Medicare Program: Request for Public Comments on Implementation of Risk Adjusted Payment for the Medicare+Choice Program and Announcement of Public Meeting

DEPARTMENT OF HEALTH AND HUMAN SERVICES

Health Care Financing Administration

[HCFA-1045-N]

RIN 0938-AJ16

Medicare Program: Request for Public Comments on Implementation

of Risk Adjusted Payment for the Medicare+Choice Program and

Announcement of Public Meeting

AGENCY: Health Care Financing Administration (HCFA), HHS.

ACTION: Solicitation of comments; announcement of meeting.

-----------------------------------------------------------------------

SUMMARY: This notice solicits further public comments on issues related

to the implementation of risk adjusted payment for Medicare+Choice

organizations. Section 1853(a)(3) of the Social Security Act (the Act)

requires the Secretary to implement a risk adjustment methodology that

accounts for variation in per capita costs based on health status and

demographic factors for payments no later than January 1, 2000. The

methodology is to apply uniformly to all Medicare+Choice plans. This

notice outlines our proposed approach to implementing risk adjusted

payment.

In order to carry out risk adjustment, section 1853(a)(3) of the

Act also requires Medicare+Choice organizations, as well as other

organizations with risk sharing contracts, to submit encounter data.

Inpatient hospital data are required for discharges on or after July 1,

1997. Other data, as the Secretary deems necessary, may be required

beginning July 1998.

The Medicare+Choice interim final rule published on June 26, 1998

(63 FR 34968) describes the general process for the collection of

encounter data. We also included a schedule for the collection of

additional encounter data. Physician, outpatient hospital, skilled

nursing facility, and home health data will be collected no earlier

than October 1, 1999, and all other data we deem necessary no earlier

than October 1, 2000. Given any start date, comprehensive risk

adjustment will be made about three years after the year of initial

collection of outpatient hospital and physician encounter data.

Comments on the process for encounter data collection are requested in

that interim final rule. We intend to consider comments received in

response to this solicitation as we develop the final methodology for

implementation of risk adjustment.

This notice also informs the public of a meeting on September 17,

1998, to discuss risk adjustment and the collection of encounter data.

The meeting will be held at the Health Care Financing Administration

headquarters, located at 7500 Security Boulevard, Baltimore, MD,

beginning at 8:30 a.m. Additional materials on the risk adjustment

model will be available on or after October 15, 1998, and may be

requested in writing from Chapin Wilson, Health Care Financing

Administration, Department of Health and Human Services, 200

Independence Avenue, S.W., Room 435-H, Washington, DC 20201.

DATES: We request that comments be submitted on or before October 6,

1998.

ADDRESSES: Mail written comments (1 original and 3 copies) to the

following address: Health Care Financing Administration, Department of

Health and Human Services, Attention: HCFA-1045-N, P.O. Box 26688,

Baltimore, MD 21207.

If you prefer you may deliver your written comments (1 original and

3 copies) to one of the following addresses:

Room 309-G, Hubert H. Humphrey Building, 200 Independence Avenue, SW.,

Washington, DC 20201, or

Room C5-09-26, 7500 Security Boulevard, Baltimore, MD 21244-1850

Because of staffing and resource limitations, we cannot accept

comments by facsimile (FAX) transmission. In commenting, please refer

to file code HCFA-1045-N. Comments received timely will be available

for public inspection as they are received, generally beginning

approximately 3 weeks after publication of a document, in Room 309-G of

the Department's offices at 200 Independence Avenue, SW., Washington,

DC, on Monday through Friday of each week from 8:30 a.m. to 5 p.m.

(phone (202) 686-7890).

FOR FURTHER INFORMATION CONTACT: Cynthia Tudor, (410) 786-6499.

SUPPLEMENTARY INFORMATION:

I. Background

Since 1985, Medicare payments to risk contracting Health

Maintenance Organizations (HMOs) for aged and disabled beneficiaries

living in a given county have been based on actuarial estimates of the

per capita cost Medicare incurs paying claims on a fee-for-service

(FFS) basis in that county. (Medicare's costs in paying claims for

beneficiaries with end-stage renal disease are not considered in these

county estimates, but are treated separately on a statewide basis.)

These county estimates have been adjusted for the demographic

composition of that county (age, gender, Medicaid eligibility status,

and institutional status) in order to produce a figure representing the

costs that would be incurred by Medicare on behalf of an average

Medicare beneficiary in the county. These county per capita payment

rates, adjusted for the average beneficiary, have been published

annually as the county rate book. Prior to January 1998, actual

payments for a given HMO enrollee were based on this county rate book

amount, adjusted by demographic factors associated with each enrollee.

Again, the demographic factors have been age, gender, Medicaid

eligibility, and institutional status. This methodology is known as the

``Adjusted Average Per Capita Cost'' (AAPCC) methodology, and HMOs with

Medicare contracts under section 1876 of the Social Security Act (the

Act) were paid on this basis between 1985 and 1997.

[[Page 47507]]

In enacting the new Part C of Title XVIII to create the

Medicare+Choice program, the Congress provided, a new section 1853 of

the Act, for a new methodology for paying organizations that enter into

Medicare+Choice contracts. Under this new methodology, the equivalent

of the above-described county rate book (that is, the county-wide

amount that is adjusted by an individual enrollee's demographic status

to determine the final payment amount) is based on the greatest of

three amounts. The first amount is a new blended payment rate

methodology that would combine the area specific amounts with national

data and would be subject to other adjustments. The second amount is a

new minimum specified rate amount (for example, $367 per month per

enrollee in 1998). The third amount is based on a 2 percent increase

over the prior year's rates, with the rate book for 1997 serving as the

baseline. As in the case of the AAPCC methodology described above, the

county rates under section 1853 of the Act, are adjusted for the

demographic status of each enrollee.

Under section 1876(k)(3) of the Act, the new Medicare+Choice

payment methodology under section 1853 of the Act applies to existing

HMO contracts under section 1876 for 1998, and to Medicare+Choice plans

beginning in 1999.

Section 1853(a)(3) of the Act requires the Secretary to develop and

implement a new risk adjustment methodology to be used to adjust the

county-wide rates under section 1853 of the Act to reflect the expected

relative health status of each enrollee. This new methodology, which

must be implemented by January 1, 2000, would replace the current

method of adjusting county-wide rates based on the four demographic

factors of age, gender, Medicaid eligibility, and institutional status.

The goal is to pay Medicare+Choice organizations based on better

estimates of health care costs of the population they enroll (relative

to the FFS population).

While the Medicare+Choice legislation mandates the implementation

of risk adjustment in general, the legislation provides the Secretary

with broad discretion to develop a risk adjustment methodology that

would ``account for variations in per capita costs based on health

status and other demographic factors.'' Because Medicare+Choice

legislation does not allow for the collection of any data other than

inpatient hospital data (in the near term), we are constrained

initially to using a model that requires only inpatient data. We are

currently receiving these data. In previous public meetings on

encounter data requirements, organizations have been briefed on the

Principal Inpatient Diagnostic Cost Group (PIP-DCG), created by HHS-

sponsored researchers at Health Economics Research, Inc., and Boston

and Brandeis. This is the only risk adjuster model that has been

developed to run solely on inpatient data. The model was recently

updated using 1995 and 1996 Medicare data.

The remainder of this notice outlines our proposed approach for

implementation of risk adjusted payments on January 1, 2000, discussing

both the risk adjustment methodology and the proposed risk adjustment

payment model. In the development of all risk adjustment payment

models, there are two tasks that must be performed: (1) The estimation

of the risk adjustment model, and (2) application of the risk

adjustment model to a payment system. The estimation of the PIP-DCG

model is described first.

A. The Principal In-Patient Diagnostic Cost Group (PIP-DCG) Model

In constructing a risk adjustment model, it is important to

determine which set of conditions should be used to adjust payments.

Under the current payment system, all enrollees are placed in a base

group paid according to demographic characteristics. In this risk

adjustment system, all conditions that appear as inpatient principal

diagnoses are candidates for adjusting payments. The base payment

category decreases as more conditions are placed into separate disease

groups. Because an inpatient hospital-based system depends on a

person's site of service, only a subset of conditions should be

recognized for changing payments. That is, the system should recognize

admissions for which inpatient care is most frequently appropriate. For

example, admissions for diseases most commonly treated on an outpatient

basis should remain in the base group and should not be used for

adjustment.

The PIP-DCG model was estimated using diagnostic information for

Medicare FFS enrollees from inpatient hospital stays during calendar

year 1995. The sample used in the estimation analyses consisted of

individuals included in the 5-percent sample of Medicare beneficiaries

who were alive and enrolled in Medicare during all of 1995, and on

January 1, 1996. Beneficiaries with certain characteristics (for

example, HMO enrollees and end-stage renal disease enrollees, new

Medicare eligibles in 1996) were excluded from the analyses. In

general, these exclusions were made to increase confidence that a

complete set of Medicare claims for each beneficiary in the sample data

set was included in the model development. The final estimation data

set included 1.4 million Medicare beneficiaries.

While the PIP-DCG model uses only inpatient diagnoses in creating

the risk adjustment classification system, the model predicts total

expected costs for the following year across multiple sites of

services. Consequently, all Medicare expenditures, other than those for

hospice care, were included in the calculation. Medicare expenditures

for hospice care were not included because Medicare+Choice

organizations are not responsible for hospice care. The model was

estimated assuming no time lag between the base year (diagnostic

information) and the predicted expenditures; that is, calendar year

1995 beneficiary diagnoses were used to predict calendar year 1996

expenditures.

1. From Diagnosis Groups (DxGroups) to PIP-DCGs

The risk adjustment model estimation process begins with a

classification system, forming the inherent logic of the model. For the

PIP-DCG model, diagnoses are classified into DxGroups based on the

principal inpatient diagnosis. The DxGroups comprise an exhaustive

classification of all valid International Classification of Diseases,

Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic codes. For

example, DxGroup 1, Central Nervous System Infections, includes ICD-9-

CM diagnostic codes for such conditions as encephalitis and meningitis.

The primary criteria in forming the DxGroups were clinical coherence

and an adequate sample size to estimate average expenditures.

Beneficiaries with multiple different inpatient diagnoses could have

multiple hospital stays, and would initially be placed in multiple

DxGroups.

Next, DxGroups were aggregated into payment groups, or PIP-DCGs,

using a sorting algorithm that ranked DxGroups based on 1996 actual

expenditures. For example, DxGroup 7 (Metastatic Cancer with a mean

future expenditure of $26,331) was placed in PIP-DCG 26. Highest

expenditure DxGroups were grouped into the ``highest'' PIP-DCG. Once

beneficiaries with the highest costs were placed into a DxGroup, those

beneficiaries and all their associated expenditures were removed from

the data for other DxGroups and then the DxGroups were re-ranked. The

DxGroups with the next most costly diagnoses were grouped into the next

highest numbered PIP-DCG, and those beneficiaries were removed from the

[[Page 47508]]

remaining DxGroups. The process was repeated until each beneficiary and

his or her expenditures were assigned to a single PIP-DCG group.

Beneficiaries with multiple inpatient diagnoses were placed in their

highest expenditure PIP-DCG group.

In this way, each PIP-DCG group was defined according to average

total expenditures for beneficiaries with inpatient diagnoses,

categorized and sorted using the DxGroups rather than diagnosis by

diagnosis. Based upon this sorting algorithm, more than 20 initial PIP-

DCGs were defined. Lower average expenditure PIP-DCG groups had lower

cost ranges (or intervals), while the highest average expenditure PIP-

DCG groups had wider ranges.1

---------------------------------------------------------------------------

\1\ The PIP-DCG groupings were further refined using a number of

criteria. First, each original PIP-DCG group remained in the final

payment model only if it contained at least 1,000 beneficiaries from

the original sample; this minimum sample size was defined to assure

stability of estimated payments in the final model. If sample sizes

were smaller than 1,000, the potential PIP-DCG was expanded to

include DxGroups with average expenditures in the next lower range

until the sample size criteria was satisfied. If at any time during

the sorting algorithm a DxGroup had fewer than 50 beneficiaries

assigned to it, it was assigned to the base payment category. This

base payment category also included all beneficiaries (and

expenditures) for whom there was no inpatient diagnosis during 1995.

---------------------------------------------------------------------------

2. Modifications to the PIP-DCG Model

After the initial sorting of DxGroups into PIP-DCG groups was

complete, a clinical panel reviewed the placement of the DxGroups and

their resulting predicted expenditures, to determine the

appropriateness of their application in a payment model. Through this

process, 75 DxGroups (covering about \1/3\ of the admissions) were

identified as: (1) Representing only a minor or transitory disease or

disorder, not clinically likely to result in significant future medical

costs, (2) rarely the main cause of an inpatient stay, or (3) vague or

ambiguous. These groups, as recommended by the clinical panel, were

identified as those most likely to result in inconsistent or

inappropriate reimbursements and were placed (with their associated

expenditures) in the base payment category (for which the payment is a

function of demographic factors). Examples of these groups include the

DxGroup for fluid/electrolyte disorders and malnutrition. Though the

treatment for individuals with this diagnoses are often quite costly in

the following year, the diagnosis is clinically vague and, therefore,

represented a likely target for coding ``creep.'' The clinical panel

concluded that many of the sickest individuals with this diagnosis were

likely to have another hospitalization that would trigger appropriate

increased reimbursements. Then, the remaining DxGroups were resorted

and placed into revised DCGs for the payment model. A total of 10 PIP-

DCGs (above the base payment category) are included in the current

model.

As a second strategy to ensure consistent and appropriate payment

levels, beneficiary diagnoses reported as a result of a short hospital

stay (1 day or less) were left in the base payment category. Since the

majority of 1-day stays are for diagnoses already assigned to the base

group, the effect on payment is small. Also, short stays are often

indicative of less serious, and, hence, less costly cases. It is

important to note that these modifications do not mean that these

expenditures have been excluded from the model. Rather, the payments

associated with these diseases are captured in increased payments for

the base payment category, where the majority of enrollees are paid

based on demographic factors.

Under the proposed PIP-DCG model, beneficiaries who are

hospitalized for chemotherapy (V58.1 and V66.2) were treated as

exceptions. These codes are indicators of a treatment method, rather

than a particular disease. Recognizing, however, that Medicare's

current inpatient coding rules require that the diagnoses for

beneficiaries who are hospitalized for chemotherapy must be coded using

these V-codes as the principal diagnoses, the most appropriate PIP-DCG

group for these beneficiaries would be assigned based on the type of

cancer, using a secondary diagnosis. A model will be estimated that

uses secondary diagnoses to determine risk scores for hospitalized

beneficiaries that were assigned chemotherapy V-codes (as defined

above). This modification could be made for payment in calendar year

2000. The model described in this notice has left these admissions in

the base group.

3. Addition of Demographic and Other Factors

The next phase in the estimation of the model was the creation of

demographic variables (age, sex, and disability status) for the PIP-DCG

groups. In this phase of the calibration, 24 age and sex groupings were

created. Separate groupings were created for males and females, by 5-

year age increments, except where numbers were too small to get good

estimates (that is, age group 0 through 34 and greater than 94 for

males and females).

Separate parameters were also included to estimate the unique cost

effects of whether an aged beneficiary was formerly eligible because of

a disability, and whether an aged or disabled beneficiary is eligible

for Medicaid. The estimated adjustments for the demographic categories

are the same irrespective of which PIP-DCG an enrollee falls into. The

Medicaid adjustment, however, depends on a person's status as aged or

disabled.

New enrollees to Medicare, for whom there are no claims history,

will be assigned a score based on a separate HCFA analysis of actual

new enrollee expenditures. At this time, a separate parameter is not

anticipated for the institutionalized because institutional status is

not needed as an indicator of high Medicare utilization. Under the

demographically adjusted system, institutional status was an indicator

of a beneficiary with relatively poor health status. It, therefore,

increased payments over the age and sex based amounts. The risk

adjuster model has health status measures built in, and on the average,

compensates for poor health status. In fact, preliminary estimates

indicate that after accounting for inpatient hospital admissions, the

institutional adjustment would be negative. Adjustments for the

working-aged will be made in a manner similar to the current system. As

a last step during the estimation, expenditures were adjusted to create

an estimate of annual payments as if each beneficiary had been alive

and enrolled for the entire year. This is equivalent to an expenditure

per month measure. Estimation of the incremental costs associated with

each of the variables (for example, demographics, DCGs) was made by the

linear regression technique, which takes account of all the variables

that apply to an individual.

4. The Current PIP-DCG Model

The current PIP-DCG model contains a total of 37 parameters (10

PIP-DCGs and 27 demographic or Medicaid factors). The model will

continue to be refined over the next few months. While there are a

number of ways to assess the ``accuracy'' of the model, payment for

different groups of beneficiaries is improved with risk adjustment

compared to the application of a demographic only model. Preliminary

coefficients for the PIP-DCG model are presented in Table 1. The

current placements of DxGroups into PIP-DCG groups are shown in Table

2. The next section of this notice details how we are proposing to use

the PIP-DCG model in the Medicare+Choice payment system as of January

1, 2000.

[[Page 47509]]

B. Proposed Payment System Application of the PIP-DCG Model

In its basic form, the PIP-DCG model is an algorithm that uses base

year inpatient diagnoses, along with demographic factors and Medicaid

eligibility, to predict total health spending in the following year. In

applying the PIP-DCG model to risk adjusted payments for the

Medicare+Choice program, however, the model will be used to determine

relative risk scores. These relative risk scores will be used, in place

of the current demographic factors, to adjust county rate book payments

for the relative health status of the individual enrollee.

1. Estimating Beneficiary Relative Risk Factors

The PIP-DCG model was developed to be ``additive'', meaning that

incremental dollars are added together based on each beneficiary's

characteristics. Referring to Table 3, the following examples

illustrate how the PIP-DCG model will be used for estimating relative

risk factors.

A beneficiary is placed in a PIP-DCG group, based on inpatient

diagnoses reported. In this example, ``Beneficiary A'' was hospitalized

twice during the base year. The diagnoses reported were Asthma (PIP-DCG

8) and Lung Cancer (PIP-DCG 18). The highest PIP-DCG category then for

this beneficiary is PIP-DCG 18, which carries with it an estimated

future year expenditure of $12,883. The beneficiary is also placed in

the appropriate demographic groups. In this case, Beneficiary A is

male, aged 82. This age group carries an estimated expenditure of

$5,617. In addition, Beneficiary A had originally been Medicare

eligible because of a disability (which carries an incremental

expenditure of $2,381), but is not eligible for Medicaid (no

expenditure increment). Adding together these increments based on the

PIP-DCG model, the predicted expenditures for this beneficiary are

$20,881.

As another example, consider ``Beneficiary B.'' Beneficiary B had

no inpatient admissions during the base year. Therefore, no specific

PIP-DCG increment is added; expenditures for non-hospitalized

beneficiaries are included in the demographic factors. Beneficiary B is

placed in the appropriate age and sex grouping; in this case, female

aged 72, which carries a predicted expenditure of $3,118. Beneficiary B

is also placed in the Aged with Medicaid eligibility group, which adds

$2,124 to her annual predicted expenditures. Since she has never been

disabled, no additional expenditures are added. Therefore, total annual

predicted expenditures for Beneficiary B are $5,242.

Because Medicare+Choice program payments are based on the county-

wide rates determined under section 1853(c) of the Act, the predicted

annual expenditures described above will be converted to relative risk

scores. This is accomplished by dividing the predicted expenditures for

each beneficiary by the national average predicted expenditure

($5,300). Individuals whose risk scores are equal to 1.00 are

``average.'' In the examples described above, Beneficiary A's relative

risk score is 3.9 (indicating a high expected cost individual), while

Beneficiary B's relative risk score is 0.99 (indicating a slightly

lower than average risk individual).

After Medicare+Choice organizations submit inpatient hospital

encounter data, we will use the demographic information and diagnostic

information from all Medicare+Choice organizations a beneficiary may

have joined and from FFS to determine the appropriate risk factor for

each beneficiary. When a Medicare+Choice organization forwards

enrollment information to us, we, in turn, will send the

Medicare+Choice organization the appropriate risk factor, as well as

the resultant payment. Because the risk factor is computed for each

individual beneficiary, the factor follows that beneficiary. In

addition, since all beneficiaries will have risk factors, information

will be immediately available for payment purposes as beneficiaries

move among Medicare+Choice organizations.

Risk adjustment factors for new Medicare beneficiaries (for whom

health status information) is not available will be based on

demographic information only. Examples of persons using the demographic

model are new 65-year-olds and new Medicare disabled individuals.

Similar to the current system, a ``demographic only'' model is being

developed that will be used to determine the risk adjustment factors

for these beneficiaries.

2. Risk Adjusted Payment Model

To determine risk adjusted monthly payment amounts for each

Medicare+Choice enrollee, individual risk factors (described above)

will be multiplied by the appropriate payment rate for the county

determined under section 1853(c) of the Act. Beginning with the

implementation of risk adjustment, the separate aged and disabled rate

books (incorporating combined Medicare Parts A and B) will be combined.

Risk adjusted payments will be made using a single, combined

Medicare+Choice county rate book. This change will be made because

there is a single risk adjustment methodology for the entire Medicare

population (excluding persons with end-stage renal disease).

In addition to combining the current aged and disabled county rate

books into a single combined county rate book, an adjustment to these

rate book amounts will be required before applying the risk adjustment

factors discussed above. This adjustment, or re-scaling factor, is

necessary in order to account for the fact that the existing county

rate book already accounts for demographic factors that are addressed,

in a more precise way, in the risk adjustment factors we will be using.

If the PIP-DCG model risk adjustment factors were applied to unadjusted

county rate book amounts, this would create unintended distortions that

would produce adjustments inconsistent with Congress' mandate in

section 1853(c) of the Act. The application of the rescaling factor we

are proposing would in effect translate the rate book amounts into the

same language used under the risk adjustment methodology, so that we

are not comparing ``apples to oranges.'' As a result of rescaling,

payment for a person with the average risk score in a county would be

the same as payment for a person with the average demographic score in

that county. (However, a person with the average demographic score does

not necessarily have the average risk score.) To the extent that an

organization enrolls sicker people, the organization will receive

higher payments.

C. Summary of HCFA's Proposed Approach for 2000

The proposed approach we will use to meet the year 2000 mandate for

risk adjusted payments will--

(1) Be based on inpatient data;

(2) Utilize a prospective PIP-DCG risk adjuster to estimate

relative beneficiary risk scores;

(3) Apply a re-scaling factor to address inconsistencies between

demographic factors in the rate book and new risk adjusters;

(4) Apply individual enrollee risk scores in determining fully

capitated payments;

(5) Include the auditing of medical records to validate encounter

data;

(6) Implement processes to collect encounter data on additional

services; and

(7) Continue to refine the risk adjustment system based on ongoing

research.

[[Page 47510]]

D. Other Issues

In addition to comments on the proposed risk adjustment approach,

we are interested in receiving responses to the following questions:

(1) Under one possible implementation approach we have considered, a

Medicare+Choice organization would be paid initially based on estimates

of the number of enrollees the organization has in a given risk factor

category. These estimates would be based on the most recently available

data (probably July 1998 through June 1999). Once more current data

(from January 1999 through December 1999) became available in July

2000, a retroactive adjustment would be made pursuant to section

1853(a)(2) of the Act ``to take into account any difference between the

actual number of individuals enrolled'' in a given risk category, and

the ``number of such individuals estimated to be so enrolled when the

advance payment was determined.'' These adjustments would be made

retroactive to January 2000. This would be consistent with our

longstanding practice of making retroactive adjustments to reflect the

actual number of enrollees in a current demographic category (such as

institutional status, end-stage renal disease status, dual eligible

status, or working aged status) when this number differs from the

number of enrollees estimated to be in any such category at the time

payments were initially made.

An alternative approach is to use data from an earlier period (for

example, July 1, 1998 through June 30, 1999) to determine the risk

factor for enrollees and payments to Medicare+Choice organizations for

calendar year 2000. Using data from an earlier time period introduces

some error into the estimates, but we do not believe it introduces any

systematic bias. Note that implementation of this alternative model

solves the problem of basing the payments to a plan on the estimated

number of enrollees in a given risk factor category, which would

require a retroactive adjustment as described above. Assuming a

relatively large and stable population for a plan, aggregate payments

under this approach are not likely to differ from aggregate payments

using a method requiring this type of retroactive payment adjustment.

However, on an individual basis, using data from an earlier time period

lengthens the time between a hospital stay for an enrollee and

compensation to the organization for the future predicted cost of that

illness.

Given these issues, what problems are Medicare+Choice organizations

likely to encounter with retroactive payment adjustments? Conversely,

if data from an earlier time period were used, what problems are

organizations likely to encounter?

(2) The Secretary is required to announce the annual

Medicare+Choice capitation rate for each Medicare+Choice payment area

and the risk and other factors to be used in adjusting such rates by

March 1 of the year preceding the payment year. In addition, at least

45 days prior to the annual announcement of capitation rates, the

Secretary shall provide notice to Medicare+Choice organizations of

proposed changes to be made in the methodology from the methodology and

assumptions used in the previous announcement.

The implementation of risk adjustment will alter the methodology

for calculating rates for each Medicare+Choice payment area. Given the

proposed changes, what types of information should be included in the

45-day notice and the annual announcement to assist Medicare+Choice

organizations in planning for risk adjusted payments?

(3) What types of problems are Medicare+Choice organizations likely

to encounter as capitation payments are changed from a demographic only

basis to a health status adjusted basis? How should we address these

problems, especially for small plans, rural plans, and start up plans?

While we are currently processing the inpatient hospital data for

managed care enrollees, we note that we will be unable to model the

financial impact of the risk adjustment methodology until we have

completed the processing of these data and have assigned risk scores to

plans enrollees.

II. September 17, 1998, Public Meeting

In addition to seeking written comments from the public, we will

hold a public meeting on September 17, 1998, at HCFA, 7500 Security

Boulevard, Baltimore, MD. The purpose of this meeting will be to

discuss issues and concerns from potential Medicare+Choice

organizations, organizations contracting under section 1876 of the Act,

providers, beneficiaries, and other interested parties on the

implementation of risk adjusted payment. The collection and auditing of

encounter data, which was described in the Medicare+Choice interim

final rule published on June 26, 1998, in the Federal Register, will

also be addressed in this meeting. The agenda for the meeting is likely

to cover the following topics:

Background on the Principal Inpatient Diagnostic Cost

Group (PIP-DCG) risk adjustment model.

Changes to the payment rates.

Application of the risk adjustment model for payment in CY

2000.

Description of the overall risk adjustment implementation

process.

Auditing of encounter data.

Collection of additional encounter data.

Comments on the proposed agenda are welcome. Further information on

the meeting can be obtained from Chapin Wilson, (202) 690-7874.

In accordance with E.O. 12866, this notice was reviewed by the

Office of Management and Budget.

Table 1.--Current PIP-DCG Model

------------------------------------------------------------------------

Number of Observations.................................. 1,401,274

R-Squared............................................... 0.058718

Dependent Variable Mean................................. $5,300

Root Mean Square Error.................................. 14,256

Model Parameters........................................ 37

------------------------------------------------------------------------

Base Payment Categories Payment

Increment

------------------------------------------------------------------------

Male: Aged 0-34......................................... 1,255

Male: 35-44............................................. 1,940

Male: 45-54............................................. 2,654

Male: 55-59............................................. 3,350

Male: 60-64............................................. 3,970

Male: 65-69............................................. 2,792

[[Page 47511]]

Male: 70-74............................................. 3,702

Male: 75-79............................................. 4,738

Male: 80-84............................................. 5,617

Male: 85-89............................................. 6,562

Male: 90-94............................................. 7,209

Male: 95+............................................... 7,189

Female: 0-34............................................ 1,345

Female: 35-44........................................... 2,167

Female: 45-54........................................... 2,763

Female: 55-59........................................... 3,647

Female: 60-64........................................... 4,673

Female: 65-69........................................... 2,439

Female: 70-74........................................... 3,118

Female: 75-79........................................... 3,994

Female: 80-84........................................... 4,768

Female: 85-89........................................... 5,592

Female: 90-94........................................... 5,855

Female: 95+............................................. 5,466

------------------------------------------------------------------------

Other Demographic Factors

------------------------------------------------------------------------

Previously Disabled..................................... 2,381

Medicaid, Medicare Aged................................. 2,124

Medicaid, Medicare Disabled............................. 1,744

------------------------------------------------------------------------

PIP-DCGs

PIP-DCG 6............................................... 2,265

PIP-DCG 8............................................... 4,406

PIP-DCG 10.............................................. 5,829

PIP-DCG 12.............................................. 7,950

PIP-DCG 14.............................................. 9,946

PIP-DCG 18.............................................. 12,883

PIP-DCG 20.............................................. 16,346

PIP-DCG 23.............................................. 18,950

PIP-DCG 26.............................................. 21,881

PIP-DCG 29.............................................. 29,317

------------------------------------------------------------------------

Notes: PIP-DCG 4 is combined with the demographic factors, and includes

those with no hospitalizations, modified or certain low-cost

admissions. Diagnoses from hospital stays of less than two days are

not used in assigning PIP-DCGS.

Table 2.--Diagnoses (DxGroups) Included in Each PIP-DCG--Current Payment Model

----------------------------------------------------------------------------------------------------------------

PIP-DCG 6:

DxGroup.............................. 18 Cancer of Prostate/Testis/Male Genital Organs.

14 Breast Cancer.

PIP-DCG 8:

DxGroup.............................. 82 Acute Myocardial Infarction.

146 Pelvic Fracture.

145 Fractures of Skull/Face.

77 Valvular and Rheumatic Heart Disease.

86 Atrial Arrhythmia.

84 Angina Pectoris.

80 Coronary Atherosclerosis.

92 Precerebral Arterial Occlusion.

16 Cancer of Uterus/Cervix/Female Genital Organs.

79 Hypertension, Complicated.

36 Peptic Ulcer.

110 Asthma.

96 Aortic and Other Arterial Aneurysm.

153 Brain Injury.

1 Central Nervous System Infections.

39 Abdominal Hernia, Complicated.

64 Alcohol/Drug Dependence.

PIP-DCG 10:

DxGroup.............................. 109 Bacterial Pneumonia.

42 Gastrointestinal Obstruction/Perforation.

143 Vertebral Fracture Without Spinal Cord Injury.

21 Other Cancers.

4 Tuberculosis.

97 Thromboembolic Vascular Disease.

59 Schizophrenic Disorders.

[[Page 47512]]

11 Colon Cancer.

116 Kidney Infection.

83 Unstable Angina.

94 Transient Cerebral Ischemia.

81 Post-Myocardia Infarction.

150 Internal Injuries/Traumatic Amputations/Third Degree

Burns.

32 Pancreatitis/Other Pancreatic Disorders.

147 Hip Fracture.

158 Artificial Opening of Gastrointestinal Tract Status.

PIP-DCG 12:

DxGroup.............................. 91 Cerebral Hemorrhage.

93 Stroke.

56 Dementia.

98 Peripheral Vascular Disease.

41 Inflammatory Bowel Disease.

22 Benign Brain/Nervous System Neoplasm.

48 Rheumatoid Arthritis and Connective Tissue Disease.

49 Bone/Joint Infections/Necrosis.

19 Cancer of Bladder, Kidney, Urinary Organs.

45 Gastrointestinal Hemorrhage.

87 Paroxysmal Ventricular Tachycardia.

133 Cellulitis and Bullous Skin Disorders.

57 Drug/Alcohol Psychoses.

PIP-DCG 14:

DxGroup.............................. 66 Personality Disorders.

29 Adrenal Gland, Metabolic Disorders.

70 Degenerative Neurologic Disorders.

2 Septicemia/Shock.

144 Spinal Cord Injury.

58 Delirium/Hallucinations.

61 Paranoia and Other Psychoses.

63 Anxiety Disorders.

73 Epilepsy and Other Seizure Disorders.

10 Stomach, Small Bowel, Other Digestive Cancer.

12 Rectal Cancer.

26 Diabetes with Acute Complications/Hypoglycemic Coma.

113 Pleural Effusion/Pneumothorax/Empyema.

60 Major Depression.

PIP-DCG 18:

DxGroup.............................. 34 Cirrhosis, Other Liver Disorders.

72 Paralytic and Other Neurologic Disorders.

108 Gram-Negative/Staphylococcus Pneumonia.

111 Pulmonary Fibrosis and Bronchiectasis.

89 Congestive Heart Failure.

105 Chronic Obstructive Pulmonary Disease.

95 Atherosclerosis of Major Vessel.

13 Lung Cancer.

8 Mouth/Pharynx/Larynx/Other Respiratory Cancer.

PIP-DCG 20:

DxGroup.............................. 112 Aspiration Pneumonia.

76 Coma and Encephalopathy.

75 Polyneuropathy.

17 Cancer of Placenta/Ovary/Uterine Adnexa.

55 Blood/Immune Disorders.

PIP-DCG 23:

DxGroup.............................. 134 Decubitus and Chronic Skin Ulcers.

33 End-stage Liver Disorders.

9 Liver/Pancreas/Esophagus Cancer.

88 Cardio-Respiratory Failure and Shock.

27 Diabetes with Chronic Complications.

115 Renal Failure/Nephritis.

PIP-DCG 26:

DxGroup.............................. 7 Metastatic Cancer.

PIP-DCG 29:

DxGroup.............................. 3 HIV/AIDS.

15 Blood, Lymphatic Cancers/Neoplasms.

20 Brain/Nervous System Cancers.

----------------------------------------------------------------------------------------------------------------

[[Page 47513]]

Table 3.--Estimating Prospective Beneficiary Expenditures Mean Predicted Expenditures = $5300

--------------------------------------------------------------------------------------------------------------------------------------------------------

Demographic factors base PIP- + PIP-DCG + Other factors

--------------DCG---------------------------------------------------------------------------------------------------------------------------------------

Aged Population

--------------------------------------------------------------------------------------------------------------------------------------------------------

Male 65-69..................... $2792 PIP-DCG 6 $2265 Previously Disabled....................... $2381

Male 70-74..................... 3702 PIP-DCG 8 4406 Medicaid, Medicare Aged................... 2124

Male 75-79..................... 4738 PIP-DCG 10 5829

Male 80-84..................... 5617 PIP-DCG 12 7950

Male 85-89..................... 6562 PIP-DCG 14 9946

Male 90-94..................... 7209 PIP-DCG 18 12,883

Male 95+....................... 7189 PIP-DCG 20 16,346

Female 65-69................... 2439 PIP-DCG 23 18,950

Female 70-74................... 3118 PIP-DCG 26 21,881

Female 75-79................... 3944 PIP-DCG 29 29,317

Female 80-84................... 4768

Female 85-89................... 5592

Female 90-94................... 5855

Female 95+..................... 5466

--------------------------------------------------------------------------------------------------------------------------------------------------------

Disabled Population

--------------------------------------------------------------------------------------------------------------------------------------------------------

Male 0-34...................... 1255 PIP-DCG 6 2265 Medicaid, Medicare Disabled............... 1744

Male 34-44..................... 1940 PIP-DCG 8 4406

Male 45-54..................... 2654 PIP-DCG 10 5829

Male 55-59..................... 3350 PIP-DCG12 7950

Male 60-64..................... 3970 PIP-DCG 14 9946

Female 0-34.................... 1345 PIP-DCG 18 12,883

Female 34-44................... 2167 PIP-DCG 20 16,346

Female 45-54................... 2763 PIP-DCG 23 18,950

Female 55-59................... 3647 PIP-DCG 26 21,881

Female 60-64................... 4673 PIP-DCG 29 29,317

--------------------------------------------------------------------------------------------------------------------------------------------------------

(Sec. 4002 of the Balanced Budget Act of 1997 (Public Law 105-33)

Dated: August 26, 1998.

Nancy-Ann Min DeParle,

Administrator, Health Care Financing Administration.

Dated: September 1, 1998.

Donna E. Shalala,

Secretary.

[FR Doc. 98-24085 Filed 9-2-98; 4:10 pm]

BILLING CODE 4120-01-P