HHS Logo: bird/facesU.S. Department of Health and Human Services

A Profile of Medicaid Institutional and Community-Based Long-Term Care Service Use and Expenditures Among the Aged and Disabled Using MAX 2002: Final Report

Audra T. Wenzlow, Robert Schmitz and Kathy Shepperson

Mathematica Policy Research, Inc.

January 18, 2008

PDF Version (74 PDF pages)


This report was prepared under contract #HHS-100-97-0013 between the U.S. Department of Health and Human Services (HHS), Office of Disability, Aging and Long-Term Care Policy (DALTCP) and Mathematica Policy Research, Inc. For additional information about this subject, you can visit the DALTCP home page at http://aspe.hhs.gov/_/office_specific/daltcp.cfm or contact the ASPE Project Officer, John Drabek, at HHS/ASPE/DALTCP, Room 424E, H.H. Humphrey Building, 200 Independence Avenue, S.W., Washington, D.C. 20201. His e-mail address is: John.Drabek@hhs.gov.

The opinions and views expressed in this report are those of the authors. They do not necessarily reflect the views of the Department of Health and Human Services, the contractor or any other funding organization.



TABLE OF CONTENTS

EXECUTIVE SUMMARY
I. BACKGROUND AND OBJECTIVES
Federal and State Long-Term Care Policies
Recent Literature
The Goals of This Study
II. DATA AND METHODS
Using MAX PS Files to Analyze Long-Term Care
Data Strengths and Limitations
Analysis Methods
III. USING MAX TO EXAMINE INSTITUTIONAL AND COMMUNITY-BASED LONG-TERM CARE SERVICE USE AND EXPENDITURES IN 2002
Measuring the Balance of Institutional and Community-Based Long-Term Care: Expenditure Versus Utilization-Based Measures
Analyses by Subgroup: Comparing Aged and Disabled Enrollees
Examining Service-Level Detail for Community-Based Services
Other Medicaid Service Use and Expenditures Among Long-Term Care Users
IV. SUMMARY AND CONCLUSIONS
Summary of Key Results and Their Implications
The Future of MAX for the Study of Medicaid Long-Term Care
REFERENCES
APPENDICES
APPENDIX A: MAX Long-Term Care and Other Service Types
APPENDIX B: Glossary of Terms
APPENDIX C: State Long-Term Care Data Anomalies
APPENDIX D: Tables
NOTES
LIST OF FIGURES
FIGURE III.1: Composition of Medicaid Long-Term Care Users and Expenditures Among Aged and Disabled Enrollees in 2002
FIGURE III.2: Composition of Medicaid Long-Term Care Expenditures Among Aged and Disabled Enrollees in 2002
LIST OF TABLES
TABLE I.1: Optional Long-Term Care Services Covered by State Medicaid Programs in 2002

TABLE II.1: Definitions of Community-Based Long-Term Care MAX Service and Program Types

TABLE III.1: Expenditure and Utilization-Based Measures of the Balance of Institutional and Community-Based Long-Term Care for Aged and Disabled Enrollees in Select States in 2002, Ranked by Community-Based Service Share
TABLE III.2: Composition of Medicaid FFS Long-Term Care Expenditures Among Aged and Disabled Enrollees for Select States in 2002, Ranked by Community-Based Service Share
TABLE III.3: Percentage of Medicaid FFS Long-Term Care Users Who Used Community-Based Services for Select States in 2002, Ranked by Percentage Using Community-Based Services
TABLE III.4: Characteristics of Aged and Disabled Enrollees Using Medicaid Fee-for-Service Long-Term Care Services Compared with Those of All Full-Benefit Enrollees in 2002
TABLE III.5: FFS Expenditures for Personal Care, Residential Care, Home Health, and Adult Day Care per Aged or Disabled Medicaid Enrollee Using Long-Term Care Services: Top 10 States
TABLE III.6: Average Medicaid Expenditures for Aged or Disabled Enrollees Using Medicaid Fee-for-Service Long-Term Care Services in 2002, by Type of Service and Program Type
TABLE III.7: Average Medicaid Expenditures for Aged or Disabled Duals Using Medicaid Fee-for-Service Long-Term Care Services in 2002, by Type of Service and Program Type

TABLE A.1: Type-of-Service Codes in MAX, by Claim File Type
TABLE A.2: Medicaid Analytic eXtract Long-Term Care Service and Program Type Definitions

TABLE C.1: Medicaid Analytic eXtract Long-Term Care Anomalies
TABLE C.2: Expenditures for Waiver Services, Personal Care, and Home Health Reported in CY MAX, FY MSIS, and FY Form 64 Data for All States and the District of Columbia in 2002

TABLE D.1: Percentage of Medicaid FFS Long-Term Care Expenditures for Community-Based Services and Percentage of Long-Term Care Users Using Community-Based Services Among Aged and Disabled Enrollees for Select States in 2002
TABLE D.2: Average Expenditures and Expenditures per User for Medicaid FFS Institutional and Community-Based Long-Term Care Services Among Aged and Disabled Enrollees for Select States in 2002
TABLE D.3: Composition of Medicaid Long-Term Care Expenditures Among Aged and Disabled Enrollees in 2002, by Demographic and Enrollment Characteristics
TABLE D.4: Composition of Medicaid FFS Long-Term Care Expenditures Among Aged and Disabled Enrollees for Select States in 2002, by Detailed Type of Service
TABLE D.5: Average Medicaid FFS Long-Term Care Expenditures per Aged or Disabled Enrollee Using Any Long-Term Care Service for Select States in 2002, by Detailed Type of Service
TABLE D.6: FFS Expenditures per Enrollee Among All Full-Benefit Enrollees and Among Users of Long-Term Care Services in 2002, by Service Type
TABLE D.7: Percentage Using Specified FFS Among All Full-Benefit Enrollees and Among Users of Long-Term Care Services in 2002, by Service Type
TABLE D.8: FFS Expenditures per User Among All Full-Benefit Enrollees and Among Users of Long-Term Care Services in 2002, by Service Type


EXECUTIVE SUMMARY

Since 1982, states have increasingly utilized Section 1915(c) waivers and optional state community-based programs to shift long-term care for the aged and disabled from institutions to the community. New rules introduced under the Deficit Reduction Act (DRA) of 2005 provide states with even more flexibility to provide home and community-based long-term care services to their low-income populations. Two overarching goals underlie these policies: (1) to provide long-term care services more cost-effectively; and (2) to give aged and disabled people more options in how they receive their care. As baby boomers enter their senior years and increase the need for long-term care services nationally, information about how Medicaid community long-term care programs have functioned in the past will be critical for assisting states in choosing how to utilize the new options provided under the DRA. Until recently, only limited aggregate data and some national surveys have been available to examine Medicaid community-based long-term care service use and compare it with use of institutional care. The Medicaid Analytic eXtract (MAX) data system produced by Centers for Medicare & Medicaid Services now enables much more detailed analyses of long-term care utilization and expenditures at the person level.

This study evaluates the potential of using MAX Person Summary files to examine how successfully states have rebalanced their long-term care systems and how Medicaid enrollees who utilize community-based long-term care services differ from people in institutions. For 37 states we believe have reliable MAX long-term care data, we: (1) compare utilization-based measures of the balance of community versus institutional long-term care with traditional expenditure-based measures; (2) contrast patterns of long-term care service utilization and expenditures of aged and disabled subgroups; (3) examine the detailed service types that compose community-based long-term care; and (4) summarize other services used and costs incurred by long-term care users. Each analysis highlights the utility of using person-level data available in MAX to extend our knowledge of how home and community-based long-term care services are used across the country.

Key Findings About Medicaid Long-Term Care Service Use in 2002

The findings presented in this report suggest that there is significant variation across measures, across states, and across population subgroups in patterns of institutional and community-based long-term care use and expenditures. While our national estimates are based on MAX data from only 37 states, our results suggest that further person-level analyses are warranted. In comparing expenditure with utilization-based measures, we find that:

Our subgroup analyses for aged and disabled enrollees suggest that:

Service type decompositions suggest that MAX data could be used to gain much better understanding of the types of community-based services that are used in Medicaid:

Finally, our examination of non-long-term care Medicaid service use provides a broader perspective of the types of individuals that use Medicaid long-term care:

The Potential of MAX Data for Understanding Patterns of Long-Term Care

While more current data are available at the aggregate level, MAX 2002 provides the most detailed and current person-level information on all Medicaid enrollees and the services they used during a calendar year. As illustrated by our study findings, MAX can be used to address Medicaid long-term care policy questions that require knowledge about patterns of utilization and expenditures incurred by individuals. MAX could potentially be used to examine whether new enrollees, new long-term care users, or others are utilizing community-based services, and with the use of claims files, how individuals’ use of community services is associated with their future patterns of care.

Several factors that limit the usefulness of MAX data--its timeliness, its completeness, and reporting errors--have greatly improved in recent years and are expected to continue to do so. However, the utility of MAX to examine Medicaid institutional and community-based long-term care services nationally will greatly depend on improvements made by states in the accuracy and detail of long-term care data they report to the Medicaid Statistical Information System (the source data for MAX) and enhancements to MAX coding of community long-term care. We recommend the development of more detailed service-level information for services covered under Section 1915(c) waivers and state plans in future editions of MAX.


I. BACKGROUND AND OBJECTIVES

Medicaid is the largest insurer for long-term care services in the United States, covering over 60 percent of long-term care users and accounting for 45 percent of nursing home expenditures in 2002 (CMS 2006b). As health care and nursing home costs have risen nationally, states have made significant efforts to limit these costs within their state Medicaid programs (Grabowski et al. 2004). One key method used to try to control long-term care costs has been to shift long-term care provision from traditional institutional care to the community.

Studying the effects of these policy changes has been challenging because most data measuring Medicaid long-term care--for example, data reported in Form 64--are at the aggregate state level. Because expenditures depend on both utilization and service costs, aggregate expenditure measures may mask important differences in the cost and utilization rate of community relative to institutional services. Furthermore, aggregate summaries cannot be used to examine patterns of long-term care use for subgroups of Medicaid enrollees, which is essential for targeting new interventions. Meanwhile, survey data typically enable the estimation of national but not state-level summaries of use of and expenditures for Medicaid long-term care services. Because state Medicaid programs vary widely in the people they cover, services they provide, and cost of care, national measures alone provide little information about how people utilize the long-term care services that are available to them.

Since 1999, Centers for Medicare & Medicaid Services (CMS) has produced and made publicly available Medicaid administrative data collected from each state and the District of Columbia. The Medicaid Analytic eXtract (MAX) provides calendar year, person-level, eligibility and claims information for all Medicaid enrollees in the United States. In this report we examine the potential for using these data to examine long-term care use and expenditures in ways not possible using aggregate or survey data. The statistics presented in this report provide detailed information about patterns of institutional and community-based service use and expenditures for 37 states that we believe have reliable long-term care information in MAX 2002.

Our results have potential policy implications but must be interpreted in the context of Medicaid policies in place in 2002. This chapter briefly describes the federal and state long-term care policies that affect Medicaid long-term care service provision, summarizes what is known to date about utilization of these services, and describes our goals to examine Medicaid community and institutional long-term care patterns in 2002 using MAX in this report.

Federal and State Long-Term Care Policies

Historically, Medicaid had a reputation of having an “institutional bias” stemming from the requirement that all states provide nursing home care to their enrollees whereas most community-based long-term care services can be provided at state option. Like nursing home care, provision of basic home health services has also been mandatory but there is a great deal of variability in other types of institutional and community-based long-term care services that states choose to cover in their programs. Table I.1 shows the optional long-term care services covered by state Medicaid programs in 2002, including those provided via waivers and managed care programs. (For detail, see Doty 2000, Schneider et al. 2002, and Smith et al. 2000).

In general, most aged or disabled people receiving Supplemental Security Income (SSI) benefits are eligible for full Medicaid benefits including long-term care services covered in their state.1 However, states have some flexibility regarding additional populations eligible for Medicaid services. States have the option to cover all aged and disabled at higher income thresholds up to 100 percent of the federal poverty level, to cover medically needy individuals who have higher incomes but may have significant medical expenses, and/or cover only institutional care for people up to 300 percent of the SSI limit (known as the “300 percent rule”).

Since 1982, states have increasingly utilized Section 1915(c) waivers to expand community-based service coverage for their aged and disabled enrollees eligible for institutional care. The goals of the Section 1915(c) waivers and related legislation were to provide states with the flexibility to find more cost-effective ways to provide long-term care and to give aged and disabled people more options in the type of long-term care services they use. States can choose to cover only certain services under waivers. The can also limit the populations covered under waivers to specific age groups or people with specific conditions. Eligibility for waiver services is limited to people who meet the clinical criteria for institutionalization and are eligible for Medicaid institutional care. However, waivers can be used to cover people who would typically not be eligible for Medicaid services unless they resided in an institution (for example, people eligible for only institutional care services under the 300 percent rule).

Rules stipulated under the Deficit Reduction Act (DRA) of 2005 provide states with new options and even more flexibility to provide home and community-based long-term care services to their populations (Crowley 2006; Shirk 2006). Two key components of the DRA relevant to long-term care include: (1) new state options to cover community-based services to aged and disabled enrollees without waivers; and (2) the introduction of the Money Follows the Person Demonstration in January 2007 that provides selected states with enhanced federal matching funds to assist institutionalized individuals to transition to and live in the community.

Long-term care managed care programs are another tool that states can use to manage long-term care services. Except for Arizona, few states with managed care long-term care programs in 2002 covered a significant number of individuals in the state (Saucier 2005). Arizona’s long-term care program covered almost all the state’s elderly population. A number of states had Programs of All-Inclusive Care for the Elderly (PACE) but these were typically small and covered only several hundred people in each state with one or more PACE programs (National Pace Association 2004).

Recent Literature

Several state Medicaid reporting requirements have enabled analyses of Medicaid community and institutional long-term care service use nationally and by state. CMS Form 64 data are aggregate summaries of state Medicaid program expenditures that are submitted quarterly by states for services reimbursable by the Federal Government. Form 64 contains information about both institutional and community-based long-term care expenditures, which have been summarized yearly by Burwell and colleagues (see, for example, Burwell, Sredl, and Eiken 2003). Information about waiver expenditures by program type is reported in CMS Form 372 and has been used to summarize trends in expenditures for community-based waiver services nationally and by state (Eiken, Burwell, and Selig 2006; Kitchener et al. 2005a, 2005b, 2007; Reester, Missmar, Tumlinson 2004). One key result from these analyses is the finding that three-fourths of waiver services have gone to support programs for people with mental retardation and other developmental disabilities (MR/DD) (Eiken, Burwell, and Selig 2006).

The limitations of Form 64 and Form 372 data are that they provide little information about the types of people utilizing long-term care services and they do not provide any information on the other types of Medicaid services that they use. They are also limited in detail about the composition of expenditures by the type of service specified in the form. Expenditures can be examined only by the program types and service types reported. While claims data contain procedure codes that enable investigation of the services provided under waivers, for example, reporting in Forms 64 and 372 is much more aggregated. As a result, states that have investigated their Medicaid long-term care programs have often examined their state’s claims records to gain insight into how their long-term care programs function (see, for example, Robison et al. 2007). However, national data are required to understand how state experiences compare to those of other states and to examine implications of federal policies.

The Goals of This Study

The goals of this study are to examine how person-level data in the MAX data system can be used to better understand Medicaid long-term care service use and expenditures, and to evaluate the utility of MAX data for further study of long-term care. In our analyses we:

We use MAX data for calendar year 2002 for the analysis. It should be kept in mind that these data represent the long-term care legislative environment in 2002 and do not reflect changes in utilization and expenditures expected with the implementation Medicare Part D in 2006 and the DRA in 2007. In the next chapter we describe MAX Person Summary (PS) file data, how they can be used for long-term care analyses, and their strengths and limitations. Chapter III presents our analysis results. Chapter IV discusses policy implications and the utility of MAX for future long-term care research.

TABLE I.1. Optional Long-Term Care Services Covered by State Medicaid Programs in 2002
Table: Optional Long-Term Care Services Covered by State Medicaid Programs in 2002
SOURCE: Centers for Medicare & Medicaid Services 2006a, with the exception of footnoted columns.
  1. Kitchener, Ng, and Harrington 2006. All eligibles covered implies state has waivers that cover the MR/DD, aged, disabled, and children.
  2. Saucier 2005.
  3. National PACE Association 2004 (includes pre-PACE programs).


II. DATA AND METHODS

The MAX and its source data--the Medicaid Statistical Information System (MSIS)--are the primary sources of information about each of the over 50 million people enrolled in Medicaid each year. CMS produces both MSIS and MAX and makes them available publicly (with a data use agreement) for research purposes. Because Medicaid is the largest insurer of long-term care in the United States, these data provide the most detailed information currently available about people using long-term care services nationally.

Most MAX data are derived directly from MSIS.2 MSIS contains fiscal year (FY) Medicaid enrollment and claims paid information for each state and the District of Columbia. Submission of data to MSIS became a requirement as of January 1, 1999 for all states under the Balanced Budget Act of 1997, replacing form HCFA-2082 reporting requirements. The MAX data system is a cleaned version of MSIS that enables analyses of enrollment, utilization, and expenditures at the person level.

Unlike MSIS, which reflects claims as of the date they were paid, MAX reflects the services used by Medicaid enrollees during a calendar year. The MAX PS files for 2002 were used for the analyses presented in this report.3 The MAX PS files are person-level files that contain information on enrollee demographic and eligibility characteristics and summary information on claims paid for services used by each enrollee in 2002.

This chapter summarizes the long-term care use and expenditure measures in the MAX 2002 PS files, discusses the strengths and limitations of these data, and describes the methods used to analyze institutional and community long-term care in this study. Important components of this chapter are three appendices that elaborate on the definitions of MAX long-term care measures (Appendix A), provide a glossary of terms used in this report (Appendix B), and detail state MAX data anomalies (Appendix C).

Using MAX PS Files to Analyze Long-Term Care

Measures contained in the MAX 2002 PS files generally consist of: (1) demographic and enrollment measures; and (2) summary claims information that has been aggregated from claims files. Demographic and enrollment measures include age, race, gender, basis of Medicaid eligibility, dual Medicaid and Medicare enrollment, months enrolled, managed care enrollment, and other enrollment and eligibility information that can be used to characterize people using long-term care services. Summary claims information includes utilization and expenditure measures by service type, and when relevant, by program type. Both service type and program type measures in MAX PS files can be used to identify the types of Medicaid institutional and community long-term care services used and their overall expenditures in 2002.

Medicaid institutional long-term care services can generally be identified solely using service type information. There are 33 service type categories that are summarized in the MAX PS files. (See Appendix A for a listing of all MAX service types.) The four institutional long-term care types of service (TOS) include nursing facility services (TOS 07), services provided in intermediate care facilities for persons with mental retardation (ICFs/MR) (TOS 05), mental hospital services for people age 65 and older (TOS 02), and inpatient psychiatric facility services for people under age 21 (TOS 04).

Compared with institutional long-term care, Medicaid community-based long-term care services are more difficult to identify because home or community-based long-term care may include a variety of services--for example, transportation or targeted case management--that may also be used for reasons unrelated to long-term care. We consider five MAX service type measures as capturing solely home or community-based long-term care services: personal care (TOS 30), residential care (TOS 52), home health (TOS 13), adult day (TOS 54), and private duty nursing (TOS 38). Table II.1 summarizes these service types. (More detailed information about each of the institutional and community long-term care measures is provided in Appendix A.)

An important component of Medicaid community-based long-term care is waiver services. All states except Arizona cover some community-based long-term care services under Section 1915(c) provisions of the Social Security Act that enable states to waive certain federal regulations to provide home and community-based services (HCBS) to people who otherwise would require institutional care. Because services provided under waivers make up a significant portion of community long-term care expenditures--about 65 percent (Kitchener et al. 2006; authors’ computations using MSIS 2002)--waiver service information is critical to measuring community long-term care. MSIS FY 2002 data suggest that over 75 percent of waiver expenditures were for claims coded in the “Other” (or unspecified service type TOS 19), which implies that MAX type of service codes cannot be used to capture a significant share of community long-term care expenditures. A mode of identifying community-based long-term care in MAX is through program types. Total expenditures for an individual under Section 1915(c) waivers are measured by program type codes 6 and 7 in MAX PS files.4

TABLE II.1. Definitions of Community-Based Long-Term Care MAX Service and Program Types
Description of Measure
Personal Care (TOS 30). Personal services such as bathing and toileting, sometimes expanded to include light housekeeping furnished to an individual who is not an inpatient or a resident of a group home, assisted living facility, or long-term facility such as a hospital, nursing facility, ICF/MR, or institution for mental disease. Personal care services are those that individuals would typically accomplish themselves if they did not have a disability.a
Residential Care (TOS 52). Although room and board services provided in residential care facilities are not covered by Medicaid, other components of residential care--for example, personal care, 24-hour services, and chore services--can be covered. Residential care includes group, family or individual home residential care; cluster residential care; and therapeutic residential care services, assisted living, supported living, and night supervision.a
Home Health (TOS 13). Services provided at a patient's place of residence (typically a patient’s home), in compliance with a physician's written plan of care that is reviewed every 62 days--including nursing services, as defined in the State Nurse Practice Act, home health aide services, physical therapy, occupational therapy or speech pathology, and audiology services--that are provided by a home health agency or by a facility licensed by the state to provide these medical rehabilitation services.a
Adult Day Care (TOS 54). Day care services including day health and rehabilitation care, day treatment, habilitation, psoriasis day care, services in an adult day care center, and transportation to adult day care.a
Private Duty Nursing (TOS 38). Services provided by registered nurses or licensed practical nurses under direction of a physician to recipients in their own homes, hospitals, or nursing facilities as specified by the state.a
HCBS Waivers (Program Types 6 and 7). Services provided under Sections 1915(c) of the Social Security Act that enable states to provide Medicaid-financed community-based long-term care for people who would otherwise require Medicaid-covered hospital care, nursing facility care, or care in an ICF/MR. These programs can be designed to target individuals in specific age groups and with specific conditions, and the services can be restricted to certain areas of the state.
  1. Durable medical equipment and services for mental health or substance abuse treatment are excluded.

Unfortunately, summary expenditure measures in the MAX PS files are reported by service type and by program type, but not by service and program types in MAX 2002.5 As a result, using both measures to identify community long-term care can lead to double-counting because waiver expenditures may also be reported under one or more service type categories. This also implies that waiver service expenditures cannot typically be disaggregated by service type using MAX PS data. (See Appendix A and Table A.1 for the composition of waiver service expenditures reported in MSIS for FY 2002 by TOS.)

To avoid duplicate counting while incorporating waiver expenditures, we use the following method to compute community long-term care expenditures:

This computational method underestimates community long-term care expenditures because: (1) overall expenditures for people using waiver services are underestimated if they also use non-waiver community long-term care services provided at state option; and (2) services provided at state option (non-waiver services) but not included among the five community long-term care service types are not captured in the cost measures for either subgroup identified above. The first cause of the underestimate could potentially be resolved using MAX claims data, which were not available for this study. The second cause results from lack of information about which services are provided for long-term care purposes when not categorized within a specific long-term care service type or program. This data limitation could potentially be improved in future years of MAX data with more detailed reporting of service type in MSIS or with new data elements that enabled researchers to identify those services that were for long-term care but coded as more general service types.

Data Strengths and Limitations

MAX provides the potential to examine patterns of long-term care use and expenditures in much more detail than is possible with aggregated data. MAX enables researchers to analyze utilization-based rather than just expenditure-based measures and to examine service use and expenditures by subgroup. Whereas aggregate data only provide total Medicaid long-term care expenditures, MAX data can be used to additionally estimate total Medicaid expenditures among people using long-term care services. When claims data are available, MAX can also be used to examine the composition of waiver expenditures as well as variation in services used over time. MAX also provides more extensive service type information than is available in other data sources. For example, residential care, adult day care, and durable medical equipment service types are not reported on Form 64 or Form 372. Procedure codes available on claims can also be used to further examine the types of Medicaid long-term care services that are provided by individual states.

Such detailed data about over 50 million enrollees comes with some limitations. In addition to the underestimation of MAX community long-term care when using MAX PS files, there are a number of other caveats that should be considered when using MAX data to analyze patterns of long-term care:

Analysis Methods

The analyses presented in the following chapter capitalize on the strengths of MAX while taking into account the limitations of the data discussed above. Our analyses represent Medicaid enrollees from 37 states we believe have reliable data. We exclude states with extensive missing data or data that varies significantly from summary measures reported for other data sources. Specifically, we exclude Arizona, Hawaii, Massachusetts, Michigan, Montana, North Carolina, Oregon, Pennsylvania, Rhode Island, South Carolina, Tennessee, Texas, Virginia, and Washington. (See Appendix C for details regarding data anomalies in each state.) Exclusion decisions were based on comparisons made with statistics prepared by Burwell, Sredl, and Eiken (2003) from CMS Form 64, comparisons with waiver statistics reported by Kitchener, Ng, and Harrington from CMS Form 372, and knowledge about the structure of state Medicaid programs in terms of their institutional and community long-term care service provisions.

We note that not all excluded states are known to have problematic long-term care data in MAX. We excluded states from the analysis when statistics obtained using MAX varied substantially from published reports, and the accuracy of MAX data could not be confirmed (see Appendix C for more detail). MAX data for such states may be indeed be accurate but are not included in the results presented in the following chapter.

We also limit the population of long-term care users to those eligible for Medicaid as a result of age or disability and those eligible for comprehensive benefits at some point during the entire year.7 Aged enrollees include all enrollees age 65 and older in 2002. Disabled enrollees include people of all ages who are under 65 in 2002 and became eligible for Medicaid as a result of their disability. These two groups include almost all enrollees using Medicaid long-term care services. See Appendix B for further detail about basis of eligibility groups.

While we present national averages based on 37 states, the exclusion of 14 states from our analyses may bias our results. Some excluded states--Oregon and Washington, for example--are known to have strong community-based long-term care programs. However, others--Pennsylvania and Tennessee, for example--have typically spent less on community-based services than institutional care in the past. Overall, half the excluded states had lower-than-average and half had higher-than-average percentage of long-term care expenditures spent on community-based services in 2004 (O’Brien 2005). As a result, it is plausible that our national totals based on the 37 represented states closely match true averages for all Medicaid enrollees, although it is also possible that significant biases are present. Our national estimates should be interpreted with caution.


III. USING MAX TO EXAMINE INSTITUTIONAL AND COMMUNITY-BASED LONG-TERM CARE SERVICE USE AND EXPENDITURES IN 2002

In this chapter we use data from MAX 2002 to illustrate four key ways in which MAX data can be used to gain better understanding of Medicaid long-term care service use and expenditures. We first compare utilization-based measures with traditionally-computed expenditure-based measures of the balance of institutional versus community-based long-term care. We then examine differences in Medicaid long-term care use and expenditure for two important subgroups of long-term care users: aged and disabled enrollees. In the next section, we summarize use and expenditures by detailed community-based long-term care service types that are not available in other data sources. In the final analysis presented in this chapter, we investigate total and service-specific Medicaid expenditures for three groups of long-term care: users of only institutional care, users of only community-based care, and users of both TOS. Each analysis highlights the potential of using person-level or service-specific data available in MAX to better understand patterns of long-term care use in Medicaid. The chapter also provides detailed estimates of Medicaid institutional and community-based long-term care utilization and expenditures in 37 states, with potential policy or future research implications. We discuss the implications of our results in Chapter IV.

Measuring the Balance of Institutional and Community-Based Long-Term Care: Expenditure- Versus Utilization-Based Measures

To capture the extent to which community-based programs have shifted long-term care from institutional care to the community, researchers have typically relied on aggregate measures summarizing community-based long-term care expenditures as a share of total long-term care expenditures. The problem with this approach is that total expenditures depend on the number of users that utilize services as well as the cost of those services. If community-based services are less costly alternatives to institutional care, expenditure share values would naturally be skewed towards institutional long-term care. Utilization-based measures provide an important alternative measure of the balance of institutional and community-based long-term care in Medicaid.

MAX data enable the examination of utilization-based as well as expenditure-based measures. They also enable state-level analyses. Table III.1 shows our 37 sample states ranked by two alternative measures of balance of community-based care: (1) an expenditure measure defined as the percentage of expenditures that were for community-based services in 2002; and (2) a utilization-based measure defined as the percentage of long-term care users that used community-based services in 2002. (See Appendix Table D.1 for information in Table III.1 ordered alphabetically by state.) The two measures can lead to different inferences about long-term care service balance nationally and across states.

TABLE III.1. Expenditure and Utilization-Based Measures of the Balance of Institutional and Community-Based Long-Term Care for Aged and Disabled Enrollees in Select States in 2002, Ranked by Community-Based Service Share
Rank States Ranked by Percentage of Expenditure for CLTC States Ranked by Percentage Using CLTC
$ # State Total LTC Expenditures Percentage of Total LTC Expenditures for CLTC State Number of LTC Users Percentage of Total LTC Users Using CLTC
1 3 New Mexico 542,600,816 65.1 Alaska 5,351 86.5
2 1 Alaska 170,053,768 62.1 California 442,150 77.2
3 12 Vermont 188,769,963 54.8 New Mexico 20,436 67.7
4 6 Maine 523,719,228 52.7 Colorado 40,666 67.6
5 16 Wyoming 134,720,415 51.9 Idaho 15,065 67.1
6 9 Minnesota 2,156,559,593 48.3 Maine 21,126 67.1
7 4 Colorado 789,483,284 47.3 New York 376,597 66.6
8 13 Kansas 721,488,818 47.0 Alabama 63,540 65.1
9 2 California 6,474,012,070 45.7 Minnesota 83,752 63.1
10 28 New Hampshire 347,963,248 43.1 Illinois 182,040 62.8
11 7 New York 14,262,218,168 39.4 Missouri 86,969 62.3
12 27 Utah 254,038,802 38.4 Vermont 8,200 61.6
13 11 Missouri 1,269,674,117 35.0 All 37 States 2,470,774 58.8
14 20 Oklahoma 840,695,921 34.0 Kansas 37,474 58.8
  All 37 States 55,853,564,420 33.8 West Virginia 26,331 58.6
15 5 Idaho 271,976,684 33.3 Arkansas 46,856 56.8
16 14 West Virginia 538,084,789 32.9 Wyoming 5,316 56.2
17 31 South Dakota 208,020,237 32.3 Iowa 42,299 54.4
18 24 Maryland 1,229,324,525 32.1 New Jersey 94,780 53.3
19 26 Connecticut 1,845,597,817 30.9 Florida 158,986 52.1
20 18 New Jersey 2,545,888,294 30.8 Oklahoma 49,223 51.6
21 30 Nebraska 479,154,285 30.5 Nevada 9,478 50.8
22 25 Delaware 242,973,383 28.9 Kentucky 51,946 50.2
23 33 Wisconsin 1,434,230,775 27.8 District of Columbia 9,222 49.7
24 29 Georgia 1,294,811,205 27.0 Maryland 45,071 49.6
25 19 Florida 3,091,499,328 26.4 Delaware 6,851 49.1
26 21 Nevada 188,035,239 26.2 Connecticut 53,933 49.0
27 15 Arkansas 737,779,689 23.6 Utah 11,196 48.2
28 17 Iowa 797,603,243 23.5 New Hampshire 12,465 46.3
29 22 Kentucky 911,441,510 22.2 Georgia 72,476 44.9
30 35 North Dakota 252,311,287 21.8 Nebraska 19,525 44.0
31 10 Illinois 2,724,087,990 21.4 South Dakota 10,049 42.6
32 8 Alabama 949,247,839 20.6 Ohio 139,378 40.9
33 32 Ohio 3,936,270,515 20.3 Wisconsin 62,026 40.3
34 36 Louisiana 1,149,262,277 17.1 Mississippi 36,131 39.6
35 37 Indiana 1,402,625,664 16.9 North Dakota 8,838 37.7
36 23 District of Columbia 267,951,360 11.6 Louisiana 57,558 31.1
37 34 Mississippi 679,388,274 8.6 Indiana 57,474 23.0
SOURCE: Medicaid Analytic eXtract, 2002 for a sample of 37 states with valid long-term care data.
CLTC = community-based long-term care; LTC = long-term care.

The utilization-based measure far exceeded the expenditure-based measure overall and in almost all states, suggesting far greater use of community-based services than the expenditure measure might suggest. Overall, only 33.8 percent of expenditures in our 37 sample states were for community-based services whereas a majority, 58.8 percent, of long-term care users utilized community-based services.8 The share of long-term care expenditures that were used for community-based services ranged from 65.1 percent in New Mexico to 8.6 percent in Mississippi. Aside from New Mexico, other states with high expenditure-based balance measures included Alaska (62.1 percent), Vermont (54.8 percent), Maine (52.7 percent), and Wyoming (51.9 percent). In comparison, over half the sample states had a utilization-based balance measure of more than 50 percent. The community-based service utilization rate ranged from 86.5 percent in Alaska to 23.0 percent in Indiana.

While the utilization-based measure was larger than the expenditure-based measure in every state, there was significant variation across states in how the two measures compared. For example, Illinois and Alabama ranked 31st and 32nd out of the 37 states in the percent of expenditures that were for community-based services but ranked 10th and 8th, respectively, in the percent of long-term care users who used community-based services. In these two states, the utilization rate of community-based services was much higher than their share of expenditures. In contrast, the two measures of long-term care balance were most similar in New Mexico and New Hampshire.

It should be kept in mind that states differ greatly in their Medicaid programs, service costs within the state, and the scope of covered services. Community-based waiver programs, for example, can cover a large number of individuals but provide only a limited set of long-term care services. Similarly, states vary greatly in the costs of institutionalization within the state as illustrated by the two highest ranked states in terms of community-based expenditure share--New Mexico and Alaska. While community-based service expenditures per user were only slightly lower ($25,547) than costs for institutional care ($26,507) in New Mexico, they were only a third of per-user institutional care expenditures in Alaska (see Appendix Table D.2). However, less than a thousand individuals were institutionalized in Alaska at the cost of $66,934 per user. Per-user expenditures for community-based care in Alaska, at $22,792, are not so different from those in New Mexico. (See Appendix Table D.2 for total expenditures, number of users, and per-user expenditures for institutional, community-based, and both types of long-term care services by state.)

In addition to variation in cost of care and program characteristics, many other factors can affect expenditure and utilization-based measures of long-term care balance. Utilization of community-based services at some point in the year does not preclude use of institutional services during the same year; states may differ in the proportion utilizing both services. Other factors, such as the enrollee age distribution, can also affect summary measures by state. Subgroup analyses investigating these factors provide insight into how long-term care services are used across states.

Analyses by Subgroup: Comparing Aged and Disabled Enrollees

Figure III.1 illustrates how aggregate numbers can mask considerable variation in the overall balance of community and institutional long-term care expenditures by subgroup. Community and institutional long-term care service expenditures were far more balanced among young disabled enrollees than among enrollees aged 65 or older who traditionally use nursing facility care. The percentage of total long-term care expenditures accounted for by community-based services ranged from 50.4 for people under age 65, 30.8 for people between ages 65 and 74, 21.0 for people between ages 75 and 84, and 12.9 for those age 85 and older. (See Appendix Table D.3 for these and other expenditure balance statistics by population subgroup.) Community-based services accounted for an average of 19.5 percent of long-term care expenditures among all enrollees age 65 and older.

FIGURE III.1. Composition of Medicaid Long-Term Care (LTC) Users and Expenditures Among Aged and Disabled Enrollees in 2002
Bar Chart: Composition of Medicaid Long-Term Care Users and Expenditures Among Aged and Disabled Enrollees in 2002
SOURCE: Medicaid Analytic eXtract, 2002 for a sample of 37 states with valid long-term care data.
CLTC = community long-term care; ILTC = institutional long-term care.

Table III.2 shows the composition of Medicaid expenditures and Table III.3 shows the counterpart utilization-based measures separately for aged and disabled enrollees by state (also see Appendix Table D.1 for both types of information compared by state). Each table is ranked by the overall balance of long-term care service that is for community-based care. The measures depend on the people eligible, services covered, and the population composition in each state. To capture the long-term care cost burden to states, Table III.2 and Table III.3 also present the percentage of long-term care users who were aged and the percentage of Medicaid enrollees who were aged or disabled.

TABLE III.2. Composition of Medicaid FFS Long-Term Care Expenditures Among Aged and Disabled Enrollees for Select States in 2002, Ranked by Community-Based Service Share
Rank State Total LTC Expenditures Community-Based LTC Expenditures as a Percentage of Total LTC Expenditures % of LTC Users Who Were Ageda % of All Medicaid Enrollees Who Were Aged or Disabled (Burden)
$ # Overall Among Ageda Among Disabled
1 3 New Mexico 542,600,816 65.1 44.3 83.6 57.8 16.6
2 1 Alaska 170,053,768 62.1 36.8 81.4 46.0 15.4
3 12 Vermont 188,769,963 54.8 20.6 91.7 57.5 18.8
4 6 Maine 523,719,228 52.7 23.0 78.4 56.2 29.9
5 16 Wyoming 134,720,415 51.9 11.6 79.8 51.4 17.9
6 9 Minnesota 2,156,559,593 48.3 14.2 74.9 50.9 23.9
7 4 Colorado 789,483,284 47.3 18.6 77.9 54.8 23.9
8 13 Kansas 721,488,818 47.0 20.4 72.2 56.0 25.5
9 2 California 6,474,012,070 45.7 38.1 54.1 60.1 26.1
10 28 New Hampshire 347,963,248 43.1 12.4 86.8 67.3 22.9
11 7 New York 14,262,218,168 39.4 30.9 48.8 58.7 25.0
12 27 Utah 254,038,802 38.4 6.8 54.1 44.7 16.1
13 11 Missouri 1,269,674,117 35.0 16.1 60.1 62.6 22.4
14 20 Oklahoma 840,695,921 34.0 11.3 57.0 66.6 20.2
  All 37 States 55,853,564,420 33.8 19.5 50.4 59.5 25.0
15 5 Idaho 271,976,684 33.3 19.4 46.2 52.2 20.2
16 14 West Virginia 538,084,789 32.9 13.1 59.4 59.3 31.4
17 31 South Dakota 208,020,237 32.3 6.7 61.6 64.2 20.8
18 24 Maryland 1,229,324,525 32.1 10.9 62.4 56.4 22.7
19 26 Connecticut 1,845,597,817 30.9 11.3 55.8 66.0 23.0
20 18 New Jersey 2,545,888,294 30.8 21.5 43.2 65.9 28.0
21 30 Nebraska 479,154,285 30.5 9.7 55.9 63.1 19.5
22 25 Delaware 242,973,383 28.9 8.9 53.2 57.8 16.2
23 33 Wisconsin 1,434,230,775 27.8 10.1 52.6 65.8 26.1
24 29 Georgia 1,294,811,205 27.0 9.8 53.6 64.0 22.8
25 19 Florida 3,091,499,328 26.4 9.4 52.6 59.1 27.7
26 21 Nevada 188,035,239 26.2 11.8 42.2 59.2 19.8
27 15 Arkansas 737,779,689 23.6 14.0 36.3 62.0 27.7
28 17 Iowa 797,603,243 23.5 9.0 37.5 59.9 26.6
29 22 Kentucky 911,441,510 22.2 9.1 42.9 64.5 33.6
30 35 North Dakota 252,311,287 21.8 5.0 43.7 63.9 25.8
31 10 Illinois 2,724,087,990 21.4 9.5 31.4 50.5 21.4
32 8 Alabama 949,247,839 20.6 6.7 52.5 60.4 33.3
33 32 Ohio 3,936,270,515 20.3 8.8 37.4 63.5 23.3
34 36 Louisiana 1,149,262,277 17.1 2.0 28.1 49.9 25.1
35 37 Indiana 1,402,625,664 16.9 2.8 33.3 65.3 21.8
36 23 District of Columbia 267,951,360 11.6 5.6 19.1 47.3 25.6
37 34 Mississippi 679,388,274 8.6 5.8 13.1 67.9 32.9
SOURCE: Medicaid Analytic eXtract, 2002 for a sample of 37 states with valid long-term care data.
  1. Aged enrollees include all enrollees ages 65 or older.

In every state in our sample, expenditures for and utilization of community-based services relative to institutional care were significantly higher for disabled enrollees than for aged enrollees. On average, only 44.9 percent of aged but 79.3 percent of disabled used community-based care (reflecting 19.5 percent and 50.4 percent of costs, respectively). As a result, the states with the highest balance of community versus institutional long-term care expenditures and service use overall were those with the highest community-based service expenditures or use among the aged.

TABLE III.3. Percentage of Medicaid FFS Long-Term Care Users Who Used Community-Based Services in Select States in 2002, Ranked by Percentage Using Community-Based Services
Rank State Number of LTC Users Community-Based LTC Users as a Percentage of Total LTC Users % of LTC Users Who Were Ageda % of All Medicaid Enrollees Who Were Aged or Disabled (Burden)
# $ Overall Among Ageda Among Disabled
1 2 Alaska 5,351 86.5 80.1 92.0 46.0 15.4
2 9 California 442,150 77.2 71.6 85.7 60.1 26.1
3 1 New Mexico 20,436 67.7 53.4 87.2 57.8 16.6
4 7 Colorado 40,666 67.6 49.6 89.3 54.8 23.9
5 15 Idaho 15,065 67.1 56.1 79.2 52.2 20.2
6 4 Maine 21,126 67.1 48.0 91.6 56.2 29.9
7 11 New York 376,597 66.6 52.5 86.7 58.7 25.0
8 32 Alabama 63,540 65.1 49.1 89.7 60.4 33.3
9 6 Minnesota 83,752 63.1 38.6 88.4 50.9 23.9
10 31 Illinois 182,040 62.8 47.0 78.9 50.5 21.4
11 13 Missouri 86,969 62.3 50.6 81.9 62.6 22.4
12 3 Vermont 8,200 61.6 37.7 93.8 57.5 18.8
  All 37 States 2,470,774 58.8 44.9 79.3 59.5 25.0
13 8 Kansas 37,474 58.8 36.8 86.8 56.0 25.5
14 16 West Virginia 26,331 58.6 42.9 81.4 59.3 31.4
15 27 Arkansas 46,856 56.8 47.0 72.9 62.0 27.7
16 5 Wyoming 5,316 56.2 28.3 85.6 51.4 17.9
17 28 Iowa 42,299 54.4 39.1 77.1 59.9 26.6
18