Copyright 1999 Federal News Service, Inc.
Federal News Service
FEBRUARY 25, 1999, THURSDAY
SECTION: IN THE NEWS
LENGTH:
4064 words
HEADLINE: PREPARED STATEMENT OF
WILLIAM
J. SCANLON
DIRECTOR, HEALTH FINANCING AND PUBLIC HEALTH
GAO
BEFORE
THE HOUSE COMMITTEE ON COMMERCE
SUBCOMMITTEE ON HEALTH AND
ENVIRONMENT
BODY:
Mr. Chairman and Members of
the Subcommittee:
We are pleased to be here today as you address the
question of adjusting Medicare's payments to managed care plans in the
Medicare+Choice program. Although the subject matter is technical, its
implications are significant for Medicare's greater use of managed care. The
Balanced Budget Act of 1997 (BBA) includes provisions designed to slow the
growth of Medicare payments overall. BBA also encourages the expansion of
managed care in its creation of Medicare+Choice, designed to offer beneficiaries
more health plan options beyond those available through Medicare's health
maintenance organizations (HMO). BBA provisions modify the method used to pay
health plans, and it is the details for implementing these provisions-
representing billions of dollars in savings-that are under discussion here
today.
Managed care plans receive from Medicare a fixed monthly payment,
called a capitation payment, for each beneficiary they enroll. Because the
payment is fixed per enrollee, regardless of what the plan spends for each
enrollee's care, health plans lack the incentive to provide unnecessary
services. However, the enrollment of beneficiaries in managed care plans has not
saved the government money as expected, mainly for two reasons. First, as we and
others previously determined, Medicare's capitation rates are excessive because
payments are based on health care spending for the average non-enrolled
beneficiary, while the plans' enrollees tend to be healthier than
average.Medicare HMOs: HCFA Can Promptly Eliminate Hundreds of Millions in
Excess Payments (GAO/HEHS-97-16, Apr. 25, 1997). Second, instead of diminishing
as more beneficiaries enrolled in managed care, excess payments per enrollee
continued to grow. To correct these problems, BBA changed the rate-setting
formula used by the Health Care Financing Administration (HCFA), the agency
responsible for administering Medicare. It required that most of the
rate-setting provisions be in place in 1998 and required that HCFA replace
Medicare's current risk adjuster-the mechanism that modifies a plan's average
capitation rate to better reflect an enrollee's expected medical costs-with a
new one to be implemented in 2000. The risk adjuster in place has been widely
criticized as a major factor in the HMO overpayment problem.
In considering
Medicare's new rate-setting method, my comments today will focus on (1) the
importance of improving the current risk adjustment method, (2)
the implications of rate-setting changes implemented in 1998, and (3) the
advantages and drawbacks of HCFA's proposed new interim risk adjuster. My
comments are based on information drawn from our issued work on this subject,
supplemented by relevant published studies and interviews with HCFA officials.
In summary, Medicare's current risk adjuster has failed to protect
taxpayers, certain plans, and beneficiaries, underscoring the urgency of
replacing it with a health-based risk adjuster.
Studies by us and others
show that methodological flaws have led to billions of dollars in excess
payments and inappropriate payment disparities.
BBA provisions now in place
may reduce, but not eliminate, excess payments; and payment disparities persist
that could jeopardize plan participation and access to managed care for costlier
seniors.
The new risk adjuster required to be in place by 2000 is intended
to improve estimates of health plan enrollees' medical costs. Better cost
estimates producing fairer rates could reduce the unnecessary spending of
taxpayer dollars while minimizing the financial disincentive for plans to serve
a costly mix of beneficiaries.
The use of the new risk adjuster, while not
perfect, is an interim step and improves on the one now in place. In addition,
HCFA plans to phase in the use of the new adjuster, thereby recognizing the need
to avoid sharp payment changes that could affect plans' offerings and diminish
the attractiveness of the Medicare+Choice program to beneficiaries.
BACKGROUND
The long-term financial condition of Medicare is now one of
the nation's most pressing problems. As the nation's largest health insurance
program, Medicare's size and impact on all Americans is significant. The program
covers about 39 million elderly and disabled beneficiaries at a cost of more
than $193 billion in fiscal year 1998. About 83 percent of the program's
beneficiaries receive health care on a fee-for-service (FFS) basis, in which
providers are reimbursed for each covered service they deliver to beneficiaries.
The rest, about 6.8 million people, are provided care through more than 450
managed care plans, as of December 1, 1998.About 90 percent of the 6.8 million
Medicare beneficiaries are enrolled in managed care plans that receive fixed
monthly capitation payments. The remainder are enrolled in plans that are
reimbursed for the costs they incur, less the estimated value of beneficiary
cost-sharing.
To extend the solvency of Medicare's Hospital Insurance Trust
fund beyond 2008, BBA provided for substantial reforms in both the FFS and
managed care components of Medicare. BBA provisions are expected to achieve
estimated Medicare savings that reduce the program's average annual growth rate
by more than 3 percent, representing over $100 billion over 5 years.
One way
in which BBA seeks to restructure Medicare is to encourage greater participation
in Medicare+Choice. Under this program, BBA permits the creation of new types of
Medicare health plans, such as preferred provider organizations and
provider-sponsored organizations. BBA's emphasis on Medicare+Choice reflects the
perspective that increased managed care enrollment will help slow Medicare
spending while expanding beneficiaries' options in choosing health plans.
BBA also sought to improve the method for setting managed care plans'
payment rates. In general terms, the pre-BBA rate-setting methodology worked as
follows. Every year, HCFA estimated how much it would spend in each U.S. county
to serve the "average" FFS beneficiary. It would then discount that amount by 5
percent under the assumption that the managed care plans provided care more
efficiently than the unmanaged FFS system. The resulting amount constituted a
base county rate to be paid to the plans operating in that county. Because some
beneficiaries were expected to require more health services than others, HCFA
"risk adjusted" the base rate up or down for each beneficiary, depending on
certain beneficiary characteristics-specifically, age; sex; eligibility for
Medicaid; employment status; and residence in an institution, such as a skilled
nursing facility.Separate rates, using the same demographic traits, are
calculated for beneficiaries who qualify for Medicare because of a disability
(under age 65). Separate rates are also set for beneficiaries with end-stage
renal disease (kidney failure).
BBA's new payment rate method seeks to
address the two main factors contributing to excess payments: (1) the disparity
in expected health costs between Medicare's FFS and managed care populations
built into each county's base capitation rates and (2) the failure of the risk
adjuster to correct for that disparity on an individual enrollee level.
BBA required that a county's capitation rate equal the highest of a
blended capitation rate, which reflects a combination of local and national
average FFS spending from 1997, updated for increases in national spending;
the previous year's county rate increased by 2 percent; or
a minimum
payment amount, called a floor, set equal to $367 in 1998 and updated each year.
Loosening the link between the current cost of Medicare's FFS population and
counties' base rates helps prevent the excess payments from continuing to
increase as more beneficiaries join managed care plans. BBA also acknowledges
the need for individual enrollee adjustments by requiring the development of a
risk adjustment method based on health status. The law requires
that HCFA develop and report on the new risk adjuster by March 1 of this year
and the method be in place by January 2000.Technically, the law requires the
Secretary of the Health and Human Services to develop, report, and implement the
health-based risk adjustment method.
MEDICARE'S CURRENT
RISK ADJUSTMENT METHOD FAILS TO PREVENT OVERPAYMENTS AND
APPROPRIATELY TARGET PAYMENTS TO PLANS
Risk adjustment is a
tool to set capitation rates so that they reflect enrollees' expected health
costs as accurately as possible. This tool is particularly important given
Medicare's growing use of managed care and the phenomenon of favorable
selection-the tendency of managed care plans to attract a population of Medicare
seniors whose health costs are generally lower than those of the average program
beneficiary. Our 1997 study on payments to California HMOs, which enrolled more
than a third of Medicare's managed care population, found that Medicare overpaid
plans by about 16 percent because HMO enrollees had costs that were lower than
the average beneficiary's.GAO-HEHS-97-16, Apr. 25, 1997. This is consistent with
a 1996 study by HCFA researchers finding that health plan enrollees had costs
estimated at 12 to 14 percent below the average beneficiary's. (Riley and
others, HCFA Review, 1996.)
Medicare's current risk adjuster cannot
sufficiently lower rates to be consistent with the expected costs of managed
care's healthier population. The reason is that Medicare's risk adjuster relies
on demographic factors such as age and sex, which alone are poor predictors of
an individual's health care costs. For example, two beneficiaries can be
demographically identical (same age and sex), but one may experience occasional
minor ailments while the other may suffer from a serious chronic condition.
Without the use of health status factors to make that distinction, Medicare's
risk adjuster produces excessive payments in compensating plans for their
relatively lower cost enrollees.
The financial consequences of a poor risk
adjuster are huge. In our 1997 study of California's payment rates, we estimated
that Medicare paid about $1 billion in excess to health plans operating in the
California in 1995. Shortly before we issued our report, the Physician Payment
Review Commission (PPRC), now a part of the Medicare Payment Advisory
Commission, estimated that annual excess payments to Medicare HMOs nationwide
could total $2 billion.
Some analysts have speculated that, with growing
enrollment, health plans would necessarily enroll a substantially larger share
of less healthy beneficiaries, which would raise plans' costs and reduce
Medicare's excess payments. Our 1997 analysis, however, showed that- rather than
shrinking excess payments-the rapid growth in Medicare managed care enrollment
actually exacerbated the situation. The counties with higher managed care
enrollment had higher, not lower, excess payments. Data indicated that the
sickest beneficiaries tended to remain in FFS while the healthier beneficiaries
joined managed care plans. Excess payments grew with managed care enrollment
partly because HCFA based the payment rates on average FFS spending, which
increased as the pool of FFS beneficiaries shrank and, as a group, became less
healthy.
Better risk adjustment is also important for plans
that may not be adequately compensated for serving higher cost beneficiaries who
enroll. Having enrollees who are sicker than the average mix of Medicare
beneficiaries can alter a plan's costs significantly. About 10 percent of
Medicare beneficiaries account for 60 percent of Medicare's annual expenditures.
Without adequate risk adjustment, plans with more than their
share of the costly beneficiaries are at a competitive disadvantage.
BBA
PROVISIONS MAY REDUCE OVERPAYMENTS, BUT SUBSTANTIAL EXCESS LIKELY REMAINS
BBA contains several provisions, implemented in 1998, that are designed to
improve Medicare's rate-setting method. Certain provisions seek to reduce excess
payments and inappropriate geographic disparities. These changes represent steps
in the right direction but do not eliminate the need for a health-based risk
adjuster. Substantial excess payments likely persist, in part, because other BBA
provisions tended to incorporate the excess that existed in 1997 into the
current rates.
Certain BBA Provisions May Reduce Excess Payments but Are Not
Substitutes for Improved Risk Adjustment BBA aims to reduce the
excess in Medicare's managed care payments in two ways. First, BBA holds down
managed care per capita spending increases for 5 years. Specifically, BBA sets
the factor used to update managed care payment rates equal to national per
capita Medicare growth minus a specified percent: 0.8 percent in 1998 and 0.5
percent in each of the following 4 years.
BBA also provides for a
methodological approach known as "blending," which may help reduce excess
payments. The blended rate set for each county combines that county's 1997 rate,
updated for increases in national Medicare spending, and a national average. The
blending formula is currently weighted heavily toward local rates but will
gradually change so that local and national rates will be weighted equally in
2003. Over time, blending will reduce the substantial variation in county
payment rates that now exist. For, example, county rates ranged from a low of
$380 to a high of $798 in 1999. Because of BBA-mandated budget neutrality and
minimum payment constraints, no county received a blended rate in 1998 or 1999.
Blending is expected to occur for the first time in 2000.
Blending may help
reduce excess payments because high-rate counties (where excess payments are
estimated to be concentrated) will receive smaller annual increases relative to
low-rate counties. Evidence on the relationship between county payment rates and
excess payments is provided in a 1997 PPRC study. PPRC reported that county
payment rates tend to overestimate beneficiaries' health care costs in
high-payment- rate areas and underestimate their costs in low-payment-rate
areas.Physician Payment Review Commission, 1997 Annual Report to the Congress.
PPRC found that a comprehensive health-based risk adjustment
methodology would have lowered, for example, the average Miami-area payment rate
from $616 to $460 in 1995. The same methodology would have raised the average
payment rate in rural Minnesota from $263 to $310.
Blending is a rather
blunt tool for addressing the excess payment problem, however, and does not
obviate the need for improved risk adjustment. As the PPRC
results indicate, not all high-rate counties have rates that are too high and
not all low-rate counties have rates that are too low. For example, PPRC's
risk-adjustment methodology would have reduced the average
payments in rural Michigan (a relatively low-payment-rate area) from $346 to
$334. Furthermore, not all plans in high-rate counties may receive excess
payments. Because payment rates are based on the expected costs of beneficiaries
in average health, plans that attract costly beneficiaries may be underpaid by
the current risk adjustment method.
Some BBA Provisions
Have Tended to Incorporate Excess Payments From 1997 Into Current Rate Structure
BBA specified that 1997 county rates be used as the basis for all future
county rates beginning in 1998. Although the law changed many aspects of the
rate-setting formula, this BBA provision had the effect of incorporating the
excess payments that existed in 1997 into all future rates.
As we testified
before this Subcommittee in February 1997, HCFA's then current rate-setting
methodology resulted in county rates that were generally too high. Simply put,
instead of setting rates based on the expected cost of the average beneficiary
in each county, the agency set rates based on the expected costs of serving FFS
beneficiaries. If the agency had included the expected costs of serving managed
care beneficiaries-who as a group tend to be healthier than FFS
beneficiaries-the overall county average would have been lower.
About
one-quarter of the $1 billion in overpayments we estimated in our California
study resulted from flaws in developing the county rate.
Excess payments are
also built into current rates because BBA did not allow HCFA to adjust 1997
county rates for previous forecast errors-a critical component of the
rate-setting process. Although the process for setting rates was extremely
complex and involved separate adjustments for each county, annual payment rate
updating was straightforward. Each fall, HCFA would forecast total Medicare
spending for the following year; the estimated percentage spending increase,
from the current year to the following year, was used to update the county
rates. Before applying the increase, however, HCFA corrected any forecast errors
from previous years. If HCFA discovered that previous forecasts had
overestimated or underestimated the current spending, the update was
appropriately adjusted.
HCFA actuaries now estimate, based on FFS claims
data, that the 1997 managed care rates were too high by 4.2 percent. BBA, in
establishing a new methodology for setting rates in 1998 and future years,
specified that HCFA use the 1997 rates as the basis for the new rates. While the
law permits HCFA to correct forecasts in future years, it did not include a
provision that would have allowed HCFA to correct its forecast for 1997.
Consequently, about $1.3 billion in overpayments were built into plans' annual
payment rates beginning in 1998.
HCFA'S PROPOSED RISK
ADJUSTMENT APPROACH IMPROVES ON CURRENT METHOD AND MINIMIZES DISRUPTION
FOR PLANS AND BENEFICIARIES
HCFA's proposed interim health-based
risk adjustment method-to be implemented in 2000-represents a
major improvement over the current method. For the first time, Medicare managed
care plans can expect to be paid more for serving beneficiaries with serious
health problems and less for serving relatively healthy ones. The interim method
relies exclusively on hospital inpatient data to measure health status. Although
it would be better to measure health status with complete and reliable data from
other settings, such as physicians' offices, these data are not yet available.
In addition, HCFA's decision to phase in the new method will likely minimize
disruptive plan pull-outs and altered benefit packages, which could occur if
payment rate changes were implemented too suddenly. Proposed Risk
Adjustment Method Based on Available Hospital Inpatient Data
The
proposed method, known as the Principal Inpatient Diagnostic Cost Group
(PIP-DCG) method, would use hospital inpatient data to more accurately match
managed care payments to beneficiaries' expected total Medicare costs. PIP-DCG
would assign each individual to 1 of 15 categories if during the prior year they
had been hospitalized for certain diagnoses. For example, a beneficiary who had
been hospitalized for congestive heart failure would be placed in one category,
while a beneficiary who had been hospitalized for a kidney infection would be
placed in another. Those beneficiaries who were not hospitalized and those who
were hospitalized for diagnoses not included in PIP-DCG-about 88 percent of all
beneficiaries-would be placed in the base category. The next year's payment rate
for each enrollee would be determined by the category the individual was placed
in and by certain demographic data, such as age and sex. Rates for enrollees
placed in one of the 15 prior hospitalization groups would be higher than rates
for those in the base category with the same demographic characteristics.
HCFA anticipated potential concerns about a risk adjustment
methodology based on hospital inpatient data. Such an approach could reward
plans that hospitalize patients unnecessarily or, conversely, penalize efficient
plans that provide care in other, less costly settings. HCFA has attempted to
address these concerns in several ways.
First, PIP-DCG would assign
individuals to prior hospitalization categories only when the diagnosis is for a
condition that normally requires hospitalization and is linked to further
medical costs in the following year. To determine which specific diagnoses to
include, HCFA relied on the advice of a clinical panel. The panel recommended
that diagnoses associated with about one-third of hospital admissions be
excluded because they (1) could be ambiguous, (2) were for conditions that were
rarely the main cause for an inpatient stay, or (3) were not good predictors of
future health care costs. For example, a beneficiary hospitalized for
appendicitis would not be assigned to a higher cost category because that
condition generally is not linked to further medical costs in the next year.
Also, HCFA's proposal does not permit enhanced payments for hospital diagnoses
associated with 1-day stays. These admissions may be more discretionary than
admissions for longer stays.
Second, delaying an adjustment in payment until
the following year discourages unnecessary hospitalizations that would trigger
an enhanced payment. Further, the payment delay dampens any incentive to
encourage higher cost enrollees who have been hospitalized to switch plans,
since the plan in which the beneficiary is a member the following year receives
the payment.
The PIP-DCG method assumes that admission rates for
beneficiaries of similar health status are the same for FFS and managed care
providers. Although the evidence on managed care admission rates is limited,
findings presented by the American Association of Health Plans last month
support this hypothesis. A study conducted for the Association found that
hospital admission rates for managed care plans and FFS plans were comparable.
These findings are consistent with those of a 1993 Mathematica Policy Research
study on hospital admissions rates.
Gradual Implementation of Interim Method
Will Minimize Impact on Health Plans and Beneficiaries
HCFA proposes to
phase in the new interim risk adjustment method slowly. In
2000, only 10 percent of health plans' payments will be based on the new system.
This percentage will be increased each year until 2003, when 80 percent of
plans' payments will be based on the PIP-DCG risk-adjusted rate. In 2004, HCFA
intends to implement a more accurate risk adjuster that uses medical data from
physicians' offices, skilled nursing facilities, home health agencies, and other
health care settings and providers--in addition to inpatient hospital data.
Although a gradual phase-in of the interim risk adjuster delays the full
realization of Medicare savings, it also minimizes potential disruptions for
both health plans and beneficiaries. Rapid payment rate changes could strain the
financial soundness of some plans. Rapid rate changes could also adversely
affect beneficiaries if plans respond by suddenly altering their benefit
packages or reconsidering their commitment to the Medicare+Choice program.
If HCFA had comprehensive patient-level data from Medicare managed care
plans, it could adjust the PIP-DCG methodology to reflect any differences in
practice patterns between managed care and FFS providers. Although plans
currently are required to submit only hospital inpatient data, the agency
intends to begin collecting more comprehensive data shortly. Therefore, it may
be possible to refine the PIP-DCG methodology before the implementation of the
full risk adjustment in 2004.
Conclusions
The
implementation of a new health-based risk adjustment system
will lead to major changes in Medicare managed care payments and will create
more desirable incentives. Plans attracting healthier beneficiaries will be paid
less, whereas those attracting costlier beneficiaries will be paid more. In more
fairly compensating individual plans for the beneficiaries they enroll, the new
method will reduce excess payments and produce savings for taxpayers. The new
method represents an interim step in the use of health-based risk
adjustment. We believe that to facilitate the introduction of an
improved risk adjuster in 2004, plans should aggressively pursue the collection
and reporting of more comprehensive data on beneficiaries' medical conditions. -
- - - - Mr. Chairman, this concludes my statement. I will be happy to answer any
questions you or other Members of the Subcommittee may have.
END
LOAD-DATE: February 27, 1999