A ‘Goldilocks’ Theorem of Shared Savings and ACOs

Mar 22, 2013, 9:00 AM, Posted by

Brendan Saloner, PhD, is a Robert Wood Johnson Foundation (RWJF) Health & Society Scholar in residence at the University of Pennsylvania and a senior fellow at the Leonard Davis Institute of Health Economics. This is the first in a series of essays, reprinted from the Leonard Davis Institute of Health Economics’ eMagazine, in which scholars who attended the recent AcademyHealth National Health Policy Conference reflect on the experience.


Like Goldilocks wandering through the house of the Three Bears, policy-makers in search of a health care payment model have found it difficult to settle on an option that is "just right."

Fee-for-service—paying doctors separately for each service they provide—leads to too much unnecessary and duplicative care (too hot!). Capitation—paying doctors a fixed fee for caring for patients—leads doctors to skimp on care and avoid costly populations (too cold!). A "just right" payment model should give providers incentives to provide all the clinically necessary care to patients while keeping costs low.

Shared savings models—allowing providers to keep a portion of the money they save caring for patients—have been touted as one method for aligning the incentives of providers and payers. Most prominently, shared savings is a central element of the Affordable Care Act's Accountable Care Organizations (ACOs).

An ACO is a network of providers that have agreed to accept a bundled payment for treating patient populations, and in return stand to gain incentive payments for meeting performance targets (or to lose money for missing targets). In the "happily ever after" version of ACOs, groups of providers will finally have a business case for coordinating patient medical records, reducing costly visits to the emergency room, and improving patient compliance with chronic disease therapies without leading to excessive procedures or gaps in care. Healthy patients, healthy profits.

But will it work?

That was the core question of the AcademyHealth National Health Policy Conference session on "New Models to Pay for Health Care," which offered serious reasons for optimism but also serious grounds for concern. Moderated by Harold Miller, MS, executive director of the Center for Healthcare Quality & Payment Reform and the former associate dean of Carnegie Mellon University, the panel's members were Susan Dade, MPA, the deputy director of Puget Sound Health Alliance; Robert Mechanic, MBA, a senior fellow at Brandeis University's Heller School of Social Policy and Management; and Karen Van Wagner, PhD, executive director of North Texas Specialty Physicians, an Independent Practice Association of 600 doctors.

Encouraging Changes in the Delivery System

Let's start with the optimist's case for shared savings. Although not widely recognized, the organizational infrastructure for creating ACOs has been greatly enlarged over the past few years. More physicians are organizing themselves into multispecialty practice groups, with linkages to primary care and to hospitals. These linkages have taken many forms, including shared data infrastructure and joint investments in care coordination through the use of clinical social workers. These help patients stay out of the hospital and stay on their complex medication regimens.

A total of 32 ACOs, each covering a minimum of 15,000 Medicare beneficiaries, now participate in the Centers for Medicare and Medicaid Services' Pioneer demonstration project, designed to test alternate payment arrangements in organizations with experience coordinating care across settings. Relatedly, there have been dramatic improvements in the number of providers that have met meaningful use requirements for electronic medical records, one of the main tools that will be necessary to enable networks of providers working together in shared savings programs.

These developments provide an encouraging sign that the delivery system is making the transition from the older model of solo practitioners working independently of specialty and hospital care, without adequate record-keeping. Panelist Van Wagner provided examples of some of the information and care coordination tools being used by physicians within her North Texas Specialty Physicians network.

Looming Organizational and Payment Challenges

Transforming the delivery system is a necessary, but not sufficient, condition, however. To make these models work, an elaborate contracting and oversight mechanism needs to be established.

The Puget Sound Health Alliance's Dade spoke of these challenges in the context of a multi-payer shared savings program that encourages primary care practices to reduce avoidable emergency department visits. While seemingly straightforward in its objective, the pilot experienced multiple challenges bringing together different health plans around common data collection and reporting goals. This led to delays in providing the primary care practices with real-time data that they could use to inform their quality improvement efforts, a major hindrance to achieving the pilot's targets.

Preliminary results from the Puget Sound initiative suggest that it has not yet produced significant reductions in emergency department visits overall (except, interestingly, for the populations enrolled in the two Medicaid managed care plans). It is possible that a longer implementation period and further tweaking will improve these results, but ongoing challenges around reporting and data requirements remain.

In addition to the basic data requirements, there are also problems around setting benchmarks for providers to meet. The simple and intuitive idea is that providers should receive savings relative to how they performed prior to the implementation of the initiative, since within-provider comparisons can better account for variations in how sick or costly certain patient populations are to treat.

However, Brandeis' Mechanic described the practical challenges of making these adjustments in the context of Medicare's shared savings program. He pointed out that some hospitals have high readmission rates for conditions like heart failure and chronic obstructive pulmonary disease. An ongoing debate between Medicare and hospitals pivots around how to risk adjust and establish targets between the baseline year and the year after the intervention. To take a simple example, if a hospital happens (by chance) to have a handful of very costly cases in the baseline year versus the post-intervention year, this can dramatically impact how little or how much it stands to gain in future incentive payments.

A Way Forward

The implementation challenges described above are daunting, but they will not necessarily need to be resolved in order for ACOs and shared savings programs to be successful in the future. The unfortunate reality is that current provider performance in certain areas, such as medication reconciliation for patients recently discharged from the hospital and chronic disease counseling, is so dysfunctional, that even incremental improvements can translate into large savings for payers and for health systems.

Ultimately, finding the "just right" solution will require some compromise between financial risks and rewards for providers and for payers. To entice lagging providers to participate, payers can consider various "sweeteners," such as upfront investments in information and record-keeping systems, patient registries, and subsidies for care coordinators. These are not direct monetary payments, but can help smooth the way to better patient care.

On the other hand, providers should be expected to take on some meaningful new financial risk for the opportunity to participate in shared savings programs, such as the prospect of losing payments for substantial increases in hospitalization rates.

Not too hot, not too cold: We may yet be able to have our porridge and eat it too.

Reprinted from the Leonard Davis Institute of Health Economics' eMagazine.

This commentary originally appeared on the RWJF Human Capital Blog. The views and opinions expressed here are those of the authors.