Pay-for-Performance in Health Care

Pay-for-Performance in Health Care
Updated December 12, 2006
Jim Hahn
Health Economist
Domestic Social Policy Division

Pay-for-Performance in Health Care
Summary
In recent years, many health care industry leaders and policy makers have joined the
call to pay health care providers different amounts based on variation in the quality of
their services as determined through their achievement on quality performance measures.
Proponents of these pay-for-performance systems in health care assert that such programs
could help improve the quality of care while also helping to control the rate of growth in
health care costs. Pay-for-performance systems have been implemented for some managed
care plans that cover Medicaid beneficiaries and recently for Medicare physician
payments.
A pay-for-performance system is a remuneration arrangement in which a portion of
the payments is based on performance assessed against a defined measure. Typically,
there is another component of the remuneration that is independent of the amount at risk.
While most of the current discussions about pay-for-performance in the health care
industry address quality-based measures, performance objectives and metrics could target
any of a number of variables, including profitability, volume, or customer or patient
satisfaction.
The elements common to all pay-for-performance programs are (1) a set of targets
or objectives that define what will be evaluated, (2) measures and performance standards
for establishing the target criteria, and (3) rewards — typically financial incentives — that
are at risk, including the amount and the method for allocating the payments among those
who meet or exceed the reward threshold.
For a pay-for-performance program to be successful, there needs to be agreement and
buy-in among those being evaluated that the objectives are fair and the measures
appropriate, that performance is accurately measured, and that the incentives make the
effort worthwhile. Possible shortcomings and unintended consequences of a pay-for-
performance program include having inappropriate measures and objectives, competing
or uncoordinated efforts, insufficient or inappropriate incentives, and placing excessive
focus on the reward.
There are few rigorous objective evaluations of the effect of pay-for-performance
programs. Initial studies suggest that pay-for-performance programs can change
performance on quality measures that are used for the basis of bonus payments, but
assertions that pay-for-performance programs are cost-saving in the long run are largely
unsubstantiated.
Most of the pay-for-performance legislation that Congress has considered recently
has centered on the Medicare program, although Medicaid and private insurers have also
garnered some attention. A number of pay-for-performance demonstration projects have
been authorized, and Congress issued a directive to the Secretary of Health and Human
Services to develop a value-based purchasing program for hospital payment by 2009. A
provision in H.R. 6111, the Tax Relief and Health Care Act of 2006, creates a voluntary
pay-for-performance system that will reward physicians an additional 1.5% of their
Medicare payments for care provided between July 1 and December 31, 2007, if they met
reporting requirements on quality measures. This report will be updated as needed.

Contents
In troduction ....................................................... 1
Background ....................................................... 2
Recommendations and Calls for Action.............................3
Institute of Medicine........................................3
MedPAC ................................................. 3
Theory and Conceptual Foundation.....................................4
Objectives and Targets...........................................5
Measures and Performance Standards...............................6
Incentives and Bonus Payments....................................7
Competitive and non-competitive rewards.......................8
Performance Unit for Measurement and Reward......................10
Possible Shortcomings and Unintended Consequences.....................10
Inappropriate Measures and Objectives.............................11
Competing or Uncoordinated Efforts...............................11
Insufficient or Inappropriate Incentives.............................12
Excessive Focus on the Reward...................................13
Pay-for-Performance and Cost Control.................................13
Pay-for-Performance Initiatives.......................................15
Public Sector.................................................16
Medicare ................................................ 16
Medicaid ................................................ 21
Private Sector.................................................21
Bridges to Excellence......................................21
Leapfrog ................................................. 22
United Kingdom’s Quality and Outcomes Framework.................23
Physicians and Pay-for-Performance...............................23
Legislative Activity................................................25
S. 1932, the Deficit Reduction Act of 2005..........................25
H.R. 3617, the Medicare Value-Based Purchasing
for Physicians’ Services Act of 2005...........................27
H.R. 6111, the Tax Relief and Health Care Act of 2006................27
List of Figures
Figure 1. Growth in Pay-for-Performance Programs by Sponsor Type........15
List of Tables
Table 1. Pay-for-Performance Program Characteristics
and Effects on the Competitiveness of the Reward.....................8
Table 2. CMS Pay-for-Performance Initiatives...........................17

Pay-for-Performance in Health Care
^If^{a phys}ⁱ^ci^{an m}^a^{ke a l}^a^r^g^{e i}ⁿ^ci^sⁱ^{on w}ⁱ^t^h^{an oper}^a^tⁱ^{ng kni}^f^e^{and cur}^eⁱ^t^,^orⁱ^f^{he open}
^{a tumor}⁽^o^ver^{the eye)}^w^{ith an}^oper^a^ting^kn^{ife, and s}^a^ves^{the eye,}^he^s^hall^r^eceive^ten
^s^h^ekels^{in money.}
^{If a phys}ⁱ^{cian make a lar}^g^{e incis}ⁱ^on^w^ith^{the oper}^a^{ting knife, and kill}^him,^or^open
^{a tumor}^w^{ith the oper}^a^{ting knife, and cut out}^{the eye, his}^hands^s^{hall be cut off.}
^{— C}^{ode of}^H^a^m^m^u^r^abi^{, c. 1750 B.C}^.
Introduction
In recent years, many health care industry leaders and policy makers have joined
the call to pay health care providers different amounts based on variation in the quality
of their services as determined through their achievement on quality performance
measures. Proponents of these pay-for-performance systems in health care assert that
current payment systems are typically neutral with regard to differences in quality, and
that pay-for-performance programs could help improve the quality of care while they
also help to control the rate of growth in health care costs. Additionally, pay-for-
performance has been touted as a mechanism that can accelerate the rate of adoption
of health information technology and electronic medical records, and promote the
delivery of more preventive services. Pay-for-performance systems have been
implemented for some managed care plans that cover Medicaid beneficiaries, and
proposals for establishing pay-for-performance systems for Medicare services continue
to come before Congress.
This report examines the factors that need to be considered when evaluating the
advantages and disadvantages of a pay-for-performance system in health care. The
report first discusses the conceptual foundation for pay-for-performance systems and
then addresses their strengths, potential shortcomings, and unintended consequences.
Following a discussion of the potential for pay-for-performance systems to control the
growth in health care costs, the report examines the track record of pay-for-
performance programs in health care, keeping an emphasis on experiences in the
private sector and in public demonstration projects. The report then summarizes the
pay-for-performance proposals contained in bills that have been introduced before
Congress, and examines potential legislation likely to come before the Congress in the
coming year.

Background
A pay-for-performance system is a remuneration arrangement in which a portion
of the payments is based on performance assessed against a defined measure.
Typically, there is another component of the remuneration that is independent of the
amount at risk. The terms merit and bonus pay are also used to describe similar
systems.¹ While most of the current discussions about pay-for-performance in the
health care industry address quality-based measures, performance objectives and
metrics could target any of a number of variables, including profitability, volume, or
customer or patient satisfaction.
Pay-for-performance systems are not new. The Code of Hammurabi (c. 1750
B.C.), the set of statutes from ancient Mesopotamia that ranks among the oldest
collections of extant laws in history, clearly established a payment system that
rewarded practitioners differentially based on the quality of the outcome of their
services. More recently, pay-for-performance systems for executive compensation
have become more common in many industries.² Health insurance company
representatives have testified that pay-for-performance efforts have been implemented
in the private market since the late 1990s, and that pay-for-performance programs are³

growing in number and scope.
¹ Although often used interchangeably, pay-for-performance can also be considered one
dimension of a broader set of concepts collectively known as “value-based purchasing.”
Proponents of value-based purchasing emphasize the focus of each decision maker, such as
an insurer, employer, or patient, on assessing differences in perceived value — whether it
be in efficiency, quality, cost, or some other measure — when choosing among options. The
value assessment can be specific to each decision maker, and proponents of the concept
prefer the positive connotation of maximizing value, typically emphasizing quality when
making health care purchasing decisions. Value-based purchasing approaches emphasize
the collection and analysis of data on quality, the dissemination of quality information to
providers and beneficiaries, and the selective rewarding of identified high-quality achievers
through contracts, partnerships, or incentives. (See, for example, the testimony of Robert
Berenson, M.D., before the Subcommittee on Health of the House Committee on Ways and
Means, September 29, 2005.) Critics of pay-for-performance approaches object to the
explicit assessment of health care providers against a specified standard and to linking
provider reimbursement to these measures, even though many of the activities are similar
to those used for value-based purchasing. The version of S. 1932 originally passed by the
Senate in 2005 included provisions for establishing “value based purchasing” under
Medicare, with details that reflected a pay-for-performance structure.
² For example, some corporate officers may have part of their compensation tied to the value
of company stock or some other measure of performance.
³ See, for example, the testimony of Karen Ignagni, President and CEO, America’s Health
Insurance Plans, before the House Committee on Education and the Workforce
Subcommittee on Employer-Employee Relations hearing on “Examining
Pay-for-Performance Measures and Other Trends in Employer-Sponsored Health Care,”
May 17, 2005.

Recommendations and Calls for Action
Many organizations, both private and public, have recommended changes in how
health care providers are reimbursed, and have suggested moving to payment methods
that are at least partially based on quality differences.
Institute of Medicine. The Institute of Medicine (IOM) has issued several
reports recommending differential payments based on quality. In a seminal 1999
report, To Err is Human: Building a Safer Health Care System, the IOM called for
initiatives to reward quality. In the 2001 report Crossing the Quality Chasm: A New
Health System for the 21^st Century, the IOM called for an increase in payments to⁴
providers of high quality care. This was followed in 2006 by the report Performance
Measurement: Accelerating Improvement, mandated as part of the Medicare
Modernization Act (P.L. 108-173), which recommends the creation of a national
system for performance measurement and reporting to be financed annually through
funds appropriated from the Medicare Trust Funds. One recommendation urges
purchasers and insurers to redesign payments to encourage providers to make positive
changes to their care processes. In its most recent report, Rewarding Provider
Performance: Aligning Incentives in Medicare, the IOM came to two conclusions
regarding pay-for-performance as a payment strategy for Medicare: (1) using payment
incentives to reward quality and quality improvement “can serve as a powerful
stimulus to drive institutional and provider behavior toward better quality,” but (2)
pay-for-performance incentives alone would be insufficient without “certain operating
conditions ... such as the use of electronic health records, public reporting, beneficiary
incentives, and education of boards of directors.”⁵ These IOM reports have been
influential in bringing attention to the need for quality improvement in health care and
calling for variable payments based on quality.
MedPAC. For many years, Medicare Payment Advisory Commission (MedPAC)
has recommended paying providers different rates based on differences in quality. In
its March 2004 report, MedPAC recommended that Congress “establish a quality
incentive payment policy for all Medicare Advantage plans” and “establish a quality
incentive payment policy for physicians and facilities providing outpatient dialysis⁶
services.” MedPAC added hospitals, home health agencies, and physicians to its
recommendations in its March 2005 report.⁷ In the March 2006 report, MedPAC also
recommends gathering information about resource use and providing information
about practice patterns confidentially to physicians as “important steps to improving

⁴ Institute of Medicine, To Err is Human: Building a Safer Health Care System (1999);
Crossing the Quality Chasm: A New Health System for the 21^st Century (2001);
Performance Measurement: Accelerating Improvement (2006); and Rewarding Provider
Performance: Aligning Incentives in Medicare (2006). National Academy Press,
Washington, DC.
⁵ Institute of Medicine, Rewarding Provider Performance: Aligning Incentives in Medicare.
National Academy Press, Washington, DC, 2006.
⁶ MedPAC, Report to the Congress: New Approaches in Medicare, June 2004.
⁷ MedPAC, Report to the Congress: Medicare Payment Policy, March 2005.

quality for beneficiaries and laying the groundwork for obtaining better value in the
Medicare program.”⁸
MedPAC’s recommendations are most fully set forth in the 2005 report, which
recommends that Congress give the Medicare program the ability to pay differentially
based on performance. MedPAC also recommends that the Secretary of Health and
Human Services (HHS) design a program that rewards both relative and absolute
standards. MedPAC further recommends that the pay-for-performance system be
budget-neutral, with the incentive pool to be funded by setting aside 1% or 2% of
budgeted payments.⁹
Also of note is an open letter published in 2003 from health care industry leaders
who recommend that “payment for performance should become a top national priority
and that Medicare payments should lead in this effort.” The signatories represented
many research, academic, and insurance groups, and included several former Center
for Medicare and Medicaid Services (CMS) administrators. The group noted that it
was “not ... suggesting that such an initiative be dominated by government,” but that
a “major initiative by Medicare to pay for performance can be expected to stimulate
similar efforts by private payers.”¹⁰
Theory and Conceptual Foundation
Pay-for-performance systems can be structured in many ways. Incentives,
bonuses, and especially performance criteria can be based on any number of
dimensions. However, the elements common to all pay-for-performance programs are
(1) a set of targets or objectives that define what will be evaluated, (2) measures and
performance standards for establishing the target criteria, and (3) rewards — typically
financial incentives — that are at risk, including the amount and the method for
allocating the payments among those who meet or exceed the reward threshold.
The relationship between value and price has long engaged economists, who share
no agreement on the precise relationship. However, when considering an economic
commodity such as a good or service, the price, which is actual and observable,
typically increases systematically, but not proportionately, with the value, which is¹¹
latent and unobservable. Economic theory holds that individual decision makers
compare their implicit assessment of value against the explicit price to make optimal
purchasing decisions. In most markets, this comparison works reasonably well, with
knowledgeable buyers having sufficient information about competing products and
services to keep overall prices in line with aggregate valuations of the goods or
services.

⁸ MedPAC, Report to the Congress: Medicare Payment Policy, March 2006.
⁹ MedPAC, Report to the Congress: Medicare Payment Policy, March 2005, p. 187.
¹⁰ Berwick, DM et. al. “Paying for Performance: Medicare Should Lead,” Health Affairs
Volume 22, Number 6, November/December 2003. pp. 8-10.
¹¹ Desai, M., “Value and Price,” in The New Palgrave Dictionary of Economics, Eatwell,
J., et al., ed. The Stockton Press, New York, 1988.

In health care, this relationship between value and price has been tenuous or
nonexistent because buyers and payers are not typically the patients who receive the
care. The asymmetry of information on both the part of the providers (with the medical
expertise) and the buyers (with knowledge of personal or group health status) has
meant that neither the buyer nor seller has transparent access to information. In
addition, the presence of insurers as third-party payers has removed much of the
incentive for price consciousness.
Payments to health care providers historically have not been determined in a
classical free market through the interaction of supply and demand, and consequently,
most insurers have not made any distinctions in payments to providers who exhibit
differences in quality. Reimbursements have been tied to other dimensions of care —
for instance, efficiency or low cost. Pay-for-performance programs are an attempt to
bring this relationship between prices and value, as reflected in quality care, into a
closer balance. In theory, pay-for-performance programs in health care would be less
compelling if at-risk buyers were fully informed about quality differences.
Objectives and Targets
The objectives and targets specified in a pay-for-performance system are the
explicitly identified priorities and goals against which progress and achievement are
to be measured. While most current discussions about pay-for-performance address
quality improvement and quality-based measures, these are only one possible
dimension. The objectives could address any number of goals, including costs of care,
quantity of services delivered, or some combination including quality.
There are four types of performance measures typically used in measuring the
quality of health care activity: clinical outcome measures, process measures, structural
measures, and patient satisfaction measures.¹² Methods of defining and collecting data
to address these different objectives vary in complexity and difficulty. Clinical
outcomes are the preferred standard, but are rarely possible to collect in health care.
Accurately assessing the consequences of a particular treatment upon the eventual
outcome may require a long time. For instance, the ultimate outcome for patients with
chronic conditions such as diabetes or hypertension might be the number of additional
years of life, adjusted for quality, following treatment.¹³
Process measures often address the proper delivery of health care services and
practice patterns (e.g., the use of diagnostic screening for those who present with a set
of symptoms or the dispensing of medications such as beta-blockers to patients with

¹² Donabedian developed the quality measurement model targeting structure, process, and
outcome measures that has become widely accepted and used as the basis for much of the
work addressing quality and outcomes. For further details, see Donabedian, A., Explorations
in Quality Assessment and Monitoring: Vol 1. The Definition of Quality and Approaches to
Its Assessment. Ann Arbor, MI: Health Administration Press, 1980.
¹³ A quality adjusted life year (QALY) is a construct developed to adjust for differences in
the quality or value of a year of life because of morbidities. A year in perfect health would
have a value of 1.0 QALY, while the value of a year in ill health would be discounted to
some fraction less than one, depending on the condition.

heart attacks.) Process measures are easier to gather but are not necessarily consistent
with clinical outcomes; research on the evidence base for clinical practices is ongoing,
but much of the variation in care is likely to remain unexplained for the foreseeable
future.¹⁴ Quite often, process measures tend to focus on underuse of services, and may
thus be cost-increasing in the short run.
Structural measures that have been used in health care pay-for-performance
programs include, for example, addressing whether organizations have adopted health
information technology (HIT) that allows electronic medical records to be created and
sent. Some of the Medicare demonstration projects include incentives to promote the
use of HIT through pay-for-performance programs.¹⁵
Finally, patient satisfaction measures are sometimes used in pay-for-performance
programs, but can be controversial. Proponents point out that patient satisfaction
measures are relatively easy to collect, and some research suggests that patient
satisfaction is positively linked to patient compliance and better clinical outcomes.
These measures often reflect the value that patients place in the service or
“consumption” dimension of health care, and physician communication is the factor
most strongly associated with a high patient satisfaction score.¹⁶ However, critics
question the usefulness of patient satisfaction measures. For example, one study of
elderly patients found no relationship between patient satisfaction and the technical
quality of the care.¹⁷ Partly as a consequence of this difference in perspectives, patient
satisfaction measures are typically given less weight (usually less than 20%), when
they are used at all.
Measures and Performance Standards
Ideally, performance is evaluated using metrics that capture and assess the
behavior that has been determined to be causally related to the objectives. The
performance reward standards are the benchmarks or thresholds established to
determine which of the eligible bonus recipients actually qualify for extra payments.
Standards can be either absolute or relative, with different implications for what
behavior is being rewarded. Absolute thresholds, such as requiring that 95% of patients
brought to a hospital following an acute myocardial infarction receive aspirin within
an hour, or that 100% of pneumonia patients are counseled about smoking cessation,

¹⁴ For example, an index of 20 articles on variations in clinical practice patterns can be
found at [http://content.healthaffairs.org/webexclusives/index.dtl?year=2004].
¹⁵ For more discussion of health information technology and quality, see CRS Report
RL32858, Health Information Technology: Promoting Electronic Connectivity in
Healthcare, by C. Stephen Redhead.
¹⁶ See, for example, Sleath, B., et al., “Physician-Patient Communication About Over-the-
Counter Medications,” Social Science and Medicine, 53(3):357-9. August 2001, and
Markoul, G., et al., “Health Promotion in Primary Care: Physician-Patient Communication
and Decision Making About Prescription Medications,” Social Science and Medicine.

41(9):1241-54, November 1995.

¹⁷ Chang, et al., “Patients’ global ratings of their health care are not associated with the
technical quality of their care.” Annals of Internal Medicine. May 2, 2006. 144(9):665-72.

would reward actual levels of achievement. One advantage of having specific, targeted
objectives is that there is no uncertainty on the part of those being evaluated about
whether they have met the standard, whereas a relative ranking is not revealed until a
later time. However, absolute thresholds may offer less of an incentive for those
already above the criteria, who may not see any motivation to improve beyond their
current practice, or for those far below the criteria who may feel that any effort to
improve may be futile. Absolute standards may not allow for flexibility when trends
outside the control of health care providers shift the experience for a large portion of
the population, such as when insurers substantially change their coverage policies or
when epidemics strike.
Relative standards or rankings focus on the position of each provider when
compared with peers. Typically, relative rankings can be used to reward improvement,
such as when a provider moves from the third quartile into the second, but high
achievers have little opportunity to improve their already favorable relative portion. If
incentives are successful in altering provider behavior and practice patterns, the
distribution of providers along the quality measure is likely to compress. As the spread
contracts, differences in ranking and quantiles may represent little practical
significance — for instance, if the difference between being in the top decile or the
second becomes the difference between a score of 98.5% or 99% — and relative
rankings and quantiles may not be a useful distinction.
Many existing pay-for-performance systems in health care incorporate both types
of standards. Applying both types of standards offers the possibility of both rewarding
high achievers and stimulating improvement by lower-ranked providers. For example,
an assessment prior to the implementation of a pay-for-performance program may
produce a distribution across providers, from which a baseline can be established. This
baseline, perhaps the cutoff point that defines the bottom quantile, might then be made
the absolute criteria that all providers must meet over a defined period. Providers
would then be eligible for bonuses if they improved their relative ranking — e.g.,
moved from the bottom half of the distribution to the top half — or if they exceeded
a minimum or qualifying threshold.
Incentives and Bonus Payments
The size and source of the incentives and bonus payments may have the most
significant role in determining how much of a change in behavior eventually results.
Laudable objectives accurately measured and fairly evaluated may produce very little
or no change if the payoff is insignificant. On the other hand, disproportionately large
incentives can lead to substantial changes in behavior, even if the objective is
misguided or misaligned with the incentives.
In addition, whether the payments at risk are perceived as a potential gain or a
potential loss may affect the likelihood of behavior change. The incentive pool can be
funded either by infusing new funds or by redistributing existing funds. Adding new
money can allow for the status quo to be the worst-case scenario, but the consequence
in the short run will be a more expensive total compensation package. With no other
redistribution of existing payments, such additional funds would simply be a potential
gain, and there would be less resistance to such a change on the part of providers.
However, if the bonus payments were funded by withholding a fraction of existing

payments, then the zero-sum situation would create winners and losers, and there
would be greater potential for objections to the implementation of such a system from
those being evaluated.
Competitive and non-competitive rewards. Differences in the structure
of pay-for-performance programs can lead to rewards that are earned either
competitively or non-competitively, with different implications for budgets and for
those being evaluated. Two dimensions are critical: (1) whether the performance
criteria are absolute or relative, and (2) whether the bonus pool contains a fixed or an
open-ended, uncapped amount of funds. Varying these two dimensions creates four
possible scenarios. (See Table 1.)
Table 1. Pay-for-Performance Program Characteristics and
Effects on the Competitiveness of the Reward
^B^o^{nus P}^o^o^l
^Fⁱ^xed^O^pen-^ended
^C^e^rtain^t^y^abou^t^C^e^rtain^t^y^abou^t
^q^u^alif^icatioⁿ^q^u^alif^icatioⁿ
^Abso^lute
^Un^cer^tain^ty^ab^o^u^{t th}^e^C^e^rtain^t^y^abou^{t th}^{e (pre-}
^a^m^o^unt^d^e^ter^mⁱⁿ^ed⁾^am^o^uⁿ^t
^rformance^S^t^andards^Un^cer^tain^ty^ab^o^u^t^Un^cer^tain^ty^ab^o^u^t
^P^e^q^u^alif^icatioⁿ^q^u^alif^icatioⁿ
^Rela^tiv^e
^C^e^rtain^t^y^abou^{t th}^{e am}^ouⁿ^t^C^e^rtain^t^y^abou^{t th}^{e (pre-}
^d^e^ter^mⁱⁿ^ed⁾^am^o^uⁿ^t
^Source:^CRS.
Absolute performance standards combined with bonus pools that are not capped
lead to non-competitive bonus awards. Participants will be fully informed about
whether they have reached the bonus threshold as well as the amount of the bonus,
assuming that the reward is clearly identified in advance. Participants only need to
know how they performed against the qualifying threshold and do not need to know
how others performed to determine whether they qualify for the reward. If the amount
of the reward is previously determined, participants will also have knowledge about
how much they might gain, should they achieve the standard. This is the least
competitive type of pay-for-performance program, exemplified by the Quality and
Outcomes Framework in place in the United Kingdom (see below), and typically
requires additional monies to fund the bonus pool.
If the bonus pool is fixed but the performance standards are absolute, then the
award becomes competitive.The number of providers who meet the performance
standard and qualify for the extra payments could determine the size of the bonuses.
When the amount of money provided for incentives is fixed, the bonus amount for
each provider decreases as the number of providers who qualify increases. For

instance, if the bonuses are funded by a fractional withholding and everyone qualifies,
then the best a provider can hope for is to recover the full amount before the
contribution to the incentive pool. As the number of providers who qualify for the
incentive payments falls, the amount of bonus per provider increases because the pool
is divided among fewer recipients.
If the amount of the bonus pool is not capped but the qualifying standard is a
relative measure, then the rewards are also competitive. Again, assuming the reward
is clearly agreed to in advance, participants would know how much is at stake but
would not know whether they qualify to receive the bonus until their performance is
assessed against their peers. For instance, a pay-for-performance program may reward
only the top 10% but will give those who qualify a predetermined bonus. Participants
would then balance the potential gain against their likelihood of achieving top 10%
status. The Premier Hospital demonstration uses this reward structure (see below).
Finally, if the amount of the bonus pool is capped and the qualifying standard is
relative, the rewards are also competitive. Participants will not know whether they
qualified for the bonus until their relative standing is known but may know how much
the bonus will be if only a pre-determined percentage of peers will receive bonuses
from a fixed pool or, alternatively, if the reward is a pre-determined fixed amount.
The relationship between the number of qualifying recipients and the amount of
the bonus creates a tradeoff between a larger incentive in response to a bigger bonus
payment and the proportion of the pool who achieve the standard and are therefore
eligible for the extra payments. From a policy standpoint, if the objectives are truly
desirable, e.g., accurate measures that truly improve the quality of patient care, then
the more who meet a stringent standard the better. However, because the size of each
payment decreases as more recipients qualify, the incentive effect becomes more
diluted with wider success. If there is a distribution along the performance scale, then
one parameter available to balance this effect is to set the qualifying standard at a level
high enough so that not everyone meets the threshold but low enough so that it is
potentially achievable for enough of the population targeted to encourage widespread
effort. The pay-for-performance program proposed as part of S. 1932 in the 109^th
Congress most resembled this structure (see below). The bill would have required all
the payments withheld to create the bonus pool to be distributed as awards, but allowed
for an uncertain number of recipients.
One advantage of a pay-for-performance framework is that it can be incorporated
into all types of existing provider payment schemes, from fee-for-service to capitation
to salary. This flexibility has helped the popularity of pay-for-performance programs
as payers do not have to completely undo their existing payment structure to add a pay-
for-performance component. In essence, a pay-for-performance system functions as a
secondary source of payment that operates in conjunction with the main remuneration
arrangement. If new monies are introduced to the payment system to fund the bonus
payments, then the original payment system remains intact. If the funds for bonus
payments are created by withholding some portion of the primary payment system,
then this can either be done prospectively, as in the case of prepaid capitation or
salaried arrangements, or retrospectively through adjustments with fee-for-service,
after total service quantities and payments are known.

Performance Unit for Measurement and Reward
Ideally, pay-for-performance programs would reward the agent or decision maker
directly responsible for changes in the standard being measured. Accountability
without responsibility is inappropriate and potentially counter-productive. Similarly,
unmerited rewards are inefficient and inequitable.
Identifying and attributing cause and effect with regard to improvements in health
can be difficult or even impossible. Health improvements may result from a multitude
of factors in addition to health care. Furthermore, health care interventions are often
collaborative efforts that may not be attributable to a single individual or health care
provider. Therefore, determining the marginal contribution of each team member can
be difficult and may engender unnecessary and undesirable conflicts of interest.
In practice, pay-for-performance programs vary substantially in whom they hold
accountable and reward. Individual physicians, physician groups, hospitals, managed
care plans, and even patients have been the subject of performance measures,
evaluation and reward. Because it is possible for more than one pay-for-performance
program to be in effect at the same time, it is conceivable that some patients and
providers might earn bonuses from a health plan that in turn might be evaluated under
a pay-for-performance program.
This wide ranging experience in pay-for-performance program design has not yet
produced evidence nor even a consensus on the most effective structure for health care
pay-for-performance programs. Recently initiated pay-for-performance programs take
on many forms, as discussed in a later section. Some argue that managed care plans are
best suited to pay-for-performance programs because their wide if not comprehensive
range of health care services most closely match their responsibilities for the overall
health care for its enrolled beneficiaries.¹⁸ Until more pay-for-performance evaluation
studies are completed that produce evidence identifying characteristics of the most
effective and efficient programs, the variation in pay-for-performance designs is likely
to continue.
Possible Shortcomings
and Unintended Consequences
While pay-for-performance offers the potential for improving quality and
efficiency, the success of such programs depends critically on the design and
implementation. The complex interactions between incentives, rewards, and other
market factors can produce perverse incentives and counterproductive results if
measures and objectives are not appropriately established or are poorly constructed.
The following discussion identifies a number of considerations that will determine the
effectiveness of a pay-for-performance system.

¹⁸ See, for example, Cannon, M.F. “Pay-for-Performance: Is Medicare a Good Candidate?”

7 Yale Journal of Health Policy, Law & Ethics (forthcoming 2007).

Inappropriate Measures and Objectives
Appropriately matched objectives and incentives, expressed through the chosen
measures, metrics, and standards for achievement, are critical to the success of any
pay-for-performance system. If incentives for performance are not well aligned with
the objectives, resources will be inappropriately diverted. Providers may lose faith in
a program that doesn’t accurately capture their honest efforts, undermining the
credibility of the program and the confidence of those being evaluated. With this
understanding, the IOM has recommended that Congress create a National Quality
Coordination Board (NQCB) and that this board, in collaboration with federal agencies
and private-sector stakeholders, “formulate and promptly pursue a research agenda to
support the development of a national system for performance measurement and
reporting.” This agenda would include
^!development, implementation, and evaluation of new measures to
address current gaps in performance measurement,
^!applied research focused on underlying methodological issues, such
as risk adjustment, sample size, weighting, and models of shared
accountability,
^!design and testing of reporting formats for consumer usability, and
^!valuation of the performance measurement and reporting system.
Competing or Uncoordinated Efforts
The overlap between health care providers and insurers is complex, with most
providers accepting patients who are covered by a myriad of insurers. If pay-for-
performance systems differ across insurers, providers will be striving to satisfy
multiple performance standards simultaneously, and the effects may be diluted. While
some of the performance measures may be common, particularly around clinical areas
for which the evidence base is convincing and there is a consensus, there may be areas
where different measures can develop. Some proponents have argued that competition
in this area will be beneficial because the insurance companies will be stimulated to
innovate and consumers — either employers selecting among insurance companies or
employees selecting among plans — will make decisions based in part on which pay-
for-performance program is most successful, however it may be measured.¹⁹
Competition across performance standards might also lead to initiatives that address
issues other than quality. These approaches may not fully acknowledge the potential
for inefficiency brought about through near redundancy or even conflicting objectives.
In contrast, a representative from a large insurance company has stated that the
company does not wish to develop its own pay-for-performance system and is waiting
for a consensus to develop on a set of the most appropriate measures and standards that²⁰

it can then adopt.
¹⁹ Cannon, M.F., op. cit.
²⁰ Statement by Lisa M. Latts, MD, MBA, MSPH, Vice President, Programs in Clinical
Excellence, WellPoint, Inc. during the 4^th Annual Quality and Pay-for-performance
Conference, Boston, MA, Aug. 6-8, 2006.

A related concern is that any pay-for-performance system proposed and
implemented by Medicare, as the market leader, would crowd out other innovations.
As the largest payer and market leader that sets trends often followed by private
insurers, Medicare’s pay-for-performance activities will be closely watched and
possibly copied.
Insufficient or Inappropriate Incentives
The incentives under a pay-for-performance system must be both appropriate to
modify behavior toward the desired objective, and sufficient to induce the effect.
Mismatched activities and targets and bonuses that don’t elicit the desired results will
lead to inefficiencies.
As with most incentives, there is a range of possible responses including no
change in behavior. Those for whom the costs of achieving the bonus threshold
exceed the potential gains may not exhibit any behavioral changes and become
frustrated. At the other extreme, bonus payments to those who already meet the
standard may not produce any changes in behavior either but will reward them for
standards they already have reached or exceeded. In this case, the justification would
be to provide incentive for those providers to continue to maintain a high level of care.
The design of incentives is most critical for the group that would change its
behavior towards the desired outcome if incentives were sufficient. The difference
between the costs of meeting the threshold and the potential gains to be realized will
be different for different providers under different circumstances. Consequently, the
policy implications will vary depending on the population, the activity targeted and the
size of the potential bonus.
The appropriate size of the ideal incentive will vary, depending on what group
and which objectives are being targeted. While MedPAC has suggested that 1% or 2%
of total payments be withheld to fund the bonus pool, this figure may reflect very
different incentives across providers. For example, a physician in solo practice who
has annual gross revenues of $250,000, half of which comes from Medicare, would
face a potential bonus payment of $2,500 while a hospital with annual gross revenues
of $300 million, 60% of which is Medicare related, will have over $3.6 million at risk
under the pay-for-performance system. In the case of the physician, the amount may
not come close to the costs of meeting the pay-for-performance criteria, considering
the cost of acquiring the necessary information systems (hardware and software) that
may be needed, in addition to the hiring and /or training required of the staff. On the
other hand, the hospital not only has much of the fixed costs of compliance in place
(including automated systems and clerical staff), but the amount of the payment at risk
would almost certainly make the effort worthwhile. Similarly, some have argued that
pay-for-performance is best suited for managed care companies who not only would
have significant amounts at stake, but would stand to recoup the benefits of any
potential cost savings because of the vertically integrated nature of the care.

Excessive Focus on the Reward
Incentives can be powerful motivators and, ceterus paribus, the behavior being
targeted will increase. The caution is that, “You get what you measure,” and non-
targeted areas may suffer as a result. Some other positive efforts that may be more
difficult to measure, are not fully understood, or simply not targeted, could end up
being neglected or underappreciated.
Pay-for-Performance and Cost Control
Pay-for-performance systems are not inherent vehicles for cost containment.
Advocates of pay-for-performance believe that successful pay-for-performance
programs will lead to improvements in quality and provider efficiency which in turn
will lower cost.²¹ However, this assertion is not necessarily true, nor is there
substantial evidence to support this position. One review of the existing research finds
that there are only nine randomized controlled trials of the effectiveness of pay-for-
performance systems, and many of the studies are limited to a single indicator or aspect
of care.²² While a randomized controlled trial may be the gold standard for isolating
and identifying the true effectiveness of an intervention, such experiments are rarely
possible in social policy contexts. Pay-for-performance systems could potentially be
cost-saving but also may be cost-increasing, and to examine how this might occur
requires a more systematic analysis of the mechanisms for change.
Considered simply, quality improvement involves doing the wrong things less
often and/or doing the right things more frequently. In practice, quality can improve
by reducing or eliminating overuse or misuse, while increasing underuse. The
difficulty comes in translating these generally innocuous concepts into identifiable
actions and activities in health care.
To the extent that quality improvement measures in pay-for-performance systems
incorporate measures of underuse, then the short-term consequence of quality
improvement would be higher cost of care. Whether the long-term costs would be
lower as a result of this short-term increase would depend on the ability of the care to
offset potentially more expensive care later on. Examples might include the increased
administration of vaccines, or the early detection of treatable cancers. If the quality
improvement is achieved by eliminating overuse or misuse, the potential for cost-
savings is more direct. Not only would the direct costs of the inappropriate care be
avoided but the negative consequences that lead to complications and additional care,
such as nosocomial infections, would be prevented.

²¹ Trude, S. Au, M., and Christianson, J. “Health Plan Pay-for-Performance Strategies,”
American Journal of Managed Care. 2006; 12:537-542.
²² Dudley, R.A., Frolich, A., Robinowitz, D.L., Tlavera, J.A., Broadhead, P., and Luft, H.S.
Strategies to Support Quality-Based Purchasing: A Review of the Evidence. Rockville,
Maryland: Agency for Healthcare Research and Quality, publication 04-0057, 2004.

In practice, the current state of medical knowledge, management practices, and
policy environment is insufficient for the identification of what the appropriate level
of care is in most clinical situations. There is a movement towards more evidence-
based medicine, but the state of the knowledge base is still in its early stages.²³
The existing examples of pay-for-performance in health care and their evaluations
reflect some of the shortcomings of the absence of a true experimental design and the
limitations that call for reservations in generalizing the findings. Even if pay-for-
performance produces the successes claimed in the sample population, the differences
between the sample and non-sample populations mean that study findings cannot be
simply inflated and extrapolated to represent population results. For instance, it would
not be reasonable to assume that the results obtained in the sub-sample of hospitals that
participated in the Premier demonstration would be reproducible in the remaining
Premier hospitals that did not participate in the demonstration, let alone all non-
Premier hospitals. Similarly, measures selected for the demonstration project and other
current pay-for-performance efforts are likely to include the “low-hanging fruit” in the
orchard of possible measures, and further successes are likely to become more difficult
to achieve.
The design of pay-for-performance programs can also affect the potential for cost
savings. The incentive fund for bonus payments can be created by adding more money
to existing payments or by withholding a fraction of payments. While larger bonus
payments may create more of an incentive, they also create an additional hurdle to
achieving overall cost savings. If new money is used, the pay-for-performance program
would need to generate program savings that equal or exceed the new bonus payments.
For these and other reasons, most of the claims of cost-control based on
extrapolations of current pay-for-performance experiences are difficult to quantify and
may be overstated. While pay-for-performance systems that emphasize quality
objectives and measures have the potential for improving health and health care
delivery, the degree of absolute and relative improvement possible in disparate settings
and the cost-effectiveness of the efforts remains unknown.

²³ Institute of Medicine, Rewarding Provider Performance: Aligning Incentives in Medicare.
National Academy Press, Washington, DC, 2006.

Pay-for-Performance Initiatives
Pay-for-performance programs are growing in number, with most of the current
programs and growth occurring among commercial health plans. According to an
industry survey, 107 pay-for-performance programs were in place as of November,
2005, up from 84 the year before. (See Figure 1.) Similar growth is expected for the
period since the survey and for the next few years. The IOM claims that more than 100
reward and incentive programs are in operation in the private sector.²⁴
Figure 1. Growth in Pay-for-Performance Programs by Sponsor Type

^Source:^N^a^tⁱ^oⁿ^a^l^Pay^-^f^o^r-^Perf^orm^aⁿ^{ce S}^u^rv^ey^{, 2005, Med-}^V^aⁿ^t^ag^e.
²⁴ Institute of Medicine, Rewarding Provider Performance: Aligning Incentives in Medicare.
National Academy Press, Washington, DC, 2006.

Pay-for-performance programs are relatively recent developments in U.S. health
care and there are few rigorous objective evaluations of the effect of the programs.
Initial studies suggest that pay-for-performance programs can change performance on
quality measures that are used for the basis of bonus payments, but despite the
declarations of some proponents, pay-for-performance as a cost-saving initiative is
largely unsubstantiated. There is a scarcity of evidence that long-term cost savings are
created through pay-for-performance programs, in part because many programs have
only been in existence for a few years, and the long-term outlook for chronic
conditions such as diabetes and heart disease, clinical conditions that are commonly
the focus of pay-for-performance programs, will not be known for years after any
incremental intervention. A 2004 survey of the research found that there had been only
nine randomized controlled trials measuring the effects of pay-for-performance in
health care at the time of the study.²⁵
Public Sector
Medicare. The Centers for Medicare and Medicaid Services (CMS) currently
has a number of pay-for-performance initiatives underway that encourage quality care
in physicians’ offices and ambulatory care facilities, hospitals, nursing homes, home²⁶
health care agencies and dialysis facilities. Many of these demonstrations were
mandated by the Benefits Improvement and Protection Act (BIPA, 2000) or the
Medicare Modernization Act (2003). The programs are summarized in Table 2.

²⁵ Dudley, et al., 2004.
²⁶ See the CMS website for additional information: [http://new.cms.hhs.gov/apps/
media/press/release.asp?Counter=1343].

Table 2. CMS Pay-for-Performance Initiatives
Pay-for-performanceSummary Description
Program
Hospitals
Hospital QualityAll hospitals are required to report a set of quality measures in
Initiative [MMAorder to receive the full Medicare DRG update. Under MMA,
Section 501(b) andhospitals that didn’t report would have received the market
Deficit Reduction Actbasket update less 0.4% in 2005 and 2006. DRA increased the
(DRA) Sectionamount at risk to 2% in 2007 and subsequent years and

5001(a)]expanded the measure set.

Premier HospitalFinancial incentives are offered to 260 hospitals for high quality,
Quality Incentiveas demonstrated on 34 quality measures relating to five clinical
Demonstrationconditions. Hospitals scoring in the top 10% will receive a 2%
bonus payment. Those scoring in the next highest 10% will
receive a 1% bonus. In the third year of the demonstration, those
hospitals that do not meet a predetermined threshold score on
quality measures will be subject to reductions in payment.
Physicians or Integrated Health Systems
Physician GroupTen large (200+ physicians) groups earn performance-based
Practice (BIPA 2000)Medicare fee-for-service payments on quality results. The
demonstration seeks to encourage coordination of Part A and
Part B services, promote efficiency through investment in
administrative structure and process, and reward physicians for
improving health outcomes.
Care ManagementA three-year pay-for-performance demonstration with
Performance Demophysicians to promote the adoption and use of health
(MMA Section 649)information technology to improve the quality of patient care for
chronically ill Medicare patients. Focused on small and
medium-sized physician practices, this demo will be
implemented in four states: Arkansas, California, Massachusetts,
and Utah, with the support of the Quality Improvement
Organizations in those states.
Health Care QualityA five-year demonstration program under which eligible
Demo (MMA Sectionphysician groups, integrated health systems, or regional
646)coalitions of the same, aim to enhance quality by improving
patient safety, reducing variations in utilization by appropriate
use of evidence-based care and best practice guidelines,
encouraging shared decision making, and using culturally and
ethnically appropriate care.
Disease Management/Chronic Care Improvement
Chronic CareParticipating organizations are paid a monthly per beneficiary
Improvement Programfee for managing a population of chronically ill beneficiaries
(MMA Section 721)with advanced congestive heart failure and/or complex diabetes
to test a population based model of disease management.
Payment of fees is contingent upon performance on quality
measures and satisfaction of both beneficiaries and providers.

Pay-for-performanceSummary Description
Program
After two years, pending successful interim results, this pilot
may be expanded more broadly, possibly nationally.
ESRD DiseaseThis 3-year demonstration will test a fully case-mix adjusted
Managementpayment system for an expanded bundle of end stage renal
Demonstration (MMAdisease (ESRD) services. A portion of the payment will be
Section 623)linked to ESRD-related quality measures.
Disease ManagementThis demonstration, which began enrollment in February 2004,
Demonstration foris designed to test whether applying disease management and
Severely Chronicallyprescription drug coverage in a fee-for-service environment for
Ill Medicarebeneficiaries with illnesses such as congestive heart failure,
Beneficiaries (BIPAdiabetes, or coronary artery disease can improve health

2000)outcomes and reduce costs.

Disease ManagementDisease management services are being provided to dually
Demonstration for(Medicare & Medicaid) eligible beneficiaries in Florida who
Chronically Ill Dualsuffer from advanced-stage congestive heart failure, diabetes, or
Eligible Beneficiariescoronary heart disease. The demonstration combines the
resources of what had previously been the state’s Medicaid
pharmacy benefit with a disease management activity funded by
Medicare to coordinate the services of both programs, with the
goal of improving quality while lowering total program costs.
The demonstration organization is being paid a fixed monthly
amount per beneficiary and is at risk for 100% of its fees if
performance targets are not met.
Care Management ForThis demonstration will test models of care management in a
High Costpopulation of high-cost and high-risk Medicare fee-for-service
Beneficiariesbeneficiaries. Participating providers are required to meet
relevant clinical quality standards as well as guarantee savings
to the Medicare program.
^Source:^C^M^S^[^h^t^t^p^:^/^/^www.^c^m^s^.^h^h^s^.^g^o^v^/^a^p^p^s^/^m^e^dⁱ^a^/^p^r^e^s^s^/^r^e^l^e^a^s^e^.^a^s^p^?^C^o^uⁿ^t^e^r⁼^1343].
Most of the demonstrations are still in the development stage or too early in the
implementation phase to be properly evaluated. However, two projects have been in
place long enough to produce preliminary results.
The Hospital Quality Initiative was mandated in MMA section 501(b). Eligible
hospitals that submitted data for a “starter set” of ten quality measures in 2004 would
receive the full annual payment update, while those that did not submit performance
data would receive a 0.4 percentage point reduction in their annual payment update for
FY2005, 2006 and 2007.²⁷ Although participation was voluntary, 98.3% of the
hospitals reported the quality measures in the first year of the program. The Deficit
Reduction Act of 2005 (DRA) increased the amount at risk to a 2% reduction, “for

²⁷ CMS, Hospital Quality Initiative Overview, December 2005. [http://www.cms.hhs.gov/
HospitalQualityInits/downloa ds/HospitalOverview200512.pdf].

fiscal year 2007 and each subsequent fiscal year” and instructs the Secretary to expand
the measure set.²⁸
This high participation figure has been cited as evidence that pay-for-performance
(or more accurately in this case, pay-for-reporting) works well and that incentive
payments do not need to be very substantial in order to achieve significant results.²⁹ A
more realistic assessment of this result might be that hospitals would have been
imprudent to “leave money on the table” since the quality measures are widely used
and already reported for other purposes, e.g., for Joint Commission on Accreditation
of Healthcare Organizations (JACHO) accreditation and Quality Improvement
Organization (QIO) programs, so the marginal cost of reporting to CMS was exceeded
by the revenue gains.³⁰ Had hospitals not already had the infrastructure in place
(information systems and staff) or if the measures had been previously uncollected or
reported or less universally accepted, the participation rate almost certainly would have
been lower.
In October 2003, CMS and Premier, Inc., a nationwide alliance of not-for-profit
hospital facilities and healthcare systems, began a three-year pay-for-performance
demonstration to test whether monetary incentives for improvements in quality
measures would lead to measurable differences in quality and costs. The demonstration
included 260 hospitals in 38 states. The top hospitals on 34 quality measures across
5 clinical conditions were eligible for a bonus of up to 2% of the Medicare DRG
payment, which amounted to $8.85 million to the top 123 performers in the first year.³¹
The demonstration ended in September 2006 and some preliminary results are
available. The first year results showed increases in composite quality scores for all
five clinical areas:
^!from 87 to 91 percent for patients with acute myocardial infarction
(heart attack)
^!from 65 to 74 percent for patients with heart failure,
^!from 69 to 79 percent for patients with pneumonia,
^!from 85 to 90 percent for patients who received a coronary artery
bypass graft, and
^!from 85 to 90 percent for patients with hip and knee replacements.

²⁸ P.L. 109-171, section 5001(a).
²⁹ See, for example, testimony of Herb Kuhn, Director, Center for Medicare Management
before the Senate Finance Committee, July 27, 2005 [http://www.cms.hhs.gov/
apps/media/press/testimony.asp?Counter=1537].
³⁰ A full list of the hospital quality measures can be obtained from the source listed in the
prior footnote. Measures include whether aspirin and beta-blockers are given at arrival and
discharge to acute myocardial infarction (heart attack) patients; whether patients admitted
for pneumonia receive an initial antibiotic within 4 hours of arriving at the hospital and are
counseled about smoking cessation; and whether prophylactic antibiotics are received within
1 hour prior to surgical incision and discontinued within 24 hours after the end of surgery
to prevent surgical infection.
³¹ CMS press release, “Medicare Demonstration Shows Hospital Quality of Care Improves
With Payments Tied to Quality,” Nov. 14, 2005.

Premier representatives reportedly claim that the improved care achieved through
the demonstration has led to 3,000 fewer deaths, 6,000 fewer medical complications,
6,000 fewer hospital admissions and 500,000 fewer days in the hospital for the
participating population. Furthermore, “improvement in evidence-based quality
measures is expected to save Medicare money over time because [of] the demonstrated
relationship to improved patient health, fewer complications and fewer hospital
readmissions.”³²
While the achievements are admirable, the results and implications for adopting
the program more widely should be interpreted with caution. There is self-selection at
work both in the study sample of hospitals and in the measures chosen for the program.
Not all Premier hospitals participated in the study, and there likely would have been
a difference in the experience of organizations who voluntarily chose to participate
compared to those that did not. Similarly, extending the program to more difficult
clinical areas or to objectives and measures where the evidence base is not so settled
would also produce more tempered results.
Medicare Quality Improvement Organizations (QIOs) have been given a role in
the Medicare Care Management Performance Demonstration (MCMP), mandated by
the MMA and set to be launched in four states (Arkansas, California, Massachusetts,
and Utah). The MCMP seeks to encourage the dissemination and adoption of HIT and
the electronic reporting of quality measures through a pay-for-performance program.
As part of their scope of work, QIOs are offering assistance to physicians’ offices in
the adoption of health information technology, patient-focused care processes, and
clinical measures reporting through tools and methods developed in the Doctors Office
Quality - Information Technology (DOQ-IT) pilot project.³³ Physicians can earn an
annual clinical bonus of up to $10,000 per year over three years ($50,000 per practice)
based on composite scores of quality measures over four clinical areas as well as an
additional 25% of the clinical bonus (up to $2,500 per physician, $12,500 per practice)
for reporting the data electronically.³⁴
While not a pay-for-performance project by itself, the Physician Voluntary
Reporting Program (PVRP) may lay the groundwork for establishing the types of
reporting pathways that will be critical to any future CMS pay-for-performance
program. Launched in October 2005 with data collection beginning in January 2006,
the goal of the PVRP is to gather data about, “the quality of care provided to Medicare
beneficiaries in order to identify the most effective ways to use the quality measures
in routine practice and to support physicians in their efforts to improve quality of care.”
CMS states that, “the objective of the PVRP is to help physicians obtain information

³² CQ Healthbeat, “Premier P4P Demo Shows Better Care Cuts Health Costs,” June 20,

2006.

³³ The Doctor’s Office Quality - Information Technology (DOQ-IT) program is a national
initiative that promotes the adoption of Electronic Health Record (EHR) systems to improve
quality and safety for Medicare beneficiaries in small- and medium-sized physician offices.
³⁴ Presentation by Jody Blatt, Medicare Demonstrations Program Group, Office of Research,
Development & Information, Centers for Medicare & Medicaid Services at the 4th Annual
World Congress Leadership Summit on Healthcare Quality and Pay-for-Performance,
Boston, MA, Aug. 7, 2006.

they can use to improve quality and avoid unnecessary costs,” and will provide
feedback to physicians as early as July 2006 on their level of performance based upon
the data submitted.
Medicaid. Pay-for-performance systems have recently been introduced to state
Medicaid programs either directly by the state agencies or indirectly through Medicaid
managed care plans. Several states, including California, Iowa, Massachusetts and
New York, have incorporated financial or other bonuses based on achieving quality
measures into their managed care contracts. These quality measures are often adopted
from the National Committee for Quality Assurance’s Health Plan Employer Data and³⁵
Information Set (HEDIS).
The pay-for-performance features have varied across states. For example, health
plans participating in the New York Medicaid program who qualified to receive bonus
payments were awarded an additional 1% of monthly capitation payments. In
California, budget restrictions prevented the awarding of bonus payments, however
plans that scored higher on quality measures were given preference by the state when
assigning Medicaid beneficiaries who failed to select a health plan on their own.
Finally, CMS is also conducting a pay-for-performance demonstration project for
Medicare and Medicaid dual eligibles. The Disease Management Demonstration for
Chronically Ill Dual Eligible Beneficiaries targets eligible beneficiaries in Florida who
suffer from advanced-stage congestive heart failure, diabetes, or coronary heart disease
and provides disease management services and drug coverage under a capitation
framework. The demonstration organization is at risk for 100% of its fees if
performance targets are not met, with any potential savings shared equally between³⁶
CMS and the demonstration organization.
Private Sector
Several pay-for-performance programs have been implemented by insurers and
employers in the private sector and others are continuing to be developed and
introduced. Two of the earliest and most widespread efforts, one that focuses on
physicians and the other on hospitals, are described here.
Bridges to Excellence. Bridges to Excellence (BTE) is a not-for-profit
multi-state, multi-employer coalition developed by employers, physicians, healthcare
services researchers and other industry experts created to encourage and reward quality
in health care.³⁷ BTE currently has three pay-for-performance programs in operation:

³⁵ Foubister, V. “Pay-for-performance in Medicaid,” Quality Matters: April Update, The
Commonwealth Fund, April 2005.
³⁶ CMS Office of Public Affairs, Fact Sheet released Monday, Jan. 31, 2005
[ ht t p: / / www.cms.hhs.gov/ a pps/ medi a / pr e ss/ r e l e ase.asp?Count er =1343] .
³⁷ Bridges to Excellence participants include groups affiliated with the National Business
Coalition on Health (NBCH), large employers, health plans, the National Committee for
Quality Assurance, the American Board of Internal Medicine, and several Quality
Improvement Organizations. Participants share the goal of “improving health care quality
(continued...)

the Physician Office Link, the Diabetes Care Link and the Cardiac Care Link. Each
program rewards financial bonuses to physicians who satisfy the performance criteria.
The Physician Office Link program encourages the adoption and use of health
information technology. Incentives are based on the physician office’s implementation
of specific processes to reduce errors and increase quality, including the adoption and
use of electronic medical records that have been certified by the Certification
Commission for Health Information Technology. Physicians can earn up to $50 per
year for each patient covered by a participating employer or plan.
The Diabetes Care and Cardiac Care Link programs reward physicians who
demonstrate high performance in diabetes or cardiac care over a one- or three-year
period. Physicians can receive a bonus (up to $80 for each diabetic patient and $160
for each cardiac patient) covered by a participating employer and plan. Patients are
provided with products and tools “to help [them] get engaged in their care, achieve
better outcomes, and identify local physicians that meet the high performance
measures.” BTE claims that the cost to employers is no more than $175 per diabetic
patient per year with savings of $350 per patient per year and no more than $200 per
cardiac patient per year with savings up to $390 per patient per year.
Leapfrog. The Leapfrog Hospital Rewards Program is a pay-for-performance
program that rewards hospitals “that demonstrate excellence and/or sustained quality³⁸
and efficiency improvement.” Data are collected and performance is measured in five
clinical areas that account for 20% of commercial inpatient spending and 33% of
commercial admissions: coronary artery bypass graft (CABG); percutaneous coronary
intervention (PCI); acute myocardial infarction (AMI); community acquired
pneumonia (CAP); and deliveries/newborn care. The program claims to produce
savings from reduced lengths of stay due to complications and reductions in
re-admission rates as a result of improvements in quality. Because the data is already
collected for surveys needed by the Joint Commission on the Accreditation of
Heathcare Organizations (JCAHO), a national accrediting organization, the program
claims that there is no additional reporting burden for participating hospitals. The
program is standardized and marketed commercially.

³⁷ (...continued)
through measurement, reporting, rewards and education.” For further information, see
[http://www.bridgestoexcellence.org/ bte/about_us/home.htm] .
³⁸ The Leapfrog Group is a voluntary program consisting of organizations that buy health
care. The group applies employer purchasing power to recognize and reward “big leaps” in
health care safety, quality and customer value. The consortium includes Fortune 500
companies and other large private and public health care purchasers. Additional information
is available at [http://www.leapfroggroup.org/about_us/leapfrog-factsheet].

United Kingdom’s Quality and Outcomes Framework
There have been several attempts over the last few decades to pay health care
providers differentially according to quality in the United Kingdom, but it was only in
2003 that a new “Quality and Outcomes Framework” (QOF) was put in place that
systematically reimbursed providers for successfully achieving pre-defined quality
improvement standards.³⁹ Under the QOF, general practitioners can earn up to 25% of
their incomes in incentive payments by meeting certain clinical and access indicators.
Up to 1000 points can be earned for achievement against a set of 136 performance
measures for clinical care, practice organization, and patient experience, with another
50 points available for providing prompt access to services. The points are directly
translatable to bonuses (in pounds sterling £) paid annually.
While apparently successful, the experience in the UK may not be very indicative
of pay-for-performance systems under consideration in the U.S. First, the U.K. QOF
applies only to family practitioners, who comprise about half of the medical workforce
of the National Health Service (NHS) and derive the majority of their income from
NHS patients.⁴⁰ In contrast, only 10.4% of the practicing physicians in the U.S. are in
family medicine or general practice.⁴¹
Second, this program was successful in large part because additional funds were
available for the government to provide bonus payments. The Ministry of Health
originally expected that the average score would be around 750 but instead, the
average score in the first year was over 950. This required the NHS to provide £1.0
billion ($1.8 billion) of new money to cover the bonus payments.⁴²
Third, practitioners in the NHS may omit certain patients who meet exclusionary
criteria from the calculation of the performance measures (both the numerator and the
denominator).⁴³ While this may lead to the manipulation of the system, the evidence
is inconclusive as to whether substantial “gaming” is occurring.
Physicians and Pay-for-Performance
Ideally, a successful pay-for-performance program would include general
agreement and buy-in among those being evaluated that the objectives are fair, that
their performance is being measured accurately, and that the incentives make the effort
worthwhile. Having the support of physicians is critical; clinical measures will differ

³⁹ Roland M. “Linking Physicians’ Pay to the Quality of Care — A Major Experiment in the
United Kingdom.” New England Journal of Medicine, 351: 1448-54, 2004.
⁴⁰ Ibid.
⁴¹ American Medical Association, Physician Characteristics and Distribution in the U.S.,

2006 edition.

⁴² Roland, M. presentation to the 4th Annual World Congress Leadership Summit on
Healthcare Quality and Pay-for-Performance, Boston, MA, Aug. 7, 2006.
⁴³ Patients could be excluded from the QOF performance calculations if they refused to be
included, if they were not clinically appropriate, if they were newly diagnosed or recently
registered, or if they were already on maximum doses of medication.

for each specialty and will need to be developed in conjunction with practitioners to
insure that they are appropriate and practical.
Pay-for-performance issues have created an uneven relationship between the
professional medical societies and Congress over recent months. The AMA initially
objected to the CMS PVRP, claiming that the PVRP would create additional
administrative burdens and lead to an eventual mandate on performance measures. In
turn, “physician concerns about the initial CMS Physician Voluntary Reporting
Program proposal were interpreted on Capitol Hill as a sign of opposition to quality
reporting.”⁴⁴ In response to these and other comments, CMS revised the program to
address some of the concerns about administrative burden, for instance, by reducing
the number of core measures from 36 to 16.⁴⁵
The various medical societies also have had difficulties coming to agreement on
developing performance criteria that could potentially be used in pay-for-performance
programs. On December 5, 2005, the American Medical Association (AMA) sent a
letter to CMS stating that the organization would oppose efforts to adopt a
pay-for-performance system under Medicare unless Congress repeals the current
method for calculating yearly physician payment increases.⁴⁶ On December 16, 2005,
the AMA leadership and chairs of three committees or subcommittees of jurisdiction
co-signed a working agreement to develop “a starter set of evidence-based quality
measures” for “a total of approximately 140 physician performance measures covering
34 clinical areas” by the end of 2006 that would be used “to implement ... reforms to
address payment and quality objectives.” The agreement also stated that physicians
would voluntarily report on “at least 3 to 5 quality measures per physician” and
“receive an additional quality update to offset administrative costs.” Furthermore,
physician groups “will have developed performance measures to cover a majority of
Medicare spending for physician services” by the end of 2007.⁴⁷
This working agreement was criticized by some for being premature, especially
by specialty physicians. In a January 26, 2006 letter to the Committee chairs, the
Alliance of Specialty Medicine, a coalition of 13 national medical specialty societies,
expressed its belief that, “through a collaborative and thoughtful process, a value-based
purchasing system is possible,” and emphasized that, “[t]o succeed, any national
quality measurement initiative must work with the realities of specialty medicine, be
specialty specific, and be developed by the specialty societies with expertise in the area
of care in question.” The Alliance stated that, “this [AMA] agreement binds

⁴⁴ AMA memo to State Medical Associations and National Medical Specialty Societies, Feb.

7, 2006. Copy obtained from [http://www.insidehealthpolicy.com/secure/data_extra/

dir_06/he2006_0548_1.pdf].
⁴⁵ CMS, Physician Voluntary Reporting Program Overview [http://www.cms.hhs.gov/
PV RP/01_Ove rview.asp].
⁴⁶ The current formula for determining yearly increases in physician payment under
Medicare is based on the Sustainable Growth Rate (SGR), which mandates decreases in
payments for physicians for the next several years. See CRS Report RL31199, Medicare:
Payments to Physicians, by Jennifer O’Sullivan.
⁴⁷ “Joint House-Senate Working Agreement With the AMA,” signed on December 16, 2005
[ ht t p: / / www.i nsi deheal t hpol i c y.com/ secur e / dat a_ext r a/ di r _06/ he2006_0548_2.pdf ] .

organizations to time lines and processes that may not be able to be accomplished by
all medical societies,” and expressed concern that there was no assurance that doctors
would be adequately paid for treating Medicare patients.⁴⁸
The AMA claimed a misunderstanding and denied that it had over-promised. In
a memo to State Medical Associations and National Medical Specialty Societies, the
AMA stated that, “The commitment to develop 140 physician performance measures
and to cover a majority of Medicare spending represents work either already completed
by the [Physician Consortium for Performance Improvement] or was in the planning
stages at the end of [2005],” and that the AMA “did not commit any individual state
or national specialty society to the activities outlined in the agreement.” These events
illustrate the sensitivity of the issue and the difficulty in achieving consensus on
developing appropriate quality metrics.⁴⁹
Legislative Activity
In recent years, Congress has considered legislation that would implement pay-
for-performance in health care. Most of the efforts have centered on the Medicare
program, although Medicaid and private insurers have also garnered some attention.
A number of pay-for-performance demonstration projects have been authorized, and
Congress issued a directive to the Secretary of Health and Human Services to develop
a value-based purchasing program for hospital payment by 2009. H.R. 6111, the Tax
Relief and Health Care Act of 2006, creates a voluntary pay-for-performance system
for physicians who serve Medicare patients that will pay bonuses based on the
satisfactory reporting of quality measures.
S. 1932, the Deficit Reduction Act of 2005
In the fall of 2005, during the budget reconciliation deliberations that led to the
Deficit Reduction Act of 2005 (P.L. 109-171, DRA), the Senate passed S. 1932, a bill
that contained language to establish a pay-for-performance program under the
Medicare program.⁵⁰ The proposals in S. 1932 would have established value-based
purchasing systems for each of the different Medicare providers. There would be
separate value-based purchasing programs for hospitals, physicians and other
practitioners, Medicare managed health care plans and prescription drug plans, ESRD
providers and facilities, home health agencies, and skilled nursing facilities. Although
the specifics of each program differed in the details, they all shared some general
principles.

⁴⁸ Alliance of Specialty Medicine letter to Committee chairs Barton, Grassley, and Thomas,
Jan. 26, 2006. Copy available from [http://www.insidehealthpolicy.com/secure/
data_extra/dir_06/he2006_0569.pdf].
⁴⁹ AMA memo from Maves, M. to State Medical Associations and National Medical
Specialty Societies, “Joint House-Senate Working Agreement with the AMA,” Feb. 7,

2005.

⁵⁰ S. 1932 contained many elements of S. 1356, the Medicare Value Purchasing Act of 2005,
a bill introduced by Sen. Grassley at the end of June 2005.

^! The value-based purchasing programs would begin collecting data on
quality measures in the initial year of establishment, with incentive
payments disbursed in subsequent years. Data from the initial year
would be used to inform providers what their payments would have
been for the year had the value-based purchasing program already
been in place.
^!Following MedPAC’s recommendation, each value-based purchasing
program would create an incentive pool funded by withholding up to
2% of total payments to that category of provider. The percentage of
funds that goes towards the incentive pool would not decrease over
time, and all funds collected for the year incentive pool must be paid
to qualifying providers as incentive payments under the program for
that year.
^!Participation in the value-based purchasing program would be
voluntary, but providers would be required to report quality data in
order to be eligible for incentive payments.
^!Incentive payments would be paid to providers who meet certain
thresholds for quality measurement. These thresholds would be based
on either relative or absolute standards.
^!The quality measures would be specific to each category of providers
and would be revised over time, but the measures would be required
to be evidence-based, easy to collect and report, address process,
structure, outcomes, beneficiary experience, efficiency, over- and
underuse of health care, and to address disparities in health care
provided and health outcomes between majority and minority groups.
In the initial year, the measures would include at least one measure of
health information technology infrastructure.
^!Because all the funds collected under the value-based purchasing
programs would be paid out as incentive payments, the total payments
over time would not change as a result of these provisions, but the
timing of the incentive payments would be delayed a year compared
to payments made in the absence of the value-based purchasing
programs.
Eventually, only two pay-for-performance provisions from those deliberated as
part of the DRA passed and became law: section 5001(b) requires that the Secretary
of Health and Human Services (HHS) develop a plan to implement a value based
purchasing program for Medicare hospital payments beginning with FY2009, and
section 5201(d) requires MedPAC to develop recommendations for a value based
payment system for Medicare home health services.⁵¹

⁵¹ Another provision establishes quality adjustments in Medicare inpatient hospital payments
for certain hospital acquired infections, but not in a pay-for-performance framework.

The requirement for the development of a hospital pay-for-performance system
for Medicare payments does not include detailed specifics about such a program. The
general guidelines contained in the statute require the value-based purchasing program
to include the following: on-going development, selection, and modification of
measures of quality and efficiency in hospital inpatient settings; the reporting,
collection, and validation of quality data; the establishment of thresholds and standards
that would substantiate value based payments and the source of such funds; and the
disclosure of information on hospital performance.⁵²
DRA also included a provision that requires MedPAC to develop
recommendations for a detailed structure of value based payment adjustments for
home health services under the Medicare program by June 1, 2007. Specifically, the
report is to include recommendations concerning the determination of thresholds for
payments, the size of value based payments, the sources of funds for these payments,
and the relationship of payments to the improvement and attainment of quality.
H.R. 3617, the Medicare Value-Based Purchasing
for Physicians’ Services Act of 2005
In July 2005, Ways and Means Health Subcommittee Chair Johnson and several
co-sponsors introduced the Medicare Value-Based Purchasing for Physicians’ Services
Act of 2005. This bill would establish a value-based purchasing program under
Medicare based on measures of quality (“Q-measures”) and efficiency (“E-measures”).
Among other things, these measures were to be, “evidence-based, if pertaining to
clinical care, ... consistent, valid, practicable, and not overly burdensome to collect, ...
and relevant to physicians and other practitioners [and Medicare beneficiaries].” Each
physician specialty organization would have been requested to submit measures to a
“consensus-building organization,” such as the National Quality Forum, that would
then submit recommendations to the HHS Secretary. The measures would be used to
rate physicians to determine whether the physicians’ billing unit would be eligible for
the performance incentive (an increased annual update).⁵³
H.R. 6111, the Tax Relief and Health Care Act of 2006
In December 2006, Congress passed legislation implementing a voluntary quality
reporting system that ties a portion of the payments for Medicare professional services⁵⁴
to the reporting of claims data. Physicians and other eligible professionals who
provide health care services to Medicare beneficiaries between July 1 and December
31, 2007, and who satisfactorily report the quality information, would be eligible for
a bonus payment based on their Medicare reimbursement for the same period.

⁵² P.L. 109-171, Section 5001(b).
⁵³ This bill also would have eliminated the method of calculating annual Medicare physician
payment updates in accordance with the Sustainable Growth Rate (SGR).
⁵⁴ The legislation also overrides the 5% cut that would have taken effect due to adjustments
based on the SGR and provides for a 0% update, on average, to Medicare physician
payments for 2007.

The pay-for-performance measures to be used as the basis for this bonus are
borrowed from an existing CMS program. For covered professional services furnished
beginning July 1, 2007, and ending December 31, 2007, the quality reporting measures
are those identified as physician quality measures under the CMS Physician Voluntary
Reporting Program (PVRP). The set of quality measures may be modified until April
1, 2007; however, no measures may be added or removed after that date, although
modifications or refinements to previously published quality measures are allowed
(without notice or opportunity for public comment) up until July 1, 2007.
To qualify for the bonus, providers must meet reporting guidelines. Eligible
professionals who (1) furnish services for which there are established quality measures
as described above and (2) satisfactorily submit quality measures would be paid a
single additional bonus payment amount equal to 1.5% of the allowed charges for
covered professional services furnished during the reporting period. The bonus
incentive payments would be paid from the Supplemental Medical Insurance Trust
Fund (Part B).⁵⁵
Satisfactory reporting of data determines whether the provider is eligible for the
bonus payment. If there are no more than three quality measures that are applicable to
the professional services provided, the provider must report each measure for at least
80% of the cases to meet the criteria. If there are four or more quality measures that are
applicable, the provider must report at least three of the quality measures for at least

80% of the cases.

CMS would have the authority to determine whether providers qualify for and
receive the bonus. This would include the ability to validate (by sampling or other
means) to determine if the reported measures are applicable to the professional services
provided. If CMS determines that an eligible professional has not reported applicable
measures, the provider would not receive the bonus. The provision also places a limit
on bonus payments.⁵⁶
The pay-for-reporting program established by H.R. 6111 might be transitory. For
2008, the quality measures would change to a set of measures adopted or endorsed by
a consensus organization (such as the National Quality Forum or AQA, originally
known as the Ambulatory Care Quality Alliance) that includes measures that have
been submitted by a physician specialty organization developed through a
consensus-based process. The CMS administrator would publish a proposed set of
quality measures for 2008 in the Federal Register no later than August 15, 2007, with

⁵⁵ These bonus incentive payments would not be taken into account in the calculations and
determination of payments for providers in health professional shortage areas or Physician
Scarcity Areas.
⁵⁶ CMS would estimate the average per measure payment amount (APMPA) equal to (the
total amount of allowed charges under Medicare part B for all covered professional services
furnished during the reporting period on claims for which quality measures are reported)
divided by (the total number of quality measure for which data are reported during the
reporting period under the physician reporting system). No provider would receive
payments in excess of the product of the total number of quality measures for which data
are submitted and three times the average per measure payment amount.

a public comment period. The final set of measures appropriate for eligible
professionals to use to submit quality data in 2008 would be published no later than
November 15, 2007. H.R. 6111 does not specify what bonus, if any, would be
obtainable in 2008.
Proposals that incorporate pay-for-performance components for reforming public
health care programs such as Medicare are likely to continue to come before Congress.
While the potential for improving the quality and efficiency of health care is
encouraging, the claims of improved quality with simultaneous cost savings will
undoubtedly be carefully weighed against the evidence of proven results that can be
generalized to larger populations and other settings and situations in the health care
industry.