Copyright 1999 Federal Document Clearing House, Inc.
Federal Document Clearing House Congressional Testimony
June 09, 1999
SECTION: CAPITOL HILL HEARING TESTIMONY
LENGTH: 5026 words
HEADLINE:
TESTIMONY June 09, 1999 STEVEN M. LIEBERMAN EXECUTIVE ASSOCIATE DIRECTOR
SENATE FINANCE HEALTH CARE OVERSIGHT OF RISK ADJUSTMENT
METHODOLOGY AND OTHER IMPLEMENTATION ISSUES
BODY:
Statement of Steven M. Lieberman Executive Associate Director Office of the
Director Congressional Budget Office on Medicare+Choice before the Committee on
Finance United States Senate June 9, 1999 Mr. Chairman, Senator Moynihan, and
Members of the Committee, it s a pleasure to appear before you today to discuss
the enrollment and payment issues confronting the Medicare+Choice program. The
growth in that program s enrollment is closely linked to the adequacy and
appropriateness of Medicare s capitated payments. The recent withdrawal of plans
from Medicare+Choice, coupled with reduced growth in payments, has prompted some
observers to worry about the future of the Medicare+Choice program. My testimony
discusses the Congressional Budget Office s (CBO s) projection of enrollment in
Medicare+Choice plans over the next 10 years and the factors influencing growth
in that enrollment. Financial incentives play a critical role in determining
whether plans participate in Medicare+Choice, whether beneficiaries enroll, and
whether providers deliver appropriate services in an efficient manner. For
Medicare+Choice to be a viable program, beneficiaries must have incentives to
relinquish traditional fee-for-service and enroll instead in competing health
plans. The challenge is to have a system that yields greater returns when it
efficiently provides necessary, high-quality services and smaller returns when
it provides inefficient, low-quality, or unnecessary services. Meeting that
challenge requires that plans, providers, and beneficiaries each bear some
degree of financial risk. Serious problems can result if Medicare payments do
not bear a reasonable relationship to the costs of care for each group of
beneficiaries for which plans and providers accept risk. Payments to providers
must be fair and, ideally, give incentives to control costs while rewarding
quality. If consumers have a choice of health plans offering various
combinations of benefits and premiums, they can select the plan that best meets
their needs. Enrollment in Medicare+Choice plans would grow if those plans
offered better benefits or lower costs than traditional Medicare. If consumers
have no choice of plans or if those plans offer unattractive benefits, high
costs, or poor quality, beneficiaries will remain in fee-for-service Medicare.
ENROLLMENT IN THE MEDICARE +CHOICE PROGRAM CBO projects that growth in
Medicare+Choice enrollment will average 9 percent annually between 1999 and
2009. Though quite rapid, that rate of increase represents a sharp reduction
from earlier trends. The Balanced Budget Act of 1997 (BBA) established
Medicare+Choice and changed payment provisions for both health maintenance
organizations (HMOs) and fee-for-service providers. CBO had assumed that
Medicare+Choice enrollment would continue to grow at the dramatic rates of the
program it replaced. The annual rate of growth in enrollment in Medicare s
risk-based plans peaked at 36 percent in fiscal year 1996, however, and slowed
in subsequent years. CBO projects that 31 percent of all Medicare beneficiaries
will join Medicare+Choice plans in 2009, up from 16 percent this year (see Table
1). HMO Withdrawals Last year, 99 HMOs announced they were either terminating
or, far more commonly, scaling back their Medicare+Choice operations in certain
counties. The potential disruption involved 407,000 enrollees, accounting for 7
percent of all Medicare+Choice enrollment. Plan withdrawals occurred in 406
countiesC42 percent of the counties covered by Medicare managed care.
Nonetheless, the overwhelming majority of the affected beneficiaries had the
option to switch to a competing Medicare+Choice plan. The unanticipated
withdrawal of plans from the Medicare market has heightened awareness that plans
can leave the market. That perception is likely to reduce the willingness of
some Medicare beneficiaries to enroll in plans in the next few years. Although
the effects of plans withdrawal on Medicare+Choice enrollment seem relatively
clear, explaining why plans withdrew appears more controversial. In a recent
report, the General Accounting Office concluded that most likely more than one
factor was responsible for the withdrawals. No one factor can explain why plans
choose to participate in particular counties. Although plans obviously consider
payment rates, many other factors also influence their business decisions. The
current movement of plans in and out of Medicare may be primarily the normal
reaction of plans to market competition and conditions. . . . Other factors
associated with plan withdrawalsCrecent entry in the county, low enrollment, and
higher levels of competitionCsuggest that a number of Medicare plans withdrew
from markets in which they had difficulty competing. By contrast, the HMO trade
group, the American Association of Health Plans (AAHP), attributes the
withdrawals to inadequate payment rates, exacerbated by the administrative
burdens imposed by the Health Care Financing Administration s (HCFA s) AMegaReg
for implementing the BBA s provisions. AAHP believes that without substantial
revisions to Medicare+Choice, additional plans will withdraw from the program.
Adverse publicity associated with the health plans withdrawal from
Medicare+Choice is likely to temporarily slow growth in enrollment. But over the
longer term, that growth depends critically on the size of payment increases and
the ability of plans to offer attractive additional benefits, such as
prescription drugs. Constraining Medicare+Choice Payments Health plans, as
businesses, will participate in Medicare+Choice markets only if they have an
expectation of an adequate returnCat a minimum, if they can reasonably expect at
least to cover costs. If payments are perceived as being inadequate, health
plans will tend not to participate in Medicare+Choice, especially if they
foresee little prospect of Medicare payments becoming adequate. A similar
dynamic applies to providers. Regardless of mission or not-for-profit status,
physicians and other providers cannot afford to participate indefinitely when
their enterprises are losing money. In addition to causing plans to withdraw,
inadequate Medicare+Choice payments have another, compounding effect on
enrollment growth. Reducing payment increases to Medicare+Choice plans will
impede their ability to offer extra benefits or limit beneficiary cost sharing.
Taking steps such as eliminating prescription drug benefits or requiring hefty
monthly premiums instead of Azero premiums will make Medicare+Choice plans less
attractive to consumers. As a result, fewer beneficiaries will choose to join
those plans. Are Medicare+Choice payments inadequate? The adequacy of payments
can be evaluated from five often-competing perspectives. - Are plans able to
provide appropriate services while remaining financially stable? - Are payments
fair, permitting (if not encouraging) plans and providers to serve sicker
patients? - Is there an adequate choice of health plans in both urban and rural
parts of the country? - Do the payments offered by Medicare+Choice plans attract
physicians, hospitals, and other providers to participate in their networks? -
Do the payments help keep Medicare affordable for both beneficiaries and
taxpayers? Having well-established plans A vote with their feet and withdraw
from their key Medicare+Choice markets is an indication that payment and other
conditions of participating in Medicare+Choice may be too stringent. But health
plans have powerful incentives to convince policymakers that Medicare+Choice
payments need to be increased without having to withdraw from the program.
CHANGES TO MEDICARE +CHOICE PAYMENTS UNDER THE BALANCED BUDGET ACT The BBA
enacted six policies that affected Medicare+Choice payments. - The BBA
significantly reduces fee-for-service spending, which also slows the growth of
payments to health plans because annual updates to Medicare+Choice payment rates
are tied to the growth in per-enrollee spending in the traditional Medicare
program. - The BBA sets the annual increases in Medicare+Choice payment rates
below the growth in fee-for-service spending from 1998 through 2002. - The
portion of Medicare+Choice payment rates that is attributable to fee-for-service
spending for graduate medical education will be gradually
eliminated. - HCFA will withhold about 0.2 percent of payments to
Medicare+Choice plans to pay for dissemination of information to beneficiaries
about their coverage options. - A blend of local and national payment rates will
be phased in for Medicare+Choice plans. That blending provision redistributes
money from areas with high payment rates to those with low payment rates. - New
payment risk adjusters will be implemented in two stages. Those adjusters are
intended to more accurately reflect the expected costs of providing health care
to enrollees in Medicare+Choice plans. The first four policies were enacted with
the expectation that they would slow the growth of Medicare spending. Those
policies reduce the cumulative growth in Medicare+Choice payment rates relative
to fee-for-service payments by 6 percent. The blending of local and national
payment rates is purely redistributive, but particular counties will see
substantial changes in payment rates. The new risk adjusters were not
necessarily expected to lower average payments to Medicare+Choice plans but, as
discussed below, they could yield substantial program savings when they are
implemented. Impact of the Payment Blend Because of the blending of national and
local payment rates, payment increases are projected to vary enormously from
county to county. For example, some counties would experience such large
increases in payment rates from 1997 to 2000 that the theoretically available
Medicare+Choice payment rates C if any plans operated in the areas C would
exceed 180 percent of the 1997 (pre-BBA) payment rates. In contrast, some
counties with high payment rates would see only a 6.1 percent increase in their
rates over the same period. Historically, both the level of and increase in
Medicare spending per beneficiary varied dramatically in different counties.
HCFA, however, no longer produces those data on county-specific spending trends.
If past trends continue, some Medicare+Choice plans will face payment rates that
are projected to be substantially below both per capita fee-for-service spending
and 1997 (pre-BBA) amounts. Over half (52) of the 100 counties with the most
Medicare+Choice enrollees are projected to have payment rates fall by 5 percent
or more using as the standard of comparison the rates that Medicare would have
paid if 1997 payments were increased by the national average growth in per
capita fee-for-service spending and the BBA payment provisions were fully in
effect. Using that methodology, the steepest reduction is estimated to be 12
percent. In the top 100 counties, 88Chome to 78 percent of the enrolleesCwould
experience declines in payment rates, compared with 1997 rates. These estimates
do not include the lower payments resulting from HCFA s implementation of risk
adjustment. Impact of Risk Adjustment Until 1999, CBO had assumed that
Medicare+Choice payments would be adjusted for risk without changing total
outlays. In January, the Administration published plans to phase in risk
adjustment in a manner that would reduce payment rates for enrollees in
Medicare+Choice plans. The first stage of risk adjustment would be based on the
use of inpatient hospital services by individual enrollees. That change would
reduce payments for existing enrollees by 7.6 percent when fully phased inC by
2004. The Administration also announced a second stage of risk adjustment that
would be based on use of services in all settings. The Administration expects
that such an adjustment would reduce payments by another 7.5 percent, beginning
in 2004. If both plans are implemented as announced, the combined effect could
reduce payments by about 15 percent. Payment reductions related to risk
adjustment on the order of 15 percent would be likely to cause plans to drop out
of the program and enrollment in Medicare+Choice to drop sharply. Because of the
magnitude of the planned reduction and the discretion retained by the
Administration in implementing the adjusters, the CBO baseline does not assume
the full savings from risk adjustment. For the same reason, the projections of
Medicare+Choice enrollment discussed in my testimony today explicitly do not
reflect the full savings. Instead, CBO assumes that risk adjustments will
ultimately reduce payments by lesser amounts. RISK SELECTION AND RISK ADJUSTMENT
Risk selection occurs when groups of beneficiaries, such as those who enroll in
a Medicare+Choice plan, have average costs that are systematically different
from the average costs of beneficiaries who are treated as similar by the risk
adjuster. When monthly payments are made on a fixed, prospective (or capitated)
basis, those groups of enrollees are referred to as Arisk pools. If
Medicare+Choice enrollees tend to have lower costs than comparable
fee-for-service beneficiaries, the result is known as A favorable risk
selection. Conversely, A adverse risk selection occurs when groups or risk pools
have costs that are higher than those of comparable fee-for-service
beneficiaries. Risk selection is incompletely understood and imperfectly
measured. It can arise from many different sources. If unchecked, risk selection
can destroy an insurance system. Systematically selecting people who are
healthier than average pays off handsomely: the returns on favorable selection
can overwhelm any potential savings from operating an efficient system for
managing care. Health insurance systems in which biased selection segments the
risk pool are said to enter a Adeath spiral if the problem is not fixed. One
goal of risk adjustment is to pay more fairly. In a fair system, the amounts
paid for different risk pools would closely approximate the average cost of
providing services to their members. Under that framework, a good risk adjuster
would pay groups with sicker, more expensive people proportionately more and
groups with healthier, less expensive beneficiaries proportionately less.
Medicare+Choice Risk Adjuster There are a wide variety of potential approaches
to mitigating the effects of risk selection. HCFA has adopted a mechanism for
risk adjustment that relies on inpatient hospital admissions for specific
diagnoses to trigger higher capitated payments in the following year. That
mechanism, which is known as the principal in-patient/diagnostic cost group (or
PIP/DCG), attempts to adjust payments statistically to account for individuals
with persistently high costs. On average, PIP/DCGs would reduce payments
somewhat for most beneficiaries but increase them significantly for the minority
of beneficiaries who were hospitalized in the prior year for specific conditions
(such as congestive heart failure). HCFA has had to overcome significant
analytical and operational obstacles in setting up the PIP/DCG system. The
agency appears to be successfully implementing that complex system, for which it
deserves recognition. But it is important to understand the limitations of that
system for adjusting payments. Developing a Medicare Risk Adjuster Although the
PIP/DCG system is a significant improvement over demographic adjusters, it has
had limited success in achieving the goal of A fair payments C payments that are
closely related to the costliness of beneficiaries (based on their health
status). Two factors contribute to the difficulty of developing an adequate
Medicare risk adjuster. First, the health care costs for individuals are
enormously difficult to predict. That difficulty is compounded when the
predictions are based on the administrative data available from processing
claims. Second, Medicare spending is extremely skewed C that is, the sickest
beneficiaries are extraordinarily costly. The most expensive 5 percent of
Medicare beneficiaries cost almost as much as the remaining 95 percent of all
Medicare beneficiaries. On average, those in the top 5 percent cost over $70,000
annually C more than 10 times the average annual cost for all Medicare
beneficiaries. The variation in cost per beneficiary has two critically
important implications. On the one hand, it highlights the potential financial
consequences associated with both risk selection and inadequate risk adjustment.
On the other hand, assuming neutral risk selection C that a risk pool has an A
average population C the skewness of the distribution of costs may require
relatively large numbers of participants for a risk pool to be stable. Very
large risk pools are unlikely to be undermined by having one too manyCor too few
C million-dollar cases in a year. Small risk pools, however, could be seriously
disrupted by having just one person who incurs catastrophic health care costs.
Large health plans may be able to assume full financial risk for their
enrollees. Even without risk selection, small plans may not be well positioned
to assume full financial risk. In many large Medicare+Choice markets, health
plans base payments to physicians or other providers on a percentage of
premiums, thereby passing risk on to the providers. These compensation
arrangements do not directly connect HCFA to provider payments. Yet HCFA remains
vitally involved for two reasons. First, HCFA regulates the terms and conditions
under which physicians may be placed at substantial financial risk, approving
their contracts with Medicare+Choice plans. Second, HCFA has a vital interest in
and regulatory responsibility for assuring that beneficiaries have adequate
access to sufficient providers and receive high-quality care. The numerous
Medicare+Choice providers who are paid on a capitated, percentage-of-premium
basis subdivide a health plan s risk pool. As a result, even relatively large
risk pools at the health plan level may become too small at the provider level.
PIP/DCGs may not be a desirable system for adjusting payments to small risk
pools. Problems with Using an Inpatient Risk Adjuster The first phase of the
PIP/DCG relies solely on inpatient hospital admissions and excludes care
delivered in other settings. One can argue that the reliance on inpatient
hospital admissions hurts managed care plans, many of which have reduced their
use of inpatient hospital services. Some plans have implemented effective
disease management and other protocols that may alter the pattern of care,
possibly minimizing the specific admissions that are rewarded by the PIP/DCG
methodology. What are the implications of the inpatient PIP/DCG payment system
for a Medicare+Choice plan that has invested in developing sophisticated disease
management systems for chronic conditions? Unlike acute episodes of care,
chronic conditions, such as congestive heart failure, can frequently have high
and recurring costs. Paradoxically, that makes such conditions ideal for both
disease management interventions and for creating a PIP/DCG payment adjustment.
With chronic conditions, an HMO can identify who is at risk and develop
intervention strategies to improve outcomes. Typically, successful interventions
stress prevention, investing in patients education, and gaining their compliance
with protocols. Although such strategies do not Acure chronic conditions, they
improve patients outcomes and frequently save money by avoiding
hospitalizations. Success in avoiding hospitalizations, however, means that the
Medicare+Choice payment rate is never increased to compensate for the
beneficiary with high-cost, chronic conditions. Without a hospitalization for
congestive heart failure, for example, the PIP/DCG system does not recognize
that the beneficiary has the condition. Is this A Catch 22 real? Preliminary
findings from an analysis being conducted by John Bertko, a principal in the
actuarial consulting firm of Redden & Anders, provide some guidance. A
highly sophisticated Medicare+Choice plan appears to have implemented effective
disease management protocols for several conditions, including congestive heart
failure. By investing about $3,000 annually in each patient, that HMO has
apparently managed to avoid about half the expected hospital inpatient
admissions for congestive heart failure. Such an HMO could become the victim of
its own success in managing care. In cases in which a beneficiary with
congestive heart failure avoids hospitalization because of better medical
management, for example, the HMO would forgo over $12,000 in higher PIP/DCG
payments in the subsequent year if the system was fully phased in. Not only
would the HMO s success in avoiding hospitalization preclude its receiving the
higher revenues, but the plan would also have incurred higher expenses to
finance the disease management program. These findings are preliminary. But even
if the completed analysis confirms the initial findings, it is unclear how many
Medicare+Choice plans have the sophistication to implement comparable programs.
It is also unclear how many conditions would be susceptible to disease
management interventions that avoided hospitalizations that trigger higher
PIP/DCG payments. However, sophisticated disease management programs for
conditions such as diabetes with complications or chronic obstructive pulmonary
disease might generate similar ACatch 22s. Problems with Refining PIP/DCGs The
successful development of the second stage of PIP/DCG risk adjusters faces
formidable obstacles. Relying on hospital inpatient data means that the data
sets are, compared with the total volume of Medicare claims, relatively
manageable. Expanding the adjustment system to include outpatient procedures
markedly increases the number of claims to be analyzed. Including all Medicare
services could further increase the number of claims by an order of magnitude.
Simply manipulating the data will pose significant challenges. Hospitals have
long had strong incentives to precisely code inpatient admissions, making the
claims and diagnostic information relatively reliable. HCFA may encounter
significant problems with the reliability and validity of some of the data that
would be used in the second stage of PIP/DCGs. The accuracy of hospital
outpatient data, for example, might prove problematic for use in the more
comprehensive risk-adjustment system. ALTERNATIVE APPROACHES TO RISK ADJUSTMENT
The discussion earlier in my testimony highlighted some of the problems
associated with devising and improving an adequate mechanism for adjusting
payments for risk. HCFA and others have funded extensive research in efforts to
develop viable mechanisms. The inability to devise more effective tools
underscores how difficult the challenge actually is. An alternative to using a
statistical approach to adjust payments is to alter the level of risk borne in
the payment pool. Some payers, such as state Medicaid agencies, are using a
variety of approaches that, in effect, adjust the risk pool, not the payments.
Under fee-for-service, physicians and other providers can be viewed as revenue
centers: the more services they provide and bill, the more they get paid. That
arrangement provides strong incentives to use more, rather than fewer, services.
In stark contrast, under capitated payment arrangements, providers are cost
centers: their revenue is fixed, so that providing services adds only to costs,
not to payments. One explanation for the differing utilization patterns between
fee-for-service and (capitated) managed care is that providers are converted
from Arevenue centers to Acost centers. In a Health Affairs article, Joseph
Newhouse and colleagues have argued in favor of partial capitation. They raise
concerns about stinting on needed care when a provider must bear 100 percent of
the marginal cost of providing services. That concern may be strongest where
providers risk pools are too small to be stable or where providers are thinly
capitalized. Payment systems that combine attributes of fee-for-service and
capitation create incentives to avoid unnecessary services but not stint on
needed care. Many such approaches are possible. I will describe four generic
types of hybrid payment systems that combine some capitation with additional
payments as services or costs increase. Those approaches are currently used in
commercial markets, Medicaid, or Medicare demonstrations. They all limit the
amount of risk assumed by a risk pool by paying extra for high- cost cases; that
permits smaller risk pools to be more stable, lessening their volatility and
susceptibility to big financial swings. To keep such systems budget neutral, the
average capitation payments must be reduced by the amount being Acarved out for
separate payment. First-Dollar Partial Capitation. HCFA is experimenting with
partial capitation payments in a demonstration project with an academic health
center at the University of California at San Diego (UCSD). For inpatient
hospital services, HCFA pays the UCSD health plan half of the Medicare
fee-for-service payment plus a capitated amount. In part because of the reduced
risk associated with this payment system, UCSD chose to offer a managed care
plan that permitted direct access to the specialists on its medical school
faculty. Condition-Specific Carve-Outs. Pregnancy, acquired immunodeficiency
syndrome (AIDS), solid organ transplants, and end-stage renal disease (ESRD) are
all examples of disease or condition-specific carve-outs being employed by
Medicaid agencies, HMOs, or Medicare. Some Medicaid agencies remove AIDS or
other high-cost conditions from their capitation rates. Others exclude
pregnancy-related costs from their normal capitated payments. Instead, special
payments are made for each case or each delivery. Such payment systems can
easily be adjusted to promote specific objectives. For example, if a goal was to
promote prenatal care and limit caesarian deliveries, a flat Abundled payment
could be made for all hospital and physician services. In contrast, paying
separate, higher rates for C-sections and lower rates for vaginal deliveries
would instill fewer incentives to avoid C-sections. For decades, Medicare has
separated individuals with ESRD into a distinct risk pool. Now, Medicare is
experimenting with paying for ESRD beneficiaries on a capitated basis.
Similarly, some HMOs carve out solid organ transplants from their capitation
payments to providers, retaining the risk (and payment responsibility) at the
plan level. Individual (Specific) Stop-Loss Coverage. Many providers and health
plans purchase private reinsurance to limit the costs of specific individuals or
cases, which is often referred to as A specific stop-loss coverage. Coverage
thresholds, known as Aattachment points," vary considerably. Some entities
choose very high reinsurance thresholds, seeking to handle only catastrophically
expensive cases. Others choose lower attachment points, seeking to reduce their
financial exposure. The lower the attachment point, the higher the reinsurance
premium the amount carved out of the capitation rates necessary to finance the
costs. Like the attachment points, the amount of excess costs reimbursed can
also vary. In some cases, reinsurance pays 50 percent of costs in excess of the
first threshold and 80 percent of costs above a second, higher threshold. Other
policies pay 100 percent of costs in excess of a threshold. By varying both the
attachment point(s) and the share of costs paid, specific stop-loss policies can
significantly moderate risk. At the extreme, certain stop- loss policies
approach first-dollar partial capitation. (That occurs if the initial payment
threshold is the first dollar.) Aggregate Stop-Loss Coverage. Aggregate
stop-loss coverage is also a commercially available product. Typically, that
coverage presupposes the existence of an underlying specific stop-loss policy.
If the cost of services for all members of the risk pool exceeded a specific
level, the aggregate reinsurance policy could reimburse those excessive costs.
For example, assume that a physician has 300 capitated Medicare beneficiaries in
his or her risk pool and buys both specific and aggregate reinsurance. Any costs
of physician services for an individual in excess of $7,500 would be paid by
specific reinsurance. None of the amounts above the attachment point would be
counted when calculating aggregate costs. However, all costs up to $7,500 would
be included in calculating whether aggregate reinsurance payments would be
triggered. In this example, two individuals might require extensive cardiac
services and open- heart surgery, generating physician fees in excess of $10,000
each. The specific reinsurance policy would pay the costs over $7,500 in each
case. Assume further that the average cost of physician services for each member
of this physician s Medicare risk pool equals $1,800 (after excluding the
catastrophic costs over the threshold) but that the physician only averaged a
capitation payment of $1,440 per patient per year. Any costs averaging in excess
of $1,728 per patient per year, which is 120 percent of the annual capitation
payment, would qualify for aggregate reinsurance. CONCLUSION The success of
Medicare+Choice is tied to how much, and how, Medicare pays. Low rates of
increase in payments will tend to cause health plans to withdraw from or limit
their presence in the Medicare+Choice market. Constrained payment rates will
make benefit offerings less attractive to consumers, which will further slow
growth in enrollment. Even though it is an improvement over the prior
demographic adjuster, the PIP/DCG is a flawed mechanism for adjusting for risk
selection. HCFA is working to develop an improved method for implementing stage
two that would take account of service use in all settings. Because of the
difficulty in markedly improving mechanisms that adjust payments, however, the
Congress may wish to consider other approaches that would limit the risk borne
by a pool.
LOAD-DATE: June 9, 1999