Four years of nation-wide testing by The Centers for Medicare and Medicaid Services (CMS) has now proven that the current shared savings payment models do not work effectively for low-cost Accountable Care Organizations (ACOs). High-cost ACOs have more room to improve and therefore more opportunity for savings.

For those organizations that have already developed efficient, low-cost care delivery systems, Shared Savings is retrogressive. In this model, inefficiency is rewarded and lower-cost systems are penalized. In fact, we have lost about 40 percent of the members of the first cohort of ACOs—the Pioneer program—due mostly to the payment system design. Even the recently announced Next Generation ACO Model, with its welcome improvements, does not address a number of significant fundamental issues that must be addressed in order to create a sustainable model.

Overview Of Our Model

After being challenged by members of CMS and others to design a better payment scheme, we studied the issues and are proposing a global risk-adjusted payment system that we will detail in this post. At a high level, CMS would pay ACOs a set amount to cover all health care costs for a population of enrollees. This would be a per-member, per-year payment with appropriate adjustments for risk and geography. ACOs would be expected to improve their performance over time while CMS compresses the variation in per-member, per-year payments being made to participating ACOs. While all ACOs would be expected to improve, the least efficient ACOs would be expected to make the greatest improvements.

Non-participants would remain on the Medicare fee-for-service payment system with progressive downward adjustments in fees. In essence, the model would reward provider systems on the path to improvement and reward those that are achieving the goals of the triple aim — lower cost, higher quality, and a better patient experience.

Issues With The Current CMS Model

The core fundamental issue is CMS reluctance to pay providers a competitive amount based on the entirety of care delivered rather than the history of payments. CMS payments in all of the ACO models remains tied to historical fee-for-service payments rather than the value of care delivered.

For example, if an ACO can perform a knee replacement and produce a good outcome, it should be paid a risk- and region-adjusted amount for the outcome delivered. Once risk and region are adjusted, the payment should be nearly the same across the country. Under this model, health care systems would simply compete on providing the best outcomes at the greatest efficiency.

In the current CMS models, if an ACO is producing good outcomes and has already developed coordinated care, it is not being rewarded. CMS continues to underpay the advanced ACO while actually paying the less-efficient competitor a higher rate. CMS payment for health services must become more market based and payments should be made for delivering outcomes.

Some may argue that advanced ACOs have already proven that they can deliver care at that low price. This is simply not true and this underpayment is covered by cost-shifting from other patients. What appears as “low cost” in CMS is actually just low total reimbursement and oftentimes an indicator of an advanced-practice model operating in a fee-for-service system.

CMS must develop a competitive global payment that is adjusted for risk and geography and is not linked to prior total reimbursement. Failure to acknowledge historical improvements and the true cost of delivery will eventually drive the best performers out of the market altogether.

Key Components Of A Successful Model

There are six key components of a global risk-adjusted payment system that need to be in place for all to succeed.

  1. Payment should be adjusted for risk, geography, and outcomes, such that it is market competitive wherever that care is delivered. This would set up a competitive market that would drive more efficient and higher-quality care.
  2. ACOs should be expected to put processes in place to improve the health of Medicare enrollees. This will produce savings for the organizations that keep their populations healthy and out of crisis. CMS should not risk-adjust those savings away from ACOs. This will require development of a flexible risk adjustment model that rewards health systems for improvements in care delivery that result in lower-risk scores.
  3. The system must provide flexibility for patient mobility and properly assign accountability to the system that manages the care in a given time period. In certain regions of the country, ACOs lose control over a sizable fraction of Medicare patients every year as they move seasonally. When a patient will be absent from the ACO’s geographic region for three months or more, we must have the ability to transfer that person’s enrollment and accountability to another provider, using appropriate geographic adjustments to the payment.
  4. Patient benefits must also be aligned. Benefits should be aligned using a mechanism such as reference pricing that allows for patient choice and patient responsibility. Payment for a total knee replacement, for instance, should be set at an appropriately adjusted price across markets and regions. Patients who choose to have an elective procedure outside of the ACO at a higher-priced facility should pay the price difference. Patients should also be given benefit enhancements to encourage participation in their health care choices.
  5. CMS will need to continue to act as insurer and provide secondary reinsurance in order to account for catastrophic events to ACOs. Small and mid-size ACOs could be bankrupted by caring for a small cancer cluster and a single rare blood disorder that costs $1 million to treat annually. Even the biggest ACOs need a cushion against events over which they have no control.
  6. Every ACO should be rewarded for efficient management of all patients, including outlier patients. We recommend rewarding the successful management of high-cost patients through an upside-only type mechanism. If we include all patients in the design it will encourage retention and management of outliers.

Risk-Adjusted Payment

Setting the initial per-member, per year (PMPY) payment will be critical for ACOs in the system. Ideally, the PMPY payment would be set around the current national average per-person adjusted for geography and risk of the population with Hierarchical Condition Categories. We may need a stepwise approach toward the ideal system. Initially, the PMPY payments could take into account the current regional average per-person cost, adjusted for risk. Then, in following years, the risk-adjusted PMPY base would be modified to keep in step with the economy (e.g. GDP plus 1 percent) and establish the year’s new PMPY target for all ACOs. Over time, the variation in payments across the country would diminish.

For those ACOs in the lowest-cost percentile, expected financial improvement would be less than expected financial improvement for higher-cost systems. An ACO with a PMPY cost of $8,000, for instance, might have a 0 percent expected improvement target while ACO’s costing $11,000 PMPY would be expected to improve by 1 percent and an ACO charging $16,000 PMPY might have a 1.8 percent improvement target. Ultimately, payment should only vary across the nation due to risk and region.

For patients, the biggest change would be the required selection of a primary care provider within their ACOs. The primary care physician will be the main point of contact for the patient and will manage the patient’s plan of care and work with the patient to establish a wellness approach. Typically a family practice or internal medicine doctor will take on this role. This relationship will be crucial for improving health of the patient while controlling costs.

During the transition to global payment, we will still need to reference a fee schedule. Since the role of the primary care provider is crucial in managing care, we should immediately establish a case-management fee similar to the recently introduced chronic disease management fees, but applicable to the entire attributed population. This will eventually be a part of the global PMPY payment but should be introduced in the fee schedule now in order to support current and future infrastructure development.

Population Health

Health risk and socioeconomic assessments would be part of the commitment to wellness and coordinated care to identify those at risk before they incur poor outcomes and expense. Wellness visits should not be a separate benefit, but designed into the benefits of all Medicare beneficiaries. Patients may require education on healthy cooking, stress management, or exercise, and providers could seamlessly integrate those elements into a plan of care. If socioeconomic barriers are identified—such as transportation issues or inability to safely store medicines—those those would be addressed as well.

Further work is needed to improve access to data necessary to improve care. As we transition to systems of care managing a population of patients, ACOs will need complete access to unblinded data held at the Health Insurance Marketplace and state-run Health Information Exchanges in addition to the data provided directly from CMS. Timely data is needed to reduce duplication of services, identify the best interventions in patient care, and to help coordinate overall across organizations and throughout a patient’s life.

In order to compare the effectiveness of health interventions across organizations, we will need a common set of quality metrics that drive all reporting. This is the only way we can provide transparency and comparison data to consumers in a market-driven system. Since we know that what we measure drives behavior—and improvement—these measures must be few and focused. In the absence of a national measurement set, state organizations such as those that participate in the Network for Regional Healthcare Improvement (NRHI) could provide comparative measurements for ACOs.

Patient Mobility And Reference Pricing

ACOs will not be working exclusively within their own boundaries, of course. Some percentage of patients will need specialty care or will elect to take certain procedures to other providers. They will get sick while traveling or go south for the winter and have a myocardial infarction. Therefore, CMS needs to create reference pricing and complete definitions of an episode of care that can transcend regional marketplaces.

For example, if the ACO provider reimbursement for a total knee replacement is $28,000 including pre-work, surgery, and four weeks of integrated physical therapy, that should be the patient’s benefit as well. Should the patient decide to seek that knee replacement outside of her ACO and there is a difference in cost or included therapies, the beneficiary would be responsible for the difference.

ACOs should provide bundled services for common episodes of care and publish outcome measures. ACOs could collaborate with CMS and private payers to establish universal criteria for care appropriateness and revisit those criteria as medical knowledge advances. Should a requested procedure—a liver transplant, for instance—not be appropriate given the patient’s comorbidities, the ACO would establish a care plan to help reduce a patient’s risk factors and help him qualify at a future date, if realistically attainable. The patient should earn additional credits against his deducible when he follows through on care plans established in these informed or shared-decision-making sessions. This care plan should, of course, be based on clinical best practices.


Recognizing that the incidence of illnesses in health care is not always predictable, CMS must offer reinsurance. ACOs cannot be expected to act as insurance companies and most will not have sufficient populations over which to spread risk. CMS should retain the risk and carve out those costs when setting the ACO total PMPY fees. Even though the ACO is not at risk for catastrophic loss, it should be eligible for financial rewards for effectively managing outlier cases. This will create incentives for ACOs to stay engaged in efficiently treating these cases.

For CMS beneficiaries whose claims are in excess of the reinsurance threshold, a national average Hierarchical Condition Category (HCC) cost per unit should be established. For example, the total of all medical costs over the CMS reinsurance threshold is $2.5 billion and the total HCC units associated with these beneficiaries is 250,000 units. This would mean that the population costs per HCC unit is $10,000. Let’s say an ACO has a beneficiary that incurs $125,000 and the reinsurance threshold is $100,000. The patient’s HCC score is 15. In comparing the ACO’s actual cost ($125,000) to expected costs of $150,000 (HCC score of 15 x $10,000 per unit) the ACO saves CMS $25,000 and should share in that savings.

Patient Management

CMS should also continue working towards having ACOs manage pharmaceutical costs. Pharmacy is an integral component of patient care. And management of this benefit has significant downstream effects on health outcomes and cost of care. Pharmacy information also provides valuable inputs to help identify gaps in care, compliance with medication plans, and identification of disease states.

The top performing ACOs should expect additional credits from CMS and increases in PMPY awards for performers in the top quartile of all ACOs. The ACOs should be in open competition with one another to provide the best patient outcomes at the lowest cost. Incentives can spur this kind of positive competition.

This will require CMS to develop new capabilities of support, but it could also reduce the administrative burden by moving away from a purely claims-based program to one that is focused on overall cost and health outcomes for beneficiary populations. CMS will need to improve real-time data sharing capability and work toward greater compatibility between legacy systems.

We hope that this draft of a global risk-adjusted payment system sparks debate and collaboration, and then significant movement on the part of CMS toward a first experiment. Providers such as Bellin-Thedacare HealthPartners and others are willing to step forward and experiment with these payment changes that ultimately support the very reason we chose this profession to begin with — improving the health of our patients.