Skip Navigation

R E C O M M E N D A T I O N 4 B
CMS should require hospitals to identify which
secondary diagnoses were present on admission on
their claims forms.

R A T I O N A L E 4 B
Currently, a diagnosis recorded on the discharge summary
that may have been present on admission cannot be
distinguished from one that developed during the hospital
stay. This additional information would significantly
enhance the ability to identify which complications are
avoidable. It would improve risk-adjustment of mortality
and complications measures. Several quality organizations
have supported this concept, and it should not significantly
increase hospital burden.

I M P L I C A T I O N S 4 B

Spending
• This recommendation would not affect federal
program spending relative to current law.
Beneficiary and provider
• This recommendation is expected to improve the
quality of care for beneficiaries.
• This recommendation is expected to result in some
increase in training for hospital coders.

When hospital coders are using the original history and
physical documentation to determine what diagnoses to
record on the claim (a task they must do anyway), they
also need to flag whether the diagnoses were present at
admission. California and New York already require
hospitals to report this information, and researchers have
found it very helpful for identifying patient characteristics
that may affect the likelihood they would die or
experience an adverse event.

The quality subcommittee of the National Committee on
Vital and Health Statistics, the Agency for Healthcare
Research and Quality, and the Consumer/Purchaser
Disclosure project have all supported including this type
of information in claims to better measure quality. In
addition, the National Uniform Billing Committee has
included a field for this information in the UB04 hospital
billing form.

Some have suggested that in addition to including these
types of complications as measures in a pay-forperformance
initiative, Medicare could also identify a
subset of events that should never happen (for example,
wrong site of surgery) and either deny payment or pay less
for care associated with the event. One health plan in
Minnesota has implemented this policy using data from a
state sentinel events reporting system. This
recommendation would also help Medicare identify those
events. MedPAC will continue to explore the feasibility of
identifying “never events” for purposes of revisions to
payment policy.

Structural measures
Measures of structures that ensure a provider is capable of
delivering high-quality care can apply to all types of
hospitals and patients. Assuring safety is one goal of these
types of measures. A survey designed to assess hospital
progress on implementing 27 of the NQF-endorsed safe
practices is used by large purchasers (Leapfrog Group
2004).5 Proponents of the survey and hospitals themselves
say that the survey creates opportunities for hospital
leaders and staff to discuss strategies and priorities for
decreasing medical errors and poor quality in their
hospitals. The Leapfrog Group worked with the Texas
Medical Institute of Technology (TMIT) to develop the
survey, and in its first year of use, more than a thousand
hospitals have assessed their internal systems. Hospitals
fill out the survey on a Web-based tool designed to score
the hospitals electronically. The TMIT has audited surveys
that appear as outliers, in which hospital scores are very
high, low, or are out of the normal range. It plans to
institute a more formal random audit process in the next
round of surveys, later in 2005.

The survey provides information on a variety of aspects of
hospitals’ operations, including simple ones like handwashing
practices, and more complex ones, such as
whether pharmacists are active in setting medication use
policies. The scores on the responses about the practices
are weighted to signal the relative importance and
comprehensive nature of each. For example, out of a
possible 1,000 points, ensuring that patient care
information and orders flow to all necessary providers is
worth 84 points, and hand-washing practices is worth 33
points. Other examples include (with associated points):
• verbal order readback (36)
• pharmacists active in medication use (32)
• pressure ulcer prevention (28)

page 191
-------------------------------------------
page 196

Where is quality measurement for home health heading?

The Robert Wood Johnson Foundation and AHRQ are
both funding work to guide efforts to enhance quality
measurement and improvement in home health (Rosati
2004). The potential to move forward on quality
measurement depends upon the development of clinically
tested, evidence-based best practices. MedPAC has
recommended such research in previous reports and
reiterates its importance elsewhere in this report. Work has
begun on gathering protocols and exploring their
applications in home health (Peterson 2004). One
potentially useful step would be to adopt a common
vocabulary to describe the processes of home health care.
Combined with the already widespread use of a common
patient assessment tool (OASIS), a common vocabulary
could help to focus providers’ efforts to improve and
stimulate the necessary research.

Home health provides an opportunity that policymakers
may wish to consider to take a step toward bridging the
setting-by-setting orientation of the current quality
measures. Policymakers could consider creating an
incentive for the physicians who are responsible for
reviewing, approving, and amending the plan of care for
home health care patients. This incentive would be similar
to the incentives for physicians who are responsible for
dialysis patients that MedPAC recommended in its
proposal to pay for performance in the end-stage renal
disease benefit (MedPAC 2004a). Physicians only use
three codes to bill Medicare (two for certification and
recertification of the plan of care and one for care
oversight); from these, a small pool could be formed and
redistributed to physicians whose patients achieve better
outcomes.

Home health as a setting is perhaps uniquely positioned to
take a larger step toward the quality of transitions or
“hand-offs” as patients move through the medical system.
Home health agencies are the front line for patients who
are making the difficult transition from the highly
structured environment of inpatient settings, such as a
hospital or rehabilitation facility, to their own home or
perhaps an assisted living facility. The quality of the
transition can improve a patient’s ability to recover from
an acute illness or injury or to prevent another
exacerbation of the condition (Forster et al. 2003).
Measures that transcend single settings would encourage
better management of patients as they move among
different sites of care.

Physicians
Physicians are central to the delivery of health care. They
evaluate and manage patients in their offices; decide when
hospitalization is necessary; perform surgery in hospitals
and ambulatory settings; prescribe drugs; and direct nurses
and others in nursing homes, home health agencies, and
dialysis facilities. The quality of the care they provide has
a tremendous effect on Medicare beneficiaries. Improving
Medicare quality will require their active participation.
Physicians are highly trained and knowledgeable
professionals who are expected to apply their training,
experience, and the most current research to decisions
regarding uniquely different individuals with serious
health problems. Without electronic means to store,
retrieve, and assist the physician in managing the
information regarding patients, this task is very difficult
(Crane and Raymond 2003, Bates and Gawande 2003).
MedPAC has stated that information technology is one of
the key organizational changes necessary to improve
quality (MedPAC 2003). However, the Medicare program
includes no incentive for physicians to adopt clinical IT.
To consider whether it is feasible to base a portion of
physician payment on quality, the Commission evaluated
the available measures and measurement activities for
physicians by our criteria and found useful structural,
process, and patient experience indicators. Outcomes
measures could be used with additional data and research.
The Commission concludes that it is feasible to base a
portion of physician payment on quality.

R E C O M M E N D A T I O N 4 E
The Congress should establish a quality incentive
payment policy for physicians in Medicare.

R A T I O N A L E 4 E
Physician participation is essential to improving quality.
Well-accepted measures of quality do exist, and the data
for many can be collected with minimal additional burden.
By focusing on measures of quality-enhancing functions
and outcomes associated with IT use, the quality
incentives in a pay-for-performance program could spur
physicians to adopt information technology that improves
care and helps build the infrastructure for further
assessment efforts. Condition-specific process measures
are also available, and those based on physician claims
would add no burden to physicians and apply to many
different conditions of importance to Medicare
beneficiaries.

I M P L I C A T I O N S 4 E
Spending
• Because this recommendation redistributes resources
already in the system, it would not affect federal
program spending relative to current law.
Beneficiary and provider
• This recommendation should improve the quality of
care for beneficiaries.
• This recommendation will result in higher or lower
payments for individual providers depending on the
quality of their care.

Which physician measures could be used?

The experts whom we consulted said that a wide variety of
measures exist for many types of physician specialties.
However, they also said that measuring physician quality
is more complex than measuring quality in other settings
because of the lack of sufficient data infrastructure, the
wide variety of often specialized services, and the sheer
number of physicians. Further, although Medicare requires
other providers to submit information on how they ensure
or improve quality, the primary data Medicare receives
from physicians are claims.

This lack of an infrastructure for measuring the quality of
physicians does not argue against a pay-for-performance
program. However, this program may require a transition
strategy because of these challenges.

Although some have argued that pay for performance
should be applied to only those types of physicians for
whom many measures are available, exempting some
physicians from the program would undermine the ability
to improve care for as many beneficiaries as possible.
Including all physicians will build the incentive for
different physician specialties to develop and improve
measures.

The Commission finds that two types of measures are
ready to be collected. The starter set of measures for
physicians reflects the need to balance two priorities:
building capacity and minimizing burden. First, we
recommend measuring quality-enhancing functions and
outcomes associated with information technology use,
such as whether a physician office tracks whether its
patients receive appropriate follow-up visits. These types
of measures apply to all types of physicians and address
important components of physician care—appropriate
monitoring, follow-up, and coordination of patient care.
Further, as physicians adopt IT in response, the capacity to
move toward more sophisticated and complete measure
sets will grow.

We also find that claims-based process measures provide
important information and are the least burdensome
approach to collecting condition-specific information.
Current research is finding that these measures are
available for a broad set of conditions of importance to
Medicare beneficiaries and some of them correlate well
with measures drawn from medical records. However, the
depth of information they provide on each individual
clinician is still the subject of research, as is the ability to
expand the set to measure an even greater number of
physicians. These measures will be greatly enhanced by
information on prescriptions and laboratory values, which
can be added over the next few years. Finally, we note that
patient experience measures will be available soon for
physicians and should be considered for this program.
Because these claims-based process measures do not
currently apply to every physician and we wish to ensure
that all physicians who see Medicare patients have the
incentive to improve, a transition strategy is necessary.
From the beginning of the program, CMS should collect
information on both structural measures—functions and
outcomes associated with IT use—and the claims-based,
condition-specific measures that are available, but it
should only base rewards on the IT structural measures.
The information on each physician’s performance on the
condition-specific measures could be given to the
physician for quality improvement purposes. To
encourage specialty societies to work with others to
continue to develop measures, CMS or the Congress
should establish a date certain, perhaps two to three years,
when the claims-based process measures would be
included in the pay-for-performance program.

Structural measures
To provide optimal care, physician offices need systems to
track numerous patient interactions over multiple settings
of care, pharmaceutical use, test results, and continually
evolving clinical guidelines. While tracking patients could
be achieved without information technology, it would be
far easier with IT. The ultimate goal is use of electronic
health records to improve quality. The Commission,
however, has concluded that it is important to reward
physician offices that have put systematic processes in
place to improve care management even with more limited
IT functions. This strategy would base payment on the
physician’s ability to produce information clearly related
to quality, rather than on the physician’s purchase of an IT
system. (We discuss the relationship in more detail in the
section of this chapter on IT.)

Measures of quality-enhancing activities associated with
IT use assess central functions of patient care: appropriate
monitoring, follow-up, and coordination. This is important
both for primary care physicians, who must manage
patients with chronic conditions, and for surgeons and
other specialists, who must follow patients after acute
events and coordinate care across settings. In addition to
the potential for improving care, encouraging physician
adoption of IT increases physicians’ ability to report on
quality and allows the pay-for-performance program to
apply to nearly all types of physicians.

This strategy will not require physicians to purchase fully
operational electronic health records. Various forms of
information technology enable these types of functions to
be performed much more easily (Table 4-4). For example,
NCQA finds that in its Physician Practice Connections
recognition program, some physician offices use their
patient management systems to satisfy the function of a
patient registry, while others receive credit through use of
an electronic health record. This flexibility makes it more
likely that all types of physician practices, large groups
and small offices, will participate in the program.

Data collection to report achievement on these types of
measures would add some burden to physician offices.
However, some physicians are already participating in a
recognition program that uses similar ones and CMS is
also planning to use them to measure physician quality.
The National Committee for Quality Assurance (NCQA)
has a recognition program that uses these types of
measures to encourage the adoption of IT and improve
patient care and has recognized 450 physicians in 38
practices. The Integrated Healthcare Association, a
California-based group of health plans, health systems,
and physician groups, and several other large purchasers
use these measures in its pay-for-performance programs.
Currently, physician offices applying for recognition
report data on their practices, including printouts of the
results on a Web-based data collection tool. For example,
if an office reports that it has a patient registry, it must
identify patients with different chronic conditions (the
function) and report whether the office sent reminders
prompting office visits or other necessary follow-up (the
outcome of the use of the registry). NCQA allows
physician offices to receive credit without actually using
IT, but reports that physician offices that use information
technology to perform the functions have a far easier time
complying.

Note: IT (information technology). Source: MedPAC analysis, using some examples from the National Committee for Quality Assurance Physician Practice Connections recognition program.

CMS is working with NCQA to revise this set to use in the
QIO 8th scope of work.10 The draft 8th scope of work
requires every QIO to work with physicians to adopt and
use electronic health records, electronic prescribing, and
reminders to better manage patient care on these and other
functions (CMS 2004a). In addition, CMS is planning to
use these types of measures in the Medicare Care
Management Performance Demonstration mandated by
the MMA to test pay-for-performance strategies for
physicians.

Two other structural measures—certification and
education—could eventually be part of a measure set, but
the link with improved care would need to be clear.
Certification measures could include whether a physician
was board certified in his or her specialty or other types of
certification or education that help keep physicians’
clinical knowledge current. Continuing education
measures could include physician participation in courses
on strategies for improving quality or enhancing physician
clinical knowledge.

Most hospitals, health plans, the JCAHO, and the NCQA
use board certification as one measure of physician
quality. However, the linkage with quality is unclear. A
recent systematic review found that more than half the
studies of this relationship showed an association between
board certification status and positive clinical outcomes
(Sharp et al. 2002). However, few published studies used
methods appropriate for the research question.

As of 2002, 85 percent of licensed physicians were board
certified (Brennan et al. 2004). Because so many
physicians are board certified, the American Board of
Medical Specialties (ABMS) has begun to revise its
process to better measure physician quality. Physicians
now must recertify periodically. In addition, several
member boards have begun to incorporate data about
current physicians’ performance into the recertification
process. The ABMS recently announced that all 24
specialty boards had agreed to develop a “maintenance of
certification” requirement, including measures of patient
care, practice-based learning, and interpersonal skills
(Romano 2004). Board certification could be part of a payfor-
performance program, but the specific requirements
need to be clearly linked with quality.

Condition-specific process measures
Process measures are the most widely used and accepted
for physician quality and apply broadly to different types
of conditions and physicians. Clinicians use these
measures to assess their performance and at the same time,
identify necessary improvements. For example, the
percentage of diabetic patients who have had their
hemoglobin A1c tests at appropriate intervals is a measure
of quality, but it also tells the physician what steps are
needed for improvement. While a wide variety of
physician measures are available from medical records,
flow sheets, or electronic health records, some are also
available through claims. Claims-based measures put little
new burden on physicians, and efforts are under way to
develop a broader set.

MedPAC is sensitive to the potential burden of data
collection. Therefore, while acknowledging the quality of
information collected from other sources, we conclude
that, at least initially, the program should use currently
collected data, such as claims and other administrative
data to derive condition-specific process measures. We
also recommend improving the information stream CMS
could use to link with claims data. (This is discussed in
more detail below.)

Although measures derived from physician claims are not
an extra burden for physicians, they do not provide as
detailed information as other data collection sources. For
example, a physician claim tells us whether a certain test
was performed, but information on the outcome of that test
resides in medical records.

Some researchers have tested whether the detailed
information derived from medical records provides a more
accurate picture of physician quality by observing the
correlation between rankings based on claims-based scores
and those based on medical record abstraction. Recent
unpublished research shows a strong correlation between
the relative rankings of physicians based on information
from claims and those based on information from medical
records for a set of conditions (Greenfield and Kaplan
2004). While this research focused on measures for a few
conditions, including diabetes and heart care, those
conditions affect many Medicare beneficiaries and,
therefore, the care of many types of physicians. RAND is
currently testing the ability to use claims-based process
measures on physicians in many different conditions of
importance to Medicare beneficiaries, including:
• asthma
• atrial fibrillation
• breast cancer