Medicare Program; Prospective Payment System and Consolidated Billing for Skilled Nursing Facilities-Update

AGENCY:

Health Care Financing Administration (HCFA), HHS.

ACTION:

Notice of proposed rulemaking.

SUMMARY:

This proposed rule sets forth updates to the payment rates used under the prospective payment system (PPS) for skilled nursing facilities (SNFs), for fiscal year 2001. Furthermore, it specifically proposes changes to the SNF PPS case-mix methodology. Annual updates to the PPS rates are required by section 1888(e) of the Social Security Act, as amended by the Medicare, Medicaid and State Child Health Insurance Program Balanced Budget Refinement Act of 1999, related to Medicare payments and consolidated billing for SNFs. In addition, this proposed rule sets forth certain conforming revisions to the regulations that are necessary in order to implement amendments made to the Act by section 103 of the Medicare, Medicaid and State Child Health Insurance Program Balanced Budget Refinement Act of 1999.

DATES:

We will consider comments if we receive them at the appropriate address, as provided below, no later than 5 p.m. on June 9, 2000.

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-1112-P, P.O. Box 8013, Baltimore, MD 21244-8013.

If you prefer, you may deliver your written comments (1 original and 3 copies) to one of the following addresses:

Room 443-G, Hubert H. Humphrey Building, 200 Independence Avenue, SW., Washington, DC 20201, or

Room C5-15-03, 7500 Security Boulevard, Baltimore, MD 21244-8150.

Because of staffing and resource limitations, we cannot accept comments by facsimile (FAX) transmission. In commenting, please refer to file code HCFA-1112-P. 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 443-G of the Department's office at 200 Independence Avenue, SW., Washington, DC, on Monday through Friday of each week from 8:30 to 5 p.m. (phone: (202) 690-7061).

FOR FURTHER INFORMATION CONTACT:

Dana Burley, (410) 786-4547 or Sheila Lambowitz, (410) 786-7605 (for information related to the case-mix classification methodology).

John Davis, (410) 786-0008 (for information related to the Wage Index).

Bill Ullman, (410) 786-5667 (for information related to consolidated billing).

Steve Raitzyk, (410) 786-4599 (for information related to the facility-specific transition rates).

Bill Ullman, (410) 786-5667 and Susan Burris (410) 786-6655 (for general information).

SUPPLEMENTARY INFORMATION:

Copies: To order copies of the Federal Register containing this document, send your request to: New Orders, Superintendent of Documents, P.O. Box 371954, Pittsburgh, PA 15250-7954. Please specify the date of the issue requested and enclose a check or money order payable to the Superintendent of Documents, or enclose your Visa or Master Card number and expiration date. Credit card orders can also be placed by calling the order desk at (202) 512-1800 (or toll free at 1-888-293-6498) or by faxing to (202) 512-2250. The cost for each copy is $8. As an alternative, you can view and photocopy the Federal Register document at most libraries designated as Federal Depository Libraries and at many other public and academic libraries throughout the country that receive the Federal Register.

To assist readers in referencing sections contained in this document, we are providing the following table of contents.

Table of Contents

I. Background

A. Current System for Payment of Skilled Nursing Facility Services Under Part A of the Medicare Program

B. Requirements of the Balanced Budget Act of 1997 for Updating the Prospective Payment System for Skilled Nursing Facilities

C. The Medicare, Medicaid and State Child Health Insurance Program (SCHIP) Balanced Budget Refinement Act of 1999

D. Skilled Nursing Facility Prospective Payment—General Overview

1. Payment Provisions—Federal Rates

2. Payment Provisions—Transition Period

3. Payment Provisions—Facility-Specific Rate

II. Update of Payment Rates Under the Prospective Payment System for Skilled Nursing Facilities

A. Federal Prospective Payment System

1. Cost and Services covered by the Federal Rates

2. Methodology Used for the Calculation of the Federal Rates

B. Case-Mix Adjustment and Options

C. Wage Index Adjustment to Federal Rates

D. Updates to the Federal Rates

E. Relationship of RUG-III Classification System to Existing Skilled Nursing Facility Level-of-Care Criteria

III. Three-Year Transition Period

IV. The Skilled Nursing Facility Market Basket Index

A. Facility-Specific Rate Update Factor

B. Federal Rate Update Factor

V. Consolidated Billing

VI. Provisions of the Proposed Rule

VII. Collection of Information Requirements

VIII. Response to Comments

IX. Regulatory Impact Analysis

A. Background

B. Impact of this Proposed Rule

X. Federalism

Regulations Text

Technical Appendix A

A. Creation of the Analytic Sample

B. Characteristics of the Sample

C. Test and Validation Samples

D. Creation of Measure of Non-Therapy Ancillary Charges from SNF Claims

1. Cost-to-Charge Multiplier

E. Analysis and Findings—RUG-III Refinements

1. Costs for Beneficiaries Who Qualify for Both Extensive Services and Rehabilitation

2. Non-Therapy Ancillary Index Models

F. Model Performance

1. RUG-III CMI Adjustment

2. RUG-III (proposed, version 2001)

3. Weighted Index Model (WIM1)

4. Weighted Index Model 2 (WIM2)

5. Unweighted Index Model (UWIM)

G. RUG-III Medications Data

1. Creation of MDS-Based Cost Measures

2. RUG-Based Imputation Method

3. State and Year-Based Imputation Method

In addition, because of the many terms to which we refer by abbreviation in this rule, we are listing these abbreviations and their corresponding terms in alphabetical order below:

ADL—Activity of Daily Living

BBA—Balanced Budget Act of 1997

BBRA—Balanced Budget Refinement Act of 1999

BLS—(U.S.) Bureau of Labor Statistics

CPI—Consumer Price Index

HCFA— Health Care Financing Administration

HCPCS—HCFA Common Procedure Coding System

IFC—Interim Final Rule with Comments

MDS—Minimum Data Set

MSA—Metropolitan Statistical Area

PPI—Producer Price Index

PPS—Prospective Payment System

PRM—Provider Reimbursement Manual

RUG—Resource Utilization Group

SCHIP—State Child Health Insurance Program

SNF—Skilled Nursing Facility

I. Background

A. Current System for Payment of Skilled Nursing Facility Services Under Part A of the Medicare Program

Section 4432 of the Balanced Budget Act of 1997 (BBA) (Pub. L. 105-33) mandated the implementation of a per diem prospective payment system (PPS) for skilled nursing facilities (SNFs), covering all costs (routine, ancillary, and capital) of covered SNF services furnished to beneficiaries under Part A of the Medicare program, effective for cost reporting periods beginning on or after July 1, 1998. The SNF PPS payment methodology features a case-mix adjustment that utilizes data from the comprehensive assessment process required for every SNF beneficiary in order to group them clinically in terms of their degree of resource intensity. The case-mix adjustment is designed to ensure that the amount of the PPS per diem payment is appropriate to the individual beneficiary's actual condition, and is sufficient to purchase the full range of care and services that a beneficiary with a particular clinical profile would typically be expected to require. We are setting forth this proposed rule in accordance with section 1888(e)(4)(H)(ii) of the Social Security Act (the Act), which requires us to publish each year in the Federal Register any changes in the case-mix classification system that we use to make the case-mix adjustment. Although we are not proposing any other changes in the overall PPS payment methodology at present, we are nonetheless including a detailed discussion of the overall payment methodology in section I.C. below, in order to provide a context for the proposed changes to the case-mix classification system. In addition, we are incorporating revisions based on the Medicare, Medicaid and State Child Health Insurance Program (SCHIP) Balanced Budget Refinement Act of 1999 (BBRA). Major elements of the system were implemented in an interim final rule that was published in the Federal Register on May 12, 1998 (63 FR 26252), and in a final rule that was published in the Federal Register on July 30, 1999 (64 FR 41644). These elements are discussed in greater detail in section I.C. below, and include:

• Rates: Per diem Federal rates were established for urban and rural areas using allowable costs from fiscal year (FY) 1995 cost reports. These rates also included an estimate of the cost of services that, before July 1, 1998, had been paid under Part B but furnished to Medicare beneficiaries in a SNF during a Part A covered stay. Rates are case-mix adjusted using a refined classification system (Resource Utilization Groups, version III (RUG-III)) based on beneficiary assessments (using the Minimum Data Set (MDS) 2.0). The proposed refinement to the RUG classification system is based on critical analysis which examined various options to account more precisely for the variation in non-therapy ancillary services in our payments and the care needs of medically complex patients. The proposed RUG refinement includes the addition of new categories and incorporation of an ancillary index, as discussed in further detail in section II.B. In addition, the Federal rates are adjusted by the hospital wage index to account for geographic variation in wages. At this time, data for the FY 2001 hospital wage index is not yet available; therefore, the index applied in this proposed rule is the same index used in the July 30, 1999 update notice. We will be updating the wage index in the final rule using the latest hospital wage data. Further, the rates are adjusted annually using an SNF market basket index. Lastly, as a result of section 101 of the BBRA, for SNF services furnished on or after April 1, 2000, and before the later of October 1, 2000, or implementation by the Secretary of Health and Human Services of a refined RUG system, per diem adjusted payments are increased by 20 percent for 15 RUGs falling under categories for Extensive Services, Special Care, Clinically Complex, High Rehabilitation and Medium Rehabilitation. This 20 percent increase serves solely as a temporary, interim adjustment to the payment rates and RUG-III classification system as published in the final rule of July 30, 1999, until we have had the opportunity to implement the case-mix refinements proposed in this rule. At that point, the temporary adjustment afforded by the 20 percent increase will no longer be applicable, as payment will be made in accordance with the newly-refined RUGs. The RUG-III groups to which this adjustment applies are: SE3, SE2, SE1, SSC, SSB, SSA, CC2, CC1, CB2, CB1, CA2, CA1, RHC, RMC and RMB. In addition, for FY 2001 and FY 2002, the adjusted Federal per diem payment to a facility is increased by 4 percent in each year, calculated exclusive of the 20 percent RUG rate increase.
• Transition: The SNF PPS includes a 3-year, phased transition that blends a facility-specific payment rate with the Federal case-mix adjusted rate. The blend used changes for each cost reporting period after a facility migrates to the new system. For most facilities, the facility-specific rate is based on allowable costs from FY 1995. As a result of section 102 of the BBRA of 1999, SNFs may elect immediate transition to the Federal rate on or after December 15, 1999 for cost reporting periods beginning on or after January 1, 2000. There is no such election for cost reporting periods beginning before January 1, 2000. SNFs may elect immediate transition up to 30 days after the start of their cost reporting period.
• Coverage: The PPS statute did not change Medicare's fundamental requirements for SNF coverage. However, because RUG-III classification is based, in part, on the beneficiary's need for skilled nursing care and therapy, we have attempted where possible to coordinate claims review procedures with the outputs of beneficiary assessment and RUG-III classifying activities. For example, we believe that when an initial Medicare required (5-day) assessment, properly completed, places the beneficiary in one of the upper RUG-III classifications that we designate as representing a covered level of SNF care (see section II.E. of this preamble), this provides the basis for us to assume that the beneficiary needed such care upon admission and at least up until the assessment reference date for the initial Medicare-required assessment. We will, however, continue to make individual review determinations for claims of those individuals who classify in one of the lower RUG-III categories.
• Consolidated Billing: The statute includes a billing provision that requires a SNF to submit consolidated Medicare bills for its beneficiaries for virtually all services that are covered under either Part A or Part B. The statute excludes a small list of services (primarily those of physicians and certain other types of practitioners). As discussed later in this preamble, section 103 of the BBRA has identified certain additional services for exclusion, effective April 1, 2000.

As noted above, an interim final rule implementing the SNF PPS was published in the Federal Register on May 12, 1998, for which the comment period was initially scheduled to close on July 13, 1998. A subsequent notice extended the public comment period for an additional 60 days (July 13, 1998, (63 FR 37498)), and a second notice reopened the comment period for another 30 days (November 27, 1998 (63 FR 65561)). In addition, a correction notice was published October 5, 1998 (63 FR 53301) that made a number of minor technical and editorial corrections to the interim final rule. In the July 30, 1999, final rule we responded to the public comments received on the interim final rule and made a number of modifications in the regulation. This final rule was followed by a correction notice published on November 4, 1999 (64 FR 60122), which made a technical correction to the final rule's preamble. Also on July 30, 1999, we issued an update notice (64 FR 41684), followed by a correction notice published on October 5, 1999 (64 FR 54031). We have also issued several Program Memoranda on claims processing and billing under the SNF PPS that are available on the SNF PPS home page at the HCFA website on the Internet, at the following location: <www.hcfa.gov/Medicare/snfpps.htm>

B. Requirements of the Balanced Budget Act of 1997 for Updating the Prospective Payment System for Skilled Nursing Facilities

As described above, section 1888(e)(4)(H) of the Act requires that we publish in the Federal Register:

1. The unadjusted Federal per diem rates to be applied to days of covered SNF services furnished during the FY.

2. The case-mix classification system to be applied with respect to these services during the FY.

3. The factors to be applied in making the area wage adjustment with respect to these services.

In addition, in the July 30, 1999 final rule, we indicated that we would announce any changes to the guidelines for Medicare level of care determinations related to Part A SNF services or to the RUG-III classifications.

This proposed rule updates the rates as mandated by the Medicare statute.

C. The Medicare, Medicaid and State Child Health Insurance Program (SCHIP) Balanced Budget Refinement Act of 1999

As a result of enactment of the BBRA, there are several new provisions that result in adjustments to the PPS for SNFs. The following highlights the major provisions involving the PPS for SNFs:

Temporary Increase in Payment for Certain High Cost Residents

As noted previously, section 101 of the BBRA provides for a temporary, 20 percent increase in the per diem adjusted payment rates for 15 specified RUGs, falling under categories for Extensive Services, Special Care, Clinically Complex, High Rehabilitation and Medium Rehabilitation. The specific RUG-III groups to which this adjustment applies are: SE3, SE2, SE1, SSC, SSB, SSA, CC2, CC1, CB2, CB1, CA2, CA1, RHC, RMC, and RMB. The statute provides that the 20 percent increase takes effect with SNF services that are furnished on or after April 1, 2000, and continues until the later of October 1, 2000, or implementation by the Secretary of a refined RUG system. Thus, the 20 percent increase serves solely as a temporary, interim adjustment to the payment rates and RUG-III classification system as published in the final rule of July 30, 1999, until we have implemented the case-mix refinements that we now propose elsewhere in this document, which we expect to accomplish by October 1, 2000. Once we have implemented the case-mix refinements, the temporary adjustment afforded by the 20 percent increase will no longer be applicable, as we will then make payment in accordance with the newly-refined RUGs.

For FY 2001 and FY 2002, section 101 of the BBRA also provides for an across-the-board increase in the adjusted Federal per diem payment rates by 4 percent in each year, calculated exclusive of the 20 percent RUG rate increase discussed above. Unlike the 20 percent increase, which is targeted at certain particular RUG-III groups, this 4 percent increase will apply equally to all RUG groups.

Election For Immediate Transition to Federal Rate

As noted earlier, under section 102 of the BBRA, all SNFs may now elect to bypass the transition and be paid based upon 100 percent of the Federal rate. This election applies to cost reporting periods beginning on or after January 1, 2000. There is no such election for cost reporting periods beginning before January 1, 2000. SNFs may make this election beginning on or after December 15, 1999 and up to 30 days after the start of their cost reporting periods. An election to bypass the transition is effective for all subsequent periods and cannot be rescinded once it is effective. Further information can be found in Program Memorandum A-99-53.

Special Payment Adjustment for Certain SNFs

Section 155 of the BBRA provides that PPS payments to certain SNF providers located in Baldwin or Mobile County, Alabama, will be based on 100 percent of their facility specific rates for cost reporting periods that begin in FY 2000 or FY 2001. In addition, it requires that the facility specific portion of their payment rate be calculated using data from their cost reporting period beginning in FY 1998. In order to be eligible for this special payment, a SNF must meet the following criteria: began participation in the Medicare program before January 1, 1995; have at least 80 percent of the total inpatient days of the facility in the cost reporting period beginning in FY 1998 comprised of persons entitled to Medicare; and, be located in Baldwin or Mobile County, Alabama.

Special SNF PPS Payment Provisions for SNFs with Certain Types of Patient Populations

Section 105 of the BBRA adds paragraph (12) to section 1888(e) of the Act and permits certain SNFs to receive 50 percent of the facility specific rate and 50 percent of the Federal per diem rate, effective from November 29, 1999, until September 30, 2001. In order to be eligible, a SNF must: have been certified as an SNF under Medicare prior to July 1, 1992; be a hospital-based facility; and, in the cost reporting period beginning in FY 1998, have had a patient population, eligible for Part A benefits, of which at least 60 percent were “immuno-compromised secondary to an infectious disease,” with “specific diagnoses specified by the Secretary.” The statute gives the Secretary the authority to specify the diagnosis associated with this provision, and we believe the legislative history provides some guidance concerning the application of this provision. The House Ways and Means Committee report (H. Rep. 106-436, Part 1 at 47) indicates that this provision is directed at facilities that serve “* * * very specialized patients * * * whose medical conditions are not well-accounted for in the RUG classification system.” The Senate Finance Committee Report (S. Rep. 106-199 at 8) indicates the need to study “* * * alternative payment methods for skilled nursing facilities that specialize in providing care to extremely high cost, chronically ill populations * * *” such as “a facility that exclusively specializes in caring for AIDS patients * * *” In light of this general Congressional intent, we believe that the scope of this provision should be limited and propose that this provision be applied to human immunodeficiency virus (HIV) as coded in ICD-9-CM with the following code: 042.

Provision for Part B Add-Ons for Facilities Participating in the Nursing Home Case-Mix and Quality (NHCMQ) Demonstration Project

Under prior law, section 1888(e)(3) of the Act provided for an add-on to the payment rates for Part B services furnished during the course of a Part A covered stay for those facilities that did not participate in the demonstration that preceded SNF PPS. However, the Act did not provide for a similar add-on for facilities that did participate in the demonstration project. Therefore, section 104 of the BBRA amended section 1888(e)(3) to provide that SNFs that had participated in the Nursing Home Case Mix and Quality Demonstration (NHCMQ) project are eligible for the inclusion of a Part B add-on amount in their facility specific PPS rates. This provision is effective as if included in the enactment of the BBA and, therefore, applies to all cost reporting periods subject to the PPS transition.

For the purpose of computing facility specific rates, the base year for providers participating in the NHCMQ demonstration project is calendar year 1997 rather than FY 1995 (which is the base year for SNFs not participating in the demonstration project). Therefore, the Part B add-on amounts for the demonstration SNFs will be calculated using data from the appropriate periods in 1997. Because of the time period necessary for us to compute these amounts, existing Part B data from 1995 will be updated for inflation and used as the bases for payment on an interim basis until we can develop the final amounts using the 1997 data, at which point earlier payments will be adjusted to reflect the correct data.

Exclusion of Certain Additional Services from the SNF PPS Bundle and Consolidated Billing

The original SNF PPS legislation in the BBA identified several service categories that were excluded from the SNF consolidated billing requirement, as well as from the bundled Part A payment made under the SNF PPS itself. Effective with services furnished on or after April 1, 2000, section 103(a) of the BBRA has amended section 1888(e)(2)(A) to exclude certain additional types of services from the consolidated billing requirement, thus allowing these services to be billed separately to Part B. Section 103(b) of the BBRA has also amended section 1888(e)(4)(G) to provide for a corresponding proportional reduction in Part A SNF payments, beginning with FY 2001. We discuss these additional excluded service categories in section V. of this preamble, on consolidated billing.

D. Skilled Nursing Facility Prospective Payment—General Overview

The Medicare SNF PPS was implemented for cost reporting periods beginning on or after July 1, 1998. Under the PPS, SNFs are paid through per diem prospective case-mix adjusted payment rates applicable to all covered SNF services. These payment rates cover all the costs of furnishing covered skilled nursing services (that is, routine, ancillary, and capital-related costs) other than costs associated with approved educational activities. Covered SNF services include posthospital SNF services for which benefits are provided under Part A and all items and services that, before July 1, 1998, had been paid under Part B (other than physician and certain other services specifically excluded under the BBA) but furnished to Medicare beneficiaries in a SNF during a Part A covered stay. (For a complete discussion of these provisions, see the May 12, 1998 interim final rule (63 FR 26252)).

1. Payment Provisions—Federal Rate

The statute sets forth a fairly prescriptive methodology for calculating the amount of payment under the SNF PPS. The PPS utilizes per diem Federal payment rates based on mean SNF costs in a base year updated for inflation to the first effective period of the PPS. We developed the Federal payment rates using allowable costs from hospital-based and freestanding SNF cost reports for reporting periods beginning in FY 1995. The data used in developing the Federal rates also incorporate an estimate of the amounts that would be payable under Part B for covered SNF services to individuals who were receiving Part A covered services in an SNF. In developing the rates for the initial period, we updated costs to the first effective year of PPS (15-month period beginning July 1, 1998) using a SNF market basket index, and standardized for facility differences in case-mix and for geographic variations in wages. Providers that received “new provider” exemptions from the routine cost limits were excluded from the database used to compute the Federal payment rates. In addition, costs related to payments for exceptions to the routine cost limits were excluded from the database used to compute the Federal rates. In accordance with the formula prescribed in the BBA, we set the Federal rates at a level equal to the weighted mean of freestanding costs plus 50 percent of the difference between the freestanding mean and weighted mean of all SNF costs (hospital-based and freestanding) combined. We compute and apply separately the payment rates for facilities located in urban and rural areas. In addition, we adjust the portion of the Federal rate attributable to wage related costs by a wage index.

The Federal rate also incorporates adjustments to account for facility case-mix using a classification system that accounts for the relative resource utilization of different patient types. This classification system, RUG-III, utilizes beneficiary assessment data (from the Minimum Data Set or MDS) completed by SNFs to assign beneficiaries into one of 178 groups. The May 12, 1998 interim final rule (63 FR 26252) has a complete and detailed description of the original (44 group) RUG-III classification system. A detailed discussion of the proposed changes to the RUG classification system is found in Section II.B. of this proposed rule.

The Federal rates reflected in this notice update the rates in the July 30, 1999 update notice (64 FR 41684) by a factor equal to the SNF market basket index minus 1 percentage point. According to section 1888(e)(4)(E)(ii) of the Act, for FYs 2001 and 2002, we will update the rate by adjusting the current rates by the SNF market basket change minus 1 percentage point. For subsequent FYs, we will adjust the rates by the applicable SNF market basket change.

2. Payment Provisions—Transition Period

Beginning with a provider's first cost reporting period beginning on or after July 1, 1998, there is a transition period covering three cost reporting periods. During the transition period, SNFs receive a payment rate comprising a blend between the Federal rate and a facility-specific rate based on each facility's FY 1995 cost report. Under section 1888(e)(2)(E)(ii) of the Act, SNFs that received their first payment from Medicare on or after October 1, 1995 receive payment according to the Federal rates only.

For SNFs subject to transition, the composition of the blended rate varies depending on the year of transition. For the first cost reporting period beginning on or after July 1, 1998, we make payment based on 75 percent of the facility-specific rate and 25 percent of the Federal rate. In the next cost reporting period, the rate consists of 50 percent of the facility-specific rate and 50 percent of the Federal rate. In the following cost reporting period, the rate consists of 25 percent of the facility-specific rate and 75 percent of the Federal rate. For all subsequent cost reporting periods, we base payments entirely on the Federal rates.

As noted earlier, in accordance with section 102 of the BBRA, SNFs that would otherwise be subject to the statutory three-year, phased transition from facility-specific to Federal rates, may elect to bypass the transition and go directly to the full Federal rate. This amendment applies to elections made on or after December 15, 1999, except that no election will be effective for a cost reporting period beginning before January 1, 2000; an election is effective for a cost reporting period beginning no earlier than 30 days before the date of the election.

3. Payment Provisions—Facility-Specific Rate

For most facilities, we compute the facility-specific payment rate utilized for the transition using the allowable costs of SNF services for cost reporting periods beginning in FY 1995 (cost reporting periods beginning on or after October 1, 1994 and before October 1, 1995). Included in the facility-specific per diem rate is an estimate of the amount that would be payable under Part B for covered SNF services furnished during FY 1995 to individuals who were beneficiaries of the facility and receiving Part A covered services. The facility-specific rate, in contrast to the Federal rates, includes amounts paid to SNFs for exceptions to the routine cost limits. In addition, we also take into account “new provider” exemptions from the routine cost limits, but only to the extent that routine costs do not exceed 150 percent of the routine cost limit.

We update the facility-specific rate for each cost reporting period after FY 1995 to the first cost reporting period beginning on or after July 1, 1998 (the initial period of the PPS) by a factor equal to the SNF market basket percentage increase minus 1 percentage point. For FYs 1998 and 1999, we updated this rate by a factor equal to the SNF market basket increase minus 1 percentage point, and in each subsequent year, we will update it by the applicable SNF market basket increase.

Appeals Rights

In enacting SNF PPS, Congress imposed limitations on the rights of SNFs to appeal their new payment rates (section 1888(e)(8) of the Social Security Act). Similar to the hospital PPS, the new SNF system begins with a transition period, wherein a portion of the payment rates (that is, the facility-specific rate) is based upon the facilities' costs in a base period (cost reporting periods beginning in 1995). The facility-specific portion of the rate phases out over the course of a three year cost reporting transition period, after which the SNFs will be paid on a fully Federal rate. The statutory language removes the Federal portion of the rate from administrative and judicial review, while allowing for a limited review of the facility-specific portion of the rate related to an SNFs Part A historical costs from the 1995 base year. The language of the interim final rule with comment and the Medicare Provider Reimbursement Manual (PRM) contemplate situations where adjustments are made to the reimbursement amounts allowable in the base year that are used to set the facility-specific portion of a provider's PPS rate. Adjustments may be made in the cost report settlement process and/or providers may have appealed specific cost report adjustments. Where adjustments are made to the base year costs either through final settlement of the cost report or as a result of an appeal of the base year Notice of Program Reimbursement (NPR), such adjustments may be applied to the facility-specific portion of the PPS rate for any cost years that are open or are within the time periods subject to reopening under the regulations at 42 CFR 405.1885. Additionally, providers may challenge the facility-specific portion of their rates by appealing the facility-specific rate notice they receive from their fiscal intermediary before the start of SNF PPS. The fiscal intermediaries will apply any adjustments resulting from a successful challenge to this rate notice to all open transition years. Providers may also challenge their facility-specific rates by appealing their transition year NPRs. Adjustments obtained through a NPR challenge will only be applied to the year under appeal. Moreover, in accordance with the judicial review prohibitions contained in section 1888(e)(8)(B) of the Act, all reviews of facility-specific rates are limited to challenges relating to specific Medicare Part A costs in the base year.

II. Update of Payment Rates Under the Prospective Payment System for Skilled Nursing Facilities

A. Federal Prospective Payment System

This rule sets forth a proposed schedule of Federal prospective payment rates applicable to Medicare Part A SNF services beginning October 1, 2000. The schedule incorporates per diem Federal rates designed to provide Part A payment for all costs of services furnished to a beneficiary of an SNF during a Medicare-covered stay.

1. Cost and Services Covered by the Federal Rates

The Federal rates apply to all costs (that is, routine, ancillary, and capital related costs) of covered SNF services other than costs associated with operating approved educational activities as defined in § 413.85. Under section 1888(e)(2) of the Act, covered SNF services include posthospital SNF services for which benefits are provided under Part A (the hospital insurance program), as well as all items and services (other than those services excluded by statute) that, before July 1, 1998, were paid under Part B (the supplementary medical insurance program) but furnished to Medicare beneficiaries in a SNF during a Part A covered stay. (These excluded service categories are discussed in greater detail in section V.B.2. of the May 12, 1998 interim final rule (63 FR 26295-97). Also, as mentioned previously, section 103 of the BBRA has identified certain additional types of services for exclusion from the SNF PPS bundle, and has provided for a corresponding proportional reduction in Part A SNF payments beginning with FY 2001.).

2. Methodology Used for the Calculation of the Federal Rates

The methodology to compute the unadjusted Federal rates incorporates several changes since we published the final rule on July 30, 1999 (64 FR 41684). First, to facilitate the incorporation of our proposed refinement to the case mix classification system, we are creating a new component of the payment rates to account for non-therapy ancillary services. This component is being created by moving the non-therapy ancillary costs used in establishing the nursing case-mix component of the payment rates to a separate component. For the payment rates associated with urban areas, 43.4 percent of the nursing case mix component is related to non-therapy ancillary services (including Part B services). For the payment rates associated with rural areas, 42.7 percent of the nursing case mix component is related to non-therapy ancillary services (including Part B services). These percentages were previously identified in a Federal Register notice dated November 27, 1998 (63 FR 65561). This new component of the payment rates is presented in Tables 1 and 2 of this proposed rule.

In addition, in accordance with section 103 of the BBRA, the Federal rates will be adjusted to reflect the exclusion of certain items and services from consolidated billing, as explained previously. The complexity and time necessary for computing the numeric adjustment itself does not allow us to present it in this proposed rule. However, we describe the general methodology that we plan to use later in this preamble (in the discussion of the PPS Rate Tables). As required by the statute, the rates are updated using the latest market basket percentage minus 1 percentage point. For a complete description of the multi-step process, see the May 12, 1998 interim final rule. In addition, based on section 101 of the BBRA, we have provided for a 4 percent increase in the adjusted Federal rate for FY 2001. This 4 percent adjustment is not reflected in the rate tables (Tables 1, 2, 5, and 6 of this proposed rule). In accordance with the statute, it is applied after all adjustments (wage and case-mix). See the example in Section III; Table 9, of this proposed rule.

The SNF market basket is used to adjust each per diem component of the Federal rates forward to reflect cost increases occurring between the midpoint of the Federal FY beginning October 1, 1999 and the midpoint of the Federal FY beginning October 1, 2000, and ending September 30, 2001, to which the payment rates apply. In accordance with section 1888(e)(4)(B) of the Act, the payment rates are updated between FY 2000 and FY 2001 by a factor equivalent to the annual market basket index percentage increase minus 1 percentage point. This factor is equal to 1.01833. Tables 1 and 2 below reflect the updated components of the unadjusted Federal rates.

B. Case-Mix Adjustment and Options

As required by the BBA, HCFA must publish the SNF PPS case-mix classification methodology applicable for the next Federal FY before August 1 of each year. This proposed rule discusses options for refinements to the RUG-III system, describes ongoing research and analyses, shares the initial results that we propose be incorporated into the Medicare PPS system effective October 1, 2000, and solicits comments from all interested parties. During the next 60 days, comments will be reviewed and considered, additional analyses will be conducted, and final decisions will be made on the need for, and types of, RUG-III refinements to be implemented. A final rule will then be promulgated before August 1, 2000.

Research Goals

We commissioned a study to review the RUG-III classification system with particular emphasis on the care needs of medically complex Medicare beneficiaries and the variation in non-therapy ancillary services within RUG-III categories. This project is a major priority for us, the provider industry, and others. The initial research identified potential refinements to the system that we propose to implement effective October 1, 2000.

A key part of this research was the exploration of potential refinements to the Extensive Services category. Previous research showed that the Extensive category is associated with the highest per diem non-therapy ancillary costs of any of the RUG-III categories. The research also indicated that, while the Extensive Services category did capture a disproportionate share of high cost beneficiaries, there was considerable variance in costs within this category as well as within other categories. In the current project, additional studies were conducted to extend the analysis of non-therapy ancillary costs and within-group variance to other RUG-III categories.

The researchers focused on the following analyses to identify options, and the results were used to develop the proposed RUG-III refinements discussed in this rule:

1. Evaluate the ability of the current RUG-III system to predict variance in drug, respiratory or other non-therapy ancillary costs.

2. Evaluate the ability of specific MDS items to predict variance in non-therapy ancillary costs, and identify the MDS items most closely associated with differences in non-therapy ancillary costs.

3. Design/test potential refinements to the RUG-III methodology.

A detailed description of the methodology used to conduct these analyses is included in the Technical Appendix A to this proposed rule.

Data Sources

Since ensuring the equity and accuracy of the SNF PPS has been, and continues to be, a major HCFA priority, the studies were initiated shortly after the introduction of the new payment system. In fact, the research was conducted before actual PPS claims and acuity data became available. For this reason, the analyses described here were conducted using a large cross-linked research data base that included clinical assessment data collected from the Federally-mandated MDS, drug information, our claims data, and organizational data on nursing home providers. The data sets used in the analyses are described below:

Minimum Data Set (MDS)

MDS data were collected from 6 states: Kansas, Maine, Mississippi, Ohio, South Dakota, and Texas. (As explained in Technical Appendix A, we were unable to utilize data from a seventh state, New York, due to that state's use of an all-inclusive payment rate.) These states were selected because the MDS data had been collected and used for rate-setting purposes prior to the start of the Medicare SNF PPS (either through the HCFA Case-Mix Demonstration Project or for state Medicaid payment systems), and provided a greater number of MDS records over a longer period of time than available from any other source. In addition, previous demonstration project reliability studies and state validation activities indicated a generally high level of data accuracy.

MDS data used in this study were for calendar years 1995, 1996 and 1997 (except for Texas, where data were only available for 1997), and included assessments for Medicare beneficiaries, Medicaid recipients and private pay patients. While some states required MDS assessments for all beneficiaries admitted to the SNF regardless of the length of stay, most of the assessments were prepared following the Federal guidelines in effect at the time; that is, assessments required by day 14 of the SNF admission.

MDS Drug Data

Facilities participating in the HCFA Case-Mix Demonstration project submitted medications data as part of their MDS assessments. In addition, several of the states, including Maine, South Dakota, and Ohio, required the medications data with every MDS, regardless of payor source. The medications reported on the MDSs were collected from seven states, the six states used for this study, plus New York (see Technical Appendix A for details on the use of New York data).

Up to 18 medications administered during the assessment reference period can be reported on an MDS record. The MDS drug data were cleansed and verified through a combination of manual examination (by either a clinical pharmacist or physician) and computerized reclassification of National Drug Codes (NDC). The data were then ordered into therapeutic groups for easier analysis.

SNF Claims

All SNF Medicare claims spanning the years 1995 through 1997 were downloaded from the HCFA Data Center and matched to MDS files. The files were constructed so that there are multiple observations per SNF stay if multiple MDS assessments were performed.

Staff Time Measurement (STM) Study Data

This analysis incorporated HCFA STM Study data (combined 1995 and 1997). The May 12, 1998 interim final rule described the STM Study, and the methodology used to incorporate the STM data into Medicare PPS rate-setting. These data were used to impute staff time costs for the observations used in this study.

On-Line Survey Certification and Reporting System (OSCAR) Data

The OSCAR data provide facility-level information, such as the results from annual survey inspections and information regarding facility type. OSCAR data from 1991 through 1998 were linked serially into a longitudinal file. The analytic database constructed for this research has been merged to this longitudinal OSCAR file through the linking of facility identifiers, using the OSCAR information from the survey dates closest to the MDS assessment data.

Case Mix Research Findings

While maintaining the general structure of RUG-III, we found that the two most viable ways to refine the system are by adding new categories and end splits to the system, and by developing a new index system to reflect the variation of non-therapy ancillary service costs. Adoption of these refinements will add additional groups to the case-mix system, somewhat increasing its complexity. This proposed change also may introduce some initial uncertainty for providers, who would have to become familiar with the refined system and modify existing operational and support systems.

In evaluating a particular change, we first identified the drawbacks of that change (for example, added complexity of the RUG-III model and time and effort required by providers, contractors, and beneficiaries to assimilate the change). Then, to evaluate the overall desirability of the potential change, we weighed these drawbacks against the benefits, such as the expected improvement in payment and clinical accuracy. In addition, we evaluated potential refinements in terms of possible incentives and disincentives related to access, quality and cost-effectiveness of SNF care. We incorporated this analysis into our evaluation of potential RUG-III refinements.

After careful review and extensive analysis, we then identified several possible RUG-III refinements that will improve the accuracy of SNF PPS payments. One such refinement is the development of new categories for beneficiaries who qualify for both the RUG-III Rehabilitation and Extensive Services categories. As expected, our analyses indicated that ancillary costs were much higher for Medicare beneficiaries in the Extensive Services category than for those in other categories. There are also a significant number of beneficiaries who would classify into the Extensive Services category based on clinical conditions but who, because they are also receiving rehabilitation services, classify into one of the Rehabilitation categories instead (due to the hierarchical logic of the RUG-III classification system). These beneficiaries carry with them the same non-therapy ancillary costs associated with their complex clinical needs even though they are classified into a RUG-III Rehabilitation category.

The high costs for beneficiaries in the Extensive Services category suggest that the payment rate for Extensive Services should be increased. However, increasing the payment rate without further adjustments could adversely affect provider incentives to provide therapy to beneficiaries requiring Extensive Services. Therefore, we expanded the scope of the proposed refinement to include a new category for beneficiaries who qualify for both Extensive Services and a RUG-III Rehabilitation category.

Our research findings showed little or no correlation between the groups within the Extensive Services category (that is, SE1, SE2, SE3) and the level of rehabilitation services used. For this reason, the structure for the new hierarchy level proposed here would mirror that of the existing Rehabilitation categories. Thus, we would add to the current RUG-III model fourteen (14) new “Rehabilitation and Extensive Services” sub-categories that use the same Rehabilitation sub-category and ADL splits as the current system (See Table 4 for the proposed RUG-III structure).

The second component of the proposed refinement is the development of a separate “non-therapy ancillary” index based on clinical variables on the MDS. We tested MDS items to identify clinical conditions and services that are predictive of non-therapy ancillary costs. First, we analyzed each MDS variable independently, and identified all MDS items that had a significant positive relationship (at the 5 percent level) with per diem non-therapy ancillary costs. Next, we identified combinations of MDS items that were associated with significant cost differences. We then evaluated variables for clinical validity and potential incentive effects. For example, we rejected consideration of indwelling catheters as case-mix adjustors due to the potential negative incentive factors associated with their use in the index. See Table 3 for a list of MDS items that were found to be associated with significant differences in ancillary costs.

Once we identified the critical predictive variables, we investigated a number of index model approaches. We developed weighted and unweighted versions of a non-therapy ancillary index. Both versions improved the variance prediction of the case-mix system. The unweighted index model assigns a non-therapy ancillary level based on a count of the variables (selected MDS items) associated with non-therapy ancillary costs. Under the weighted index model, different weights are assigned to the selected MDS items based on the difference in costs associated with the item. In this study, the researchers assigned the weights based on quantitative analysis of the data. With both indices, thresholds were determined to form subgroups which vary logically in cost. However, these cost variations relate to the research data base, and need to be verified against the national MDS/Medicare claims data base.

The grouping logic used for the refined RUG-III is very similar to that currently used. The same 108 MDS items that are used to classify beneficiaries into the 44 RUG-III groups will be used to classify beneficiaries into the refined RUG-III subcategories in either the unweighted or weighted index models. It is only at the last level of classification that additional MDS items are considered. The MDS items used for the last step of classification include some of the 108 items that are used for the first level of classification in addition to some others, either alone or in combinations.

The last step to grouping using the unweighted index model (UWIM) that we are proposing is based on a count of clinical variables, up to a maximum of 11. There are 11 “domains,” some of which are comprised of multiple MDS clinical variables. The clinical conditions and services that define the domains are shown in Table 3. Within a domain, any one clinical variable, or combination of variables, satisfies the criteria for being included in the count for classification into one of the refined RUG-III groups. For example, the first domain is “Parenteral/IV feeding with greater than 76 percent total calories.” In order for the domain to be counted for determining the final step in RUG-III classification in the UWIM, the MDS items K5a and K6a must be coded to reflect the receipt by the beneficiary of at least 76 percent of total nutrition received via parenteral or IV feeding in the previous 7 days.

Other domains are comprised of many more MDS items than the parenteral/IV feeding domain. An example of this is the domain entitled, “Oxygen and either pneumonia or respiratory infection with fever, or pneumonia or respiratory infection, chronic obstructive pulmonary disease, congestive heart failure, coronary artery disease with shortness of breath.” This domain will only count once toward classification even though it is possible for a beneficiary to have values for all of these clinical conditions. As soon as the grouper software identifies that one combination of MDS items' values is present on the MDS that satisfies this domain, it will credit the case with a count of 1 in addition to whatever other domain criteria are satisfied by the MDS.

The identified clinical variables are used for classification of every Medicare MDS in the Clinically Complex category and above, regardless of the other qualifying conditions and services reported on the MDS. This means that a beneficiary who has a count of 2 of the relevant clinical variables, will classify into the “3” level of the particular refined RUG-III subcategory for which he or she qualifies. As described above, the “3” level signifies a count of 1 or 2 of the clinical variables used for determining the non-therapy ancillary end split.

For example, a beneficiary who has pneumonia, an ADL sum score of 8, dehydration, a fever, and a surgical wound that requires twice daily dressing changes, will classify to the Special Care category. Within the Special Care category, the ADL score of 8 will classify this beneficiary into the “SC” subcategory. The count of the items that are used to make the final classification is 2, as the pneumonia and the wound care with dressing changes are the two clinical variables that will affect classification of this beneficiary to the SC3 group.

Using the selected MDS items, we calculated a non-therapy ancillary index score for each MDS and classified them to the appropriate non-therapy ancillary level. We are including a more detailed description of the non-therapy ancillary index methodology in Technical Appendix A.

An index model can differ with respect to the RUG-III categories to which the model is applied. Two options that we considered were to apply the index model only to the Extensive Services category (including beneficiaries in rehabilitation who also qualify for Extensive Services) or to apply the index option to a broader group of RUG-III categories. The research indicated very little difference in ancillary costs for beneficiaries in the Impaired Cognition, Behavior and Physical Function categories. Differences in ancillary costs were identified within the Rehabilitation, Clinically Complex, Special Care, and Extensive Services groups. For this reason, we propose to apply the non-therapy ancillary index model to all residents in the Clinically Complex category or above (where over 90 percent of Medicare patients fall). In addition, we propose to apply a single non-therapy ancillary index factor to each of the lower levels of the RUG-III model (that is, Impaired Cognition, Behavior, and Physical Function).

Index models can also be applied differently across RUG-III levels. The most straightforward method is to apply a fixed dollar amount for each level of the index. In this case, the add-on for a non-therapy ancillary index score of 3 would be the same regardless of the beneficiary's RUG-III group. Separate indices can also be calculated for each level of the hierarchy. In this case, the dollar amount of the non-therapy ancillary index level of 3 would be different for beneficiaries in different levels of the RUG-III hierarchy, for example, clinically complex, special care, rehabilitation, etc. Separate indices are more appropriate when there is significant inter-group variance. Using the research data base, we found significant variation. In projecting rates for both the UWIM (Tables 5 and 6) and WIM 2 (Technical Appendix A, Tables 6.1 and 6.2) models, we calculated separate index values for each of the 8 proposed hierarchy levels. This approach will be analyzed and evaluated using the national PPS/MDS data base.

Finally, index models can also differ with respect to the number of non-therapy ancillary index groups that are used. Six groups were developed for the weighted index model. Four groups were used for the unweighted model. The weighted index model performs slightly better than its unweighted counterpart. However, it adds a significant level of complexity both in terms of the number of additional RUG-III variations and the addition of a new type of MDS scoring methodology based on cost instead of clinical criteria. In addition, as stated above, the weighted index model break points are not representative of national ancillary costs.

On the other hand, the unweighted index model relies on a count of MDS items to differentiate among index levels, an approach similar to that used currently in RUG-III for classification into the Extensive Services category. At this phase of our analysis, we have concluded that the added complexity of the weighted model offsets any benefits gained. Therefore, we are proposing the unweighted non-therapy ancillary index model that will be applied to the combined Rehabilitation/Extensive Services, Rehabilitation, Extensive Services, Special Care and Clinically Complex categories of the RUG-III hierarchy.

Adopting a new Extensive Services with Rehabilitation category and adding a non-therapy ancillary index component will require modifications to the naming conventions used to identify each RUG-III group. Based on these recommendations, we have updated the RUG-III structure to incorporate the proposed refinements, as displayed in Table 4. These proposed RUG-III groups are based upon the existing 3 digit RUG-III coding structure, but will designate the non-therapy ancillary level as well as the RUG-III category.

The first letter of the RUG-III code defines the hierarchy level. First, a new hierarchy level is being added to recognize beneficiaries needing a combination of Extensive and Rehabilitation Services. The codes used to reflect the hierarchy level are also being expanded to identify separately each level of Rehabilitation (that is, Ultra High, Very High, High, Medium and Low) either in combination with Extensive Services or separately.

The second letter of the proposed RUG-III coding structure is an alpha character that indicates the final group assigned after the RUG-III end-splits (that is, ADLs, depression, restorative nursing) have been calculated.

The third digit of the proposed RUG-III coding structure will indicate the non-therapy ancillary index level. In the unweighted non-therapy ancillary model, there are 4 levels determined by the number of MDS non-therapy ancillary qualifying items (See Table 4 for the complete list of qualifiers.)

For example, under the current RUG-III model, a beneficiary whose MDS reflects an ADL sum score of 11, a tracheostomy, suctioning, pneumonia, IV medications and receipt of 380 minutes per week of physical therapy, would group into the RHB rehabilitation group.

In the refined RUG-III model with the unweighted non-therapy ancillary index, this beneficiary would group into the LB4 group with the first digit, L, indicating a combination of Extensive Services and High Rehabilitation, the second digit, B, indicating the ADL level of 11, and the third digit, 4, indicating the non-therapy ancillary level for a beneficiary with 4 qualifiers. See Table 4 for a crosswalk from the current RUG-III groups to the new groups.

In Example 2, we will show the proposed classification for a beneficiary who receives no rehabilitation services. This beneficiary is a quadriplegic, who has an ADL sum score of 17, a stage 4 pressure ulcer, treatment for the pressure ulcer, pneumonia, and daily respiratory therapy. This beneficiary currently classifies into the Special Care category, into the SSC group. In the refined classification system he or she will group into the SA4 group, showing that he or she is in the Special Care category, with an ADL sum score of 17-18, and 3-5 of the MDS non-therapy ancillary qualifiers.

A naming convention has also been established for the weighted model. The first 2 digits are the same as for the unweighted model. The third digit, the non-therapy ancillary indicator, uses alpha characters A through F, with “F” as the lowest ancillary level.

Additional Research Plans

As noted above, we performed the RUG-III refinement analyses on a research data base rather than on PPS Medicare claims and MDS data. The research data base was appropriate and extremely useful in testing hypotheses, and identifying areas where refinements could be introduced. However, research data always have limitations, and HCFA and contractor staff have identified several areas of concern. Fortunately, since actual PPS claims and MDS data are now available, we are already conducting additional analyses of the unweighted and weighted models to address these concerns and validate the research findings.

For this proposed rule, we have developed Tables 5 and 6 to illustrate the application of the proposed refinement to the RUG-III classification system on the FY 2001 Federal per diem rates. In addition, for comparison purposes, we have developed rate tables for the WIM2 model that are shown in Technical Appendix A (Tables 6.1 and 6.2). However, in reviewing these tables, it is important to recognize the following limitations:

The nursing index is a critical factor in accurately calibrating the system to link payment to acuity levels. The nursing indices shown in Tables 5 through 6 assume that the distribution of the actual Medicare population is the same as the distribution of the research data base. We are now reworking these calculations using national PPS data to ensure accurate calibration of the system.

Using the actual PPS data base also adjusts for a second data limitation: the extent to which MDS data reflects short stay patients. The research data base utilized MDS assessments from 1995 through 1997, a period when MDSs were often not completed for beneficiaries who were in a SNF for less than 14 days. By contrast, the PPS data base includes short-stay beneficiaries, and we will take any special needs of this population into account by using actual PPS data to validate the initial findings.

In addition, the methodology used to adjust non-therapy ancillary charges to cost used the older, non-therapy ancillary charges and facility cost-to-charge ratios. In developing the PPS data base, we will use PPS claims data and the latest available cost-to-charge ratios.

Using the smaller research data base, it was not always possible to obtain a large number of observations in some of the RUG-III groups to fully determine ancillary costs with the necessary level of precision. For that small number of RUG-III groups, the researchers imputed ancillary costs, and applied these imputed costs to the non-therapy ancillary index used in the rate-setting projections. Using the national PPS data base will allow better differentiation between the non-therapy ancillary index levels for the new, combined Rehabilitation and Extensive Services categories, particularly in index levels 2 and 3 of the unweighted model (and B and C of the weighted model.) (See Tables 5 and 6 for the UWIM model and Technical Appendix A Tables 6.1 and 6.2 for the WIM2 model.)

Finally, we will continue the process of identifying possible negative incentives associated with MDS items used in the non-therapy ancillary index. We will carefully evaluate each item before incorporating it into the final index. Then, we will develop methods to monitor coding practices and to identify changes in coding patterns for use in medical review, quality assurance and program integrity activities. We will issue clarifications, through Program Memoranda and other appropriate means, of MDS requirements needed to maintain the integrity of the RUG-III system.

Using the national PPS data base, we will recalculate the distribution of the beneficiary population across RUG-III categories, including the proposed combined Rehabilitation and Extensive Services category. Then, we will perform the necessary analyses and sensitivity tests to compare the results with those derived from the research data base. We will reevaluate program options (for example, unweighted vs. weighted non-therapy ancillary index, etc.) based on the additional analyses, and modify the proposed refinements as needed. We expect these final analyses to be available in late Spring 2000, and we plan to incorporate them in the final rule to be issued before August 1, 2000.

PPS Rate Tables

We are confident that the additional analyses based on national data will confirm the need for refinements in the RUG-III model by adding the new combined Extensive and Rehabilitation Service groups and by creating a new non-therapy ancillary index. However, it is very likely the values of some of the model components (for example, average ancillary cost by RUG-III group, frequency distribution by RUG-III group, relative weights, etc.) will be further refined through use of the national data base. For this reason, it is important to understand that the values contained in these tables will likely change in the final rule.

While we are confident that these research findings are based on sound methodology, it is certainly possible that additional testing will identify new issues or support variations of the models to those presented here. We remain open to suggestions during the comment period and will carefully evaluate the validation analyses before proceeding to final rulemaking. To illustrate the impact of these proposed changes based on the best data currently available, we have developed rate Tables 5 and 6 using the unweighted model. (For an additional discussion of the weighted model, including a schedule of rates, see Technical Appendix A.) These projections should not be viewed as final nursing indices, non-therapy ancillary indices, or payment rates.

Further, as noted above, we based the non-therapy ancillary indices on the mean adjusted derived cost (that is, charges adjusted by facility ancillary cost-to-charge ratios) of non-therapy ancillary services. Mean costs were calculated separately for each of the eight proposed levels of the RUG-III hierarchy. For the research data base, we used the cost-to-charge ratio applicable to the service date of the claim. For the follow up analyses using actual PPS claims data, we are using the most recent available cost-to-charge ratio. We expect that using the newer cost-to-charges ratios will enhance the accuracy of the calculations. However, due to the lag time between SNF PPS claims submission and cost report processing, it is impossible to match the claims service dates perfectly with the cost report period used for the cost-to-charge ratios. For the SNF PPS data base, we are proposing to use approximately 9 months of claims data starting from January 1, 1999, the date almost all providers became subject to PPS. The cost reports for calendar year 1999 are not due until April 2000.

Finally, the research findings in this proposed rule include the use of “imputed” data in situations where the cell size (for example, number of records meeting the criteria for a specific RUG-III group, etc.) was too small for accurate measurement. When using the national data base, we expect that the relevant data cells will be adequately populated and that all analyses used in developing the final rule will be based on actual rather than imputed data.

These tables reflect two adjustments in particular. First, our nursing and therapy staff time indices (combined 1995 and 1997 staff time data) currently used to establish PPS rates have been adjusted to reflect the new combined Extensive Services with Rehabilitation categories. Second, we have adjusted the nursing case mix component of the rate to remove the non-therapy ancillary component that is part of the current nursing index used in PPS rate-setting. We will need to adjust one or both of these components based on the additional analyses.

We integrated these proposed refinements into the rate-setting methodology, and we list the estimated per diem Federal rates for 178 separate RUG-III classification groups in Tables 5 and 6. We list the case-mix adjusted payment rates separately for urban and rural SNFs (178 each), with the corresponding case-mix index values. These tables list the rates in total and by component. The application of the wage index, described later in this section, is the final adjustment applied to the projected Federal rates in these tables.

In accordance with section 101 of the BBRA, we will make a four percent upward adjustment to the adjusted per diem Federal rate for FY 2001. This estimated adjustment is shown in Table 9.

Finally, these projected rates do not reflect the BBRA requirement (section 103) to reduce the Part A SNF payment rates to account for those services that are newly excluded from consolidated billing and, thus, will be separately billable to Part B by the supplier. As mentioned in section II.A.2. above, because of the complexity of the process and the amount of time needed to implement this requirement, we are unable at present to adjust the proposed rates to reflect this. However, we will make these adjustments prospectively in the final rule establishing payment rates for FY 2001, using the methodology described below.

In order to compute the level of this adjustment, we propose to determine the per diem amount of allowed charges associated with the specific HCPCS codes identified in the statute (and later in this rule) using the same 1995 data on Part B services used in establishing the Federal rates. These data are described in detail in section II.A.2.b of the May 12, 1998 interim final rule (63 FR 26251) and final rule (64 FR 41644) associated with the implementation of the SNF PPS. The per diem amount will be subtracted from the non-therapy ancillary component of the Federal rates shown in Tables 5 and 6 of this rule. We expect this adjustment to be minimal.

Summary of Proposed RUG-III Refinements

Based on the research described here, we are proposing the addition of new RUG-III groups to recognize the needs of Medicare beneficiaries with both heavy medical and rehabilitation needs and the development of an unweighted index model that would account more precisely for the variation in non-therapy ancillary services. Since the research shows substantial ancillary cost variation in the Rehabilitation and Extensive Services, Rehabilitation, Extensive Services, Special Care, and Clinically Complex categories, we have proposed four ancillary index levels to capture variation in ancillary costs accurately. Since beneficiaries in the Impaired Cognition, Behavior, and Physical Function categories exhibited a much smaller ancillary cost variation, we calculated a single ancillary add-on amount. The ancillary add-on amounts were calculated separately for each of the eight proposed RUG-III categories.

The refinements will achieve important improvements in the PPS model, and allow for more accurate payment rates. In addition, after further analysis and review of public comments, we may adjust these proposed refinements further to reflect actual PPS experience.

Collection of Medication Data

In the interim final rule published in the Federal Register on May 12, 1998, we stated that we would require facilities to complete and include MDS Section U with their Medicare MDS submissions beginning October 1, 1999. Subsequently, in the final rule published in the Federal Register on July 30, 1999, we announced a delay of that requirement and stated our intention to require completion of Section U beginning October 1, 2000. However, we are currently unable to implement the collection of medication data on the MDS beginning October 1, 2000. Accordingly, we will not require completion and submission of Section U of the MDS beginning October 1, 2000, as we had planned. We are currently examining issues related to the implementation of this requirement and we plan to address this matter when we implement the SNF PPS payment update for FY 2001.

C. Wage Index Adjustment to Federal Rates

Section 1888(e)(4)(G)(ii) of the Act requires that we provide for adjustments to the Federal rates to account for differences in area wage levels using an “appropriate” wage index as determined by the Secretary. In addition, it is our intent to evaluate a wage index based specifically on SNF data once it becomes available. The SNF wage data are currently being collected and evaluated to determine if we can utilize them in the future. If a wage index based on SNF data is developed, we will publish it for comment. However, in the interim, many commenters urged us to incorporate the latest wage data available. We continue to believe that, until a wage index based on SNF wage data is collected and analyzed, the hospital wage index's wage data provide the best available measure of comparable wages that should be paid by SNFs. We believe, since hospitals and SNFs compete in the same labor market area, that the use of this index's wage data results in an appropriate adjustment to the labor portion of SNF costs based on an appropriate wage index, as required under section 1888(e) of the Act.

For rates addressed in this proposed rule, we are using wage index values that are based on hospital wage data from cost reporting periods beginning in FY 1996, the same wage data as used to compute the FY 2000 wage index values for the inpatient hospital PPS. We will incorporate updated wage data in the final rule for the FY 2001 SNF PPS update.

The computation of the wage index is similar to past years in that we incorporate the latest data and methodology used to construct the hospital wage index (see the discussion in the May 12, 1998 interim final rule (63 FR 26274)). The wage index adjustment is applied to the labor-related portion of the Federal rate, which is 77.663 percent of the total rate. The schedule of Federal rates below shows the Federal rates by labor-related and non-labor-related components.

As discussed above and in the interim final rule, until an appropriate wage index based specifically on SNF data is available, we will use the latest available hospital wage index data in making annual updates to the payment rates. In making these annual updates, section 1888(e)(4)(G)(ii) of the Act requires that the application of this wage index be made in a manner that does not result in aggregate payments that are greater or less than would otherwise be made in the absence of the wage adjustment. In this third PPS year (Federal rates effective October 1, 2000), we are updating the wage index applicable to SNF payments using the most recent hospital wage data and applying an adjustment to fulfill the budget neutrality requirement. This requirement will be met by multiplying each of the per diem rate components by the ratio of the volume weighted mean wage adjustment factor (using the wage index from the previous year) to the volume weighted mean wage adjustment factor, using the wage index for the FY beginning October 1, 2000. The same volume weights are used in both the numerator and denominator and will be derived from 1997 Medicare Provider Analysis and Review File (MedPAR) data. The wage adjustment factor used in this calculation is defined as the labor share of the rate component multiplied by the wage index plus the non-labor share. The budget neutrality factor for FY 2001 is multiplied by each of the Federal rate components. This factor will be established when the updated wage data for the FY 2001 hospital wage index is available and set forth in the final rule establishing the FY 2001 SNF PPS rates.

For any RUG-III group, to compute a wage-adjusted Federal payment rate, the labor-related portion of the payment rate is multiplied by the SNF's appropriate wage index factor. The wage index factor has not been updated since the publication of the July 30, 1999 update notice (64 FR 41684). The product of that calculation is added to the corresponding non-labor-related component. The resulting amount is the Federal rate applicable to a beneficiary in that RUG-III group for that SNF.

D. Updates to the Federal Rates

In accordance with section 1888(e)(4)(E) of the Act, the proposed payment rates listed here have been updated by the SNF market basket minus 1 percentage point, which equals 1.01833 percent. For each succeeding FY, we will publish the rates in the Federal Register before August 1 of the year preceding the affected Federal FY.

For the current FY (FY 2001), and for FY 2002, section 1888(e)(4)(E)(ii) of the Act requires the rates to be increased by a factor equal to the SNF market index change minus 1 percentage point. For subsequent FYs, this section requires the rates to be increased by the applicable SNF market basket index increase.

E. Relationship of RUG-III Classification System to Existing Skilled Nursing Facility Level-of-Care Criteria

As discussed in II.B above, we are proposing a number of refinements in the RUGs classifications in this notice. Further, regulations at § 413.345 provide that the information included in each update of the Federal payment rates in the Federal Register will include the designation of those specific RUGs under the classification system that represent the required SNF level of care, as provided in § 409.30. Accordingly, we hereby propose to designate the following RUG-III classifications for this purpose: all groups within the Rehabilitation and Extensive category; all groups within the Ultra High Rehabilitation category; all groups within the Very High Rehabilitation category; all groups within the Medium Rehabilitation category; all groups within the Low Rehabilitation category; all groups within the Extensive Services category; and, all groups within the Clinically Complex category.

III. Three-Year Transition Period

Under sections 1888(e)(1) and (2) of the Act, during a facility's first three cost reporting periods that begin on or after July 1, 1998 (that is, the transition period), the facility's PPS rate will be equal to the sum of a percentage of an adjusted facility-specific per diem rate and a percentage of the adjusted Federal per diem rate, as discussed in Section I.D.2. above. After the transition period, the PPS rate will equal the adjusted Federal per diem rate. The transition period payment method will not apply to SNFs that first received Medicare payments (interim or otherwise) on or after October 1, 1995 under present or previous ownership, or to those facilities choosing to bypass the transition in accordance with section 102 of the BBRA; these facilities will be paid based on 100 percent of the Federal rate.

The facility-specific per diem rate is the sum of the facility's total allowable Part A Medicare costs and an estimate of the amounts that would be payable under Part B for covered SNF services for cost reporting periods beginning in FY 1995 (base year). The base year cost report used to compute the facility-specific per diem rate in the transition period may be settled (either tentative or final) or as-submitted for Medicare payment purposes. Under section 1888(e)(3) of the Act, any adjustments to the base year cost report made as a result of settlement or other action by the fiscal intermediary, including cost limit exceptions and exemptions, or results of an appeal, will result in a revision to the facility-specific per diem rate. The instructions for calculating the facility-specific per diem rate are described in detail in the May 12, 1998 interim final rule. In order to implement section 104 of the BBRA, for providers that received payment under the RUG-III demonstration during a cost reporting period that began in calendar year 1997, we will determine their facility-specific per diem rate using the methodology described below.

It is possible that some providers participated in the demonstration but did not have a cost reporting period that began in calendar year 1997. For those providers, we will determine their facility-specific per diem rate by using the calculations outlined in the May 12, 1998 Federal Register interim final rule (63 FR 26251, section III. (A)(1)(a), (b), or (c)). As with the facility-specific per diem applicable to other providers, the allowable costs will be subject to change based on the settlement of the cost report used to determine the total payment under the demonstration. In addition, we derive a special market basket inflation factor to adjust the 1997 costs to the midpoint of the rate setting period (October 1, 2000 to September 30, 2001.)

Step 1—Determine the aggregate payment during the cost reporting period that began in calendar year 1997—RUG-III payment plus routine capital costs plus ancillary costs (other than occupational therapy, physical therapy, and speech pathology).

Step 2—Divide the amount in Step 1, by the applicable total inpatient days for the cost reporting period.

Step 3—Adjust the amount in Step 2, by 1.094828 (inflation factor).

Step 4—Add the amount determined in step 3, to the appropriate Part B add-on amount determined according to Program Memorandum transmittal no. A-99-53 (December 1999).

The amount in Step 4 is the facility-specific rate that is applicable for the facility's first cost reporting period beginning on or after October 1, 2000.

Computation of the Skilled Nursing Facility Prospective Payment System Rate During the Transition:

For the first three cost reporting periods beginning on or after July 1, 1998 (the transition period), an SNF's payment under the PPS is the sum of a percentage of the facility-specific per diem rate and a percentage of the adjusted Federal per diem rate. Under section 1888(e)(2)(C) of the Act, for the first cost reporting period in the transition period, the SNF payment will be the sum of 75 percent of the facility-specific per diem rate and 25 percent of the Federal per diem rate. For the second cost reporting period, the SNF payment will be the sum of 50 percent of the facility-specific per diem rate and 50 percent of the Federal per diem rate. For the third cost reporting period, the SNF payment will be the sum of 25 percent of the facility-specific per diem rate and 75 percent of the Federal per diem rate. For all subsequent cost reporting periods beginning after the transition period, the SNF payment will be equal to 100 percent of the Federal per diem rate. An example is given below computing the SNF PPS rate and SNF payment.

Example of computation of adjusted PPS rates and SNF payment:

Using the XYZ SNF described in Table 9, the following shows the adjustments made to the facility-specific per diem rate and the Federal per diem rate to compute the provider's actual per diem PPS payment in the transition period. XYZ's 12-month cost reporting period begins October 1, 2000. (This is the provider's second cost reporting period under the transition.)

Step 2

Compute Federal per diem rate:

IV. The Skilled Nursing Facility Market Basket Index

Section 1888(e)(5)(A) of the Act requires the Secretary to establish an SNF market basket index (input price index) that reflects changes over time in the prices of an appropriate mix of goods and services included in the SNF PPS. This rule incorporates the latest estimates of the SNF market basket index at the time of this proposed rule. The final rule will incorporate updated projections based on the latest available projections as of that point in time. Accordingly, as described below, we have developed a SNF market basket index that encompasses the most commonly used cost categories for SNF routine services, ancillary services, and capital-related expenses. In the May 12, 1998 Federal Register, we included a complete discussion on rebasing the SNF market basket to FY 1992, and revising the index to include capital and ancillary costs. There are 21 separate cost categories and respective price proxies. These cost categories were illustrated in Tables 4.A, 4.B, and Appendix A, found in the May 12, 1998 Federal Register.

Each year we calculate a revised labor-related share based on the relative importance of labor-related cost categories in the input price index. Table 10.A below summarizes the updated labor-related share for FY 2001.

The forecasted rates of growth used to compute the projected SNF market basket percentages, described in the next section, are shown in Table 10.B.

Released by HCFA, OACT, National Health Statistics Group

Use of the Skilled Nursing Facility Market Basket Percentage:

Section 1888(e)(5)(B) of the Act defines the SNF market basket percentage as the percentage change in the SNF market basket index, described in the previous section, from the midpoint of the prior FY (or period) to the midpoint of the current FY (or other period) involved. The facility-specific portion and Federal portion of the SNF PPS rates addressed in this proposed rule are based on cost reporting periods beginning in the base year, Federal FY 1995. For the Federal rates, the percentage increases in the SNF market basket index will be used to compute the update factors occurring between the midpoint of FY 2000 and the midpoint of FY 2001. We used the Standard & Poor's DRI CC, 4th quarter 1999 historical and forecasted percentage increases of the revised and rebased SNF market basket index for routine, ancillary, and capital-related expenses, described in the previous section, to compute the update factors. Finally, the update factors, as described below, will be used to adjust the base year costs for computing the facility-specific portion and Federal portion of the SNF PPS rates.

A. Facility-Specific Rate Update Factor

Under section 1888(e)(3)(D)(i) of the Act, for the facility-specific portion of the SNF PPS rate, we will update a facility's base year costs up to the corresponding cost reporting period beginning October 1, 2000, and ending September 30, 2001, by the SNF market basket percentage. We took the following steps to develop the 12-month cost reporting period facility-specific rate update factors shown in Table 10.C.

For the facility rate, we developed factors to inflate data from cost reporting periods beginning October 1, 1994, through September 30, 1995, to the corresponding cost reporting period beginning in FY 2001. According to section 1888(e)(3)(D) of the Act, the years through FY 1999 were inflated at a rate of market basket minus 1 percentage point, while FY 2000 and FY 2001 are to be inflated at the full market basket rate of increase.

1. We first determined the total growth from the midpoint of each 12-month cost reporting period that began during the period from October 1, 1994, through September 30, 1995, to the midpoint of the corresponding period beginning in FY 2001.

2. From this total growth, we determined the average annual growth rate for each time span.

3. We subtracted 1 percentage point from each average annual growth rate through FY 1999.

4. These reduced average annual growth rates were converted to cumulative growth rates, using market basket minus one for the first four years, and with full market basket for the final two years. (For example, if the time span were for 9 years, we would inflate at the market basket minus 1 percentage point annual rate for 7 years and at annual market basket rate for 2 additional years).

B. Federal Rate Update Factor

To update each facility's costs up to the common period, we:

1. Determined the total growth from the average market basket level for the period of October 1, 1999 through September 30, 2000 to the average market basket level for the period of October 1, 2000 through September 30, 2001.

2. Calculated the rate of growth between the midpoints of the two periods.

3. Calculated the annual average rate of growth for number 2, above.

4. Subtracted 1 percentage point from this annual average rate of growth.

5. Using the annual average minus 1 percentage point rate of growth, determined the cumulative growth between the midpoints of the two periods specified above.

This revised update factor was used to compute the Federal portion of the SNF PPS rate shown in Tables 1 and 2.

V. Consolidated Billing

Section 4432(b) of the BBA sets forth a consolidated billing requirement applicable to all SNFs providing Medicare services. SNF consolidated billing is a comprehensive billing requirement (similar to the one that has been in effect for inpatient hospital services for well over a decade), under which the SNF itself is responsible for billing Medicare for virtually all of the services that its beneficiaries receive. As with hospital bundling, the law contains a list of services (primarily those of physicians and certain other types of medical practitioners) that are excluded from SNF consolidated billing and, thus, can be separately billed to Part B directly by the outside entity that furnishes them to the Medicare beneficiary (see section 1888(e)(2)(A)(ii) of the Act).

Section 103(a)(2) of the BBRA added section 1888(e)(2)(A)(iii) to the Act to provide for the exclusion of certain additional types of services from SNF consolidated billing, effective with services furnished on or after April 1, 2000. The original statutory exclusions enacted by the BBA consisted of a number of broad service categories, and encompassed all of the individual services that fall within those categories. By contrast, the additional exclusions enacted in the BBRA apply only to certain specified, individual services within a number of broader service categories that otherwise remain subject to consolidated billing. Within the affected service categories—that is, chemotherapy items and their administration, radioisotope services, and customized prosthetic devices—the exclusion applies only to those individual services that are specifically identified by HCPCS code in the legislation itself, while all other services within those broader categories remain subject to consolidated billing. See Table 11, Post-BBA Consolidated Billing Exclusions. We have issued Program Memorandum (PM) no. AB-00-18 (March 2000), which lists the HCPCS codes of those particular services identified by the BBRA as excluded from consolidated billing.

The BBRA Conference report (H.R. Conf. Rep. No. 106-479 at 854) characterizes the individual services that this legislation targets for exclusion as “* * * high-cost, low probability events that could have devastating financial impacts because their costs far exceed the payment [SNFs] receive under the prospective payment system * * *.” According to the conferees, section 103(a) “is an attempt to exclude from the PPS certain services and costly items that are provided infrequently in SNFs * * *.” Some chemotherapy drugs, which are relatively inexpensive and are administered routinely in SNFs, were excluded from this provision [and thus continue to be subject to consolidated billing requirements]. Id.

Further, we note that the exceptionally costly and intensive outpatient hospital services, such as magnetic resonance imaging (MRIs) and cardiac catheterization, that we identified previously under the regulations at § 411.15(p)(3)(iii) (see the preamble discussion in the May 12, 1998 interim final rule at 63 FR 26298-99, and in the July 30, 1999 final rule at 64 FR 41675-76) are excluded from consolidated billing only when furnished in the outpatient hospital setting. By contrast, as indicated in Table 11, the services identified in section 103 of the BBRA are excluded regardless of whether they are furnished in a hospital or nonhospital setting.

In addition, section 103(a)(2) of the BBRA excludes from consolidated billing those ambulance services that are furnished to an SNF beneficiary in conjunction with dialysis services that are covered under Part B. We note that Part B dialysis services themselves are already excluded from consolidated billing (see regulations at 42 CFR 411.15(p)(2)(vii)), as are those ambulance services that are furnished to a beneficiary who is not considered an SNF “resident” for consolidated billing purposes (see § 411.15(p)(2)(x))—for example, a beneficiary who receives one of the excluded outpatient hospital services under § 411.15(p)(3)(iii). The BBRA Conference Committee report further indicates that the newly excluded ambulance services (that is, those needed to transport a SNF resident who receives Part B dialysis services offsite at a certified dialysis facility) still remain subject to the overall medical necessity requirement that applies to ambulance services generally; that is, that ambulance coverage is available only in those situations where the use of other means of transportation is medically contraindicated. (H.R. Conf. Rep. No. 106-479 at 854.)

Further, we note that the statutory exclusion of those ambulance services that are furnished to SNF residents in conjunction with Part B dialysis services does not extend to ambulance services furnished to SNF residents in conjunction with any of the other types of services that this section of the BBRA identifies as excluded. For example, when a SNF resident is temporarily transported offsite via ambulance to receive a type of chemotherapy that is excluded by the BBRA, the ambulance services themselves remain subject to the SNF consolidated billing provision, and are not separately billable to Part B.

Section 103 of the BBRA also gives the Secretary the authority to designate additional, individual services for exclusion within each of the specified service categories. The BBRA Conference report notes that “* * * [n]ew, extremely costly items may come into use or codes may change over time”, H.R. Conf. Rep. No. 106-479 at 854 and the discretionary authority provided in the BBRA affords the Secretary the flexibility to revise the exclusion list as warranted by changing conditions that may occur in the future. For example, we note that the BBRA's conference agreement requests the GAO to conduct a review, by July 1, 2000, of the appropriateness of the codes that this legislation has designated for exclusion from consolidated billing. We will carefully consider the GAO's findings to determine whether further refinements in the exclusion list are warranted.

Also, we note that the BBRA made a number of technical corrections in the provisions of the BBA. One of these corrections, section 321(g)(2) of the BBRA, has revised the statute at section 1833(h)(5)(A)(iii) of the Act to make it clear that clinical diagnostic tests furnished to a SNF resident are subject to the consolidated billing requirement.

Finally, while we have implemented consolidated billing in connection with services furnished to SNF residents during Medicare-covered stays, we have not yet implemented so-called “Part B” consolidated billing, in connection with services furnished to SNF residents who are in noncovered stays. As we explained in the July 30, 1999 final rule, the overriding need to accomplish systems renovations in time to achieve Year 2000 (Y2K) compliance forced us to delay certain other projects that involved significant systems modifications of their own, including the implementation of this aspect of consolidated billing. Now that the Y2K-related systems changes have been completed, we have been able to resume work on these other projects. In this context, we have been reexamining some of the operational implications of consolidated billing that are specific to implementing the “Part B” aspect of this provision.

For example, under regulations at § 411.15(p)(3)(iv), if a beneficiary leaves the SNF and then returns within 24 hours of departure, his or her status as an SNF “resident” (for consolidated billing purposes) continues during the absence, regardless of whether the SNF has effected a formal discharge. This would make the SNF responsible for billing Medicare for any services that a beneficiary receives during a temporary absence of up to 24 hours, other than those that are specifically excluded (see the preamble discussion in the SNF PPS interim final rule (63 FR 26298 through 26299, May 12, 1998)). Since consolidated billing is currently in effect only for those SNF stays that are covered by Part A and paid by the PPS, this essentially means that such a beneficiary remains a SNF “resident” after leaving the SNF only if he or she then returns to the SNF by midnight, thus making the day of departure a covered Part A day. However, once consolidated billing is fully implemented, this will effectively convert the policy regarding services furnished during a beneficiary's temporary absence from the current “midnight rule” to the full “24 hour rule” described in the regulations.

As explained in the SNF PPS interim final rule, we initially established a 24-hour window in the regulations in order to prevent a SNF from being able to unbundle a particular service merely by sending a beneficiary offsite briefly to receive the service as an outpatient of a hospital or clinic. However, we note that SNFs basically have a financial incentive to unbundle such services only in connection with a resident whose stay is covered under Part A, since unbundling the service would mean that it could be paid separately under Part B, rather than out of the global per diem amount that Part A pays the SNF for the covered stay itself. By contrast, a resident who is in a noncovered stay does not qualify for comprehensive coverage of the entire institutional package of care under Part A, but only for Part B coverage of the individual medical and other health services specified in section 1861(s) of the Act. This means that when a SNF resident is in a noncovered stay, Part B would pay individually for each covered medical or other health service furnished to that resident, regardless of whether the SNF or an outside supplier submits the bill.

Thus, as the financial incentives for unbundling are associated with covered stays, we believe that it may be appropriate to have a standard with regard to SNF “resident” status that, in actual practice, is not more stringent for noncovered stays. We could revise the regulations at § 411.15(p)(3)(iv) to provide for continuing a beneficiary's “resident” status during a temporary absence only if he or she returns by midnight of the day of departure. This would, in effect, utilize the same standard that currently applies to covered stays for noncovered stays as well, and we invite comments on the appropriateness of such a revision.

As a point of clarification, we note that the phrase “midnight of the day of departure” refers to the midnight that immediately follows the actual moment of departure, rather than to the midnight that immediately precedes it (see, for example, the discussion of a “leave of absence” in section 3103.3 of the Medicare Intermediary Manual, Part 3 (HCFA Pub. 13-3), which indicates that the day a patient returns to the hospital from a leave of absence “* * * is counted as an inpatient day if he is present at midnight of that day” (emphasis added)). Thus, under this policy, a patient “day” begins at 12:01 A.M., and midnight of a particular day occurs at the very end of that day rather than at the very beginning. For example, under the “midnight rule,” if a beneficiary begins a leave of absence from the SNF at 10:00 A.M. on July 1 but subsequently returns to the SNF by 12:00 A.M. that night, the beneficiary would continue to be considered a “resident” of the SNF, for consolidated billing purposes, during his or her absence. By contrast, if the beneficiary does not return to the SNF until 1:00 A.M. on the morning of July 2, his or her “resident” status, for consolidated billing purposes, would end as of 10:00 A.M. on July 1, and would not resume until the actual point of readmission to the SNF (that is, as of 1:00 A.M. on July 2).

VI. Provisions of the Proposed Rule

The provisions of this proposed rule are as follows:

• In § 411.15, paragraph (p)(2)(vii) would be revised to exclude from consolidated billing those ambulance services that are furnished to an SNF resident in conjunction with dialysis services that are covered under Part B.
• In § 411.15, paragraph (p)(2) would also be revised to list the additional services that the BBRA has excluded from consolidated billing.
• In § 411.15, paragraph (p)(3)(iv), the phrase “within 24 consecutive hours” would be revised to read “by midnight of the day of departure”.
• In § 489.20, paragraph (s) would be revised to list the additional services that BBRA has excluded from consolidated billing, and a conforming change would be made in § 489.21(h).
• In § 489.20, paragraph (s)(7) would be revised to exclude from consolidated billing those ambulance services that are furnished to an SNF resident in conjunction with dialysis services that are covered under Part B.
• Section 489.20(s)(11) and § 411.15(p)(2)(xi), would be revised to reflect editorial revisions in the paragraphs concerning the transportation costs of electrocardiogram equipment.

VII. Collection of Information Requirements

This document does not impose information collection and recordkeeping requirements. Consequently, it need not be reviewed by the Office of Management and Budget under the authority of the Paperwork Reduction Act of 1995 (44 U.S.C. 3501 et.seq.).

VIII. Response to Comments

Because of the large number of items of correspondence we normally receive on Federal Register documents published for comment, we are not able to acknowledge or respond to them individually. We will consider all comments we receive by the date and time specified in the DATES section of this preamble, and, if we proceed with a subsequent document, we will respond to the comments in the preamble to that document.

IX. Regulatory Impact Analysis

We have examined the impacts of this rule as required by Executive Order (EO) 12866, the Unfunded Mandates Reform Act (UMRA) (Pub. L. 104-4), the Regulatory Flexibility Act (RFA) (Pub. L. 96-354), and the Federalism Executive Order (EO) 13132.

Executive Order 12866 directs agencies to assess costs and benefits of available regulatory alternatives and, when regulation is necessary, to select regulatory approaches that maximize net benefits (including potential economic, environmental, public health and safety effects, distributive impacts, and equity). A regulatory impact analysis (RIA) must be prepared for major rules with economically significant effects ($100 million or more annually). This notice is a major rule as defined in Title 5, United States Code, section 804(2), because we estimate its impact will be to increase the payments to SNFs by approximately $900 million in FY 2001. The update set forth in this notice applies to payments in FY 2001. Accordingly, the analysis that follows describes the impact of this one year only. In accordance with the requirements of the Act, we will publish a notice for each subsequent FY that will provide for an update to the payment rates and include an associated impact analysis.

The UMRA also requires (in section 202) that agencies prepare an assessment of anticipated costs and benefits before developing any rule that may result in an annual expenditure by State, local, or tribal governments, in the aggregate, or by the private sector, of $100 million or more in any given year. This rule will have no consequential effect on State, local, or tribal governments. We believe the private sector cost of this rule falls below these thresholds as well.

Executive Order 13132 (effective November 2, 1999), establishes certain requirements that an agency must meet when it promulgates regulations that impose substantial direct compliance costs on State and local governments, preempts State law, or otherwise have Federalism implications. As stated above, this rule will have no consequential effect on State and local governments.

The RFA requires agencies to analyze options for regulatory relief of small entities. For purposes of the RFA, small entities include small businesses, nonprofit organizations, and governmental agencies. Most SNFs and most other providers and suppliers are small entities, either by virtue of their nonprofit status or by having revenues of $5 million or less annually. For purposes of the RFA, all States and tribal governments are not considered to be small entities, nor are intermediaries or carriers. Individuals and States are not included in the definition of a small entity. The policies contained in this rule would update the SNF PPS rates by increasing the payment rates published in the July 30, 1999 notice, but will not have a significant effect upon small entities.

In addition, section 1102(b) of the Act requires us to prepare a regulatory impact analysis if a rule may have a significant impact on the operations of a substantial number of small rural hospitals. This analysis must conform to the provisions of section 604 of the RFA. For purposes of section 1102(b) of the Act, we define a small rural hospital as a hospital that is located outside of a Metropolitan Statistical Area and has fewer than 50 beds. We are not preparing a rural impact statement since we have determined, and the Secretary certifies, that this notice will not have a significant economic impact on the operations of a substantial number of small rural hospitals.

A. Background

This notice sets forth proposed updates of the SNF PPS rates contained in the update notice, published on July 30, 1999. Table 13 below, presents the projected effects of the policy changes in the SNF PPS update notice, as well as statutory changes effective for FY 2001, on various SNF categories. We estimate the effects of each policy change by estimating payments while holding all other payment variables constant. We use the best data available, but we do not attempt to predict behavioral responses to our policy changes, and we do not make adjustments for future changes in such variables as days or case-mix.

This analysis incorporates the latest estimates of growth in service use and payments under the Medicare SNF benefit based on Medicare claims from 1998. Some of the data used for this analysis are the same data used to develop the impact analysis associated with the SNF PPS update notice promulgated on July 30, 1999 (64 FR 41684). These data were used to estimate the effects of changing only one payment variable at a time. We have also utilized MDS 2.0 data from the States used for the RUG-III refinement research (described in section 2.B earlier) to illustrate the effect of case mix refinements on the classification of the patient population in the study States. In addition, we are unable at this time to demonstrate the distributional impact of these case mix refinements on facility payments but anticipate doing so in the final rule planned for later this year.

We have used the best avaliable data on SNF case mix in calculating the FY 2001 impact for this proposed rule; however, we note that the data currently available on Medicare SNF claims and MDS 2.0 do not reflect the refined case mix classification system and case-mix indices proposed in this rule. While we still have only a partial database of SNF PPS claims and MDS 2.0 data at the present time due to the phased-in manner in which SNFs came into the PPS, we are confident that sufficient national data reflecting the distribution of payments and service days under the new RUG-III classification model can be assembled before promulgation of the final rule associated with this update. While the refinement to the case-mix classification system results in no greater or lesser aggregate payments to SNFs under the Medicare SNF PPS, we believe it is important to estimate the potential distributional impact of incorporating the refined RUG-III case-mix groups and indices. Consequently, for the final rule implementing the FY 2001 SNF PPS rates, we anticipate using such a national data base of SNF PPS claims and MDS 2.0 data to estimate more accurately the impact of this update, including the distributional effect of the case-mix refinements on payments for different facility types and locations. However, based on the data currently available, we believe that the method we have used to develop the impact analysis for this proposed rule offers the most accurate estimate of the FY 2001 update to the SNF PPS.

For this proposed rule, we have attempted to convey a sense of the effect of the case-mix refinements on the classification of residents in SNFs. Below, we have prepared Table 12 which displays the distribution of patients in the six-state sample used to develop the case-mix refinements, as shown for both the existing RUG-III groups and for the refined model proposed in this rule. This table details a comparison of the distribution of an identical group of Medicare patients across both the existing and proposed RUG-III classification models. In addition, Table 6, in Technical Appendix A accompanying this rule, illustrates a comparison of the distribution of this same group of patients across the existing RUG-III system and the alternate ancillary index refinement approach (WIM2) discussed earlier in this proposed rule.

We note that certain events may combine to limit the scope or accuracy of our impact analysis, because such an analysis is future-oriented and, thus, very susceptible to forecasting errors due to other changes in the forecasted impact time period. Some examples of such possible events are newly legislated general Medicare program funding changes by the Congress, or changes specifically related to SNFs. In addition, changes to the Medicare program may continue to be made as a result of the BBA. Although these changes may not be specific to SNF PPS, due to the nature of the Medicare program the changes may interact, and the complexity of the interaction of these changes could make it very difficult to predict accurately the full scope of the impact upon SNFs.

B. Impact of This Proposed Rule

As stated previously in this preamble, the aggregate increase in payments associated with this update is estimated to be $900 million. There are three areas of change that produce this increase for facilities—

1. The effect of the Federal transition, that results in many facilities being paid 75 percent at the Federal rate and 25 percent at the facility-specific rate instead of the current 50 percent Federal rate and 50 percent facility-specific rate. There is also the additional effect of the BBRA option to bypass the transition and be paid according to 100 percent of the Federal rate;

2. The implementation of various other provisions in the BBRA; and,

3. The total change in payments from FY 2000 levels to FY 2001 levels. This includes all of the previously noted changes in addition to the effect of the update to the rates.

As seen in table 13 below, some of these areas result in increased aggregate payments and others tend to lower them. The breakdown of the various categories of data in the table are as follows:

In column one, the first row of the table includes the effects on all facilities. The next six rows show the effects on facilities split by hospital-based versus freestanding and urban versus rural. The rest of the table shows the effects on urban versus rural status by census region.

The second column in the table shows the number of facilities in the impact database. The third column shows the effect of the transition to the Federal rates. It includes the impact of the normal progression of facilities in the transition to new cost reporting periods and, therefore, blended payment amounts (that is, facility-specific versus Federal rates) as well as those facilities that, as a result of the BBRA, elect to bypass the transition and go immediately to the full Federal rate). This change has an overall effect of raising payments by .3 percent, with most of the increase coming from freestanding facilities. There are several regions that have decreased payments due to this provision, but the majority (and most populous) of the regions evidence higher payments, with the largest increase being in the New England and mid-Atlantic regions for both urban and rural facilities.

We estimate that approximately 51 percent of SNFs currently under the transition will elect to be paid based on 100 percent of the Federal rate. Of these facilities, we estimate 22 percent are hospital-based and 78 percent are freestanding.

The fourth column shows the projected effect of the 4 percent add-on to the adjusted Federal rate mandated by the BBRA. As expected, this provision results in an increase in payments for all facilities. However, as seen in the table, the varying effect of the SNF PPS transition results in a distributional impact of this provision. In addition, since this increase only applies to the Federal portion of the payment rate, the effect on total expenditures is less than 4 percent.

The fifth column of the table shows the effect of the update to the Federal and facility-specific payment rates. It reflects an update to the Federal rates of 1.833 percent, which is equivalent to the market basket increase minus 1 percentage point, as required by law. In addition, it reflects an update to the facility-specific rates of 2.833 percent, which is equivalent to the full market basket increase for this period. For this analysis, it is assumed that payments will increase by 2.0 percent in total if there are no behavioral changes by the facilities. As can be seen from this table, the effects of the update itself do not vary significantly by specific types of providers or by location.

The sixth column of the table shows the effect of all of the changes on the FY 2001 payments. This includes all of the previous changes, including the update to this year's payment rates by the market basket. Therefore, it is assumed that payments will increase by 5.8 percent in total, assuming facilities do not change their care delivery and billing practices in response. As can be seen from this table, the combined effects of all of the changes vary much more widely by specific types of providers and by location. For example, freestanding facilities enjoy more significant payment increases due to the policy changes, while the effects of the transition tend to diminish the increase for hospital-based providers.

In the final rule implementing the SNF PPS update for FY 2001, we will revise the estimates listed in Table 13 to reflect the final FY 2001 payment rates as well as the latest available data on estimates of program growth in services and expenditures. Table 13 will also incorporate two additional columns showing the projected distributional effect of the refined case-mix classification system based on actual MDS 2.0 data and updated wage index across the various facility types and locations, as discussed earlier. We will also indicate the impact of the reduction in the Federal rates to account for the new services excluded from consolidated billing under section 103 of the BBRA.

As discussed earlier in this rule, Section 101 of the BBRA provides for a 20 percent positive adjustment to the adjusted Federal rates associated with 15 RUG-III groups for the period of April 1, 2000 through October 1, 2000. In addition, it provides for a four percent positive adjustment to the Federal rates associated with all RUG-III categories for FY 2001 and FY 2002, regardless of whether refinements to the case-mix adjustment are implemented. However, were we not to implement case-mix refinements such as those proposed in this rule for FY 2001, the Federal rates for this period would be based on the existing RUG-III model currently in use and maintain the 20 percent adjustments to the 15 specified RUG-III groups. As indicated in Table 13, the effect of this proposed rule will be an increase in expenditures of 900 million dollars (or +5.8 percent) over the payment rates and policies as described in the SNF PPS Notice published on July 30, 1999 (64 FR 41684). However, were we not to implement case-mix refinements, the effect of this BBRA provision would be a larger increase in expenditures equaling 1.9 billion dollars (or +12.5 percent). At the present time, we are unable to illustrate the distributional impact of maintaining this 20 percent add-on, but will attempt to develop the data to allow us to do so for the final rule associated with the FY 2001 update. It is important to note that such a result would also have negative consequences for the beneficiary. Section 101 of the BBRA provides the 20 percent add-on for certain RUG-III rehabilitation groups, resulting in higher payments for such groups even though they are associated with a lower intensity of service than other rehabilitation groups. This results in a perverse incentive where some facilities may choose to provide less rehabilitation services to beneficiaries in order to receive the higher payments. Because this provision of the law takes effect on April 1, 2000, it may already be resulting in a reduction of needed services. Adoption of the refinements proposed in this rule would eliminate this perverse incentive.

As noted previously, we are proposing the addition of new RUG-III categories to recognize the needs of Medicare beneficiaries with both heavy medical and rehabilitation needs and to account more precisely for the variation in non-therapy ancillary services. The refinements will achieve important improvements in the PPS and allow for more accurate payment rates, thus meeting our responsibility to provide for equitable payments to providers while ensuring access to quality SNF care for Medicare beneficiaries. In evaluating the different options, it is important to analyze the overall impact of implementing a refined case-mix system. Adoption of any of these refinements will increase the complexity of the PPS and may introduce some initial uncertainty for providers, who would have to become familiar with the refined system and modify existing operational and support systems. As discussed in section II.B of this proposed rule, we propose adoption of the UWIM model because we believe it best represents an appropriate balance between improvements in the accuracy of our payments and the complexity and uncertainty which results from changes of this nature.

Finally, in accordance with the provisions of Executive Order 12866, this notice was reviewed by the Office of Management and Budget.

X. Federalism

We have reviewed this final rule under the threshold criteria of Executive Order 13132, Federalism, and we have determined that it does not significantly affect the rights, roles, and responsibilities of States.

List of Subjects

42 CFR Part 411

• Kidney diseases
• Medicare
• Reporting and recordkeeping requirements

42 CFR Part 489

• Health facilities
• Medicare
• Reporting and recordkeeping requirements

For the reasons set forth in the preamble, 42 CFR chapter IV would be amended as follows:

PART 411—EXCLUSIONS FROM MEDICARE AND LIMITATIONS ON MEDICARE PAYMENT

A. Part 411 is amended as set forth below:

1. The authority citation for part 411 continues to read as follows:

Authority:Secs. 1102 and 1871 of the Social Security Act (42 U.S.C. 1302 and 1395hh).

Subpart A—General Exclusions and Exclusion of Particular Services

2. Section 411.15 is amended by:

A. Republishing the introductory text.

B. Revising paragraphs (p)(2)(vii) and (p)(2)(xi).

C. Adding new paragraphs (p)(2)(xii), (p)(2)(xiii), (p)(2)(xiv), and (p)(2)(xv).

D. Revising paragraph (p)(3)(iv).

PART 489—PROVIDER AGREEMENTS AND SUPPLIER APPROVAL

B. Part 489 is amended to read as follows:

1. The authority citation for part 489 continues to read as follows:

Authority:Secs. 1102 and 1871 of the Social Security Act (42 U.S.C. 1302 and 1395hh).

Subpart B—Essentials of Provider Agreements

2. Section 489.20 is amended by:

A. Republishing the introductory text and paragraph (s) introductory text.

B. Revising paragraphs (s)(7) and (s)(11).

C. Adding new paragraphs (s)(12), (s)(13), (s)(14), and (s)(15).

Note:

The following appendix will not appear in the Code of Federal Regulations.

Technical Appendix A—Technical Features of the RUG-III Refinements Analyses

The purpose of the research discussed in this proposed rule is to develop potential refinements to the PPS that would better ensure accurate and equitable payment. An analytic (or research) data base consisting of linked MDS assessments and Medicare claims data was developed, and used to perform the analyses described in this proposed rule.

A. Creation of Analytic Sample

In creating the analytic sample used to develop and test potential refinements, we were guided by the desire to have a large, representative sample and the need to exclude assessments likely to contain reporting errors. Our original sample included 733,300 MDS assessments from seven States, representing the years 1995 through 1997. We then reduced this sample through implementation of the following exclusion criteria:

1. Exclude all assessments from New York. All assessments from New York were excluded from analyses that used Medicare claims data because many facilities in the State billed SNF stays using an all-inclusive rate. Because these facilities did not use the revenue codes that we used to measure prescription drug, respiratory therapy or other non-therapy ancillary charges, measured ancillary charges for most New York beneficiaries were zero in some or all of the revenue codes analyzed for this study. The exclusion of New York results in the removal of 525,215 of the 733,300 total MDS assessments from our analytic sample.

2. Exclude all assessments for which a cost-to-charge ratio could not be calculated. Medicare cost report data were used to calculate the facility-specific ratio of Total Part A allowed cost to total Part A charges for each facility in each year. Facilities missing Medicare cost reports for at least two years between 1995 and 1997 were excluded because we were not able to calculate cost-to-charge ratios for the facility. This resulted in the exclusion of 93,314 additional assessments.

3. Exclude all facilities for which the correlation between a measure of drug costs calculated from Section U and one calculated from Medicare claims data was less than zero. We used drug charge data derived from Medicare claims in the refinement analyses, but used the Section U data to identify facilities with unreliable drug cost data. For facilities that have a negative correlation between the two drug cost measures, there is a concern about inaccurate reporting on either claims or MDS assessments at the facility level, and these facilities were excluded. This step resulted in the exclusion of 10,915 MDS assessments.

4. Exclude all beneficiaries with per diem ancillary charges greater than $1,000. Two hundred fifty-three (253) observations with per diem total ancillary charges greater than $1,000 were excluded from the refinement analyses. Summary measures of statistical performance such as R-squared are typically sensitive to outliers, and these extreme values were judged unlikely to be accurate. In addition, such values have disproportionate leverage in the design of potential refinements. The exclusion of extreme outliers in refinement analyses does not mean that their costs cannot be considered when determining payment rates.

The resulting analytic sample included 103,603 assessments, which were assigned randomly to either the test or validation samples. We assigned approximately 60 percent of this sample—61,929 assessments—to the test sample which was used to develop and test potential refinements. The remaining 41,674 assessments comprised the validation sample.

B. Characteristics of the Sample

Table 1 shows the sociodemographic characteristics of the sample stratified by an aggregate of the RUG-III categories. The majority of beneficiaries were female (65 percent), with little variation in the proportion across the RUG-III categories. Beneficiaries classified in the Behavior category were less likely to be male (37 percent) and those in the Physical Function categories were the least likely to be male (30 percent). The majority of beneficiaries were white, of non-Hispanic origin (84 percent). Approximately nine percent of beneficiaries were black and 2 percent were Hispanic. Overall, nearly one quarter of the beneficiaries were severely cognitively impaired. Among beneficiaries classified in a Rehabilitation category, 35 percent were moderately impaired and 14 percent were severely cognitively impaired. The distribution of cognitive impairment among those classified as Reduced Physical Function was similar to that of the Rehabilitation category. Beneficiaries classified as Extensive Services or Special Care also had a similar distribution of cognitive impairment level. Approximately one third of each were moderately impaired. Thirty-nine percent of beneficiaries were classified as dependent in activities of daily living and only 7 percent with no limitations. Beneficiaries in the Behavior category were most likely to have only minimal limitations in physical functioning (28 percent). Beneficiaries classified in the Clinically Complex (14 percent), Cognitively Impaired (13 percent), or Physical Function (14 percent) categories were also more likely to have minimal limitations relative to the other RUG-III categories. Beneficiaries in the Extensive Services (58 percent) and Special Care (56 percent) categories were most likely to be classified as dependent in activities of daily living.

The active clinical diagnoses documented for beneficiaries in the sample are shown stratified by RUG-III group on Table 1.1. Cardiovascular diseases were common in beneficiaries. Overall, 20 percent of beneficiaries had coronary artery disease. Cardiac arrhythmia was present in 14 percent of beneficiaries. Overall, nearly one quarter of beneficiaries had congestive heart failure and 9 percent had peripheral vascular diseases. On average, 43 percent of beneficiaries had documented hypertension. While the distribution of beneficiaries with coronary artery disease appeared similar across RUG-III groups, congestive heart failure and arrhythmia were more common in the Extensive Services, Special Care, and Clinically Complex categories. For most of the cardiovascular conditions, beneficiaries in the Impaired Cognition category were less likely to have these diseases relative to other RUG-III categories. A similar, but attenuated pattern was noted for beneficiaries in the Behavior category.

Neurological diseases were also common. Overall, 9 percent of beneficiaries had Alzheimer's disease documented. Twenty-eight percent had other dementia documented. Nearly one quarter of beneficiaries had an active clinical diagnosis of stroke and 6 percent had Parkinson's disease. While the proportion of beneficiaries with Parkinson's disease did not vary by RUG-III group, the proportion with other neurological conditions varied substantially by RUG-III group. Beneficiaries in the Impaired Cognition group were more likely to have Alzheimer's disease (22 percent) and other dementia (54 percent) documented and less likely to have had a stroke (15 percent) compared to other RUG-III groups. Similar to the Impaired Cognition group, beneficiaries in the Behavior category were more likely to have other dementia (41 percent) and less likely to have had a stroke (12 percent) compared to other RUG-III groups, but this category had a similar proportion of beneficiaries with Alzheimer's disease. The distribution of neurological conditions among beneficiaries classified as Extensive Services, Special Care, and Clinically Complex was similar. A third of beneficiaries classified as Extensive Services and Special Care had non-Alzheimer's dementia and one quarter had suffered a stroke.

Only 5 percent of beneficiaries had anxiety and 16 percent had depression documented as a diagnosis on the MDS. Across RUG-III groups, the proportion of beneficiaries with anxiety and depression was similar. However, the prevalence of anxiety (8 percent) and depression (22 percent) was higher in the Behavior category. Twelve percent of beneficiaries had cataracts and 7 percent had glaucoma. These conditions did not vary substantially by RUG-III group. Overall, septicemia was rare (1 percent), and only 8 percent of beneficiaries had pneumonia, while 17 percent had urinary tract infections. Beneficiaries in the Extensive Services category were more likely to have septicemia (2 percent), pneumonia (17 percent), and urinary tract infections (24 percent) compared to other RUG-III categories. Other diagnoses and conditions were common. Twenty-one percent of beneficiaries had allergies, 19 percent had anemia, 22 percent had arthritis, 22 percent had diabetes, and 12 percent had cancer. Beneficiaries in the Rehabilitation, Extensive Services, Special Care, and Clinically Complex categories were more likely to have these conditions relative to the Impaired Cognition and Behavioral Problem categories. The prevalence of hypothyroidism (10 percent) did not vary by RUG-III group.

Pooling across all States and the three years, there is little variation by RUG-III group in total daily drug cost as measured by Section U. Median costs within the Rehabilitation groups range from approximately $6.50 (Low Rehabilitation) to approximately $9.00 (Ultra-high Rehabilitation) whereas the lowest costs of medications were experienced by the Impaired Cognition category (approximately $3.00). The groups with the higher interquartile range (approximately $13) were the Extensive Services categories and some of the Rehabilitation groups (for example, RVC was approximately $12). The Impaired Cognition category also demonstrated the least variation in costs of medications, with an interquartile range of approximately $5.

To better understand which classes of drugs may be driving costs, we classified the drugs according to fourteen major therapeutic classes. The most expensive therapeutic drug classes are anti-infective agents (Median: $6.53) and biologics (Median: $9.73). The least expensive therapeutic drug classes are analgesics (Median: $0.10) and nutritional products (Median: $0.18). The proportion of beneficiaries within each of the major RUG-III categories are shown in Table 1.2. Variations in medication use across RUG-III groups were apparent for many medication classes and corresponded to observed variations in the active clinical diagnoses shown by RUG-III group in Table 1.1. Beneficiaries were least likely to be on biologics (1 percent) and anti-neoplastics (2 percent), regardless of RUG-III class. The majority of beneficiaries were on at least one cardiovascular medication, with substantial variation across RUG-III groups. Beneficiaries in the Rehabilitation category (67 percent) and in the Clinically Complex category (64 percent) were the most likely to be receiving at least one cardiovascular medication. Beneficiaries in the Impaired Cognition (47 percent) and Behavior (53 percent) categories were the least likely to be receiving cardiovascular medications.

Similar trends were observed across RUG-III groups for both gastrointestinal agents and endocrine/metabolic agents. More than half of beneficiaries had taken at least one gastrointestinal agent with beneficiaries in the Rehabilitation categories (67 percent) the most likely to use gastrointestinal products and beneficiaries in the Impaired Cognition or Behavioral Problem categories the least likely to receive these drugs (approximately 50 percent). With endocrine and metabolic agents, over one third of beneficiaries in the Rehabilitation, Extensive Services, Special Care, and Clinically Complex categories received these drugs, relative to approximately 25 percent of other RUG-III groups. Beneficiaries in the Rehabilitation, Extensive Services, Special Services, and Clinically Complex categories were most likely to be on anti-infective agents, with over 25 percent of beneficiaries in each on these medications. Among these RUG-III groups, beneficiaries in the the Extensive Services categories were the most likely to be taking anti-infective agents (39 percent). Less than 15 percent of beneficiaries in other RUG-III groups received these drugs.

Overall, 47 percent received at least one analgesic. Impaired Cognition (32 percent) and Behavior beneficiaries (39 percent) were less likely to receive analgesics than those in the Rehabilitation category (60 percent). Similar trends were apparent with hematological agents (approximately 20 percent Impaired Cognition vs. approximately 35 percent in the Rehabilitation groups), and topical agents (approximately 20 percent vs. approximately 37 percent in the Special Care groups). Conversely, beneficiaries in the Impaired Cognition (approximately 46 percent) and Behavior (over 50 percent) categories were more likely to receive CNS drugs relative to the other RUG-III groups (approximately 33 percent).

The highest proportion of total costs due to anti-infective use is found in the Extensive Services and Clinically Complex groups, with approximately 50 percent of drug costs attributable to the anti-infective agents. Use of biologics was relatively infrequent (approximately 1.2 percent) and the proportion of drug costs due to these agents was highly variable among the users, regardless of RUG-III group. Among people receiving anti-neoplastic medications (approxmiately 2.2 percent of beneficiaries), these agents accounted for one quarter of their total daily drug cost (Median: 27 percent; 25th percentile: 13 percent; 75th percentile: 49 percent). Regardless of RUG-III group, this measure is highly variable. While nearly one third of all beneficiaries received an endocrine medication, these agents only accounted for 8 percent of the total daily drug costs among users. Cardiovascular medications accounted for 18 percent of the total daily drug cost, which varies slightly across RUG-III group (+/− approximately 4 percent). There appears to be slightly less variation in this measure among the Extensive Services, Special Care, and Clinically Complex groups as compared to other RUG-III categories. Among the 19 percent of beneficiaries using respiratory medications, 12 percent of their drug costs were due to these agents. Higher median proportions and greater variability occurred at the end splits within the aggregate RUG-III categories. A similar pattern is observed among users of gastrointestinal agents. These medications accounted for only 13 percent (median) of the total daily costs. This measure is highly variable, regardless of RUG-III group. Only 5 percent of beneficiaries had used a genitourinary medication, accounting for only 13 percent of total drug costs (median value). This measure varied slightly across RUG-III groups.

C. Test and Validation Samples

The recursive strategies employed by stepwise regression, AID, and other fitting techniques may produce over-optimistic measures of variance explanation. For that reason, assessment of the explanatory power of alternative models required use of data that were not used in forming the models themselves. We selected at random 60 percent of the sample for use as a test sample and the remaining 40 percent for use as a validation sample. Refinements to RUG-III were developed based solely on analysis of the test sample and evaluated solely on their performance with the validation sample. Since aberrations in the test sample that may have influenced the design of refinements were absent in the validation sample, any unsupported features of the proposed models should be exposed by this approach.

D. Creation of Measure of Non-therapy Ancillary Charges From SNF Claims

Medicare Part A SNF claims were used to measure the perdiem ancillary charges. For ancillary charges developed using Medicare claims data, it was not possible to identify items with a date of service that corresponds to the period covered by the MDS assessment (used to establish the RUG-III classification). Per diem charges were calculated using Medicare claims with a covered date within a specified range of a date covered by MDS assessment. Operationally, per diem charges are derived by the sum of the charges of the ancillary therapies divided by the number of days covered by claims.

We then estimated the costs of non-therapy ancillaries, using revenue codes as extracted from the claims data. First, we identified target revenue codes and categorized charges into these conceptually meaningful categories. The categories and their related revenue codes included the following: prescription drugs/pharmacy (250-259), drugs requiring ID (630-639), IV therapy (260-269), medical and surgical supplies (270-270; 620-622), respiratory services (410-419), laboratory (300-309), oxygen (600-604), and dialysis (820-829, 830-839, 880-889).

1. Cost-to-Charge Multiplier

It is important to note that the actual ancillary costs for beneficiaries in the sample are not observed. The covered charges reported in claims are routinely discounted by the intermediary responsible for processing on the basis of audited reasonable cost. Inclusion of ancillary charges without further adjustment in our measure of per diem ancillary charges would overstate the true level of reimbursable costs, since these charges are routinely discounted before payment under the present system.

Using the appropriate annual SNF cost report (that is, the cost report for the service period covered by the claim), conversion factors were computed for each SNF included in the research data base. To be as consistent as possible, we calculated one average discount factor (the ratio of total Part A allowed cost to total Part A charges) for each facility in each year. This discount factor was applied to the facility's ancillary charges before analysis to approximate the costs of ancillary services.

E. Analysis and Findings—RUG-III Refinements

As shown by previous research and confirmed in this study, the RUG-III Extensive Services groups are associated with the highest per diem non-therapy ancillary charges of any of the RUG-III classifications, including the rehabilitation categories. For the purposes of this project, ancillary costs were divided into three categories: medications (by far the most critical predictor of overall ancillary costs), respiratory therapy, and other ancillaries. This research also showed significantly higher non-therapy ancillary costs and intra-group variance related to the variety of ancillary supplies and services needed to treat the various acute and severe health conditions characterizing beneficiaries who classify into the Extensive Services category. Figures 1 through 3 compare the mean, per diem costs of ancillary services for beneficiaries in the Extensive Services category with those of beneficiaries in other RUG-III categories.

Another key to more accurate accounting of the cost(s) associated with treating Extensive Services beneficiaries is disentangling some of the overlap between the Extensive Services and Rehabilitation categories. Under the current PPS system, the payment rate (under an index maximization approach) is the same for beneficiaries who qualify for both Extensive Services and one of the top three rehabilitation categories (Ultra High, Very High and High Rehabilitation) as for those beneficiaries who qualify only for one of the top three rehabilitation categories. Using this research data base, we found a significant number of beneficiaries qualifying for both Extensive Services and Rehabilitation.

1. Costs for Beneficiaries Who Qualify for Both Extensive Services and Rehabilitation

As shown in Figures 4 through 7, across all three ancillary categories, costs were significantly higher for beneficiaries who qualified for both Extensive Services and Rehabilitation compared to those who qualify only for a Rehabilitation category. Therefore, we considered whether those qualifying for both categories should be separately identified.

• Across all five Rehabilitation categories, mean prescription drug costs were approximately double for beneficiaries who qualified for both Extensive Services and Rehabilitation, compared to those who qualified only for Rehabilitation. (See Figure 4 for comparison of drug charges across all five Rehabilitation categories based on whether the beneficiary also qualified for Extensive Services.)
• A similar pattern was observed for respiratory therapy. Across all five rehabilitation categories, respiratory therapy costs were more than twice as high for beneficiaries who also qualified for Extensive Services as for those who qualified only for Rehabilitation (Figure 5).
• Other non-therapy ancillary costs were considerably higher for beneficiaries who qualified for both Rehabilitation and Extensive Services than for those who qualified for Rehabilitation but not Extensive Services (Figure 6).
• Total average ancillary charges for beneficiaries who qualified for both Rehabilitation and Extensive Services were also significantly higher than for those qualifying only for rehabilitation (Figure 7).

Based on these results, it makes sense, for statistical, incentive-related, and clinical reasons, to consider potential refinements which reflect the higher costs of beneficiaries in the Rehabilitation categories who also qualify for Extensive Services.

These cost differences suggested that a potential refinement could be based on interactions between existing RUG-III categories. Such a change could be implemented in either of two ways:

• A new terminal split within the current RUG-III Rehabilitation groups based on whether the beneficiary also qualified for Extensive Services. These changes would be reflected in changes in the Case Mix Index (CMI) for nursing in calculating payments for the Rehabilitation categories.
• A new RUG-III category for beneficiaries who qualify for both Extensive Services and Rehabilitation. The new category (which could be called “Rehabilitation and Extensive Services”) would be at the top of the hierarchical case-mix system.

2. Non-Therapy Ancillary Index Models

In addition, variations in non-therapy ancillary costs could be addressed through several types of index model-based refinements. There are a number of ways that index model-based refinements can be implemented:

• The models can be based on an unweighted count of the number of index model variables present or on a weighted index that assigns a relative cost factor to each of the index model variables.
• The index models can differ with respect to the RUG-III categories to which the model is applied.
• The index models can differ with respect to the number of index groups that are used.
• The index models can also vary based on the thresholds used to define groups. For the weighted index model, beneficiaries were classified based on their predicted costs.
• The index model can be applied separately to each major category; that is, each level of the RUG-III hierarchy.

In our analysis of ancillary costs, the results did not indicate strong interaction effects. There were two implications of this finding. First, the variables effects were principally additive and models which develop indexes are indicated. Second, the appropriate approach was to use regression analysis to form indexes, rather than PC-Group to identify tree models. (It should be noted that PC-Group still has some unique capabilities, employed later, to help identify optimal thresholds for an index.)

One way an index model could be used is in an “add-on” system for predicting non-therapy ancillary charges. RUG-III could be used for predicting staff time costs and the non-therapy ancillary index would be “added-on” to determine the total payment rate for beneficiaries with given characteristics. The motivation for this approach is that RUG-III has been well tested and validated for predicting staff time costs, but was not designed to capture variance in non-therapy ancillary charges. Although such a system can be described as consisting of two components, it could easily be implemented as an integrated system, as though the non-therapy ancillary component defined a new set of end-splits to RUG-III.

The index model approach allowed for a large number of items to be considered simultaneously in determining payment rates, including additional measures of severity that are not reflected in RUG-III. We designed both weighted and unweighted versions of a non-therapy ancillary index for each level of the RUG-III hierarchy, and showed that both versions resulted in large improvements in the proportion of the variance predicted by the case-mix system and some improvement in the system's ability to identify high-cost beneficiaries. The weighted version allowed items that predict much higher costs (such as receipt of IV medications) to have more impact on predicted costs than less-influential items such as shortness of breath. For this study, the weights were assigned by the researchers based on a combination of expert opinion and a comparison of cost data for the various MDS items. The weighted index model exhibited enhanced explanatory power, but at the cost of additional complexity and subjectivity.

F. Model Performance

We tested a number of potential refinements, but selected only the most powerful alternative from each type for presentation here. The most promising types of potential refinements are summarized in Table 2, and discussed below.

1. RUG-III CMI Adjustment: This potential refinement improved the ability of the case-mix system to capture variance in ancillary and total costs. Changes to the CMI alone (that is, changes to the payment rates associated with different groups but no changes to the case-mix system) will reduce the proportion of beneficiaries for whom costs are greater than payment, but will not affect the proportion of variance in costs captured by the case-mix system. The current RUG-III methodology accounted for about 6 percent of the variance in ancillary charges and 11 percent of the variance in total costs (See Table 2).

2. RUG-III (proposed, version 2001): Adding the new Extensive Services and Rehabilitation categories resulted in small improvements in statistical performance. The validation sample R-squared increased to 7.8 percent for ancillary charges, an increase of about 2 percent relative to RUG-III, and to 13.7 percent for total costs. However, the improvements associated solely with a change in the RUG-III (proposed, version 2001) methodology were substantially less than those produced by the other potential refinements that incorporated a combination of RUG-III and index model-based refinements.

In conducting this analysis, new CMIs had to be constructed. For this research, the CMIs were developed from the same 1995 through 1997 staff time measurement studies that were used to construct the indices used under the current RUG-III methodology. (See Table 3)

3. Weighted Index Model (WIM1): Under WIM1, Extensive Services beneficiaries (including those in the new Extensive Services and Rehabilitation categories) would receive an ancillary “add-on” based on the beneficiary's predicted, per diem ancillary costs for the index model qualifiers. The ancillary index has 6 groups with break points at costs at the 50th percentile or below, from the 51st through 75th percentile, from the 76th through 90th percentile, from the 91st through 95th percentile, from the 96th through 98th percentile, and the 99th percentile. The break points were calculated separately for each level of the RUG-III hierarchy.

Application of WIM1 resulted in some improvement relative to RUG-III (proposed, version 2001). For the validation sample, the model accounted for 11 percent of the variance in ancillary charges and 17 percent of the variance in total costs. Nearly 32 percent of beneficiaries in the top 10 percent of ancillary charges were also in the top 10 percent in terms of predicted costs, compared to 27.8 percent for RUG-III (proposed, version 2001).

4. Weighted Index Model 2 (WIM2): Model WIM2 extends the use of the non-therapy ancillary index to 40 RUG-III (proposed, version 2001) groups (14 Rehabilitation/Extensive Services, 3 Extensive Services, 14 Rehabilitation, 3 Special Care and 6 Clinically Complex groups), and accounted for 19 percent of the variance in total costs and 13 percent of the variance in ancillary charges. This was more than twice the R-squared of the existing RUG-III or the proposed RUG-III (version 2001) alone. The range of payments was similar to that of WIM1. Using WIM2, 32 percent of beneficiaries in the top 10 percent in terms of actual ancillary charges were also in the top 10 percent in terms of predicted ancillary charges.

Table 4 shows the distribution of Medicare beneficiaries in the 6 non-therapy ancillary index levels by RUG-III (proposed version 2001) category. The cut-off points used to define these groups are the same as for WIM1.

In the Regulatory Impact Analysis, we showed the distributional impact of these case mix refinements using the UWIM model proposed in this rule. Table 6 shows the distributional shifts of beneficiaries between the existing RUG-III model and the WIM2 Option. In addition, Tables 6.1 and 6.2 show the projected rates using the WIM2 model. (See Table 12 in the Proposed rule for the UWIM model.)

5. Unweighted Index Model (UWIM): This model is the unweighted counterpart to WIM2. While this model performed better than the current RUG-III and proposed RUG-III (version 2001) models, it was slightly outperformed by WIM2. However, we regard the unweighted model as preferable to WIM2, for two reasons. First, it is relatively simple, and employs a more familiar methodology similar to that used in classifying beneficiaries into the Extensive Services groups. Second, in developing the weighted models, the researchers had to rely more heavily on imputed data to develop the number of index levels, and the cut-off points. Therefore, even though the WIM models appear to have slightly more predictive power, they are based upon more subjective criteria. However, the WIM models are subject to additional testing using the full PPS data base, and, based on the results, this model may be reconsidered.

UWIM accounted for 11 percent of the variance in ancillary charges and 17 percent of the variance in total costs. The sensitivity and specificity of the model were slightly less than for WIM2. Using UWIM, beneficiaries are split into four groups based on the number of index model variables present.

Table 5 shows the distribution of Medicare beneficiaries in the 4 non-therapy ancillary index levels by RUG-III (proposed, version 2001) category.

The models described here focus on those upper RUG-III categories that are reflective of the skilled care needs of Medicare beneficiaries. However, since there are a small number of beneficiaries in the research data base who may be classified into one of the lower RUG-III levels, we also applied the WIM and UWIM models to the Impaired Cognition, Behavior, and Physical Function categories. Almost all the beneficiaries in these three levels of the RUG-III hierarchy grouped into the two lowest non-therapy ancillary index levels. In fact, in the UWIM model, 90 percent of the Impaired Cognition, 87.8 percent of the Behavior and 85 percent of the Physical Function observations fell into the lowest level of the non-therapy ancillary index. In these analyses, we did find a relationship between costs and the index value for these beneficiaries. However, including these groups in the model resulted in minimal additional improvement in statistical performance (See Table 7).

While these groups have not been included in the refinements proposed in this rule, we will include these RUG-III categories in additional analyses using the full PPS data base. Based on the results, we will review the applicability of the non-therapy ancillary index to the Impaired Cognition, Behavior, and Physical Function categories.

G. RUG-III Medications Data

Although the bulk of the development and analysis of potential RUG-III refinements to date have been based on Medicare claims data, the Section U drug cost data holds unique promise as a source of detailed information on the drug use of particular beneficiaries. In the coming months, once the characteristics of these new data are more fully understood, we plan to use Section U drug cost data to analyze the behavior of high-cost individuals as well as the potential effects of case mix refinements.

1. Creation of MDS-Based Drug Cost Measures

The following types of pricing are available in the Medispan Master Drug Data Base: Average wholesale price (AWP), Direct Price, Wholesaler Acquisition Cost, HCFA Federal Financial Participation (FFP) limit price, Average AWP, and the generic equivalent average price. While we translated the medications listed on the MDS with NDC codes to therapeutic classes and sub-classes, we needed to cross-link the two data systems to identify the cost of the medications. We used the average wholesale price (AWP) for medication costs for several reasons. The AWP is a national figure and not subject to regional influence resulting from purchasing contracts and other local market factors. This helps to account for the cost of dispensing. Using AWP is conservative when the price of a medication is relatively low or high, and AWP is not subject to institutional cost-shifting. Additionally, AWP, compared to other pricing options, was found to yield the lowest amount of missing cost data.

In evaluating the drug regimens of beneficiaries in our sample, we realized that because of the way some drugs are packaged, the AWP price may reflect a price for multiple doses. Examples include injectables, inhalants, elixirs, and other drugs that indicated a multi-day supply in the drug description. We generated a printout of all potential problems of this sort. A clinical pharmacist reviewed the potential appropriateness of multiple use and long-acting dosage forms and unique treatment regimens for bundling. The Physician Desk Reference, the Red Book and other sources were used in addition to the documented AWP to determine a likely constant by which to divide the cost for each potential problem. In many instances, not enough information was available to make an appropriate estimate. In these cases, the drug cost remained as indicated by the AWP.

While we were able to successfully map NDC codes to drug names (nested within therapeutic classes and sub-classes), successfully matching to a drug cost required more information. Specifically, assigning an AWP to a drug requires both the strength of the drug administered and complete information regarding the frequency with which the medication was administered. Unfortunately, many of the NDC codes included in the MDS data did not include information regarding strength. For example, we may know that a beneficiary received aspirin, but we do not know if it was 80 mg, 325 mg, or some other strength. As a result, we have substantial missing cost data. Because of the extent of missing data, we opted to impute the drug costs as opposed to excluding cases for which we did not have complete drug cost information. Analyses of the extent of missing data revealed that missing data did not vary by RUG group, State, year, or type of medication.

Nonetheless, by imputing missing drug costs, we have introduced random variations in the data that were not generated by the underlying process that we are attempting to model. Consequently, variables that explain variance in non-missing data will have no explanatory power for imputed data. The coefficients on these variables will, therefore, be biased toward zero. This bias will be small if the proportion of total variance attributable to imputation is small. However, variables explicitly or implicitly used in the imputation process may have explanatory power with regard to the imputed values. For example, if the RUG group is implicitly used as part of the imputation process, it theoretically could explain more of the variance in the dependent variable simply because RUG was used as part of the imputation algorithm. The coefficients of the variables used to impute cost data may be amplified relative to other coefficients in the explanatory models. Depending on the correlation between the RUG groups and other variables, these coefficients will also be biased in unpredictable ways. This problem could be small if the between-group variance is small (overall variance can be broken down into between-group and within-group components). Given the potential for introducing bias in our models, we opted to create two imputation algorithms.

2. RUG-Based Imputation Method

We assigned drug costs based on NDC codes recorded on Section U of the MDS evaluation forms using the following algorithm. First, if the NDC code was listed among the approximately 150,000 codes tracked by Medispan, we used the pricing information collected by Medispan. If the NDC code was not listed, but the exact name of the generic drug was listed, we calculated pricing as follows. In those instances where the RUG code (as calculated for our recording purposes and provided on the “raw” data files) was observed among beneficiaries using the drug, if only one cost was associated with the drug, it was used. If multiple costs were associated, the most likely cost was chosen based on the distribution of observed costs among beneficiaries. If the RUG code was not observed, we applied the process to a pooled distribution over all of the medication codes observed among all of the MDS records for all of the beneficiaries. If we could not match the exact generic name, we sought a match for the leading words in the generic name, and if matched, we applied the same approach (that is, selecting the most likely drug cost based on the RUG distribution). In cases where no reasonable match could be found, no price was assigned to the medication. This algorithm was iterative over the observed distribution among beneficiaries.

3. State and Year-Based Imputation Method

Because of our concerns regarding bias, we implemented a similar, but alternative algorithm to estimate the drug costs based on data contained in Section U of the MDS. We thought that missing data might vary systematically by State owing to differing data collection procedures (and software) among States. Further, we considered that coding of drugs might have improved over time. If both assumptions were true, the pattern of missing data would vary systematically through time and place. It follows that an imputation method based on time and place would be reasonable. If the NDC code was not listed among the 150,000 Medispan codes, but the exact name of the generic drug was listed, we calculated pricing as follows. If only one cost was associated with the drug within a given State and year, it was used. If multiple costs were associated, we chose the most likely cost based on the distribution of observed costs among beneficiaries. If we could not match the exact generic name, we sought a match for the leading words in the generic name, and if matched, we applied the same approach (that is, selecting the most likely drug cost using the State and year). In cases where no reasonable match could be found, no price was assigned to the medication. As with the RUG-based imputation measure, this algorithm was iterative over the observed distribution among beneficiaries.

During the course of initial analyses, we noted discrepancies between costs as measured by MDS Section U and costs as measured by SNF claims. The discrepancies between the Section U-based drug cost measure and the drug cost measure estimated from SNF claims may be due to several factors. The pharmacy cost detail codes used from the SNF claim include treatments that would not necessarily be included on the Section U according to the MDS instructions. For example, radiation treatment supplies and other procedure-related drug supplies are clearly not included on Section U. Furthermore, while applying the cost to charge ratio for pharmacy charges might appear to estimate “costs”, this adjustment may only capture the administrative step-down from the facility cost report since, in all but the largest facilities, consultant pharmacy firms supply all drugs to beneficiaries. The charge to the facility includes both its “cost” (from the pharmaceutical firm or supplier) as well as the value-added labor of the facility's consultant pharmacists who perform its drug utilization review, along with any mark-up that the consultant pharmacy contractor applies. These charges for services provided represent “costs” to the facility, and so applying the facility cost to charge ratio only discounts its administrative step-down. Finally, in most States and areas, the typical practice in nursing homes is for a new admission to have a 30-day blister pack ordered for each specified drug the resident was taking upon admission to the nursing home. Since most residents came from the hospital where drugs are dispensed daily, they generally arrive at the nursing home with less than a one-day supply of medications. As a result, the transition and ordering of medications must be very quick. In turn, the “charge” for the drug will, in many instances, include drugs that may have already been changed by the 14th day of the stay, when the MDS Section U would be completed. The net result of this practice of delivering and billing for a full 30-day supply is a higher observed cost than would be produced by estimating per diem drug cost based on an enumeration of the drugs received.

Thus, we believe that Section U-based drug cost measures may eventually provide further insight into drug utilization patterns in the SNF population as these potential sources of data inconsistency yield to further analysis. However, in view of the delay in implementing the collection of medication data on the MDS, and given the current need to address and resolve these issues before proceeding, the analysis of potential RUG-III refinements described in this report was based on SNF claims data.