Chairman Bilirakis, Congressman Brown, distinguished Committee members, thank you for
inviting me here to discuss our efforts to pay health plans accurately and fairly. The Balanced
Budget Act of 1997 requires Medicare to "risk adjust" payments to Medicare+Choice
organizations, starting January 1, 2000. That means we must base payment to Medicare+Choice
plans on the health status of their enrollees.
Risk adjustment is an essential component of the Medicare+Choice program, and represents a
vast improvement over the current payment method. It helps assure that payments are
appropriate and curtail the disincentive for plans to enroll sicker beneficiaries.
Under risk adjustment, data on individual beneficiaries use of health care services in a given year
will be used to adjust payment for each beneficiary enrolled in a Medicare+Choice plan the
following year. The payment adjustments are based on the average total cost of care for
individuals who had the same diagnoses in the previous year. In order to prevent disruptions to
beneficiaries and health plans, we will phase this change in over five years. Initially, we will use
data on inpatient hospital stays and move in an orderly fashion, as envisioned in the Balanced
Budget Act, to use of data from other health care settings.
We would like to thank plans for their cooperation in providing the data needed to implement
this important advance.
Currently, some 6 million of Medicare's 40 million beneficiaries have chosen to enroll in
Medicare+Choice plans. Risk adjustment will increase payment to plans for their sickest
patients, and thus curtail the disincentive for plans to enroll these beneficiaries. It also will lower
payment to plans for their healthier patients. Risk adjustment is an essential step forward for
beneficiaries, taxpayers, and health plans.
Risk adjustment will help beneficiaries feel confident in all their Medicare+Choice
options. It will assure beneficiaries that Medicare pays plans the right amount to provide
all necessary care because payment to plans will take each enrollee's health status into
account. That will help people with serious illnesses, such as cancer or cardiovascular
disease, who can benefit most from the coordination of care health plans can provide.
Risk adjustment will help taxpayers by addressing the main reason that Medicare has
lost rather than saved money on managed care. Many studies show that health plans
enroll Medicare beneficiaries who, on average, are much healthier and therefore less
costly than those who remain in traditional Medicare. This "favorable selection" of
healthy beneficiaries has cost taxpayers $2 billion a year, according to a 1997 report by
Congress' Physician Payment Review Commission (now part of the Medicare Payment
Risk adjustment will help level the playing field among Medicare+Choice plans. It
will temper the risk of significant financial loss when plans enroll beneficiaries who have
expensive care needs, and focus competition more on managing care than on avoiding
risk. Risk adjustment also will help plans by alleviating concerns among beneficiaries
that plans have financial incentives to deny care.
Phasing-In Risk Adjustment
The law requires us to proceed with risk adjustment starting January 1, 2000, and does not call
for a transition. However, we believe we must implement these changes in an incremental and
prudent fashion, as was done with other new major payment systems. We are, therefore, using
flexibility afforded to us in the law to phase in risk adjustment over 5 years to prevent disruptions
to beneficiaries or the Medicare+Choice program.
In the first year, only 10 percent of payment to plans for each beneficiary will be calculated
based on the new risk adjustment method based on inpatient hospital diagnoses. The remaining
90 percent will be based on the existing method for calculating plan payments, which are flat
amounts per enrollee per month based on the average cost to care for Medicare fee-for-service
beneficiaries in each county and adjusted for basic demographic factors like age and sex. In
2001, 30 percent of payment amounts will be risk adjusted. In 2002, 55 percent of payment
amounts will be based on risk adjustment. In 2003, 80 percent of payment amounts will be based
on risk adjustment. By 2004, we and health plans will be ready to use data from all sites of care,
not just inpatient hospital information, for risk adjustment. Then, and only then, will payment to
plans be 100 percent based on risk adjustment.
Using Inpatient Data
During the first year of data collection for risk adjustment, both the statute and practical issues
require that we use hospital inpatient data alone. About one in every five Medicare beneficiaries
is hospitalized in a given year. Data on these hospitalizations are relatively easy to gather, easy
to audit, and highly predictive of future health care costs. We will use the data to pay plans more
for beneficiaries hospitalized the previous year for conditions that are strongly correlated with
higher subsequent health care costs. While we will eventually be using a broader data base for
risk adjustment, that is simply not feasible at this time.
The Balanced Budget Act clearly stipulated that more comprehensive data on outpatient,
physician, and other services could be collected only for services provided on or after July 1,
1998. That was prudent, because it has been no small task for plans to learn how to gather the
inpatient data we are using for the initial phase-in of risk adjustment. Requiring plans to provide
additional data on outpatient, physician and other services would have been unduly burdensome
at this time.
This year, we will issue a schedule and guidance to plans for reporting other encounter data, such
as outpatient information. The schedule will provide sufficient time for plans to gather accurate
data and for HCFA to analyze and incorporate the data into accurate risk adjusted payments. We
are now confident that by 2004 we will be using data on all health care encounters to assess
beneficiary health status for risk adjustment. If we could base risk adjustment on more
comprehensive data now, we would. But we cannot. The law requires us to move forward.
And, even with its limitations, this initial risk adjustment system based on inpatient data alone
will increase payment accuracy 5-fold.
The initial risk adjustment system uses only the approximately 60 percent of inpatient hospital
diagnoses that are reliably associated with future increased costs. For example, beneficiaries
hospitalized for conditions such as heart attacks in aggregate are at higher risk of subsequent
cardiovascular problems, and they consistently have higher health care costs in the subsequent
year. Hospitalizations for such diagnoses will lead to higher payments to plans in the following
year under risk adjustment. Hospitalizations for acute conditions such as appendicitis, however,
rarely lead to increased subsequent care costs. They will not lead to higher payments under risk
The 60 percent of hospital admission diagnoses that are clearly associated with increased
subsequent care costs account for about 30 percent of all Medicare spending the following year.
It is important to note that, while risk adjustment is initially based only on inpatient data, the risk
adjustment payments account for all costs of care associated with each diagnosis. It is also
important to note that risk adjustment is not cost-based reimbursement; it is reimbursement
adjusted for projected need based on health status in the previous year.
Determining Diagnosis Groups
The relevant diagnoses will be used to classify beneficiaries into 15 different cost categories.
One category is for beneficiaries who were not hospitalized the previous year with relevant
diagnoses. For beneficiaries included in any of the other categories, plans will receive an
additional payment to cover the increased risk associated with diagnoses in that category.
Payment will continue to be adjusted for demographic factors, such as age, gender, county of
residence, and whether a Medicare beneficiary is also a Medicaid beneficiary. We have revised
these demographic factors for use with risk adjustment, for example, by no longer including
institutional status because the risk adjustment methodology itself does a good job of predicting
expenses for nursing home residents.
Medicare will calculate a score for each beneficiary to determine the payment that will be made
if they choose to enroll in a Medicare+Choice plan. For example, Medicare's average payment
per year to health plans is $5,800. Under risk adjustment, payment for an 85-year-old man will
on average be $6,414. It will be an additional $2,060 if he is on Medicaid, another $1,207 if he
is disabled, and $8,474 more if he was admitted to the hospital for a stroke the previous year, for
a total of $18,155. The score for each beneficiary will be calculated annually, and will follow
them if they move from one health plan to another.
Protecting Program Integrity
Most health plans operate with integrity and play by the rules, and we doubt that plans will
compromise successful medical management programs that keep patients out of the hospital in
order to game the risk adjustment system. However, plans themselves have raised concerns that
risk adjustment based on inpatient data alone could create perverse incentives for unnecessary
hospitalizations. We, therefore, have taken solid steps to prevent gaming of the system with
inappropriate hospital admissions or attempts to inflate the data submitted for use in risk
The risk adjustment system does not include hospital stays of just one day, in order to help guard
against inappropriate admissions. And it excludes diagnoses that are vague, ambiguous, or rarely
the principal reason for hospital admission. In addition, we will use independent experts to
assess the validity and completeness of data plans submit to us by conducting targeted medical
record reviews and site visits. This will help ensure that plans do not "upcode," or claim that
hospital admissions were for more serious conditions that would result in higher payment.
It is essential to stress that risk adjustment will not and cannot be budget neutral if we intend to
protect the Medicare Trust Fund and be fair to the taxpayers who support our programs. The
whole reason for proceeding with risk adjustment and specifically with risk adjustment that is
not budget neutral is that Medicare has not been paying plans properly.
There is considerable evidence that we have overpaid plans and continue to overpay plans, in
large part because payments are not adjusted for risk.
The Physician Payment Review Commission, in its 1997 Annual Report to Congress,
estimated that Medicare has been making up to $2 billion a year in excess payments to
managed care plans. This Congressional advisory body notes that, unlike the private
sector where managed care has slowed health care cost growth, managed care has
increased Medicare program outlays. The Commission's 1996 Report found that those
who enroll in managed care tend to be healthy and those who disenroll tend to be
unhealthy, exacerbating Medicare losses.
Mathematica Policy Research, which has conducted several studies on Medicare HMOs,
says care of Medicare beneficiaries in HMOs costs only 85 percent as much as care for
those who remain in traditional fee-for-service Medicare. That is 10 percent less than the
95 percent of the average fee-for-service costs plans were being paid.
The Congressional Budget Office has said managed care plans could offer Medicare
benefits for 87 percent of Medicare fee-for-service costs, even though they were paid 95
Congress also recognized that plans have been paid too little for enrollees with costly conditions,
and too much for those with minimal care needs. The simple demographic adjustments made
now for age, gender, county of residence, Medicaid and institutional status, do not begin to
accurately account for the wide variation in patient care costs. Risk adjustment will.
The vast majority of beneficiaries enrolled in Medicare+Choice cost far less than what Medicare
pays plans for each enrollee. Medicare fee-for-service statistics make clear why risk adjustment
must not be budget neutral. More than half of all Medicare fee-for-service beneficiaries cost less
than $500 per year, while less than 5 percent of fee-for-service beneficiaries cost more than
$25,000 per year, according to the latest available statistics for calendar year 1996. The most
costly 5 percent account for more than half of all Medicare fee-for-service spending.
Since Medicare+Choice enrollees tend to be healthier than fee-for-service Medicare
beneficiaries, the ratio of high to low cost beneficiaries in health plans is even more stark.
Clearly, care for the overwhelming majority of Medicare enrollees costs plans much less than
what Medicare pays because our payments are predicated on the average beneficiary cost of care,
calculated by county. This average includes the most expensive beneficiaries in fee-for-service,
who generally do not enroll in managed care.
If risk adjustment was budget neutral, Medicare and the taxpayers who fund it would continue to
lose billions of dollars each year on Medicare+Choice. Accurate risk adjustment inevitably and
appropriately must change aggregate payment to plans.
Budget neutral risk adjustment would cost taxpayers an estimated $200 million in the first year
of the phase-in, and $11.2 billion over 5 years if health plans maintained their current, mostly
healthy mix of beneficiaries. It is important to stress that actual savings to taxpayers from risk
adjustment will vary to the extent that less healthy beneficiaries enroll in Medicare+Choice
plans, resulting in higher payments than health plans receive today.
The amount of payment change will vary among plans and depend on each plan's individual
enrollees. Total payment may be higher for some plans as they enroll a mix of beneficiaries that
is more representative of the entire Medicare population. As part of our Medicare+Choice March
1 rate announcement, we will send a letter to each health plan with an estimate of how payment
will differ from what they are paid now, based on their current mix of enrollees.
Overall, we project that payment to Medicare+Choice plans on average will change by less than
one percent in the first year. How it will change over time depends on the mix of beneficiaries in
each plan. Risk adjustment significantly changes incentives for plans and could well lead to
enrollment of beneficiaries with greater care needs. That could result in plans receiving higher
payments than they do now. Phasing in risk adjustment also substantially buffers the financial
impact on plans. The federal government is forgoing $1.4 billion in savings in the first year and
as much as $4.5 billion over the full 5 years because of the phase in.
Payment changes will be further buffered by an annual payment update for 2000 that our
preliminary estimate suggests will be 5.2 percent. This is substantially larger than projections
that were made last year. The final figure will be released March 1, 1999. This annual update is
based on formulas set in law and projected expenditures for Medicare that are included in the
President's fiscal year 2000 budget.
Risk adjustment is an essential step forward for Medicare, beneficiaries, taxpayers and the
Medicare+Choice program. It will help Medicare pay plans fairly and accurately. It will curtail
disincentives to enroll less healthy beneficiaries. It will help taxpayers and the Medicare Trust
Fund start saving, rather than losing, money on managed care. It will help level the playing field
among plans. And it is required by law.
We are aware of the magnitude of the impact of risk adjustment and are, therefore, phasing in
implementation to avoid undue disruptions. We are also taking proactive steps to prevent
potential gaming of the system. We will closely monitor the impact on beneficiaries and plans.
We will continue to consult with beneficiary groups, health plans and academic experts.
Adjustments can be made each year as we proceed.
But, clearly, we must proceed. Risk adjustment is too important to postpone and too important
to implement without a prudent phase-in that allows time for any necessary refinements. Again,
I thank you for inviting us here today to discuss this, and I am happy to answer your questions.