These are unedited comments on proposed Federal regulation. They aren’t short and they aren’t ACO 101. 

Dear Administrator Slavitt,

Today Aledade, Inc partners with 111 primary care physician practices, FQHCs and RHCs in value-based health care. Spanning eight accountable care organizations in 11 states these primary care physicians are accountable for over 90,000 Medicare beneficiaries. We are committed to outcome based approaches to determine the value of health care. We are committed to using technology, data, practice transformation expertise and most importantly the relationship between a person and their primary care physician to improve the value of health care in the United States.

We believe this proposed rule addresses some of the most fundamental questions facing physician led ACOs and their place in the future of health care during and after the transition to alternative payment models. The move to regional benchmarking is the only sustainable course for a long-term financial model for population health. This is also an opportunity for CMS to right size two-sided risk in preparation for the transition to alternative payment models. Below is a summary of our positions on issues followed by an in-depth commentary of those positions. Thank you very much for your consideration as we move together through this exciting time in health care. Please feel free to follow up with me or Travis Broome (travis@aledade.com) if you or your staff have questions or would like to explore these positions further

Sincerely,

/s/

Farzad Mostashari, MD

CEO and Co-Founder, Aledade, Inc

 

Summary

Defining Value in Accountable Care

We believe ACO value is “did a person in the ACO get better care than if the person had not in the ACO,” a “difference in difference” approach. In today’s health care system, better care leads to lower costs. Financial models like the Medicare Shared Savings Program (MSSP) create a way to reward doctors for better care and the resulting lower costs. But better care is hard. Creating a financial model that rewards lower costs due to better care, but not due to avoiding care (stinting) is hard. Our comments all strive to get closer to measuring ACO value, but on a path that is good for physicians, good for patients and good for society. We recently published on value and sustainability of ACOs in the American Journal of Managed Care (http://www.ajmc.com/contributor/travis-broome/2016/03/creating-sustainability-in-accountable-care), which is also included as Attachment B.

Regional Benchmarking

We applaud CMS for recognizing in this rule that a policy of continually resetting historical benchmarks has a limited shelf life as improvements get harder and harder and lower costs get dearer and dearer over time. We also recognize that a pure regional benchmark creates such obvious financial winners and losers that it is not good for society or patients. Physicians above the regional benchmark would have no incentive to join the program and physicians below the regional benchmark would have little incentive to continually improve. This creates the need for balance. We agree with CMS that the balance is achieved by blending the two benchmark methodologies to create a path for long-term sustainability for an ACO.We encourage CMS to adopt this approach in Next Generation ACO as well and to make it available to 2012/2013 start MSSP ACOs. This blend requires continual improvement from the ACO, but at the same time rewards the ACO more and more for the true measure of value in a difference in difference approach.

Sustaining Successful, High Cost ACOs

After the first contract period all ACOs will fall into one of four buckets: above regional benchmark (or high cost) and successful; high cost and unsuccessful; below the regional benchmark (or low cost) and successful; low cost and unsuccessful. The inclusion of a regional benchmark benefits directly low cost ACOs. However, we believe there should also be an incentive for high cost and successful ACOs to continue in the MSSP. For this reason, we recommend that CMS include earned savings into the historical component of benchmark for the second and third contract.

Accurate Benchmarking for Population Health

We have advocated, along with many others, for the last two years that to measure real ACO value risk scoring must accurately reflect the measured population. The artificial cap imposed by CMS on risk scoring turns ACOs into mini-insurance companies. This in turn scares ACOs away from two-sided risk because they are no longer responsible for just population health, but also statistical anomalies. Nothing keeps ACOs out of two-sided risk more than the cap on risk scores. It is also the only area where CMS is not leading the accountable care movement, but falling behind commercial health plans.

The next important step CMS can take to measuring ACO value is to include regional inflation update factors instead of national inflation update factors in all contract years. CMS should seek to reward ACOs for the work they do in creating difference in difference ACO value not because they happen to be in a low cost or high cost area on any given year. CMS’s own analysis for this proposed rule shows that very few ACOs can individually impact their area’s cost curve. Every ACO can impact whether the person got better care in the ACO than they did out of it. A regional inflation update ensures that the work of the ACO impacts the ACO’s financial future instead of regional cost arbitrage.

Right Sizing Risk

While not directly addressed in the proposed rule, we believe CMS must consider today whether current MSSP Tracks work in the era of MACRA. Congress created the requirement for alternative payment models to have more than nominal financial risk. Congress could have said substantial financial risk or significant financial risk, but instead they said more than very small in amount. We translate this into motivating, but never ruinous. Enough to create concern, but not enough to instill fear. None of CMS’s current two-sided risk programs pass this test. All three (Track 2 and Track 3 of MSSP and the Next Generation ACO) create significant financial risk for a physician led ACO, potentially even ruinous financial risk. While across the program only a few will face such losses, the fear of being one of the few who are ruined is real and cannot be what Congress intended by the phrase “more than nominal.”

Our solution is for CMS to replace the current loss protection in the least risky Track 2 with loss protection that ensures no ACO will be liable for more the 15 percent of the Medicare revenues received by those who can share in the savings generated by the ACO. Two-sided risk coupled with this right-sized loss protection creates a concerning situation not a ruinous one. It motivates physician-led ACOs without making them fearful.

We support the inclusion of the optional fourth year as a mechanism to move ACOs to risk. CMS should consider whether it would benefit APM policies to have that year count as the 4th year of the current contract or the first year of a new risk-bearing contract.

Changes in ACO Composition Over Time

The benchmark must be reset using the same calculations and not proxies. ACOs live and die on single percentage points. If the proxy calculation is off by even a single percentage point an ACO could lose all their revenue if they dip below the MSR. Until CMS can clarify what the actual correlation of the proxy to current methodology is, no ACO can possibly support this change simply to alleviate the burden of calculation. CMS should not let expediency threaten the accuracy of the program.

Defining Value in Accountable Care

Value is a simple equation, benefit minus cost. Value is also in the eye of the beholder, defined by who gets the benefit and who bears the cost and in what proportions. In health care this is especially true because rarely is value determined in a two party transaction. Nearly every transaction has three parties: the health care provider, the person who receives health care services and the party paying for those services (society at large or a more limited group of people in the same insurance pool). The key to accountable care’s sustainability is to ensure that all three parties benefit as there can be no value without any benefit.

We believe the best definition of ACO value is “did a person in the ACO get better care than if the person had not in the ACO,” a “difference in difference” approach. In today’s health care system, better care leads to lower costs. This creates an opportunity for all three parties to see value creation. The person receiving the care gets more benefit in health and financially, the physician gets more benefit in satisfaction and financially and society gets more benefit financially. Higher quality leading to lower costs is a proven equation inside and outside of health care, accountable care gives us the vehicle to equitably share the value creation among the parties.

Since you have to start from where you are, the path towards the definition of ACO value is not necessarily measuring it directly in a classic difference in difference approach from day one. Our comments layout how to get from where we are today to a place where ACO value is measured most directly that is our view good for physicians and health care providers, good for people receiving their services and good for society.

Regional Benchmarking

ACO’s Regional Service Area

We agree that county is the best available geographic unit on which to base a regional benchmark. States simply vary too much in both geographic and population size to be used in a national program. CBSA, MSA, and CSA would generally be analogous to the counties that make them up as they are economically linked together by definition. There is one case where this may not be true and that brings us to our one recommendation to change the proposed ACO’s regional service area.

There maybe cases where an ACO covers such a high proportion of the residents of a given county that comparison population begins to suffer from small numbers. In those cases, we recommend including the contiguous counties in the ACO’s regional service area. The weight of those contiguous counties should reflect a weight necessary for the validity of the comparison.

Establishing the Benchmark Population for an ACO’s Service Area

In keeping to our definition of ACO value and a difference and difference approach, we must recommend that CMS exclude an ACO’s own population from the regional benchmark calculation. To include an ACO’s own assigned population would include historical benchmarking into regional benchmarking. CMS reported the median county penetration of ACOs as 12 percent. This means that 12 percent of the regional benchmark is actually historical benchmark if CMS does not remove an ACO’s assigned beneficiaries from the population. We appreciate the complexities of doing so, but we maintain that removing the ACO’s assigned population is imperative to an accurate measurement of the ACO’s value.

We fully support the following proposals:

  • Weighting the counties by the proportion of the ACO’s assigned population that resides in the county
  • Only including assignable beneficiaries in all ACO calculations

Determining County FFS Expenditures

We fully support CMS’s proposals for determining expenditures. We particularly note the importance of accurate risk adjustment between the two populations.

Applying Regional Expenditures to the ACO’s Rebased Benchmark

As a voluntary program, CMS should look at policies through the eyes of an individual ACO making a decision on whether to join or remain in the program. From a financial standpoint every ACO will fall into one of four groups at the end of their first contract.

ACO Status

We support the blending of the regional benchmarking into the ACO model as it creates the only truly sustainable path for accountable care and moves us closer to a difference in difference approach. However, we acknowledge that this is a financial disincentive for high cost ACOs. Of particular concern is the financial plight of individual ACOs who are successful at earning shared savings, but still above the regional benchmark at the end of the first three-year contract.

In considering the hundreds of different points on this graph that an individual ACO may find themselves we believe more flexibility is needed. First, hang on to good ACOs. We recommend that for the historical component of the benchmark continue to incorporate earned savings back into the benchmark as is done in the current rules. The adjustment for savings is a critical bridge for these individual ACOs as they work towards getting done to the regional average. The adjustment for savings would only apply to the historical portion of the rebased benchmark lowering its impact over time. Second, allow limited flexibility in the move to a blended benchmark over the first two contracts. We recommend the spectrum of that flexibility to be:

  • Year over year ramp up in the first contract (10% 1st yr, 20% 2nd yr, 35% 3rd yr) second contract is at 35% regional benchmark
  • Historical benchmark for the first contract leading to 35% regional benchmark in the second contract
  • Historical benchmark in the first contract with a year over year ramp up in the second contract (10% 1st yr, 20% 2nd yr, 35% 3rd yr)

This will increase the financial viability of individual ACOs while still moving all ACOs to the same place by the end of the second contract.

We support the move to 70 percent in the third contract period. If the true measure of ACO value is to be whether a person received better care at a lower cost in the ACO then they would have outside of the ACO then regional benchmarking must become the dominate measure. Moving only to 50 percent in the third contract would not be enough to move ACOs along in value.

While we understand that the Next Generation ACO is a model under the Innovation and therefore not subject to rulemaking, we take this opportunity to point out that the same arguments made here apply to Next Generation ACO. The arguments are arguably strengthening by the advanced two-sided risk aspect of the Next Generation ACO. We encourage CMS to adopt an in-contract ramp up of the inclusion of regional benchmarking in the Next Generation ACO model.

Updates to the Rebased Historical Benchmark

We support the move to regional update factors for resetting the benchmark. Furthermore, CMS should use regional inflation update factors instead of national inflation update factors in all contract years. CMS should seek to reward ACOs for the work they do in creating difference in difference ACO value not because they happen to be in a low cost or high cost area on any given year. CMS’s own analysis for the regional benchmarking rule shows that very few ACOs can individually impact their area’s cost curve with a median of 12 percent of the county. Every ACO can impact whether a person got better care in the ACO than they did out of it. A regional inflation update ensures that the work of the ACO impacts the ACO’s financial future instead of regional cost arbitrage. We disagree with CMS’s current view that having a higher hill to climb serves as motivation for the ACO in a high cost growth area to get even more savings. We believe that most ACO’s do all they can at any given moment and artificially raising the bar with inaccurate national updates is discouraging to ACOs not encouraging.

Updates Based on Assignable Beneficiaries

We support the proposed change to only include assignable beneficiaries in all calculations. This gets us closer to accurately measuring ACO value by comparing like beneficiaries to like beneficiaries.

Timing of Revised Rebasing and Updating Methodology

These updates should be made available to ACOs that started in 2012/2013. While we understand the complexities of rule making that appears retroactive, the ACOs that started in 2012/2013 are the pioneers of the MSSP. CMS should give them the option to enter a new 3-year contract at the start of 2017 that includes the regional benchmarking regulations that are finalized in this rule. It is only just that the pioneering ACOs be given the same opportunities as ACOs that came latter.

Risk Adjustment and Coding Intensity Adjustment

Because the same person cannot be both in the ACO and not in the ACO, risk adjustment is the most critical component of accurate measure of an ACO’s difference in difference value. While we understand CMS’s concern with coding intensity we continue to disagree that fear of the unknown justifies the obviously inaccurate comparisons and benchmarks created due to the artificial cap on continuously assigned beneficiaries. Any adjustments to coding intensity should reflect actual coding intensity and not perceived coding intensity. ACOs lack the mechanism of Medicare Advantage (MA) plans to submit codes independently. This will massively reduce coding intensity over what is done in MA because the primary tool is not available. Any residual coding intensity, if it is exists, should be accounted for by a measured adjustment not an arbitrary ceiling be it zero percent as in MSSP or even three percent as in Next Generation ACO. We therefore support the use of risk adjustment in regional benchmarking and encourage CMS to improve risk adjustment across the MSSP.

Adjusting Benchmarks for Changes in ACO Participant Composition

The benchmark must be reset using the same calculations and not proxies. ACOs live and die on single percentage points. If the proxy calculation is off by even a single percentage point an ACO could loss all their revenue if they dip below the MSR. Until CMS can clarify what the actual correlation of the proxy to current methodology is, no ACO can possibly support this change simply to alleviate the burden of calculation. CMS should not let expediency threaten the accuracy of the program.

In order for us to truly consider such a proxy, CMS would need to publish the findings of the modeling. CMS would also need to quantify the reduction in operational burden to CMS and the benefits to ACOs of that reduction.

Facilitating Transition to Performance-Based Risk

Regarding the proposal made, we agree with CMS that a transition year based on the original benchmark and one-sided risk would be beneficial to ACOs as they consider the move to two-sided risk. Only through the lenses of MACRA implementation do we see any difference between whether the transition year should be the 4th year of the first contract or the 1st year of the new, 4-year two-sided risk contract. Despite lack of details on APMs, we can foresee a situation were it would be beneficial for the transition year to be the first year of a new two-sided risk contract that could qualify as “more than nominal financial risk.” While no silver bullet (there is no single silver bullet) to moving ACOs to two-sided risk, it would be a step in the right direction as would floating risk adjustment, regional update factors and a new way of thinking about stop loss that we present below.

We would also like to take this opportunity to comment on why so few ACOs are moving to two-sided risk. Nearly everything about the future of risk in Medicare FFS hangs on the phrase “more than nominal financial risk.” One of the first questions we get is will your ACO qualify as an APM. Our answer is we are not sure. Current two-sided risk models expose primary care physicians to too much risk too quickly. As explained below, the stop loss happens after financial ruin for a physician led ACO rendering the provision nearly meaningless for a physician led ACO while serving as an effective financial safeguard for large hospital led ACOs. In addition, inflation and risk adjustment policies transfer to much “insurance” risk from Medicare to the ACO. In this paper, we make the case that the purpose of two-sided risk should be motivational for all ACOs and not be financially ruinous for any ACO. We believe this view of risk aligns perfectly with Congressional intent for “more than nominal financial risk.” Between the financial investments ACOs make in the work of population health and the 15% of Medicare revenues that represent our new, proposed stop-loss we believe it is clear that ACOs would be taking more than nominal financial risk under our proposal. We include a detailed proposal in Attachment A to our comments.

 

Attachment A: Framing the Purpose of Risk

There are two reasons for a payer like Medicare to desire that accountable care organizations participate in risk under a total cost of care models:

  • Recoupment of losses
  • Motivation

From the provider perspective there are only two reasons to participate in risk:

  • Only way to access the accountable care model
  • A greater financial reward if they achieve savings

In a voluntary program such as the MSSP, no ACO will remain in the program year over year if they are generating losses so the potential recoupment of losses is limited to at most two years. In addition to losses from of total cost of care, every unsuccessful ACO (one-sided or two-sided) experiences investment and opportunity costs. Estimates are all over the map, but at a minimum an ACO will be spending $500,000 a year. There is also the opportunity cost of seeing few patients to work on population health and participate in other initiatives. For example, ACO providers cannot achieve 2x in the Value Based Modifier and therefore potentially left a significant portion of their fee for service revenue behind. Therefore, we encourage CMS to view risk as a motivating factor to generating shared savings not as a mechanism to transfer money from providers to the CMS trust funds.

A focus on motivation raises a different set of questions that a focus on recoupment of losses. In recoupment of losses, it is simply a financial balance between greater shared savings rates and risk that must be fair to both sides. However, CMS must make this decision at the programmatic level while ACOs make the decision at an individual level and with their specific circumstances in mind. Essentially creating an adverse selection situation just like you have with individuals and the purchase of health insurance. With a focus on motivation, CMS goal is not financial balance, but rather how to get as many folks into two-sided risk as possible so they will be more motivated to achieve savings, and to reduce the number of unmotivated “ACO squatters” in non-performing one-sided risk models.

From the provider perspective, taking on the possibility of losses is taking a gamble. The providers will implement ACO best practices that should generate savings. They believe that if losses were to result instead it would be due to factors beyond their control. Hospital consolidation in their area drives up local health care inflation at three times the rate of national health care inflation as we saw in one of our ACOs. Or the benchmark year was a statistical fluke of a year for risk scoring and since risk scores can not rise they will always be fighting a headwind. Providers, rightly in our opinion, view losses as mostly beyond their control and savings as within their control. Because of this they focus on the worst case scenario. Like someone entering a casino only willing to loose $200, they want to know that the gamble won’t financially ruin them.

Yet in today’s MSSP program even the lowest stop loss of 5% losses in the first year of Track 2 could bring a physician led ACO and its providers to ruin in a single year. For example, let’s explore a 10,000 beneficiary ACO with a $10,000 benchmark or $100,000,000 in costs. The physicians in that ACO account for $4,000,000 of those costs. It is quite possible for those physicians to get a late start and generate a risk adjusted, regional adjusted savings of only 1% in the first year, but be hit with a 7% headwind due to the lack of risk adjustment and regional inflation. This ACO is now responsible 40% of $5,000,000 or $2,000,000. This is ½ of all the revenue they received in Medicare revenue. Practices are potentially out of business. The ACO is completely forgotten. This is a realistic scenario. Again the providers worry about the worst case scenario in year 3 when the stop loss is 10% and they have to pay $4,000,000 if they generate 10% losses wiping out their entire annual Medicare revenue despite having the top quality scores needed for the 40% shared losses rate. At the practice level an individual physician seeing 300 Medicare patients a year bringing in $120,000 from Medicare could individually owe Medicare under the ACO $60,000.

From a behavioral economics viewpoint, it is true that incentives and fear of losses can be motivating, but the literature indicates that excessive fear of losses actually reduces performance on complex tasks. Putting primary care physicians in a position of overwhelming losses also increases the probability that some small number of them will take excessive risks with patient care- stinting on sending a patient for needed emergency or acute care, or surgeries, and putting the public perception of the entire program and movement at risk.

You can see why an extra 10% in the shared savings rate is not enticing physician led ACOs to Track 2 in large numbers.

The Motivation Focus and Alternative Payment Models

Congress highlighted the motivation focus of risk with the phrase “more than nominal financial risk”. The risk needs to motivate, it needs to be enough to matter, it needs to be more than nominal, but more than nominal clearly does not translate into financial ruin.

We propose a new stop loss in MSSP that would make risk motivating for ACOs of all make ups while being financially ruinous to none. After all, what is more than nominal to one entity might be downright miniscule for another entity. To set one fixed dollar or fixed percentage of total cost of care standard for more than nominal financial risk for all entities is to give large organizations a pass on financial risk and to expose small organizations not to more than nominal financial risk, but massive financial risk. Rather than a stop loss of a percentage of total costs, we propose to replace a total cost of care stop loss with a stop loss on the amount an ACO will be liable to Medicare as a percentage of Medicare revenue received by the participants that make up the ACOs. MIPS can create a loss in Medicare revenue of 9% couple that with the 5% bonus for being in an APM and you have a MIPS risk of 14% of revenue. We propose a new stop loss set at 15% – 20% of the fee for service payments made by CMS to all of the participants in the ACO. This sets risk at levels that motivate a specific ACO rather than the current situation where a “fair” policy to recoup losses creates massive disparities in financial risk from ACO to ACO. In this new scenario, our individual physician is only having to write Medicare a stiff, but survivable $18,000 check. See Attachment 1 for a table breakdown where you can see physician-led ACOs are always at more risk than hospital-led ACOs, but this proposal levels the playing field.

 

Table of Proposed Stop Loss Provision

Best Case Scenario that Involves the Stop Loss

(6% losses in Year 1 Track 2 with a 5% Loss Rate)

Physician-Led ACO Individual Physician Hospital-Led ACO
Beneficiaries 10,000 300 10,000
Benchmark $10,000 $10,000 $10,000
Total Cost $100,000,000 $3,000,000 $100,000,000
Losses (6% ) $6,000,000 $180,000 $6,000,000
Current Best Case Total Cost of Care Stop Loss (5% – Year 1 Track 2) $5,000,000 $150,000 $5,000,000
Current Best Case Losses
(40% Shared Losses Rate)
$2,000,000 $60,000 $2,000,000
Proposed ACO Share Stop Loss
(15% of Medicare Revenue)
$600,000 $18,000 $9,000,000
Proposed Best Case Losses
(40% Shared Losses Rate)
$600,000 $18,000 $2,400,000
Medicare Revenue $4,000,000 $120,000 $60,000,000
Current % of Medicare Revenue Lost 50.00% 50.00% 3.33%
Proposed % of Medicare Revenue Lost 15.00% 15.00% 4.00%

 

 

 

Worst Case Scenario that Involves the Stop Loss in Track 2 –

(11% losses in Year 1 Track 2 with a 10% Stop Loss)

Physician-Led ACO Individual Physician Hospital-Led ACO
Beneficiaries 10,000 300 10,000
Benchmark $10,000 $10,000 $10,000
Total Cost $100,000,000 $3,000,000 $100,000,000
Losses (11% ) $11,000,000 $330,000 $11,000,000
Current Best Case Total Cost of Care Stop Loss (10% – Year 1 Track 2) $10,000,000 $300,000 $10,000,000
Current Best Case Losses
(40% Shared Losses Rate)
$4,000,000 $120,000 $4,000,000
Proposed ACO Share Stop Loss
(15% of Medicare Revenue)
$600,000 $18,000 $9,000,000
Proposed Best Case Losses
(40% Shared Losses Rate)
$600,000 $18,000 $4,400,000
Medicare Revenue $4,000,000 $120,000 $60,000,000
Current % of Medicare Revenue Lost 100.00% 100.00% 6.67%
Proposed % of Medicare Revenue Lost 15.00% 15.00% 7.33%

 

 

 

Attachment B: AJMC Article on ACO Value and Sustainability

Farzad Mostashari, Travis Broome

Accountable Care Organizations (ACOs) are the cornerstone of policymakers’ pledge to convert over half of all healthcare dollars to alternative payment models by 2018. The Centers for Medicaid and Medicare Services (CMS) has led the way by creating new models like the Medicare Shared Savings Program (MSSP) which rewards doctors for delivering better care and lowering the total cost of care (Disclosure: Aledade partners with primary care practices to create and operate ACOs). Under the MSSP, providers share in savings realized over a projected benchmark if they meet quality targets. A study byLeavitt Partners shows the number of Americans in ACOs currently at around 23 million will soar to over 100 million by 2020. But is “accountable care” merely a transitional pathway from totally unmanaged “Fee for Service” towards capitated payments and HMO-style managed care? Or will accountable care prove to be a third durable payment and delivery destination offering more choice than managed care and better value than fee for service, with softer incentives for both providers and patients.

I don’t believe that we know the answer to that question yet, and perhaps wisely, policymakers have been somewhat cagey about committing to one strategy or another. The Next Generation ACO program ingeniously created a pathway towards capitation or all-inclusive population-based payments as CMS refers to them, by allowing ACOs – with full immunity from what would normally be considered violations of anti-kickback regulations—to negotiate capitated Medicare payments to “affiliates” in essence using CMS as their claims administrator. Meanwhile, the Medicare Shared Savings Program (MSSP) also continues to evolve towards sustainability with continual improvements- none more significant than the recent proposed rule issued by CMS around how the “benchmark to beat” is to be calculated in the future. The comment period on this rule closes on March 28, and I assure you that this is no boring technical detail, it is in fact a fascinating example of the delicate balancing act of good policymaking that merits our attention and input.

The goal of good policy in a market economy is to align individual incentives with societal good. Successful policy creates an environment where public welfare is furthered not only by direct government action and grants, but also by private organizations acting in their own interests. In health care, the “volume to value” movement seeks to align the interests of healthcare providers with the societal triple aim of better care, better health and lower costs. But how should that “value” be measured and rewarded? How to establish the counterfactual “expected costs” as the benchmark for ACO doctors to beat?

The policy goal is to reward both improvement and attainment, without creating perverse incentives. If the ACOs started with a regional benchmark (“attainment”) then providers with higher than average benchmark have no incentive to participate, taking away the biggest societal gain, while providers who are already lower cost are paid more automatically without anything changing. So the program began by setting the first three-year contract’s benchmark by projecting forward from recent historical costs (“improvement”). Perhaps wisely, policymakers punted on the inherent and obvious flaw in this approach. The most common question asked by sophisticated investors and policy neophytes alike has been unanswered until now: “if the organization has to keep reducing costs below historical benchmark, won’t the returns dwindle to nothing? What happens in the long run?”

Continually resetting a purely historical benchmark gets harder and harder over time, and will lead the most successful ACOs to drop out of the program. Last year’s rules provided a temporizing stop-gap, replacing a “full ratchet” with a “partial ratchet” whereby the pace of adjustment was slowed but leaving the fundamental inexorable unsustainability unaddressed.

Regulators have struggled to strike a balance between maintaining incentives for private organizations to continue to seek profit in reducing cost, while also seeking to ensure that there continues to be value created for society and the taxpayer. In keeping with comments submitted by us and others, CMS is now proposing to gradually move away from a purely historical benchmark (beat your own past performance) to a regional benchmark (beat your neighbor’s performance). In this way, value created by ACOs is defined commonsensically as “did people in the ACO get better care than they would have if the ACO had not existed.”

We acknowledge that getting the transition right – from rewarding improvement to rewarding attainment– is devilishly difficult. Transition to a regional benchmark too fast, and ACOs that still have high costs will leave the program. Transition to a regional benchmark too slow, and ACOs with low costs will lose their financial viability. We support the proposed transition of 35 percent regional comparison in an ACO’s 2nd contract (years 4, 5, and 6) and 70 percent regional in its 3rd contract (years 7, 8, and 9).

But this is only part of the solution. With healthcare costs rising every year, the benchmark must also be projected forward. CMS currently uses national inflation for annual updates of the benchmark within a contract period, but this creates an imperfect view of the counterfactual (what would have occurred had the ACO not existed). Regional updates using county weighted, risk adjusted costs (as is done in Medicare Advantage) create more accurate benchmarks and therefore more accurate measure of ACO value. CMS is indeed proposing to switch to such regional inflation in future contracts but is adding complexity and reducing predictability by proposing to continue to use national inflation in the first contract (first three years). An ACO should be rewarded because a person in Dover, Delaware got better care in the ACO than a like population outside of the ACO. Regional updates do that, while national inflation updates dilute that difference.

The benchmark must also account for differences between people’s health in order to accurately compare one person’s costs to another. To be sure, this adjustment creates opportunities to shift money around in the health care system without creating value, a practice CMS calls “coding intensity” in the proposed rule. Currently, CMS using a blunt instrument to combat this practice. Simply put, they do not let risk scores to go up for the same population year over year. This is done without consideration of whether that population had a bad run of cancer or an unusually high number of unavoidable accidents. This prevents accurate comparison and therefore prevents accurate measurement of value. This transfer of insurance risk to ACOs is one of the biggest barriers to ACO sustainability over the long run, and increases the risk of providers dumping patients whose true risks are rising. People do in fact tend to get sicker over time, and pretending otherwise can only last so long.

2016 is shaping up to be the most pivotal year in health care policy in a long time, possibly the most pivotal year since 1965. CMS is actively listening and working with private sector partners to iterate and improve the alignment between what’s good for society, what’s good for patients, and what’s good for doctors entering these new payment models. All of us in health care, and particularly those of us who are population health, must make the most of every opportunity to inform the changes that will be happening this year, and will illuminate the path forward.