interventions were applied to chronically ill patients get diluted down to nothing-much-to-write-home-about when they’re spread out over an entire population.

    In regard to your comment that studies in JAMA and elsewhere that claimed to risk-adjust the expenditures on (simulated) ACO and control groups, keep in mind that CMS’s risk aduster (the HCC method) is very crude. It explains only 12 percent of the variation among individuals, and that’s only when it is applied to huge groups (in the millions). We can be sure that the risk adjustment done in those studies was not accurate.

  3. Allan, your first argument may have a small kernel of truth to it but it is basically wrong; your second argument is completely inaccurate.

    Your first claim is that Medicare’s 2 percent overhead would be higher if it were measured as a percent of the smaller payments made by private insurers. But the reported average overhead of Medicare Advantage plans is 13 to 14 percent of expenditures, or approximately seven times FFS Medicare’s 2 percent overhead (see for example p. 7 and p. 27 of this US GAO report http://www.gao.gov/new.items/d08359.pdf). MA plans are paying per-patient claims that are much higher than those they pay for their non-elderly insured, and still they have much higher admin costs than FFS Medicare. (The actual overhead of MA insurers is higher than 13-14 percent if they are counting much of their meddling in the doctor-patient relationship as “medical care.” I don’t have time at the moment to dig up a citation, but I believe that’s what all the Medicare Advantage plans do now).

    Your claim that FFS Medicare’s overhead would rise if Medicare paid the much smaller bills of the non-elderly was based on your assumption that sicker populations do not generate higher administrative costs for insurers. Here is how you stated that assumption: “Overhead doesn’t rise that much with increasing bills.” How do you know that to be true of health insurance companies and programs? You don’t. You’re just guessing. Right? Can you post a link to any research that supports your assumption that administrative costs do not rise as per-enrollee bills rise?

    Let’s treat this issue as an empirical issue, not one to be determined by ideology. Unfortunately, we have very little research to turn to. That’s because the US health policy community gives very low priority to understanding the administrative costs of our system (they have to save all their time to study overuse by greedy doctors). The only research on this issue you raise that I know of was done by the Congressional Budget Office in 1993. The CBO considered and rejected your assumption that the overhead costs of insurers do NOT rise as per-enrollee medical claims rise. It appears that as patients get sicker, they file more claims that have to be processed, they file more appeals of denied claims, more money has to be spent on quality and fraud control, etc. For further discussion of your claim and of the 1993 CBO paper, please go to pp 492-493 of my paper, “Thinking clearly about Medicare administrative costs,” the link to which is in the footnote you’re commenting on.

    Turning to your second argument — that the Medicare Board of trustees, the source for the 2-percent-overhead figure for FFS Medicare, was too dumb to include administrative expenses other than those incurred in processing claims. That claim is flat wrong. Every conceivable administrative expense is included in the 2 percent figure (revenue collection, fraud control, the cost of heating buildings, etc.) with one possible exception — the cost of treating Congress as if it were the board of directors of Medicare (a trivial sum). Please read my paper. There’s no point in my taking up space here repeating what I said there. To my knowledge, no one has attempted to refute that paper in any peer-reviewed journal or even in a sloppy paper for a biased think tank.

    You might want to double check the paper by Litow you refer to. If it’s the same paper I read years ago, Litow didn’t even inform his readers where the 2 percent overhead figure came from. How does anyone credibly attack the Medicare trustees’ estimate of Medicare’s overhead if he doesn’t even know who he is criticizing, what document he is criticizing, or what methodology the people who doesn’t know employed to derive their estimate?

    As for your arguments that Medicare is victimized by $60 billion worth of fraud, could I ask, Compared to what? Are private-sector insurers never defrauded? Could you post your estimate of how much fraud is costing the private-sector?And while you’re at it, could you post information on how much the private sector spends on fraud control per insured individual versus FFS Medicare?

    Finally, I suggest that when we’re comparing FFS Medicare overhead with that of United Healthcare and their ilk that we tack onto their 13-14 percent minimum MA overhead whatever it costs the US government to prosecute MA insurers for defrauding the American taxpayer plus the additional cost to CMS of regulating hundreds of MA insurers. Sparse evidence suggests that it is more costly for governments to manage privatized programs (Medicare, Medicaid) than good old-fashioned, skip-the-middleman programs like FFS Medicare.

    I appreciate your reading my diatribes here on THCB. But I do want to close by quoting you back to you: “I think we should we more careful when attempting to compare Medicare overhead with private insurer overhead….”

  4. Fee For Click – fantastic!

    So the AAFP and other primary care societies are supporting ACOs/PCMH/P4P/value based-payment because they’re brain dead morons. That’s obvious.

    But why are the specialty dominated AMA and the other specialty societies supporting these Rube Goldbergesque payment schemes? What’s the payoff for them?

  5. Good, Allan.
    Another thing to remember is that administrative costs are isually calculated as a fraction: adminstrative expenses for the patient divided by total costs for the patient. Total costs go way up in the aged, so that the percentages will look much smaller. Indeed, some authors have written that administratve costs per patient in Medicare actually are higher than those with other insurance. A guy in Forbes keeps saying this. You have to include the services of the other agencies involved in the Medicare program, eg collecting FICA.

  6. “Second, the traditional program devoted a much smaller percent of its expenditures to overhead (2 percent since the early 1990s)”

    I liked your discussion. Very valuable, but when numbers such as the 2% Medicare overhead is used, as if that is a real number, things become quite confused. Comparing Medicare overhead to private insurer’s overhead is like comparing apples to oranges. They are both round, but don’t taste the same.

    There is a bias in the way that 2% is measured. Per patient Medicare payments are paid at least twice as much as private insurer payments. (overhead divided by total cost (payments): 100/1000 = 10% overhead 100/2000 = 5% overhead). We know from most businesses that this is a poor way of calculating overhead. Overhead doesn’t rise that much with increasing bills. If it did then every time Medicare costs doubled their overhead would be cut in half. The solution to the overhead problem for Medicare would be to pay an infinite amount of money per patient and overhead would trend towards zero.

    That is just a part of the problem assuming the 2% represents Medicare overhead. (Based upon what year or average of what years that number obtained. I have heard higher numbers.) Mark Litow in his paper explains that the over head generally provided by Medicare involves only the payment of claims. There is also the problem of generating revenues, the costs of which are not included in the 2%, nor are capital costs, nor are taxes etc. Part of the insurer’s overhead is spent to prevent fraud. CBS 60 minutes discussed $60 Billion worth of Medicare fraud a number of years ago. $60Billion of fraud in a $600 Billion program is 10%.

    I think we should be more careful when attempting to compare Medicare overhead with private insurer overhead because that leads to some very bad conclusions. I don’t think it matters in this discussion, but it does matter in the way others perceive how financing of healthcare should be managed in the future and this 2% number seems to be a set up for another debate.

  7. Sure. Thanks for engaging, Kip.

    You made references to how ACOs have a healthy enrollee bias. I pulled a quote from part one below:

    “In its first evaluation of the Pioneer program released in May 2015 (with a March 2015 date on it) in which L&M Policy evaluated the program’s first two years, L&M suggested that CMS’s method of assigning Medicare recipients to ACOs could result in favorable selection, that is, the assignment of healthier patients to ACOs. In this final evaluation, L&M did not mince words: They clearly stated that CMS’s assignment algorithm causes highly favorable selection”

    In NEJM and JAMA when investigors examine ACO savings, and I know that term ‘savings’ has many subtexts–they compare control and intervention cohorts based on both CMS and more detailed methods–to take patient characterics into account. While the savings may be harder to obtain on healthier group, the spend easier to control as well. Also, we know CMS bases performance off of a baseline of predicated costs.

    Again, researchers outside of Mcare doing the statistical leveling on the same groups as CMS and the outcome indicates ACOs are having difficulty eeking out profits. I would think its not the population of folks and sorting then but the fixed costs of running and ACO and other variables as major drivers. Not patient selection. Thats being controlled for in the data. Now if the statistical adjusting off by unmeasured confounders the outside groups cant control for, and I am sure there is a some of that, that would endorse L&Ms conclusions. But i cant say.

    I hope that is clearer.


  8. It’s not that FFS is so terrible, but the current manifestation is very unfair to Primary Care, based on the corrupt RVU system. Example: If a radiologist does a CT scan of the chest, abdomen and pelvis, those are considered 3 separate procedures. If I see a patient and address hypertension, diabetes, and diagnose a new rash, that is called “office visit”. Add a few more services like reviewing their health maintenance and it can rise to the occasion of slightly more “office visit”. Additionally there is bias on what is or is not a “service”. Much of what Primary Care does is behind the scenes and remains unrecognized. This desperate trap that Primary Care is in has led to a quest for anything other than the current FFS system. So medical homes and ACOs will save us, right? Another example of where fools rush in. The same inherent problems in our current system are baked into the new models and are contributing to the failures we are witnessing. Kip has made some incredible points. These models continue to fail for these additional reasons: 1) They are routinely an overlay on the same broken iteration of RVU based FFS. 2) Despite “moving to value”, most providers remain on production. 3) Moving from FFS to FFC (fee for click) does absolutely nothing. 4) The model of largely consolidated entities funneling patients up through Primary Care so they can be relentlessly churned by hospitals and specialists has not changed at all. 5) Controlling cost and use in medicine would require controlling all sources of cost and all reasons for use. So unless someone wants to give physicians deity like control over all components in our health system, including human behavior, no one should expect remarkable results through managed care. While CMS is trying to figure out new ways to punish and rip off physicians for taking care of the sick, drug companies are participating in a price gouging bonanza. Now, who is going to “manage” that? The ACO?

  9. Kip,
    Another outstanding Post from your pen. Medicare actuarial traditions have never recognized the potential value of Primary Healthcare to represent the importance of long-term continuity for a person’s healthcare, even if only based on a supportive source of responsive accessibility. The remaining attribute of this Medicare bias can be demonstrated by the fact that 9 of the Flexnor-Era (@ 80) legacy medical schools still do not sponsor a Family Medicine residency. And, for a long-time, Medicare did not cover a Tetanus immunization on a Routine basis, only with a specific injury.
    A recurring theme of governance in large institutions of our modern times is a lack of connection between the upper levels of governance and the front-line governance to honor the Vision, Mission and Values of these institutions. The definition of an institution formulated by Nobel Prize Winner, Elinor Ostrom, says it best. As follows, she has said: “An INSTITUTION may be defined as the rules that humans use to organize all forms of repetitive and structured interactions within families, neighborhoods, markets, firms, sports leagues, churches, private associations, and governments, at all scales. Individuals interacting within rule-structured situations face choices regarding the actions they may take, leading to consequences for themselves and for others. The opportunities and constraints individuals face in any particular situation, the information they obtain or are excluded from, and how they reason about the situation are all affected by the rules or absence of rules that structure the situation. IF THE INDIVIDUALS who are crafting and modifying the rules do not understand how a particular combination of rules affects the actions and outcomes in a particular ecological or cultural environment, RULE CHANGES may produce unexpected, and at times, DISASTROUS RESULTS.” (capitalization added by me)
    I would also cite a quotation from President Thomas Jefferson about institutional adaptation that ends with historically augmented, but similar, two words: “…barbarous ancestors.”
    see http://www.nationalhealthusa.net/innovation/epilogue/

  10. Another great post.

    Any thoughts on why the medical societies, from the AMA on down (and the primary care societies most of all) have swallowed the FFS bad/ACO good meme hook, line and sinker?

  11. Kip
    On the healthy patient spillover: in the published peer review Pioneer ACO lit, the studies have been risk adjusted. The results parallel CMS findings rather closely. Can you comment on that?