Understanding the True Costs of ACOs and Medical Homes

flying cadeuciiOne of the privileges of being a managed care advocate is that you never have to discuss the unpleasant question of how much your proposed intervention will cost. Whether your proposed intervention is HMOs, report cards, pay-for-performance, ACOs, “medical homes,” or electronic medical records, you never have to estimate what your bright idea will cost. With this privilege comes another: You are free to criticize doctors and hospitals for being “cost unconscious.”

Over the last decade, CMS has become a proponent of this double standard – cost consciousness for doctors and hospitals and cost unconsciousness for the health policy illuminati
. Beginning with the Physician Group Practice Demonstration, which ran from 2005 to 2010, and running through today’s ACO and “medical home” demos, CMS has assiduously avoided reporting the costs that clinics and hospitals incur to participate in these demos. Jeff Goldsmith and Nathan
Kaufman  have described CMS’s behavior as “sunny obliviousness to provider economics.” [1]

Cost unconsciousness permeates CMS’s MACRA rule

CMS’s cost unconsciousness is on vivid display in its proposed MACRA rule . About two-thirds of the way into the rule, CMS admits:

it does not know what it costs doctors to implement electronic medical records nor whether the benefits of EMRs exceed the costs (p. 669);

it does not know what it costs doctors to participate in “medical homes” (p. 670);

it will allow ACOs and “homes” to participate in MACRA’s Alternative Payment Model (APM) program even if those entities cannot cut costs (pp. 460-461); and

it cannot say whether the benefits of its MACRA rule will outweigh the costs. (p. 679 and 684)

Moreover, by maintaining throughout the 962-page rule its long-standing silence on the subject of what it costs to start and run ACOs, CMS concedes it doesn’t know or doesn’t care what that cost is either.

And yet CMS gives every indication it intends to impose its rule on the doctors and nurses who treat Medicare patients. Is that not the textbook definition of cost unconsciousness?

All this cost unconsciousness spells trouble for MACRA. CMS’s refusal to acknowledge the costs that doctors incur to join ACOs and “homes,” coupled with CMS’s willingness to let ACOs and “homes” into the APM program regardless of whether they cut Medicare’s costs, suggests MACRA’s APM program will fail one of two ways: Either doctors won’t participate in APMs because their costs will exceed payments they’ll receive from CMS, or they will participate but the APMs they join will raise Medicare spending. [2]

CMS is in this bind because ACOs and “homes” do little or nothing to cut Medicare’s costs despite boasts to the contrary by ACO and “home” proponents. If those entities did save money, the physician’s share of the savings might be big enough to make participation in ACOs and “homes” financially attractive. But that’s not what CMS’s demonstrations have shown.

CMS’s three “home” demos and three ACO demos indicate that the gross savings these entities can achieve (savings before taking into account payments CMS makes to ACOs and “homes”) are at most one percent of CMS’s claims costs, and are sometimes negative (CMS’s costs go up). The small or nonexistent savings cause tiny or zero payouts from CMS to ACOs and “homes.” Those payouts, in turn, are too small to cover the start-up and maintenance costs ACOs and “homes” incur. And yet, to complete my description of the bind CMS is in, those tiny payouts help reduce CMS’s small gross savings to zero or, worse, raise Medicare’s net costs.

CMS has gone out of its way to make it difficult for the public to understand how poorly Medicare ACOs and “homes” have performed. You most definitely will not find evidence on this issue in CMS’s MACRA rule. Allow me, then, to review the limited evidence.

ACOs and “homes” are not cutting Medicare costs

As I reported in a previous comment here , the three “medical home” demonstrations CMS has conducted have shown “homes” cannot cut Medicare’s costs. Evaluations of two of the demos (the [CPCI]  and the Federally Qualified Health Centers demo ) suggest they are raising Medicare’s costs when CMS’s “care-coordination-fee” payments to “homes” are taken into account (the evaluation of the third demo made no attempt to estimate savings). The second-year evaluation of the CPCI by Mathematica is the only one that reports data on the gross savings achieved by the “homes” and the payments CMS sent back to them. CPCI “home” clinics cut Medicare spending by 1 percent but that was swamped by payments CMS made to the “homes” of 1.6 percent. [3]

The results from the three ACO demos are very similar:

PGP demo: According to the final evaluation of the Physician Group Practice Demonstration  (which was a test of the ACO), “[T]he demonstration saved Medicare .3 percent of the claims amounts, while performance payments were 1.5 percent of the claims amounts” over the five years the demo ran, for a net loss of 1.2 percent (p. 64).

Pioneer ACO demo: According to an April 2015 report from CMS, Pioneer ACOs reduced Medicare spending by 1.2 percent in 2012 and 1.3 percent in 2013, but these savings were offset by payments to the ACOs of 1.0 and .8 percent respectively, for net savings of  .2 and .5 percent. (p. 4)

MSSP ACO demo: According to the same April 2015 CMS report, MSSP ACOs reduced Medicare spending by 0.5 percent for the period April 2012 through the end of 2013, but these savings were more than offset by payments to the ACOs of 0.7 percent, for a net loss of 0.2 percent. (p. 4) [4]

Pioneer and MSSP demos together: According to an analysis  done by Kaiser Health News, CMS broke even on the two ACO demos combined after taking into account bonus payments to the ACOs. [5]

To sum up, the evidence indicates Medicare ACOs and “homes” are achieving a gross savings rate of no more than 1 percent, and on a net basis (that is, when CMS’s payments to those entities are taken into account) ACOs and “homes” are saving no money for Medicare or are raising Medicare’s costs.

Are ACOs and “homes” making any money?

So how do the CMS payouts compare with the cost to providers of starting and running ACOs and “homes”? The limited evidence on ACO start-up costs and overhead is easy to summarize, so I’ll start there.

The only evidence I’m aware of appears in transcripts of two MedPAC meetings. According to MedPAC staff, ACOs spend 1 to 2 percent of their Medicare revenues creating and administering ACOs (I’ll refer to these costs hereafter as administrative or overhead costs). The staff claimed they got these estimates from interviews with a few ACOs. These estimates sound low to me. [6] But even if ACO overhead is no more than 1 or 2 percent, the economics don’t work for the doctors and hospitals who finance the ACOs. If ACOs are at most cutting Medicare’s gross costs by less than 1 percent, and if Medicare splits that 50-50 with the ACOs (the split is not necessarily 50-50; I use 50-50 for the sake of illustration), that would mean ACOs are losing somewhere between half a percent and one-and-a-half percent on their Medicare patients. MedPAC’s staff agree. [7]

I am unaware of research (or even conversation at MedPAC meetings) that tells us what the administrative costs of “homes” are as a percent of their Medicare revenues. We do have research on those costs in absolute dollar terms. According to the first-year report on CMS’s Comprehensive Primary Care Initiative by Mathematica, CMS and insurance companies participating with CMS in the initiative paid “homes” $70,000 per doctor, an amount that Mathematica said equaled 19 percent of each doctors’ “total practice revenue.” CMS’s share of that $70,000 came to $45,000 (p. 64 of the Mathematica report). A paper by Magill et al. found the cost of partial implementation of “homes” is much higher; they concluded it is around $105,000.

These numbers are astonishing. Mathematica’s report that insurer payments to CPCI doctors equaled 19 percent of their entire income suggests that “homes” are costing doctors 100 percent or more of their entire Medicare income (Medicare is the source of about 20 percent of physician income in the US). Could it be that “homes” eat up 20 percent or more of physician total revenue and 100 percent or more of their Medicare revenue? If so, something is very wrong with the “home” experiment. It can’t possibly save money for Medicare.

In any event, it is not clear that CMS payments to the clinics in its “home” demos cover the overhead costs those clinics incur to participate in those demos.

CMS ignores some demos, spins others

With the exception of the conclusions of the Magill paper, CMS fails to mention in its proposed rule any of the facts I just gave you. Although CMS acknowledges that the administrative costs of ACOs and “homes” are “substantial” (p. 481), CMS refuses to make any effort to measure those costs and weigh them against the revenues doctors are likely to get from ACOs and “homes” under MACRA. [8] Consequently, CMS has no idea whether doctors will join ACOs and “homes” and, if they do, whether ACOs and “homes” will cut Medicare costs.

CMS’s glib treatment of the administrative costs of ACOs and “homes” is aggravated by its misrepresentation of the results of its own demos. CMS cherry-picks which of the demos it discusses (it discusses only two of its three ACO demos and just one of its three “home” demos) and inflates the results of the three demos it does discuss.

Here, for example, is how CMS misrepresents the results of the Pioneer and MSSP ACO demos. Citing an August 2015 CMS press release, CMS states: “Taken together, Pioneer and Shared Savings Program ACOs yielded $411 million in cost savings in 2014.” (p. 678) But that $411 million figure conveniently failed to take into account the payments CMS made to ACOs. Here is how Kaiser Health News  described the same data: “In August [2015], Medicare officials released 2014 financial details showing that the so far the ACOs have not saved the government money. The 20 ACOs in the Pioneer program and the 333 in the shared savings program reported total savings of $411 million. But after paying bonuses, the ACOs recorded a net loss of $2.6 million to the Medicare trust fund.”

CMS did not have to play the role of enabler

This concludes my three-part comment on CMS’s proposed MACRA rule. In part 1 and part 2  I argued that both programs within MACRA, the Merit-based Incentive Payment System and the APM program, will fail because CMS cannot measure physician value accurately. In this part I have argued that the APM program will fail for an additional reason – it will rely on ACOs and “homes,” neither of which has shown an ability to cut Medicare costs.

I realize that Congress put CMS between a rock and a hard place when it handed the MACRA mess to CMS to decipher and implement. No agency, no matter how dedicated, can make MACRA work. When we as a society finally get around to agreeing MACRA failed, we should blame Congress first, not CMS.

But CMS does deserve to be blamed for acting as a congressional accomplice, a congressional enabler. If CMS had stated clearly – in its proposed rule and in its testimony to Congress and in all other public statements – that it does not have the ability to measure physician cost and quality accurately, and that its ACO and “home” demos have demonstrated that ACOs and “homes” have yet to demonstrate an ability to lower Medicare’s costs, I for one would have been far more reserved in my criticism of CMS.

But CMS has done just the opposite. CMS staff have sold us hype and half-truths when they should have given us facts and the whole truth. For that, they deserve severe criticism.

[1] CMS’s sunny obliviousness to ACO and “home” costs extends to CMS’s own costs as well. CMS often fails to mention the payments it makes to ACOs and “homes,” and it never reports the costs it incurs to administer its ACO and “home” demos.  MedPAC, another proponent of cost conscious for doctors and cost unconscious for managed care advocates, is also totally uninterested in the costs CMS incurs to administer its demos. At MedPAC’s September 11, 2014 meeting, commissioner Mary Naylor inquired what all of CMS’s “efforts” in running ACO programs are costing. “Do we have a sense of the overhead cost for the Medicare program as we’re doing this model development?” she asked. “It’s just something that I think is really important.” (p. 165 of the transcript ) None of the other 16 commissioners supported her. The staff remained silent.

[2] CMS says ACOs and “homes” will qualify as APMs even though they cannot cut costs. In its discussion of what types of entities will qualify as APMs (CMS calls them “advanced APMs”), CMS declared: “[A]n Advanced APM Entity that bears more than nominal financial risk for monetary losses … would be an Advanced APM Entity regardless of whether it actually earns shared savings or generates shared losses under the Advanced APM …. We do not intend to add additional performance assessments on top of existing Advanced APM standards. …” (pp. 460-461)

[3] Peikes et al., authors of the Mathematica evaluation  , reported that the CPCI cut Medicare costs by $11 per “member” per month (PMPM) during its first two years, and that this equaled 1 percent of Medicare spending. Peikes et al. also reported that CMS payments to CPCI “homes” amounted to $18 PMPM (p. xv), but did not translate $18 into a percentage term. I did, and derived 1.6 percent. Last April, Dale and others from Mathematica  reported slightly different results for the first two years of the CPCI. They reported “home” clinics were still not saving money, but the ratio of payments by CMS to savings changed slightly.

[4] In that same April 2015 report , CMS went on to “certify” that Pioneer ACOs will save money in the future, possibly as much as 5 percent per year gross (CMS didn’t bother to say what the net would be). CMS accomplished this magic by ignoring the actual performance of Pioneer ACOs in 2012 and 2013 and relying instead on a simulation of what might happen in the future if CMS measured ACO success using different prices and different Medicare beneficiaries in the ACOs and FFS comparison groups than CMS has actually used. CMS used this same tactic to produce rosy results in a paper published in the Journal of the American Medical Association on May 4, 2015 (see my comment here ).

[5] Here is how Kaiser Health News put it in its October 2015 article : “In August, Medicare officials released 2014 financial details showing that so far the ACOs have not saved the government money. The 20 ACOs in the Pioneer program and the 333 in the shared savings program reported total savings of $411 million. But after paying bonuses, the ACOs recorded a net loss of $2.6 million to the Medicare trust fund….”

[6] One- or 2-percent overhead for ACOs sounds low to me. The typical health insurance company has overhead costs of 15-to-20-percent of revenues. ACOs are risk-bearing entities that must perform many of the functions insurance companies carry out (such as forming and maintaining provider networks, profiling doctors and hospitals, and distributing revenues from ACO activities to participating clinics and hospitals). Is it really possible that they can perform those insurance-related functions and hire more staff to provide the extra medical and social services ACOs allegedly provide with no more than 1 or 2 percent of ACO revenues?

[7] At the September 11, 2014 MedPAC meeting, commissioner David Nerenz asked MedPAC staffer Jeff Stensland if “we know anything about” ACO “overhead.” Stensland replied, “[P]eople we talk to and the data we have seen, it looks like maybe 1 to 2 percent of your spend, that that’s what they’re spending on their ACO

to operate it….” (p. 133 of the transcript A few minutes later, Nerenz asked: “[I]f I could go back to the overhead cost issue, I just want … to see if I’m thinking about this correctly. If, say, Jeff, as you said, two percent of overall spend might be administrative – you said one to two, but let’s just use two – in order, then, for the ACO to make money net, it would have to achieve four percent savings if the  shared savings ratio is about 50 percent. …. [Y]ou’d have to have a savings of four percent to get two percent back and at that point you break even…. [A]re there ACOs who have actually made money net of overhead cost? Do we know that yet?” (pp.143-144) Stensland replied, “[I]f you averaged everybody [that is, all ACOs] so far, at least in the first year of the program, the share of savings, on average, that they get is going to be less than their administrative costs of being in it….”(p. 144)

[8] For example, in its discussion of what should constitute “nominal risk” (bearing “nominal risk” is one of the three criteria MACRA requires APMs to meet), CMS rejected the suggestion that the high cost of starting up and running ACOs and “homes” should count as money at risk. CMS’s rationale was, “The amount of financial investment made by APM Entities may vary widely and may also be difficult to quantify….” (p. 481). Note the double standard: Physician “merit” is not “difficult to quantify” even with patient pools as small as 20, but the money ACOs and “homes” spend to meet CMS’s criteria is just too difficult to measure.

Categories: Uncategorized

9 replies »

  1. Agree with the comments below. Some things to include in a better model:
    1. Stop punishing physicians for providing care to sick people.
    2. Stop punishing physicians for taking care of people with limited resources.
    3. Understand that raising the bar raises overhead, this should be funded accordingly. CMS has to stop thinking all this stuff comes for free. It is why physicians are not performing. Volume based care continues to rule in MACRA. People need to quit talking about aligning reimbursement with quality and actually start doing it. There is nothing in MACRA that gets me more time to address chronic illness…nothing!
    4. A better model will take a few steps back and reassess when it is going down the wrong path. CMS has difficulty with this concept as noted in it’s simple renaming of MU.

  2. “The best-designed model will fail if physicians do not cooperate, just as the very best healthcare does little if patients do not comply.”

    Well, first you have to design something that will make sense to practicing physicians that will not be too cumbersome and intricate. They are busy enough taking care of complex patients in a complex system they have limited control over. And realistically, physicians’ fees are a small portion of the huge medical costs incurred by patients.
    As for patients, if you’ve ever treated them, they are complex as well, some following recommendations to the tee and getting healthier, some doing the best they can with what they have, and some keep up the bad habits. And now doctors are expected to put out large amounts of money for questionable return, the only entities that can afford to do so are large organizations or hospitals. The small private practice can’t keep up with all this stuff.
    Apparently we can’t pay docs just for doing things because they “do too much”, yet we want to pay them for doing less, how do we keep them from doing “too little”?
    I recommend you go to the website for comments on MACRA and see how docs feel about it. I have serious reservations that CMS will listen.
    And by the way, here’s an article showing small practices can do well. But I guess that’s too simple for CMS to understand:


  3. “But we have to do SOMETHING” doesn’t excuse poorly executed schemes that haven’t been properly thought out.

    Identifying something as clearly broken is an important part of the process. Sometimes the answers come later, or from others. Kip Sullivan has done an excellent job of explaining why these solutions are not actual solutions.

  4. Jim, you said ” so what? We must do SOMETHING to change the trajectory of our healthcare delivery system which remains gluttonously expensive and or poor quality.”

    No need to worry any longer (or ask Kip to redirect his prodigious analytical skill!) There is a well tested answer, studied and endorsed by no less than Rand! The only problem…..power and money flows away from the administrative class and the bureaucrat class back to patients and docs…..so it tends to be ignored. Here is the link to Rand:


  5. Kip, you’ve demonstrated your ability to critically analyze CMS’ activities and what you describe as misinformation or lack. Very impressive. But at some point, I start to think: so what? We must do SOMETHING to change the trajectory of our healthcare delivery system which remains gluttonously expensive and or poor quality. So while your devastating “cross examination” is impressive, I’d be even more impressed by you using your skills to design a better model. I’m guessing you could do better than CMS has done. And of course, CMS can design, but physicians must perform (so to speak). The best-designed model will fail if physicians do not cooperate, just as the very best healthcare does little if patients do not comply.

  6. Agree this is a good summary of the data to date. However, I would not discount the progress that has been made in advancing primary care to a more comprehensive model.

    Our practice has been participating in CPC, NS are in our final year of the 4 year demonstration, – but plan to continue in CPC+.

    Although savings have not been demonstrated in a global sense for the program. You do point out that there were savings if you do not include the “cost” to the program that has been paid to practices to develop more comprehensive primary care models. However consider that any savings over trend must be considered as a positive benefit for the patient’s served. Fewer readmissions, fewer unnecessary ER visits, fewer avoidable hospitalizations. whatever the reason, costs were lower than trend. All of these activities were possible because of increased resources in these primary care practices. This includes care managers, integrated behavioral health, health coaches.

    It will likely take more time for the benefits of this type of work to be achieved. For most practices, such as ours, we are really just getting her feet under us with the additional staff we have added since the program began. We are refining workflow’s and care pathways every year.- we are learning as we go.

    I’m encouraged by the results we have seen in our practice and would anticipate over the course of the next several years, as we become more sophisticated, with our delivery models, we will see greater savings.

  7. Agree these articles have been awesome. Someone, somewhere, who has some power regarding these issues is hopefully reading.

    I hope you are working on another series.

  8. Good read. But the problem isn’t that the new orgs can’t make money – they can’t keep up with the old health care business models

    via TweetBot

  9. Kip, your articles are awesome. I hope that you are leaving these comments on the MACRA proposed rule. I have been in an on going battle with our hospital about the “costs” of implementing a bundled care program, the billing rework, the workflow changes, the staff that needs to be hired. That side of the cost, the implementation, programming, on going costs, workflow changes, staff hires, etc. for ACO, AAPM, Homes, MACRA, etc, does not exist. That somehow all that is “free” or doesn’t count. Its ridiculous. When I tell non-medical people in real business about this, they cannot believe what I am telling them. They get a big look of fear in their eyes as they know that the system will break with this kind of treatment and regulation. I hope everyone that is willing, will pound the proposed rule with these issues, that they will send tweets to the tweet master Andy Slavitt, to let him know that he is making a HUGE miscalculation and mistakes with MACRA, etc.