READ MORE…

READ MORE…
Toolkit
CONTENTS
PREFACE
4
PART 1: Overview and Key Principles of Accountable Care Organizations
7
Elliott Fisher, Mark McClellan
PART 2: Organization and Governance 15
29
Part 3 Appendix: Data and Health Care Analytics – Understanding Health Risk, Measuring Performance, and Assessing
Opportunities for Improvement 69 David Chin, Brett Hickman, Sandy Lutz, Craig McKnight, Timothy Ray,
Warren Skea, David Zielke
2.1
Organizational Models
2.2 Roles of Individual Providers Included in the ACO Model
2.3
Governance Models
2.4
Moving Forward
PART 3: Accountability for Performance
David Axene, John Bertko, Dan Dunn, Bela Gorman, Steve Lieberman,
Joachim Roski, Mark Zezza
3.1
Attributing Patients to an ACO
3.2
Developing a Budget
3.3
Payment Models and Incentives
3.4
Distribution of Shared Savings
3.5
Performance Measurement
Dan Dunn
PART 4: ACO Infrastructure
90
113
129
John Bertko, Paul Katz, Bob Power
4.1
Data Exchange
4.2
Tools for Timely Communication with Physicians and Patients
4.3
Tools for Tracking Performance and Costs
4.4
Assessing the Appropriate Level of Resources and Tools Needed for an ACO
PART 5: Health Care Delivery Transformations for Achieving
High-Value Health Care
Mark McClellan, Andrew Kruegar, Sue Podbielski
5.1
Care Coordination
5.2
Condition or Population Specific Interventions
5.3
Patient Engagement in Care
5.4
Infrastructure and Organizational Redesign
PART 6: Legal Issues for ACOs Katherine Funk, Chris Janney
6.1
Federal Antitrust Law
6.2
Federal Physician Self-Referral (or “Stark”) Law
6.3
Federal Health Care Program Anti-Kickback Law
6.4
Federal Services Reduction Civil Monetary Penalty Law
6.5
Federal Tax Law
6.6
Other Potential Legal Issues Copyright 2010 © The Brookings Institution
ACO TOOLKIT
PREFACE
Our nation’s health care system is at a critical
crossroads; unsustainable growth in health care
costs poses a serious threat to our national fiscal
integrity, as well as to individual’s and small
businesses’ abilities to receive affordable, highquality health care. The current fee-for-service
payment system undermines health care providers’
efforts to invest in real clinical transformation, while
incentivizing the provision of an ever increasing
volume of services. Payers are similarly concerned
that current attempts by health care providers to
better coordinate care and invest in new health
information technology (IT) infrastructure will
continue to add to already rapidly growing health
care costs. In an attempt to address this challenge, historic
national health care reform legislation was passed
in March of 2010. This legislation, among other
initiatives, includes the creation of a Medicare
Shared Savings program, to be implemented no
later than 2012, that allows for Accountable Care
Organizations (ACOs) to participate in Medicare. ACOs are being recognized as a promising new
payment model that could successfully start to
realign our current payment system to better reward
improvements in the efficiency of care delivery by
supporting health care providers’ efforts to improve
quality and bend the cost curve with the distribution
of shared savings. Over the past few years a number of organizations
committed to higher quality and lower costs have
come together ahead of the Medicare Shared
Savings program to try and form ACOs in the
commercial sector and through Medicare and
Medicaid pilot programs. The experiences of these
organizations, as well as the research of health
care policy experts from across the country, have
led to the development of one of the first attempts
to lay out succinct guidance for the successful
implementation of an ACO – the ACO Toolkit. The following work provides a path forward for the
implementation of ACOs across the country. By
attempting to walk a fine line between being both
specific enough to allow organizations to clearly
understand the steps needed to become an ACO,
and also staying broad enough to make sure the
path put forward for implementation is possible for
a diverse range of health care provider groups, this
toolkit is designed to be both helpful and applicable
to a wide array of health care organizations. There is no one organizational make-up that will
predetermine a successful ACO. Rather, we believe
success will be determined by leadership and
dedication to improving the quality and lowering
the cost of health care for a community and by
alignment and commitment to the guiding ACO
principles described in this work.
We want to acknowledge and express our deep
gratitude to John Bertko and and Steve Lieberman
for their many contributions to this project, without
which this toolkit would not have been possible. In addition to a laser focus on “making this
real” for practitioners, John and Steve provided
leadership, dedication, and creativity to the
substantive roadmap that became our collective
thinking about this toolkit. Additionally, we want
to express our special appreciation to John and
Steve for helping to recruit such a talented group of
experts, who, both through their own careers and
their efforts on this work, have been instrumental
in advancing more accountable care across the
country. David Chin, Brett Hickman, Sandy Lutz,
Craig McKnight, Timothy Ray, Warren Skea, and
David Zielke all made tremendous contributions
to help define the best organizational models and
governance structures needed for an ACO to be
successful. Similarly, David Axene, Dan Dunn, Bela
Gorman, Joachim Roski, and Mark Zezza all put in
tremendous work to clearly articulate the varying
technical features required for an ACO to be truly
accountable for its performance. ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 4
ACO TOOLKIT
Paul Katz and Bob Power have provided invaluable
insights on the type of analytical resources and tools
needed for an ACO to meet its goals, and Andrew
Krueger and Sue Podbielski made considerable
efforts to lay out opportunities for ACOs to truly
transform their clinical practice while simultaneously
reducing total costs. Finally, Katherine Funk and
Chris Janney helped us grapple with the necessary
legal issues health care providers need to consider
before beginning ACO implementation.
A number of other dedicated individuals spent
incalculable hours editing, researching, revising
and generally making this publication possible. Specifically, without the tireless efforts of Larry
Kocot, Sean McBride, Todd Wintner, Aaron
McKethan, Julie Lewis, Peter Basch, Tony
Hammond, Kristina Lowell, Niall Brennan, Nadia
Nguyen, Erin Weireter, Annalia Glenn, Emily
Carnahan, and April Choi, this toolkit would not
have come to fruition.
The health care landscape is rapidly changing and
ACOs are very much in their infancy. Therefore,
this toolkit will be continually updated and
supplemented over the coming months and years
to ensure it stays relevant and helpful to health care
providers as they face new challenges in advancing
more accountable care. Mark B. McClellan, MD, PhD
Director, Engelberg Center for Health Care Reform
Leonard D. Schaeffer Chair in Health Policy Studies
The Brookings Institution
Elliott S. Fisher, MD, MPH
Director, Center for Population Health
The Dartmouth Institute for Health Policy and
Clinical Practice
James W. Squires, MD, Professor
Dartmouth Medical School
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 5
Part 1
Overview and Key Principles of Accountable Care Organizations
Elliott Fisher, Mark McClellan
PART 1: OVERVIEW AND KEY PRINCIPLES OF ACCOUNTABLE CARE ORGANIZATIONS | ACO TOOLKIT
PART 1: OVERVIEW AND KEY PRINCIPLES OF
ACCOUNTABLE CARE ORGANIZATIONS
The Accountable Care Organization (ACO) Toolkit
is designed to serve as a reference guide for those
in the health care industry who are interested in
learning more about ACOs and how they can
prepare to participate – as provider or payer – in
moving toward implementation of an ACO. Most
ACOs will likely “learn by doing,” given that each
will face unique circumstances and challenges
when implementing this new model for fostering
greater accountability and improving value in the
health care system. At the same time, a common
set of technical and policy issues are likely to arise
throughout the implementation process.
This Toolkit provides an overview of the ACO model
and its role in the current policy context, highlights
the key technical and policy issues that will need
to be addressed by virtually all ACOs, and provides
guidance on how to overcome some of the key
challenges that will face those implementing ACOs.
This initial version of the guide will be updated and
revised on a regular basis through feedback from
provider and payer organizations that are moving
forward with implementation of ACOs.
Achieving Better Health and Lower Costs
The U.S. health care system faces unprecedented
challenges – rising costs that threaten the
affordability of care and fiscal outlook of the U.S.
government, and a widening gap between the
promise of biomedicine and the reality of care that
is often impersonal, inaccessible, unsafe, unreliable,
and poorly coordinated.
Proposals to address these problems typically focus
on only particular aspects of health care reform. For
example, some proposals provide more financial
support for certain health professionals because
evidence suggests that greater access will reduce
complications; others suggest paying more for
better processes of care because studies have
shown that “evidence-based” care can improve
outcomes or lower costs.
The ACO approach builds on these reform
efforts that often focus on specific groups of
providers, such as the medical home model, or
on a discrete episode of care, such as bundled
payments. On their own, these other initiatives
may help strengthen primary care and improve
care coordination, but they do not address
the problem of supply-driven cost growth. A
comparison of these payment reform models is
provided in Appendix 1. The table summarizes
the key differences between the ACO model and
other payment methods: medical home, bundled
payment, partial capitation, and full capitation.
Even though many of the other efforts are helpful
elements of health care reform, they are often
unlikely to achieve significant and lasting impacts
on quality and cost. If we want health care reform
to provide better care at a lower cost, we must
implement reforms that focus on this goal directly.
The core principle of accountable care is aligning
payments, benefits, and other health care policies
with measurable, meaningful progress in improving
health care while lowering costs. Exhibit 1.1
illustrates how accountable care addresses the
underlying causes of a poorly coordinated health
care delivery system.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 7
PART 1: OVERVIEW AND KEY PRINCIPLES OF ACCOUNTABLE CARE ORGANIZATIONS | ACO TOOLKIT
EXHIBIT 1.1. KEY PRINCIPLES OF ACCOUNTABLE CARE
Underlying Causes of Poor Performance
Principles of Accountable Care
Lack of clarity about aims, and about whose
perspectives are most relevant.
Clear aims: better overall health through higherquality care and lower costs with a focus on
patients.
Providers are fragmented and unable to coordinate
care well; providers accept responsibility only for
what they directly control.
Establish provider organizations accountable for
achieving better results for all of their patients at a
lower cost.
Payment system drives fragmentation, rewards
unnecessary care, and penalizes care coordination
and overall efficiency.
Align financial, regulatory, and professional
incentives with the aims of better health through
higher-quality care, lower costs.
Inadequate information to support provider and
patient confidence about the value of reforms.
Valid, meaningful performance measures that
support provider accountability for aims and support
informed and confident patient care choices.
Overview of the ACO Model
Accountable care is based on the principles of clear,
patient-focused aims (better overall health through
higher-quality care and lower costs for patients),
provider accountability through transparent
performance measures that reflect those aims, and
payment reforms that use the measures to align
provider support with the aims. Accordingly, ACOs
are provider organizations that are directly and
meaningfully focused on these aims. They are able
to monitor and report their performance in improving
health and lowering costs, and are supported by
financial and professional incentives that are aligned
with achieving better health and lower costs for their
patients. In summary, ACOs are:
• Collaborations of primary care professionals
•
•
and other health service providers, such as
other physicians and hospitals;
Organized around the capacity to improve
health outcomes and the quality of care
while slowing the growth in overall costs for
a population of patients cared for by a welldefined group of primary care professionals;
and
Capable of measuring improvement in
performance and receiving payments that
increase when such improvements occur.
Delivery Systems
ACO configurations can vary, reflecting the diversity
of local health care markets. As such, potential
provider configurations include – but are not limited
to – individual practice associations, physician
hospital organizations, and regional collaborations,
as well as integrated delivery systems. They may
involve non-traditional health providers such as
public health and wellness programs with different
payer participants. They also can feature various
payment incentives, ranging from “one-sided”
shared savings within a fee-for-service environment
to an array of limited or substantial capitation
arrangements with quality bonuses and “two-sided”
risk.
While a broad range of organizational models can
be ACOs, most health care organizations today
do not meet the core principle of accountability –
for health, quality, and costs of care over the full
continuum of patient care. Such accountability
requires a degree of integration or collaboration
among providers that is currently not often
found outside of integrated systems; however,
technological trends and economic and public
health pressures are changing this. As effective
medical practices, health information technology,
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 8
PART 1: OVERVIEW AND KEY PRINCIPLES OF ACCOUNTABLE CARE ORGANIZATIONS | ACO TOOLKIT
and performance measurement all continue to
advance, the potential for providers to take more
steps to achieve improvements in overall care and
receive better support in doing so also will continue
to increase.
• Legal Structure: ACOs must have a formal
•
Financial Incentives – Shared Savings
The ACO model establishes a spending benchmark
based on expected spending. If an ACO can
achieve quality targets while slowing spending
growth, it receives shared savings from the payers.
This model is well aligned with many existing
payment reforms, but also offers additional support
and accountability to provider organizations to
enable them to deliver more efficient, coordinated
care. This approach has been implemented in
programs like Medicare’s Physician Group Practice
(PGP) demonstration, which has shown significant
improvements in quality and savings for several
large group practices.
Because the groups receive a share of the
savings beyond a threshold level, steps like care
coordination services, wellness programs, and other
approaches that achieve better outcomes with
less overall resource use result in bonuses to the
providers who may be reducing up-front revenue.
These steps thus “pay off” and are sustainable in a
way that they are not under current reimbursement
systems. In addition, the shared savings approach
provides an incentive for ACOs to avoid expansions
of health care capacity that are an important
driver of both regional differences in spending
and variations in spending growth and that do not
improve health.
Key Design Features
Regardless of specific organizational form, the ACO
model has several key design features:
• Local Accountability: ACOs must aim to
be accountable to their patients and the
community they serve. They should also strive
to improve patient health and overall care and
reduce costs for their patients and community.
•
•
•
•
legal structure with a governing board
responsible for measuring and improving
performance.
Primary Care Focus: ACOs must be
established on a strong foundation of primary
care to impact care at the patient level. The
patient population for which an ACO is
accountable must be selected based on
patients’ use of outpatient evaluation and
management services, with primary care given
the highest priority.
Sufficient Size in Patient Populations: ACOs
must have a sufficient number of patients to
ensure that quality and cost impacts at the
patient level can be reliably benchmarked and
evaluated.
Investment in Delivery System
Improvements: ACOs must implement
meaningful and identifiable reforms in care
delivery, patient engagement, and other aspects
of health care that will credibly improve health
and costs.
Shared Savings: ACOs must offer a realistic
and achievable opportunity for providers to
share in the savings created from delivering
higher-value care. The incentive system must
reward providers for delivering efficient care as
opposed to the current volume-driven system.
Performance Measurement: ACOs
must participate in ongoing performance
measurement that provides meaningful
evidence of health and cost impacts. Results
must be transparent and accessible by patients.
With greater experience and further technical
progress, ACO care improvements are expected
to become more sophisticated. Examples include
more comprehensive care improvement activities,
better performance measures – such as more
meaningful outcome measures, including patient
experience measures – and payment systems and
other incentives that rely more on performance than
volume, intensity, or other factors unrelated – or
often inversely related – to value.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 9
PART 1: OVERVIEW AND KEY PRINCIPLES OF ACCOUNTABLE CARE ORGANIZATIONS | ACO TOOLKIT
Better Evidence to Help ACOs Succeed
Given that this is a new model, many activities
related to accountable care are still at an early
stage. The combination of care delivery and
payment reform is significantly different from
current practice outside of currently capitated
health systems and will require time and effort to
implement. Providing better evidence of “best
practices” to various types of health care providers,
communities, and patient populations will ensure
that accountable care achieves its goal as quickly
and efficiently as possible. Evaluation and learning
activities that enable better technical support for
ACO implementation are therefore essential to
support the deployment of ACOs.
Overview of the Toolkit
The process of setting up and implementing an
ACO will involve a multitude of technical, legal,
and analytic challenges. These include issues
such as the form and management of the ACO,
which specific providers to include and how any
shared savings will be distributed among them
and participating payers, and significant technical
and analytic challenges, such as the calculation
of spending benchmarks and the selection of
appropriate quality measures.
This Toolkit is meant to serve as a guide for ACOs
through these issues and the implementation
process. It aims to facilitate the process by
identifying the challenges that may arise along
the way and highlighting the key decisions and
functions that will be critical to the formation and
implementation of an ACO. This edition is Version
1.0; the Toolkit will be edited and improved based
on lessons learned by early adopters and by those
moving in this direction in the months and years
ahead. The other sections of this guide include:
Part 2. Organization and Governance
ACOs must have a formal legal structure with
a governing board to take steps to improve
care and measure performance. ACOs can be
developed from many of the existing organizational
models, such as Integrated Delivery Systems
(IDSs), Physician Hospital Organizations (PHOs),
Independent Practice Associations (IPAs),
Multispecialty Group Practices, and Regional
Collaboratives, with some models requiring more
organizational and operational changes than others.
This section discusses these models and includes
examples of the governance model and required
infrastructure.
Furthermore, this section discusses insights on
how ACOs can develop the appropriate blend of
providers with the right management structure to
redesign health care delivery in order to achieve
higher quality care at lower costs. This includes a
discussion on which providers are most likely to be
essential partners, such as primary care physicians
and certain specialists and institutions. It provides
examples of a management structure to provide
administrative and care coordination support, and
the resources to manage the operational functions.
Part 3. Accountability for Performance
ACOs are eligible for financial performance
bonuses by achieving measured quality targets and
demonstrating real reductions in overall spending
growth for a defined population of patients. In
this section, we outline the processes involved in
the ACO cost-of-care budget development and
the financial performance evaluation, including a
description of the patient attribution process used
to determine the ACO population upon which the
budget is based. Furthermore, we provide some
insights on how shared-savings may be distributed
within an ACO.
In an accountability-payment structure, the
financial benefits of achieving cost targets should
be contingent on quality metrics being met or
exceeded. As such, this section discusses the key
components and considerations for implementing a
quality measurement program, including selecting
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 10
PART 1: OVERVIEW AND KEY PRINCIPLES OF ACCOUNTABLE CARE ORGANIZATIONS | ACO TOOLKIT
a broad range of measures, collecting data,
developing a standard set of measures, developing
targets, calculating performance results, and
validating and reporting measures.
Part 4. ACO Infrastructure
For an ACO to successfully deliver the level of
integrated and efficient care that would allow
it to achieve shared savings, it must have the
appropriate information and resources to effectively
carry out care delivery functions and to improve
the quality of care. The ACO also must be able to
monitor progress, evaluate performance against
targets, and take appropriate actions to stay on
track. This section provides an overview of the
essential information and analytical resources
needed to achieve a level of clinical integration that
improves quality and reduces costs, focusing on
the following key elements: data warehouse and
data sources, using disease registries to provide
physicians and their care teams with meaningful
information, and examples of reports for tracking
financial and clinical performance. Factors to be
considered in assessing the appropriate level of
tools and resources for an ACO also are covered.
Part 5. Health Care Delivery Transformations for
Achieving High-Value Health Care
To help achieve the aims of better patient outcomes
and satisfaction at lower health care costs, ACOs
will need to explore opportunities for savings and
care delivery improvements. This section provides
examples of demonstrated savings opportunities
for ACOs through both public and private reform
efforts.
ACOs should consider both the short-term and
long-term opportunities to qualify for shared
savings. In the short term, ACOs should consider
interventions that can quickly generate savings
and return on investment, such as interventions
designed to reduce hospital readmissions
or relatively simple interventions that correct
inefficiencies in care delivery. ACOs should
also consider long-term investments, such as
interventions aimed at better managing chronic
disease. There are many shared savings
opportunities for ACOs. In this section, we discuss
a few examples from four key transformation areas:
1) care coordination; 2) population or condition
specific treatments; 3) patient engagement in care;
and 4) infrastructure and organizational redesign.
Furthermore, we discuss how ACOs should also
aim to layer reforms so that they build on each
other. Multi-faceted reforms that integrate different
silos of care, including inpatient services, outpatient
services, and patient self-management, have
the greatest chance of reducing spending while
improving quality. Lessons learned from successful ACO models will
expand this section dramatically in the future.
Section 6. Legal Issues for ACOs
An overview of legal considerations in setting up
an ACO is provided in this section. Federal level
issues include anti-trust issues, physician selfreferrals (Stark law), anti-kickback law, service
reduction, civil monetary penalty law, and tax
exemptions, while state issues focus on state antitrust laws, state fraud and abuse laws, false claims
acts, government-managed care regulations, and
corporate practice of medicine and state insurance
laws.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 11
Yes – Strong incentive
to coordinate and take
other steps to reduce
overall costs.
Yes – Significant
incentive to
coordinate among
participating
providers.
Fosters
coordination among
all participating
providers
Yes (for those
included in the
bundled payment)
– Depending on
how the payment
is structured, it
can improve care
coordination.
No – Specialists,
hospitals and other
providers are not
incentivized to
participate in care
coordination.
Strengthens primary
care directly or
indirectly
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE Yes – Strong incentive
to coordinate and
take other steps to
reduce overall costs.
Yes – It gives
providers “upfront”
payments and
changes the care
delivery model
for primary care
physicians.
Provides “upfront”
payments for
infrastructure and
process improvement
and makes providers
accountable for percapita costs. Requires
patient “lock-in.” May
be viewed as risky by
many providers.
By combining FFS
and prospective fixed
payment, it provides
“upfront” payments that
can be used to improve
infrastructure
and process, but
provides accountability
only for services/
providers. May be
viewed as risky by many
providers.
Yes/No – Only for
bundled payments
that result in greater
support for primary
care physicians.
Full Capitation
Partial Capitation
Yes – When primary care
services are included
in a partial capitation
model, it can allow
for infrastructure and
process improve-ment,
and a new model for care
delivery.
Yes – Provides
incentive to focus on
disease management.
Can be strengthened
by adding medical
home or partial
capitation payments
to primary care
physicians.
General strengths
or weaknesses
Promotes efficiency
and care coordination
within an episode,
but does not provide
accountability for
total per-capita costs.
Bundled Payments
Yes – Changes
care delivery model
for primary care
physicians, allowing
for better care
coordination and
disease management.
Supports new
efforts of primary
care physicians to
coordinate care, but
does not provide
accountability for
total per-capita costs.
Primary Care
Medical Home
Providers are
accountable for total
per-capita costs.
Does not require
patient “lock-in.”
Reinforced by other
reforms that promote
coordinated, lowercost care.
Accountable Care
Organization
(Shared Savings)
APPENDIX 1: COMPARING ACCOUNTABLE CARE ORGANIZATIONS WITH OTHER PAYMENT REFORMS
PART 1: OVERVIEW AND KEY PRINCIPLES OF ACCOUNTABLE CARE ORGANIZATIONS | ACO TOOLKIT
January 2010 | 12
No – Incentives are
not aligned across
providers. No global
accountability.
No – No risks for
providers who
continue to increase
volume and intensity.
Yes – In order to give
providers a PMPM
payment, patients
must be assigned.
Yes – Incentives are
based on value, not
volume.
Yes – In the form of
shared savings based
on total per-capita
costs.
Limited risk – While
there might be risksharing in some
models, the model
does not require
providers to take
risks.
No – Patients are
assigned based
on previous care
patterns. There
are incentives to
provide services
within participating
providers.
Fosters
accountability for
total per-capita
costs
Requires providers
to bear risks for
excess costs
Requires “lockin” of patients to
specific providers
No – There is no
incentive in the
medical home to
decrease volume.
Primary Care
Medical Home
Removes payment
incentives to
increase volume
Accountable Care
Organization
(Shared Savings)
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE Likely – Depending on
the model, patients might
need to be assigned to a
primary-care physician.
Yes – To the degree
that prospective fixed
payment is weighted in
overall payment.
Yes, within the
episode – Providers
are given a fixed
payment per episode
and bear the risk
of costs within the
episode being higher
than the payment.
No – Bundled
payments are for a
specific duration or
procedure and do not
require patient “lockin.”
Yes – Very strong
accountability for percapita cost.
Yes – Strong efficiency
incentive to the degree
that prospective fixed
payment is weighted in
overall payment.
No –for payments
outside the bundle.
No accountability for
total per-capita cost.
Yes – To calculate
appropriate
payments, patients
must be assigned.
Yes – Providers are
responsible for costs
that are greater than
the payment.
Yes – Very strong
efficiency incentive.
Yes – Strong efficiency
incentive to the degree
that prospective fixed
payment is weighted in
overall payment.
No – For payments
outside the bundle.
There are strong
incentives to increase
the number of
bundles and to shift
costs outside the
bundle.
Full Capitation
Partial Capitation
Bundled Payments
PART 1: OVERVIEW AND KEY PRINCIPLES OF ACCOUNTABLE CARE ORGANIZATIONS | ACO TOOLKIT
January 2010 | 13
Part 2
Organization and Governance
David Chin, Brett Hickman, Sandy Lutz, Craig McKnight,
Timothy Ray, Warren Skea, David Zielke
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
PART 2: ORGANIZATION AND GOVERNANCE
ACOs are likely to encompass multiple health
care providers, so organizational alignments will
be critical for ACOs to better manage care across
the full spectrum of services provided to their
patients. Momentum towards alignment already
exists. Hospitals and other health care providers
have frequently formed linkages to strengthen
their primary care capabilities, offer new delivery
models to communities, innovate care management
and delivery programs, recruit new providers, and
leverage resources more effectively.
There are many issues potential ACOs will need
to consider when assessing alignment options,
including who to partner with, how to navigate
legal and financial barriers, and how to manage the
venture. Achieving success as an ACO may require
health care organization executives to understand
the various models for better aligning medical
practice and how leaders can move an organization
toward the goal of providing better care that is more
integrated and coordinated.
As many ACOs are still in a formative stage,
definitive lessons on how new ACO organizations
can or should be structured are lacking. Drawing
on experience from other related efforts to better
align incentives and the organization of health care
providers, this section highlights potential ACO
organizational and governance models to help
identify core issues and potential solutions.
We first discuss the organizational features of the
care delivery system needed to ensure that the
right mix of providers is available to support an
ACO framework. That is, an ACO needs to have
access to an appropriate mix of key providers to
ensure that the full spectrum of services can be
delivered to their attributed patients, and, in turn,
be able to impact patients’ health and overall cost.
Additionally, providers may need to band together
to have enough patients to produce statistically
meaningful data on cost and quality. Meaningful
measurement of performance on cost and quality
is key to achieving accountability and necessary for
determining shared savings.
The discussion of organizational models presented
in section 2.1 relates integrally to the discussion
of provider involvement in ACOs presented in the
next section (2.2), and the governance models,
leadership, and decision-making presented in the
third section (2.3). The governance model can
often include entities that provide leadership for the
ACO – such as foundations or hospital-physician
joint venture corporations – but may not be
directly involved in delivering health care. Provider
involvement must focus on alignment between
any new shared-savings or payment reform
arrangements and steps that the ACO will take to
achieve measurable improvements in the delivery of
care.
We also note that sections on organizational and
governance models are intentionally high-level.
Building on this initial discussion, subsequent
sections will provide more operational detail on key
issues, such as assuring an appropriate flow-offunds or organizational capability for performance
measurement.
BASIC FUNCTIONS
While some functions of an ACO will vary depending
on the organizational and governance model, it is
expected that most, if not all, provider entities will
need to make arrangements for key capabilities
to succeed as an ACO. In many cases, additional
administrative processes will be needed to provide
the following basic functions and to facilitate
effective collaboration:
• Develop a patient care process that crosses
•
•
•
service settings;
Negotiate ACO payment models with payers;
Develop a methodology for shared savings
disbursement;
Enhance information technology and data
analysis infrastructure;
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 15
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
• Build capacity to continually learn and enhance
•
•
care processes;
Develop prospective budgets and resource
planning; and
Calculate performance metrics.
Health care organizations should seek arrangements
that will take advantage of each participant’s
resources to efficiently deliver the functions
necessary for operating an ACO. For example,
a physician group may be able to enhance its
health IT capabilities by aligning with a hospital
that already has an established, effective health IT
infrastructure. Hospitals may find similar synergies
in driving quality improvement through evidencebased medicine initiatives initiated by physician
leaders.
Given the new level of coordination required in both
patient management and administration to achieve
greater efficiency in care, prospective ACOs will
need to consider the various existing delivery care
models that can be adapted to meet their needs.
2.1: ORGANIZATIONAL MODELS
An ACO requires a level of organization and
governance that is rooted in formal arrangements,
such as contracts between providers who are
members and the ACO. In addition, an ACO must
have the capacity to improve care, with more than a
shared budget as the only element in common. The
diversity of local markets requires accommodating
multiple ACO models. Nonetheless, all ACO
models strive to redesign the delivery of health
care to achieve greater value in both quality of
outcomes and reduction in total costs. To achieve
this objective, an ACO requires an appropriate
blend of physicians, physician extenders, care
managers, and facilities working together to align
incentives, share patient information, and apply
evidence-based medicine protocols. The selected
approach should ensure greater appropriateness
and efficiency in the use of high-cost services and
improvement in maintaining beneficiary health.
ACOS will vary in the extent to which key providers
and facilities are members of the ACO or, for nonmembers, have formal contractual or informal
relations with the ACO.
Today’s market includes several types of
organizations that can support the types of
functions required for an ACO, including the
following:
Integrated Delivery Systems. Integrated
delivery systems (IDSs) typically involve one or
more hospitals and a large group of employed
physicians. In some cases, these health care
systems can also include an insurance plan,
even though they typically contract with multiple
health insurers. Eight of the 10 participants in
the Medicare Physician Group Practice (PGP)
shared-savings demonstration, for example, are
identified as belonging to an IDS (see Exhibit
2.1). These systems generally feature aligned
financial incentives, relatively advanced health
IT infrastructures, including the use of electronic
health records (EHRs), and have well-coordinated
team-based care. As a fully integrated model, IDSs
may already be capable of implementing some
of the more advanced payment models under
consideration for ACOs, such as accepting partial
capitation payments.
Multispecialty Group Practices. Multispecialty
group practices typically have strong physician
leadership through a committed group of
physicians who work closely with each other. They
usually do not own a health plan but contract
with multiple health plans. Most have highly
developed mechanisms for providing coordinated
care. Additionally, systems to coordinate care and
the related costs may be developed or arranged
through another partner. Two multispecialty
group practices, Marshfield Clinic and The Everett
Clinic, are participating in the Medicare PGP
demonstration.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 16
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
Physician-Hospital Organizations. PhysicianHospital Organizations (PHOs) are joint ventures
between one or more hospitals and at least one
physician entity. PHOs can vary from focusing
on contracting with payers to functioning like
multispecialty group practices but with an explicit
link to one or more hospitals. PHOs with the
latter characteristic are more likely to focus on
improving the delivery of care to enhance quality
and reduce costs. Many PHOs may require
stronger management and governance focused
on clinical integration and care management in
order to succeed as ACOs. The Tucson Medical
Center, a pilot site for the Dartmouth-Brookings
ACO Collaborative, is an example of a PHO
experimenting with the ACO model.1
Independent Practice Associations. Independent
practice associations (IPAs) consist of individual
physician practices that work together as a
corporation, partnership, professional corporation,
or foundation. In many instances, the primary
motivation for the formation of an IPA has been
for purposes of contracting with health plans in a
managed care setting. While financial risk in some
managed care contracts may be shared among IPA
members, the individual practices typically serve
non-HMO clients on a standalone basis, and bill
under their individual provider numbers. Stemming
from their experiences with capitation over the
past decades, many successful IPAs have already
evolved into more organized networks of practices
that are actively engaged in practice redesign,
quality improvement initiatives, and implementation
EXHIBIT 2.1. PHYSICIAN GROUP PRACTICE DEMONSTRATION PARTICIPANTS:
ORGANIZATIONAL CHARACTERISTICS
Part Of
Participant
Includes
Owns Or
Academic
Owned An
Medical Center?
HMO?
Yes
Yes
Organizational
Integrated
Structure
Delivery
System?
Dartmouth-Hitchcock Clinic
Faculty/Community
Group Practice
Not For
Number Of
Profit?
Providers
Yes
Yes
907
Billings Clinic
Group Practice
Yes
No
No
Yes
232
Geisinger Clinic
Group Practice
Yes
No
Yes
Yes
833
Middlesex Health System
Network Model
Yes
No
No
No
293
Marshfield Clinic
Group Practice
No
No
Yes
Yes
1039
Forsyth Medical Group
Group Practice
Yes
No
Yes
Yes
250
Park Nicollet Clinic
Group Practice
Yes
No
Yes
Yes
648
St. John’s Clinic
Group Practice
Yes
No
Yes
Yes
522
The Everett Clinic
Group Practice
No
No
No
No
250
Faculty Practice
Yes
Yes
Yes
Yes
1,291
University of Michigan Faculty
Group Practice
Note: HMO = Health Maintenance Organization
Source: RTI International, as cited in: Report to Congress. 2006. Physician Group Practice Demonstration: First Evaluation Report. <Available as of
July 21, 2010 at https://www.cms.gov/reports/downloads/Leavitt1.pdf>.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 17
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
MEDICARE DEMONSTRATION PROJECTS
While the Patient Protection and Affordable Care Acts (ACA) may have given an unprecedented push to
shared-savings models by incentivizing the creation of ACOs, federal initiatives to create shared-savings
projects are hardly new. Two ongoing Medicare demonstrations, the Physician Group Practice (PGP)
and Medicare Health Care Quality (MHCQ) demonstrations, already involve mechanisms for providers to
receive shared-savings bonus payments when their efforts slow health spending and meet quality targets.
PGP Demonstration. Launched in 2005, this five-year demonstration was designed to test whether care
management initiatives, when implemented under a shared-savings payment model, could generate cost
savings by reducing avoidable hospital admissions, readmissions, and emergency department visits, while
at the same time improving the quality of care for Medicare beneficiaries. The ten PGP sites tend to be
tightly integrated with numbers of participating providers that vary from just over 200 to more than 1,000
physicians.
MHCQ Demonstration. Established through Section 646 of the Medicare Prescription Drug,
Improvement, and Modernization Act (MMA) of 2003, MHCQ builds on the PGP Demonstration by testing
a similar payment and quality improvement model in multi-stakeholder organizations that include – but
are not limited to – physician groups. After several years of development, two organizations launched the
demonstration in January 2010: the Indiana Health Information Exchange and North Carolina Community
Care Networks, Inc.
of EHRs. Monarch Healthcare in California is an IPA
participating in the Dartmouth-Brookings ACO pilot
collaborative that is becoming an ACO. For other
IPAs, stronger leadership and clearer governance
structures may be required to better coordinate
information sharing, track costs, and manage care
across most IPA practices.
Regional Collaboratives. Another potential
model exists when independent or small
provider practices organize to become regional
collaboratives. This model may be particularly
useful for providers interested in forming an ACO
that are located in rural areas. Anti-trust and other
policy considerations might limit the scope of
regional approaches in many metropolitan areas.
Leadership in this model may come from a variety of
sources, including providers, medical foundations,
non-profit entities, or state government, operating
through its Medicaid agency or the Legislature. In
these cases, the impetus for forming an ACO may
come in conjunction with regional collaboratives
implementing initiatives such as health information
exchanges or public reporting, which can help
small practices share information and better
coordinate care. The Medicare Health Care Quality
Demonstration (MHCQ) includes two sites that can
be considered a regional collaborative.
2.2: ROLES OF INDIVIDUAL PROVIDERS INCLUDED IN
THE ACO MODEL
ACO success depends on a variety of providers
working more effectively together as a system to
provide appropriate, coordinated care for ACO
patients. Different providers may play various
roles that involve directly providing services to
patients, as well as ACO governance and operation.
The ACO governance roles and responsibilities
are explored in more detail in the next section,
but generally include legal, fiscal, and clinical
responsibilities. Not all providers serving ACO
beneficiaries need to be members of the ACO or
involved in its governance. In fact, some ACO
members – which could include employed or
contracted providers – may not participate in shared
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 18
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
savings (or losses). However, providers who have
patients attributed to them for the purpose of
defining the ACO population can only participate in
a single ACO.
ACOs also do not change the underlying
characteristics of a patient’s insurance coverage.
For example, if a patient is enrolled in a preferred
provider organization (PPO), the ACO does not
impose a “gatekeeper” or a closed provider
network, over and above the normal requirements
of the insurance benefit. Patients attributed to an
ACO will not be limited to seeking care from the
participating providers of the ACO. ACOs remain
accountable for the total cost and quality of care for
their attributed patients, no matter where or what
services are provided. This responsibility includes
care delivered across the full spectrum of services
required for patient’s health care needs. Inevitably,
some of this care will be delivered by non-ACO
providers.
In this section, we discuss the various roles
that ACO providers must fulfill, including which
providers are most likely to be a part of the ACO
governance team, an exclusive ACO provider used
for assignment, and/or eligible for shared savings.
We also touch upon the potential relationships with
providers who are not contractual members of the
ACO.
The specific configuration for an ACO is expected to
vary depending on the local practice environment,
patient needs, and infrastructure required to
support providers for services such as IT, analysis,
performance improvement, and finance. Exhibit
2.2 summarizes the range of possible provider
arrangements.
PROVIDERS TO WHOM PATIENTS ARE ASSIGNED
As a core principle, ACOs must rely on primary care
as a foundation for achieving improved care at lower
cost. Evaluation and measurement (E&M) services
provided by primary care providers (PCPs) are the
preferred basis for assigning patients to ACOs.2
As such, core ACO providers almost certainly will
include primary care providers such as internal
medicine, family practice, and pediatric providers.3
They may also include specialists who are likely to
have frequent contact with patients, particularly for
serious chronic diseases, and handle many primary
care services. In addition to being members exclusively of a
single ACO (as discussed in more detail in Part
3), these ACO members would have central
responsibility for the overall cost and quality of
care for their attributed patients. Patients often
regard these providers as their personal physician,
and the physicians generally provide the entry or
coordination point for referring patients to more
specialized treatment on a non-emergent basis.
The pattern of referrals from personal physicians
is likely to have a critical impact on the overall
cost and quality of care received by ACO patients.
These “high touch” providers ideally develop longterm relationships with their patients through a
combination of regularly scheduled and episodic
visits.
Core providers may encounter an increased
volume of patient contacts if they assume a larger
role in overall care coordination. The widespread
consensus that primary care is generally
undercompensated may help increase likelihood
that PCPs and other core providers will participate
in bonus arrangements – either shared savings or
losses – as members of an ACO.4
Specialists Likely To Have an Ongoing Relationship
with Patients
Many providers other than primary care providers
also play critical roles in delivering higher-quality
and lower-cost care to patients. This category of
providers includes those who manage or perform
mainly episodic events but with significant resource
and quality implications. This group may include
physicians in specialties such as cardiology,
oncology, urology, neurology, or gastroenterology.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 19
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
Each ACO will need to determine prospectively
whether these providers are members and, if
so, whether they participate in its governance
structure, shared savings, or other programs.
For example, these providers might view having
preferred referral status or participating in bundled
payment arrangements, and accompanying further
support to improve patient care, as sufficiently
attractive without being eligible for shared savings.
Specialists who are members of an ACO may have
a role in governance without participating in the
shared-savings program.
characteristics. While hospitals are expected to
be a key provider for all populations, other entities
might be more important in delivering care to
specific ACO populations, such as Medicare or
Medicaid patients. For example, skilled nursing
facilities and home health agencies may play a more
important role in serving Medicare patients than
commercial patients. Discussions of governance
and shared-savings participation should identify
clear opportunities for collaboration to improve
quality and lower costs.
Non-Contracted Providers
Other Specialists
Other specialists that typically do not have an
ongoing relationship with patients may be members
of an ACO, be members of multiple ACOs, or simply
not be members of any ACO, but provide services
to ACO patients. Examples of such specialties
include anesthesiology, radiology, and emergency
medicine. In many cases, these providers are less
involved in the coordination of care. For greater
participation in shared savings, individual ACOs
would need to evaluate the circumstances to make
a well-informed business decision.
Facilities
Hospitals play a major role in health care delivery,
as they provide the most intensive and highest
cost care. Their potential inclusion in an ACO
should be carefully considered based on several
factors, including local market conditions,
alignment of interests, leadership in creating an
ACO, capital, organizational resources, and patient
As mentioned above, ACOs are financially
accountable for all the care delivered to their
attributed patients, whether delivered by ACO
members, contracted providers, or non-contracted
providers.
Being assigned to an ACO does not change a
patient’s benefits or access to providers. Therefore,
unless ACO patients are enrolled in health plans
with prior authorization by PCPs or closed networks
– such as HMOs – ACO patients will have some
care provided by non-ACO providers. These
patients are not locked in to any providers and can
receive care from any provider within the insurer
network without prior authorization or extra cost.
There may also be other reasons to seek treatment
from non-contracted ACO providers. For example,
ACO patients may seek care from out-of-area
providers while traveling. The ACO may have no or
very limited ability to influence care from these nonACO providers.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 20
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE providers
Non-contracted
Facilities
Other specialists
with patients
ongoing relationship
likely to have an
Specialists more
ACO)
if unique to a single
(assigned patients
Core providers
Role in ACO
January 2010 | 21
work with the ACO providers.
network but who do not routinely
provider) or providers in the payer’s
payer’s network (e.g., an out-of-area
No
No
governance
facilities.
These can be providers out-of-the
be involved in ACO
attributed to
Facilities, and Home Health Agencies
If an ACO member,
a hospital is likely to
No, Patients
are not
Not Likely
Possible
Governance
to Participate in
providers most likely
These are the
Governance Role
No
Not likely
Yes
Assigned
Patients
e.g., Hospitals, Skilled Nursing
Emergency medicine
e.g., Anesthesiology, Radiology, and
Oncology, Urology, Gastroenterology
Orthopedics, Otolaryngology,
e.g., General Surgery, Hospitalists,
Gynecology
Pulmonary, Obstetrics and
(non-interventional), Pediatrics,
Family Practice, Cardiology
e.g., InternaL Medicine and
Provider Types
Non-member
Possible member
Possible member
Member
Member
ACO Member
or nonMember
EXHIBIT 2.2. TYPES OF PROVIDERS; ROLES AND RESPONSIBILITIES
No
individual ACOs.
circumstances of
on specific
Depends
No
bonuses
shared-savings
participate in
Possible to
in shared-savings.
likely to be included
providers most
These are the
Shared Savings
No
Yes
Yes
participation
factor determining ACO
expected to be key
Other factors are
Yes
Other Incentives
(e.g., Bundled
payments, health
IT, Administrative
Assistance,
preferential referrals,
quality bonuses)
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
Organizations with Services in Multiple Categories
There are many organizations or health systems
that include more than primary care providers and
offer services reflecting some or all of the categories
discussed above. For example, multispecialty
group practices will often include primary care
as well as specialist physicians. To better align
incentives and minimize administrative complexity,
ACO arrangements with such organizations may
prefer to include the full scope of services offered by
a group to avoid splitting an existing organization.
One model for creating an ACO would incorporate
an entire health care system, making it the basis
for the ACO. For example, where an enterprise
encompasses both hospitals and physicians,
including PCPs, these providers could form an ACO.
If sufficiently extensive, the enterprise’s network
could be the only providers affiliated with the ACO.
2.3: GOVERNANCE MODELS
The success of an ACO may require substantial
cultural changes that complement changes in
operations. For example, to succeed under the
shared-savings model, the business practice
of maximizing revenue will need to shift toward
a greater focus on efficiency and eliminating
unnecessary care. Also, unlike the traditional
payer–payee relationship, which gives providers and
plans little reason to cooperate, the ACO sharedsavings framework rewards coordination and
alignment, and may offer the potential to improve
relations. Achieving cultural and operational
changes will require strong ACO leadership.
Identifying the right leaders is a critical step for
enabling an aspiring ACO to succeed. Successful
ACOs will require strong, capable clinical, fiscal,
and administrative operations leaders. Candidates
drawn from existing health system or provider
participants may provide a strong group of ACO
leaders. Based on their assessment, those leaders
may choose to create a free-standing administrative
structure to manage and implement the ACO.
Alternatively, ACOs may recruit outside candidates
as leaders, or choose to implement essential
functions by outsourcing services to vendors with
proven capability.
ROLES AND RESPONSIBILITIES
The functions of an ACO can generally be divided
into two categories – management (administrative)
and clinical – which must be implemented together.
Management
Management roles within an ACO include oversight
for all strategic and high-level operational issues.
Responsibilities may include:
• Oversight of interaction with payers;
• Oversight of coordination with ACO participant
and partners;
• Review of data and financial trends; and
• Setting policies for resource distribution
including setting criteria of the shared savings
and its allocation methodology.
Clinical Roles
The overall clinical role of an ACO is to ensure that
the best care is provided in the most economical
setting for ACO patients. Clinical responsibilities
may include:
• Establishment of care coordination guidelines;
• Review and dissemination of best practice
•
•
information, including evidence-based
guidelines;
Practice improvement, including identification
of providers with outlier practice patterns and
strategies for clinical improvement; and
Handling clinical issues involving ACO patients.
Both roles must be coordinated; for example,
performance measures used in ACO financial
arrangements should be integrated with clinical
strategies to achieve improvements on these
measures.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 22
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
SAMPLE GOVERNANCE MODELS
How ACO management and clinical leadership roles
are assigned and organized depends ultimately on
the type of organizational model – whether IDS,
PHO, multispecialty group practice, IPA, or regional
collaborative – as well as the specific infrastructure
and market characteristics in play. For example, an
IPA model in a rural area with limited infrastructure
may look different from a similarly conceived model
among suburban physician groups.
Below, we describe two models that illustrate the
variety of ACO governance models for entities
that include both physicians and hospitals. The
Foundation Model might best suit more integrated
health systems, while the Clinical Co-Management
Model might better suit separate entities interested
in forming a virtual organization. These models
are by no means exhaustive. Rather than being
blueprints, these examples highlight how the degree
of integration alters governance considerations, and
how governance must promote financial and clinical
alignment. The discussion draws on how parties
have handled related issues in the past.
Foundation Model
The first approach relates to organizational systems
that already exhibit some degree of integrated
care delivery. This “foundation model” typically
addresses clinical and management issues through
two entities:
1. The Physician Group. The physician group
is fully self-managed and wholly owned by
physicians. Physicians in medical groups may
be partners or employees.
2. The Medical Foundation. The foundation is
usually established as a tax-exempt 501(c)(3)
non-profit public benefit corporation.
The foundation contracts with the physician
group(s) through a Professional Services Agreement
“PSA” specifying that the physician group provides
professional services to the Foundation’s patients.
The foundation typically owns and operates the
facilities, equipment, and supplies of a practice
employing non-physician personnel.
The foundation can serve as a vehicle to make
hospital capital available to the group of physicians
to expand clinical services, staff, or operations. The
infusion of resources can also provide administrative
systems that might not be available to the medical
practices separately, as well as comprehensive
patient services to help manage care and assume
financial risk. The availability of tax-exempt status
also brings the foundation access to less expensive
capital and exemption from sales and income taxes.
This approach is typically used by an IDS as
a strategy to approximate direct physician
employment without actually having to directly
employ the physicians. It is particularly relevant
in states such as California, where strict corporate
practice of medicine laws restrict employment of
physicians.
Establishing a foundation can be legally
complicated, time consuming, and costly. As an
organizational and governance strategy, these
factors make it less attractive for smaller and
financially weaker health care entities. There can
also be requirements on the minimum number of
physicians contracted through the foundation, as
well as a requirement to conduct medical research
and health education. The effective barrier posed
by these requirements may preclude using a
foundation model in rural and other areas.
Clinical Co-Management Models
Another model that may be more widely applicable
is the clinical co-management model, which strives
to align physician-hospital interests. This approach
allows physicians to align with a hospital yet retain
their independent practices.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 23
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
In the past, “co-management” companies have
tended to focus on specific service lines, such
as cardiovascular “centers of excellence” or
orthopedic “joint centers.” Within this model,
participating providers buy shares in a separate
management company, usually a Limited Liability
Corporation (LLC), incorporated in the state where
the physicians and hospital are licensed.
provide oversight and specific services. The
company has the authority to negotiate and
disburse financial incentives to the physician
investors and hospitals. Often, a third party
consultant facilitates the co-management model
to ensure timeliness and success. This includes
facilitating the structure, governance, incentives,
and management services.
The co-management organizational structure
enables physicians to hold leadership positions
that are designed to influence and improve day-today clinical management and operations. Equity
interests can be equal among the parties, or one
party can have a controlling interest. Independent
of how ownership is allocated, special rules and
procedures can be established to protect certain
rights, such as the number of board seats allocated
to a hospital that supplies the initial capital.
Under the traditional model, the hospital
typically retains all legal, regulatory, and fiduciary
responsibilities and requirements that it has under
its license to operate. When adapting this model
to an ACO, these responsibilities could be held
by either the hospital or the physicians. The LLC
provides a flexible framework that can be tailored
to meet the clinical, operational, and other needs of
the ACO participants.
A co-management company under this LLC
structure may frequently contract with the hospital
through a management services agreement to
Under this model, the LLC would be governed by a
board of directors expected to include both hospital
and physician representatives. Exhibit 2.3 provides
an illustrative example of a co-management model.
EXHIBIT 2.3. POTENTIAL ACO CO-MANAGEMENT MODEL
Hospital
Example:
Service Contract
to Manage
Cardiovascular
Services
XX%
Physicians
XX%
Management
Company
LLC
Management
Contract
$
Equity Return
Class A
Physicians
XX%
Class B
Hospital
XX%
Cardiovascular
Services
Management
Company
Compensation
¥ Base management fees
¥ Expense reimbursement
¥ Incentive compensation,
subject to limitations
including:
Quality
New Program Dev
Operational
Efficiency
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 24
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
Co-management structures tend to be very flexible,
in part because it is relatively inexpensive to
amend LLC agreements as conditions change. If
necessary, LLC arrangements can be undone. The
flexibility of the clinical co-management model helps
make it an attractive approach where hospitals and
physicians want to build a sustainable partnership
and effective ACO.
allow sufficient time to put in place the necessary
infrastructure and processes, many of which are
described in later sections of this toolkit. Aside
from all of the internal administrative and clinical
reorganization that an emerging ACO may require,
there are also many important decisions and
processes where leaders will need to work with
external partners. Foremost are the agreements
THE BENEFITS OF LIMITED LIABILITY PARTNERSHIPS
The Limited Liability Corporations (LLCs) have several advantages as a structure for some ACOs. This
structure carries potential tax benefits to the hospitals and physicians; limits personal risk and liability; and
enables funds to be distributed to the investors (hospital and physicians). Moreover, the LLC structure is
easy to legally set up and register, requiring little capital investment from individual investors.
POTENTIAL GOVERNANCE STRUCTURE DIAGRAM UNDER AN LLC
LLC Board of Directors
Sub-Committee #1
(if necessary)
Sub-Committee #2
(if necessary)
Ad Hoc Committee(s)
(if necessary)
As a joint venture LLC, governance can be customized to meet the specific needs of each ACO. The
figure above provides one illustrative example. A board of directors can be elected to oversee
development of strategic, financial methodologies that include payment terms and risk sharing formulae,
and to oversee distributions to its participating providers. The board of directors can also coordinate the
implementation of the critical clinical operational changes required for the ACO. Meanwhile, specific tasks
that are time-intensive can be assigned to appropriate standing sub-committees or to an ad hoc task force
for recommendations. The recommendations can then be presented back to the LLC Board of Directors
for approval and implementation.
2.4: MOVING FORWARD
FIRST STEPS IN SETTING UP AN ACO
A key consideration in setting up an appropriate
ACO organization and governance is developing
a clear and executable timeline. Organizations
interested in forming an ACO should plan to
that will need to occur with payers, which will
include negotiations on the parameters of the
shared-savings contracts, as well as determining
how data will be shared.
As an immediate first step, aspiring ACOs should
take stock of their internal capacities to successfully
implement the ACO initiative. This includes
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 25
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
identifying the key leaders of the initiative, as well
as determining which services can be effectively
and efficiently provided internally versus being
outsourced. As part of this process, the leaders
will need to identify partnering payers and test
the willingness of potential provider groups to
participate. The ACO providers will need to engage
payers at an early stage for the ACO to succeed.
Suggesting specific timeframes for ACO
implementation is difficult, as there is still no single
roadmap to becoming an ACO, and much needs to
be learned about their development and successful
implementation. Nevertheless, one way for an
ACO to set up its operations relatively quickly –
particularly if it is not already a quite sophisticated,
well-developed system – is to rely initially on
contractual or leased services from entities with a
proven track record of assisting providers achieve
efficiency gains. Services can range from those
provided by management service organizations
(MSOs) or other similar organizations that serve
physician-hospital entities to more narrowly focused
data reporting and analysis vendors.5
For example, Marshfield Clinic is a large
multispecialty group practice that is participating
in the PGP demonstration. It is supported by an
MSO and hospital partner. The MSO provides
quality improvement, medical management, public
reporting, contracting, and information management
services to the multiple practices it has participating
in the demonstration.6 Part 4 touches upon data
reporting and analysis services and the potential for
outsourcing such services.
Phased Implementation
ACOs will vary in their degree of integration,
scope of services provided by ACO members,
sophistication of care coordination and data
analytics, and ability to bear risk. Consequently,
successful organizational and governance models
will both differ and likely evolve over time. An
advanced ACO capable of successfully assuming
two-sided risk with partial capitation, for example,
requires a far more sophisticated and proven
infrastructure than an ACO operating under a
simple, one-sided shared-savings model based
on fee-for-service payments. In general, new
ACOs should take a phased approach to assuming
more risk and more demanding ACO levels,
progressing based on their level of experience
and demonstrated capacity. This is particularly the
case if providers have not been members of a long
established and highly functioning, systems of care.
An organization need not reach the stage of full
integration – an IDS – to be a successful ACO.
In many cases, the optimal level of integration
may be other than that displayed in current IDSs.
To determine what level of integration is most
appropriate for an organization, participants will
need to examine their tolerance for financial and
insurance risk as well as other infrastructure and
patient population considerations that are explored
more fully later in this toolkit.
CONCLUSION
The emergence of ACOs will largely depend on a
mix of leadership and financial, cultural, and other
considerations. The financial benefits must be
significant enough to allow for the development of
the required organizational entity; a successful ACO
requires implementation of necessary patient care
and administrative re-engineering. The specific
characteristics and preferences of providers and
patients in a particular local market will increase
the challenges associated with implementing
a successful ACO, a process which starts with
identifying key leaders, selecting an appropriate
ACO model, and developing required organizational
and governance features. Like any business, the
distribution of financial benefits must be skillfully
handled and properly aligned with responsibilities to
avoid becoming highly contentious.
ACOs are a classic case of “the devil is in the
details.” As with many complicated undertakings
that involve both process and cultural change,
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 26
PART 2: ORGANIZATION AND GOVERNANCE | ACO TOOLKIT
ACOs may learn about essential details as they gain
a year or two of actual experience. For that reason,
newly formed entities may want to start as relatively
simpler ACOs, designed with flexible operational,
governance, and rewards structures that can
rapidly evolve in response to data and analysis of
changes in payments, practice patterns, and patient
behavior.
ENDNOTES
1. At the time that this publication was being written, the Tuscon Medical Center was actively in the ACO planning phase
and aiming to begin operating as an ACO in early 2011.
2. As discussed more fully in Part 3, the Dartmouth patient attribution algorithm assigns second priority to E&M codes
provided by medical specialists, with third priority for E&M services from surgical specialists.
3. It is also expected that non-physicians such as nurse practitioners, would be eligible to be assigned patients in the
ACO model.
4. Whether they are employees (or contracting with the ACO as independent practices), may influence how directly
compensation of core ACO physicians is directly tied to shared savings realized by the ACO.
5. MSOs offer a range of support services to independent physician and ancillary service groups, such as purchasing or
leasing equipment, contracting, billing, human resources, and regulatory compliance. MSOs typically include a hospital
ownership interest and hence are often used with hospital and physician alignments.
6. Report to Congress. 2006. Physician Group Practice Demonstration: First Evaluation Report. <Available as of July 21,
2010 at https://www.cms.gov/reports/downloads/Leavitt1.pdf>.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 27
Part 3
Accountability for Performance
David Axene, John Bertko, Dan Dunn, Bela Gorman,
Steve Lieberman, Joachim Roski, Mark Zezza
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
PART 3: ACCOUNTABILITY FOR PERFORMANCE
The shared-savings concept is based on the
notion of holding providers accountable for the
cost and quality of care delivered to a defined
population of patients. More specifically, providers’
reimbursement will be linked to their ability to
achieve greater value in care – lower costs and high
quality. This requires that ACOs be able to produce
meaningful evidence of their cost and quality
effects. In this section, we discuss several key
functions that an ACO will need to perform in order
to provide such evidence, including:
•
•
•
•
Patient Attribution;
Budget Development;
Payment Models and Incentives; and
Performance Measurement.
An initial step for an ACO and its prospective payers
is to agree upon a patient attribution methodology
in order to determine the patients and providers
that will be participating in the ACO. Once an
ACO has identified participating providers and
assigned patients, a budget can be developed to
calculate the actual and the projected benchmark
cost of care for the ACO. The establishment of this
budget will allow an ACO to monitor costs in order
to start identifying potential inefficiencies, as well
as to understand the value of differing elements of
attempted clinical transformation actions.
An ACO will also have to determine – with its
prospective payers – the amount of risk it is willing
to take in relation to the percentage of the shared
savings it hopes to receive in order to determine an
appropriate payment model. An ACO will also have
to decide how to distribute its shared savings to the
various providers to best incentivize the highestquality care at the lowest price. Finally, an ACO will
have to establish and track performance measures
that ensure an ACO’s efforts to reduce costs occur
through real improvements in the delivery of care.
Each organization will have to decide on how to
approach these key design features in a way that
best fits its organizational capacity and needs,
as well as its position in the marketplace. The
approaches and insights discussed below were
gathered from the Centers for Medicare & Medicaid
Services’ (CMS) demonstrations, experience with
the Brookings-Dartmouth ACO pilot sites, and the
expertise of thought leaders and practitioners;
however, these approaches do not represent the
only feasible tactics for carrying out these critical
functions. Rather, the suggested approaches below
serve as a guide on how to start thinking about the
development of elements required to become a
functioning ACO.
Exhibit 3.1 is an overview of the relationship of
these key functions with determining ACOs’ cost
impacts. The process discussed in Part 3 includes
the full progression of identifying the patients for
which ACOs will be held accountable, projecting
spending benchmarks, measuring financial
performance, distributing incentive payments,
and quality measures with a direct link to provider
reimbursements.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 29
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
EXHIBIT 3.1. OVERVIEW OF ACO BUDGET PROCESS AND FINANCIAL PERFORMANCE
Patient Attribution
Patients are assigned to an ACO provider if they receive the plurality of non-inpatient care for
evaluation and management services from that provider within a recent historical period. The ACO
is responsible for all of the costs and quality of care delivered to patients attributed to providers
who are exclusively members of that ACO.
ACO Budget Development
The historical spending amount is calculated and the projected benchmark spending amount is
estimated. There are several important subtasks involved in this step.
Baseline Historical Data
Trend Estimates
Adjustments
This includes claims, exposure,
utilization, and cost per service
statistics for a recent historical
period specific to the ACO
population.
Actual spending is trended
to the ACO contract period
based on historical trends,
to determine the projected
spending amount.
Several adjustments are made
to ensure actuarial reliability
and representativeness of the
data to finalize the projected
spending amount. The
benchmark spending amount is
then determined based on the
threshold.
Performance Monitoring
Once the budget is finalized, the ACO should use a set of reports that will monitor the actual costs
and compare them to the budget.
Payment Incentives
There are many different payment incentive arrangements. They range from “one-sided” shared
savings within a fee-for-service environment to partial capitation arrangements with quality
bonuses. The example described in this section is based on the one-sided shared savings model.
ACOs would be eligible to receive financial performance incentive payments if the actual spending
on their patients is below the benchmark, and they have met the quality targets.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 30
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
3.1: ATTRIBUTING PATIENTS TO AN ACO
The first step towards developing the ACO financial
framework is to determine the patient population
that an ACO will assume accountability for care.
Attributing patients to providers is an essential
component of the ACO model for purposes of
performance measurement (both cost and quality)
and payment incentives. It is important to note that
accountability for assigned patients lies with the
ACO, not the individual provider. Therefore, patient
attribution to providers is the way patients are
assigned to an ACO.
There are several ways to associate patients with
providers. One method, which is typically employed
by health maintenance organizations (HMOs) or
point of service (POS) plans, is for the payer or
enrollee to select a primary care physician from the
HMO or POS panel as the main source of care. This
assigns all patients within a plan to a provider, even
new enrollees without utilization history. However,
this method is typically associated with plans that
have a “closed” provider network.
An alternative approach is to use beneficiary
attribution methods to identify the patients for
which ACOs are accountable. This approach
has the benefit of relying on patient preferences
based upon where they have sought care in the
past. Therefore, it is based on established patientprovider relationships. This approach does not
require patients to select a provider. Furthermore,
this approach supports “open” access to providers.
This is particularly important to ensure that patients
enrolled in preferred provider organization (PPO)
or indemnity plans will not be restricted in their
choice of providers or access to care. A potential
downside of this approach is that individuals insured
by a plan who have limited utilization may not be
assigned. Also, patients may not know they have
been attributed to a particular ACO. Some ACOs
may find this problematic due to concerns that
attribution approaches lessen their ability to oversee
the care received by their attributed beneficiaries if
they are free to seek care from providers outside of
their ACO network.
Patient attribution methods have been used by
several shared-savings programs in the past. For
instance, CMS use patient attribution methods in
the Physician Group Practice (PGP) and Medicare
Health Care Quality (MHCQ) demonstrations.
Below, we discuss the patient attribution process in
more detail.
Once the providers and the payers decide to form
an ACO, a first step of the attribution model is for
the ACO to determine who the eligible providers
will be. More specifically, providers who are both
members of and exclusive to the ACO will be used
to assign patients to the ACO for accountability
purposes.1
The ACO gives this list of ACO providers to each
participating payer. Payers may include Medicare,
Medicaid, and multiple private payers. At this point,
each payer runs the patient attribution algorithm
on their covered patient population to determine
which of their enrollees are assigned to the ACO.
The assignment of the patients to the ACOs should
be empirically developed based on where the
individuals are receiving the plurality of evaluation
and management (E&M) physician services.
The Dartmouth Patient Attribution Model
Various patient attribution methodologies exist.
We will describe the Dartmouth patient attribution
process in more detail below. The Dartmouth
method emphasizes a longitudinal provider-patient
relationship and limits the likelihood of patients who
turn out to be high-cost from being excluded from
the ACO (i.e., taking steps so that high-cost patients
who had established a longitudinal relationship with
the provider are not referred out or discontinued
after implementation of the ACO).
Using the Dartmouth patient assignment
methodology, patients are empirically assigned to
a provider based on the patient’s historical care
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 31
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
PGP PATIENT ATTRIBUTION MODEL
The PGP Demonstration uses a retrospective annual patient attribution model that assigns beneficiaries to
a PGP based on where they receive their largest share of outpatient evaluation and management (E&M) services. Certain E&M services were excluded from the attribution methodology (e.g., emergency department
visits), as not all services relate to a PGP’s ability to manage and coordinate the health care of its beneficiaries.
One key distinction with the Dartmouth model is the retrospective feature. The retrospective approach
taken helps ensure that providers treat all patients the same, as providers are unaware of which beneficiaries are participating in the Demonstration until the end of the performance period. Yet, while this approach
prevents bias in patient treatment, it does not allow providers to easily target future care delivery interventions based on specific needs of their assigned patient population. To the extent the cost of care management capabilities increases with the number of patients, it also lessens economic benefits.
It was found that the PGPs in the Demonstration provided approximately 80 to 90 percent of the outpatient
E&M services for their assigned beneficiaries and retained nearly two-thirds of these beneficiaries from
year-to-year.2 This finding alleviates some of the concern over not being able to properly oversee their
patients care.
patterns, specifically the plurality of outpatient E&M
visits. Each payer does this by using the most
recent two years of claims data to assign patients
to a provider based on their visits and the specialty
priority of the provider. Each provider (both ACO
and non-ACO providers) is classified into one of the
three categories: primary care providers, medical
specialists, and surgical specialists, based on their
specialties. Exhibit 3.2 provides an overview of the
patient attribution algorithm.
Primary care is given the highest priority, so even
a single visit to a primary care provider trumps any
number of visits to a medical specialist or surgical
specialist. Thus, if a person had at least one visit
to a primary care provider, he or she would be
assigned to a primary care provider. If the patient
visited more than one primary care provider, he
or she would be assigned to the one whom the
patient visited the most. If the number of visits
among multiple primary care providers were equal,
the patient would be assigned to the provider with
the greatest number of days between the first and
the last visit, to choose the one with the longest
relationship. If the patient had only one visit with
multiple primary care providers, the patient would
be assigned to the provider with the most recent
visit.
The same methods are used when a patient has
had no primary care visits. If the patient had at least
one visit to a medical specialist, he or she would
be assigned to the medical specialist regardless of
the number of visits he or she had with a surgical
specialist. If the patient had no primary care or
medical specialist visits, the patient would be
assigned to the surgical specialist with whom he
or she had the most visits. If a patient had no valid
outpatient E&M visits within the two-year window,
that patient would not be assigned.
Once each patient is assigned to a provider, the
payers consult the list provided by the ACO to
determine which providers are the ACO providers
to whom patients can be assigned. Those patients
assigned to a participating ACO provider are then
assigned to that provider’s ACO. Patients may be
reassigned to the ACOs on an annual basis, with the
exception of large groups of patients signing on to a
private plan in mid-year.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 32
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
EXHIBIT 3.2. OVERVIEW OF A PATIENT ATTRIBUTION PROCESS
Start with two years of
Only use
Merge with Physician to Speciality Crosswalk
claims data using 100
claims that reflect an
file for all physicians, including non-ACO
percent data and all
outpatient visit and have allowed
providers; e.g., physician A —> Cardiologist
age groups
charges > $0.00
—> Medical Specialist (MDSP)
Eliminate
For each enrollee, find:
claims with physician
• Number of visits per provider
• Date of last visit to each provider
• Number of days between first and last visit to each provider
speciality indicating an ancillary care
provider (e.g., pathology, radiology, or nuclear
medicine)
• If no PRIM visits occurred,
attempt first to assign patient to a
Check if
No
to a Primary Care Provider (PRIM)
Medical Specialist (MDSP)
•
enrollee had at least one visit
If no Medical Specialist visits,
assign to a Surgical Specialist
(SURG)
•
Yes
In both cases, use the same
algorithm as for PRIM assignment
Check if
(at right)
Yes
only one PRIM visit
Assign
to PRIM
No
Check
if enrollee visited one
Yes
PRIM more than any other
Assign
to this PRIM
No
Assign
to PRIM with whom
patient had most recent visit
No
Is the
largest number of
Assign
Yes
visits to a given PRIM > 1?
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE to PRIM with
greatest length of time
between first and last visit
January 2010 | 33
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
For the Brookings-Dartmouth pilot sites, the
participating payers and providers must agree to
use this patient attribution methodology for ACO
incentives. We recognize that different attribution
models are in place for other delivery system
reforms, such as the Patient-Centered Medical
Home. While more testing is necessary, it appears
reasonable for a provider system to use multiple
attribution models for the various delivery system
reforms they are participating in. For example, a
provider system could use one attribution method
for the medical home delivery system and a
separate, distinct attribution method for the ACO
delivery system. However, it is beneficial for all
ACOs in an area to use the same attribution method
to avoid having conflicts in patient assignment.
Changes in core practice patterns and patient care
management are expected under the ACO model.
To support and reinforce the practice pattern
changes, a sufficiently large percentage of the
providers’ patients should be enrolled in the ACO.
Based on anecdotal discussions with industry
experts, estimates of the desired critical mass can
be expected to vary widely, ranging from 20 percent
to 60 percent of the patients at an office or clinic
location to be enrolled in the ACO.
Reaching critical mass may require the inclusion
of both Medicare beneficiaries and commercial
members. While Medicaid should be considered,
the complexity, payment rates, and other unique
attributes may add obstacles including to Medicaid
in some states.3
Exclusivity
It is important to note that the providers used
for patient attribution should be exclusive to the
ACO, while the providers not selected for patient
attribution are free to participate in multiple ACOs.
The exclusivity criterion allows for clearer evaluation
of an ACOs performance and also alleviates
concerns over gaming. For example, imagine if a
Medicare provider is used for patient attribution
in two ACOs. It may become difficult to keep
track of which ACO is accountable for his or her
patients. Furthermore, if the patients happen to be
high cost, each ACO may be incentivized to try and
disassociate itself from them.
Critical Mass
Another critical issue with patient attribution is the
number of patients that are attributed to an ACO.
An ACO should have a sufficiently high number
of patients attributed for two reasons. The first
involves the ability to obtain statistically or even
practically meaningful results on the cost and
quality impacts of ACOs. The statistical issue will
be discussed in more detail with the benchmarking
and performance measurement discussion later in
this section. The second reason involves increasing
the likelihood of achieving a critical mass of patients
to incentivize providers to change care processes.
It should also be noted that many ACOs will be
contracting with multiple payers. Arrangements
with commercial payers will vary based on the
location and the market of the ACO partners.
Potential contracts may be negotiated with a
commercial carrier for fully insured business, selfinsured employer plans, or both. Although the
characteristics of differing payers and membership
may vary, the expected provider practice pattern
changes and payment incentives should be aligned
across all patients in the ACO.
It is also expected that ACOs will continue
treating non-ACO patients. To the extent that
non-participating payers benefit from the practice
pattern changes adopted by ACO providers, their
costs may drop without sharing the savings with
the ACO providers, creating a “free rider” problem.
There is thus an incentive for both ACOs and payers
to achieve as much broad participation as possible.
Changes in Membership
After a patient has been attributed to a physician
and the ACO, the assignment generally remains for
a period of one year even if the patient changes his/
her care to providers outside of the ACO. Rules
will be needed to make potential adjustments for
patients who relocate, die, or lose coverage within
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 34
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
the year. Although ACOs may consider alternative
timeframes for handling enrollment issues, any
approach adopted must limit the ability to “dump”
high-cost patients while giving providers the
opportunity to impact their patients’ care.
It is expected that ACO membership would remain
relatively stable on an annual basis. Over time,
membership would grow through the addition of
new payers, ACO providers, or new groups of
patients from existing payers.
3.2: DEVELOPING A BUDGET
Budget development within the context of the ACO
framework entails analyzing historical utilization
and cost data for the purposes of identifying areas
for performance improvements, and developing
benchmark spending targets that will determine
shared-savings eligibility. Essentially, this process
is used to measure the financial performance – the
ability to control cost growth – of the ACO.
There are many different approaches to measuring
the performance of an ACO. In the PGP and
MHCQ shared-savings demonstrations, a quasiexperimental design has been used to determine
shared savings. This approach compares spending
and quality trends from the population served by
intervention providers, ACOs, with similar trends in
control populations, those not being treated by the
ACO.
This cohort approach has several advantages. For
instance, it more easily allows for the inclusion of
trends, such as a sudden surge in utilization due to
an unforeseen local disease outbreak, which would
have been difficult to anticipate prior to the budget
development process.
There are also several drawbacks with the control
group approach. For example, there is often a
lengthy lag between the end of the performance
period and when the results can be analyzed due
to the slow availability of claims information. As a
result, providers often do not find out the results
of their intervention until months – even years –
after the performance period ends. Consequently,
they would not receive shared-savings until much
after they have made initial investments. The PGP
Demonstration, for example, was already at the
end of performance year five before results were
available for the first three years.
Another challenge with using the cohort approach
is ensuring that the control and intervention groups
share similar risk characteristics. For example,
in the PGP demonstration, diagnoses and other
health information indicated that claims data are
used to control for relative risk differences between
the intervention and control cohorts. However,
there are incentives for intervention providers to
improve diagnostic coding practices from how they
coded prior to participating in the demonstration.
Financially, there is incentive to more fully document
diagnoses as higher-risk scores ultimately translate
into higher payments. Participating providers may
also improve documentation to better target quality
improvement initiatives. This issue can make it
difficult to distinguish cost and quality impacts
based on improved performance in care delivery
from coding changes.
Another major drawback of the cohort approach
is that it will become less viable over time. More
specifically, as interest in ACOs grows – as well
as many other value-based payment initiatives –
especially with the start of the Medicare Shared
Savings program in 2012, it will be increasingly
difficult to identify control populations without
significant intervention activity. Therefore, a budget
projection approach using historical data appears
to be a more sustainable approach.
The budget projection model builds on historical
spending and utilization data from the ACO specific
population to project budget benchmarks for future
performance periods. The ACOs’ actual spending
in the performance period is compared to the
spending benchmark to determine whether savings
were achieved.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 35
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
Aside from addressing the increasing difficulty
in finding adequate external control populations,
there are a number of other benefits to the budget
projection approach. Having prospective budget
benchmarks gives providers the ability to judge their
ongoing performance and set course corrections
throughout the performance period, rather than
having to wait until after the period to learn about its
performance relative to a control cohort about which
they had no comparative information. As discussed
later on in more detail, the budget projection
approach may also mitigate difficulties involved in
controlling for relative risk characteristics, since the
intervention population itself is used to develop the
spending benchmarks.
While the budget projection approach has
advantages, there are also several technical
and theoretical problems. For example, budget
projection models build historical trends into the
projected benchmarks. This could unintentionally
favor and reward ACOs with historically high
spending growth from years of inefficient health
care practices, as this trend extrapolates into higher
benchmark spending targets that provide easier
opportunities to achieve cost reductions. Setting
benchmarks on a prospective basis under the
budget projection framework also makes it more
difficult to take into account unanticipated shifts in
spending patterns from historical trends. Using a
control group approach with retrospective spending
benchmarks would better account for system-wide
changes in health care delivery.
Besides the fact that the cohort approach will
become less viable over time, another major
factor for the budget projection model to become
the predominate method for measuring financial
performance in an ACO construct is that the
Patient Protection and Affordable Care Act (ACA)
has legislated such an approach for the Medicare
shared-savings program. Furthermore, the cohort
approach is not as feasible in the private sector due
to limitations in the amount of data that would be
available on non-ACO patients. Therefore, we focus
on the budget projection in Part 3, which is also the
method of choice for the providers participating in
the Brookings-Dartmouth ACO pilot program. The
following provides a discussion of the key issues
involved in developing a budget projection model
in the ACO framework, while also providing an
illustrative example.
In order to develop these spending benchmarks,
it is necessary to determine the baseline spending
amount and appropriately project spending for the
contract period, assuming there are no behavioral
changes in practice patterns (i.e., projecting what
spending would have been without implementation
of an ACO). Developing an accurate ACO budget
requires the actual claims and exposure data,
significant data analysis and interpretation, and
an understanding of ACO operations. Additional
factors that need to be taken into consideration
include the projected time period, the type of
data to be used, any data anomalies, changes in
the population, and development of appropriate
assumptions and adjustments to be used. Detailed
below are the key steps involved in developing an
ACO budget, organized into four broad sections:
baseline data, trend estimates, adjustments, and
performance monitoring.
Baseline Data
Ideally, the ACO budget will be developed using
baseline data, including existing claims, utilization,
and exposure data. Claims information is reported
based on either a paid basis or an incurred and
paid basis. For example, paid claims in calendar
year (CY) 2008 represent all the claims that were
paid during that year, regardless of when the
services occurred. Incurred claims in CY 2008
represent the claims for services rendered in CY
2008, regardless of when the claims were paid. For
budget development purposes, incurred claims
are more appropriate. However, when using
incurred claims, the significant lag time between
when a claim is incurred and when it is actually
reported by the provider to the insurance carrier or
Medicare Administrative Contractor (MAC) must be
considered.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 36
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
Generally, insurance carriers report incurred claims
for a 12-month period with a three-month lag (runout), which is then adjusted for an incurred but
not yet reported (IBNR) factor. For example, CY
2008 incurred claims paid through March 2009
would represent 12 months of data with three
months of run-out. These incurred claims would
then be increased by an IBNR factor to account
for additional incurred claims in CY 2008 that have
not yet been reported to the insurance carrier.
Usually, the insurance carrier’s actuary will develop
an IBNR factor to “complete” the claims. Using
claims development models, actuaries review the
claims payment patterns and the historical trends to
determine the appropriate IBNR factor to apply to
claims. The IBNR factor decreases as the number
of months of run-out increases.
For budgeting purposes, incurred claims should be
reported by broad service or expense categories.
The categories may vary by payers since each
payer has different reporting mechanisms. When
determining these categories, special attention
should be given to the plan designs chosen by
the ACOs members. For example, if there were
a specific copayment for CT-Scans, it would
be helpful to track the CT-Scans as a separate
category. This will help the ACO to account for
the impact of plan designs (and future plan design
changes) on their budget. Some payers may not
include certain services in their ACO contracts;
for example, Medicare fee-for-service (FFS) ACOs
may not include the outpatient pharmacy service
(since they are paid separately under Part D).
Some services are more important for specific
populations; for example, home health is an
important service for the Medicare population, but
this is not typically the case for commercial plans
covering the non-elderly population. Suggested
categories are shown below:
•
•
•
•
•
Hospital Inpatient
Hospital Outpatient
Lab/X-Ray
Advanced Imaging
Physicians (Primary Care and Specialty Care may be
broken out separately)
The above discussion focused on a more traditional
approach of developing the spending amount based
on the service categories. As new payment methods
evolve, such as the patient-centered medical
home model, partial capitation payments, and
bundled episode payments, additional analyses and
modifications are needed. For example, with partial
capitation payments, the capitation amount should
be captured, and we would expect a cost and
utilization reduction in the corresponding service
categories.
The ACO needs to capture the total claims costs for
each category. This includes the claims the payers
are responsible for financially, and the members’
cost-share amount. The total costs are defined as
the “allowed claims.” Note that these claim costs
should also include the out-of-network claims,
which will also count against the benchmark when
determining the ACO’s eligibility for its financial
performance payments.
In addition to the claims, developing the budget
requires exposure data, which are typically reported
in the form of member months. Exposure is
defined as the number of members enrolled in the
ACO each month. For example, if there were 50
members enrolled in January, 49 in February, and
47 in March, the total member months for January
through March would be 146. The time period
of the exposure data must match the time period
of the claims information. For CY 2008 incurred
claims, one would need CY 2008 member month
information. Exposure data are extremely important
when there are significant fluctuations in enrollment.
One cannot develop an accurate budget without
it. For ACOs, member months are determined from
the attributed members. As stated in the patient
attribution section, patients are assigned once
•
•
•
•
•
Outpatient Pharmacy
Mental Health/Substance Abuse
Durable Medical Equipment (DME)
Emergency Room
Other4
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 37
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
a year. Thus, normally there will not be any new
patients, unless there is a large group joining during
the mid-year. However, there are deaths, so the
number of members will typically only decrease.
It is best to use the most recent 12-24 months of
historical experience to develop the baseline data.
Incurred claims and member cost sharing should
first be adjusted for IBNR, and then divided by the
member months (exposure) to calculate the claims
costs per member per month (PMPM) and costshare PMPM for the various service categories
listed above. An illustrative example is shown in
Exhibit 3.3.
After the baseline claims PMPM are developed,
the next step is to review the baseline utilization
data. Typically, the utilization data is reported in
important to understand the utilization assumptions
that are reflected in the budget. Outpatient
pharmacy utilization is typically measured in
prescriptions per member per year. All other
services are reported as visits per member per year,
visits per 1,000 members per year, or services per
member per year. Also, as mentioned above, there
is a lag time between when a service is incurred and
when a service is reported. Utilization data should
be adjusted for the IBNR just as the incurred claims
are adjusted. It is important to ensure that the
reporting metrics are consistent among reports and
from year to year. Changes in utilization reporting
will impair an ACO’s ability to monitor and measure
their progress over time.
With the utilization data and the allowed claims
PMPM, one can impute the average cost per
EXHIBIT 3.3. BASELINE ALLOWED CLAIMS PMPB
CY 2008
Hospital Inpatient
Hospital Outpatient
Lab/X-Ray
Imaging
Physicians
Primary Care
Specialty Care
Incurred
Incurred Claims
Member Cost
Member Cost
Total Allowed
Claims
PMPM
Sharing
Sharing PMPM
Claims PMPM
$3,600,000
$4,500,000
$1,260,000
$360,000
$60.00
$75.00
$21.00
$6.00
30,000.00
15,000.00
-
$0.50
$0.25
$60.50
$75.25
$21.00
$6.00
-
$1,500,000
$2,000,000
$25.00
$33.33
180,000.00
270,000.00
$3.00
$4.50
$28.00
$37.83
$3,060,000
$51.00
780,000.00
$13.00
$64.00
$720,000
$12.00
120,000.00
$2.00
$14.00
$180,000
$90,000
$730,000
$3.00
$1.50
$12.17
30,000.00
-
$0.50
Total
$18,000,000
$300.00
1,425,000
$23.75
Member Months CY 08
60,000
Pharmacy
Mental Health/Substance
Abuse
Emergency Room
DME
Other
-
$3.50
$1.50
$12.17
-
various forms, such as services PMPM, or services
per 1,000 members per year. There are several
measures for hospital inpatient utilization: days
per 1,000, average length of stay, and admits or
discharges per 1,000. It is best to receive all the
measures of hospital inpatient utilization, since it is
$323.75
service. When performing this calculation, it is
best to translate all utilization information into a
PMPM metric. For example, if the hospital inpatient
days per 1,000 members are 300, the PMPM is
calculated by dividing 300 by 12,000 to get a
resulting PMPM of 0.025. In other words, for every
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 38
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
1,000 members, there are 300 hospital days in a
year, or 2.5 percent of a day for every member each
month. To calculate the average costs per day, one
would take the allowed claims PMPM (i.e., $60.50),
and divide it by the utilization .025, resulting in
$2,420 per day. The cost and utilization report
includes all the baseline data the ACO needs to
begin developing the ACO budget. A hypothetical
cost and utilization report is shown in Exhibit 3.4.
Several adjustments can be applied to the baseline
allowed claims PMPM to project to the applicable
projection period. Since the baseline information
is being projected to a future time period, several
questions must be addressed, such as:
• What are the historical cost trends, and do we
expect the same cost trends in the future?
• Would our reimbursement of services change,
and if so, how?
EXHIBIT 3.4. COST & UTILIZATION REPORT
Reported
Cost/Day or Cost/
Utilization
Allowed Claims
CY 2008
Utilization
Service
PMPM
PMPM
Hospital Inpatient
300 days per 1000 members per year
$2,420.00
0.025
$60.50
Hospital Outpatient
7250 services per 1000 members per
$124.55
0.604
$75.25
$75.00
0.280
$21.00
$248.28
0.024
$6.00
year
Lab/X-Ray
3360 services per 1000 members per
year
Imaging
290 services per 1000 members per
year
Physicians
Primary Care
3.6 visits per member per year
$93.33
0.300
$28.00
Specialty Care
2 visits per member per year
$227.00
0.167
$37.83
Pharmacy
10 scripts per member per year
$76.20
0.833
$64.00
Mental Health/Substance
2.25 visits per member per year
$74.67
0.188
$14.00
Emergency Room
150 visits per 1000 members per year
$280.00
0.013
$3.50
DME
90 services per 1000 members per
$200.00
0.008
$1.50
$132.73
0.092
$12.17
Abuse
year
Other
1100 services per 1000 members per
year
Total
$323.75
Trend Estimates
• What are the historical utilization trends, and are
The budget projection model could be implemented
in a number of ways, ranging from complex
regression-based forecasts to simpler movingaverage trend extrapolations. It is also possible to
use a negotiated approach that takes into account
budget projections but also leaves room for ACO
input. Below, we describe key considerations for
projection calculations.
• How has the mix of services changed, and how
these trends expected to continue?
would it change in the future?
Answers to the above questions will facilitate the
development of the adjustment assumptions to be
applied to the baseline claims PMPM.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 39
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
EXHIBIT 3.5. THREE-YEAR COST & UTILIZATION REPORT
CY 2008
CY 2007
CY 2006
Allowed
Service Category
Allowed
Cost/Day
Utilization
Claims
Cost/Day
Utilization
or Service
PMPM
PMPM
or Service PMPM
Allowed
Claims
Cost/
Utilization
Claims
PMPM
Day or
PMPM
PMPM
0.025
0.545
0.240
0.021
$51.59
$57.71
$16.49
$4.58
Hospital Inpatient
Hospital Outpatient
Lab/X-Ray
Imaging
$2,420.00
$124.55
$75.00
$248.28
0.025
0.604
0.280
0.024
$60.50
$75.25
$21.00
$6.00
$2,251.16
$114.27
$71.43
$232.03
0.025
0.570
0.257
0.022
$56.28
$65.13
$18.35
$5.19
Service
$2,084.41
$105.80
$68.68
$218.90
Physicians
Primary Care
Specialty Care
$93.33
$227.00
0.300
0.167
28.00
37.83
$90.61
$216.19
0.291
0.159
$26.39
$34.32
$87.98
$207.88
0.286
0.154
$25.12
$32.04
Pharmacy
Mental Health/
$76.80
$74.67
0.833
0.188
$64.00
$14.00
$73.14
$72.49
0.801
0.175
$58.61
$12.70
$69.00
$70.72
0.763
0.165
$52.66
$11.69
Substance Abuse
Emergency Room
DME
Other
$280.00
$200.00
$132.73
0.013
0.008
0.092
$3.50
$1.50
$12.17
$260.47
$190.48
$126.41
0.012
0.007
0.089
$3.10
$1.37
$11.25
$242.29
$179.69
$119.25
0.011
0.007
0.086
$2.77
$1.25
$10.30
Total
$323.75
The “raw” or unadjusted trend factor is one of the
more difficult assumptions to determine, and there
are many approaches to its development, from a
simplified approach to ones that are more complex.
All approaches should begin with a historical cost
and utilization trend analysis, typically on the patient
population expected to be treated by the ACO. The
ACO should obtain the historical cost per service,
utilization, and allowed claims data for each service
category for a series of time periods. In addition,
the ACO would need exposure or member month
data for the various time periods. Ideally, the more
time periods or data points one has, the more
information there is to perform a thorough trend
analysis. For example, retrieving data for the most
recent 36 months and then grouping the data
into 12-month, six-month, and three-month time
periods can provide many data points, and through
regression modeling, can result in a robust trend
analysis. However, it may be difficult to retrieve this
amount of data and the ACO may receive only a few
data points, such as the annual data for the past
three years. The data should be put in the form
of a cost and utilization report as shown in Exhibit
$292.69
$266.20
3.5, which illustrates an annual three-year cost and
utilization report.
Once the ACO has the historical cost and utilization
report, it is easy to calculate the annual trends.
This example involves the annual data for only
the past three years and thus calculating annual
trends is a simple division exercise. The above
cost and utilization report can be used to derive
the cost and utilization trend factors, as illustrated
in Exhibit 3.6. As shown, for CY 2008, the hospital
inpatient utilization trend factor is 1.00, or a zero
percent trend. This means the utilization for this
service did not increase from CY 2007 to CY 2008.
The hospital inpatient cost per day trend factor is
1.075 or 7.5 percent, which means the cost per day
increased 7.5 percent from CY 2007 to CY 2008.
For all services, the overall claims PMPM increased
10.6 percent from CY 2007 to CY 2008.
If the ACO is able to retrieve more data points
such as the six-month and the three-month time
periods, annual trend calculation becomes a little
more complex, and the resulting trends may need
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 40
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
to be annualized. In order to determine how to
annualize a trend, one must calculate the number
of months between the midpoints of each of the
time periods being compared. For example, if the
two time periods being compared are July 2006 –
December 2006 and January 2006 – June 2006,
the midpoint of the first time period is Oct 1, 2006,
and the mid-point of the second time period is
March 1, 2006. The number of months between
these two midpoints is six months and thus the
resulting calculated trend is a six-month trend. To
annualize this six-month trend, one would determine
the monthly trend and then translate it to an annual
historical pattern and to determine whether there
is a trend toward the use of more expensive or less
expensive services. It is important to consider all
the factors that could affect the historical pattern
and the projected mix of services. For example,
it could be driven by a change in the physician
practice patterns due to the newly implemented
medical management or disease management
programs. It could also be driven by a change in
the member utilization due to the change in their
benefit coverage. The result of this analysis should
be reflected in the trend assumptions or in the
adjustment factors.
EXHIBIT 3.6. ANNUAL TREND ANALYSIS
CY 2008
Cost/Day or
Utilization
Allowed Claims
CY 2007
Cost/Day or
Utilization
Allowed Claims
Hospital Inpatient
Service
1.075
PMPM
1.000
PMPM
1.075
Service
1.080
PMPM
1.010
PMPM
1.091
Hospital Outpatient
Lab/X-Ray
Imaging
1.090
1.050
1.070
1.060
1.090
1.080
1.155
1.145
1.156
1.080
1.040
1.060
1.045
1.070
1.070
1.129
1.113
1.134
Physicians
Primary Care
Specialty Care
1.030
1.050
1.030
1.050
1.061
1.103
1.030
1.040
1.020
1.030
1.051
1.071
Pharmacy
Mental Health/
1.050
1.030
1.040
1.070
1.092
1.102
1.060
1.025
1.050
1.060
1.113
1.087
Substance Abuse
Emergency Room
DME
Other
1.075
1.050
1.050
1.050
1.040
1.030
1.129
1.092
1.082
1.075
1.060
1.060
1.040
1.040
1.030
1.118
1.102
1.092
Total
1.106
trend. If the six-month trend is five percent, the
calculation to annualize this trend is (1.05^(12/6)),
with a resulting annual trend of 10.25 percent.
Once the cost and utilization trend information is
calculated, the ACO must analyze the historical
trend and understand the drivers, and then consider
how the historical trends may change in the
future. Separate analyses should be performed
to consider possible changes in the costs and
utilization patterns. For example, the ACO should
perform a “mix-of-services analysis” to assess the
1.100
In addition, the ACO should review and reflect
expected changes in reimbursements. For
example, with the implementation of the patientcentered medical home model or the bundled
episode payment model, the ACO would expect
greater efficiencies while improving the quality of
care. Further, analyzing the data at the episode
or condition level could provide the ACO a clearer
actionable path to reducing costs and increasing
the quality of care delivered. Before the ACO
determines the trend assumptions to be used in the
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 41
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
budget development, it must perform all the relevant
analyses. Below is a suggested list of analyses:
•
•
•
•
•
Cost and Utilization Trend Report
Mix of Services Analysis
Provider Reimbursement Analysis
Utilization Management Analysis
Disease Management and other Medical
Intervention Analysis
The ACO is now equipped to finalize the annual
cost and utilization trend assumptions by service
category. Before applying the trend assumptions
to the baseline data, one must consider the
time periods between the baseline data and the
applicable ACO contract period. For example, if the
baseline data represent CY 2008 and the contract
is for CY 2010, the annual trend must be applied for
two years. If the ACO has information that warrants
different annual trends for CY 2009 and CY 2010,
this should be reflected in the trend assumptions.
Exhibit 3.7 illustrates the application of the trend
factors to the baseline data to calculate the
unadjusted projected costs for the ACO contract
period, using service categories. As shown, for
the primary care physicians, the utilization PMPM
for CY 2008 is 0.30, the cost per service is $93.33,
and the allowed claims PMPM is $28 (0.30 x
$93.33). The cost and utilization trend assumptions
for CY 2009 are three percent and four percent
respectively, and for CY 2010, three percent and
4.5 percent. The CY 2010 allowed claims PMPM
is $32.28. The utilization trend factors are applied
to the utilization PMPMs and the cost trend factors
are applied to the cost/day or service. The allowed
claims PMPM is calculated by multiplying the
utilization PMPM and the cost/day or service. In
this example, the projected CY 2010 allowed claims
PMPM is $391.83, which is 21 percent higher than
the baseline allowed claims PMPM of $323.75. The
average annual trend is 10 percent (1.21^0.5).
In the above discussion, we have described an
illustrative budget development process. Providers
and payers may consider alternative trend
projection/budget development methodologies for
the purpose of setting benchmark spending levels.
This is particularly important if reliable baseline data
is not available or if there are reasons to believe
that the future trends would significantly diverge
from the historical trends. In the text box below, we
discuss the application of national or regional trends
in the context of Medicare FFS ACOs. Providers
and payers may also want to agree on other trend
factors such as the use of the rate of growth in
general inflation.
Adjustments
The projected spending estimates should be
representative of the patient population that is
treated by the ACO providers. It is possible that
there may be anomalies in the baseline data that
need to be adjusted accordingly, in order to ensure
that the baseline estimates remain representative,
when trended forward. Also, it is possible that
there may be a significant population change in the
ACO membership that could cause deviations from
historical experience. In this section, we specifically
address the risk adjustment factors and the highcost claimant adjustments.
Risk Adjustment
A population shift can occur between the time
period of the baseline data and when one begins
to develop a budget (e.g., from the recession
recession). This shift can dramatically change the
risk profile of the ACO’s population and must be
reflected in the budget. There are many ways to
assess the risk profile of a population, such as using
risk-adjustment models or actuarial age/sex factors.
Using actuarial age/sex factors as an example,
one should first review recent demographic data
and then compare them to the demographics of
members that represent the baseline data. For
example, if the baseline data reflect CY 2008 and
budget development begins in March or April of
2009, the ACO should request demographic data
as of July 2008 and as of March or April of 2009.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 42
Service
$2,420.00
$124.55
$75.00
$248.28
$93.33
$227.00
$76.80
$74.67
Category
Hospital
Inpatient
Hospital
Outpatient
Lab/X-Ray
Imaging
Physicians
Primary Care
Specialty Care
Pharmacy
Mental Health/
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE $200.00
$132.73
Room
DME
Other
Total
$280.00
Abuse
Emergency
Substance
Day or
Service
CY 2008
Cost/
0.008
0.092
0.013
0.833
0.188
0.300
0.167
0.280
0.024
0.604
0.025
PMPM
Utilization
$323.75
$1.50
$12.17
$3.50
$64.00
$14.00
$28.00
$37.83
$21.00
$6.00
$75.25
PMPM
$60.50
Claims
Allowed
1.050
1.050
1.075
1.050
1.030
1.030
1.040
1.050
1.070
1.090
Factor
1.060
Trend
CY 2009
Cost
1.040
1.030
1.050
1.040
1.070
1.040
1.040
1.080
1.070
1.050
Factor
1.000
Trend
Utilization
$210.00
$139.36
$301.00
$80.64
$76.91
$96.13
$236.06
$78.75
$265.66
$135.76
Service
$2,565.20
Day or
Cost/
0.008
0.094
0.013
0.867
0.201
0.312
0.173
0.302
0.026
0.634
0.025
PM/PM
Utilization
$355.92
$1.64
$13.16
$3.95
$69.89
$15.43
$29.99
$40.92
$23.81
$6.87
$86.12
PMPM
$64.13
Claims
Allowed
EXHIBIT 3.7. UNADJUSTED PROJECTED SPENDING
1.050
1.050
1.075
1.050
1.030
1.030
1.040
1.050
1.080
1.090
Factor
1.060
Trend
CY 2010
Cost
1.040
1.030
1.050
1.040
1.070
1.045
1.040
1.080
1.070
1.050
Factor
1.000
Trend
Utilization
Cost/
$220.50
$146.33
$323.58
$84.67
$79.21
$99.02
$245.52
$82.69
$286.91
$147.98
Service
$2,719.11
Day or
0.008
0.097
0.014
0.901
0.215
0.326
0.180
0.327
0.028
0.666
0.025
PMPM
Utilization
$391.83
$1.79
$14.23
$4.46
$76.32
$17.00
$32.28
$44.26
$27.01
$7.94
$98.57
PMPM
$67.98
Claims
Allowed
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
January 2010 | 43
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
USING A NATIONAL TREND FACTOR
In the example described in the text, baseline data from the ACO specific population is used to derive
a trend factor. This data reflects the practice patterns of the local market. It is well documented that
among the FFS Medicare population, there are significant variations in practice patterns and thus, cost
disparities across geographic areas. Given these geographic variations, a national or regional trend factor
may be more appropriate for developing the spending benchmarks when working with a Medicare FFS
population. In fact, ACA suggests that projected national spending growth be used as the basis for setting
benchmarks in the Medicare Shared Savings program.
There are reasons to consider a national trend in both historically low- and high-spending growth areas.
In the low-growth areas where providers have already demonstrated success in limiting cost growth, it
may not be realistic to expect ACO providers to continue the same level of cost-control in the future. In
contrast, areas with historically high-spending growth may have spending patterns reflective of poor quality
and inefficient care. There may also be a perverse incentive for providers to drive up costs before starting
an ACO. In these cases, ACOs would not want to incorporate that inefficiency into the trend factors, as
inflated factors would make it easier for providers to earn rewards for care that is not necessarily reflective
of better quality.
A national trend factor, which averages the high- and low-growth areas, has the advantage of mitigating
the influence of geographic variation. This approach reduces the likelihood of rewarding historically high
levels of spending growth that are indicative of low quality and inefficient care. It can also provide areas
that have demonstrated success in controlling costs more achievable financial performance goals.
While a national trend factor may be sensible for a Medicare FFS population, in the non-Medicare market,
there are additional variations in areas such as reimbursement design, benefit design, management
protocols, and member risk composition. These additional variations make it less likely that a national
trend factor is applicable or achievable for a local non-Medicare market.
Typically, demographic data are reported in five-year
age intervals by sex for adults and varying intervals
for children. Once the information is received, the
ACO should calculate a weighted average age/sex
adjustment factor for each of the time periods, using
a standard set of actuarial age/sex factors. The
ratio of the two age/sex adjustment factors from the
two periods represents the change in the risk profile
of the population between the two points in time.
This example further assumes the risk profile will
remain stable throughout the projection period, as
the March/April period.
Exhibit 3.8 and Exhibit 3.9 illustrate the calculation
of the age/sex factors using hypothetical
demographics and age/sex factors. The standard
age/sex factors in the illustration represent a
typical commercial population (these factors and
age groupings would be different for a Medicare
population). The first table calculates the average
age/sex factor of a hypothetical population as
of June 2008. As shown, the weighted average
factor is 1.05. The second table calculates the
average age/sex factor for the population as of April
2009. As shown, the population grew from 4,413
members to 5,115 members, and the average age
factor changed from 1.05 to 1.10. This suggests
that the risk profile of the population has changed.
The population became older and the average age
factor moved from 1.05 to 1.10, suggesting that
costs will increase by approximately 4.8 percent
(1.10/1.05-1) from the baseline. With this data, the
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 44
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
EXHIBIT 3.8. JUNE 2008 AGE/SEX FACTOR CALCULATION
Age Cohort
0-18
19-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65+
Total
Male
Actuarial
June 08
Weighted
Female
Actuarial
June 08
Weighted
Total
June 08
Weighted
Age
Enrollment
Factor
Age Factor
Enrollment
Factor
Enrollment
Factor
275
275
275
228
210
250
175
150
75
125
75
2,113
146.39
102.58
116.78
119.32
136.57
201.58
181.01
210.20
142.23
318.25
249.08
1,924
0.532
0.846
0.997
1.102
1.072
1.119
1.281
1.512
1.812
2.170
2.646
1.18
250
225
250
275
300
250
275
175
150
100
50
2,300
133.08
190.43
249.25
302.96
321.60
279.75
352.28
264.66
271.85
216.97
132.28
2,715
525
500
525
503
510
500
450
325
225
225
125
4,413
279.48
293.00
366.03
422.28
458.17
481.33
533.28
474.86
414.08
535.22
381.36
4,639
Factor
0.532
0.373
0.425
0.523
0.650
0.806
1.034
1.401
1.896
2.546
3.321
Weighted
0.91
Factor
1.05
EXHIBIT 3.9. APRIL 2009 AGE/SEX FACTOR CALCULATION
Male
Actuarial
Apr 09
Age Factor
Enrollment
0-18
19-24
0.532
0.373
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65+
Total
Weighted
0.425
0.523
0.650
0.806
1.034
1.401
1.896
2.546
3.321
Age Cohort
1.18
Weighted
Female
Actuarial
Weighted
Total
Apr 09
Apr 09
Factor
Age
Enrollment
Weighted
Factor
Enrollment
Factor
275
250
146.39
93.25
Factor
0.532
0.846
250
225
133.08
190.43
525
475
279.48
283.68
290
300
225
375
250
300
150
125
100
2,640
123.15
157.00
146.33
302.38
258.58
420.40
284.45
318.25
332.10
2,582
0.98
0.997
1.102
1.072
1.119
1.281
1.512
1.812
2.170
2.646
1.23
250
275
300
275
300
200
175
125
100
2,475
249.25
302.96
321.60
307.73
384.30
302.47
317.16
271.21
264.57
3,045
1.23
540
575
525
650
550
500
325
250
200
5,115
372.40
459.96
467.93
610.10
642.88
722.87
601.61
589.46
596.67
5,627
1.10
Factor
ACO should adjust its projected cost by 4.8 percent.
The simplest approach is a bottom-line adjustment
to the total claims PMPM. As shown in Exhibit
3.7, after trend, we have projected the CY 2010
total allowed claims to be $391.83 PMPM. This
unadjusted projected cost would then be increased
by 4.8 percent to reflect the change in risk profile.
employer group within the ACO’s service area.
Although it is critical to monitor the population
make-up over time, it may not be necessary to
make risk adjustments to the spending estimates if
the population is fairly stable. ACOs are expected
to have a large enough membership to ensure a
stable risk profile over time.
It should be noted that large shifts in the risk profile
of the population are expected to be rare and
typically would only occur if a payer adds a large
Additional discussion on risk adjustment included
more complicated models which control for specific
diseases is included in the Part 3 Appendix.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 45
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
High-Cost Claimant Adjustment
In the course of analyzing the baseline data
and performing the trend analyses, the ACO
may observe data anomalies. This may warrant
further investigation and a review of the high-cost
claimants for the baseline time period. Insurance
carriers or third-party administrators should be able
to provide a high-cost claimant report that shows
every individual in the ACO population that has
incurred a claim amount over a certain threshold,
such as $50,000. The report normally includes a
member identifier, a claim amount, and primary
diagnoses. The ACO can then ascertain if there
were any shock claims or long-term illnesses during
the base period.
In the case of a shock claim – for example, the
birth of sextuplets – the ACO may want to adjust
for this data anomaly, especially if the probability
of a similar claim occurring is low. One approach
to smooth out the shock claim is to remove all but,
say the first $50,000 of claims from the baseline
claims PMPM. The percentage reduction in the
baseline claims PMPM should then be applied to
the resulting projected budget claims PMPM. As
shown in Exhibit 3.10, $400,000 was removed from
the baseline allowed claims, which resulted in a
2.1 percent reduction in the claims PMPM. This
2.1 percent reduction should then be applied as
an adjustment factor of 0.979 (1-2.1 percent), to
arrive at the final CY 2008 allowed claims PMPM of
$323.75.
EXHIBIT 3.10. SHOCK CLAIM ADJUSTMENT
Baseline Allowed Claims
Member Months
Allowed Claims PMPM
CY 2008
$19,425,000
60,000
$323.75
Shock Claim
Claim Adjustment
$450,000
($400,000)
Adjusted Baseline Claims
Adjusted Baseline PMPM
Percentage Adjustment
$19,025,000
$317.08
-2.1%
If the high-cost claimant report indicates some
long-term illnesses, the ACO may want to add back
in any claim amounts over $50,000 to account for
them. The exercise is similar to the shock claim
adjustment shown in Exhibit 3.10, except that
claims in excess of $50,000 are being added in
rather than taken out. The resulting adjustment
would be an increase to the projected cost.
The ACO and the payer could also set up a full
or partial reinsurance type of arrangement, where
the expected costs associated with the highcost individuals in excess of a given threshold
are converted into a PMPM and expressed as an
adjustment factor to be applied to the projected
cost. The level of the reinsurance threshold could
vary with the size of the ACO membership (e.g.,
$30,000 for 15,000 members and $50,000 for
25,000 members). Smaller ACOs may warrant
lower thresholds, as there is less ability to spread
the risks associated with the high-cost cases.
This arrangement would help to smooth out the
costs associated with the high-cost but relatively
infrequent cases, such as solid organ transplants
and very expensive low-birth weight babies.
Summary of Adjustments
In this illustration, the baseline claims costs are
adjusted by a trend adjustment, an age/sex
adjustment, and a high-cost claimant adjustment.
Exhibit 3.11 summarizes the adjustments to the
baseline and the resulting CY 2010 spending
projection.
EXHIBIT 3.11. CY 2010 SPENDING PROJECTION
CY 10 PMPM Spending Projection
CY 08 Allowed Claims PMPM
Age/Sex Adjustment
High Cost Claimant Adjustment
CY 09 Trend Factor
CY 09 Allowed Claims PMPM
CY 10 Trend Factor
CY 10 Allowed Claims PMPM
$323.75
1.048
0.979
1.0994
$365.18
1.1009
$402.01
Note: This example assumes the age/sex adjustment
and the high cost claimant adjustment were 1.0 for the
historical period.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 46
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
The adjusted projected spending ($402.01 PMPM
in our example) would be used to calculate the
spending benchmark for an ACO beginning in CY
2010. Its calculation is discussed below, under
the Payment Incentives section. The spending
benchmark is used to measure the performance
of the ACO spending growth. When the ACO
achieves the quality goals and has lower spending
levels relative to the benchmark, it will receive the
incentive payments.
in developing the spending benchmark for future
years. A “benchmark to actual” report compares
the benchmark claims PMPMs to the actual claims
PMPMs by service category. If there is a significant
variance, one can drill down by the cost and
utilization breakdown between the benchmark costs
and actual costs. A member demographic report
can indicate the risk profile of the population and
how it is changing. Finally, a high-cost claimant
report can identify any shock claims or long-term
illnesses that were not planned for in the budget.
Sensitivity Analysis
It is important to perform sensitivity analyses on the
assumptions used in setting the budget to gain an
understanding of the potential range of variations.
The results of the sensitivity analysis will eliminate
many surprises when the actual claims costs are
not within the calculated budget. Additionally,
it will draw attention to which assumptions are
more sensitive than others to the overall claims
projections. Since the trend assumptions are
the most important factor when determining the
overall spending, the sensitivity analysis should
be focused on this assumption. In the example
used throughout, the ACO-specific trend was used
in the budget projections, meaning that the data
specific to the ACO were reviewed and used to
determine a trend assumption. Another approach
may be to review the national data and determine
a national trend assumption. If the national trends
are significantly different from the ACO trends, it
is important to understand why. There could be
specific reasons for the variation, and the ACO
budget should account for this.
Performance Monitoring
Once the spending benchmark, or target, is
finalized, the ACO should design a set of reports to
monitor the actual costs and compare them to the
benchmarks. At a minimum, the ACO should review
these reports on a quarterly basis. Information
from these reports can highlight problem areas
and allow the ACO to intervene and manage the
claim costs. These reports can also assist the ACO
Sometimes, the variances between the benchmark
and the actual claims are due to random
fluctuations, particularly when one is looking at
monthly data or comparisons by service categories.
A rolling three-month average or a rolling six-month
average would be more stable than monthly data.
Pro Forma Budget Development
It is expected that ACOs will use a variety of welldefined interventions to improve the quality of care
provided to their enrollees while reducing the cost of
care. As these interventions are likely to cost ACOs
money and resources to implement, it is critical
that ACOs develop a pro forma budget detailing
their expected costs and savings. This step is
also necessary for ACOs in projecting the total
amount of incentive payments they could potentially
achieve.
For example, suppose an ACO implements a
patient navigator program for patients with chronic
conditions that are recently discharged from
a hospital. This program increases the ACO’s
administrative costs by $50,000 and reduces the
claim costs through reduced readmissions by
$100,000. The ACO would need to share in at least
50 percent of the claims savings ($50,000) with the
payer in order to break even on the investment for
the new program in one year.
Due to the variation of each intervention, it is difficult
to offer specific guidance in the development of a
pro forma budget. In general, each of the services
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 47
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
in the budget needs to be considered for a potential
change due to the intervention. Both additional
expenses, as well as potential cost savings, will
need to be considered for the contract period.
As many of the interventions may be newly
developed with limited or no historical evidence for
effectiveness, pro forma budgets may essentially
be best guesses or extrapolations from limited
experiences. Therefore, these estimates should
continually be updated as part of the regular
budget monitoring process, using any new and
actual financial information that becomes available.
Monthly or quarterly updates will help assure
that projections are as close to being accurate as
possible.
Adjusting Benchmarks Over Time
Providers and payers are likely to consider multiyear ACO contracts. Multi-year agreements will
help foster long-term, lasting improvement impacts.
The PGP demonstration began as a five-year
program, as did the MHCQ demonstration. ACA
requires a three-year commitment for providers
interested in joining the Medicare Shared Savings
program beginning in 2012. Private sector
initiatives such as the Blue Cross Clue Shield of
Massachusetts Alternative Quality Contract (BCBS
MA AQC) and the Brookings-Dartmouth ACO pilot
program also require multi-year participation.
Given that there is still only limited evidence on
the budget impact of specific interventions, and
given that reliable baseline data used to establish
benchmarks is not always available, participating
providers and payers should consider reviewin –
and if necessary updating – benchmarks at regular,
predefined intervals. For example, the BCBS MA
AQC program rebases its budget benchmarks
annually.
There are a number of reasons why both payers
and providers may want to consider rebasing
benchmarks based on more recent data. For
example, recent data could suggest that exogenous
factors – such as an economic recession or
development of new medical technology – could
cause future spending to diverge significantly from
past trends, which could make current benchmarks
less reflective of expected spending under ceteris
paribus assumptions. Additionally, payers may
want to recognize increases in efficiency and
consider making the spending target more difficult
for ACOs to achieve in proceeding years.
How payers and providers choose to negotiate
whether and how benchmarks should be updated
will require important consideration. Payers and
providers should anticipate ways to make rebasing
adjustments in their initial contracts. For instance,
if payers and plans could agree on a particular
updated factor – such as zero percent, general
inflation, or a national or regional growth rate – they
could also agree to a partial adjustment to past
trends based on national or regional trends, or some
sort of rolling average approach.
3.3: PAYMENT MODELS AND INCENTIVES
In general, ACOs would be eligible to receive
financial performance incentive payments if the
actual spending on their patients is below the
benchmark, provided that the quality performance
standards are met or exceeded. In this section,
we discuss some of the key issues that must be
considered when developing such a payment
system.
There are many different payment models available
to an ACO. They range from a “one-sided” shared
savings within an FFS environment, to a range of
limited or substantial capitation arrangements with
quality bonuses, as shown in Exhibit 3.12.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 48
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
EXHIBIT 3.12. CONTINUUM OF PAYMENT METHODOLOGIES
Level 1 ACO
•
•
•
•
•
•
•
“One-sided” or Asymmetric
Model
Continue operating under
current insurance contracts
and coverage models (e.g.,
FFS reimbursement)
Provider groups have no risk
for losses if spending exceeds
budget benchmarks
Provider groups receive
relatively modest percentage
of any earned savings due to
limited risk
Most incremental approach
with least barriers to entry
Minimal requirements for
health IT infrastructure and
governance structure
Limited to no experience with
alternative to FFS payments
Attractive to new entities, riskadverse providers, or entities
with limited organizational
capacity or experience
coordinating care across
providers
Level 2 ACO
•
•
•
•
•
“Two-sided” or Symmetric
Model
Payment still predominantly
FFS, but may include some
alternative systems such as
bundled payments
Provider groups are at risk for
losses if spending exceeds
projected benchmarks
Increased incentive for
providers to decrease costs
due to risk of losses
Provider groups receive higher
percentage of any earned
shared savings in line with
increased risk
Attractive to providers with
some health IT infrastructure,
care coordination capability
and demonstrated track
record managing care
Level 3 ACO
Partial Capitation Model
• Provider groups receive mix
•
•
•
•
•
of FFS and prospective fixed
payment
Provider groups share costs
if expenditures exceed the
projected benchmarks; may
even have first dollar risk under
a global budget model
If successful at meeting budget
and performance targets,
greater financial incentives
If ACO exceeds target, more
risk means greater financial
downside
Only appropriate for providers
with robust health IT
infrastructure, demonstrated
track record in finances and
quality
May need to comply with state
regulatory oversight to take on
financial risk
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 49
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
Ideally, an ACO will transition over time to
implement payment models with increasingly
more risk, resulting in the ability to retain a
higher percentage of shared savings for being
more accountable for cost and quality, as shown
in Exhibit 3.12. As organizations grow more
comfortable managing risk and become more
clinically integrated through new care strategies
and more sophisticated health IT, they can progress
from a “one-sided” payment model (Level 1) to a
“two-sided” payment model (Level 2) where an
organization is liable to share in the costs if their
spending exceeds their projected benchmarks.
Once an ACO becomes experienced and
comfortable managing the added risk of a
symmetric payment model, they could then advance
to a partially capitated payment model, where some
services remain on a FFS reimbursement basis,
while others are reimbursed by a fixed amount per
patient. For example, an ACO could received a
pre-paid risk-adjusted capitated amount to cover
all ambulatory care services for its ACO patients,
and then have a bonus/withhold payment (i.e., a
shared-savings model) based on the traditional
discharge level reimbursement rates and spending
targets for their inpatient services. Progressively
moving towards a payment model that accepts
more risk will help further incentivize an ACO to be
more accountable for the cost and quality of their
provided care.
Regardless of the payment model level an
ACO is operating in, a necessary step to any
financial performance model is to calculate the
actual expenditures incurred during the financial
performance period. In our illustrative example
below, we define the financial performance period
as a 12-month period. The period should be long
enough to obtain a large enough sample of claims
in order to estimate the representative spending
amounts. Periods longer than a year run the risk of
displacing the financial incentive from the time when
services are being performed.
The methodology for calculating the actual
spending amount is similar to the above discussion
on calculating the baseline incurred claims. It may
be necessary to adjust the data for anomalies such
as high-cost claimants. Also, it may be necessary
to determine if there has been a shift in the risk
profile of the population.
Once the actual spending amount has been
calculated, and assuming quality benchmarks are
met or exceeded, a comparison to the benchmark
spending would determine if the ACO is eligible
for the bonus payments. For incentive payment
purposes, the comparison between the actual
and the spending benchmark is performed on the
aggregate claims level, not by the service category
level.
Three key features to consider when formulating
the bonus payments are (1) the use of a savings
threshold, (2) the percentage of the savings to
be shared, and (3) whether it should accept an
asymmetric or symmetric risk model. These
features can essentially be used as tools to balance
the levels of risk that providers and payers are
prepared to take in their ACO efforts.
Savings Threshold
In Exhibit 3.13 below, the savings threshold is
two percent, meaning that financial performance
bonuses are only distributed if the actual spending
growth is lower than the projected spending
growth by more than two percent. The target – or
the benchmark spending growth – is therefore
calculated as the projected spending growth less
two percent.
Small fluctuations in actual spending amounts are
to be expected. The use of the two percent savings
threshold is meant to avoid making bonus payments
for savings that essentially happen by chance.
Thresholds can also be used in two-sided models to
protect providers from financial risk against random
losses.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 50
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
ACOs may want to consider setting the level of
the threshold as a factor of the size of the ACO
membership. For example, smaller ACOs are
more susceptible to larger variation and therefore
a higher threshold, such as four percent, may
be more appropriate. In fact, ACA establishes a
minimum Medicare beneficiary panel of 5,000 for
ACO participation in the Shared Savings program
beginning in 2012. The greater the population size,
the less variation that would be expected and the
more predictable spending becomes, so a smaller
threshold would be required.
The level of the threshold should also vary with
the level of risk that the providers are willing to
take. For instance, under a Level 3 global budget
approach, providers may participate in first-dollar
savings, as well as first-dollar losses.
Percentage of the Savings to be Shared
In our illustrative example, 50 percent of the
savings between the actual spending growth and
benchmark spending growth is eligible to be shared
with the ACOs. The other 50 percent is retained by
the payer as savings. ACOs may want to select a
different distribution level such as 80/20, with 80
percent going to the providers and 20 percent to
the payer. Considerations that may factor into this
decision are (1) the amount of expected savings that
an ACO could feasibly accrue, and (2) the cost of
the cost-savings interventions that the ACO would
implement. Ideally, the bonus payments should
at least offset the intervention costs. Additionally,
the percentage should also vary with the payment
model used. Models where the providers have
greater financial risk should be associated with
higher percentages for the providers to balance the
risk-reward relationship.
PAYMENT MODELS IN MEDICARE SHARED SAVINGS PROGRAMS
ACA legislates that savings thresholds and percentages be used for the Medicare Shared Savings program
beginning in 2012, the parameters of which have yet to be determined as of the writing of this document.
Legislation also suggests the potential for using both one-sided and two-sided models.
In the PGP demonstration, the shared savings threshold is two percent. PGPs qualify to receive up to
80 percent of the total savings based on how they perform on quality benchmarks. The other 20 percent
reflects savings to the Medicare program.
In addition to the opportunity to share in any measurable savings, PGPs are also somewhat accountable
for potential losses. More specifically, if spending exceeds the spending benchmark by two percent in a
performance period, the excess spending is carried forward as losses and are deducted from any bonuses
earned in future years.
The shared savings thresholds differ for the two sites in the MHCQ demonstration that have a sharedsavings model based on total patients costs. For the Indiana Health Information Exchange, the savings
threshold is set by a formula that is dependent upon the size of the intervention and comparison
populations. Initial estimates place the threshold at around 1.5 percent. For the North Carolina
Community Care Networks, Inc., the threshold is set at 2.9 percent for the first two performance years
when a relatively small panel of dual-eligible beneficiaries is included in the demonstration. From year
three to five, the total savings threshold is reduced to 1.5 percent, as the general Medicare FFS population
is included in the demonstration.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 51
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
Asymmetric or Symmetric Risk
savings model. In this example, the projected
allowed PMPM is $402 in 2010, followed by
$434 and $469 in 2011 and 2012, respectively.
After taking into account the two percent saving
threshold, the PMPM targets (or benchmark) for the
shared-savings are calculated to be $394, $426,
and $460 for 2010, 2011, and 2012, respectively.
In our example, we assume that the ACO can only
share in the savings if its actual spending growth is
below the benchmark spending growth and that the
payer assumes full costs of the spending in excess
of the benchmarks. This scenario is known as a
one-sided risk or asymmetric payment model.
In 2010, the ACO attains an actual PMPM amount
of $394. In this case, there is no shared savings
because the ACO did not reduce the costs by more
than the two percent threshold.
An alternative is a two-sided or symmetric risk
situation, where the ACO shares in the costs if
they exceed the benchmark spending amounts.
A symmetric risk model may provide stronger
incentives for the ACO providers to achieve more
efficient care, but could also deter many providers
from deciding to participate in an ACO payment
model.
In 2011 and 2012, the ACO is able to reduce the
costs by three and five percent, respectively.
Therefore, shared savings is achieved in both
years. In this example, the ACO would receive
the incentive payments of $2.50 (50 percent of $5)
PMPM in 2011, and $7.50 PMPM (50 percent of
$15) in 2012.
Example of A Shared Savings Model
Exhibit 3.13 provides an example of how these
payments would work in a one-sided shared-
EXHIBIT 3.13. EXAMPLE OF A ONE-SIDED ACO PAYMENT MODEL
$470
$450
$430
$410
$390
2% Threshold
Projected = $469
Target = $460
Shared Savings
$9
Actual Spending
Projected= $402
Target = $394
$8
Projected = $434
Target = $426
$8
$15
$5
$445
$0
$421
$394
$370
2010
2011
2012
Note: Numbers may not add up due to rounding. The target represents the projected spending less
the 2 percent savings threshold.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 52
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
As mentioned above, the one-sided model is only
one of the methods for implementing an ACO
financial performance payment. An advantage of
this model is that it demands the least disruption
in the current administrative and delivery systems;
however, a disadvantage is that the payments are
still tied to a FFS system, which provides incentives
for overutilization of health care services. It also
does not provide additional flexibility for ACOs to
alter the way services are reimbursed. For example,
high-value services, such as preventative care
that have low reimbursement rates, may still be
underutilized.
There is no single ACO model that is the best
for all the situations, as each has its advantages
and disadvantages. As more ACOs begin
implementation, it will be important to learn what
works and doesn’t work. Given the wide variation
in local circumstances, such as organizational and
governance structure, experience with previous risk,
and historical spending trends, it can be expected
that there will also be variation in the ACO payment
models that are being utilized.
3.4: DISTRIBUTION OF SHARED SAVINGS
The distribution of the shared-savings bonuses will
be an ACO-specific decision. ACOs will need to
consider the incentives necessary to motivate the
providers in making the required practice pattern
changes.
In the discussion below, we lay out three
potential “pools” for distribution. In general, the
incentive pools could focus on rewarding either
those providers making the key practice pattern
changes or those affected by the changes. The
methodology of the incentive pool allocation should
be established up front as part of the process of
organizing the ACO. The pools described below
are just examples of how this allocation could
be accomplished. The portion of the aggregate
incentive funds directed to the pools will vary
by the individual ACO and by year of operation.
For example, ACOs that are established by fully
integrated delivery systems may already have
internal financial incentives established with
participating physicians and other providers. In
this case, any shared savings bonuses could
presumably be used by the ACO to invest in further
care improvements or for other purposes.
Shared Savings to Offset Revenue Reduction
Some of the ACO partners may see a significant
reduction in revenue due to the change in practice
patterns. An ACO may choose to use a portion of
its shared savings to partially compensate providers
who are affected by these changes, such as
hospitals or some specialists.
It should be noted that while the per patient per year
revenue amount may decline for some providers
such as hospitals and specialists, not all hospitals
and specialists will experience an overall revenue
reduction, as the reduction could be offset by
ACO’s market share expansion and patient volume
increase. In evaluating the effects on particular
providers resulting from practice pattern changes, it
is also important to focus on both “top line” revenue
and “bottom line” net earnings. If revenues are
reduced, but costs are reduced by a greater factor,
profitability can be increased. Shared savings
allocations should take into account cost savings
that help offset the revenue reductions.
Shared Savings for Cost Savings
The objective of this pool is to recognize the core
physicians – those used for patient assignment –
who generate savings by improving management
of the patient’s health resources and the other
physicians who utilize episodic resources effectively.
Allocation of funds should consider the relative
contributions of each of the bonus eligible ACO
providers.
Incentive Pool for Return of Capital
Essentially, this pool is to distribute the remaining
net income back to the principle ACO investors or
partners based on their capital contributions to the
ACO.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 53
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
ACO Group Operating Cost
3.5: PERFORMANCE MEASUREMENT
The ACO is likely to incur organizational costs and
expenses from developing innovative methods
for coordinating care of the ACO patients. As
part of the feasibility analysis before starting
operations (e.g., a pro forma), these costs need to
be estimated, and ACOs need to have a plan for
funding them. These costs may need to be covered
before distributing shared savings. If there are no
savings or if the savings are not sufficient to pay for
the ACO costs, contributions by the ACO partners
will be required to cover the costs.
Another critical ACO design feature is the
implementation of a quality measurement strategy
to ensure the financial benefits of achieving
cost targets are contingent on meeting health
care quality performance targets. Performance
results provide ongoing information and feedback
to providers to help improve patient care, to
incorporate patient’s feedback and insights into care
delivery strategies, and to assure the public that any
cost savings coincide with improvements in care.
“Spillover Effects” To Non-ACO Patients
Physicians and other providers tend to treat all
patients in a similar manner of practice. To the
extent that ACOs succeed in generating savings,
total revenues paid for health services will decline
and ACO incentive bonus payments may be used to
mitigate this reduced revenue.
To the extent that non-participating payers benefit
from the practice pattern changes adopted by ACO
providers, their costs may drop without sharing the
savings with the ACO providers, creating a “free
rider” problem. Therefore, it is imperative for an
ACO to try and involve multiple payers, constituting
the majority of the ACO’s served patients.
In this next section, we discuss several major
challenges to implementing a comprehensive
performance measurement system. Initially, many
organizations will only be able to track a basic set
of measures, but their ability to track patient-centric
results is assumed to improve significantly over
time. Summarized below in Exhibit 3.14 are some
of the key components and considerations involved
with measuring health care quality in the ACO
framework.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 54
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
EXHIBIT 3.14. KEY CONSIDERATIONS FOR MEASURING HEALTH CARE QUALITY
Selecting
Measures
Measures should track the results along the continuum of care, covering a wide range of
services and a broad range of quality of care goals, including care coordination, population
health, overuse, and patient engagements. Measures should be well established and
preferably nationally endorsed.
Data
Sources and
Collection
Quality measurement relies on multiple data sources. Sources include administrative
claims data (e.g., medical and pharmacy), laboratory and clinical records, electronic
medical/health records, registries and patient-generated information, such as patient
surveys.
Standard Set
of Measures
We propose a starter set of standard ACO measures that are based on administrative
claims data. We discuss a pathway to expand performance measurement to be based on
clinical and other data sources over time.
Targets
Under the accountability-payment framework, financial incentives are contingent on
providers meeting or exceeding performance targets. We describe a general framework of
tying performance to financial rewards for the ACO pilot sites.
Performance
Calculation
We discuss various methods of calculating performance results, including the use of risk
adjustment and composite scores.
Validation of
Measures
Accuracy and consistency are both important. For accuracy, verification processes should
ensure all calculations are done in accordance with technical specifications. To evaluate
the effectiveness across ACOs, the validation process should verify data collection and
aggregation methods are implemented consistently.
Public
Reporting
A core principle of ACOs is to be accountable for the quality of care provided. As such,
public reporting of the quality performance is a key aspect of implementing an ACO quality
improvement program.
Consistency
with Other
Reforms
There is a wide range of payment reform initiatives, including expanded use of pay-forperformance programs, medical homes, and ACOs. Each of these requires the use of
performance measures. Having consistent or standardized measurements across these
initiatives would greatly assist in the evaluation and implementation of these programs.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 55
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
Staging the Implementation of
Performance Measures
We envision phasing in the implementation of
performance measurement to align with the ability
of accessing multiple data sources. In Exhibit 3.15,
we describe three phases of implementation. We
expect that over time, those at the beginning phase
will reach the advanced phase, where performance
measures can effectively address multiple priorities
spanning the continuum of care and are outcomeoriented. The various phases are described below.
• Basic Phase. ACOs with a “basic” health
IT infrastructure predominantly rely on
administrative data, with limited access by
providers in their ACOs. Health care quality
performance measures for these organizations
will be limited to those that can be computed
reliably using claims data.
• Advanced Phase. ACOs with an “advanced”
health IT infrastructure will have comprehensive
access to clinical data collected through their
widely deployed and interoperable ACO-wide
EHRs. Moreover, these ACOs also will have
the ability to directly collect patient-generated
information about their care experience.
These ACOs will be able to track and measure
patients’ outcomes and experience across
multiple care settings, such as inpatient care,
tertiary/long-term care, specialty outpatient
care, and primary care.
• It is expected that over time, with additional
incentives available through the Federal
Government and others, ACOs will quickly
expand their health IT infrastructure to
measure care quality more comprehensively,
and reap associated rewards of being able to
demonstrate superior quality and outcomes
along the care continuum.
• Intermediate Phase. ACOs with an
“intermediate” health IT infrastructure will
utilize clinical data in addition to the claims
data, particularly for primary care and chronic
care management purposes. These ACOs
may be able to routinely access and receive
electronic laboratory results from their
contracted laboratories. Some clinical data
may also be available from specific registries
(e.g., immunizations) maintained for their
organizations or from nationally-maintained
registries (e.g., interventional cardiology).
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 56
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
EXHIBIT 3.15. EXAMPLES OF POTENTIAL PERFORMANCE MEASURES FOR ACOS
WITH BASIC, INTERMEDIATE, OR ADVANCED HEALTH IT INFRASTRUCTURE
Advanced Phase:
Basic Phase:
PRIORITY
AREAS
Intermediate Phase: LIMITED CLINICAL
COMPREHENSIVE PATIENT-FOCUSED
CLAIMS-BASED MEASURES
AND SURVEY MEASURES
MEASURES
ACOs have access to medical,
ACOs use specific clinical data (e.g.,
ACOs use more complete clinical data
pharmacy, and laboratory
electronic laboratory results) and limited
(electronic records, registries, etc.) and
claims from payers
survey data
robust patient-generated data (Health
Risk Appraisals, functional status)
•
•
Cancer Care
Screenings
Care
•
Effectiveness/
and H1c tests, eye
Population
Health
Diabetes care (LDL
exams, etc.)
•
•
•
Coronary Artery
Disease care (LDL
test)
Immunization rates for children
summary score (BMI, blood
and adolescents
pressure, cholesterol, smoking,
Patients with diabetes whose
exercise, alcohol)
blood sugar (H1c) are in control
•
•
Stage-specific quality of life
Patients with diabetes or ischemic
and functional outcomes for
vascular disease whose lipids
common cancers
(LDL) are in control
•
Comprehensive health risk
•
Quality of life and functional
Patients with hypertension whose
outcomes for common
blood pressure are in control
conditions (e.g., AMI, hip
replacement, diabetes)
•
High-risk medication
•
for the elderly
Safety
•
Appropriate testing for
•
“Never events” in hospitals
patients using high-
Hospital infection and risk
adjusted mortality rates
•
Outpatient medication errors
•
Care plans – patient activation
risk medications
and engagement in chronic/
other conditions
•
Patient
Engagement
•
----
•
Physician instructions understood
•
Preference sensitive conditions
(CAHPS)
– level of information
Care received when needed
communicated regarding patient
(CAHPS)
choice (e.g., knee surgery)
•
Patient preferences – adherence
to design and execution of care
plan (e.g., advanced directives)
•
Imaging for low back
•
pain (in absence of
“red flags”) during first
Overuse/
Efficiency
•
•
•
Episode-based resource
Episode-based resource use –
use – linked to quality of
30 days
linked to quality measures for
life, functional and patient
Inappropriate
common medical (e.g., diabetes,
engagement measures for
antibiotic prescribing
AMI) and common surgical
common medical (e.g., diabetes,
Utilization rates of
conditions (e.g., hip replacement)
AMI) and surgical conditions
select services (e.g.,
(e.g., hip replacement)
C-section)
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 57
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
Selecting Measures
While primary care is a critical element of the ACO
model, ACOs are accountable for all care delivered
to their patients, including specialty care and
services provided in non-ambulatory settings –
inpatient hospitals, home health, and skilled nursing
homes – regardless of whether or not the services
are delivered by ACO providers. Developing a
broad set of measures covering the full spectrum
of these services and addressing multiple priorities
will provide incentives for providers to focus on
population health as well as interventions to improve
care.
The above characteristics are very helpful for
providing momentum for the quality improvement
and financial reforms implemented by the ACO.
To reduce overall measurement burden, selected
measures should ideally be aligned with the use of
measures for other purposes, policy objectives, and
payment reform initiatives across the public and
private sector.
Use Well Established and Validated Measures
Finally, selected measures should rely on nationally
consistent specifications (e.g., as contained in
NQF endorsed performance measures) as well as
nationally consistent rules for data collection and
aggregation (e.g., as being defined through efforts
of the Quality Alliance Steering Committee).5
Consensus around selected measures can aid
with acceptance of measures by all stakeholders,
including participating providers in the ACOs.
Entities such as the National Quality Forum (NQF)
have endorsed useful performance measures for
selected priorities.
While the NQF has endorsed more than 500
measures, many gaps covering critical priorities
for care improvement – such as care coordination,
proximal and long-term outcomes for many critical
conditions – remain. In addition, detailed data
THE NATIONAL QUALITY FORUM CONSENSUS PROCESS
Using its multi-stage ‘consensus development process’ – designed to call for input and take into considerations
the interests of stakeholder groups from across the health care industry – NQF fosters consensus among a
wide variety of stakeholders around specific standards that can be used to measure and publicly report health
care quality. From this, the NQF has developed a portfolio of endorsed performance measures that can be
used to measure and quantify health care processes, outcomes, patient perceptions, and organizational
structure and/or systems that are associated with the ability to provide high-quality care.
These endorsement processes ensure that
measures:
• Characterize important processes and
outcomes of care;
• Produce scientifically sound and statistically
reliable results;
• Are feasible to collect efficiently; and
• Are deemed useful by payers, consumers/
patients, and others for taking action.
collection and aggregation details are not available
for many measures. With the recent passage
of health care reform legislation it is expected
that significant federal funds will be invested to
develop and test needed performance metrics
for high priority areas. It is expected that these
newly developed metrics and their associated data
collection processes will be integrated rapidly for
tracking the ACO performance. Where relevant
“endorsed” measures are not available, it may be
possible to use measures that are in process for
multistakeholder endorsement.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 58
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
Data Collection
Performance measurement efforts are supported
by a number of different data sources, including
administrative data, electronic clinical data, and
patient-generated data (e.g., through surveys).
Some of the performance measures can be
calculated using data from a single source, and
other measures require data elements from multiple
sources. There are three main types of data
sources that should be accessible for ACO quality
measurement.
1. Administrative data – enrollment, as well as
medical and pharmacy claims. Enrollment and
administrative claims data can be obtained from
payers. Claims data covers large populations.
Several performance measures can be reliably
calculated using administrative claims data,
whereas other performance measures often rely
on clinical data that is not as readily available as
claims data. Therefore, measures for the initial
phase of ACO implementation are often based on
administrative data.
2. Electronic clinical data. Certain performance
measures require rich clinical data – from electronic
medical or health records, laboratories, and standalone clinical data systems such as clinical registries
– and cannot be computed using only administrative
data. The availability and ability to access such
clinical data is expected to significantly increase
over the next few years. Many organizations are
now able to access and “process” clinical data for
measurement purposes (e.g., receipt and integration
of laboratory results provided by contracted
laboratories). However, the lack of interoperability
and data exchange has significantly hampered
the utility of these data to date. Depending on
the ability to effectively exchange information and
overcome other constraints, electronic medical
record systems may not be capturing the data from
multiple specialists who have rendered care to
patients.
3. Patient-generated information (e.g., care
experience, health and functional status). In
addition to claims-based and clinically-enhanced
data, other data are generated directly by patients.
Examples of such data include patient assessments
of care experience, patient understanding of care
instructions/plans, and patient health and functional
status. Such information is typically collected
through surveys deployed in the clinical setting
(e.g., functional status) or by other organizations.
The Consumer Assessment of Healthcare Providers
and Systems (CAHPS®) is the leading tool for
measuring patient experience. It has been carefully
constructed and tested, endorsed by the NQF, and
nationally accepted by various stakeholders.
Efficient and effective data collection mechanisms
are still being developed. While this information
is largely collected through paper-and-pencil
surveys today, alternative data collection methods
– such as through electronic survey kiosks – are
being tested to allow for quicker and less costly
integration of data for care improvement and
performance measurement purposes. Several large
measurement initiatives targeting medical groups
(e.g., through the Integrated Healthcare Association
in California and Massachusetts Quality Partners)
have been implemented through a collaborative
data collection model and are yielding patientgenerated results at modest costs.
Standard Set of Measures
ACOs participating in the Brookings-Dartmouth
ACO pilot program have agreed to implement an
initial set of standardized measures that will be
produced in a consistent fashion across all payers
and sites. Several criteria were considered in
selecting the starter measures. First, the program
sought to identify a nationally consistent measure
set based on widely accepted and endorsed
measures that the vast majority of payers and
providers are familiar implementing. Second,
the measures should cover key aspects of
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 59
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
primary care, preventive care, and chronic care
services. Lastly, in recognition of the health IT
infrastructure capability, the program agreed to rely
on administrative data only for the computation of
these measures, so that they could be implemented
within the first performance year of any ACO. The
thirteen measures in Exhibit 3.16 represent the
starter set of measures that will be used across the
pilot sites.
Each of the measures in the starter set can be
computed using administrative data and has
been endorsed by the NQF. Also, because these
measures are part of the Healthcare Effectiveness
Data and Information Set (HEDIS), payers are
experienced in calculating them at the health plan
level.
This initial set of measures focuses on key aspects
of primary care and chronic care management.
In the future, this set will expand to incorporate
measures that cover services more comprehensively
across the care continuum and settings. For
hospital quality performance assessment, there are
several nationally recognized sources, including
the Agency for Healthcare Research and Quality
(HCAHPS® and Quality Indicators), the Centers for
Medicare & Medicaid Services (Quality Measures
Management Information System and Hospital
Compare), the Hospital Quality Alliance, The Joint
Commission, and Leapfrog. The program strives to
integrate these measures into the expanded set of
measures as these pilot sites mature.
EXHIBIT 3.16. STARTER SET OF MEASURES
Priority Areas a/
Initial Measures
Overuse
Use of imaging studies for low back pain
Overuse
Appropriate testing for children with pharyngitis
Overuse
Avoidance of antibiotic treatment for adults with acute bronchitis
Overuse
Appropriate treatment for children with upper respiratory infection
(URI)
Population Health
Breast cancer screening
Population Health
Cervical cancer screening
Population Health
Colorectal cancer screening
Population Health
Diabetes: HbA1c management (testing)
Population Health
Diabetes: cholesterol management (testing)
Population Health
Cholesterol management for patients with cardiovascular conditions
(testing)
Population Health
Use of appropriate medications for people with asthma
Population Health
Persistence of Beta-Blocker treatment after a heart attack
Safety
Annual monitoring for patients on persistent medications
a/
The priority areas identified by the National Quality Forum.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 60
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
ACOs that are able to produce performance results
and outcomes above and beyond the measures
listed in the starter set are encouraged to do so.
The Brookings-Dartmouth collaborators also plan
to build on the starter measures by implementing
clinically enhanced performance measures, which
rely on data gleaned from clinical data systems
in conjunction with administrative claims data.
Candidates for additional measures are listed in
Exhibit 3.17. Prior to their full adoption, they will
be tested to ensure computation is feasible and
results are reliable and valid. The future phases
will include outcome measures, measures covering
non-ambulatory services, and measures addressing
critical priorities such as patient engagement and
care coordination.
In addition to these quality measures, ACOs
will benefit by including measures of health
care utilization, such as inpatient length-ofstay, emergency room utilization, and use of
generics. These measures can provide insight into
potentially unnecessary use patterns that could be
ameliorated with access management or chronic
care management. As the health IT infrastructure of
ACOs mature and more potentially useful measures
become available, it will be necessary to ensure that
required and optional performance measures are
updated on a regular basis.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 61
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
EXHIBIT 3.17. POTENTIAL ACO QUALITY MEASURES USING CLINICALLY ENHANCED DATA
Measure
Details
HbA1C Control - Percentage of adult patients with diabetes who had HbA1c control (<8.0
percent).
Diabetes
Measures
LDL Control - Percentage of adult patients with diabetes with most recent LDL-C <130 mg/
dL; LDL-C <100 mg/dL.
BP Control - Percentage of patient visits with blood pressure measurement recorded, for
patients with diagnosed hypertension.
Eye Exam - Percentage of adult patients with diabetes who received a dilated eye exam.
Kidney Disease Screen - Percentage of adult patients with diabetes who had at least one
test for microalbumin or who had evidence of medical attention for existing nephropathy.
Aspirin Prophylaxis - Percentage of diabetes patients who are taking aspirin on a daily
basis.
CAD Measures
Drug therapy for lowering LDL - Percentage of patients with CAD who were prescribed a
lipid – lowering therapy.
Aspirin Prophylaxis - Percentage of vascular disease patients who are taking aspirin on a
daily basis.
Persistence of Beta-Blocker Treatment after a Heart Attack - The percentage of
patients 18 years of age and older during the measurement year who were hospitalized
and discharged alive, from July 1 of the year prior to the measurement year to June 30 of
the measurement year, with a diagnosis of acute myocardial infarction (AMI) and received
persistent beta-blocker treatment for six months after discharge.
CHF Measures
Beta-Blocker Treatment after a Heart Attack - The percentage of patients 35 years of age
and older during the measurement year, who were hospitalized and discharged alive from
January 1 – December 24 of the measurement year, with a diagnosis of AMI and received an
ambulatory prescription for beta-blockers upon discharge.
IVD: Blood Pressure Management - The percentage of patients 18 years of age and older
who had blood pressure <140/90 mmHg.
IVD: LDL-C <100 - Percentage of patients 18 years and older with IVD whose most recent
LDL-C screening <100.
Hypertension
Measure
BP Control - The percentage of patients 18-85 years of age who had a diagnosis of
hypertension (HTN) and whose blood pressure (BP) was adequately controlled (<140/90)
during the measurement year.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 62
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
Advising Smokers To Quit - The number of patients in the denominator who responded to
the survey and indicated they had received advice to quit smoking from a doctor or other
health provider, during the measurement year.
Discussing Smoking Cessation Medication - The number of patients in the denominator
who responded to the survey and indicated that medication to assist with quitting smoking
was recommended or discussed.
Discussing Smoking Cessation Strategies - The number of patients in the denominator
who responded to the survey and indicated that their doctor or health care provider
recommended or discussed methods and strategies other than medication, to assist with
quitting smoking.
Population
Health
Measures
Childhood immunizations - Percentage of children two years of age who had four
diphtheria, tetanus and acellular pertussis (DTaP), three polio (IPV), one measles, mumps
and rubella (MMR), three H influenza type B (HiB), three hepatitis B, one chicken pox vaccine
(VZV),four pneumococcal conjugate vaccines (PCV), two hepatitis A (Hep A), two or three
rotavirus vaccine (RV) and two influenza (flu) vaccines by their second birthday. The last
three were added in 2010.
Adult Body Mass Index (BMI) Assessment - Percentage of patients 18-74 years old who
had an outpatient visit and who had their BMI documented during the measurement year.
BMI records / Children (WCC) - Percentage of patients 2-17 years old who had
an outpatient visit with a PCP or PB/GYN and who had evidence of BMI percentile
documentation, counseling for nutrition and counseling for physical activity during the
measurement year. Because BMI norms for youth vary with age and gender, this measure
evaluates whether BMI percentile is assessed rather than the absolute BMI value.
Flu Shots for Adults Ages 50-64 - The percentage of patients 50-64 years of age as of
September 1 of the measurement year who received an influenza vaccination.
Influenza vaccine - Percentage of patients 65 years of age and older as of January 1 of the
measurement year who received an influenza vaccination.
Pneumovax vaccine - Percentage of patients with pneumonia, age 65 and older, who have
ever received the pneumococcal vaccine.
Medication reconciliation - Percentage of discharges from January 1 – December 1 of
the measurement year for patients 65 years of age and older, for whom medications were
reconciled on or within 30 days of discharge.
Establishing Performance Benchmarks
and Targets
There are several ways to take performance
metrics into account under an ACO accountability
framework. Multiple models are currently in use
around the country tying performance attainment to
financial incentives.
We have laid out a basic framework for linking the
performance targets to shared savings based on the
starter set of measures. These principles could be
applied against the identified starter set of measures
as well as additional, desirable performance
measures that payers and ACOs find useful. The
basic framework is described below:
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 63
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
• Each performance measure will have an
associated threshold. A minimum level of
performance attainment (e.g., achieving the 50th
percentile of a national or regional distribution
of provider performance) could be required
to “earn” performance points, with more
points earned based on how far the minimum
threshold has been exceeded.
• A minimum number of points are needed across
the performance measure set in order for the
ACO to become eligible for shared savings. An
ACO could achieve a sufficient number of points
by significantly exceeding performance targets
for most but not all measures.
• In addition to – or instead of – earning points
by achieving certain performance levels, ACOs
could also earn points by demonstrating
significant improvement in their performance
as compared to the last time their performance
was measured. Because there are variations in
current performance across ACOs, the use of
improvement thresholds – such as reducing the
gap between current and national benchmark
performance by 10 percent – may be seen as
more equitable by some.
Details on how performance measurement will be
tied to bonus payments are being developed for
the Brookings-Dartmouth pilot sites and will be
discussed more broadly when additional information
is available.
Performance Calculation
There are several issues that need to be considered
when determining how to calculate quality measures
in the ACO framework. Foremost, one needs to
determine the eligible patient population for whom
the ACO providers assume accountability for the
costs and quality of care. In order to have reliable
and valid results, a sufficient population size is
required.
Below discusses several considerations for
performance measurement calculation, including
patient attribution, performance period, sample size,
composite measures, and risk adjustment.
Physician/Patient Attribution
Previously, we describe the patient attribution
process which aligns patients with ACOs, and these
patients become the basis for determining whether
the cost benchmarks are met. The same population
should be used to determine eligibility for quality
measurement purposes.
It is possible that not all patients in the ACO
would be included in each of the performance
measurement calculations, as not all patients will
have relevant conditions. Also, each measure may
require a certain period of enrollment (e.g., a “lookback” period) in order to calculate the measure.
Most of the performance measures in the starter
set require access to at least a full year of data to
determine if the quality criterion was met. However,
some measures may require less time, such as the
use of imaging studies for low back pain.
Performance Period
The same performance period should be used for
evaluating the quality measures and the financial
performance. For example, if financial performance
is evaluated on an annual basis, quality measures
should be calculated on the same annual basis
as well. This will permit quality and cost of care
to be evaluated together, helping to guard against
reductions in costs that reflect stinting on care.
Performance calculations should also be done
periodically to verify data integrity and to ensure
timely actions are taken when performance goals
are not met.
Reliability (Sample Size)
While other factors, such as measurement error, can
contribute to the issue of reliability, sufficient sample
size is a major determinant. There are few hard and
fast rules on an appropriate sample size, as different
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 64
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
stakeholders have different comfort levels on
margins of errors. For example, HEDIS measures
are only publicly reported when a health plan has at
least 30 observations for the denominator.
Composite measures, which are discussed below,
can also be used to deal with small sample size
issues. In this case, a larger sample size is obtained
by combining data from various measures.
Several options are available for ensuring the
sample size is sufficient to produce statistically
valid measures, including not using measures with
small populations, expanding the timeframe of the
measures, or aggregating data at the ACO level for
all participating payers. Not using certain measures
has the drawback of throwing out data that could
provide useful insights on the quality of care
provided by ACOs. Expanding the timeframe can
add to the sample size, but it would take longer for
ACOs to observe changes in the quality measures.
Composite Measures
Aggregating measures at the ACO level across
payers involves summing up the numerators and
denominators for each of the payers within an
ACO. The ACO-level measure would be the basis
for the comparisons to targets. In order to have
a valid measure at the ACO level, it is essential
that measures are calculated consistently across
all payers and all provider groups participating in
ACOs.
In considering whether to aggregate measures,
particularly for bonus-payment determination
purposes, providers and payers may choose to
make the decisions on a measure-by-measure
basis. That is, not all measures require aggregation
across payers to reach the minimum sample size
to be statistically reliable. Furthermore, ACOs may
choose different aggregation rules depending on
the market characteristics. For example, assume
that an ACO contracts with three payers in the
market with one payer having a dominant market
share. In this scenario, the large payer may not
need to aggregate its results, while the two smaller
payers may need to aggregate in order to achieve
statistically reliable results.
There are several ways to assess the value of the
measure being used for performance determination.
The simplest approach is to evaluate each measure
separately in determining whether the benchmark
is met. This means that each measure is given the
same weight of importance.
One alternative is to rely on composite measures.
Composite measures provide a comprehensive
view of the overall quality of care delivered by
combining individual measures into a single
measure. Composite measures offer several
advantages. It offers a simple way to identify and
reward providers who are delivering high-quality
care comprehensively. It also provides an easy
way to rank provider performance.6 Furthermore,
composite measures can improve the statistical
reliability of quality measures, which is a particular
problem when assessing care for small patient
panels or relatively rare outcomes measures. In
order for composite measures to not impede
actionability, it is recommended that underlying
details for all measures making up the composite be
provided to ACO providers.
Risk-Adjustment
Risk-adjustment of measures takes into
consideration the underlying risk and severity of
the patients, and supports more equitable and
consistent comparisons. Risk-adjusted measures
can provide meaningful comparisons across
different ACOs or over different periods for an ACO.
Since factors other than the quality of care
rendered – such as patients’ age, gender, severity
of illness, and comorbid conditions – can affect
patient outcomes, risk-adjusted measures allow the
analysis to focus on the quality of care rendered
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 65
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
and not the risk characteristic of the patient or mix
of patients. However, since only measurable and
reported risk factors can be accounted for, the
extent to which data can be risk adjusted is limited.
It is critical for measures to be risk adjusted
appropriately, especially when using national or
regional “norms” to determine targets. ACOs
should conform to the standard risk adjustment
methodologies applicable to endorsed measures.
Additional information on best practices for risk
adjustment is provided in the Part 3 Appendix.
Validation of Measure Results
Validation should be incorporated into the quality
measurement process. It is critical to ensure that
both the payers and providers are confident in
the measure results. Validation should ensure all
calculations are done in accordance with technical
specifications. Consistency in the implementation
of data collection/aggregation methods will assist
in evaluations of the effectiveness across various
ACOs. Consistency also allows aggregating the
measures to increase the sample size.
The validation process should ensure:
• Complete data are used in measure calculation;
• Programming algorithms are used accurately;
• Data checks (e.g., logic checks to identify if
calculated results are plausible) are available;
and,
• There is statistical precision and reliability.
A major component of data verification could be a
full data audit. Audit ensures the validity of reported
data and addresses data accuracy concerns. Audit
programs typically assure the measure results for
all parties are computed in accordance with predefined rules using comprehensive data.
Operationally, each ACO will need to determine how
and who will be conducting the validation process,
as well as verifying the measure calculation. One
option is to rely on the payers to perform these
functions for some of the measures. In Part 4, we
discuss the pros and cons an ACO should consider
in determining whether to perform these tasks
in-house or to contract with an external vendor.
Engaging a third party that is agreeable to both
the payer and the ACO can alleviate concerns of
gaming by either party. For ease of administrative
burden, ACOs may want to standardize the
processes of performance reporting and validation
across different payers.
Public Reporting
A core ACO principle is to be accountable for
the quality of care provided. As such, publically
reporting the measures is a key aspect of
implementing an ACO performance measurement
program. Publically reporting the measures is
intended to equip consumers with quality of care
information that would help them make more
informed decisions about their health care, while
encouraging hospitals and clinicians to improve the
quality of care provided to all patients.
Consistency is an important attribute of the
measures to ensure that they are comparable
across providers and understandable to consumers.
Quality Measurement in Other Reform Models
Quality measurement in accountability payment
systems should send consistent signals to providers
regarding priority areas for improvement. There is a
wide range of payment reform initiatives, including
expanded use of pay-for-performance programs
and medical homes, as well as ACOs. Each of
these requires the use of performance measures.
Having consistent and standardized measures
across these models will greatly assist in the
evaluation and implementation of these programs.
Consistency also needs to extend to incentive
payments made by CMS to providers, in promoting
the “meaningful use” of health IT, in particular EHRs.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 66
PART 3: ACCOUNTABILITY FOR PERFORMANCE | ACO TOOLKIT
The use of consistent measures across these
initiatives can help stimulate market-wide
movement towards accountability payment reforms
and accelerate improvement in the priority areas
identified by the community. Consistency should
also reduce the burden on providers and payers
involved with participating in these initiatives.
THE “MEANINGFUL USE” REGULATION
The American Recovery and Reinvestment Act of 2009 includes the Health Information Technology for
Economic and Clinical Health Act (HITECH), which established programs under Medicare and Medicaid
to provide incentive payments for the “meaningful use” of certified EHR technology. The HITECH act will
make up to $27 billion available in incentive payments for EHR use until 2020.
On July 13, 2010, the U.S. Department of Health and Human Services (DHHS) released a final regulation
defining “meaningful use” for EHRs to be applicable through 2012. The regulation includes multiple sets of
clinical quality measures and requires providers serving Medicare and Medicaid beneficiaries to report on
a minimum of: (1) three measures from their defined core set of measures, and (2) three measures from the
additional set of clinical measures. There are an additional set of 15 quality measures required for eligible
hospitals and critical access hospitals participating in Medicare and Medicaid programs. After 2012, the
DHHS plans to require more advanced clinical measures. The current set of measures can be found at:
http://www.ofr.gov/OFRUpload/OFRData/2010-17207_PI.pdf
ENDNOTES
1. There may be some providers in an ACO, such as anesthesiologists, that would not be used for attribution purposes.
See Part 2 for more details on individual provider roles within an ACO, including those that would be most likely to be
used for patient attribution or assignment.
2. Kautter J, Pope GC, Trisolini M, Grund S. 2007. Medicare Physician Group Practice Demonstration Design: Quality
and Efficiency Pay-for-Performance. Health Care Financing Review. 29(1): 17-20.
3. We note that there are some emerging ACO initiatives involving Medicaid beneficiaries. For example, the Colorado
Medicaid program is developing a regional accountability payment program. Also, the Patient Protection and Affordability
Act included funding for a pediatric Medicaid ACO demonstration.
4. Other categories could include therapy (e.g., occupational therapy, physical therapy, and speech therapy) or home
health services.
5. For information from the Quality Alliance Steering Committee on Data aggregation, please visit: http://www.
healthqualityalliance.org/hvhc-project/data-aggregation-and-integration
6. Shaller D, Sofaer S, Findlay S. Consumers and quality-driven health care: A call to action. Health Affairs. 2003;22(2):95101.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 67
Part 3 Appendix
Data and Health Care Analytics – Understanding Health Risk,
Measuring Performance, and Assessing Opportunities for Improvement
Dan Dunn
PART 3 APPENDIX | ACO TOOLKIT
PART 3 APPENDIX: DATA AND HEALTH CARE ANALYTICS – UNDERSTANDING HEALTH RISK, MEASURING
PERFORMANCE, AND ASSESSING OPPORTUNITIES FOR IMPROVEMENT
OVERVIEW
To assure chances of success, ACO managers
will need sophisticated data and analytic tools
in order to assess financial and clinical health
risk, identify care opportunities, and to measure
the cost and quality of the care delivered by the
organization. These tools encompass a wide
range of methodologies, and leverage different
types of clinical and financial data. Assessing
health risk, measuring performance, and assessing
opportunities for improvement underpin the three
key elements or principles of the accountable
care model. Under the “local accountability”
principle, risk assessment is applied to adjust the
ACO spending benchmarks to reflect patient risks.
Under the “shared savings” principle, ACOs need to
understand the health risk of their patients, monitor
the cost and quality of the care they receive, and
find actionable opportunities for improvement to
achieve savings. Under the “performance
measurement” principle, measures of cost and
quality and comparisons with benchmarks will be
provided to payers, providers, and consumers.
organizations practice proactive care and case
management, using various approaches to
manage patients with a wide array of diseases
and conditions. Accurately identifying higher-risk
individuals and designing appropriate interventions
can improve patient outcomes.
Measuring and improving medical care is an
important focus for ACOs. Significant opportunities
exist to improve the quality and efficiency of health
care. Understanding the value delivered in health
care and identifying and rewarding excellence
are key steps in addressing these opportunities.
These objectives can only be achieved through
valid, actionable, and transparent health care
measurement.
Understanding member health risk will provide
significant advantages for an ACO, as differences
in risks across payers or groups of providers can
impact contracting, budgeting, and the assessment
of financial performance. Individuals generally do
not select health plans and medical care providers
randomly. Some members require more resources
due to their health status and ACO or provider costs
are affected by the particular combination of the
risks patients represent. A key challenge for ACOs
is to assess this risk across the organization and to
adjust accordingly.
Increasingly, commercial payers, state agencies,
and the federal government are evaluating the
performance of physicians and hospitals. The same
evaluations are applicable to ACOs. The results of
these assessments are used in a number of ways.
They include sharing findings with purchasers
and consumers, offering incentives to providers,
and rewarding best practices. There is increased
interest in using performance measures to drive
value-based payments and network design – with
the federal government following a road map to
link Medicare payments to the cost and quality
delivered by providers and becoming a more active
purchaser of higher quality and affordable care.1
Performing well against these standards provides
the opportunity for an ACO to benefit financially and
to distinguish itself as a high-performing and highvalue organization. Measurement can of course
also be used to drive improvements in care and
outcomes.
Assessing health risk can also facilitate coordination
of patient care for an ACO. Many health care
This appendix covers: 1) the analytic tools and
methods used to address health risk assessment 2)
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 69
PART 3 APPENDIX | ACO TOOLKIT
the analytic tools and methods used to measure the
cost and quality of care and 3) data and resources
required to support these advanced analytics.
Examples of the tools used in each area and their
applications are provided.
UNDERSTANDING AND APPLYING MEASURES OF
HEALTH RISK
Health risk can be defined as the expected health
care costs or utilization of an individual or groups of
individuals. Risk assessment is the measurement of
that risk, linking the characteristics of an individual
to their current and future resource use or clinical
outcome. Risk adjustment is the mechanism
that adjusts payment rates or measure results to
reflect the differences in risks as measured by the
risk assessment process. Risk assessment and
adjustment tools have a number of applications for
an ACO. They include setting payment rates more
accurately, adjusting financial performance to reflect
differences in population health status, measuring
provider performance fairly across patient
populations, and identifying high-cost patients for
care management.
Basics of Risk Assessment Models
Models of health risk assessment vary along a
number of dimensions, including the type of data
required, the applications for which the models are
being used, the modeling techniques, the outcomes
to be measured, and the model outputs. Many risk
models used in health care provide an assessment
of cost or utilization outcomes for an individual –
particularly when applied for financial analysis or
rate setting. A number of health care models also
have been developed that focus on specific patient
events, such as the likelihood of an adverse event
or mortality related to a clinical intervention. This
section describes modeling approaches that are
primarily used in assessing the risk of financial
outcomes.
The two basic components of most health risk
models are risk markers and risk weights. Risk
markers describe demographic clinical and other
characteristics that distinguish one individual from
another. Risk weights translate those markers into
a measure of risk. Equation 1 below illustrates
the basic structure of such a model. Riski is the
measure of risk for individual i, Markeri,m describes
the presence of risk marker m for individual i, and
Wtm is the weight assigned by the model to Marker
m. The risk for an individual is the sum of the
weights for all of their risk markers observed – often
expressed as a relative score centered around
1.00, where 1.00 represents the average risk for a
reference population. Using this approach, a risk
score of 0.50 represents a level of risk one half of
that average, a risk score of 2.0 twice that average,
and so on. The risk for a group of individuals can
be expressed as the average risk for all individuals
in that group.
(1)
Riski = S Wtm*Markeri,m
Modeling and Data – Clinically-Based and
Demographic Risk Models
Models of health risk assessment can be simple
or quite complex. A demographic or age-sex
model is an example of a simple model, where the
markers describe the age and gender category
for the individual and the weights represent the
relative expected costs or resource utilization of
individuals in that category. Demographic models
are straightforward to administer, but offer little
clinical information and perform poorly in terms
of predictive accuracy for most applications
(i.e., a demographic risk model does not predict
costs well for an individual). Clinically-based
models employ markers that leverage patient
diagnoses and, in some cases, the use of medical
services. In these models, the markers describe
the presence of a clinical diagnosis or utilization
event for an individual, and the weights represent
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 70
PART 3 APPENDIX | ACO TOOLKIT
the incremental contribution to the risk of having
that marker. Given the richness of the clinical
information employed and the strong link between
patient health status and expected resource
use, the clinically-based models provide greater
predictive accuracy, versus models using only
demographic information.2
Many health risk assessment models leverage
information readily available from administrative
medical and pharmacy claims and enrollment data.
Some models use clinical lab results, in particular
where high-risk prediction is the objective. Models
that employ pharmacy data or results from member
surveys can also provide value when medical claims
data are not available or incomplete.
Most risk models use data for a 12-month period
to identify markers for an individual. The risk
weight assigned to a marker is typically predefined
by the model developer and can vary depending
on the outcome being measured and available
data. Software that encapsulates the health risk
methodology and weightings is often employed to
produce the risk assessment results. Relevant data
are prepared and processed using the software
to produce the risk scores by individuals and to
develop information that is useful in understanding
the measured risk.
Applications for Health Risk
In selecting a health risk assessment model, it is
important to recognize the intended business use
of the model. For example, risk assessment can
be applied either retrospectively or prospectively.
Both types of models have importance for ACOs.
Retrospective or concurrent models use risk
markers for an individual in a base year to measure
risk for that same period of time. A prospective
application uses markers in a base year to measure
risk for a future time period. Retrospective models
are most often used for comparing provider and
health plan performance. Prospective models are
often applied when setting payment rates and to
stratify populations for care intervention and disease
management.
The intended business use is an important
consideration when selecting a model – using the
right tool for the right purpose. As described above,
prospective models are often applied when setting
payments. The risk assessment model used by the
Centers for Medicare & Medicaid Services (CMS)
to reimburse health plans for serving Medicare
beneficiaries is one example.3 Many state Medicaid
programs use similar models.4 As a third example,
an ACO’s target benchmark may be risk-adjusted
prospectively based on future risk expectations. In
addition to being prospective, risk models used in
payment most often include risk markers based on
patient diagnoses and exclude markers describing
utilization events. Where discretion is present, the
risk assessment formula will not reward or penalize
treatment decisions, such as the decision to admit
a patient to the hospital, to perform a surgery, or to
prescribe a medication. In this way, the payment
systems provide appropriate incentives for medical
practice.
A second business use for health risk assessment
is high risk prediction. “Predictive models” are
designed to identify patients of the highest risk in
a population and to provide information useful in
supporting care and health management. These
models leverage all available, useful information
– including diagnoses, history of medical service
use, utilization events, and lab results – to identify
patients who are expected to consume significant
resources in the future and are good candidates
for care management. As a result, these models
provide enhanced predictive ability relative to
models based exclusively on diagnoses.
Example – Assignment of Health Risks Using Three
Different Models
A few examples can help illustrate how risk
assessment models work and how they can be
applied by an ACO. Table 1 summarizes the
calculation of risk for three individuals using three
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 71
PART 3 APPENDIX | ACO TOOLKIT
different approaches to health risk assessment. The
first two models are clinically-based and describe
retrospective and prospective applications. The
third model is a demographic-only, age-sex model.
The risk markers observed for each individual
are shown along with the weights assigned to
each marker for each model. As shown, the
clinical models use markers based on diagnostic
information and also give some weight to age and
gender in calculating prospective risk. The agesex model uses only the individual’s age and sex to
assess risk.
The first individual, a 58-year-old male, is observed
to have diabetes, congestive heart failure (CHF),
ulcers, and a dermatology condition. Each of these
markers receives a numeric weight describing the
contribution of that marker to risk.5 The sum of
the weights across the markers observed is the
risk score for the individual. The total risk scores
of 6.632 and 6.741 for the retrospective and
prospective applications suggest a level of risk for
this 58-year-old male is more than six times that of
the average individual in the reference population.
The risk score based on the age-sex model
provides a different assessment and is markedly
lower than that for the clinical models. In an agesex model, all individuals in the category of male
age 55-64 are assigned the same risk factor of 2.25,
indicating that the expected cost of health care for
individuals in this category is more than twice that
of the average individual in the reference population.
A comparison of these models suggests that using
an age-sex model alone would likely underestimate
their health risk. Finally, note that the retrospective
and prospective risk scores for this example are
similar, both driven by the presence of two chronic,
ongoing conditions (diabetes and CHF) that
comprise the majority of the patient’s risk.
The calculation of risk for the second and third
individuals can be interpreted in a similar manner.
The observed difference between retrospective and
prospective risk for the 35-year-old female illustrates
the impact of an acute event (pregnancy) on the
two clinical models. The pregnancy is a significant
driver of expected costs for the current year, but has
a negligible impact on risk for the future year.
TABLE 1. EXAMPLES OF THE ASSIGNMENT OF HEALTH RISK
Risk Marker
Retrospective
Risk
Prospective
Risk
Age-Sex Risk
2.268
3.028
1.231
0.105
0.000
6.632
2.532
2.842
0.606
0.103
0.658
6.741
n/a
n/a
n/a
n/a
2.250
2.250
0.352
0.066
0.000
0.418
0.360
0.045
0.326
0.731
n/a
n/a
0.495
0.495
0.767
2.253
0.000
3.020
0.591
0.000
0.326
0.917
n/a
n/a
1.224
1.224
Male, Age 58
Insulin dependent diabetes, with co-morbidity
Congestive heart failure, with co-morbidity
Ulcer
Lower cost dermatology
Males, 55 to 64
Total Risk Score
Female, Age 14
Asthma
Lower cost infectious disease
Females, 12 to 18
Total Risk Score
Female, Age 35
Non-cranial nerve inflam, incl carpal tunnel
Normal pregnancy, delivery
Females, 35 to 44
Total Risk Score
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 72
PART 3 APPENDIX | ACO TOOLKIT
Example – Applying Risk Adjustment to Capitation
Payments
Table 2 provides an example of using risk results
to adjust primary care capitation payments to
medical groups. A base capitation rate of $30 is
assumed, an amount covering the cost per member
for delivering monthly services related to primary
care. Estimates of average relative health risk for
the members assigned to each group are listed
along with total membership. For this example, the
the bottom of the table. The higher risk – and higher
expected primary care costs – for Medical Group A
resulted in an increase in payments, while the lower
risks for Groups B and C resulted in a downward
adjustment in payments.
Example – Predictive Model
Table 3 presents a final example, showing scores
of individuals from a high-risk predictive model. An
ACO may decide to further investigate the health
status of these individuals and the care received, in
TABLE 2. RISK ADJUSTMENT TO SUPPORT PRIMARY CARE CAPITATION PAYMENTS
Risk Marker
Relative Health Risk
Members
Base Primary Care Capitation Rate
(Monthly)
Risk Adjusted Capitation Rate
Unadjusted Total Monthly Payments
Adjusted Total Monthly Payments
Medical
Group A
1.150
2,500
Medical
Group B
0.950
3,500
Medical
Group C
0.900
2,000
$30.00
$34.50
$75,000
$86,250
$30.00
$28.50
$105,000
$99,750
$30.00
$27.00
$60,000
$54,000
amounts have been adjusted to ensure the budget
neutrality of the calculation (the same total dollars
for all groups before and after the adjustment).
As shown, the risk-adjusted capitation rate is the
product of the group’s relative health risk and the
base rate. The adjusted and unadjusted total
monthly payments for each group are presented at
All Groups
1.000
8,000
$30.00
$30.00
$240,000
$240,000
particular where gaps in care are also observed and
actionable patient interventions can be applied. The
primary risk drivers observed for each patient are
also included. Most predictive models will provide
information beyond individual risk scores, including
a clinical profile, a summary of the key clinical
drivers of risk, and information on opportunities for
care.
TABLE 3. HIGH-RISK PATIENTS – MEASURES FROM A PREDICTIVE MODEL
Member
Member A
Member B
Member C
Member D
Member E
Member F
Member G
Member H
Member I
Member J
Member K
All Members
Relative Risk
Score
87.50
84.00
81.00
74.50
71.10
68.20
55.40
54.35
52.00
45.00
44.50
1.00
Predicted Annual
Cost
$218,750
$210,000
$202,500
$186,250
$177,750
$170,500
$138,500
$135,875
$130,000
$112,500
$111,250
$2,500
Primary Risk Drivers
Malignant neoplasm, lung
Chronic Kidney Disease, CHF, Diabetes
Hemophilia
CHF, COPD, Diabetes
Chronic Kidney Disease, Diabetes
Malignant neoplasm, bone
Neoplastic blood disease
CHF, Diabetes, Depression
Chronic Kidney Disease
Malignant neoplasm, GI system
CHF, Diabetes
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 73
PART 3 APPENDIX | ACO TOOLKIT
MEASURING PERFORMANCE – ASSESSING QUALITY
AND COST OF CARE
Public awareness of quality issues and rapidly
increasing costs have placed increased focus on
measuring the quality and cost of health care,
using measures that are meaningful, simple,
and actionable (i.e., they can be used to drive
improvement over time). Toward these ends, private
payers, federal and state agencies, consumer
groups, and even health care providers themselves
have proposed, tested, and implemented a variety
of measures to evaluate the quality and cost of
health care. Understanding and performing well
against these standards is critical for an ACO to
achieve savings, meet quality improvement targets,
and differentiate itself. Such measurement will
also enable ACO managers to identify the referral
physicians, hospitals, or other providers that provide
high-quality and cost-efficient care and the groups
that offer opportunity for improvement.
For most organizations, the primary objective in
assessing performance is to identify quality care at
a reasonable cost and delivered with good service.
Cost of care describes the relative resources used
in delivering health care or managing a patient’s
clinical condition. Quality is the assessment
of clinical outcomes or the processes used in
delivering patient care and their correspondence
to evidence-based medicine and other treatment
guidelines. Service can relate to patient satisfaction
and access to care. Efficiency is the cost of care
or resources required to deliver a given level of
quality and service. This section focuses on the
methods used to assess the quality and cost of
care delivered by providers and organizations.
Examples of how the results can be used to identify
opportunities for improvement are provided.
Methods to assess the quality of care are described
first, followed by a discussion of cost measurement.
Measuring Quality of Care
There is a well-documented gap between current
medical knowledge and actual health care practice
– a number of studies have shown a correlation
between compliance with care guidelines and
proven treatments, and the outcome and/or cost of
care.6,7,8,9,10 Across geographical regions, significant
variation exists in the use of medical services
without any evidence of improved outcomes, while
health care costs continue to grow at a high rate
nationwide.11 The increasing push to pay for quality
translates into purchasers, consumers, and patients
requiring increased transparency regarding the
quality delivered by physicians and hospitals.
Measuring and comparing the quality of care within
an organization allows ACOs to:
• Identify both high-performing providers
•
•
•
•
and areas where improved compliance with
prescribed care is recommended;
Identify diagnostic tests or treatments that are
unnecessary or potentially harmful, with the
ability to determine the pervasiveness of these
tests in their populations;
Identify care management opportunities,
including “gaps” in care for patients and
populations;
Identify patients with indications of poor disease
control, such as poor adherence to prescribed
medication regimens; and
Reduce potentially harmful drug-to-drug or
drug-to-disease interactions.
A broad range of quality measures are currently
available in the public domain, covering a range of
measurement areas such as prevention, disease
management, medication adherence, and patient
safety. These measures support assessments
of care for both chronic and acute patients and
for a range of conditions covering the breadth of
clinical medicine. As further investment is made
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 74
PART 3 APPENDIX | ACO TOOLKIT
and as electronic clinical data become more widely
available, quality measurement is likely to grow in
a significant way. Below highlights some of the
key issues and challenges involved with quality
measurement.
Challenges and Progress in Measuring Quality of
Care
Efforts to measure health care quality in the
United States have historically faced considerable
challenges. These challenges include limited
agreement on the standards used to measure
care, the need to identify valid and available data
sources to support measurement, and the lack of
tools that incorporate robust and adaptable sets of
measurement criteria to assess compliance.
technology exist that encode standards of care and
provide an efficient and robust way to assess care
compliance against these standards.
Process Measures and Outcome Measures
Most measures of health care quality can be
categorized as either process measures or
outcome measures. Process measures compare
the care received by a patient with that indicated
by research-based standards, with the idea that
increased compliance with these standards will
lead to better patient outcomes. For example, for
patients with diabetes, process measures include13:
• HbA1c Testing – an HbA1c test performed for
the patient during a 12-month measurement
period (measurement year);
Eye Exam – an eye screening for diabetic retinal
disease by an eye care professional within the
measurement year; and
LDL-C Screening – an LDL-C test performed
during the measurement year.
More recently, progress has been made to address
these challenges. Technology is better, metrics
are improving, and national programs dedicated
to the development of quality measures have
grown.12 Quality standards and metrics are derived
from published, peer-reviewed literature, as well
as guidelines from medical specialty organizations
and national quality organizations. Many physician
specialty organizations are participating in
initiatives to develop quality measures. Endorsing
organizations such as the National Quality Forum
(NQF) are playing a significant role in setting the
standards for measurement.
•
The information available to support measurement
also has improved. Administrative or transaction
data – including medical and pharmacy claims,
and selected clinical data such as laboratory
results – have increased in both availability
and comprehensiveness, thus providing a rich
and convenient information source from which
organizations can evaluate health care. Significant
investment is being made to standardize medical
records and other electronic clinical data, thereby
increasing access to this information – a key source
for valid measurement. Equally important, tools and
• Good HbA1c Control – the most recent result
•
Outcome measures describe the clinical status of a
patient or a clinical result, again reflecting published
guidelines and standards of care. Increased patient
functionality and quality of life can also be used
as a measure of outcome. Examples of outcome
measures for diabetes describing clinical status
include:
•
•
for an HbA1c test for the patient during the
measurement year is < 7.0 percent;
LDL-C Control – the most recent result
for an LDL-C test for the patient during the
measurement year is < 100 mg/dL (threshold 1);
and
BP Control – the most recent blood pressure
reading for the patient during the measurement
year is < 130/80 mm Hg.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 75
PART 3 APPENDIX | ACO TOOLKIT
Implementing Measures of Quality and Applying
Results
Quality measures are typically implemented
using either commercial software that encodes
the measure specifications or internallydeveloped computer programs that capture these
specifications. The data used to support quality
measures include member enrollment data,
administrative medical and pharmacy claims,
encounter data, lab results, and information
from medical records. The data is collected and
integrated by the user and processed using the
packaged software or program code. Outputs
include the compliance results of each measure for
each individual and information summarizing the
details behind the measurement.
ACO managers can use the quality measurement
results to support analysis at the organization,
provider, or patient level. At the ACO level, reports
can highlight the organization’s best opportunities
for quality improvement by identifying the areas
with the lowest guideline compliance. At the patient
level, quality results can be used by providers and
case managers to identify care opportunities and
insights for patient education.
At the provider and provider group level, quality
measurement can involve a number of steps. As
a first step, patients and quality measures are
attributed to those providers most responsible for
patients’ care. Attribution is a key step in valid
quality measurement and can be performed in
different ways – the approach often depends on the
providers being measured. For example, for primary
care physicians, a wide range of measures can
be attributed to the physician who is responsible
for managing the patient’s care. For specialists,
attribution may focus only on patients and measures
where the specialist contributes significantly to the
relevant care (e.g., an endocrinologist observed to
provide the majority of diabetes care to a patient
over some period of time.) For surgeons, the
physician performing the procedure can be deemed
responsible. Finally, the same quality measure
is often attributed to more than one provider,
recognizing the importance of care coordination and
the fact that for many patients, multiple physicians
contribute to their care.
Once attribution has been done, the provider’s
results can be summarized at various levels, such
as across measures for a particular condition or
across all measures and patients. Results are often
compared with internal and external benchmarks,
such as a target level of compliance or the average
results of the provider’s peers.
Examples of Quality Measurement Results
Table 4 provides an example of a report for an
individual physician, summarizing the level of
compliance with the treatment protocols prescribed
for diabetes. A similar report could be created
by the physician group or for all patients covered
by the ACO. Eight measures are included in the
example. The table lists a description for each of
the measures and a clinical synopsis of the measure
guideline.14 It shows the level of compliance for
each of the measures for the patients attributed
to “Dr. Smith.” As a comparison, the levels of
compliance for the same measures for the other
internists in the ACO are shown, along with the ratio
of Dr. Smith’s performance relative to his peers. The
last row of the table provides a composite result for
Dr. Smith across all patients and measures.15 The
value in this row for “% Peer Compliance” reflects
the composite result for Dr. Smith’s peers if they
had the same mix of opportunities across the eight
measures as was observed for Dr. Smith.16
The results indicate that Dr. Smith’s level of
compliance for these measures is somewhat less
than his peers within the ACO, with the greatest
discrepancies observed for HbA1c testing and
screening for nephropathy and retinopathy. The
ratio of Dr. Smith’s overall compliance rate (64
percent) to that of peers (74 percent) was 0.87,
indicating compliance 13 percent below peers. A
list of Dr. Smith’s patients for whom compliant care
was not observed could accompany this report.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 76
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE Patient(s) with evidence of
specific diabetic complications
that had endocrinology
consultation in last 6 reported
months.
All Measures
Patient(s) taking an ACEinhibitor or angiotensin II
receptor antagonist that had a
serum potassium (K+) in last 12
reported months.
Patient(s) that had an office
visit for diabetes care in last 6
reported months.
Adult(s) with most recent LDL
result < 100mg/dL.
Measure
Patient(s) that had at least 2
HbA1c tests in last 12 reported
months.
Patient(s) that had an annual
screening test for diabetic
nephropathy.
Patient(s) that had an annual
screening test for diabetic
retinopathy.
Adult(s) with a LDL cholesterol
test in last 12 reported months.
Composite of all measures for Diabetes
Patients with DM should have appropriate
access to care including, at a minimum,
assessment by a physician every 6 months.
Patients with suboptimal diabetic control
can be identified for additional interventions.
Patients with evidence of specific
diabetic complications would benefit from
endocrinology consultation within 6 months.
Annual screening for diabetic nephropathy
is recommended unless clinical exclusion
criteria have been met.
Annual screening for diabetic retinopathy is
recommended unless exclusion criteria have
been met.
Management of LDL cholesterol goal is
recommended for adults with DM. This
should be checked annually at minimum.
A LDL cholesterol goal of < 100 mg/dl is
recommended for adults with DM. <70mg/
dL is a therapeutic option in some patients.
Patients taking ACE-inhibitor or angiotensin
II receptor antagonists should have, at a
minimum, annual testing of specific serum
parameters.
Measure Synopsis
HbA1c monitoring is recommended for all
patients with DM.
Dr. John Smith, Internal Medicine, Medical Group A
160
200
1,205
30
200
95
774
22
190
75
65
92
200
80
80
90
200
200
# of Opportunities
Compliance
Count
TABLE 4. EXAMPLE OF PHYSICIAN QUALITY REPORT, DIABETES QUALITY MEASURES
64%
73%
95%
79%
81%
80%
46%
40%
45%
%
Compliance
74%
83%
90%
91%
90%
85%
62%
55%
60%
% Peers
Compliance
0.87
0.88
1.06
0.87
0.90
0.94
0.74
0.73
0.75
Compliance
Relative to
Peers
PART 3 APPENDIX | ACO TOOLKIT
January 2010 | 77
PART 3 APPENDIX | ACO TOOLKIT
Measuring Cost of Care
The cost of care is another important area of focus
for ACOs in measuring performance. There are
a number of steps involved in measuring costs.
These steps are described generally in Table 5.
above. Ideally, the measures should be meaningful
(i.e., they are reflective of the health care services
being measured), simple, and actionable (i.e., they
can be used to drive improvement in patient care
over time).
TABLE 5. KEY STEPS IN MEASURING COST OF CARE
Measurement objectives
and strategy
Identify the goals of measurement and how the results will be used.
Data preparation
Collect, standardize, and integrate information to support
measurement, including enrollment, medical and pharmacy service
data, and clinical records from lab results and other sources.
Units of measurement
Select and create the units of measurement, including per episode
of care, population-based, per inpatient admissions, or specific
procedures.
Providers to be measured
Select the physicians and hospitals to be compared and create
“peer groups” for use in comparisons – typically defined using
attributes such as hospital type, physician specialty, and location.
Scope of measurement
Identify the scope of measurement for each peer group; for example,
the medical condition categories for a group of specialists.
Attribution
Assign patients and episodes to individual providers and groups.
Metrics
Identify the metrics for use in comparisons, such as overall costs,
costs by type of service, or the utilization of specific services.
Risk or case mix
adjustment
Adjust for differences in patient morbidity or case mix across
providers
Communication and
improvement
Create physician and hospital results and share these findings
with providers and other stakeholders. Use the results to drive
improvements.
Some of the steps outlined above warrant further
discussion.
Measurement Objectives and Strategy
Identifying the measurement objectives is a key
consideration for all steps, including selecting
a measurement approach, the providers to
be measured, and the metrics to be applied.
Most importantly, an ACO should assess how
the information will be used – whether it is to
differentiate the organization, improve financial
performance, enhance patient care, or all of the
Given the central role of physicians and hospitals
in measurement, methods and measure results
will need to be understandable and transparent
to providers. Working with providers to obtain
consensus on methods and their intended uses is a
key step.
Units of Measurement
The unit of measurement should be aligned with the
measurement objectives. Below we discuss how
several different types of units can provide insights
on the effectiveness of health care delivery.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 78
PART 3 APPENDIX | ACO TOOLKIT
Per Capita Measures. Population, or per capita,
measurement is one approach and presents the
most complete picture for members served by an
ACO or a provider. Examples of these measures
are cost per member per month (PMPM), cost
per patient per month (PPPM), and inpatient
admissions per 1,000 per year. One advantage
of population-based measures is the ability to
capture all of the services for a defined population,
across all providers and conditions treated. This
type of measurement is most meaningful where
the measured entity has clear responsibility for a
significant portion of a patient’s care. Health plans,
provider-hospital organizations, ACOs, and primary
care physicians are examples. This approach can
also have advantages for patients with certain
chronic conditions, such as diabetes, for which the
management of a wide range of co-morbidities is of
central importance in delivering good care.
Episode Measures. Some of the advantages of
population-based measures also create challenges
for their use. In particular, patients often present
with a number of different acute and chronic
conditions – many occurring during the same
period of time. Patients can also have multiple care
providers, each contributing to the same or different
conditions. Assessing the cost of care related
to a condition or the performance of physicians
who focus on a certain area of medicine requires
a different approach – an approach that identifies
conditions for a patient and assigns services to
those conditions. Episodes of care accomplish this
and support the measurement of providers on those
parts of care for which they are most responsible.
An episode of care can be characterized as a
condition classification methodology that combines
related services into a medically relevant and
distinct unit describing a complete episode. An
episode defines a unique clinical condition for a
patient and the services involved in the diagnosis,
management, and treatment of that condition. In
addition to grouping individual services to unique
episodes, these methodologies also characterize
episodes from a clinical perspective, including the
conditions identified. Most methodologies will
further differentiate episodes based on the presence
of significant complications and/or co-morbidities,
some assigning a level of severity to each episode.
This approach enables more accurate case-mix
adjustment and valid comparisons across patients
and providers. Episodes of care describe a wide
range of acute and chronic conditions. Examples
include hypertension, diabetes, CHF, pregnancy,
leukemia, spinal trauma, and minor infectious
diseases.
In addition to condition-based episodes, some
methodologies provide a narrower focus – they
assess the services involved in delivering surgical
procedures. These methodologies have value in
evaluating performance around procedural care,
including assessing the resources used by surgical
specialists. Examples of procedural episodes
include coronary artery bypass graft (CABG), knee
replacement, and cataract surgery.
Most episode-of-care methodologies will determine
completion and outlier status for an episode. In
order to identify a complete episode, methodologies
review the timing of the episode services and
frame the episode by a start date and an end date,
often using a clean period that notes the absence
of patient care related to the episode. Assessing
the completeness of an episode is most important
for acute conditions, which by nature will have a
beginning and an end. Chronic episodes can be
assigned a start date where relevant care is first
observed. However, these episodes will continue
and are often parsed into annual intervals going
forward to define “complete” episodes for analysis.
Table 6 provides a simple example of an acute
episode of care and how episodes are built.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 79
PART 3 APPENDIX | ACO TOOLKIT
TABLE 6. RELATING PATIENT TREATMENT AND AN EPISODE OF CARE
Patient Treatment
Measured Disease Episode
1. Patient presents a specific complaint.
Episode begins with a service denoting a clinician has
evaluated the patient and determined the types of
services required to further identify and treat his or her
condition.
2. One or more physicians provide tests and
treatment.
Episode accumulates additional records, tied to
the first claim by clinical logic identifying specific
diagnosis or procedure codes.
3. Patient recovers and does not seek physician
care related to the same condition again for some
time.
End of episode.
Determining if episodes are cost outliers is typically
conducted by comparing observed costs for an
episode with assigned lower and upper bounds that
frame the normal range of costs. Episodes with
costs beyond this range are flagged as outliers.
Outliers can be the result of inappropriate treatment;
rare, extremely complicated cases; or simple data
or coding errors. The use of complete, non-outlier,
severity-adjusted episodes in health care analytics
supports valid and consistent comparisons.
Episode methodologies are delivered in the form
of grouper software that accepts administrative
enrollment and claims data and assigns individual
medical and pharmacy services to unique episodes
of care. Outputs include details on the mapping of
each service into a unique episode and an episodelevel summary describing the clinical condition,
episode severity, and other key information. Given
the potential complexity of these methodologies, a
key consideration in selecting an approach is the
transparency around the details of the methodology,
such as the measurement specifications, the
clinical rules, and the supporting output. Sufficient
transparency is necessary to support an in-depth
understanding of the episode results.
The results from applying episode-of-care tools can
support ACO managers in achieving a number of
insights about their organizations:
• Track ACO performance and cost trends around
specific diseases and episodes;
• Track the prevalence of conditions – overall
•
•
•
and by severity – and the key services involved
in diagnosing, managing, and treating those
conditions;
Track episodes and trends by population
groups, including payer, geographic area, and
provider organization;
Provide valid measures and comparison of
providers based on cost of care; and
Improve understanding of disease-specific risk
to enhance care and case management.
Inpatient Admission Measures. In addition
to using populations and episodes as units of
measurement, assessing the cost of services
provided in the context of an inpatient admission
offers another opportunity for an ACO. Grouping
the services provided by a hospital into a unique
inpatient stay is relatively straightforward. The
detailed inpatient hospital claim records can
be identified and used to create a single record
summarizing the inpatient admission. These
units can be further categorized by case-mix,
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 80
PART 3 APPENDIX | ACO TOOLKIT
using systems such as Diagnosis-Related Groups
(DRGs). These systems use available diagnostic
and procedural information and assign a DRG to the
inpatient stay, describing the clinical nature of the
admission, the level of severity and the presence of
a significant procedure. DRGs are used widely to
support inpatient hospital payment, including the
Medicare program. DRGs provide a useful tool for
an ACO to case-mix adjust the cost and utilization
data when comparing across hospitals.
Attributing Measures to Providers
Attributing patients and episodes to the appropriate
physicians and groups is a challenging step in cost
measurement. Over some period of time, a patient
can have multiple conditions and, in many cases,
multiple providers caring for the same condition.
For example, for an episode of hypertension, a
patient can be managed by their primary care
physician, an internist, and also receive services
from a cardiologist. For a patient with coronary
artery disease, an internist, a cardiologist, and a
surgeon can all play a key role in providing the
patient’s care. A methodology is required to identify
these episodes for a patient and the providers
responsible for the services performed within
episodes.17
Most attribution approaches can be categorized as
activity-based or population-based. An activitybased approach attributes a patient or episode to
the provider(s) responsible for the greatest amount
of activity during the course of the episode. Activity
can be measured using different concepts, including
the cost of services rendered by a provider, episode
clusters owned, or patient visits. “Sufficient”
evidence of the provider’s responsibility for the
episode is usually required (e.g., attribution only
taking place where a provider is responsible for 30
percent or more of the physician encounters during
the episode.) This approach prevents providers
from “winning” episodes where they have a small
amount of involvement relative to their peers or all
physicians involved in the episode. Activity-based
approaches are often used in performing attribution
for specialist physicians or for primary care
physicians where a gatekeeper, or panel-based,
model is not in place.
Population-based attribution assigns measures to
the provider who is responsible for the member’s
episodes – whether or not the provider rendered any
of the services during those episodes. Populationbased approaches are used where the measured
entity has clear responsibility for managing all, or
significant components, of a patient’s care. Primary
care physicians in a gate-keeper model would be
one example of this approach.
Case-Mix or Risk Adjustment
Measures of the cost of care for an ACO or its
providers can be impacted by the underlying risk
and severity of the patients enrolled or managed.
Case-mix or risk adjustment addresses these
differences and supports more consistent and
equitable comparisons. These approaches allow a
focus on differences in resource use deriving from
differences in the practice of medicine rather than
differences in the mix of episodes or patients.
Each of the units of measurement described above
– populations, episodes, and inpatient admissions
– have methodologies available to support good
case-mix and risk adjustment. For populations,
an ACO would adjust the measures using the
health risk assessment tools described above. For
episodes of care, systems that assign levels of
severity to an episode or classify episodes based on
the presence of complications and co-morbidities
can be used directly to support case-mix
adjustment. For inpatient admissions, DRG casemix classification systems serve the same purpose.
Examples of Cost of Care Measurement Results
Table 7 provides an example comparing the
cost of care performance of two cardiologists
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 81
PART 3 APPENDIX | ACO TOOLKIT
using episodes of care. The analysis uses only
complete, non-outlier condition episodes for CHF,
hyperlipidemia, hypertension, and ischemic heart
disease. The upper section of the table summarizes
results at the condition and severity level. A higher
severity level for a condition indicates the presence
of one or more complications and/or co-morbidities
that impact the resources required for treatment.
The middle section in the table summarizes results
for each of the conditions across all severity levels.
The findings for each physician across all episodes
are presented at the bottom of the table.
Table 7 shows the number of episodes attributed
to the cardiologist, the observed cost per episode,
peers’ cost per episode (the “expected” amount),
and the ratio of the cost per episode for the
cardiologist to his peers. By condition and severity
level, the peers’ cost per episode is the average
experience of all cardiologists included in the
measurement for those episodes. The peers’
experience is case-mix adjusted and assumes the
same mix of episodes (by condition and severity)
as the physician being measured. Notice that for
the overall summary, the peers’ cost per episode
for Dr. Jones is $3,207, while that amount for Dr.
Smith is $3,317. The higher amount for Dr. Smith
indicates a higher case-mix and greater expected
costs relative to Dr. Jones. These peer amounts,
adjusted for the specific mix of episodes observed
for the physician being measured, capture the casemix adjustment appropriate for the analysis.
In the last column, a relative cost ratio less than
1.00 indicates that the observed cost per episode
for the physician is less than his peers. As shown,
both the cardiologists’ costs per episode are below
their peers, on an overall basis, and for most of the
categories shown. An additional report using the
same measure information could summarize results
by type of service, or specific utilization such as
the use of a specific diagnostic test or treatment,
providing greater insights into the factors behind
differences in resource use.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 82
PART 3 APPENDIX | ACO TOOLKIT
TABLE 7. EXAMPLE OF PHYSICIAN COST OF CARE REPORT, EPISODE BASED
Cardiology, Medical Group A
Condition and Severity Level
Dr. Jones
CHF, Level 1
CHF, Level 2
CHF, Level 3
Hyperlipidemia, other, Level 1
Hypertension, Level 1
Hypertension, Level 2
Hypertension, Level 3
Hypertension, Level 4
Ischemic heart disease, Level 1
Ischemic heart disease, Level 2
Ischemic heart disease, Level 3
Dr. Smith
CHF, Level 1
CHF, Level 3
Hyperlipidemia, other, Level 1
Hypertension, Level 1
Hypertension, Level 2
Hypertension, Level 3
Hypertension, Level 4
Ischemic heart disease, Level 1
Ischemic heart disease, Level 2
Ischemic heart disease, Level 3
Ischemic heart disease, Level 4
Dr. Jones
CHF
Hyperlipidemia
Hypertension
Ischemic heart disease
Dr. Smith
CHF
Hyperlipidemia
Hypertension
Ischemic heart disease
Dr. Jones
Overall
Dr. Smith
Overall
Number of
Episodes
10
8
6
22
35
22
16
12
50
24
14
15
6
30
35
84
21
3
120
64
12
3
24
22
85
88
21
30
143
199
219
393
Observed
Cost per
Episode
Peers’ Cost
per Episode
Relative Cost of
Care Ratio
By Condition and Severity Level
$1,016
$1,354
$1,383
$2,128
$1,477
$3,283
$348
$536
$825
$687
$1,262
$949
$885
$1,106
$1,044
$1,492
$3,549
$3,622
$6,547
$6,356
$11,308
$16,154
By Condition and Severity Level
$1,422
$1,354
$2,659
$3,283
$590
$536
$550
$687
$835
$949
$874
$1,106
$1,790
$1,492
$3,803
$3,622
$3,877
$6,356
$12,923
$16,154
$40,419
$26,946
By Condition
$1,254
$2,094
$348
$536
$980
$947
$5,601
$6,361
By Condition
$1,776
$1,905
$590
$536
$791
$919
$4,929
$5,609
Overall
$2,804
$3,207
Overall
$2,924
$3,317
0.75
0.65
0.45
0.65
1.20
1.33
0.80
0.70
0.98
1.03
0.70
1.05
0.81
1.10
0.80
0.88
0.79
1.20
1.05
0.61
0.80
1.50
0.60
0.65
1.03
0.88
0.93
1.10
0.86
0.88
0.87
0.88
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 83
PART 3 APPENDIX | ACO TOOLKIT
DATA AND RESOURCES REQUIRED TO SUPPORT
HEALTH CARE ANALYTICS
A remaining challenge for an ACO is to put the
health care analytics into practice, including
integrating the required data to support the analytics
and identifying the solutions available to support
their applications.
Complete and consistent data is critical in
supporting valid measurement. Many of the
concepts described in this chapter require a
full record of the health care experience for the
individuals and the populations being measured.
For example, to assess compliance with diabetes
process measures, complete medical and
pharmacy claims are needed to identify diabetics
and to capture the performance of an indicated
nephropathy or retinopathy screening test or
the prescription of an ACE-inhibitor. Accurate
identification of episodes of care also will benefit
from complete information. ACOs may have more
direct access to the medical and pharmacy data or
clinical lab results for those services delivered by
the physicians and hospitals that participate in the
ACO. However, many of these methods described
above will require information beyond those services
delivered by the participating providers, such as the
medical and pharmacy services delivered by nonACO providers. ACOs under different organizational
structures are likely to encounter different issues
with data accessibility, data completeness, and data
accuracy. In general, partnering with commercial
and public payers to obtain more complete
information is an important step.
The data used to support measurement also
require accuracy and consistency around key data
elements. In general, diagnosis and procedure
codes are available from administrative claims and
encounter data. The financial amounts required
to support cost measurement are also typically
available from claims data. The assessment of
certain quality outcomes can be supported by
clinical lab results. These data will need to be
reported using appropriate standards, including
the codes used to identify lab tests and the metrics
used to report results. A process needs to be
in place to validate and test the reasonableness
of the data to ensure data integrity. Missing
data, incomplete data, and coding errors would
potentially affect the accuracy of the measurement
results. (Some of the analytic software have builtin processes to identify data completeness and
validity issues.)
Once the data have been integrated and prepared,
the next step involves the application of the health
care analytic methodologies. A number of the
risk assessment and measurement approaches
use a wide range of algorithms and weightings to
score patients, to group episodes and to assign
patient and episode severity. In some cases, an
ACO could develop an understanding of a measure
specification and create internal capacity to
apply the methodology. However, many of these
methodologies are complex. They require ongoing
maintenance and recalibration to reflect changes
in measure specifications and updates for new
diagnosis and procedure codes. As a result, most
users of health care analytics select an external
application or vendor to assist them with the
measurement task.
In general, the tools available to support health care
analytics are delivered in the form of a software
engine. These engines encapsulate the analytic
methodologies and specifications. The software
accepts administrative and clinical data and uses
the information to produce measures of risk, quality
results, or episodes of care. The output typically
includes a data mart, reports describing the results,
and supporting information to assist the ACO in
understanding the findings. Some applications
also will provide static and dynamic reports to allow
additional analysis and to support internal and
external communication of findings.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 84
PART 3 APPENDIX | ACO TOOLKIT
There are a number of analytic and software
vendors available to support ACOs in implementing
health care analytics. Table 8 provides a sample
of these companies and methodologies, organized
around the analytic concepts discussed in this
section.
SUMMARY
Health care analytics is an important area of
focus for ACOs. Measures of health risk across
the organization enable more accurate financial
analysis and contribute to better coordination of
patient care. Assessing the cost and quality of
the care delivered by an ACO presents another
opportunity. Increasingly, private and public payers
are evaluating physician and hospital performance
and using the results in a number of ways, including
public reporting, care improvement, and as a basis
for encouraging and rewarding best practice.
Performing well against measurement standards
provides an opportunity for an ACO to benefit
financially and to distinguish itself as a high-value
organization. Actionable measure results can also
be used to drive opportunities for improvement.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 85
PART 3 APPENDIX | ACO TOOLKIT
Table 8. HEALTH CARE ANALYTICS METHODOLOGIES
There are a number of analytic and software vendors available to support ACOs in implementing health care
analytics. The table below provides a sample of these companies and methodologies, organized around
the concepts discussed in this section. In addition to a description of the methodology and the company, a
website address is provided to obtain further information.
Health Care Analytic Methodologies
I. Health Risk Assessment and Predictive Modeling
Methodology
(Company)
Adjusted Clinical
Groups (CSC)
Description
The Johns Hopkins Adjusted Clinical Groups (ACG) Case-Mix System includes
solutions for a number of applications, including population-based health risk
assessment and predictive modeling. The ACG Rx Predictive model supports risk
measurement using only pharmacy data (http://acg.jhsph.edu/).
Chronic Illness and
Disability Payment
System (University of
CA, San Diego)
The Chronic Illness and Disability Payment System (CDPS) is a population-based risk
model developed to support Medicaid applications. The Medicaid RX (MRX) model
supports risk measurement using only pharmacy data (http://cdps.ucsd.edu).
3MTM Clinical Risk
Grouping Software
(3M)
The 3M Clinical Risk Grouping Software (3M CRGS) supports prediction of health
care utilization and costs on a prospective and retrospective basis (http://www.3Mhis.
com).
DxCG (Verisk)
The DxCG system includes solutions for a number of applications, including
population-based health risk assessment and predictive modeling. The DxCG
Pharmacy models support risk measurement using only pharmacy data (http://www.
veriskhealth.com).
Episode Risk Groups
(Ingenix®)
The Symmetry Episode Risk Group® (ERG) application supports prediction of health
care utilization and costs on a prospective and retrospective basis. The Symmetry
Pharmacy Risk GroupsTM (PRG) model supports risk measurement using only
pharmacy data (http://www.ingenix.com/ProductList/R/T/).
Impact Pro
(Ingenix®)
Impact ProTM is a predictive modeling suite that supports solutions for a number
of applications, including high risk prediction, care management, and underwriting
(http://www.ingenix.com/ProductList/R/T/).
Risk Navigator
Clinical (MEDai)
Risk Navigator Clinical® is a predictive modeling suite that supports solutions for
a number of applications, including high risk prediction, care management, and
underwriting (http://www.medai.com).
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 86
PART 3 APPENDIX | ACO TOOLKIT
Health Care Analytic Methodologies
II. Quality Measurement
Methodology
(Company)
Description
ActiveHealth
Management®
ActiveHealth Management applications encapsulate quality measure
specifications and support results on quality compliance at the patient and
measure level (http://www.activehealthmanagement.com).
EBM Connect (Ingenix®)
Resolution HealthTM
Symmetry EBM Connect® encapsulates quality measure specifications and
support results on quality compliance at the patient and measure level (http://
www.ingenix.com/ProductList/R/T/).
Resolution Health applications encapsulate quality measure specifications
and supports results on quality compliance at the patient and measure level
(http://www.resolutionhealth.com).
Health Care Analytic Methodologies
III. Episodes of Care
Methodology
(Company)
Description
Episode Treatment
Groups (Ingenix®)
Episode Treatment Groups® (ETG) creates episodes of care for a patient by
identifying a unique clinical condition and the services involved in diagnosing,
managing, and treating that condition. The Procedure Episode GroupsTM
(PEG) methodology creates episodes of care in the context of a surgical
procedure (http://www.ingenix.com/ProductList/R/T/).
Medical Episode
Grouper (Thomson
Reuters)
Medstat Medical Episode Grouper® (MEG) creates episodes of care for a
patient by identifying a unique clinical condition and the services involved in
diagnosing, managing, and treating that condition (http://home.thomsonhealth
care.com).
ENDNOTES
1. An overview of the key projects, programs and demonstrations being conducted by the Centers for Medicare &
Medicaid Services (CMS) can be found in “Roadmap for Implementing Value Driven Health care in the Traditional Medicare
Fee-for-Service Program,” available at: http://www.cms.hhs.gov/QualityInitiativesGenInfo/downloads/VBPRoadmap_
OEA_1-16_508.pdf.
2. A good health risk assessment model should maximize predictive accuracy (i.e., how close actual levels of costs or
utilization are to those predicted by the model). However, depending on the application, other factors should be considered,
including clinical relevance, incentives for efficient and quality care, administrative practicality, and ability to restrict
manipulation and gaming.
3. Pope GC, Kautter J, Ellis RP, Ash A, Ayanian JZ, Iezzoni LI, Ingber MJ, Levy JM, Robst J. Risk Adjustment of Medicare
Capitation Payments Using the CMS-HCC Model. Health Care Financing Review, Summer 2004, Volume 25, Number 4,
pp 119-141.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 87
PART 3 APPENDIX | ACO TOOLKIT
4. Kronick R and Llanos K. Rate Setting for Medicaid Managed Long-Term Supports and Services: Best Practices and
Recommendations for States, Center for Health Care Strategies, March 2008.
5. The weights for these models are typically estimated using databases describing the experience of a large population
of individuals.
6. Crossing the Quality Chasm: A New Health System for the 21st Century. Institute of Medicine, Committee on Quality of
Health Care in America. Academy Press, 2001.
7. Institute of Medicine Committee on Quality of Health Care in America. To Err is Human: Building a Safer Health
System. The National Academies Press; 1999.
8. McGlynn EA, Asch SM, Adams J, Keesey J, Hicks J, DeCristofaro A, Kerr EA. The Quality of Health Care Delivered to
Adults in the United States. N Engl J Med. 2003;348:2635-45
9. Osterberg, L and Blaschke, T. Adherence to medication. N Engl J Med 2005;353:487-97.
10. McDonald, HP, Garg, AG, and Haynes, RB. Interventions to Enhance Patient Adherence to Medication Prescriptions.
JAMA 2002; 288:2868-2879.
11. Wennberg, John E, and Cooper, MM, eds. The Dartmouth Atlas of Health Care in the United States. Chicago:
American Hospital Publishing, Inc., 1998.
12. Examples of these programs and organizations include the American Medical Association Physician Consortium for
Performance Improvement (PCPI), CMS Physician Quality Improvement Initiative (PQI) and National Committee for Quality
Assurance (NCQA).
13. HEDIS 2009 Specifications for Physician Measurement, NCQA, 2008.
14. Measures and measure synopses are from Symmetry EBM Connect ®. http://www.symmetry-health.com.
15. The composite result for Dr. Smith is based on an “opportunity-weighted” design, where each measure opportunity
is weighted equally. Alternatively, opportunities for different measures could be weighted differently, reflecting clinical
importance, strength of evidence for a measure, or the objective of an ACO to drive improvement in certain processes and
clinical areas.
16. A measure of statistical significance could be applied to the comparison of the physician’s results versus peers. Such
a test would account for the number of opportunities (sample size) available for the physician and the observed variation
in compliance for a measure or mix of measures.
17. The approach used for attributing patients and episodes to providers must be defensible, understandable and
accepted by all stakeholders. It must be supported by readily available information and be robust across applications –
working well for different sources of data, patient populations and over time. Flexibility in choosing an approach also has
importance, giving consideration to the characteristics of the specialists being compared and to the nature and severity of
their patients and episodes.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 88
Part 4
ACO Infrastructure
John Bertko, Paul Katz, Bob Power
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
PART 4: ACO INFRASTRUCTURE
For an ACO to successfully deliver the level of
integrated and efficient care that would allow
it to achieve shared savings, it must have the
appropriate resources to effectively carry out care
delivery functions. It also must be able to monitor
progress, evaluate performance against targets,
and take appropriate actions to stay on track.
This requires actionable information as well as
management, analytic, and financial services.
To facilitate collaboration among physicians and
other members of the care team and to coordinate
the best care for patients, each physician needs
timely access to relevant patient information,
especially information about treatment from
other providers. Typically, a patient sees multiple
physicians, and his or her care is often not
coordinated in a systematic way, which can result
in redundant services, duplicative tests, adverse
drug interactions, and in the worst-case scenario,
adverse and irreversible patient events. Current
interest in the “Patient-Centered Medical Home”
concept reflects a return to when a single primary
care physician or team was able to coordinate all
aspects of the patient’s care. Similarly, the ACO
focus is on patient-centered care, in an era of
more complex clinical management and greater
concerns about quality and efficiency. In order to
achieve greater integration and coordination within
the delivery system, an ACO needs to serve as the
information hub of its population, integrating data
and keeping track of the care provided to patients
by all its physicians, hospitals, and other providers
as they work to improve the quality and efficiency
of care. In doing so, an ACO can also generate the
kind of meaningful performance information needed
to provide better financial support for high-value,
well-coordinated care.
Integrated delivery systems, such as Kaiser
Permanente and Group Health Cooperative,
have successfully incorporated their information
technology (IT) with their care delivery systems
to improve quality and efficiency; however, other
physicians and hospitals that are less – or not
at all – integrated can also improve quality and
efficiency by creating virtual integration through
the deployment of IT and comprehensive data
management practices. In turn, these data
capabilities can provide a foundation to pay more
for better quality and lower costs, providing further
support for care coordination.
In this section, we first provide an overview of
the essential information and analytical resources
needed to achieve a level of clinical integration that
improves quality and reduces costs, and to create a
“virtuous cycle” that enables further improvements.
We then discuss the management, analytic, and
financial services that an ACO must perform with
the information it collects and analyzes in order to
be successful. As there is an abundant array of
data, tools, and services available today, we will
provide examples of a few key elements, including:
1. Data exchange and data sources for it (4.1);
2. Applications and tools that can provide
physicians and their care teams with meaningful
and useful information on a timely basis (4.2);
3. Reports for tracking financial and clinical
performance (4.3); and,
4. The level and type of resources, tools, and
services needed for a successful ACO (4.4).
4.1: DATA EXCHANGE
Creating a Data Exchange
Taken together, the full range of available data on
services, tests, and prescriptions that have been
provided to patients can provide a rich cache
of facts and useful information about patients
and physicians to achieve care improvements.
These data can show how often and for what
reason a patient seeks care and what care they
have received. It can also show what services
a physician has provided and which patient
populations they serve. It can be helpful for
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 90
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
both ACO providers and ACO management.
For example, physicians should know if any of
their patients with diabetes have gaps in care
using evidence-based guidelines, and the ACO
should know of all the patients with diabetes, the
physicians responsible for these patients, and any
gaps in their care.
Consequently, the ACO needs a capacity to
integrate and store data on its patients and
delivered services. To achieve this, the ACO should
establish regular data feeds from its various data
sources, integrate them, and form data links by
unique combinations of patient and providers. We
call this the “ACO data exchange.”
There are many possible ways to start up a data
exchange, with a number of software and service
vendors that provide data exchange and analytic
capabilities to health care organizations. In
addition, off-the-shelf database programs are
available at a low cost for beginning to build and
support ACO data exchange. Historically, many
organizations have set up static “data warehouses”
to bring together the relevant information for
supporting quality improvement and performance
tracking. More recently, virtual data networks are
emerging that use distributed data systems and
are able to pull relevant patient data when needed.
Many systems are effectively a combination, in
which either real-time or periodic data feeds and/
or summary information from various data sources
are enabled to support an organizational database.
The key issue is the functional capabilities of the
ACO data exchange; in general, exchanges that can
produce more complete and timely patient-specific
information to support patient care and summary
information to track provider and ACO performance
will enable more effective ACO implementation.
By collecting and integrating data on an ACO’s
patient population, the data exchange becomes
the source of the various analytics necessary for
both providing useful information to physicians to
manage their patients’ care, and for systematic
quality and efficiency tracking and improvement.
For example, a patient could generate a hundred
or more records of medical claims, laboratory
tests, and prescriptions annually, resulting in a
large and complicated set of information. The
data exchange would bring this large but important
volume of patient-related data together and turn it
into actionable information for ACO initiatives, either
through using off-the-shelf software or purchasing
data exchange services from a vendor for a monthly
subscription fee. In either case, provided that data
sources are available, the resulting integrated data
exchange should not be expensive for the ACO,
and would enable the ACO to perform a range of
analysis, reporting, and tracking.
Sources and Timing of Data
With regular data feeds from the various sources
that make up health care delivery, the ACO
can develop data exchanges that enable both
performance improvement and tracking. Of course,
obtaining reliable and timely source information
can be a challenge. ACOs can start with claims
information. An ACO can use hospital and
physician billing systems to track basic measures
in clinical quality and costs of care. Procedure
codes, revenue codes, diagnostic codes, and dates
of service for identifiable patients are all valuable
information to include in data gathering. Data can
be provided directly from the billing systems or
from the claims clearinghouses used by the ACO’s
physicians and hospitals. ACOs will also need
to work with their payers to get claims data on
care provided outside of their provider network.1
Physicians should be required to authorize data
sources to release this information to the ACO as a
condition for participation.
These data can and should be used retrospectively
for evaluation purposes to determine if ACO
performance criteria on quality and cost have been
met. Data that show actual payments made by
the payers (commercial payers and/or Medicare)
can be used to compare the ACO’s costs to its
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 91
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
budget. However, to support providers in meeting
and surpassing the benchmarks, evaluation reports
that examine prior performance are not sufficient.
Rather, timely, user-friendly reports must be given
to the providers involved in care, with sufficient
opportunity for them to act to close quality gaps.
The evaluation reports are a secondary outcome
for a data exchange primarily designed to support
the ACO’s activities to improve care. Using the
same data exchange system for both supporting
performance improvement and for ACO evaluation
measures helps assure accuracy and validity of the
performance measures, as well as alignment and
provider support for the ACO’s activities.
Providers should have data on their patients’
utilization in as close to “real-time” as possible.
This ensures that providers have all of the
information needed to avoid unnecessary and
potentially harmful care and helps them determine
the most appropriate clinical course. However,
claims data, particularly within the Medicare
program, historically have not been provided
on a timely enough basis to improve care. For
example, the Government Accountability Office
(GAO) reported a key shortcoming of the Medicare
Physician Group Practice (PGP) demonstration
program was Medicare’s failure to provide paid
claim information to the medical groups in a timely
manner.2 Instead, the physician groups had to use
non-payer source data (like that described above)
to support their data exchanges. More timely and
consistent data from both public and private payers
is a key priority for future ACO implementation.
Prescription data are also essential for verifying
appropriate treatment, and are relatively easy
to integrate electronically. In the event where
prescription claims data is not available (such as
Medicare Part D claims), a workaround may be
developed. For example, the ACO could utilize
e-prescribing and other electronic tools to track
prescription fills to provide the same type of
information.
ACO exchange implementation should also begin
the process of incorporating key clinical information.
Physicians should authorize laboratories to release
test results to the ACO, as laboratory test results
provide another important electronic source of
information for identifying gaps in care.
These data exchanges consisting of claims/billing
data augmented with targeted, feasible clinical
data provide a better foundation for tracking
and reporting on meaningful ACO performance
measures. For example, regular testing and
collection of lab results both facilitates the
appropriate management of chronic conditions,
such as coronary artery disease and diabetes,
as well as allows for the development of more
meaningful health outcome measures that are not
possible with administrative data alone. Rates
of generic utilization and compliance with drug
formularies can also be computed and provide
evidence related to the efficiency of care.3
Electronic Health Records
Electronic Health Records (EHRs) are expected to
become a much more important source of data
for ACO activities, because it is an electronic
version of a patient’s medical history maintained
by the provider over time.4 It may include all of the
patient’s clinical data under a particular provider,
including demographics, progress notes, problems,
medications, vital signs, past medical history,
immunizations, laboratory data, and radiology
reports. The EHR automates access to information
and has the potential to streamline the clinician’s
workflow. The EHR also has the ability to support
other care-related activities directly or indirectly
through various interfaces, including evidencebased decision support, quality management, and
outcomes reporting. However, EHRs are currently
limited in availability and completeness. Most small
practices have not adopted them consistently, and
those that have typically do not have information
beyond their practice or practice site.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 92
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
EHR implementation is likely to be gradual, with
some capabilities and practices coming online
before others. Interoperability (the ability to share
data between EHRs in different medical practices)
with other systems such as a disease registry
is often not available today, as previous CCHIT
certification did not require interoperability.5 As
practically meaningful interoperability becomes
more important in the future, data capture and
sharing will become the rule and not the exception.
Indeed, since siloed payments do not promote
coordination across settings of care, ACO payment
incentives based on patient-level performance
measures that require interoperable exchange of
data will add to the momentum for interoperability.
Consequently, planning for the incorporation of
EHR data into an ACO data exchange will be an
incremental process, but one where progress can
be synergistic with the clinical transformation and
payment reform goals of the ACO. For example,
limited electronic record capabilities could initially
be used to provide key clinical information and
support for registries in the exchange, as described
in the next section.
4.2: TOOLS FOR TIMELY COMMUNICATION WITH
PHYSICIANS AND PATIENTS
Coordinating Care for Quality and Efficiency
From the start, the ACO needs to consider ways
to communicate relevant information from its data
exchange with providers and patients, particularly
when changes are warranted to better meet the
patients’ needs. After all, such improvements in
care are essential for the ACO’s success. The ACO
should adopt, whenever possible, valid evidencebased care guidelines, and use data exchanges to
identify and intervene with physicians when the care
provided falls outside of these guidelines. As an
example, working with providers of care for lowerback pain, an ACO may establish a guideline that
advocates physical therapy be tried before referring
certain patients for a radiology imaging test or to
an orthopedic surgeon. The guideline is designed
to achieve both quality and efficiency goals. For
Medicare patients, guidelines that help physicians
determine the appropriate course of action may be
especially important, as these patients often have
several co-morbidities. Better coordination and
communication among physicians will help clarify
which conditions to treat first.
For these efforts to improve care delivery to
succeed, the ACO needs to dedicate resources
to assist physicians with appropriate patient care
management. To do this, the ACO will need to use
the data exchange to address six key areas:
1. The patients’ treatment needs based on
evidence-based guidelines;
2. The physician’s performance with respect to
meeting the patients’ needs;
3. The benchmarks for the cost of care;
4. The physician cost of care relative to these
benchmarks;
5. The organization’s effectiveness in improving
quality; and
6. The organization’s costs of care compared to its
benchmarks and budget.
The Case for Disease Registries
Disease registries, without EHRs or with limited
or more comprehensive EHRs, offer an effective,
relatively inexpensive tool for identifying and
addressing gaps in quality through an ACO data
exchange. They can serve the dual purpose of
providing timely information for care improvement,
and the generation of performance measures for
ACO payment contracts. Kaiser Permanente, the
Group Health Cooperative, and other integratedcare organizations have achieved measurable
quality improvements using disease registries,
which identify whether the evidence-based needs
of patients are being met, and track the physicians’
and the organization’s performance on quality
measures. Advocate Health Care, the largest notfor-profit delivery system in metropolitan Chicago,
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 93
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
has successfully integrated data by instituting
data registries to manage the care and track the
performance of its primary care and specialist
physicians. Advocate Physician Partners’ Clinical
Integration Program uses a registry that tracks 15
distinct diseases and preventive care populations,
and assesses its performance on 110 clinical
and efficiency measures.6 In the Medicare PGP
demonstration project, all ten participants utilized a
disease registry to track the patients with diabetes.
Note that providers do not need to be associated
with integrated delivery systems to use registries.
Several states and other regional initiatives are
supporting the development of health information
exchanges that can support disease registries.
Employed in a wide range of care settings, disease
registries today provide web-based options that
allow physicians in various locations to access
the same data. Full implementation of an EHR is
not required, but it is necessary to have sufficient,
timely administrative and clinical data exchanges
to reliably identify meaningful opportunities for
improving quality and cost. An EHR can provide
the most up-to-date information on a patient, but it
may be incomplete if it relies only on internal data
or on systems that are not comprehensive. While
potentially more limited in scope, an ACO disease
registry combines limited internal and external data
sources to address clear opportunities to improve
care, by identifying patients with common needs
– including specific diseases and/or qualifying for
preventive care services – and providing timely
information to help address unmet needs.
The data exchange should automatically populate
the disease registries with data from physician
billing, EHRs, medical claims, encounter data,
hospitalization, laboratory test results, and
pharmacy claims or e-prescribing to pinpoint any
unmet needs, gaps in data, or gaps in service
delivery. For example, a typical set of registry
tools for a Medicare population should be able
to identify and track populations with diabetes,
acute and chronic cardiovascular diseases, heart
failure, hypertension, osteoporosis, and end-stage
renal disease. The set should also be able to
track patients who are on persistent or high-risk
medications.
For preventive care, registries should track
colorectal cancer screening, breast cancer
screening, and annual flu shots. While registries
used to be maintained on paper, today’s webbased, pre-populated registries derived from
electronic data exchanges can provide physicians
and other members of the care team with point
of care and near-real time gap analysis to see if
evidence-based services are missing or needed by
their patients. Registries can also provide regularly
updated feedback to physicians indicating the
progress towards measurable goals of improvement
in patient care.
The ACO, working with its payers and providers, will
establish goals for quality improvement based on
the historical data for its population and priorities.
Capabilities to support the quality measures should
be incorporated into the disease registry. For
example, if care for patients with diabetes is an
ACO focus, data on patients with diabetes should
be captured as part of the registry, including valid
measures such as use of diabetes screening tests
and occurrence of preventable readmissions plus
clinical results (such as hemoglobin A1c levels),
if feasible. If flu shots are one of the measures,
patients who should have annual flu shots should
be an additional registry. Populating the registry
with data on occurrence of flu shots – along with
an opportunity to add data on shots not captured
through administrative systems – then supports
both improved clinical care and performance
measurement. Organizations that have used
registries for several years may be tracking patients
in ten or more distinct registries. The registry tools
available today are able to support registries for any
distinct population that needs to be tracked by the
ACO.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 94
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
To illustrate how the disease registry process
works, high-cost and high-risk patients – who need
coordinated care the most – are identified in the
data exchange either automatically from the claims
data (through predictive modeling, for example) or
by physicians adding these patients. If the ACO
utilizes non-physician care coordinators in its care
team, the coordinators then use the disease registry
to track the care against the guidelines established
Services are available to pre-populate disease
registries using established data feeds and to
organize the data so that physicians can readily
utilize the registries to manage patient care. The
registries may include a consolidated listing of
patients with special needs, such as diabetes,
cardiovascular disease, and asthma, as well as
additional preventive care registries for screenings
and immunizations.
EXHIBIT 4.1. CLINICAL DATA INTEGRATION ENABLES ALL THE
PATIENT’S PHYSICIANS TO VIEW THE SAME INFORMATION
Jane Smith, Patient
with Diabetes
OB-GYN
Mammography
Endocrinologist
Lab Test Results
Pharmacy
Primary Care Physician
ACO Data Exchange
and Disease Registries
Primary Care Physician • OB-GYN • Endocrinologist
by the ACO and work with the physicians to
address any gaps in care. Such capability provides
the entire organization with the ability to view all
the practice settings at once. For example, Exhibit
4.1 shows how viewing individual patient-level
integrated clinical data can enable the physicians
to coordinate, monitor, and provide the appropriate
treatments for the patient.
Referrals
Physician referrals to another physician, hospital,
or other service provider can also benefit from the
timely communication and information sharing
among providers afforded by an ACO data
exchange and patient registries. Implementing
evidence-based guidelines with the referred
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 95
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
providers offers additional opportunities to improve
care and avoid unnecessary costs. Decision
support tools such as electronic reference
databases can assist in identifying appropriate
guidelines given patient histories.
The goal is not to restrict access to potentially
beneficial services, but to give the providers and
patients timely access to clinical information that
can make the referral itself more effective (allowing
better care coordination), and to help the ACO
achieve a better understanding of the use of referral
services for its patients. As mentioned in the
earlier example, a physical therapy consultation is
generally recommended as the first line of treatment
for a new diagnosis of lower-back pain before
ordering imaging studies or referring to a surgical
specialist. This is different from a “utilization review”
approach of questioning every referral in a purely
volume-based payment system; in such cases, few
requests are rejected for valid reasons, creating
unnecessary administrative costs and burdens to
physicians and heightening the lack of alignment
and conflict between the providers and the available
“support” systems. The use of more clinically
sophisticated evidence-based guidelines based on
meaningful data exchange can help promote best
practices while keeping care decisions in the hands
of ACO physicians.
4.3: TOOLS FOR TRACKING PERFORMANCE AND
COSTS
Establishing and Tracking Budgets and Costs
The ACO, working with its participating payers,
needs to set an annual medical cost benchmark
based on its analysis of the organization’s patients.
The budget may be developed based on historical
costs of the ACO’s patients, adjusted for changes
in risk factors and for trends in cost of care. Each
month, the ACO can use its data exchange to
sort, count, sum up, and compare data to the
budget for monitoring and payment purposes. The
comparisons would be performed based not just on
overall benchmarks, but on benchmarks by payer
and – depending on the ACO’s planned activities –
particular types of patients and services.
The cost of care has two components: cost per
service, and frequency of the services provided.
Opportunities for improvement can result from
providing more low-cost services, such as
monitoring a patient with heart failure with at-home
care or physician’s office visits so as to avoid a
more expensive hospital admission. Thus, an ACO
should have not only the baseline costs but also
the baseline utilization rates of its various services,
particularly key services that present opportunities
for improving quality and efficiency. Part 3 of this
toolkit provides additional information on ACO
budget and financial performance.
Generating Reports to Assess Performance
To manage the information, the ACO should
establish schedules for generating routine reports,
including how frequently the reports should be
produced, who should review the high level versus
the detailed reports, and who can act on the
information. Ad hoc reports can then be generated
as needed to answer specific questions or to drill
down on particular issues raised.
The sample reports in Appendix 4.A provide
examples of the type of reports that could be
generated from the ACO’s data exchange. They
include reports that track an organization’s
costs and utilization on a high and detailed level,
and reports that track an organization’s quality
performance, with examples of the information
included in the disease registries. As more
sophisticated person-level measures become
available through increasingly detailed registries and
EHR capabilities, the data exchange will be able to
provide more sophisticated support for performance
monitoring and improvement.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 96
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
Analytical Tools and Reporting for Case-Mix and Risk
Adjustments
In addition to data exchanges and disease
registries, there are many other tools that an
ACO can utilize for analytics and reporting, and
can potentially augment the information given to
providers to support improvements in care.
The above tools are available as licensed software
applications or as services provided by vendors
that have built their own reports using these
applications. These tools can be used for casemix or risk adjustment. Performance measurement
of an ACO or its providers is impacted by the
underlying risk and severity of the patients they
manage. Case-mix or risk adjustment addresses
these differences and supports more consistent and
equitable comparisons.
See Appendix 4.A for additional information on tools
and methodologies for risk adjustment, tracking
episodes-of-care, and quality measurement.
4.4: ASSESSING THE APPROPRIATE LEVEL OF
RESOURCES, TOOLS, AND SERVICES NEEDED FOR AN
ACO
Developing a Budget
To achieve improved clinical performance and
the integration to support it as an ACO, each
organization has its own set of needs and budget
constraints. As a first step, an ACO should develop
a budget for the information, tools, and expertise
it needs to implement a data exchange and
support the use of the data for improving care and
tracking ACO performance. The budget would
vary depending on the size of the organization, the
current level of integration, the level of integration
it plans to achieve in order to improve care, the
required resources to achieve its goals, and the
options that offer the most appropriate level of
technology sophistication. For example, an ACO
with a small patient population should set more
modest goals to align resources invested with
what the ACO could ultimately earn in incentive
payments.
Budget development should consider multiple
federal sources of funding that can support these
ACO information capabilities. First, the American
Recovery and Reinvestment Act (ARRA) of 2009
included “meaningful use” incentives in Medicare
and Medicaid for physicians and hospitals to
adopt EHR systems. Under the meaningful use
program, physicians can receive up to $44,000
if they can demonstrate that they have adopted
and meaningfully used EHRs. Initial requirements
are tied to limited data exchange and tracking
capabilities very similar to those described
here, and the U.S. Department of Health and
Human Services has emphasized that more
advanced requirements in later years will be tied
to demonstrated impacts on clinical quality and
outcomes of care that also appear very consistent
with ACO performance measures. Second, as
part of its voluntary Physician Quality Reporting
Initiative (PQRI), CMS has established a mechanism
for submitting quality information via electronic
registries, which if designed properly can be very
consistent with the ACO patient registries described
above. It is likely that further payment reform pilots
to be announced by CMS will also focus on these
kinds of data exchanges for care coordination and
meaningful performance measurement.
Performing Information “Gap” Analysis
The next step is to assess the gap between the
organization’s current capability, where it needs to
be, and the specific tools and resources required to
fill the gaps. The key is to identify the information
that is missing or needs enhancement, and equally
important, to clarify the purpose of that information.
Organizations that have been successful in
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 97
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
achieving quality and efficiency improvement have
found ways to organize health care data to provide
robust, valuable information.
In performing the information gap analysis, a few
key questions an ACO may address include the
following:
1. Is the information sufficiently accurate,
complete, reliable, consistent and robust?
2. How will the information be used? What are
the objectives of using this information? Will it
be used for facilitating timely communication to
improve care, for quality benchmarking, and/or
for financial planning and monitoring?
3. How timely does the information need to be?
4. Is the information meaningful and actionable,
and if not, what additional data are required to
make it so?
5. What kind of systems does the ACO need
to create and maintain the data exchange to
produce this information?
After an ACO has identified the areas that will
need additional tools and resources, it will need to
develop a plan to work with vendors or consultants
to implement the information systems required to
develop its data into workable formats and analyze
it to support ACO activities.
Many of these services can be provided internally
by staff of the ACO. For example, financial
reporting is a standard requirement for most health
care providers today and will not be substantially
different for an ACO. Other services, however,
may be harder to provide with internal resources,
especially those related to insurance-like functions,
such as timely access to claims adjudication,
utilization management, or distributing bonus
payments associated with shared savings.
ACOs will have differing levels of capabilities,
with some needing to provide more new services
than others. Several services will be particularly
important for an ACO to provide or obtain,
including:
1. Claims Adjudication
ACOs will need timely claims data for services from
both ACO and non-ACO providers, which will need
to be adjudicated for payment by an organization
with claims payment skills. Depending on the
contracted payment methods, claims payment may
be straightforward if they are based on common
rules or complicated if they involve new payment
methods such as bundled payments. Regardless of
how actual payments are resolved, the claims data
are necessary for both tracking performance and
identifying specific areas for improvement.
ACO Administrative Services Requirements
ACOs will need to handle a variety of management,
analytic, and financial services. Many provider
groups forming an ACO will need to start accepting
and managing risk for the first time, at least for
bonuses based on shared savings. Managing this
accountability, partly through improved information
on costs and quality, is an important reason for
ACOs to have access to certain key services –
especially since the shared savings are derived
by comparing per capita spending on all services
(across all providers within an ACO) to the per
capita budget targets.
For Medicare, claims could be processed by a
Medicare Administrative Contractor (MAC) or other
data contractor, or the ACO might process its own
claims. Timely access to such claims has been
a technical challenge in the past, but efforts are
underway at CMS and its contractors to provide
data much more quickly.
2. Network Contracting
To the extent that the ACO chooses not to use
either fee-for-service (FFS) Medicare pricing for
Medicare patients or private payer contracts for
non-Medicare patients and instead establishes
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 98
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
its own payment terms for member or contracted
providers (to whom referrals may be focused), it may
need to obtain access to the necessary discounts
on its own. Although this can be done through
preferred provider organization (PPO) network
rentals for a fee, PPO network rental contracts may
not be as favorable as either that of Medicare or
private payers with dominant market shares. As
an alternative, the ACO could negotiate its own
standard contracts with its member providers and
other participating providers; for example, at 100
percent of Medicare for Medicare beneficiaries and
115 percent for commercial members.
3. Bonus Settlement Adjudication (After the Claims
are Paid)
Another type of ACO financial transaction is
the payout of any accrued bonuses. There are
two bonus settlement adjudications. The first is
the aggregate bonus pool, which involves three
determinations: the actual performance against a
target, interim determinations of financial and quality
performance (in advance of having complete, fullyear data), and deciding when to allocate funds.
The second settlement adjudication involves how
the ACO will allocate the bonus funds among
providers. Approaches to allocating the bonus
funds across providers participating in the ACO can
range from being relatively simple to quite complex.
Before the start of the year, an ACO must specify
the rules under which the bonus amounts are to be
distributed and implement a financial mechanism for
retaining and then distributing such bonuses.
ACOs will need to ensure that the right capabilities
are available either in-house or through a vendor
to provide the necessary reporting and analytical
tools needed to determine the allocations. Also, a
process for handling bonus disputes may need to
be developed.
4. Physicians Organization Management Services
Depending on the structure of the ACO, it may
need to manage emerging physician practices or
other integrated delivery systems. While hospitals
typically have well-developed management
structures (e.g., finance, IT, contracting and quality
improvement), many small physician practices may
not have these capabilities. They may want to
develop these capabilities in-house or obtain these
services from external organizations on a fee basis.
Presently, many physicians outsource their billing
and collection functions. Physicians should be
encouraged to change their referral patterns to be
based on efficiency, appropriateness of care, and
quality metrics, and to refer patients to participating
ACO providers.
5. Care Coordination and Utilization Management
Services
Care coordination is a very important function for
an ACO. The greater the effectiveness of care
coordination, the more likely an ACO is to meet or
exceed its budget savings targets. “On the ground”
care coordination will likely be accomplished in the
physician office, but assistance may be needed
with other work related to registries, post-discharge
coordination, disease management reminder calls,
and other similar activities. Hospital partners
may be able to supply assistance. New health IT
software may provide further help. However, some
practices may want to combine their efforts with
support from vendors. An ACO may want to invest
in acquiring the care coordination and utilization
management services, which can either be “owned”
by the ACO or subcontracted to other vendors. For
example, specialty benefit management companies
that provide criteria and utilization management
services for high cost services, such as advanced
imaging, may be one of the services to subcontract
to an outside vendor.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 99
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
Possible Sources of Service
The services include:
While some fully integrated delivery systems will
already have the capabilities to perform the above
services, the majority of organizations will need to
develop or purchase these capabilities. A critical
decision for many organizations will be whether to
develop the capability in-house, to outsource, or
use a combination of the two.
• Access to the payer data to identify and
Many health care organizations believe that they
must invent their own tools and solutions internally.
However, outsourcing has its benefits. The most
notable benefit is that a company whose services
and products are its core competency can offer the
products and services more cost-effectively and
efficiently than by doing them in-house. Learning
from others not only allows the organization to take
advantage of the best practices and the proven
ideas, but it could also help avoid costly mistakes.
To determine how it can use data systems to
achieve improvements in quality and cost, the ACO
must initially focus on understanding its needs,
planning for these needs in terms of resources, and
then carefully evaluating available options. It may
provide a competitive edge for the organization to
focus initially on adopting the tools and resources
that would enable the organization to show positive
outcomes early in the process. Multiple resources
and tools are available to assist in the development
of the ACO, along with the experience and lessons
learned of organizations that have managed the
transition successfully.
Below are several possible sources of services that
can help enable an ACO to function successfully:
Private Payer Services
Private payers will be able to provide a variety of
services. If the ACO is in a synergistic contract with
a major private payer, it is likely to want to make use
of many of the services that the private payer offers.
•
•
•
•
•
•
attribute beneficiaries;
Possible use of payer contract rates with ACO
participating providers;
Use of payer contract rates with providers
“outside” the ACO;
Claims adjudication;
Data analysis services performed by the payers’
actuarial or analytical staff;
Use of certain utilization management services
such as the nurse help-line or contracts with
Radiology Benefit Managers (RBM); and
Calculation of bonus amounts according to
agreed-upon formulas.
For some activities, particularly utilization
management and care coordination services, an
ACO and its management team may prefer to
handle in-house. However, it may be more efficient
to obtain other services externally through payers or
vendors, where the infrastructure is already in place
and can be operated at marginal cost rates, such as
a nurse help line.
Centers for Medicare & Medicaid Services
CMS obtains many administrative services at
relatively low per-unit cost rates, and those services
are obviously relevant to ACOs providing care for
Medicare beneficiaries. In particular, CMS contracts
with MACs to adjudicate claims using Medicare
payment rules. If an ACO is going to pay all or
some of the claims for Medicare beneficiaries at
FFS Medicare rates, it would get the most favorable
payment rates by going through these existing
arrangements. This assumes that the MedicareACO rules would permit access to these services.
One option would be for the MACs to adjudicate
the ACO claims as they do for the FFS members,
and in addition (or through another Medicare data
contractor) provide regular, detailed, and timely
electronic reports on utilization of services by
beneficiaries in the ACO.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 100
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
At the same time, the ACO should be vigilant
in the areas of fraud and abuse and in terms of
opportunities to address unnecessary utilization.
While CMS and its MACs have taken many steps to
control fraud and abuse, there are many instances
where problems have been found.
CMS may also be able to provide some basic
data services to the ACOs; however, CMS staff is
already burdened with many activities. Other than
determining shared savings and processing and
reviewing claims payments, the extent of assistance
that CMS staff or its contractors will be able to
provide is not yet clear.
Services from Large Multi-Specialty Provider Groups,
Management Service Organizations (MSOs), or Other
Similiar Organizations
Certain integrated delivery systems, such as
Intermountain Healthcare, have long and successful
histories as health systems. Large, multi-specialty
EXAMPLES OF MSO SERVICE OFFERINGS
Based on the websites of several MSO firms, the following types of services are available in today’s capitated
medical group environment. These services are the same as or very similar to those needed by ACOs. They may
include:
•
Efficient method for claims and encounter submission, processed electronically or manually.
•
Authorization workflow that provides timely application, processing, storage and emailing back to medical
group providers.
•
Web-based electronic medical management systems to connect with electronic medical records or through
fax-based data entry.
•
Compliance reports that allow providers and medical directors to measure their compliance rates and compare
them to those of their peers.
•
Provider tools and training to continuously educate providers and their staff.
•
Referral management, such as using the website and submitting requests online. Systems allow providers to
track the status of requests at any time and minimize unnecessary phone time for provider staff.
•
Referral tracking reports available to send to each primary care physician (PCPs) on a monthly basis. This
report allows the PCPs to ensure all members will obtain the services that they need. A similar report
of members with critical conditions is available for the appropriate specialists to ensure that they make
arrangements to see the members quickly.
•
Customized financial reports tailored to a medical group’s needs, including profitability by contract or by
physician, monthly reporting packages, and monthly analytical reports.
•
Risk pool reports that can be designed to monitor and analyze pool performance, and pro forma forecasting
tools to assist the medical group as needed (e.g., to monitor ACO budgets).
•
Member education programs and special classes for members.
•
Patient handbooks and newsletters.
•
Contract Management, including:
-
Health plan contracting and negotiations
-
IPA/Physician contracting and negotiations
-
Hospital contracting and negotiations
•
Credentialing functions for every contracted IPA provider, including hospital-based physicians.
•
Developing and maintaining comprehensive Quality Improvement Programs that continually evaluate, monitor,
and identify areas to improve the quality of clinical care provided to members and the provider panel.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 101
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
groups, such as Geisinger in Pennsylvania and Hill
Physicians in California, have also successfully met
their own service needs over the last 20 years. It
is possible that negotiating for services with these
provider organizations – or related organizations
such as MSOs and actuarial consulting firms –
may be beneficial, as these organizations already
perform services that are closely tailored to the
needs of ACOs. For example, some of these
groups have been paying claims and calculating
bonuses that are shared among provider partners.
Many have experience in the quality measurement
metrics, such as the California provider groups
that received incentive payments under the Pay
for Performance program led by the Integrated
Healthcare Association.
Another service that may be contracted is
management consulting support to help with the
set-up and ongoing management of complex
provider organizations. ACOs may be able to
draw on the lessons learned by already successful
groups, rather than re-inventing the wheel.
Other Third Party Vendors
There are independent vendors, unrelated to private
payers that are able to provide some of these
services. Various kinds of utilization management
services are readily available, such as services that
provide hospitalists for hospital management, nurse
hotlines, RBMs, and other similar therapy-specific
managers.
firms available to provide the necessary services.
Some of the data analysis firms on the West Coast
have provided analytical services to large multispecialty groups and PHOs for over a decade.
While ACOs are different from the capitated provider
groups, there are many similarities between the
two groups, so vendors could transition to the ACO
services quickly.
Possible Timeline for Setting Up ACO Services
Organizations interested in forming an ACO
need to build in sufficient time to put in place the
necessary support infrastructure and processes.
Each organization will require different lead time,
depending on its own circumstances. Exhibit
4.2 provides a sample timeline for setting up the
ACO services. One way for an ACO to set up
its operations quickly is to initially rely on leased
services provided by existing MSOs or other similar
organizations with a proven track record.
This timeline assumes:
• An MSO would be engaged to provide some of
•
•
•
the services;
Other ACO start-up processes (e.g.,
establishing the panel of ACO physicians,
arranging contracts with payers, etc.) would
proceed on parallel tracks;
All tasks are accomplished without significant
delays; and
The managers are dedicated to the decisionmaking process with sufficient staff support.
In addition to clinical management services and
claims adjudication, there are many actuarial/
financial consulting firms and a few data analysis
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 102
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
EXHIBIT 4.2. SAMPLE TIMELINE FOR SETTING UP ACO SERVICES
This timeline assumes a start date of 1/1. The schedule of setting up the necessary activities, from the time a
decision is made to form an ACO to the ACO operation start date, is shown below:
1/1
Decision to form an ACO is final and an implementation work group is assigned.
2/1
Inventory of the ACO service capability (e.g., utilization review, claims adjudication, etc.) is
completed.
3/1
Discussion among ACO managers, physicians, and staff regarding the services to be
provided internally and the services to be leased.
4/15
Final decision by the ACO on the type of services that are needed.
5/1
RFP to be prepared by the ACO managers/consultant.
6/15
RFP released by the ACO managers to MSO vendor candidates.
8/15
All MSO vendor proposals received.
8/15-31
MSO services reviewed and clarification requested.
9/15-9/30
Final MSO terms negotiated by ACO managers.
10/1
Final decision of MSO vendor(s) decided by ACO managers.
10/1-12/31
Implementation preparation.
1/1
ACO operation start date.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 103
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
CONCLUSION
Successfully managing an ACO requires not only
capturing significant amounts of new data, but also
effectively transforming that data into actionable
performance measures that can offer providers
timely feedback on the costs and quality of care.
Building a robust data exchange is part of the
solution to bringing together the various sources
necessary to collect and process this data. Equally,
ACOs must invest in developing a number of new
processes to ensure both timely reporting and
decision-making based on this incoming data.
Developing these support processes, even with the
use of outsourced providers or other external parties
will take time. A provider need not have the “perfect
EHR” to begin undergoing this transformation. Nor
should providers misconstrue these needs as a
prescription for structural integration. In fact, much
of the monitoring and reporting described here can
be done by introducing new data elements from
providers’ existing billing systems, intermediaries,
and plans.
It should be expected that over time, ACOs would
work towards enhancing their infrastructure. As the
information exchange and other service capabilities
advance, ACOs will be able to support more
advanced levels of performance measurement and
payment models (see Part III for a discussion of the
potential evolution of performance measurement
and payment models under the ACO framework).
In fact, all of these activities – setting performance
goals, developing supporting payment models,
and building the infrastructure for higher quality
care – should be expected to occur in coordination
with the organization’s larger development plan for
coordinating care. In this sense, identifying gaps
in capabilities will need to be a continuous process
as health care priorities evolve and infrastructure
improvements allow for greater opportunities for
improving system effectiveness and efficiency. In
Part 5, we provide specific examples of how these
enhanced support services can translate into
improved clinical performance.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 104
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
APPENDIX 4.A
The provider specialty can be categorized into:
allergy & immunology, anesthesiology, cardiology,
dermatology, emergency medicine, endocrinology,
gastroenterology, hematology/oncology, hospital,
infectious disease, internal medicine, laboratory/
pathology, mental health, neonatal, nephrology/
dialysis, neurology, obstetrics/gynecology,
ophthalmology/optometry, PCP-family practice,
PCP-internal medicine, PCP-pediatrics, PCP-urgent
care, and other.
Reports to Track Costs and Utilization
Figure 4.1 provides a high level overview of the
ACO claim costs by provider specialty. The report
compares the per member per month (PMPM)
incurred claims in the current 12-month period with
those incurred in the prior period, as well as with
the target claim costs. For example, under the
PCP Internal Medicine specialty, the current period
claims PMPM was $4.81, compared to $4.46 PMPM
in the prior period and the target of $4.73 PMPM.
Figure 4.1: Organization Cost Compared to Budget
Summary Cost by Provider Specialty
ACO
October 1, 2008 through September 30, 2009
IBNR used for current “Claims” only: 1.02
“PRIOR” indicates: Jan 08 to Dec 08 - Prior Member Months: 308,992
“CURRENT” indicates Oct 08 to Sep 09 - Current Member Months: 265,164
Prior Total PMPM
Incurred Claims
Current Total
PMPM Incurred
Claims
Target
Comparable
Hospital
$6.36
$6.60
$3.48
Laboratory/
Pathology
$2.38
$3.37
$6.55
PCP: Family/
General Practice
$9.08
$9.88
$5.93
PCP: Internal
Medicine
$4.46
$4.81
$4.73
Provider
Specialty
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 105
Period 2
$75.24
19,768
$45.67
PMPM with percent change
Annual Utilization with percent change
Per Unit Cost with percent change
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE $29,617,865
648,496
Grand Total
Period 2
Period 1
$41.24
23,114
$79.44
644,084
-10%
17%
6%
$26,562,294
Procedures & Claim $
334,380
393,668
Procedures & Claim $
IBNR: 1.0000
Member Months:
IBNR: 1.0000
Jan 01, 07 - Dec 31, 07
Member Months:
Groupings
Specialty & Procedure Code
Jan 01, 06 through Sep 30, 09
Jan 01, 06 - Dec 31, 06
by Intelligent Healthcare
Period 1
Cost & Utilization Study
ACO
Drill Down Cost & Utilization
FIGURE 4.2: COST AND UTILIZATION TRENDS – TOTAL
Period 3
$42.32
20,691
$72.96
512,716
3%
-10%
-8%
$21,696,165
Procedures & Claim $
Period 3
297,360
Member Months:
IBNR: 1.0000
Jan 01, 08 - Dec 31, 08
Period 4
$35.66
26,255
$78.03
580,164
-16%
27%
7%
$20,691,324
Procedures & Claim $
Period 4
265,164
Member Months:
IBNR: 1.0160
Oct 01, 08 - Sep 30, 09
Figure 4.2 shows the cost and utilization trends for four reporting periods – prior three calendar years and the latest 12 months (LTM) yearto-date. It shows the PMPM claim cost, annual utilization per 1000 lives, the average cost per procedure, and the corresponding trends
over the prior periods. For example, in the current period, the number of procedures performed was 580,164; the total overall claim cost
was $20,691,324; the PMPM claim cost was $78.03; annual utilization per 1000 lives was 26,255; and the average cost per procedure
was $35.66. The current period over the prior period trend for the pmpm claims costs, annual utilization, and cost per procedure were 7
percent, 27 percent, and -16 percent, respectively.
Reports to Track Costs and Utilization
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
January 2010 | 106
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE Per Unit Cost with percent change
Annual Utilization with percent change
PMPM with percent change
PROVIDER SPECIALTY: Cardiology
Specialty & Procedure Code Groupings
Jan 01, 06 through Sep 30, 09
$83.42
308
$2.14
10,112
843,521
$78.95
342
$2.25
9536
-5%
11%
5%
752,831
Claim $
Period 2
Period 1
Procedures &
334,380
393,668
Claim $
Member Months:
Member Months:
Procedures &
31, 07
IBNR: 1.0000
Jan 01, 07 - Dec
Period 2
31, 06
IBNR: 1.0000
Jan 01, 06 - Dec
by Intelligent Healthcare
Period 1
Cost & Utilization Study
ACO
Drill Down Cost & Utilization – Specialty Level
FIGURE 4.3: COST AND UTILIZATION TRENDS – SPECIALTY LEVEL
343
$2.22
8,496
-2%
0%
-1%
660,665
Claim $
Procedures &
Period 3
297,360
Member Months:
IBNR: 1.0000
31, 08
Jan 01, 08 - Dec
Period 3
$77.76
373
$2.46
8,236
2%
9%
11%
653,518
Claim $
Procedures &
Period 4
265,164
Member Months:
IBNR: 1.0160
30, 09
Oct 01, 08 - Sep
Period 4
$79.35
Figure 4.3 shows an example of the next level drill down view of the cost and utilization trend, by provider specialty. The format is the same
as the report in Figure 4.2. In this example, for the cardiology group in the current period, the number of procedures performed was 8,236;
the total overall claim cost was $653,518; the PMPM claim cost was $2.46; annual utilization per 1000 lives was 373; and the average cost
per procedure was $79.35. The current period over the prior period trends for the PMPM claims costs, annual utilization, and cost per
procedure were 11 percent, 9 percent, and 2 percent, respectively. Further drill down by procedure code groupings within the specialty is
available (not shown here). Usually reviewing drill downs helps pinpoint areas that generate anomalies or require further investigations.
Reports to Track Costs and Utilization
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
January 2010 | 107
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
Reports to Track Costs and Utilization
costs. Patient-detail information is available for
further drill down. In total, there were 273 high
cost patients incurring 175,249 procedures, with
$4,095,043 incurred claims and an average cost of
$15,000 per patient. In conjunction with the total
claims reported in Figure 4.2, it showed 20 percent
($4,095,043/$20,691,324) of the overall costs was
attributed to a relatively small number of high-cost
cases.
Figure 4.4 shows the frequency and cost of all
the high-cost cases by diagnosis codes in the
current period. This report enables the ACO to
identify the individual high-cost patients who
require extra care coordination. This sample
report includes the costs associated with the
professional services, and excludes the hospital
FIGURE 4.4: HIGH-COST CASES BY DIAGNOSIS
High Cost Cases By Diagnosis
ACO
October 1, 2008 through September 30, 2009
Member Months:
265,164
IBNR: 1.016
Diagnosis Code
Group
Annual
Outlier
Procedures
Claims
Claims
Claim Cost
Per 1,000
Cases
PMPM
Per Case
0.6
13
1,987
145,491
$0.56
11,192
0.5
11
71,631
349,896
$1.34
31,809
0.2
5
835
55,362
$0.21
11,072
0.1
2
382
41,862
$0.16
20,932
12.4
273
175,249
4,095,043
$15.69
15,000
Frequency
Injuries, Sprains,
Strains, and
Poisoning
Nephritis,
Nephrotic
Syndrome,
Nephrosis
Diseases of
the Respiratory
System
Congenital
Anomalies
Grand Total
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 108
1,000 mbr mo]
9,198
7,750
6,230
8,707
MD, Elizabeth
MD, Alan
MD, James
MD, Michael
PCP: family/gen practice
Assigned
Members
[excludes PCPs with less than
PCP Specialty and Name
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE $17.34
$20.83
$14.11
$17.05
PCP self
^ indicates greater than +1 standard deviation over the mean
^^ indicates greater than the mean +2 standard deviations
$4.42
$8.31
$4.52
$4.50
Ranking Key
Cardio
Member Months: 265,164
Hosp
Lab
$4.69
$8.13
$5.11
$5.72
$7.24
$11.14
$8.29
$8.76
Per member Per month
October 1, 2008 through September 30, 2009
Medical Group
Referred Service Costs (***excludes hosp, lab, OB/GYN & PCP)
FIGURE 4.5: REFERRAL COSTS AND REFERRAL RATES
$5.45
$3.60
$6.93
$5.05
All Other
$47.49
$62.50
$59.04
$52.72
PMPM***
Total
^
^^
^^
^
Rank
Figure 4.5 shows the referral costs to specialty and ancillary providers by PCPs. The outliers – the PCPs with high utilization rates among
physicians within the same primary care specialty – can then be further evaluated. The specialty and ancillary providers can be categorized
into: cardiology, dermatology, emergency medicine, hospital, internal medicine, laboratory/ pathology, mental health, obstetrics/
gynecology, ophthalmology, physical therapy, radiology, surgery, and all other. This report can be further enhanced by incorporating casemix and risk adjustment methodology to provide a more equitable comparison.
Reports to Track Costs and Utilization
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
January 2010 | 109
Denominator
4th Quarter Registry Data Collected through November 30, 2009
Family Medicine
Clinical Integration Requirements
Medical and Technological Infrastructure
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE Outpatient- Children Pediatric Second Hand Counseling (>= 50%)
Outpatient- Registry Smoking Patients Counseled (= 100%)
3
3
% of LDLs < 100 mg/dL (>= 56%)
Body Mass Index (>= 50%)
44
44
44
% HbA 1c performed (>= 81%)
% LDL performed (>= 79%)
Nephropathy Testing (>= 60%)
Diabetes Care
3
% of Blood Pressure Measurement (<= 90%)
Coronary Artery Disease
Outpatient- Children Pediatric Second Hand Assessment (>= 50%)
Smoking Cessation Counseling
Clinical Effectiveness
CIRRIS Physician/Office Access (>= 4)
CIRRIS Initiative Reporting Compliance (>= 75%)
2009 Clinical Integration Progress Report – INDIVIDUAL REPORT
Use of Registries
Clinical Integration “Quality” Report Card
FIGURE 4.6: QUALITY SCORECARD
37
36
33
3
3
3
Numerator
84%
82%
75%
100%
100%
100%
No Eligible Patients
No Eligible Patients
No Eligible Patients
170
Results
0.50 of 0.50
0.50 of 0.50
0.00 of 0.50
0.10 of 0.10
0.10 of 0.10
0.20 of 0.20
0.00 of 0.00
0.00 of 0.00
0.00 of 0.00
1.00 of 1.00
1.00 of 1.00
Points
The patients’ registries are available in different formats. One report may show the year-to-date scores on each of the quality measures,
with drill down that shows the patients who do not meet the clinical effectiveness requirements. One report may show the cardiovascular
patients with high LDL cholesterol value. Another report may show the key statistics of an individual patient with diabetes, such as the
clinical observations, blood pressure, dates of laboratory visits/eye exams, and current medications.
Figure 4.6 is a quality scorecard by physician. It shows a sample of the physician’s year-to-date scores indicating his/her clinical and
efficiency performance.
Reports to Track Quality and Efficiency
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
January 2010 | 110
PART 4: ACO INFRASTRUCTURE | ACO TOOLKIT
ENDNOTES
1. Claims intermediaries could potentially bypass the need to get data indirectly through a plan. Currently, the standard
way to assure complete data, at least on utilization, is to get data from the payer. However, since intermediaries (i.e.,
billing systems) are able to directly report this data to providers on much shorter timeframe through the exchange model
bypasses, thus bypassing payers and reducing the delivery time for data.
2. GAO, Report GAO-08-65 at www.gao.gov/new.items/d0865.pdf
3. For more detail on the types of measures data exchanges should support: McClellan, et al. “A National Strategy to put
Accountable Care into Practice.” Health Affairs. May 2010.
4. CMS, Electronic Health Records overview, see www.cms.hhs.gov/ehealthrecords
5. Certification Commission for Health Information Technology, see http://www.cchit.org/
6. For a copy of Advocate’s Clinical Integration Program annual report, see http://www.advocatehealth.com/documents/
app/value%20report.pdf. For more information: Mark C. Shields, Pankaj H. Patel, Martin Manning and Lee Sacks.
A Model for Integrating Independent Physicians Into Accountable Care Organizations. Health Affairs, no. (2010): doi:
10.1377/hlthaff.2010.0824
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 111
Part 5
Health Care Delivery Transformations for
Achieving High-Value Health Care
Mark McClellan, Andrew Kruegar, Sue Podbielski
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS
FOR ACHIEVING HIGH-VALUE HEALTH CARE
By definition, ACOs are held accountable to achieve
high levels of patient outcomes and satisfaction,
at lower health care costs. As part of this mission,
delivery transformations should seek to achieve
three aims, consistent with all recommendations in
this toolkit:
• Provider infrastructure to coordinate and
improve care, including:
• Organized provider activities (e.g., a strong
primary care provider foundation, and
leadership to effectively involve key types
of providers in performance improvement
activities);
• Timely data reporting and analysis; and
• Health IT to facilitate the sharing of
information across providers in an efficient
manner.
The ability to coordinate care allows ACOs
to influence the health and health care of
the patient population for whom they are
accountable. An ACO is accountable for the
overall quality and cost of care for its attributed
patient population. Responsibility extends
not just to the services provided or conditions
treated by providers affiliated with the ACO, but
to all care received by patients assigned to the
ACO. As a result, ACOs benefit from the ability
to manage and coordinate care of their patients
across the full spectrum of services delivered to
their patients.
• Performance Measurement to evaluate the
impact on patients’ care experience and quality
of outcomes on their total health. Key goals
of performance measurement are to ensure
accountability for the quality of care and to
identify and drive improvement in areas of
substandard care. For example, performance
measurement can identify the potential savings
and quality improvement for an ACO’s patient
population with low levels of both generic
prescription drug utilization and medication
adherence, as indicated by measures of generic
prescribing and refill compliance or frequency.
• Shared Savings as a funding source for
many delivery reform efforts that are often not
directly reimbursable under fee-for-service
(FFS) and other traditional payment systems.
For example, shared savings could be used
to help sustain improvements in health IT
that may allow the better coordination and
communication of care across providers. ACOs
should consider investigating both the shortterm and long-term opportunities to qualify
for shared savings. In the short term, ACOs
should aim for interventions that can quickly
generate savings and return on investment
– such as interventions designed to reduce
hospital readmissions or relatively simple
interventions that correct clearly identified
inefficiencies in care delivery. ACOs should also
consider ways to build on these investments for
longer term savings – such as further steps to
improve chronic disease management through
the availability of more sophisticated, timely
information and delivery reforms.
Reforming the delivery of care to improve health
results for patients while lowering overall costs
is both difficult to accomplish and essential for
the success of an ACO. In 2009, the Bipartisan
Policy Center (BPC) released a report identifying
interventions and reforms with the greatest potential
to make a meaningful impact on the quality and
value of health care. Specifically, they cited a
number of promising approaches that may lead to
long-term savings, including: targeting interventions
to specific patient populations or clinical areas;
introducing real accountability from providers and
patients; transitioning provider reimbursement away
from volume and towards value; and integrating
multiple delivery system reforms together.1
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 113
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
In this section, we describe a range of strategies,
including many noted by the Bipartisan Policy
Center, that have been implemented in both the
public and private sector to achieve these core
ACO objectives. The evidence on most of the
strategies discussed in this section is mixed, and
the best opportunities for a particular ACO are
likely to depend on the circumstances, ideas, and
leadership of its providers. To achieve success,
clinical reform initiatives should be part of an overall
strategy that can feasibly lead to a significant
impact on costs and meaningful performance
measures in accordance to the aims described
above. Moreover, it is important for ACOs to
implement these transformative strategies flexibly
and with willingness to experiment with a variety of
approaches.
The types of delivery reforms that ACOs should
consider fall into four dimensions:
• Care coordination (5.1);
• Population or condition specific treatments
•
•
(5.2);
Patient engagement in care (5.3); and,
Infrastructure and organizational redesign (5.4).
5.1: CARE COORDINATION
The current health care system is fragmented, which
can lead to preventable and/or high-cost medical
care. To start addressing this fragmentation,
ACOs should first consider interventions that
target high-cost delivery processes where
coordination problems lead to identifiable and
measurable opportunities to reduce complications
and costs. These types of interventions may
include: addressing avoidable hospitalizations
and readmissions through better care transitions
and case management for patients with complex
illness; more effective care coordination within
care settings; and other structural and process
interventions that can encourage greater
collaboration and teamwork between health care
providers. One study found that approximately
14 percent of elderly patients transitioning from
the hospital to the home experience medication
discrepancies, which more than doubles their
chances of readmission.2
Better care coordination and care management
can also help to ensure that patients receive care
in the most appropriate, least intensive setting
as possible, and that care is not duplicated or
conflicting. Given that the average Medicare
beneficiary visits two primary care providers and five
specialists annually, this presents a key opportunity
for savings.3
Care coordination can refer to a number
of strategies that aim to emphasize overall
responsibility for the entire care process and
provide additional resources to providers to
promote preventive care, improve care transitions,
and encourage information exchange as patients
transition from one care setting to another.
Established systems and processes that encourage
joint decision making can help support timely,
multi-directional communication within and across
provider practices. These activities are particularly
valuable for Medicare patients with multiple
chronic conditions who are at high risk for costly
complications. Typically, these activities are not
reimbursed reasonably, if at all, in FFS payment
systems.
Below we highlight several promising examples of
strategies that can improve care coordination.
Care Transitions
Without planning and careful monitoring, postdischarge patients – in particular older patients with
multiple chronic conditions – are at greater risk for
readmissions and avoidable complications that can
drive up health care costs. To address this concern,
providers and researchers have developed a
number of interventions aimed at improving followup care and communication across providers.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 114
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
For example, providers in a large integrated delivery
system located in Colorado have developed a care
transition program that targets elderly patients to
improve care transitions from the hospital to the
home. Specifically, the program includes two key
components: (1) the development of a personal
health record owned and maintained by the patient
to facilitate cross-site information sharing, and
(2) a series of visits and telephone calls with a
transitions coach trained in medication review and
reconciliation. These two components are designed
to encourage patients and their caregivers to take a
more active role in care transitions and foster more
effective care coordination. In randomized control
trials, the intervention resulted in up to a 28 percent
reduction in hospital readmissions for up to six
months after the original hospitalization.4
Another method proven to help successfully
coordinate and streamline transitions is the
Transitional Care Model (TCM). TCM is designed
to address transitions across physical settings and
in health status for older Americans with complex
needs by having an advanced practice nurse lead
patients through transitions using an evidencebased care plan to coordinate their care. This
includes a comprehensive assessment of patient
and caregiver needs and focuses on increasing
patients’ and caregivers’ ability to self-manage
their care. The nurses in the TCM each manage an
active caseload of approximately 15 to 20 patients,
helping ensure each patient gets the attention they
need during one of the most vulnerable parts of the
care process.
The TCM has been shown to consistently reduce
the time to first readmission, total number of
readmissions, and inpatient days, resulting in
significant decreases in health care costs. A recent
study found that over 52 weeks post-discharge
participating patients experienced 36 percent fewer
readmissions with an estimated mean per-patient
savings of approximately $5,000.5 Patients, family
caregivers, and providers have all seen increased
levels of satisfaction – patients have reported
improvements in the quality of life and health;
family caregivers have seen the demands on them
decrease; and nurses have found their work to be
more meaningful as they get to spend additional
one-on-one time with patients.
Hospitalists
Effective care coordination depends on adequate
knowledge about the conditions being treated,
the patient’s individual needs, the roles of the
different providers, and the resources available.
Within the hospital setting, physicians designated
as hospitalists not only have developed special
technical expertise involving conditions that require
effective inpatient care, but can also help manage
patients with complex health care needs by taking
care of, and efficiently managing milestones of their
“whole person care.” This includes managing both
the condition that the patient was admitted for as
well as co-morbidities, which can help to ensure
that the patient is sufficiently stabilized before he/
she transitions to the next care setting.
A successful hospitalist will need to efficiently
coordinate care across providers, while making
efficient determinations about the types of health
problems that require effective inpatient treatment.
For instance, a cancer patient may be treated in the
hospital by the oncologist, but see a primary care
physician for diabetes management. The hospitalist
would be responsible for ensuring that treatment
addresses both the primary condition – in this case
cancer – and all co-morbidities across levels of
care, which include diabetes. Multiple studies have
shown that hospitalist care can be associated with
shorter hospital stays and lower costs.6
Additionally, non-physician staff members may be
responsible for assisting in a patient’s care. These
staff members, sometimes called case managers
or care coordinators, are responsible for helping
to identify patients with high-risk conditions,
assisting with disease management education
and/or follow-up, helping patients navigate the
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 115
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
health care system, and collecting data on process
and outcome measures. Some, usually large,
physician offices have their own staff member, but
most practices share an external coordinator who
follows specific high-risk patients. The intensity of
a coordinator’s involvement with a patient varies on
the severity of each case, but many patients can be
empowered to manage their condition. Guided Care
A team of researchers at John Hopkins developed
the interdisciplinary Guided Care model in 2001,
which has primary care practices hire a highly
skilled nurse to track, assess, and manage patients
with multiple chronic illnesses.7 One of the nurse’s
responsibilities is to develop, in coordination with
the patient and their support team, an evidencebased care plan that includes all parties’ care
responsibilities and is distributed to the patient, the
patient’s family, and the primary care practitioner.
Additionally, the nurse will work with the patient and
their support team to help identify needed social
and community supports.
The Guided Care model has been proven to improve
care quality and reduce total costs, rates of hospital
admissions, days admitted, and emergency room
visits, resulting in a net costs savings that more then
offsets the costs of adding a nurse coordinator to
the practice. 8,9
Patient-Centered Medical Homes
The Patient-Centered Medical Home (PCMH) is a
hybrid care delivery and payment reform model
that works to integrate and coordinate care by
providing a monthly management fee to support
investments in health IT, management tools,
and care coordination activities. The medical
home is centered on a redesign of primary care
delivery supported by enhanced access to clinical
information systems that help to integrate care
across providers, quality improvement and learning,
and initiatives to encourage patient engagement in
care. Medical homes can operate independently of
ACOs; however, they may work best in conjunction
with the ACO payment model. Medical homes
receive upfront payments which can help support
initial care intervention investments. If paired with
an ACO, they would be able to receive shared
savings if their efforts at quality improvement and
cost reduction are successful (see text box for ACO
initiatives building on medical homes).
Research assessing the impact of medical homes
has shown that the model increases access to
care, promotes prevention, and engages patients
in self-management and shared decision-making.
Specifically, medical homes can be an important
intervention in reducing costly emergency
department visits. Research also shows that
patients prefer the more active involvement of their
primary care physician in disease management.
All of this can help lead to increased patient
satisfaction, improved outcomes and quality of life,
and lower costs.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 116
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
ACOS BUILDING ON MEDICAL HOME PILOTS
Vermont Blueprint for Health. Vermont is piloting enhanced medical homes in three communities as part
of its statewide Blueprint for Health Initiative. The Pilots are working to improve care quality with support from a multi-disciplinary Community Health Team (CHT), which partners with primary care offices,
hospitals, and health and social service organizations to better coordinate care across settings. In addition, sites will be aided by expanded health IT capabilities including a clinical tracking system, a statewide
health information exchange network, and an integrated prevention action plan based on assessments of
community risk factors. The goal of these efforts is to enhance patient self-management and integrate
community-wide public health prevention efforts. The multi-payer initiative, which is currently supported
by three commercial payers, Medicaid, along with some funding from the state, will begin implementing
ACO pilots in 2011.i
Community Care North Carolina (CCNC). CCNC is another example of the medical home model in
practice. CCNC is a nonprofit organization made up of 14 networks and over 4,500 primary care physicians that connects community providers – including hospitals, health departments, and department
of social services – with primary care physicians. Physicians are paid a monthly payment in return for
managing patient care for more than one million Medicaid enrollees, serving as a single access point for
patients and providing 24 hour access to care, as well as participating in quality improvement, disease
management, and prevention activities. The state Medicaid program pays physicians a per member per
month (PMPM) fee in addition to FFS payments at 95 percent of Medicare payments.
CCNC aims to improve care quality, utilization, and cost effectiveness of chronic care. Originally a Medicaid program that provided some care to uninsured populations, the program was recently expanded to
include Medicare and Medicaid dual eligibles and will expand within three years to the general Medicare
population through the Medicare Health Care Quality Demonstration. As discussed in previous sections,
under the demonstration, providers will have an opportunity to share with CMS any savings they achieve
through their physician-directed care coordination program. The demonstration will test the impact of a
physician-directed care coordination program – supported by a regional physician pay-for-performance
program, health IT, and a common set of quality measures across multiple payers – on care quality and
efficiency.
Early evaluations of the model suggest that it can result in significant cost savings and improvements in
care quality. For example, conservative analyses estimate annual savings of $60 million in fiscal year 2003
and more than $160 million in fiscal year 2006. Reductions in emergency department utilization (23 percent less than projected), outpatient care (25 percent less than projected), and pharmacy (11 percent less
than projected) were the largest drivers of these savings. The model has also shown improvements in the
quality of care, including improvements in asthma control, which was one of the first CCNC focus areas.ii
i. Hester J, Lewis J, McKethan A, “The Vermont Accountable Care Organization Pilot: A Community Health System to Control Total Medical Costs and Improve Population Health”, The Commonwealth Fund, May 2010: http://www.commonwealthfund.org/Content/Publications/Fund-Reports/2010/May/The-Vermont-Accountable-Care-Organization-Pilot-A-Community-Health-System-to-Control.aspx
ii. “CCNC/Access cost savings – State Fiscal Year 2007 Analysis,” Mercer, February 2009: www.communitycarenc.com/
pdfdocs/mercer sfy07.pdf
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 117
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
5.2: CONDITION OR POPULATION SPECIFIC
INTERVENTIONS
Interventions that are targeted to specific patient
populations and clinical areas typically have a
greater impact on quality improvement and cost
containment than broader approaches. In particular,
ACOs should consider interventions that address
the growing chronic disease burden. Chronic
diseases are responsible for 75 percent of overall
health care spending, with nine of 15 diagnoses
at hospital admission directly related to chronic
disease among older Americans.10 Given that every
portion of the population has a growing prevalence
of chronic disease, gaining control over chronic
disease is one of the most essential elements of any
health care delivery reform.
ACOs may consider using predictive modeling
such as high utilization, complexity of conditions,
or other clinical and socioeconomic characteristics
to better target their interventions and improve
return on investments. Continual development
of analytical capabilities and better evidence on
which interventions are most effective for specific
populations will be important as ACOs continue to
develop and implement their reforms.
Below we outline several promising examples
of targeted interventions that can reduce
hospitalizations for specific populations and prevent
complications associated with chronic disease.
Chronic Disease Management
The underlying goal of disease management is
to reduce the burden of disease and improve
health outcomes by preventing complications
and emphasizing better prevention and care
management. These efforts are particularly
beneficial for frail patients and those with multiple
chronic conditions and can result in improved
patient and family satisfaction with care and
reduced costs. Necessary components of
a successful disease management program
include the ability to identify and monitor highrisk individuals, apply evidence-based practice
guidelines, coordinate care between providers,
and encourage patient self-management through
education and patient tools. The range of disease
management services can include timely initiation
of ancillary health services, patient monitoring
and empowerment, and coordinating community
services. When these services are delivered in a timely
manner, they can reduce preventable complications,
emergency department visits, length and frequency
of hospitalizations, and unnecessary gaps in care.
In particular, diabetes, asthma, and congestive heart
failure (CHF) are all areas that have been shown to
be amenable to disease management activities.
For example, the Camden Citywide Diabetes
Collaborative was developed by the Camden
Coalition of Healthcare Providers to address
growing overutilization of emergency room care due
to conditions related to diabetes. Efforts include
steps to increase the capacity of communitybased primary care practices and medical day
programs, support diabetes self-management, and
improve care coordination. The collaborative has
helped convert ten community-based primary care
practices into certified Patient-Centered Medical
Homes, develop electronic health records and a
diabetes registry, and support provider education to
standardize diabetes care.11
The use of an asthma nurse specialist has also
been found to reduce total hospitalizations and
readmissions for asthma. Specifically, asthma
nurse specialists assisted primary care physicians
in simplifying asthma care programs, completing
daily “asthma care” flow sheets while the patient
was in the hospital, educating patients in asthma
self-management and developing personal asthma
care plans, and providing outpatient follow-up.
These activities resulted in a 60 percent reduction
in total hospitalizations and a 54 percent reduction
in readmissions for asthma, which amounted to
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 118
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
$6,462 in savings per patient in direct and indirect
health care costs.12 Other successful asthmafocused disease management activities have
included web-based coaching and home-based
health action plans.13
unique patient population. It will also be critical for
ACOs to develop multiple initiatives to better ensure
the ability to find those that are successful.
There are also many CHF disease management
programs. Most focus on patient education by
nurses, advanced practitioners, or pharmacists with
follow-up education over a period ranging from six
months to three years. Experienced cardiovascular
nurses at the Jewish Hospital at Washington
University Medical Center provided high-risk, elderly
CHF patients with intensive education about CHF
and supported efforts to encourage treatment
compliance, including individualized dietary
assessment and instruction by a registered dietitian,
discharge planning with social service personnel,
medication analysis and reconciliation by a geriatric
cardiologist, and intensive follow-up after discharge
with the hospital’s home care services. They found
this intervention lowered hospital readmission rates
by approximately 22 percent for all causes within 90
days of discharge and 55 percent for readmissions
related to CHF. The intervention also resulted in
a 36 percent reduction in length of hospital stay,
lowering cost of care by nine percent and producing
a return on investment of 1.37 percent. Like
disease management programs for diabetes and
asthma patients, appropriate program design and
targeting can greatly influence the success of the
intervention.14
Interventions that address non-adherence to
medications and better medication management
represent an additional opportunity for generating
savings while improving patient care. It is estimated
that patient non-adherence to medication costs the
health care system up to $290 billion a year.15
It should be noted that for all these interventions,
the evidence base has been mixed, in some
cases producing significant cost savings, while
in others resulting in no statistically significant
changes. Importantly, studies have found that
disease management programs that target higherrisk patients tend to result in a greater likelihood of
reduced costs and utilization compared to programs
that provide more modest interventions targeted to
a patient base with mixed disease severity. Thus,
ACOs should carefully consider which interventions
have the greatest chance of success given their
Medication Management
Medication management can take many forms.
It can involve a care team that includes the
prescribing physician, pharmacist, or a staff
member who keeps in contact with the patient. In
one example, hypertension patients saw an increase
in controlled blood pressure rates with web-based
pharmacist care. The program began with a
telephone visit between the pharmacist and the
patient. Then an action plan was introduced and
shared with the patient and prescribing physician.
Secure web-based communication continued every
two weeks until blood pressure was controlled.16 Newly evolving technologies can help to remind
patients to take their medication, monitor patient
adherence, and relay data back to the provider.
These technologies in support of medication
management programs can also inform physicians
of the full costs of a treatment. For instance, CCNC
(see text box above for more details) created a
prescription advantage list, which ranked drugs
based on cost to encourage the use of lower-cost
medications when appropriate. CCNC reports
lower drug spending and an estimated savings of
$1 million per year due to the use of the prescription
advantage list.17
Targeting Individuals with Multiple Chronic
Conditions and Functional Limitations
As individuals with both chronic conditions and
functional limitations represent 14 percent of
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 119
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
the population but 46 percent of health care
expenditures, targeted interventions for this group
represent a high-yield strategy.18 Individuals
with chronic conditions and functional limitations
(difficulties walking, performing activities of daily
living, etc.) on average spend three times what
others spend on health care. A great deal of this
spending is dedicated to inpatient services and
prescription drugs, demonstrating an opportunity for
greater care coordination.
CMS is currently operating a demonstration for
post-acute care (PAC) payment reform. The
demonstration developed the interoperable,
electronically-based Continuity Assessment Record
and Evaluation (CARE) tool to measure health and
functional status of Medicare PAC and hospital
discharges. CMS will use the data from the tool
and demonstration to examine differences in costs
and outcomes across PAC provider types. The
results of the demonstration will also be used to
reform payment for skilled nursing facilities, home
health, inpatient rehabilitation facilities, and longterm care hospitals. As data collection on chronic
conditions and functional limitations becomes
more standardized, robust, and electronic, and as
payment reform for post-acute settings improve
provider ability to coordinate care, ACOs may find
cost-savings by focusing on frailer individuals
leaving acute or post-acute settings.19
A pilot from Boeing (the Intensive Outpatient Care
Program) demonstrated that targeting patients
with severe chronic conditions for a medical home
intervention can yield cost-savings, improved
physical and mental functioning, and reduce
number of work days missed. Because the
pilot simply targeted the highest-cost group for
intensified coordination, it did not require longterm delivery system re-organization, large-scale
health IT implementation, or years of change to see
cost-savings. Similarly, Geisinger Health System
implemented case management programs for
their highest risk patients, drawing on nurse-case
managers. Geisinger attributes some of its success
in slowing spending growth to this program.
Another example of targeting high-cost individuals
with multiple chronic conditions and functional
limitations is the Care Level Management model
(CLM). Care Level Management, a vendor of
physician services, contracts with plans on a
combination of per member per month and FFS.
CLM patients have difficulty getting to medical
offices, and CLM specializes in physician home
visits and constant physician availability. With
its focus on frail patients in the top percentiles of
spending, CLM has found that its services can
reduce admissions by 60 percent.20
5.3: PATIENT ENGAGEMENT IN CARE
For ACOs to have a positive impact on health care
spending, they must encourage active patient
participation in their own health in addition to
rewarding effective provider efforts to improve
quality and reduce costs. ACOs in tandem with
their respective payers can play an important
role in encouraging patients to make choices that
are more consistent with high quality and efficient
care. Below we discuss promising strategies that
encourage the use of cost-effective primary care
and preventative services that together can delay
or prevent the onset of costly chronic conditions,
even under the limitations imposed by the mixed
incentive structures of the traditional FFS Medicare.
Patient Education and Shared Decision-Making
A fundamental tenet of delivering high-value
health care is ensuring that patients have access
to information that can help them make informed
decisions about their health care. Patients who
are knowledgeable about and engaged in their
treatment are more likely to continue treatment
and adhere to provider advice, which ultimately
improves overall outcomes, enhances patient
satisfaction, and reduces avoidable complications.
Patients who receive either a web-based or paperbased decision aid containing information about
prostate cancer screening concepts, benefits,
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 120
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
and risks have been found to be more informed
and more engaged in the screening decision than
patients who did not receive a pre-visit education.21
The Dartmouth-Hitchcock Medical Center’s
Center for Shared Decision Making is a leader in
engaging patients to understand treatment options,
outcomes, risks, and benefits. The Center uses
an array of condition-specific educational videos,
decision-making tools, and counseling to support
informed decisions about treatment. Such services
and tools guide patients towards high-quality, costeffective, and patient-centered care.
Wellness and behavior change programs can also
serve to lower health care spending and improve
patient well being due to reductions in emergency
room visits and inpatient stays. Such programs
encourage patients to improve their health by
providing patients with education on self-care and
health maintenance strategies, and professional
support to carry out these recommendations.
For example, for patients with chronic asthma,
a patient self-management plan developed in
collaboration with a physician can help identify
appropriate responses to specific symptoms based
on established care guidelines. Studies comparing
self-management to usual care have found that
patients monitoring their own asthma not only had
better outcomes, but saved considerably over those
in standard care with a primary doctor.22
instance, 30 percent of seriously ill patients reported
that they would rather die than live permanently
in a nursing home,24 and 28 percent of patients
with advanced heart failure would trade one day
of excellent health for another two years in their
current state.25 Increased attention on end-of-life
planning can improve patient, family, and caregiver
satisfaction, improve patient quality of life, and
reduce utilization while staying consistent with
patient wishes. Another excellent example of patient education is
the Everett Clinic, which employs hospice nurses
in its primary care clinics to provide intensive case
management and end of life planning, including
promoting palliative care. The program has resulted
in significant reductions in readmission rates and
has increased the use of hospice and home care
services among elderly patients who are chronically
ill. Evaluations of the program found that it resulted
in a 35 percent reduction in readmissions 60 days
prior to death.26
Other well-documented examples of effective
end-of-life care management include treating
cancer pain and cancer associated depression; for
instance, lung cancer patients who experienced
continuity of care across the outpatient to hospital
settings were less likely to spend time in the
intensive care unit prior to death.27
Value-Based Insurance Design
Patients informed of the likely outcomes and
overall costs of treatment options make better
and often lower-cost decisions about the care that
they need. This includes care planning for serious
illnesses near the end of life, the cost of which can
vary dramatically by region. A recent Dartmouth
study found that patients in high-spending regions
received about 60 percent more end-of-life care
due to differences in the frequency of physician
and specialist visits, tests, and hospitalizations.
Yet higher spending did not lead to higher quality
of care, greater access to care, improved health
outcomes, or higher patient satisfaction.23 For
Financial rewards and other incentives for patients
may significantly leverage ACO delivery reforms.
ACOs can work with their health plans to encourage
active patient engagement in their care by designing
benefits to reward cost-saving, high-value
behavior. One approach is for plans to restructure
their cost-sharing requirements to support the
utilization of cost-effective primary care and
discourage the use of costly and ineffective care
by lowering or eliminating co-payments for primary
care visits or instituting tiered drug benefits. In
the latter circumstance, plans can implement drug
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 121
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
benefits that make effective drugs available at low
or no cost for specific populations, as well as use
formularies to encourage patients to switch to
more cost-effective drugs (e.g., generics). These
efforts can be particularly beneficial to patients with
chronic conditions. Many chronically ill patients
incur significant out-of-pocket costs, which can
discourage the use of important, high-value health
services and can serve as barriers to seeking
preventative care.
The City of Asheville implemented benefit reforms
in conjunction with disease management programs.
Specifically, the city waived co-payments for
diabetes-related drugs and devices for patients
who agreed to participate in a diabetes education
program administrated by local pharmacists.
An evaluation of the program revealed that the
expanded pharmaceutical care services reduced
total mean direct medical costs by almost $700
per patient per year.28 Given the promising early
results, the program has expanded to other chronic
conditions, including asthma.
Similarly, benefit design may enable patients to
share in the savings when they use ACO providers
that have demonstrated better results in both quality
and cost of care. Patients who use a specialist
preferred by an ACO because the specialist has
demonstrated lower complications and lower overall
costs could potentially see significant reductions in
their out-of-pocket costs.
Geisinger Health System’s ProvenCare is an
example of a successful benefit design program
that relies on strict evidence-based practice
standards, pay-for-performance principles,
accountability through capitated payments, patient
engagement, and other checks and balances to
improve quality while lowering costs. By diligently
adhering to evidence-based practice guidelines and
building accountability into their delivery system,
ProvenCare has lessened the average total length
of stay, improved the 30-day readmission rate by 44
percent and drastically reduced complications and
infections.29
Other studies have also demonstrated that reducing
out-of-pocket costs for evidence-based treatments
results in an increase in patient adherence to
medication plans and/or reduction in overall health
care costs.30
5.4: INFRASTRUCTURE AND ORGANIZATIONAL
REDESIGN
To support the reforms discussed above, ACOs
will need to invest in infrastructure and adopt new
organizational tools and strategies. This includes
investments in health IT, which can help providers
monitor progress, evaluate performance against
targets, and make real-time program adjustments
based on these findings. Such investments need
not involve comprehensive electronic records; many
successful care transformation initiatives have made
limited, focused investments in patient registries
oriented toward closing particular gaps in quality.
As the ACO’s clinical improvement objectives
become more sophisticated, the data capabilities
can be enhanced as well. A robust health IT
infrastructure with data management systems can
also help lead to more advanced care integration
and coordination strategies, such as disease and
utilization management and Patient-Centered
Medical Homes. That said, investments in health
IT should be viewed as catalysts for the delivery
transformations described throughout this toolkit,
as health IT investments alone are rarely a means of
long-term savings.
In addition to infrastructure investment, ACOs
will also need to implement strong process
management and cultural reforms to help support
the more efficient and effective delivery of care.
These operational and organizational improvements
are particularly important for clinical transformations
in hospitals and other providers where strong
leadership will be necessary to address the
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 122
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
inefficiencies and variation in care delivery that can
lead to low-quality care and costly complications.
Strong leadership will especially be needed in
provider settings under traditional FFS payments
where incentives to reduce costs and potentially
utilization are not always clear. It is important that
organizational managers see process improvements
as a means not only of achieving the potential
savings described throughout this section, but
also for reducing overhead costs which can lead to
additional savings.
Many of these process and organizational
improvements can be implemented fairly quickly,
so ACOs may want to consider implementing these
reforms first as they also ramp up other delivery
reform activities. Some examples of organizational
redesign to improve care and increase value for an
ACO can include providing detailed management
reports for clinicians to target areas of improvement,
strengthening non-physician workforce for costeffective care coordination, and improving work
environments to reduce workforce turnover.
One specific example of a relatively simple
organizational redesign that quickly yielded higher
quality care is the Institute of Family Health’s
electronic health record (EHR)-based clinical
decision support system. Although EHRs were
already in place, practices found that they had very
low pneumonia vaccinations for patients over 65.
The Institute decided to implement an automatic
vaccination reminder for all patients over the age of
65. Vaccination rates for the targeted population
rose from less than ten percent to about 90 percent
with the decision support redesign. In addition
to automatic reminders, quick safety procedure
checklists have been demonstrated as a low-cost
organizational solution to reducing hospital-acquired
infections.
Below we discuss other infrastructure and
organizational redesign examples in more detail.
Health IT
Providers rely on timely clinical information to inform
clinical improvement and drive practice redesign.
Health IT can refer to any number of electronic
systems that facilitate the collection, organization,
and sharing of medical information electronically,
including EHRs, electronic medical record (EMRs),
health information exchanges (HIE), and personal
health records (PHRs). Implementing these systems
is only the first step; to realize cost savings and
improve quality, ACOs must actively use this
information and these systems to support disease
management and care coordination activities and
better target their efforts at high-risk patients who
can benefit the most from these efforts.
The Indiana Health Information Exchange (IHIE) is
one of many successful examples of using health
IT to improve the quality, safety, and efficiency of
the care. IHIE focuses on several core services,
including: a clinical messaging service; a clinical
repository service; and a chronic disease, preventive
care, and quality reporting service. By connecting
nearly 70 hospitals, long-term care facilities, and
other Indiana health care providers involving more
than 19,000 physicians, IHIE is able to deliver lab
results, reports, medication histories, and treatment
histories in real-time for over six million patients.
Previously, this information was stored in various
physician offices or hospitals but now is able
to follow patients, regardless of where they are
receiving their care.
The providers in Indiana are also using health IT
to drive quality improvement and adherence to
evidenced-based medicine. Through the Quality
Health First (QHF) program, which was developed
and is administered by IHIE, providers receive
quality reports comparing their performance
to quality benchmarks and to their peers. The
reports – based on information from health
insurance claims, point of care data from the
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 123
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
physician’s office, and clinical information – are
then used to calculate performance payments.
Both public and private payers, including Medicaid
and Medicare through the Medicare Health Care
Quality Demonstration Program, are contributing
data. Although the program is voluntary, QHF has
seen broad participation by physicians. This is
partly because the program provides physicians
with individual, patient-specific alerts that inform
physicians when measures are not being met in
specific patients and what action is needed.
Another example are providers at Geisinger Health
System who developed a “bundle” of best practice
measures for diabetes and are using an electronic
registry derived from a fully integrated EHR to
improve physician performance in diabetes care.
Physician performance was then compared to
national benchmarks and to peers through an audit
and feedback program that included computerized
reminders. Physicians that met or improved on
the diabetes bundles criteria received payment
bonuses. Evaluations of the program found that it
resulted in significant increases in vaccinations for
pneumococcal disease and influenza (by 43 percent
and 29 percent respectively). The percentage of
patients with ideal glucose control (HBA1c<7.0) and
blood pressure control also improved (by 8 percent
and 11 percent respectively). Finally, the number of
patients receiving all nine “bundled” measurements
increased by 170 percent, from 2.4 percent to 6.5
percent.31
Organizational and Clinical Redesign
Many health systems have real potential to achieve
savings by redesigning their organization to realize
efficiencies in both clinical and administrative
processes. By focusing on improving patient
experience through better administrative systems
(e.g., more efficient scheduling processes), more
efficient supply systems, improved data sharing for
more efficient clinical practice, and better overall
system management, ACOs may be able to improve
patient care while lowering costs. Redesigning
organizational processes should be based on
an analysis of each individual ACO’s current
organizational system and identified inefficiencies.
Models have been developed to help break down
care delivery into manageable parts to identify
the truly functional units of clinical care. These
units, described as “clinical microsystems,” could
help ACOs better target their quality and cost
interventions in a more efficient manner.32
Discussed below are some examples of systems
that have borrowed organizational designs from
other industries to make their own processes more
streamlined, efficient, and patient-focused:
ThedaCare, a community health system in Northeast
Wisconsin, uses process redesign methods to fully
integrate their clinical and administrative processes.
Lean manufacturing techniques, developed by
Toyota, have helped maintain a focus on continuous
quality improvement and a reduction in “defects,”
which can include inefficiencies in care processes or
problems in care quality. Using data to drive process
change, staff members identify problems, track
progress towards improvement goals, and ensure
that once goals are achieved the improvements can
be made sustainable.
Specifically, ThedaCare seeks to improve a
patient’s care experience by decreasing defects
and wait time by 50 percent per year and aims to
increase productivity by 10 percent per year. To
accomplish these goals, ThedaCare sponsors
rapid improvement events, which take place
over seven weeks and work to identify and
eliminate organizational inefficiencies. By working
collaboratively with staff members to develop and
test new work processes, they gain more buy-in
and commitment from staff members across the
organization. Staff members are also provided with
forms that help identify and resolve any problems as
new work processes are rolled out.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 124
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
Another organization that used organizational
redesign to achieve real savings is the Seattle
Children’s Hospital. Seattle Children’s Hospital
also used a process redesign strategy used
by Toyota, known as continuous performance
improvement (CPI). CPI focuses on the patients’
entire experience at the hospital to determine ways
to not only lower costs, but also improve patient
experience. For example, improving the process to
sterilize surgical instruments allowed the hospital
to respond to an increased need for additional
procedures without having to invest in new facilities
and staff; additionally, improving scheduling
processes helped reduce the wait time for a MRI
from nearly a month to just one or two days. Other
CPI-identified redesigns were found to significantly
improve patient satisfaction, including beginning to
identify patient goals and improving communication,
as well as assembling outpatient services upon
admission while simultaneously reducing inpatient
stays and increasing the number of possible visits.
Because no single intervention may be substantial
and robust enough to lead to the kind of populationlevel improvements in care that are needed to
achieve ACO quality and cost benchmarks, these
efforts to improve care may be most effective if they
are implemented in a coordinated and integrated
fashion.
To try and align hospital staff members behind
CPI initiatives, Seattle Children’s Hospital holds
regular workshops that include doctors, nurses,
administrators, and representatives of patients’
families to walk through a typical patient experience
to help identify areas of possible improvement.
This collective process has helped ensure there is
ownership of these initiatives by not only hospital
leadership but also by physicians and patients.
The steps to change the way care is delivered
are difficult and require not only new kinds of
financial investments but also time and effort
on the part of leaders and clinicians. In effect,
these specific steps can and should add up to
meaningful cultural change in an organization,
helping create the alignment between financing
and delivery reform that is key to successful ACO
implementation. In turn, these initial steps create
a stronger organizational foundation for further,
more sophisticated steps to continue to achieve
measurable improvements in quality of care and
efficiency.
CONCLUSION
This section discusses examples of some of the
many clinical transformation activities that can be
undertaken to achieve meaningful improvements in
the quality and efficiency of care for ACO patients.
Nearly all of the delivery transformation efforts
described in this section are enabled through
health IT – such as electronic medical records,
patient registries, and electronic decision support
systems – which can help to identify high-risk
patients, encourage better information sharing and
coordination of treatments, and assist providers
in making high-value, evidence-based health care
decisions. By layering multiple reforms, ACOs
will have a greater chance of driving system-wide
change that can lead to better quality and lower
costs and justify health IT investment, which alone
are unlikely to be long-term cost-savers.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 125
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
ENDNOTES
1. McKethan A et al., “Improving Quality and Value in the U.S. Health Care System”, Bipartisan Policy Center, August
2009, http://www.brookings.edu/~/media/Files/rc/reports/2009/0821_bpc_qualityreport/0821_bpc_qualityreport.pdf.
2. E.A. Coleman et al., “Posthospital Medication Discrepancies: Prevalence and Contributing Factors,” Archives of
Internal Medicine 165, no. 16 (2005): 1842-1847.
3. H.H. Pham and P.B. Ginsburg, “Care Patterns in Medicare and Their Implications for Pay for Performance,” New
England Journal of Medicine 356, no. 11 (2007): 1130-1139. New England Journal of Medicine 356, no. 11 (2007): 11301139.
4. E.A. Coleman et al., “Preparing Patients and Caregivers to Participate in Care Delivered Across Settings: The Care
Transitions Intervention,” Journal of American Geriatric Society 52, no. 11 (2004): 1817-1825.; E.A. Coleman et al., “The
Care Transitions Intervention: Results of a Randomized Controlled Trial,” Archives of Internal Medicine 166, no. 17 (2006):
1822-1828.
5. Naylor, M.D., et al. “Transitional Care of Older Adults Hospitalized with Heart Failure: A Randomized, Controlled Trial”
Journal of the American Geriatric Society, Volume 52, pp 675-684, 2004.
6. Howell E, Bessman E, Kravet S, Kolodner K, Marshall R, Wright S. “Active bed management by hospitalists and
emergency department throughput.”Ann Intern Med. 2008 Dec 2;149(11):804-11.; Lindenauer PK, Rothberg MB, Pekow
PS, Kenwood C, Benjamin EM, Auerbach AD. “Outcomes of care by hospitalists, general internists, and family physicians.
“N Engl J Med. 2007 Dec 20;357(25):2589-600.; Peterson MC. “A systematic review of outcomes and quality measures in
adult patients cared for by hospitalists vs nonhospitalists.”Mayo Clin Proc. 2009 Mar;84(3):248-54.
7. Guided Care Website. Baltimore (MD): John Hopkins, Bloomberg School of Public Health; 2010. Available from: http://
www.guidedcare.org/.
8. Boult C, Green AF, Boult L, Pacala J, Snyder C, Leff B. Successful Models of Comprehensive Care for Older Adults with
Chronic Conditions: Evidence for the Institute of Medicine’s ‘Retooling for an Aging America’ Report. J Am Geriatr Soc.
2009; 57(12): 2328 – 2337.
9. Engelberg Center for Health Care Reform. Health Care Reform and Older Americans: Achieving Better Chronic Care at
Lower Costs [Event transcripts]. Washington (DC): The Brookings Institution; 2010. Available from: http://www.brookings.
edu/events/2010/0128_older_health_care.aspx
10. T. Bodenheimer, E.H. Wagner, and K. Grumbach, “Improving Primary Care for Patients with Chronic Illness,” Journal
of the American Medical Association 288, no. 14 (2002): 1775-1779.
11. Camden Coalition of Healthcare Professionals, “Camden Citywide Diabetes Collaborative,” http://www.camdenhealth.
org/index/diabetescollaborative.cfm.
12. C.M. Castro et al., “Asthma intervention program prevents readmissions in high healthcare users,” American Journal
of Respiratory & Critical Care Medicine 168, no. 9 (2003): 1095-1099.
13. S. Krishna et al, “Internet enabled interactive multimedia asthma education program: a randomized trial,” Pediatrics
111, no. 3 (2003): 503-510.; J.W. Krieger et al., “The Seattle-King County Healthy Homes Project: a randomized, controlled
trial of a community health worker intervention to decrease exposure to indoor asthma triggers,” American Journal of
Public Health 95, no. 4 (2005): 652-659.
14. M.W. Rich et al., “A Multidisciplinary Intervention to Prevent the Readmission of Elderly Patients with Congestive
Heart Failure,” New England Journal of Medicine 333, no. 18 (1995): 1190-1195.
15. New England Healthcare Institute. “Thinking Outside the Pillbox: A System-wide Approach to Improving Patient
Medication Adherence for Chronic Disease” Research Brief (August 2009)
16. B.B. Green; A.J. Cook; J.D. Ralston; et al. “Effectiveness of Home Blood Pressure Monitoring, Web Communication,
and Pharmacist Care on Hypertension Control: A Randomized Controlled Trial” JAMA 299;24 (2008):2857-67
17. http://www.communitycarenc.com/PDFDocs/MercerOTC.pdf
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 126
PART 5: HEALTH CARE DELIVERY TRANSFORMATIONS FOR ACHIEVING HIGH-VALUE HEALTH CARE | ACO TOOLKIT
18. L.M. Alexcih, S. Shen, I. Chan, J. Drabek, “Individuals Living in the Community with Chronic Conditions & Functional
Limitations: A Closer Look.” Prepared for the Assistant Secretary of Planning and Evaluation, Department of Health and
Human Services. January 2009.
19. Policy Council Document. Post-Acute Care Reform Plan.
20. R. Berenson. “Challenging the Status Quo in Chronic Disease Care.” Prepared for the California Healthcare
Foundation. Sept. 2006.
21. Alex H. Krist, Steven H. Woolf, Robert E. Johnson, and J. William Kerns. “Patient Education on Prostate Cancer
Screening and Involvement in Decision Making” Annals of Family Medicine 5 (2007):112-119
22. Bodenheimer T, Lorig K, Holman H, Grumbach K.”Patient self-management of chronic disease in primary care.”
JAMA. 2002 Nov 20;288(19):2469-75.
23. ES Fisher et al., “The Implications of regional Variation on Medicare Spending, Part 1: The Content, Quality, and
Accessibility of Care,” Annals of Internal Medicine 138, no.4(2003):272-87; ES Fisher et al., “The Implications of Regional
Variation in Medicare Spending, Part 2: Health Outcomes and Satisfaction with Care,” Annals of Internal Medicine 138,
no.4(2003):288-298.
24. TJ Mattimore, NS Wenger, NA Desbiens, et al., “Surrogate and physician understanding of patients’ preferences for
living permanently in a nursing home,” Journal of American Geriatric Society 45, no 7 (1997): 818-824.
25. LW Stevenson, AS Hellkamp, CV Leier, et al., “Changing Preferences for Survival After Hospitalization With Advanced
Heart Failure,” American College of Cardiology 52, no. 21 (2008): 1702-1708.
26. AHRQ, “Palliative Care Nurses in Primary Care Clinics Reduce Hospital Admissions, Increase Use of Hospice and
Home Care for Patients Nearing End-of-Life,” September 2009, http://www.innovations.ahrq.gov/content.aspx?id=1738 .
27. Sharma G., Freeman, J., Zhang D., Goodwin, J.S. “Continuity of care and ICU utilization during end of life” Arch Intern
Med. 2009 January 12; 169(1): 81–86..
28. C.W Cranor, B.A. Bunting, D.B. Christensen, “The Asheville Project: Long-term Clinical and Economic Outcomes of
a Community Pharmacy Diabetes Care Program,” Journal of the American Pharmaceutical Association 43, no. 2 (2003):
173-184.; B.A. Bunting and C.W. Cranor, “The Asheville Project: Long-Term Clinical, Humanistic, and Economic Outcomes
of a Community-Based Medication Therapy Management Program for Asthma,” Journal of the American Pharmaceutical
Association 46, no. 2 (2006): 133–147.
29. Geisinger Health Care System, “ProvenCare by the Numbers”, October 2008, http://www.geisinger.org/provencare/
numbers.html.
30. M.E. Chernew et al, “Impact of Decreasing Copayments on Medication Adherence within a Disease Management
Environment,” Health Affairs 27, no. 1 (2008): 103-112.; H.A. Huskamp et al, “The Effect of Incentive-Based Formularies on
Prescription- Drug Utilization and Spending,” New England Journal of Medicine 349, no. 23 (2003): 2224-2232.
31. V Weber, F Bloom, S Pierdon, and C Wood, “Employing the Electronic Health Record to Improve Diabetes Care: A
Multifaceted Intervention in an Integrated Delivery System,” Journal of General Internal Medicine 23, no. 4 (2007): 279-282
32. E Nelson, P Batalden, J Lazar. Practice Based Learning: A Clinical Improvement Action Guide. Oakbrook Terrace, IL:
Joint Commission Resources; 2007.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE January 2010 | 127
Part 6
Legal Issues for ACOs
Katherine Funk, Chris Janney
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
PART 6: LEGAL ISSUES FOR ACOS
One of the often cited hurdles to the formation of
ACOs is the legality of their formation and operation.
This section focuses on five of the most likely legal
issues that elements of ACO implementation raise
(e.g., organizational model, flow-of-funds, data
sharing):
• Section 6.1: Federal antitrust law;
• Section 6.2: Federal physician self-referral (or
“Stark”) law;
• Section 6.3: Federal health care program antikickback law;
• Section 6.4: Federal services reduction civil
•
monetary penalty law; and
Section 6.5: Federal tax law.
This section also touches briefly on state antitrust
laws, state fraud and abuse laws, federal and
state false claims acts, government managed care
regulations, corporate practice of medicine, and
state insurance law.
The legal topics covered in this section are by no
means an exhaustive list of legal barriers facing
ACOs. Among other things, there are a wide variety
of actual and potential ACO “models,” each of
which raises a unique set of legal issues. Further,
each state has its own, specific set of potentially
relevant fraud and abuse and other laws and
regulations.
Readers should not construe this document as
constituting legal advice. Any organization that is
considering participating in an ACO should engage
legal counsel to review the particular aspects of the
proposed structure and operations for compliance
with all relevant federal and state laws. In addition,
the legal guidelines discussed below are not
static; new rules and precedents applicable to
ACOs are expected in the coming year. As noted
in the discussion below, potential changes may
include applicable guidance from state and federal
agencies, rulemakings, Secretary “safe harbor” rules
under ACA, and even legislative reform.
6.1: FEDERAL ANTITRUST LAW
Overview
The antitrust discussion in this chapter reflects the
antitrust analysis of the activities of ACOs within
commercial insurance markets. There are few if any
antitrust issues that arise in the Medicare/Medicaid
markets because the payer (the government) sets
the rates and the rates are transparent. Antitrust is
included in this toolkit because it is likely (indeed
it is already happening) that ACOs forming in
response to ACA will want to also transact with
commercial payers. And of course, there will likely
be ACOs that form for the primary purpose of
engaging with commercial payers.
Measured against other health care industry legal
issues, the antitrust issues arising out of ACO
formation and operation are modest. In part, the
modest level of antitrust risk is due to the high
degree of congruence between the principles
underpinning the antitrust analysis and the stated
goals of ACOs under ACA and as contemplated by
Brookings/Dartmouth. Antitrust laws are premised
on the principle that competition generally benefits
consumers by producing the best combination of
quality, goods, and services at the lowest prices.
Compare with the premise of ACOs: that patient
care integration will lead to better care at lower cost.
Additionally, the Federal Trade Commission (FTC)
and the Department of Justice (DOJ) have produced
a great deal of guidance on provider integration.
Nascent ACOs should use this guidance to
construct organizations that are both lawful and that
achieve the promised cost-savings and improved
outcomes.1 Much of this guidance is available on
the FTC website at www.ftc.gov/bc/healthcare/
index.htm. Simply put, the antitrust issues raised
by the formation and operation of ACOs are both
navigable and manageable.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 129
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
Review of Basic Antitrust Concepts2
Before undertaking an antitrust analysis of the
transactional relationships that an ACO is likely to
encounter, we first introduce and discuss the two
principal antitrust categories into which conduct is
divided: (1) “rule of reason”; and (2) “per se.”
The Rule of Reason
The vast majority of conduct is analyzed under
the “rule of reason.” A fact-finder will examine
all of the facts and circumstances surrounding
the arrangement at issue to determine whether,
on balance, the anticompetitive effects of the
arrangement substantially outweigh its procompetitive benefits. Key considerations include
the following:
• Whether the arrangement at issue has the
•
•
potential to achieve significant efficiencies
(e.g., cost savings that are passed along to
consumers, or more robust competition);
Whether the pro-competitive efficiencies of
the arrangement outweigh any anticompetitive
effects; and
Whether the arrangement involves any
agreements that usually are considered
anticompetitive (e.g., agreements on price) and,
if so, whether those agreements are reasonably
necessary to achieve the efficiencies sought.
The Per Se Rule
The “per se” rule typically is reserved for conduct
that is known to have substantial anticompetitive
consequences — such as “naked” price fixing
(arrangements whose primary purpose is to fix
prices) — and few, if any, beneficial attributes.
Because such conduct is per se illegal the parties
are not given an opportunity to justify their conduct
(e.g., “the prices we agreed upon were low and the
payer saved money”). If an ACO is constructed
and operated with the pro-competitive goals and
intentions promulgated by Brookings/Dartmouth,
the antirust risk for that ACO will likely be very
manageable. But because ACOs may contain
participants who are direct competitors of one
another, there is a risk that such participants might
share information that could adversely affect
competition in transactions outside the ACO (e.g.,
two hospitals that are members of the same ACO
allocate non-ACO patients in the same ZIP code).
Under certain circumstances, such conduct could
rise to “per se” unlawful. In many if not most
situations, however, there are relatively simple and
straightforward procedures that can be put in place
to lessen the likelihood that unlawful behavior takes
place.
Applying the Antitrust Concept to an ACO
ACOs raise two main antitrust questions:
1. Whether the ACO is an arrangement to enable
its providers to fix prices – or whether it is an
integrated joint venture that has the potential to
improve quality and lower costs.
2. Whether the formation of the ACO creates
market power or allows it to be exercised in new
ways.
These questions will generally arise under the
following headings:
• Pricing Agreements
• Provider fee negotiations with
•
commercial payers
ACO Market Power
• Market allocation issues
• Provider exclusivity
Pricing Agreements
Provider Fee Negotiations. In general, antitrust
laws prohibit competing providers from jointly
negotiating their reimbursement rates with
commercial payers.3 This type of conduct can rise
to per se illegal price fixing. There is, however, an
exception where the providers have integrated for a
valid purpose, e.g., improved outcomes, and where
joint negotiations are necessary to the achievement
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 130
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
of those efficiencies. In other words, if providers
have integrated their services to create a product
or service that none of the participants alone could
produce, or produce as efficiently, antitrust law
generally will permit that joint venture, so long
as the integration does not create or enhance
market power (see discussion below for potential
suggestions to mitigate this problem).
The FTC and the DOJ have found that, when
negotiating provider fees or remuneration from
payers, the pro-competitive aspects of the following
two types of provider networks or joint ventures
outweigh the anticompetitive effects: (1) the
substantial financial risk model; and (2) the clinical
integration model.
Substantial Financial Risk. Financial risksharing gives providers the incentive to cooperate
in controlling cost and improving quality of
care. Providers that organize around substantial
financial risk-sharing closely resemble common
organizational forms such as LLCs, partnerships,
etc. As a threshold matter, determining whether
an arrangement does, in fact, involve the sharing
of substantial financial risk requires a review of the
particular facts. At a minimum, the risk of loss must
be real and that risk must be shared across the
spectrum of providers involved in the arrangement.
“Risk pools” in which individual providers bear
risk, but do not share risk generally, will not suffice.
While risk arrangements can take a variety of forms,
most typically include some combination of the
following: (1) capitation; (2) a substantial withhold;4
(3) a percentage of premium; (4) global fees or allinclusive case rates; and (5) cost and utilization
targets. Finally, in their Guidelines, the FTC and
DOJ recognize that other types of risk-sharing
arrangements may exist, and that they will consider
“other arrangements through which the participants
. . . may share substantial financial risk in the
provision of medical services through the network.”5
Clinical Integration. The clinical integration model
perhaps most closely resembles the standard
ACO, i.e., that of financially separate providers
who come together to better treat patients by
integrating their clinical practices. A clinically
integrated network is one that can6 engage in joint
pricing or collective negotiations without sharing
substantial financial risk in the manner described
above. Such arrangements can lawfully engage
in joint negotiations with payers if they include
“an active and ongoing program to evaluate
and modify practice patterns by the network’s
physician participants and create a high degree of
interdependence and cooperation among network
participants.”7
So what does this mean? There is no single way
to structure a clinically integrated network. This
flexibility is an asset. Rather than a “one size fits
all” approach, clinically integrated networks can
conform to the demands and requirements of
their respective communities and the other legal
restrictions faced by providers, while also staying
within the broad bounds of the antitrust laws.
While there is no “cookie cutter” structure, in the
Guidelines the FTC and DOJ do provide guidance
on the structural pillars that clinically integrated
networks often have:
• Mechanisms to monitor and control utilization of
•
•
health care services that are designed to control
costs and assure quality of care;
Selectivity in choosing network physicians who
are likely to further these efficiency objectives;
The significant investment of capital, measured
in both monetary and human terms, in the
necessary infrastructure and capability to realize
the claimed efficiencies.
As clinically integrated networks become more
widespread, there will likely be a longer and more
varied list of “common” pillars. For example,
where an ACO is organized around primary care
physicians, the measure of capital investment may
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 131
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
be organization-wide rather than on an individual
doctor basis. This is important where the initial
capital required to form an ACO is significant and
the physicians do not individually have the financial
resources to make significant investments.
In addition to the pillars, the Guidelines also provide
examples of the types of processes that clinically
integrated networks have utilized to implement the
pillars:8
• Electronic health records;
• Case management, pre-authorization (at
•
•
•
least of some services), and concurrent and
retrospective review of inpatient stays;
Development of practice standards and
protocols to govern treatment and utilization of
services, and active review of the care rendered
by each doctor in light of those standards and
protocols;
• Regular evaluation of both individual
physician’s and the entity’s aggregate
performance with respect to those
goals;
• Modification of individual physician’s
actual practices, where necessary,
based on those evaluations;
• Physicians who fail to adhere to the
standards and protocols will be subject
to remedial action, including the
possibility of expulsion/termination.
Reports to payers on the cost and quantity of
services provided, and the group’s success in
meeting its goals; and
A medical director and support staff to handle
the above functions and to coordinate patient
care in specific cases.
Not every ACO will have the desire or financial
capability to implement all of these processes.9
Again, there are no hard and fast rules as to how
many of these processes a network must adopt in
order to be deemed “clinically integrated.” Clearly,
the more of these types of processes that are
adopted the more secure a network should feel that
it is operating within the boundaries of the law.
Market Power
The assumption that goes along with financially
integrated and clinically integrated networks is that
the networks intend to negotiate as a collective
with commercial payers. As stated above, where
a network achieves substantial financial integration
or is clinically integrated, it is able to avoid the
antitrust issues typically present where competitors
act in concert with one another (i.e., Section 1 of the
Sherman Act).
Recall, however, that coordinated pricing between
competitors is only the first of two antitrust
questions that are put to provider networks. Even
where the coordinated pricing is permissible, the
integration – whether financial or clinical – of a
significant number of providers implicates the
second antitrust question: whether the formation
of the ACO creates market power or allows it to
be exercised in new ways. Market power is the
ability to raise prices above the competitive level
or exclude competitors from the market. Courts
usually look to market share as an indicator of
market power. Whether a provider organization has
market power will vary depending on the size of the
organization and how many competing providers
are in a particular market.
A finding of market power first requires a definition
of the relevant product market and the relevant
geographic markets.
Examples of product markets might be:
• Primary Care Physicians
• Specialists (e.g. Orthopedists, Urologists,
Neurologists, Child Psychiatrists, etc.)
• Multi-Physician Organizations
• Hospitals
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 132
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
Examples of geographic markets include:
•
•
•
•
Hospital referral areas
Metropolitan statistical areas
City
County
The relevant product market in which an ACO
competes will depend on the range of services
offered by the ACO. There are a couple of “shortcut” questions that ACOs can utilize to make a highlevel determination of market power concerns:
• Once the ACO is created, who remains to
willingness of plans to make such changes.
The Agencies will consider not only the
proportion of the providers in any relevant
market who are in the network, but also the
incentives faced by providers in the network,
and whether different groups of providers
in a network may have significantly different
incentives that would reduce the likelihood of
anticompetitive conduct. If plans can contract
at competitive terms with other networks or
with individual providers, and can obtain a
similar quality and range of services for their
enrollees, the network is less likely to raise
competitive concerns.
compete with it?
• Who can a payer turn to for the same services
as those offered by the ACO?
The geographic market analysis focuses on where,
as a practical matter, patients could go if the
participants in the ACO raise their prices above
competitive levels. For some services (especially
complex, tertiary care services), the relevant
geographic market could include other providers
offering similar services although located some
distance from the providers in the ACO.
Once the relevant product and geographic markets
are defined, the next question is whether the ACO
has attained sufficient market power such that it
could behave anti-competitively. Note that even
assuming that a market share analysis revealed
high market shares in properly defined product
and geographic markets, it does not necessarily
mean that the joint venture is unlawful. As the
Guidelines10 note:
. . . [I]n assessing the likely competitive
effects of a multi-provider network, the
Agencies are particularly interested in the
ability and willingness of health plans and
other purchasers of health care services to
switch between different health care providers
or networks in response to a price increase,
and the factors that determine the ability and
Addressing concerns about market power:
Avoiding or limiting provider exclusivity. Provider
exclusivity (to a single network, to a single payer
or ACO) can create antitrust problems when the
providers involved in the exclusivity arrangement
constitute a substantial percentage of the providers
in the relevant geographic market. This is true even
when the exclusivity carries a substantial benefit
(e.g., a stable network of providers committed to
achieving cost-containment goals) because, in
certain markets, there may be few, if any, providers
left for the remaining networks or insurers. While
concerns about exclusivity will vary from case
to case, as a rough rule of thumb, exclusive
networks which consist of greater than 35 percent
of the providers in the market may raise antitrust
concerns. The Guidelines have established a 20
percent threshold for exclusive physician networks
seeking to qualify for “safety zone” treatment.11
Failure to adhere to the safety zone percentage
does not necessarily mean that the ACO will violate
the antitrust laws, but it may mean that the network
will be scrutinized more closely. All things being
equal, exclusivity is generally more of a concern
in rural markets than in more urbanized markets
because, in a rural market, there may be fewer
options available for payers (or networks) seeking to
contract with providers.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 133
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
Exclusivity is sometimes not explicitly stated in an
agreement. Exclusivity can be expressed in other
ways, such as in a “right of first refusal” clause,
or it can be inferred from the circumstances (e.g.,
physicians in an IPA refuse to contract directly with
payers, even though their IPA agreement states that
their participation is on a non-exclusive basis). The
Guidelines describe certain factors which tend to
indicate that a network arrangement is truly nonexclusive:12
Addressing concerns about market power:
Alternative payment models. In those situations
where an ACO may have market power, using
alternative payment models may be a way of
addressing concerns. For example:
• The ACO can be established as a separate
• Viable competing networks or managed care
•
•
•
•
plans with adequate physician participation
currently exist in the market;
Physicians in the network actually individually
participate in, or contract with, other networks
or managed care plans, or there is other
evidence of their willingness and incentive to do
so;
Physicians in the network earn substantial
revenue from other networks or through
individual contracts with managed care plans;
The absence of any indications of significant
participation from other networks or managed
care plans in the market; and
The absence of any indications of coordination
among the physicians in the network regarding
price or other competitively significant terms of
participation in other networks or managed care
plans.
Notably, the Brookings/Dartmouth ACO model
assumes that primary care physicians will be
“exclusive” to one ACO, but that specialists will
not be exclusive. Whether the exclusivity of
primary care physicians raises any market power
issues depends upon a variety of factors, including
the number of other primary care physicians in
the relevant geographic market. The prevailing
assumption, however, is that the exclusivity
of primary care physicians is not problematic.
Nevertheless, this is an issue that should be
reviewed by counsel.
•
legal entity and have providers as members.
The ACO will not collectively negotiate
reimbursement with payers, but will instead
negotiate certain financial incentives with payers
based upon benchmarking. If the ACO meets
or exceeds those benchmarks, the payer will
provide some remuneration to the ACO. The
ACO will then, pursuant to its membership
agreement, make distributions to its members.
The providers will continue to operate under
their individually established commercial
contracts.
The ACO can involve commercial payers in
the formation of the organization. By involving
commercial payers in the discussions, the
ACO can insulate itself against allegations
of exercising market power to drive up
reimbursement. Importantly, the ACO would not
want to be exclusive to any commercial payer
or involve the commercial payer in the business
operations of the ACO such that it would have
access to competitively sensitive information
regarding other payers.
Other Common Antitrust Issues that ACOs May
Confront
Market Allocation Within an ACO or Between ACOs
Section 1 of the Sherman Act also prohibits two
horizontal competitors from agreeing on the
markets, customers or territories each will serve.
If they enter into such an agreement, they have
committed a per se violation of § 1 of the Sherman
Act unless they are part of an economically
integrated (i.e., risk sharing or clinically integrated)
joint venture (e.g., ACO) and the market allocation
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 134
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
agreement is reasonably necessary to achieve
the pro-competitive benefits of the ACO. In that
case, the market allocation agreement will likely be
reviewed under the rule of reason. For example, in
Statement 9 of the Guidelines, the FTC and DOJ
state that:
. . . [C]ompeting hospitals in an integrated
multi-provider network might need to agree
that only certain hospitals would provide
certain services to network patients in order
to achieve the benefits of integration. The
hospitals, however, would not necessarily
be permitted to agree on what services they
provide to non-network patients.
As the Guidelines highlight, the antitrust concern
in this area arises from information “spillover.”
For instance, as we briefly remarked upon at the
beginning of the chapter, there could be antitrust
issues where two hospitals within an ACO attempt
to transpose their intra-ACO agreements regarding
care rationalization to non-network patient
situations.
Problems with Boycotts and Refusals to Deal
In the context of provider contracting (either in
networking arrangements or with payers), it is
usually difficult, if not impossible, for an excluded
provider to state a viable boycott or refusal to deal
claim. The antitrust laws were enacted to protect
competition, not competitors, and accordingly,
the antitrust laws generally recognize that a
network has a great deal of latitude in choosing its
members.13 The antitrust laws are generally much
more concerned about networks that have too
many providers involved (especially if there is an
exclusivity clause) rather than too few. Where the
ACO from which the provider is being excluded,
however, has market power (e.g., there is only one
ACO in the geographic area from which the provider
draws patients, or the provider has evidence
demonstrating that the defendant specifically
intends to harm the plaintiff), provider exclusion
could present a problem. Further, antitrust
problems could arise, for example, if a large group
of specialists refused to do business with one of
four competing ACOs in a metropolitan area.
Nonprofit Institutions Act (NPIA)
If an ACO contains nonprofit institutions, it may
be confronted with NPIA issues which are unique
to nonprofit institutions (i.e., hospitals, schools
and other nonprofit entities). The NPIA creates an
exemption from the Robinson-Patman Act, which
prohibits price discrimination. The text of the NPIA
is found in 15 U.S.C. § 13c, which provides:
Nothing in [the Robinson-Patman Act] shall
apply to purchases of their supplies for their
own use by schools, colleges, universities,
public libraries, churches, hospitals, and
charitable institutions not operated for profit.
NPIA issues will most often occur in the context
of a nonprofit hospital’s purchases of prescription
drugs. Many nonprofit hospitals purchase
drugs at discounts and, in order to qualify for
the discount, they must be able to certify to the
drug manufacturer that the purchase is for the
hospital’s “own use.” If the use does not qualify as
the hospital’s “own use,” the manufacturer could
be exposed to a price discrimination claim by a
purchaser who did not receive the discount. The
hospital could also face liability under the RobinsonPatman Act, which precludes sellers from giving,
and buyers from receiving, discriminatory prices.
The Supreme Court defined “own use” in its 1976
decision in Abbott Laboratories, Inc. v. Portland
Retail Druggists Association.14 The court analyzed
the definition of “own use” by examining 10 different
categories of hospitals’ sales and dispensation
of drugs purchased at preferential prices. One
such category included “walk-in” customers. In
determining that NPIA discounts should not apply to
“walk-in” customers, the Court reasoned that:
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 135
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
[t]he extension of [NPIA] to the walk-in
customer, who has no present connection with
the hospital and its pharmacy other than as
a place to have his prescription filled, would
make the commercially advantaged hospital
pharmacy just another community drug store
open to all comers for prescription services
and devastatingly positioned with respect to
competing hospital pharmacies. This would
extend the hospital’s ‘own use’ concept
beyond that contemplated by Congress by
[NPIA].15
As hospitals have expanded into health care
systems consisting of multiple entities (e.g.,
home health care, nursing homes, hospices and
physicians), it has become increasingly difficult
to define exactly what constitutes “own use” by
a hospital. For example, is it permissible for a
nonprofit hospital to purchase prescription drugs
at a discount and then resell them at cost to its
affiliated, not-for-profit long-term care facility?
The answer to this question is yes. In an Advisory
Opinion issued to Presentation Health System,16
the FTC stated that the NPIA covered a hospital’s
transfer of drugs it purchased to affiliated nonprofit
long-term care facilities. The FTC stated that, in
light of the common ownership of the hospital and
the long-term care facilities, “[t]he Presentation
organization may be regarded as a unit having
purchased the pharmaceuticals for its ‘own
use’ comprised of the use by its hospital and
its long-term care facilities.” Thus, resale of the
pharmaceuticals to the long-term care facilities
would be exempt as long as they were for the longterm care facilities’ own use.17
In an Advisory Opinion issued to Harvard Vanguard
Medical Associates, Inc. (HVMA) in December
2001, the FTC stated that HVMA, through its clinic
pharmacies, may dispense products purchased
under NPIA to the clinic’s patients. HVMA is
a nonprofit clinic composed of several health
care practitioners. HVMA also operates its own
pharmacy. The FTC stated that the dispensation
of products purchased under NPIA by HVMA
pharmacies to HVMA’s patients who are under the
continuing care of an HVMA physician is acceptable
and meets the definition of “own use.”
While it is difficult to articulate a bright line rule for
the application of NPIA to the gamut of hospital
activities, it does seem reasonably clear that if
the activity promotes the nonprofit institutions’
“intended institutional operation in the care of
persons who are its patients,” an argument exists
that the activity should be protected by the NPIA.18
Ancillary Service Referrals
Currently, hospitals’ ownership of, or affiliation
with, ancillary service businesses, such as hospice
or home health, creates the opportunity for
exclusive referral arrangements. It is likely that this
competition concern would be elevated within the
ACO model because one of the ways that costsavings can be obtained is by creating efficiencies
in the referral relationship. An exclusive referral
arrangement is potentially subject to attack under
§ 1 of the Sherman Act (as an exclusive dealing
claim, assuming the hospital is only affiliated with,
and does not own, the ancillary service business).
The critical issue under § 1 of the Sherman Act
is foreclosure from referrals. If the hospital’s
percentage of patients needing ancillary services
constitutes a substantial percentage of all referrals
to ancillary service providers in the relevant
geographic market, and those patients are “steered”
to the hospital’s affiliate or subsidiary exclusively,
the potential for antitrust problems could be
significant because there may be an insufficient
number of remaining referrals for competitors. As
a general proposition, however, control of at least
30 percent of the referral market is usually needed
before antitrust concerns arise.19 For example,
assume that an ACO in a mid-sized city has within it
all of the hospitals operated by a nonprofit religious
organization. All of these hospitals also have on
their campuses senior living facilities that provide a
continuum of care from independent living to skilled
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 136
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
nursing. If the hospitals within the ACO employ
more than 30 percent of all geriatric physicians
within the metropolitan service area, then it may be
the case that the senior living facilities associated
with the hospitals will receive a substantial number
of referrals from the geriatric physicians, thereby
potentially foreclosing referrals to senior living
facilities not associated with hospitals within the
ACO.
Conclusion
The promise of the ACO model, to reduce costs and
improve outcomes, is exactly the type of economic
conduct that the antitrust laws seek to promote.
Nevertheless, as with any collaboration between
competitors or potential competitors, it is important
for participants to be aware that the potential does
exist for anticompetitive conduct to take place.
The purpose of this overview is to alert those who
are forming ACOs and those who participate in an
ACO of the likely pitfalls, and provide a pathway
to navigate around them (see appendix for more
information on federal Antitrust statutes).
6.2: FEDERAL PHYSICIAN-SELF REFERRAL (OR
“STARK”) LAW
The federal physician self-referral law (or “Stark
Law”) has two basic prohibitions: a referral
prohibition and a billing prohibition. Pursuant to the
referral prohibition, absent an applicable exception,
a physician who has a “financial relationship” with
an “entity,” or a physician with an “immediate family
member” who has such a financial relationship,
may not make a “referral” “to” that entity for the
“furnishing” of “designated health services” (DHS)
for which payment may be made by the Medicare
program.20 For example, if a physician has a
financial relationship with a hospital, the physician
cannot refer any Medicare patients to the hospital
for inpatient or outpatient services, both of which
are DHS, unless the financial relationship fits into an
exception.
Pursuant to the billing prohibition, absent an
applicable exception, a health care provider may
not bill for improperly referred services. Specifically,
an entity that furnishes DHS pursuant to a
prohibited referral may not “present” or “cause to
be presented” a claim or bill for such services to
the Medicare program or to any other individual
or entity, including secondary insurers and the
patient.21 (Thus, in our example above, just as
the physician would be prohibited from referring
patients to the hospital, the hospital would be
prohibited from billing for any services that it
furnishes to such patients.)
According to the Centers for Medicare & Medicaid
Services (CMS), the Stark Law reflects Congress’
concern that a physician with a financial stake in
determining whether or where to refer a patient may
be “unduly influenced by a profit motive,” thereby
undermining efficient utilization, patient choice, and
competition among participants in federal health
care programs.22 More specifically, CMS believes
that:
• Physicians can “overutilize by ordering items
•
•
and services for patients that, absent a profit
motive, they would not have ordered,”23
A patient’s choice “can be affected when
physicians steer patients to less convenient,
lower quality, or more expensive providers of
health care, just because the physicians are
sharing profits with, or receiving remuneration
from, the providers,”24 and
Where referrals are “controlled by those sharing
profits or receiving remuneration, the medical
marketplace suffers since new competitors can
no longer win business with superior quality,
service, or price.”25
Where a physician has violated the referral
prohibition and an entity has violated the billing
prohibition, several sanctions may be imposed.
First, an entity that collects payment for DHS
performed pursuant to a prohibited referral must
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 137
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
refund all collected amounts on a timely basis.26
Second, any person “who presents or causes to
be presented a bill or claim” for improperly referred
DHS and “knows or should know” that the claim
is for improperly referred DHS is subject to (1) a
civil monetary penalty (CMP) of up to $15,000 per
service, (2) an assessment (in lieu of damages)
of up to three times the amount claimed, and (3)
exclusion from participation in any federal health
care program.27 Finally, any physician or entity that
knowingly participates in a “scheme” to circumvent
the Stark Law is subject to a CMP of up to $100,000
and may be excluded from participation in federal
health care programs.28
Law’s underlying policy objectives. For this reason,
Congress and CMS have created some three-dozen
separate exceptions to the Stark Law’s prohibitions.
Complexity
Before turning to a discussion of how the Stark Law
might be implicated by various ACO arrangements,
several points are worth emphasizing.
In addition to the Stark Law’s overbreadth, and the
panoply of resulting exceptions, the Law can be
difficult to navigate for a third reason: many of the
Stark Law’s elements and exceptions are complex,
counterintuitive and, in some cases, have been
defined, interpreted, redefined and reinterpreted
on multiple occasions over the past two decades.
Here are but two examples: the Stark Law definition
of the word “referral” is more than 370 words long,33
and the term “indirect compensation arrangement”
was undefined by CMS until 1998,34 was defined
by CMS in 2001,35 and was redefined by CMS in
2004,36 2007,37 and 2008.38
Overbreadth of Law
Strict Liability
The Stark Law’s prohibitions are extremely broad.
For example: (1) a physician has a “financial
relationship” with any hospital with which the
physician has a “compensation arrangement,”29
(2) a compensation arrangement includes “any
arrangement” between a physician and hospital that
“involves remuneration,”30 and (3) “remuneration”
means “any payment or other benefit made directly
or indirectly, overtly or covertly, in cash or in
kind.”31 Thus, if a hospital provides a physician with
anything of value, regardless of how small (e.g., note
pads),32 the hospital and physician have a “financial
relationship” and, in the absence of an exception,
the physician may not refer Medicare patients to
the hospital for DHS, and the hospital may not bill
anyone for DHS furnished to such patients.
Proliferation of Exceptions
Because its prohibitions are so broad, the Stark Law
is overinclusive, implicating thousands of common,
everyday provider-physician arrangements, the vast
majority of which do not offend any of the Stark
To raise the compliance stakes still higher, the Stark
Law is (generally speaking) a “strict liability” statute.
That is, unlike one of the Stark Law’s cousins —
the federal health care program anti-kickback
law,39 which is violated only if the defendant acts
“knowingly and willfully” — the Stark Law may be
violated even if the parties do not intend to violate
the Law and are not aware that they are doing
so. For example, assume that a physician and a
hospital have a “financial relationship” because the
hospital has given the physician some notepads and
that this financial relationship does not fit into an
exception. Under these circumstances, each and
every time the physician refers a Medicare patient
to the hospital for DHS, the Stark Law’s referral
prohibition may be violated; and each and every
time the hospital bills Medicare (or anyone else) for
DHS furnished to such patients, the Stark Law’s
billing prohibition may be violated — all regardless
of whether the physician or the hospital intended to
violate the Stark Law or were even aware that such
violations were occurring.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 138
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
Private Enforcement
All of the factors that make the Stark Law so
challenging from a compliance standpoint probably
would be manageable for the health care industry
if the federal government had exclusive jurisdiction
to enforce the Law. Unfortunately, this is not the
case. Although the jurisprudence in this area is
evolving, several courts have concluded that when
a provider submits a claim for services that were
furnished as a result of a referral that violated the
Stark Law, that submission may constitute a “false
claim” for purposes of the Federal Civil False
Claims Act (FCA). The FCA, in turn, has a qui
tam (or “whistleblower”) provision, which allows
private individuals and organizations to bring FCA
actions in the name of (and on behalf of) the federal
government.40 If the whistleblower prevails, he or
she is entitled to keep as much as 30 percent of the
proceeds of the litigation (which may include treble
damages and a fine of up to $11,000 per claim), as
well as reasonable expenses and attorneys’ fees.41
With all of this background and context in mind, we
turn now to whether and how the Stark Law might
be implicated by ACO arrangements.
Given the variety of physician, provider and payer
arrangements that might fall into the ACO category,
coupled with the complexity of the Stark Law —
which is widely (and accurately) viewed as among
the most complicated of the federal health care
fraud and abuse statutes — there is no “one-sizefits-all” Stark Law analysis of ACO arrangements.
That being said, several observations can be made
relating to each of the above questions and their
potential application in the ACO context.
Referrals
Assuming that the ACO in question includes some
combination of physicians and hospitals, it is likely
that the participating physicians will have occasion
to refer Medicare beneficiaries to the participating
hospitals for the furnishing of inpatient
and/or outpatient hospital services. In some cases,
these beneficiaries may be enrolled in the ACO
in question. In other cases, the beneficiaries may
be patients of participating physicians but not
ACO enrollees. In either event, their referral to a
participating hospital may implicate the Stark Law.
Moreover, in addition to inpatient and outpatient
hospitals services, DHS include:
Potential Application to ACO Arrangements
As set forth above, determining whether an
arrangement violates the Stark Law essentially is a
three-step process.
1. Referrals. Does the arrangement involve a
“physician” making a “referral” “to” an “entity”
for the “furnishing” of “DHS” covered by
Medicare?
2. Financial Relationship. If so, does the
physician (or one of his or her immediate family
members) have a “financial relationship” with
the entity furnishing DHS, either in the form of
(1) a direct or indirect ownership interest or (2) a
direct or indirect compensation arrangement?
3. Exceptions. If so, does the arrangement
qualify for protection under one or more of the
Stark Law’s exceptions?
•
•
•
•
•
•
•
•
•
•
•
Clinical laboratory services,
Physical therapy services,
Occupational therapy services,
Outpatient speech-language pathology,
Radiology services, including MRIs,
computerized axial tomography (CAT) scans,
PET scans, and ultrasound services,
Radiation therapy services and supplies,
Durable medical equipment (DME) and supplies,
Parenteral and enteral nutrients, equipment and
supplies,
Prosthetics, orthotics, and prosthetic devices
and supplies,
Home health services, and
Outpatient prescription drugs.
Thus, to the extent that entities furnishing any of
these services participate in the ACO in question,
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 139
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
referrals of Medicare beneficiaries to such entities
by participating physicians may violate the Stark
Law.
Financial Relationship
If we assume that under the ACO arrangement
in question, participating physicians will have
occasion to refer Medicare patients to participating
hospitals (and/or other participating DHS entities),
the next question is this: will the ACO arrangement
create any “financial relationships” between these
physicians and these DHS entities? Unfortunately,
there is no single (or simple) answer, other than, “it
depends.”
As a threshold matter, determining whether a
“financial relationship” exists between a referring
physician and a DHS entity is highly fact-specific
and can be complex. This is particularly so where
the potential financial relationship takes the form
of a “compensation arrangement” (as opposed to
an “ownership or investment interest”). As noted
above, compensation arrangements may be “direct”
or “indirect” and CMS’ view (and interpretation) of
“indirect” compensation arrangements has varied
widely over the years.
Notwithstanding the above, the following safely can
be said: if (1) the ACO arrangement in question
provides for shared savings (or any other payments)
to be provided to participating physicians, and
(2) the source of this remuneration is a participating
hospital — either because the hospital is the
source of the funds in the first instance or because
the hospital is exercising control over funds
provided by a payer or other third-party — then
the arrangement likely will create a “compensation
arrangement” between the hospital and the
physician for Stark Law purposes. Under these
circumstances, the ACO would not be viable from a
Stark Law standpoint unless this hospital-physician
compensation arrangement meets the requirements
of one or more Stark Law exceptions, is covered by
an ACA waiver, or is the subject of a favorable CMS
advisory opinion.
On the other hand, if (1) the source of the funds in
question is not a participating hospital, but instead
is a payer, and (2) the funds flow directly from the
payer to the physician — without passing through
(or otherwise being controlled or influenced by)
a hospital (or other DHS provider) — then the
arrangement may not create a compensation
arrangement between the physician and any DHS
entity. Under these circumstances (that is, in the
absence of a financial relationship between the
referring physician and any DHS provider), it would
not be necessary to meet the requirements of any
exception to the Stark Law, for example, in order to
avoid liability thereunder.
In July 2008, CMS proposed a new Stark Law
exception that would cover certain shared savings
and similar arrangements (“July 2008 Proposed
Rule”). (This proposed exception is addressed
in detail below.) In discussing the need for such
an exception, CMS touched on the financial
relationship issue, largely confirming the provider/
payer dichotomy discussed above:
The provision of monetary or nonmonetary
remuneration by a hospital to a physician
through a gainsharing arrangement or other
incentive payment or shared savings program
would constitute a financial relationship with
an entity for purposes of the physician selfreferral statute.
* * *
We observe that payer-based programs in
which the performance measures are set by
a wholly independent, arms-length party with
a clear financial incentive to make [pay-forperformance] payments prudently may pose
somewhat less risk than non-payer based
programs, where there is no third-party payer
that sets the performance measures and
monitors compliance. We note further that
payments made directly from a payer to a
physician, at the payer’s sole discretion, may
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 140
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
not implicate the physician self-referral statute
or other fraud and abuse statutes.42
In sum, an ACO may be able to avoid implicating
the Stark Law exposure altogether if it is structured
in such a way that any remuneration that flows to
physicians pursuant to the ACO’s shared savings
(or similar) arrangements does not come from (and
is not controlled by) any hospital or other DHS
entity. If this cannot be avoided, however, then the
arrangement will need to meet the requirements of a
Stark Law exception.
Exceptions
Although there are several Stark Law exceptions
that are commonly employed where a hospital and
physician have a compensation arrangement —
including the employment43, personal services44,
fair market value45, and indirect compensation
arrangement exceptions46 — none of these was
designed with shared savings or similar programs
in mind. As a result, it is rarely the case that one
of these programs fits neatly into one of these
exceptions. CMS recognized as much in the
preamble to its July 2008 Proposed Rule:
The Medicare program and private industry
stakeholders are increasingly exploring the
benefits of various types of gainsharing,
pay-for-performance (P4P), value-based
purchasing, and similarly-styled programs that
use economic incentives to foster high quality,
cost-effective care. Many of these programs
involve payments from hospitals to physicians.
These payments potentially implicate the fraud
and abuse laws, including the physician selfreferral statute. Existing exceptions to the
physician self-referral statute, while useful,
may not be sufficiently flexible to encourage
a variety of non-abusive and beneficial
gainsharing, P4P, and similar programs.47
“For this reason,” the agency continues, “we
are proposing a new, targeted exception to the
physician self-referral statute for such programs.”48
According to CMS, the “design of the new
exception presents a particular challenge”:
crafting an exception that offers broad
flexibility for innovative, effective programs,
while at the same time protecting the
Medicare program and beneficiaries from
abuses. In reviewing various programs and
industry suggestions, we have been struck
by the considerable variety and complexity of
existing arrangements, and the likelihood of
continued future innovation in the structure
and method of these programs. This variety
and complexity make it difficult to craft a
“one-size-fits-all” set of conditions that are
sufficiently “bright line” to facilitate compliance
and enforceability, yet sufficiently flexible
to permit innovation without undue risk of
program or patient abuse. The variety and
complexity of these programs make them
potential vehicles for the unscrupulous to
disguise payments for referrals or compromise
quality of care for patients in the interest of
maximizing revenues.49
In light of these various concerns and
considerations, CMS decided to take a “cautious”
approach, proposing a “relatively narrow” exception
and conceding that it is “unlikely to cover as many
arrangements as interested stakeholders would
like.”50 Before turning to the exception’s specific
requirements, several preliminary points are worth
emphasizing.
• First, the proposed exception, although
narrow, is exceedingly detailed, consisting of
16 separate conditions, most of which (in turn)
have multiple sub-conditions.
• Second, CMS has sought comments on virtually
every element of the proposed rule and it is
likely that any final rule will include substantial
changes and modifications.
• Third, even if the proposed exception were
both simple and set in stone, it is important to
recall that the proposed exception is just that:
proposed. Thus, an ACO could not, today, rely
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 141
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
on this exception for protection under the Stark
Law.
With these caveats in mind, the proposed “Incentive
Payment and Shared Savings Programs” exception
(“Shared Savings Exception”) provides that
“[r]emuneration in the form of cash or cash
equivalent payments, but not including
nonmonetary remuneration, provided by a hospital
to a physician on the hospital’s medical staff or to
a qualified physician organization” will not create
a “financial relationship” for Stark Law purposes,
provided the following 16 conditions are met.
1. Purpose of Program. The “remuneration”
at issue must be “provided as part of a
documented incentive payment or shared
savings program” — or “program” for purposes
of this chapter — that is designed to achieve
(1) the “improvement of quality of hospital
patient care services through changes in
physician clinical or administrative practices”
or (2) “[a]ctual cost savings for the hospital
resulting from the reduction of waste or changes
in physician clinical or administrative practices,
without an adverse effect on or diminution in the
quality of hospital patient care services.”
2. Performance Measures. The program must
identify “patient care quality measures or cost
saving measures” (collectively, “performance
measures’’) that (1) use “an objective
methodology, are verifiable, are supported by
credible medical evidence, and are individually
tracked,” (2) are “reasonably related to the
hospital’s or comparable hospitals’ practices
and patient population,” (3) “[w]ith respect
to patient care quality measures, are listed in
CMS’ Specification Manual for National Hospital
Quality Measures,” and (4) are “monitored
throughout the term of the arrangement to
protect against inappropriate reductions or
limitations in patient care services.”
3. Performance Measure Baseline/Target
Levels. The program must establish (1)
“[b]aseline levels for the performance measures
using the hospital’s historical and clinical
data,” (2) “[t]arget levels for the performance
measures that are developed by comparing
historical data for the hospital’s practices and
patient population to national or regional data
for comparable hospitals’ practices and patient
populations,” and (3) “[t]hresholds above
or below which no payments will accrue to
physicians.”
4. Participating Physician Pool. At least
five physicians (the “participating physician
pool”) must “participate in each performance
measure.” Physicians participating in the
program (“participating physicians”) (1) “must
be on the medical staff of the hospital at the
commencement of the program,” and (2) “may
not be selected in a manner that takes into
account the volume or value of referrals or
other business generated between the parties.”
“A hospital may elect to make . . . [the]
program available to physicians in a particular
department or specialty, provided that the
hospital offers the opportunity to participate
in the . . . program to all physicians in the
department or specialty on the same terms and
conditions.”
5. Independent Medical Review. The program
must require “independent medical review of the
program’s impact on the quality of patient care
services provided at the hospital and corrective
action if the independent medical review
indicates a diminution in the quality of hospital
patient care services.” This review “must be
completed prior to the commencement” of
the program “(with respect to the program’s
potential impact on the quality of patient care
services provided at the hospital) and at least
annually thereafter.” For purposes of this
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 142
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
condition, “independent medical review” means
“written review by an individual or organization
that is (1) “[n]ot affiliated with the hospital, (2)
“[n]ot affiliated with any participating physician
or any physician organization to which any
participating physician belongs,” and (3) “[a]
t the time of the review, not participating in any
incentive payment or shared savings program at
the hospital.”
6. Selection of Supplies & Devices. Under the
program: (1) “[p]hysicians must have access
to the same selection of items, supplies or
devices as was available at the hospital prior
to the commencement of the program, and
must not be restricted in their ability to make
medically appropriate decisions for their
patients, including, but not limited to, decisions
about tests, treatments, procedures, services,
supplies or discharge,” (2) the “hospital
may not make a payment to a participating
physician or a qualified physician organization
for the use of an item, supply or device if the
physician or qualified physician organization
has an ownership or investment interest in,
or a compensation arrangement with, the
manufacturer, distributor or group purchasing
organization that arranges for the purchase of
the item, supply or device,” and (3) the “hospital
may not limit the availability of new technology
that is “linked through objective evidence to
improved outcomes and is clinically appropriate
for a particular patient” and “[m]eets the same
federal regulatory standards as technology
available under the incentive payment or shared
savings program (for example, approval by the
Food and Drug Administration and Medicare or
Medicaid coverage decisions).”
7. Patient Notice. The hospital must provide
“effective prior written notice to patients
affected by the incentive payment or shared
savings program” that (1) “[i]dentifies the
physicians participating in the program,” (2) “[d]iscloses that participating physicians receive
payments for meeting targets for performance
measures,” and (3) “[d]escribes the performance
measures in a manner reasonably designed to
inform patients about the program.”
8. Documentation of Arrangement and Shared
Savings Formula. The arrangement must be
“set out in writing,” “signed by the parties,”
and “specif[y] the remuneration (or a formula
for the remuneration) in detail sufficient
to be independently verified, including a
comprehensive description of the incentive
payment or shared savings program in which
the physician is participating, the applicable
baseline measures, and the targets for
performance measures to be achieved by
the participating physician.” To satisfy this
requirement, “each specific performance
measure and the resulting payment (or a formula
for the resulting payment) to the participating
physician or qualified physician organization
must be clearly and separately identified.”
9. Reasonableness of Arrangement. The
performance measures provided for under the
arrangement must “not involve the counseling
or promotion of a business arrangement or
other activity that violates any federal or State
law” and, in the aggregate, must be “reasonable
and necessary for the legitimate business
purposes of the arrangement.”
10. Term of Arrangement. The term of the
arrangement must be between one and three
years.
11. Payment Limitations – Double Dipping,
Diminution in Care. “Payments must take
into account previous payments made for
performance measures already achieved
to ensure that the participating physician
or qualified physician organization does
not receive payment related to patient care
quality improvements or cost savings that
were achieved during a prior period of the
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 143
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
arrangement.” In addition, “[n]o payment may
be made for the achievement of cost savings
that results in a diminution in hospital patient
care quality with respect to that performance
measure.”
12. Payment Limitations - Duration, Amount.
“For purposes of calculating the actual
payments to the physician, cost savings [must
be] measured by comparing the hospital’s
actual acquisition costs for the items and
supplies or costs of providing the specified
services that are subject to the shared savings
program to the hospital’s baseline costs for the
same items, supplies or services during the
[one]-year period immediately preceding the
commencement of the program.”
13. Distribution of Payments. “The remuneration
to be paid over the term of the arrangement
(or the formula for the remuneration)” (1) must
be “[s]et in advance,” must not “vary during
the term of the arrangement,” and must not
be “determined in a manner that takes into
account the volume or value of referrals or
other business generated between the parties,”
(2) must not be “based in whole or in part on a
reduction in the length of stay for a particular
patient or in the aggregate for the hospital,”
(3) must be “[d]istributed to the physicians in
each participating physician pool or in each
qualified physician organization if the qualified
physician organization consists of at least five
participating physicians on a per capita basis
with respect to each performance measure,”
and (4) must be “[p]aid directly to participating
physicians or qualified physician organizations.”
14. Overutilization of Federal Program Services.
The “remuneration paid to a participating
physician or qualified physician organization
may not include any amount that takes into
account the provision of a greater volume
of federal health care patient procedures
or services than the volume provided by
the participating physician or qualified
physician organization during the period of
the same length immediately preceding the
commencement of the program as that covered
by the payment.”
15. Documentation of Payments. The hospital
must maintain “accurate and contemporaneous
documentation of the incentive payment
or shared savings program and make such
documentation available to [HHS] upon
request,” including, but not limited to, the
following: (1) the “written agreement between
the parties,” (2) the “basis for the selection
of the performance measures,” (3) the
“selection and qualifications of the individual
or organization designated as the independent
medical reviewer,” (4) the “written findings of the
independent medical reviewer,” (5) the
“[c]orrective actions taken by the hospital based
on the written findings of the independent
medical reviewer (or any other review indicating
that corrective action was needed),” (6) the
“amount and calculation of payments made
under the incentive payment or shared savings
program, including the hospital’s projected and
actual acquisition costs where relevant,” (7) the
“re-basing of performance measures,” and (8)
the “written notification provided to hospital
patients.”
16. Anti-Kickback Law. The arrangement must
not violate the federal health care program
anti-kickback law or “any federal or state
law or regulation governing billing or claims
submission.”
Two final notes. First, where the payment at issue
is not coming from a hospital or other DHS entity,
but instead is coming from an insurer, at least two
existing Stark Law exceptions may be available
to protect certain “downstream” referrals. Under
the Stark Law’s so-called “risk-sharing” exception,
the Law’s referral and billing prohibitions do not
apply if the compensation arrangement at issue
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 144
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
consists of “compensation pursuant to a risksharing arrangement” — including, but not limited
to, “withholds,” “bonuses” and “risk pools” —
“between” an “MCO or IPA and a physician for
services provided” to “enrollees” of a “health plan,”
provided that the arrangement does “not violate the
[Anti-Kickback Law], or any federal or state law or
regulation governing billing or claims submission.”51
In addition, the Law’s so-called “pre-paid plans”
exception protects referrals that involve Medicare
beneficiaries who are enrolled in “pre-paid plans”.52
Pursuant to this exception, the Stark Law’s referral
and billing prohibitions simply do not apply to
services furnished to enrollees of any one of nine
types of pre-paid plans, including (but not limited to)
most Medicare managed care plans.53
Second, to the extent that pursuant to §3022 of
the ACA, an ACO arrangement includes individuals
enrolled in the Medicare fee-for-service program,
§3022 specifically authorizes the U.S. Department
of Health & Human Services to “waive” the
application of the Stark Law to approved ACO
arrangements. Thus, if a §3022 waiver is obtained
by the ACO at issue, referrals involving beneficiaries
who are enrolled in the ACO should not violate the
Stark Law.
Conclusion
For all of the reasons set forth above, the Stark
Law may pose obstacles to the formation and
operations of ACOs, depending on the specific
types of providers and arrangements at issue. In
particular, ACO models that call for hospitals or
other DHS providers to fund, or control the funding
of, payments to physicians are likely to implicate
the Stark Law. These obstacles should not be
insurmountable, however. Where payments are
made by non-DHS entities, for example, the Stark
Law may not be implicated and/or several Stark
Law exceptions may be available to protect patient
referrals. Further, depending on how liberally HHS
chooses to exercise its ACA waiver authority, such
waivers could go a long way toward protecting
additional patient referrals from the Stark Law’s
prohibitions.
6.3: FEDERAL HEALTH CARE PROGRAM
ANTI-KICKBACK LAW
The federal health care program anti-kickback
law (“Anti-Kickback Law”) is an older cousin of
the Stark Law. The Anti-Kickback Law is a broad
criminal statute that prohibits one person from
“knowingly and willfully” giving (or offering to give)
“remuneration” to another if the payment is intended
to “induce” the recipient to (1) “refer” an individual
to a person for the furnishing of any item or service
for which payment may be made, in whole or in
part, under a federal health care program (i.e., a
“covered item or service”); (2) “purchase,” “order,”
or “lease” any covered item or service; (3) “arrange
for” the purchase, order, or lease of any covered
item or service; or (4) “recommend” the purchase,
order, or lease of any covered item or service.54 The
Anti-Kickback Law also prohibits the solicitation or
receipt of remuneration for any of these purposes.55
“Remuneration” includes anything of value.56 The
term “inducement” has been interpreted to cover
any act that is intended to influence a person’s
reason or judgment.57 Some courts have held
that as long as “one purpose” of the payment
at issue is to induce referrals, the Anti-Kickback
Law is implicated.58 Under this one-purpose rule,
an arrangement may implicate the Law (1) even
if inducing referrals is not the primary purpose of
the payment and (2) even where there are other,
legitimate reasons for the arrangement. Courts
have also recognized, however, that a party may
hope or expect that a particular arrangement will
result in referrals without necessarily triggering the
one-purpose rule.59
Because the Anti-Kickback Law is so broad, it
covers a variety of common and non-abusive
arrangements. Recognizing this overbreadth,
Congress and the U.S. Department of Health &
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 145
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
Human Services Office of Inspector General (OIG)
— the lead enforcement agency with respect to
the Anti-Kickback Law — have established a large
number of statutory exceptions and regulatory
safe harbors (collectively, “safe harbors”). An
arrangement that fits squarely into a safe harbor is
immune from prosecution under the Anti-Kickback
Law. The safe harbors tend to be very narrow,
however, and the OIG takes the position that
immunity is afforded only to those arrangements
that “precisely meet” all of the conditions of a
safe harbor. Material or substantial compliance is
insufficient. Moreover, safe harbors do not exist
for every type of arrangement that does (or may)
implicate the Anti-Kickback Law.
Importantly, however, the fact that a particular
arrangement does not fit within a safe harbor does
not mean that the arrangement implicates (or
violates) the Anti-Kickback Law. In other words,
although there are certain types of remuneration that
necessarily implicate the Anti-Kickback Law and,
therefore, must be safe harbored in order to ensure
immunity from prosecution, most remuneration that
flows between and among health care entities does
not fall into this category.
For example, most hospitals have a variety of
arrangements pursuant to which they provide
remuneration to physicians who are in a position
to refer patients to the hospital. When a hospital
hires a physician to serve as a medical director, for
example, the hospital normally compensates the
physician for his or her services. This compensation
is, of course, “remuneration.” Just as plainly,
however, this remuneration does not necessarily
implicate the Anti-Kickback Law. Indeed, such
remuneration will implicate the Anti-Kickback Law
only if it is intended not only to compensate the
physician for his or her services, but also to induce
the physician to refer patients to the hospital. If
the compensation is not intended to induce patient
referrals, then — whether the arrangement is
safe harbored or not — the arrangement will not
implicate the Anti-Kickback Law.
Furthermore, the OIG recognizes that there are
many arrangements that do implicate the AntiKickback Law and are not covered by a safe
harbor, but that nevertheless do not implicate
any of the Law’s principal policy objectives and,
as such, do not pose a material risk of program
abuse or warrant the imposition of sanctions. In
a nutshell, the Law’s principal policy objectives
are to (1) prevent the overutilization of health care
items and services and any concomitant increase
in federal health care program costs, (2) promote
patient freedom of choice, and (3) promote market
competition.
Finally, because the Anti-Kickback Law is so
broad and the protection offered by its safe
harbors so limited, Congress created, and the
OIG has implemented, an “advisory opinion”
program. Pursuant to this program, individuals
and organizations may submit proposed (but not
hypothetical) arrangements to the OIG and request,
in effect, a “case-specific” safe harbor. As of
September 2010, the OIG had issued more than 200
advisory opinions. In the majority of these opinions,
the requestor’s proposed arrangement arguably
implicated the Anti-Kickback Law but could not be
safe harbored. More often than not, however, the
OIG concluded that the arrangement (1) did not
implicate the statute’s principal policy objectives, (2)
did not pose a material risk of program abuse, and
(3) as such, would not be subject to sanctions.
As the above description suggests, the AntiKickback Law is similar in many respects to the
Stark Law, both in terms of its overarching policy
objectives and general prohibitions. By the same
token, there are material differences between the
two laws, including the following:
• The Anti-Kickback Law is a criminal statute,
whereas the Stark Law provides for civil and
administrative sanctions.
• The Anti-Kickback Law has a “state of mind”
(or scienter) requirement (i.e., in order to be
convicted, a defendant must have acted
“knowingly and willfully”). The Stark Law is
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 146
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
•
•
a “strict liability” statute (i.e., the Stark Law’s
referral and billing prohibitions may be violated
even if the physician, provider, or supplier did
not intend to violate them).
The Anti-Kickback Law covers all federal
health care programs (with the exception of the
Federal Employee Health Benefits Program);
the Stark Law’s referral and billing prohibitions
apply only to Medicare.
The Anti-Kickback Law may be implicated by
any type of arrangement involving any type of
health care or non-health care organization;
the Stark Law focuses on physicians and their
financial relationships with certain types of
entities (e.g., hospitals) that furnish certain types
of services (i.e., DHS).
Potential Application to ACO Arrangements
Determining whether an arrangement violates
the Anti-Kickback Law is essentially a four-step
process.
1. Remuneration. Does the proposed
arrangement provide for “remuneration” of any
kind to flow from an individual or entity in a
position to benefit from the referral of federal
health care program patients or business (e.g., a
hospital) to an individual or entity in a position to
make such referrals (e.g., a physician)?
2. Implication. If so, will (or may) the
remuneration implicate the Anti-Kickback Law?
That is, is the remuneration intended to induce
the recipient to engage in conduct (e.g., patient
referrals) that is prohibited by the Anti-Kickback
Law?
3. Safe Harbors. If the remuneration does or
may implicate the Law, can the remuneration
be protected by an Anti-Kickback Law safe
harbor?
4. Risk Analysis. If not, will the proposed
arrangement pose a material risk of program
abuse?
As was the case with the Stark Law, the AntiKickback Law is sufficiently complex, and the types
of ACO arrangements are sufficiently varied, such
that there is no “one-size-fits-all” Anti-Kickback
Law analysis for such arrangements. Once again,
however, several observations relating to each of
the above questions, and their potential application
in the ACO context, can be made.
1. Remuneration
As a threshold matter, under most (and perhaps
all) ACO arrangements, “remuneration” — in the
form of shared savings, incentive payments, and
the like — will flow from payers and/or providers
(among others) to providers and/or physicians
(among others). Thus, there is little question that
most ACO arrangements will involve the payment of
“remuneration.”
2. Inducement
The next — and probably most critical — question
is this: will any of the payments be intended to
induce, or in exchange for, the past or future referral
of federal health care program patients or business?
If the answer is “no,” then the Anti-Kickback Law
will not be implicated, the arrangement at issue will
not need to be safe harbored, and the overall risk
of the arrangement under the Law will be minimal
(to non-existent). If the answer is “yes” — that
is, if even one purpose of the arrangement is to
induce the referral of federal health care program
patients or business to a payer, provider or any
other individual or entity participating in the ACO —
then the arrangement is unlikely to qualify for safe
harbor protection and likely to pose a material risk
of program abuse under the Anti-Kickback Law.
Put somewhat differently, those who are developing
and implementing ACOs need to take steps — while
the ACO is being designed, while the underlying
agreements are being negotiated, and while the
ACO is in operation — to ensure (1) whether a
physician (or any other individual or provider) is
entitled to shared savings or other payments under
the ACO, and (2) that the amount of such payments
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 147
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
is not contingent upon the volume or value of the
recipient’s referral of federal health care program
patients or business to a participating plan or
provider. Rather, such payments should be based
exclusively on specific and objective quality of care,
costs savings, and related performance measures.
It should be emphasized that the potential AntiKickback Law risk posed by ACO arrangements
is not theoretical. In its advisory opinions on
hospital-physician “gainsharing” arrangements
(discussed further below), the OIG has repeatedly
observed that “[l]ike any compensation arrangement
between a hospital and a physician who admits or
refers patients to such hospital, we are concerned
that [gainsharing arrangements] could be used to
disguise remuneration from [hospitals] to reward
or induce referrals by [physicians].”60 These
concerns will apply in equal, if not greater, measure
to potentially more sweeping and complex ACO
arrangements that involve even greater amounts
of remuneration and, potentially, a larger and more
complex set of payers, recipients and payment
arrangements.
Finally, it should be noted that many of the
conditions set forth in the Stark Law’s proposed
exception for “Incentive Payment and Shared
Savings Programs” (discussed above) would,
if implemented, help to ensure that the ACO
arrangement at issue does not implicate the AntiKickback Law. Most notably, perhaps, pursuant to
the proposed exception:
• Participating physicians cannot be “selected in
a manner that takes into account the volume or
value of referrals or other business generated
between the parties” (although a hospital
may elect to make the program available
to physicians in a “particular department or
specialty, provided that the hospital offers the
opportunity to participate in the . . . program
to all physicians in the department or specialty
on the same terms and conditions”).
• The remuneration to be paid under the
•
arrangement cannot be “determined in a
manner that takes into account the volume or
value of referrals or other business generated
between the parties.”
The remuneration paid to participating
physicians cannot “include any amount that
takes into account the provision of a greater
volume of federal health care patient procedures
or services than the volume provided by
the participating physician or qualified
physician organization during the period of
the same length immediately preceding the
commencement of the program as that covered
by the payment.”
3. Safe Harbors
Where an arrangement involves an exchange
of remuneration between providers and referral
sources, “best practices” are to safe harbor the
arrangement if at all possible, even if there are
no indicia of a quid pro quo between the parties.
Simply put, if an arrangement fits squarely into an
Anti-Kickback Law safe harbor, arguments about
whether a particular fact is or is not evidence of an
intent to induce referrals become moot.
As is the case with the Stark Law, there are several
Anti-Kickback Law exceptions and safe harbors
that are commonly employed where a hospital
and physician have a compensation arrangement,
including those covering certain employment61 and
personal services62 arrangements. As under the
Stark Law, however, none of these was designed
with shared savings or similar programs in mind
and, as a result, it is rarely the case that one
of these programs fits neatly into one of these
exceptions or safe harbors. Further, unlike CMS,
the OIG has not promulgated any proposed safe
harbor for ACO (or ACO-like) arrangements.
As under the Stark Law, however, there are certain
managed care exceptions under the Anti-Kickback
Law that may protect ACO arrangements that are
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 148
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
considered “downstream” of an insurer. The socalled “shared risk”63 and “health plan”64 exceptions,
for example, would fall into this category. Further,
as discussed above, ACA specifically authorizes
the U.S. Department of Health & Human Services to
“waive” the application of the Anti-Kickback Law to
approved ACO arrangements. Thus, to the extent
such a §3022 waiver is obtained by the ACO at
issue, referrals of participating beneficiaries should
not violate the Anti-Kickback Law.
4. Risk Analysis
Because most ACO arrangements (1) will involve
an exchange of remuneration, (2) this exchange
will be between parties who are in a position to
refer federal health care program business to one
another, and (3) this remuneration may not be
capable of safe harbor protection, whether and
the extent to which a particular ACO arrangement
poses a material risk of program abuse under the
Anti-Kickback Law will require a careful “facts and
circumstances” risk analysis.
Second, the OIG has issued a number of advisory
opinions over the past several years addressing a
variation on the ACO theme: hospital-physician
“gainsharing” arrangements. These opinions
address the types of concerns (and associated
safeguards) that are likely to apply in the broader
ACO context as well. For example, in 2008, the OIG
opined on an arrangement between a hospital and
five physician groups (four cardiology groups and a
radiology group). In a nutshell, the arrangement at
issue in OIG Advisory Opinion 08-21 was as follows:
• The hospital agreed to pay each group “a share
•
The most important risk factor effectively relates
back to the inducement question: that is, if there
are any indicia that a purpose of the shared
savings or other payments provided under the
ACO are intended to induce future (or reward past)
referrals of federal health care program patients or
business, then the arrangement will pose potentially
substantial risk under the Anti-Kickback Law.
Assuming no such quid pro quo exists, other risk
factors — and associated safeguards — can be
gleaned from two sources. First, because the
Anti-Kickback and Stark Laws share the same
fundamental objective — preventing the increase
in federal health care program costs that can
result from the overutilization of covered items and
services — ACOs can reduce their overall risk under
the Anti-Kickback Law by complying with CMS’
proposed Stark Law Shared Savings Exception
(detailed above) or any finalized version of this
proposed Exception.
•
of cost savings directly attributable to specific
changes” in that particular group’s cardiac
catheterization procedures.
Underlying the arrangement was a study of the
historical practices of the groups with respect to
such procedures performed at the hospital. The
study identified a number of specific cost saving
opportunities and made 27 recommendations
falling into three categories:
• Product standardization (e.g.,
standardizing the types of cardiac
catheterization devices and supplies
employed by the groups, based on both
clinical and cost considerations);
• “Use as needed” devices (e.g.,
limiting the use of certain vascular
closure devices and cutting balloons
to an “as needed” basis for coronary
interventional and diagnostic
procedures); and
• Product substitutions (e.g., substituting,
as appropriate, less costly contrast
agents and anti-thrombotic medications
for other products being used by the
physicians).
The arrangement included several safeguards
against “inappropriate reductions in services.”
For example, physicians were required to make
a “patient-by-patient determination of the most
appropriate device or supply” and to ensure that
the “availability of the full range of devices and
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 149
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
supplies” was not compromised by the product
standardization, use as needed, or product
substitution policies. In addition, “[t]o minimize
the physicians’ financial incentive to steer more
costly patients to other hospitals,” a committee
monitored the “case severity, ages, and payers
of the patient population treated under the
[a]rrangement.”
In analyzing the arrangement, the OIG noted
that, on the one hand, “[p]roperly structured,
arrangements that share cost savings can serve
legitimate business and medical purposes” (i.e.,
“properly structured arrangements may increase
efficiency and reduce waste, thereby potentially
increasing a hospital’s profitability”). On the other
hand, the agency posited, “such arrangements
can potentially influence physician judgment to the
detriment of patient care.”
Our concerns include, but are not limited to,
the following: (i) stinting on patient care; (ii)
“cherry picking” healthy patients and steering
sicker (and more costly) patients to hospitals
that do not offer such arrangements; (iii)
payments in exchange for patient referrals;
and (iv) unfair competition (a “race to the
bottom”) among hospitals offering cost
savings programs to foster physician loyalty
and to attract more referrals.
Turning to the Anti-Kickback Law analysis
specifically, the OIG noted that the arrangement
“could encourage the physicians to admit federal
health care program patients to the [h]ospital, since
the physicians receive not only their Medicare Part
B professional fee, but also, indirectly, a share of the
Hospital’s payment, depending on cost savings.”
“In other words,” the agency stated, “the more
procedures a physician performs at the [h]ospital,
the more money he or she is likely to receive under
the [a]rrangement.” In addition, the OIG stated,
“[m]ultiple-year gainsharing arrangements raise a
particular concern, in that they can inappropriately
carry over earlier-accomplished savings across
years, effectively accounting for them more than
once.” The “resulting unearned duplicate payments
can amount to unlawful kickbacks from hospitals to
physicians, if accompanied by illicit intent.”
Notwithstanding these concerns, however, the OIG
concluded that the arrangement “poses a low risk
of fraud or abuse under the [Anti-Kickback Law]”
and, as a result, that the agency would not “impose
sanctions in the particular circumstances presented
here.” The OIG’s reasoning was as follows.
First, the agency noted, the “circumstances and
safeguards of the [a]rrangement reduced the
likelihood that the [a]rrangement has been used to
attract referring physicians or to increase referrals
from existing physicians.” Specifically:
• “[P]articipation in the [a]rrangement was limited
•
•
•
•
to physicians already on the medical staff, thus
limiting the likelihood that the [a]rrangement
would attract other physicians.”
The “potential savings derived from procedures
for federal health care program beneficiaries
were capped based on the physicians’ prior
year’s admissions of federal health care
program beneficiaries.”
The “period for which payments have been
calculated was limited to one year (and the
[a]rrangement was rebased at the end of the
first year).”
The “overall amount of available cost savings
payments over the entire two year term of
the contract has been capped, reducing any
incentive to switch facilities.”
“[A]dmissions were monitored for changes in
severity, age, or payer.”
In sum, “while the incentive to refer was not
necessarily eliminated, it has been substantially
reduced.”
Second, the structure of the [a]rrangement
“eliminated the risk that the [a]rrangement has been
used to reward surgeons or other physicians who
refer patients to the [g]roups or their physicians.”
That is, the groups:
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 150
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
were the sole participants in the [a]rrangement
and were composed entirely of cardiologists
and interventional radiologists; no surgeons or
other physicians are members of the
[g]roups or will share in their profit
distributions. Within the [g]roups, profits are
distributed to members on a per capita basis,
mitigating any incentive for an individual
physician to generate disproportionate cost
savings.
Third, the OIG noted, the “product standardization,
limitation on use of devices and supplies,
and product substitution each carried some
increased liability risk for the physicians” and it
“is not unreasonable for the physicians to receive
compensation for the increased risk from the
change in practice.”
Moreover, the payments to be made represent
portions of two years’ worth of cost savings
and are limited in amount (i.e., the rebasing
and aggregate cap), duration (i.e., the
limited contract term), and scope (i.e., the
total savings that can be achieved from the
implementation of any one recommendation
are limited by appropriate utilization levels).
In sum, the “payments under the [a]rrangement
do not appear unreasonable, given, among other
things, the nature of the actions required of the
physicians to have implemented the twentyseven recommended actions, the specificity of the
payment formula, and the cap on total remuneration
to the [g]roups.”
Conclusion
ACO arrangements are likely to involve the payment
of remuneration between and among individuals
and entities in a position to refer federal health care
program business to one another. Further, many
of these arrangements may not fit squarely within
any Anti-Kickback Law exception or safe harbor.
These arrangements will not implicate (or violate)
the Anti-Kickback Law, however, provided that they
are not intended — in whole or in part — to induce
the referral of federal health care program business.
In addition to ensuring the absence of any such
quid pro quo, the implementation of a variety of
safeguards — which have been identified by the
OIG in similar contexts — can help to minimize any
risks posed by ACO arrangements under the AntiKickback Law.
6.4: FEDERAL SERVICES REDUCTION CIVIL MONETARY
PENALTY LAW
In addition to laws — such as the Stark and AntiKickback Laws — aimed at addressing the potential
overutilization of covered items and services, certain
federal health care laws are intended to address the
opposite issue: the underutilization of such items
and services. For example, the federal civil money
penalty (CMP) laws provide that if a “hospital . . .
knowingly makes a payment, directly or indirectly,
to a physician as an inducement to reduce or limit
services provided with respect to individuals” who
are (1) entitled to Medicare or Medicaid benefits
and (2) “under the direct care of the physician,”
then the hospital and physician are subject to a
CMP of $2,000 for each individual with respect to
whom the payment is made.65 For purposes of
this publication, will refer to this as the “Services
Reduction CMP.”
Potential Application to ACO Arrangements
Payments By Hospitals to Physicians
As a threshold matter, the Services Reduction
CMP only applies to payments from hospitals
to physicians. Thus, to the extent that shared
savings or similar payments under an ACO are not
being provided by a hospital (e.g., they are being
furnished by a payer) or are not being paid to a
physician (e.g., they are being furnished to any other
provider or supplier), the Services Reduction CMP
will not be implicated by the arrangement.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 151
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
Reduce or Limit Services
If the payments are made by a “hospital” to a
“physician,” however, the CMP will be implicated
if they serve as an inducement to “reduce or
limit” the services provided to Medicare or
Medicaid beneficiaries “under the direct care
of the physician.” The OIG has interpreted this
element of the Services Reduction CMP quite
broadly. For example, in 1999, the OIG issued a
Special Advisory Bulletin entitled, “Gainsharing
Arrangements and CMPs for Hospital Payments
to Physicians to Reduce or Limit Services to
Beneficiaries.” As its title suggests, the Bulletin
interprets the Services Reduction CMP in the
context of so-called “gainsharing” arrangements.
According to the Bulletin, and as discussed above,
these arrangements typically involve a hospital
giving “physicians a percentage share of any
reduction in the hospital’s costs for patient care
attributable in part to the physicians’ efforts.” These
arrangements, the OIG continued:
seek to align physician incentives with those of
hospitals by offering physicians a share of the
hospital’s variable cost savings attributable to
Medicare and Medicaid reimbursement. Since
the institution of the Medicare Part A DRG
system of hospital reimbursement and with
the growth of managed care, hospitals have
experienced significant financial pressures to
reduce costs. However, because physicians
are paid separately under Medicare Part
B and Medicaid, physicians do not have
the same incentive to save hospital costs.
Gainsharing arrangements are designed to
bridge this gap by offering physicians a portion
of the hospital’s cost savings in exchange
for identifying and implementing cost saving
strategies.
The OIG recognized that “hospitals have a
legitimate interest in enlisting physicians in their
efforts to eliminate unnecessary costs.”
Savings that do not affect the quality of
patient care may be generated in many ways,
including substituting lower cost but equally
effective medical supplies, items or devices;
re-engineering hospital surgical and medical
procedures; reducing utilization of medically
unnecessary ancillary services; and reducing
unnecessary lengths of stay. Achieving these
savings may require substantial effort on the
part of the participating physicians. Obviously,
a reduction in health care costs that does not
adversely affect the quality of the health care
provided to patients is in the best interest of
the nation’s health care system.
“Nonetheless,” the OIG concluded, “the plain
language of [the CMP] prohibits tying the
physicians’ compensation for such services to
reductions or limitations in items or services
provided to patients under the physicians’ clinical
care.” The OIG emphasized that the Services
Reduction CMP “is very broad.” For example, “[t]
he payment need not be tied to an actual diminution
in care, so long as the hospital knows that the
payment may influence the physician to reduce or
limit services to his or her patients.” Further, “[t]
here is no requirement that the prohibited payment
be tied to a specific patient or to a reduction in
medically necessary care.” In short, “any hospital
incentive plan that encourages physicians through
payments to reduce or limit clinical services directly
or indirectly violates the statute.”
Since issuing the Bulletin in 1999, the OIG has
addressed the Services Reduction CMP in
connection with a number of advisory opinions.
For example, in addition to considering the
Anti-Kickback Law in Advisory Opinion 08-21
(summarized above), the OIG also analyzed the
arrangement under the Services Reduction CMP.
As a threshold matter, the OIG concluded that “all of
the recommendations implicated the CMP.”
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 152
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
Simply put, with respect to the
recommendations under the [a]rrangement
regarding standardization of devices and
supplies, limiting use of specific vascular
closure devices and cutting balloons, and
substitution of contrast agent[s] and antithrombotic medication, the [a]rrangement
might induce physicians to reduce or limit the
then-current medical practice at the [h]ospital.
We recognize that the then-current medical
practice may have involved care that exceeded
the requirements of medical necessity.
However, whether the current medical practice
reflects necessity or prudence is irrelevant for
purposes of the CMP.
Risk Management
Unfortunately, the Services Reduction CMP
(unlike the Stark and Anti-Kickback Laws) does
not have any statutory exceptions or regulatory
safe harbors. (Note once again, however, that the
§3022 waiver authority under the ACA does extend
to the Services Reduction CMP.) Thus, where an
arrangement implicates the Services Reduction
CMP, affected providers have three choices: (1)
refrain from undertaking the arrangement at issue;
(2) refrain from undertaking the arrangement until
and unless a favorable advisory opinion is obtained
from the OIG; or (3) undertake the arrangement
subject to safeguards that are consistent with the
guidance issued by the OIG over the past 10 years,
thereby lowering — but not eliminating — the
overall risk of the arrangement under the Services
Reduction CMP.
Once again, many of the risk factors and safeguards
associated with ACO-like arrangements can be
gleaned from OIG advisory opinions. In Advisory
Opinion 08-21, for example, the OIG concluded that
notwithstanding the fact that the arrangement at
issue implicated the Services Reduction CMP, it also
incorporated “several features that, in combination,
provide sufficient safeguards so that we would not
seek sanctions against the [r]equestors.”
• First, the “specific cost saving actions and
•
•
•
•
resulting savings were clearly and separately
identified.” This “transparency” allows for
“public scrutiny and individual physician
accountability for any adverse effects” of the
arrangement, “including any difference in
treatment among patients based on non-clinical
indicators.” The “transparency of the incentives
for specific actions and specific procedures also
facilitates accountability through the medicallegal professional liability system.”
Second, there is “credible medical support
for the position that implementation of the
recommendations did not adversely affect
patient care.”
Third, the amounts to be paid under the
arrangement “have been calculated based on
all procedures performed, regardless of the
patients’ insurance coverage, subject to the
cap on payment for federal health care program
procedures.” In addition, the “procedures
to which the [a]rrangement applied were not
disproportionately performed on federal health
care program beneficiaries.” Further, “the cost
savings have been calculated on the [h]ospital’s
actual out-of-pocket acquisition costs, not an
accounting convention.”
Fourth, the arrangement “protected against
inappropriate reductions in services by utilizing
objective historical and clinical measures to
establish baseline thresholds beyond which
no savings accrued to the [g]roups.” The
requestors “have certified that these baseline
measures were reasonably related to the [h]
ospital’s or comparable hospitals’ practices
and patient populations.” In addition, these
safeguards were “action-specific and not
simply based on isolated patient outcome data
unrelated to the specific changes in cardiac
catheterization practices.”
Fifth, the “product standardization portion of
the [a]rrangement further protected against
inappropriate reductions in services by ensuring
that individual physicians still had available the
same selection of devices and supplies after
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 153
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
•
•
•
implementation of the [a]rrangement as before.”
The arrangement “was designed to produce
savings through inherent clinical and fiscal
value and not from restricting the availability of
devices and supplies.”
Sixth, the hospital and groups “provided
written disclosures of their involvement in the
[a]rrangement to patients whose care might
have been affected by the [a]rrangement”
and “provided patients an opportunity to
review the cost savings recommendations
prior to admission to the [h]ospital (or, where
pre-admission consent was impracticable,
prior to consenting to the procedure).” While
“we do not believe that, standing alone,
such disclosures offer sufficient protection
from program or patient abuse, effective and
meaningful disclosure offers some protection
against possible abuses of patient trust.”
Seventh, the “financial incentives under the
[a]rrangement were reasonably limited in
duration and amount.”
Eighth, because “each of the [g]roups
distributes profits to its members on a per
capita basis, any incentive for an individual
physician to generate disproportionate cost
savings was mitigated.”
In sum, the OIG noted, the arrangement is
“markedly different from ‘gainsharing’ plans
that purport to pay physicians a percentage of
generalized cost savings not tied to specific,
identifiable cost-lowering activities.” Rather, the
arrangement “set out the specific actions to be
taken and tied the remuneration to the actual,
verifiable cost savings attributable to those actions.”
This transparency, in turn, “allowed an assessment
of the likely effect of the [a]rrangement on quality
of care and ensures that the identified actions are
the cause of any savings.” In short, “[g]iven the
limited duration and scope of the [a]rrangement,
the safeguards provided sufficient protections
against patient and program abuse.” “Other
arrangements,” however, “including those that are
longer in duration or more expansive in scope than
the [a]rrangement, are likely to require additional or
different safeguards.”
6.5: FEDERAL TAX LAW
The development and implementation of ACOs raise
two primary tax exemption questions. First, if a new
“umbrella” entity is formed (as opposed to using a
series of agreements among existing entities), can
the entity qualify for tax-exempt status? Second,
will any shared-savings or other payments between
or among ACO participants be consistent with the
tax-exempt status of any tax-exempt participants in
the ACO?
Creation of New Entity
To determine whether a new entity can qualify
for tax-exempt status, the relevant analysis
is that applied by the IRS when considering
exemption for integrated delivery systems (IDS),
such as physician-hospital organizations (PHOs)
and preferred provider organizations (PPOs). A
typical PHO is formed as a nonprofit membership
organization controlled equally by a tax-exempt
charitable hospital and a medical group, an
independent practice association, or individual
physicians who practice at, or are affiliated with,
the hospital. The PHO provides no health care
services itself. Rather, it contracts with payors,
on behalf of the hospital and physicians, for the
provision of health care services in the community.
Because the PHO’s activities substantially serve the
private interests of its member physicians, the PHO
cannot avoid the proscription against more than
incidental private benefit. Thus, tax exemption as a
charitable organization is not available to the PHO.
The IRS has taken a similarly dim view with respect
to PPOs, which, in the IRS’ view, are typically
organized by the physician and hospital members
primarily to attract additional patients and revenues
to the participating providers and to increase the
providers’ market share.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 154
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
It seems likely that the IRS would apply a similar
analysis to ACOs and ACO entities seeking
tax-exempt status. Thus, in order to qualify for
exemption, the new entity would need to find a
basis for exemption that is different from those
rejected by the IRS for PHOs and PPOs. The
malleability of the promotion of health rationale
makes it a useful basis for exemption, particularly
with ACOs’ focus on improving quality of care.
Another possibility would be the theory of
“lessening the burdens of government.” This
was the rationale seized on by the IRS as a result
of statutory language in the American Recovery
and Reinvestment Act of 2009 (ARRA) to justify
tax-exempt status for regional health information
organizations. We note, however, that recently
the IRS appears to be limiting the “lessening the
burdens of government” rationale to exemptions for
Regional Health Information organizations (RHIOs).
It should also be noted that the primary goals of the
ACO — i.e., to improve quality and lower cost —
are favorable factors for exemption. The IRS has
expressly identified these as favorable reasons for
entering into joint ventures by exempt organizations.
The ACO network is a joint venture since it involves
multiple participants sharing risk and reward.
Another possibility would be to create an ACO
coordinating entity as a tax-exempt 501(c)(4) social
welfare organization. This is a type of tax exempt
organization that enjoys many of the same benefits
of charitable status but has additional flexibility
with regard to the IRS’ community benefit standard
and that has much greater flexibility with respect to
lobbying and political campaign activity, should that
become necessary.
Payment of Incentives
With respect to the second issue — whether
any shared-savings or other payments between
or among ACO participants will be consistent
with the tax-exempt status of any tax-exempt
participants in the ACO — the IRS is likely to
analyze such payments in the same manner that
it analyzes gainsharing and pay for performance
(P4P) programs. With respect to gainsharing
arrangements, for example, the IRS has ruled
favorably where (1) the physician groups provided
valuable services needed by the hospital, (2) the
arrangements resulted in cost savings to the
hospital, and (3) the allocation of the awards was
capped to reflect fair market value, as determined
by an independent third-party appraiser. Due
to the “facts and circumstances” nature of the
payments contemplated by ACOs, however, it may
be advisable for tax exempt organizations to obtain
a private letter ruling from the IRS confirming that
participation in the ACO will not have an adverse
affect on their status.
Conclusion
Although it may be difficult for an ACO to obtain
tax exempt status, to the extent the ACO can be
distinguished from many typical PHO and PPO
models — with a focus on the ACO’s primary
purpose to lower costs and improve the health of
the community — obtaining tax exempt status may
be possible. Further, provided certain fair market
value and other safeguards are implemented,
hospitals and other organizations that participate in
ACOs should be able to preserve their tax exempt
status.
6.6: OTHER POTENTIAL LEGAL ISSUES
In addition to those discussed above, ACO
arrangements — depending on their precise
makeup and operation — may raise additional legal
issues. For example:
• State Self-Referral, Anti-Kickback and
Similar Fraud and Abuse Laws. Most states
have one or more physician self-referral, antikickback and/or related fraud and abuse laws.
In some cases, these state laws largely mirror
their federal counterparts. In other cases,
however, these laws can be narrower or broader
than the Stark and Anti-Kickback Laws. In all
events, the arrangements — and, in particular,
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 155
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
•
•
•
the payment arrangements — underlying an
ACO should be analyzed under both federal and
state fraud and abuse laws.
Civil False Claims Acts. Under the federal
civil False Claims Act (FCA), a person who
“knowingly” “presents” or “causes to be
presented” a “false” or “fraudulent” “claim for
payment” to the U.S. government is liable for
a civil penalty of up to $11,000 per claim, plus
three times the amount of damages sustained
by the government.66 FCA actions may be
brought by private “whistleblowers” who are
entitled to up to 30 percent of any recovery.
Increasingly, FCA actions are being brought
on the ground that the claim at issue is “false”
because it is for services that were furnished
pursuant to a patient referral that violated the
Stark Law or Anti-Kickback Law. This potential
collateral risk needs to be considered when
analyzing any proposed ACO arrangement. In
addition, it should be noted that many states
have their own false claims (or similar) statutes.
Private Inurement/Private Benefit. In general,
a 501(c)(3) organization cannot be organized
or operated for the benefit of private interests
and the net earnings of a 501(c)(3) organization
may not inure to the benefit of any private
shareholder or individual. If one or more 501(c)
(3) organizations is participating in an ACO,
then — depending on the nature and formula
for determining any shared savings or similar
payments — these private inurement and/
or private benefit rules could be implicated.
For example, to the extent that an ACO
arrangement calls for a shared savings payment
from a 501(c)(3) hospital to a physician group,
and the payment exceeds fair market value,
the arrangement could implicate the private
inurement and/or private benefit rules.
Government Managed Care Programs.
Congress and CMS have created a complex
web of statutes and regulations that govern
the Medicare Part C – also known as Medicare
Advantage (MA) – and similar programs. To
the extent that patients participating in ACOs
•
•
include enrollees in such programs, these
statutes and regulations could be implicated.
For example, CMS’ MA regulations provide
for the imposition of sanctions on an MA
organization that “[f]ails substantially to provide,
to an MA enrollee, medically necessary services
that the organization is required to provide .
. . and that failure adversely affects (or is
substantially likely to adversely affect) the
enrollee.”67
Corporate Practice of Medicine. Many
states have what are commonly referred to as
“corporate practice of medicine” (CPOM) laws.
In general, these laws prohibit the practice
of medicine or the employment of physicians
by business corporations. Again, depending
on (1) an ACO’s corporate form (if any), (2) its
underlying arrangements with physicians, and
(3) its location, the ACO could implicate one or
more COPM laws.
State Insurance Laws. Bundled or capitated
payments are a likely component of the ACO
model. Some states require that health care
providers assume financial risk in the provision
of health care services to consumers, and
that employer groups be regulated as health
insurers. This can entail risk-based capital
reserve requirements and other state law
obligations, making it difficult for provider
organizations (and ACO components) to enter
into risk-sharing agreements.
CONCLUSION
The intent of this chapter is to identify areas of
potential legal concern for an ACO, as well as how
organizational and other choices may ameliorate
concerns. Legal analysts are quick to stress how
little is known about what existing laws might be
altered in light of ACA or where exceptions may
be granted by their enforcers. Not only does ACA
explicitly give rulemakers authority to grant a
number of exceptions and waivers to existing legal
restrictions, but we further know from previous
cases (e.g., Massachusetts) that other regulatory
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 156
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
reforms are likely to follow in course from federal
and state regulators in response to market demand.
Furthermore, integration efforts across provider
groups is not a recent phenomenon due to ACOs.
Rather this has been a strategy used by health care
organizations for many years in an attempt to better
manage care delivery.
There is a burden on the Department of Health
and Human Services, FTC, DOJ, state regulators
and other rulemakers and enforcers to design
and interpret regulations to ensure ACOs function
effectively and are able to “ramp up” quickly. This
will require an alignment of principles across these
state and federal stakeholders. These principles
must be conveyed to health care organizations
as they strategize on the best ways to manage
health care costs and quality. Along those lines,
there will be a great deal of burden on ACOs to
be clear and comprehensive when laying out their
consolidation plans to justify how they will result in
clinical improvement. Furthermore, they will need
to be able to supply meaningful evidence (i.e., data)
on the impact of these consolidation and clinical
transformation efforts.
That said, while legal ambiguity may prompt
a “wait and see” approach to aspects of ACO
implementation among some health systems,
most of these legal barriers can be avoided with
proper legal counsel. For this reason, it is again
worth stressing that any organization considering
participation in an ACO should consider legal
counsel a necessity. Being sure to build design
elements in the context of both federal and state
limitations will prevent costly revisions down the
road and, ultimately, build sustainable business
models to improve value across the system.
ENDNOTES
1. Good starting points are the FTC/DOJ Statements of Antitrust Policy in Health Care (the “Guidelines”).
2. Whether a particular situation is likely to raise antitrust issues is a fact-based inquiry that requires the analysis of many
variables. So, while we set forth general principles, whether a particular situation raises concerns necessarily requires a
review of the facts associated with the particular conduct in question.
3. Section 1 of the Sherman Act, 15 U.S.C. § 1. The law applies to conduct involving two or more actors that are legally
capable of conspiring. A single entity is legally incapable of conspiring. See, e.g., Copperweld Corp. v. Independence
Tube Corp., 467 U.S. 755 (1984) (“Section 1”).
4. The FTC and DOJ do not define what they consider to be a substantial withhold. However, in various Business Review
Letters and Advisory Opinions, the government has approved withholds which are in the range of 15%- 20%. The
withhold is usually coupled with already discounted fees. See, e.g., Advisory Opinion to George Q. Evans, July 5, 1994;
Business Review Letter to Alan C. Nelson, M.D., July 23, 1999. A withhold of less than 15% might pass muster – it would
all depend on the facts and on whether the withhold provided sufficient incentive for providers to modify their behavior
and work more efficiently in an effort to receive the return of the withhold.
5. FTC /DOJ Statements of Antitrust Enforcement Policy in Health Care Statement 8, (“Health Care Statement 8)
6. There is no requirement that a clinically integrated network negotiate provider reimbursement agreements on behalf of
the member providers. See discussion below regarding ways to mitigate market power issues.
7. Id.
8. Id.
9. See, e.g., Advisory Opinion Letter to Christi Braun, April 13, 2009.
10. Statement 9, Multiprovider Networks.
11. See Statement 8, Physician Network Joint Ventures.
12. Id.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 157
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
13. See Brown Shoe Co. v. United States, 370 U.S. 294, 344 (1962); see also, Levine v. Central Florida Medical
Affiliates, Inc., 72 F.3d 1538, 1551 (11th Cir.) 1996).
14. 425 U.S. 1, 17 (1976)
15. 425 U.S. at 17
16. 116 FTC 1526, 1527 (1993)
17. See Staff Advisory Letter to BJC Health System (November 9, 1999), in which the Bureau of Competition concluded
that the resale of pharmaceuticals to a health system’s employees is a resale for the health system’s own use within the
meaning of the NPIA. See also FTC Advisory Opinion to Bruce J. Toppin (January 7, 1998), in which the FTC stated
that the NPIA protected a hospital’s resale of drugs that it purchased at discounted prices to indigent cancer patients
who used the hospital’s Cancer Center; FTC Advisory Opinion to Robert M. Langer (December 20, 2001), in which the
FTC stated that a hospital association’s dispensation to its retired employees with vested retirement benefits meets the
definition of “own use.” Previously, the Supreme Court in Abbott Laboratories, see discussion supra, clearly held that
dispensation by a hospital to its own employees meets the definition of “own use.” The FTC found persuasive the hospital
association’s argument that benefit plans are necessary to attract and retain quality employees “for sufficient time for
them to become eligible for retirement and pension benefits.” The retention of such employees “directly promotes the
hospitals’ intended operation in the care of its patients” and, thus, dispensation of products purchased under NPIA is for
the hospital’s “own use.” Id.
18. Abbott Labs, 425 U.S. at 14.
19. See, e.g., Jefferson Parish Hospital District No. 2 v. Hyde, 466 U.S. 2 (1984).
20. 42 United States Code (USC) §1395nn(a)(1)(A).
21. 42 USC §1395nn(a)(1)(B).
22. Stark II, Phase II Proposed Regulations (Preamble), 63 Federal Register (FR) 1659, 1662 (1998).
23. Stark II, Phase II Proposed Regulations (Preamble), 63 FR 1659, 1662 (1998).
24. Stark II, Phase II Proposed Regulations (Preamble), 63 FR 1659, 1662 (1998).
25. Stark II, Phase II Proposed Regulations (Preamble), 63 FR 1659, 1662 (1998).
26. 42 USC §1395nn(g)(2); 42 Code of Federal Regulations (CFR) §411.353(d).
27. 42 USC §1395nn(g)(3); 42 CFR §§1003.102(a)(5), 1003.102(b)(9), 1003.105.
28. 42 CFR §1395nn(g)(4); 42 CFR §1003.102(b)(10).
29. 42 USC §1395nn(a)(2)(B); 42 CFR §411.354(a)(1)(ii).
30. 42 USC §1395nn(h)(1)(A); 42 CFR §411.354(c).
31. 42 USC §1395nn(h)(1)(B); 42 CFR §411.351.
32. Stark II Proposed Regulations (Preamble), 63 FR 1659, 1699 (1998).
33. 42 CFR §411.351.
34. Stark II Proposed Regulations (Preamble), 63 FR 1659, 1705-1706 (1998).
35. Stark II, Phase I Regulations, 66 FR 856, 958-959 (2001), setting forth 42 CFR §411.354(c)(2).
36. Stark II, Phase II Regulations, 69 FR 16054, 16134 (2004), setting forth a revised 42 CFR §411.354(c)(2).
37. Stark II, Phase III Regulations, 72 FR 51012, 51087 (2007), setting forth a revised 42 CFR §411.354(c)(2).
38. 2008 Stark Regulations, 73 FR 48434 (2008), setting forth a revised 42 CFR §411.354(c)(2).
39. 42 USC §1320a-7b(b).
40. 31 USC §3730.
41. 31 USC §3730(d).
42. 73 Fed. Reg. 38502, 38549 (July 7, 2008). Under certain, narrow circumstances (e.g., where it owns and operates
the providers that are furnishing the DHS at issue), a payer may constitute a “DHS entity.” In such cases, a “financial
relationship” between the payer and physicians could create Stark Law issues. In the vast majority of cases, however,
payers are not considered DHS entities.
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 158
PART 6: LEGAL ISSUES FOR ACOS | ACO TOOLKIT
43. 42 C.F.R. § 411.357(c).
44. 42 C.F.R. § 411.357(d).
45. 42 C.F.R. § 411.357(l).
46. 42 C.F.R. § 411.357(p).
47. 73 Fed. Reg. 38502, 38548 (July 7, 2008).
48. 73 Fed. Reg. 38502, 38548 (July 7, 2008).
49. 73 Fed. Reg. 38502, 38548 (July 7, 2008).
50. 73 Fed. Reg. 38502, 38548 (July 7, 2008).
51. 42 C.F.R. §411.357(n).
52. 42 C.F.R. §411.355(c).
53. The Medicare Managed Care Plans include: (1) “[a]n HMO or a CMP in accordance with a contract with CMS
under [§]1876 of the Act and part 417, subparts J through M of this chapter”; (2) “[a] health care prepayment plan in
accordance with an agreement with CMS under [§]1833(a)(1)(A) of the Act and part 417, subpart U of this chapter”; (3)
“[a]n organization that is receiving payments on a prepaid basis for Medicare enrollees through a demonstration project
under [§]402(a) of the Social Security Amendments of 1967 (42 U.S.C. [§]1395b-1) or under [§]222(a) of the Social
Security Amendments of 1972 (42 U.S.C. [§]1395b-1 note)”; (4) “[a] qualified HMO (within the meaning of [§]1310(d) of the
Public Health Service Act)”; (5) “[a] coordinated care plan (within the meaning of [§]1851(a)(2)(A) of the Act) offered by an
organization in accordance with a contract with CMS under [§]1857 of the Act and part 422 of this chapter”; (6) “[a] MCO
contracting with a State under [§]1903(m) of the Act”; (7) “[a] prepaid inpatient health plan (PIHP) or prepaid ambulance
health plan (PAHP) contracting with a State under part 438 of this chapter”; (8) “[a] health insuring organization (HIO)
contracting with a State under part 438, subpart D of this chapter”; and (9) “[a]n entity operating under a demonstration
project under [§§]1115(a), 1915(a), 1915(b), or 1932(a) of the Act.”
54. 42 USC §1320a-7b(b)(2).
55. 42 USC §1320a-7b(b)(1). Where the anti-kickback statute has been violated, the government may proceed
criminally or civilly. If the government proceeds criminally, a violation of the law is a felony punishable by up to five years’
imprisonment and a fine of up to $25,000. Id. §§1320a-7b(b)(1)-(2). If the government proceeds civilly, it may impose
a civil monetary penalty of $50,000 per violation and an assessment of not more than three times the total amount
of “remuneration” involved, and it may exclude the defendant from participating in Federal health care programs. Id.
§§1320a-7a(a)(7), 1320a-7(b)(7).
56. 56 Federal Register (FR) 35952, 35958 (July 29, 1991). Unlike the Stark Law, the anti-kickback statute does not have
an exception for de minimis amounts and remuneration means anything of value, no matter how small.
57. Hanlester Network v. Shalala, 51 F.3d 1390, 1398 (9th Cir. 1995).
58. See, e.g., United States v. Greber, 760 F.2d 68 (3rd Cir. 1985), cert. denied, 474 U.S. 988 (1985); United States v. Davis, 132 F. 2d 1092 (5th circuit 1998); United States v. Katz, 871 F. 2d 105 (9th Cir. 1989).
59. Hanlester Network v. Shalala, 51 F.3d 1390, 1398 (9th Cir. 1995); United States v. McClatchey, 217 F.3d 823, 834.
60. See, e.g., Advisory Opinion 08-21 (November 25, 2008), at 14.
61. 42 C.F.R. § 1001.952(i).
62. 42 C.F.R. § 1001.952(d).
63. 42 C.F.R. § 1001.952(t).
64. 42 C.F.R. § 1001.952(m).
65. 42 U.S.C. §1320a-7a(b)(1).
66. Stark II, Phase II Regulations (Preamble), 69 FR 16054, 16957 (2004).
67. 42 C.F.R. § 422.752(a)(1).
ENGELBERG CENTER FOR HEALTH CARE REFORM | THE DARTMOUTH INSTITUTE 159
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement