Citation
Interstate variations in long-term care spending for the elderly

Material Information

Title:
Interstate variations in long-term care spending for the elderly
Creator:
Thomas, Patricia
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
xi, 171 leaves : illustrations ; 28 cm

Subjects

Subjects / Keywords:
Older people -- Long-term care ( lcsh )
Medicaid ( lcsh )
Medical care -- Political aspects ( lcsh )
Medicaid ( fast )
Medical care -- Political aspects ( fast )
Older people -- Long-term care ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 165-171).
Thesis:
Submitted in partial fulfillment of the requirements for the degree, Doctor of Philosophy, Public Administration
General Note:
School of Public Affairs
Statement of Responsibility:
by Patricia Thomas.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
36465527 ( OCLC )
ocm36465527
Classification:
LD1190.P86 1996d .T46 ( lcc )

Downloads

This item has the following downloads:


Full Text
INTERSTATE VARIATIONS IN LONG-TERM CARE
SPENDING FOR THE ELDERLY
by
Patricia Thomas
B.S., Northeastern University, 1978
M.S., Boston University, 1981
M.B.A., University of Vermont, 1988
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
in partial fidfillment of the requirement for the degree of
Doctor of Philosophy
Public Administration
1996


1996 by Patricia Thomas
All rights reserved.


I
This Thesis for the Doctor of Philosophy
degree by
Patricia A. Thomas
has been approved for the
Graduate School of Public Affairs
Richard Foster
^ Wv, 199/
Date


Thomas, Patricia (Ph.D., Public Administration)
Interstate Variations in Long-Term Care Spending for the Elderly
Thesis directed by Professor Peter deLeon
ABSTRACT
In 1987, Medicaid expenditure on long-term care (LTC) differed among states, on a per
elderly basis, by more than a factor of ten. This variation in state spending on LTC for the elderly
holds serious policy implications regarding horizontal, inter-generational, and taxpayer equity. The
purpose of this study was to first, determine the factors driving this varied resource allocation and,
second, clarify the role of federal, state, and local government policy and administration on state
spending on LTC for the elderly. The basic theoretical underpinning for the study is Feldstein and
Scanlons contention that LTC utilization is a function of both demand and supply, each of which is
constrained by a states ability and willingness to fund LTC services.
Medicaid expenditure on LTC for the elderly was analyzed, on a per 1,000 state elderly
population basis, for years 1981-1988, by service area (nursing home, home care, and the 2176
Waiver Program [1982-1988]) and by components of spending (total expenditure, average
expenditure, and number of elderly recipients). Private spending on nursing home care for the
elderly was analyzed across states for years 1987-1988.
These dependent variables were regressed against state socio-economic and demographic
variables alone (reduced model) and, additionally, with state-selected policy (Medicaid program and
health system characteristics) and political variables (expanded model).
Findings suggest that state socio-economic and demographic variables have a stronger
relative effect on average Medicaid expenditure per elderly LTC recipient, while state-selected
policy variables have a stronger relative effect on the number of elderly recipients of Medicaid
LTC services. Results of this study further indicate that "poor" states increase access to Medicaid
LTC services and restrict the benefit package while "rich" states enhance the Medicaid LTC benefit
package, not access.
Reform strategies that reduce Medicaid per diem nursing home reimbursement rates,
optional Medicaid eligibility categories, or target optional Medicaid service may decrease Medicaid
IV


LTC utilization. Reform strategies that further reduce average length of stay for the Medicare
population in acute care hospitals may increase Medicaid LTC utilization. Nursing home bed stock
remains a powerful state policy variable regarding both public and private spending on nursing
home care.
This abstract accurately represents the content of the candidates thesis. I recommend its
publication.
Signed
Peter deLeon
v


ACKNOWLEDGEMENTS
With gratitude and appreciation, I must acknowledge the contributions and support of the
following persons:
Carol Jacobson and Roger Carver for logistical computer support with statistical analysis;
Ellen Fryxell for editorial comments and manuscript preparation;
Dr. Thomas Grannemann for generous guidance with conceptual and design development;
Dr. William Atkinson, Linda deLeon, Richard Foster, and Sam Overman, members of my
committee, for intellectual insights and good will throughout the defense;
A very special thank you to Professor Peter deLeon, my chair, mentor, and spiritual
guide, without whom this study would not have come to fruition;
A final note of thanks to the Thomas family for epitomizing the word.
I am deeply indebted and grateful to each of you.
Thank you.
Patricia Thomas
June, 1996
Andover, Massachusetts


CONTENTS
CHAPTER Page
1. INTRODUCTION ...................................................................... 1
Problem Statement............................................................. 2
Dissertation Organization..................................................... 6
2. BACKGROUND ....................................................................... 7
Definition of Long-Term Care.................................................. 7
Key Factors Influencing Demand for Long-Term Care ........................ 7
History of Medicaid Financing of Long-Term Care........................ 11
Medicare............................................................. 11
Medicaid............................................................. 12
Medicaid Eligibility................................................. 13
Medicaid Benefits ................................................. 14
The Problems Associated with Medicaid LTC Financing..................... 14
3. LITERATURE REVIEW................................................................ 17
Theory ...................................................................... 17
The Medical Model.................................................... 17
The Cultural Model................................................... 20
The Economic Model................................................... 22
The Structural Model................................................. 25
The Political Model.................................................. 28
The Public Goods Model............................................... 29
The Incremental Budgeting Model ..................................... 31
Empirical Work............................................................... 32
Determinants of Geographic Variation in Medicaid
Expenditures and Utilization Patterns........................................ 33
Determinants of Geographic Variation in Expenditure and/
or Utilization Patterns Specific to Nursing Home Care................ 42
Determinants of Geographic Variation in Medicare Home
Health Care Utilization.............................................. 49
Summary of Empirical Work ........................................... 50
4. METHODOLOGY ................................................................... 54
Model........................................................................ 55
Hypotheses................................................................... 56
Independent Variables................................................ 57
Dependent Variables.................................................. 58
Medicaid Dependent Variables (1981 to 1988)...................... 59
Private Spending Dependent Variable (1987 and 1988).................. 59
vu


Page
Design ....................................................................... 61
Unit of Analysis ............................................................. 62
Data Collection............................................................... 63
Data Analysis................................................................. 66
5. RESULTS .......................................................................... 67
Descriptive Analysis.......................................................... 67
Dependent Variables................................................... 67
Independent Variables.................................................. 71
Correlation Coefficients.............................................. 74
Regression Analysis .......................................................... 79
Dependent Variable.................................................... 79
Model ................................................................ 79
Total Medicaid Expenditure on Long-Term Care for the Elderly
and Medicaid Expenditure on Nursing Home Care for the Elderly......... 80
Elder Medicaid Recipients of Nursing Home Care ................ 91
Average Medicaid Expenditure per Elder
Recipient of Nursing Home Care........................................ 96
Percent of Medicaid Expenditure on LTC
for the Elderly Spent on Nursing Home Care.............................101
Medicaid Expenditure for Home Health Care for the Elderly.......106
Elderly Medicaid Recipients of Home Health Care Services............113
Average Medicaid Expenditure per Elder
Recipient of Medicaid Home Health Care Services ......................118
Medicaid Expenditure on the 2176 Waiver Program for the Elderly........123
Medicaid Nursing Home Expenditure/Recipient Equations
Inclusive of Non-Institutional Medicaid Expenditure/Recipient Data...128
Private Expenditure on Nursing Home Care for the Elderly,
Per 1,000 State Elderly Population, 1987 and 1988 ............... 131
Summary Of Findings...........................................................134
6. CONCLUSIONS .......................................................................143
Policy Considerations ........................................................148
The Future of Medicaid and Long-Term Care.....................................154
Implications for Further Research..............................................160
APPENDIX...............................................................................163
A. DATA SOURCES
Independent Variables.................................................163
Dependent Variables...................................................164
BIBLIOGRAPHY...........................................................................165
VUl


FIGURES
Figure 1.1
Page
. 4
IX


TABLES
Table Page
1.1. Interstate Variation in Long-Term Care ................................................. 3
2.1. Forecast of LTC Utilization by the Elderly ............................................. 8
2.2. Income of the Elderly .................................................................. 9
4.1. Summary of Hypothesized Relationships...................................................60
5.1. Descriptive Data of Dependent Variables, By Year, 1981 1988,
in 1982 Dollars.........................................................................68
5.2. Summary of Descriptive Data of Dependent Variables, Per 1,000 State
Elderly Population, By State, 1981 1988 ............................................ 70
5.3. Descriptive Data of Independent Variables, By Year, 1981 1988,
in 1982 Dollars.........................................................................72
5.4. Summary of Descriptive Data of Independent Variables, By State,
1981 1988' "' . ..................................................................75
5.5. Correlation Matrix......................................................................76
5.6. State Medicaid Spending on Long-Term Care for the Elderly
Per 1,000 Elderly Population, 1981 1988 .............................................81
5.7. State Medicaid Expenditure on Nursing Home Care for the Elderly
Per 1,000 Elderly Population, 1981 1988 ............................................ 82
5.8. Number of Elderly Medicaid Recipients of Nursing Home Care
Per 1,000 Elderly Population, 1981 1988 ............................................ 92
5.9. Average Medicaid Expenditure on Nursing Home Care for the Elderly
Per 1,000 Elderly Population, 1981 1988 ............................................ 97
5.10. Percent of Medicaid Long-Term Spending for the Elderly or Institutional
(Nursing Home) Care Per 1,000 Elderly Population, 1981 1988 ........................ 102
5.11. Medicaid Expenditure on Home Care for the Elderly Per 1,000 State
Elderly Population, 1981 1988 ...................................................... 108
5.12. Number of Elderly Medicaid Recipients of Home Care Per 1,000 State
Elderly Population, 1981 1988 ...................................................... 114
x


5.13. Average Medicaid Expenditure Per Elderly Recipient of Home Care
Per 1,000 State Elderly Population, 1981 1988 ................................... 119
5.14. Medicaid Expenditure on the Medicaid 2176 Waiver Program
Per 1,000 State Elderly Population, 1981 1988 ................................... 124
5.15. Medicaid Expenditure on Nursing Home Care for the Elderly, Per 1,000 State
Elderly Population Basis, 1981 1988, Including Medicaid Expenditure on
Home Care and 2176 Waiver Program in the Model.....................................129
5.16. Elderly Medicaid Recipients of Nursing Home Care, Per 1,000 State Elderly
Population, 1981 1988, Including the number of Elderly Medicaid Recipients
of Home Care in the Model..........................................................130
5.17 Private Expenditure on Nursing Home Care for the Elderly, State Level,
1987 and 1988 ..................................................................... 132
5.18. Summary of Statistically Significant Findings, Inclusive of New York................136
5.19. Summary of Statistically Significant Findings, Exclusive of New York................137
xi


CHAPTER 1
INTRODUCTION
Long-term care for the elderly consumes approximately one-third of all federal and state
Medicaid dollars (Meltzer, 1988). Financial pressure is expected to intensify dramatically over the
next three decades, especially in light of recent Republican budget reduction proposals.
There are important demographic, economic, and psycho-social factors setting the
contextual framework for this rise in long-term care (LTC) spending. The essential forecasts are
that demand for formal (i.e., paid) LTC services will double in the next 30 years as informal (i.e.,
unpaid) caregiving is reduced because of geographic mobility of family members and working
patterns of female family members. The informal caregiver network currently provides about 75 %
of all LTC services in this country (Rivlin & Weiner, 1988; Shearer, 1989). Annual expenditure
for the remaining 25% of LTC services, i.e., paid services, is estimated at $45 billion in 1985 and
$108 billion in 1993 (United States General Accounting Office [GAO], 1988; Burner, Waldo, &
McKirsich, 1992).
LTC services for the elderly are financed, principally, as either personal out-of-pocket
expenses by the aged or their families or by state welfare programs, i.e., Medicaid. Medicaid
spending for LTC within and across states varies widely, not only in absolute dollars, but also in
the service mix provided (Harrington, Newcomer, & Estes, 1985).
The primary purpose of this study is to explain interstate variation in state Medicaid
spending on LTC for the elderly. A secondary purpose of this study is to approximate, then
explain, interstate variation in private spending on nursing home care for the elderly.
1


As such, the results of this study are important because they enable policy makers to:
1. Better understand the determinants of long-term care spending and utilization;
2. Identify those variables that are amenable to policy manipulation;
3. Evaluate long-term care spending and utilization patterns relative to state and federal
objectives; and, finally
4. More accurately predict long-term care spending and utilization patterns under various
policy scenarios.
The ability to more accurately predict the consequences of restructuring the LTC financing
and/or delivery system is especially relevant as a public policy issue given the recent Congressional
proposal to reduce the projected growth in Medicaid spending by approximately 20% over the next
seven years (Toner, 1995).
Problem Statement
In 1988, long-term care consumed about 47% of the national Medicaid budget, with LTC
for the elderly consuming approximately 31% of the national Medicaid budget (Meltzer, 1988).
These expenditures are particularly impressive when one considers that the elderly composed only
12% of the population in 1988 while the national and global trend is toward an increasingly elderly
population (Statistical Abstract of the United States, 1988; Rivlin & Weiner, 1988).
Within these national figures lie wide variations in state spending. Given the great latitude
states have in setting specific eligibility criteria, benefits, reimbursement policies, utilization
controls, and regulation of the local health care environment, variation in Medicaid spending on
LTC for the elderly is not surprising. Nevertheless, the degree of variation is impressive.
The following chart (Table 1.1) provides information on the kind of interstate Medicaid
variation on LTC that is commonplace:
2


Table 1.1. Interstate Variation in Long-Term Care
State # Elderly in millions (1980) Elderly as % of general population (1980) Public LTC Spending on the elderly in millions (1984) Public LTC Spending on the elderly as % of State 1985 Medicaid Budget
Alaska .012 3.0% $ 12 40%
California 2.400 10.2% $ 588 29%
Colorado .250 8.6% $ 88 45%
D.C. .075 11.6& $ 38 20%
Minnesota .480 11.8% $ 382 69%
Source: Pierce and Hansen, 1987
Although this data are comparatively old, similar disparities continue today (Burner et al., 1992).
Howell, Baugh and Pine (1988) studied patterns of Medicaid utilization and expenditures
in selected states from 1980 through 1984. Average 1981 Medicaid long-term care expenditures
per Supplemental Security Income (SSI) enrollee across the five states studied varied widely, from
$1,428 per aged enrollee in California to $4,937 in New York, a threefold difference. For
disabled enrollees, expenditures ranged from $835 per enrollee in Tennessee to $2,157 in
Michigan. Figure 1.1 was compiled using Medicaid data published by the Health Care Financing
Administration (HCFA) for fiscal year 1987. It demonstrates that on a per-elderly-person basis,
Medicaid payments for both nursing home and home health differ among states by more than a
factor of ten. New York state alone accounted for 80% of the national Medicaid expenditure on
home care in 1986 (Swan, 1990).
These interstate variations in Medicaid spending for LTC for the elderly present the
central question, why? What factors explain this great variation, and what are their relative
importance? These variations in state spending on long-term care for the elderly hold serious
policy implications regarding horizontal, inter-generational, and taxpayer equity. For example,
3


Figure 1.1
MEDICAID NURSING HOME PAYMENTS PER ELDERLY PERSON
1987
14
12
t
i/i 8
L-
0
CC (
UJ o
m
2
1 4
2
0
0 100 200 300 400 500 600 700 800 900 1000
PAYMENTS (DOLLARS)
MEDICAID HOME HEALTH PAYMENTS PER ELDERLY PERSON
1987
0- 40 80 120 160 200 240 280 320 360 400
PAYMENTS (DOLLARS)
4


Scanlon and Feder (1984) consider state variation in nursing home spending to be economic
discrimination that poses a severe access problem for the disabled elderly most in need of care.
Their research indicates that where supply of nursing home beds is most limited, smaller
proportions of the most impaired population actually reside in nursing homes, while in states with
the highest nursing home bed-to-elderly population ratios, more than 90% of persons most in need
(unmarried persons 75 or older, needing assistance in all activities of daily living) were in nursing
homes.
Real interstate variations in Medicaid spending hold serious access, equity, and efficiency
implications for both recipients and taxpayers. This study will offer empirical evidence regarding
the factors driving the varied resource allocation of public and private dollars across states on one
service area (LTC) for one group (the elderly).
Commerce Committee Chairman, Thomas Bliley, Jr., a Republican Representative from
Virginia states (1995), "Medicaid is broken and we all know it. Its a complex, bureaucratic, and
uncontrolled entitlement riddled with waste and inefficiency. As such, Medicaid is frequently
cited as a state and federal "budget buster" as budgetary controls are considered of limited success
because of the open-ended entitlement nature of the program (Wildavsky, 1988). This study is
designed to help clarify the role of state-selected policy and political variables on Medicaid
spending on LTC for the elderly. Differentiating the influence of various state-selected variables
on access, expenditures, and service mix of the state elderly population to taxpayer- subsidized
LTC services is important to state policy makers, Medicaid administrators, health planners and
regulators, and providers. Determining which state-selected variables affect various aspects of
LTC spending enables policy makers to align incentives with state and federal objectives regarding
long-term care.
5


Dissertation Organization
This dissertation is organized into five remaining chapters, namely, the background of
LTC, the literature review, the methodology, the findings of the study, and finally, the conclusions
of the study.
The next chapter defines then discusses key financing aspects of LTC in the United States.
It also summarizes the major problems associated with Medicaid funding of LTC. This
background information is critical to the context of LTC.
The review of the literature, chapter 3, organizes relevant LTC and Medicaid work around
both theoretical and empirical bases. It reflects an inter-disciplinary approach to analyzing state
variation in Medicaid spending. The essential assumptions and theoretical arguments upon which
this study is based are offered in this section.
Chapter 4 presents the method of inquiry used for this study. Critical issues of design,
modeling, variable selection, and hypothesized relationships are discussed.
The results of the study are outlined in chapter 5. The findings are organized by both
descriptive and quantitative analysis. The results of the regression analysis, which is intended to
explain interstate variation in elderly Medicaid recipients of LTC and Medicaid expenditure levels,
is presented inclusive and exclusive of state-selected variables.
The sixth chapter concludes this study with a discussion of the policy implications of the
study, the possible future of Medicaid funding of LTC, and the implications of the study for future
research.
6


CHAPTER 2
BACKGROUND
A critical assessment of the results of this study, namely determining the factors that
explain state variation in spending on LTC for the elderly, requires an understanding of four inter-
related background areas: a working definition of LTC; the key factors influencing demand for
LTC; the history of Medicaid financing of LTC; and the main problems associated with Medicaid
financing of LTC.
This background information will be drawn upon extensively in discussing both the results
of this study and its implications for further research.
Definition of Long-Term Care
Long-term care is defined as the range of health, social, emotional, and personal care
services needed by functionally disabled individuals who require assistance in performing necessary
daily activities, i.e., eating, bathing, toiletry, dressing, or mobility. These activities are
appropriately called activities of daily living (ADLs). Long-term care services also encompass
instrumental activities of daily living (IADLs) such as housekeeping, shopping, meal preparation,
laundry and household chores (Pierce & Hansen, 1987).
Kev Factors Influencing Demand for Long-Term Care
Many psycho-social, economic, demographic and epidemiological factors influence the
demand for LTC services.
Demographic projections indicate that the number of elderly (age 65 or older) in the
United States will grow from 31.3 million persons in 1986 to 50.3 million in 2016, an increase of
7


61%. The over-85 population is expected to grow from 3.5 million in 1986 to 5.9 million in 2000,
to 7.2 million in 2020. It is the over-85 age group that is most important to the LTC issue, since
22% of this population currently reside in nursing homes and 35% of the remaining non-
institutionalized group require personal care assistance (Rivlin & Weiner, 1988).
Coupling these demographic projections with current LTC utilization, epidemiological, and
medical effectiveness patterns provides the following forecast of LTC usage, drawing a distinction
between nursing home and home-care services (Table 2.1).
Table 2.1. Forecast of LTC Utilization by the Elderly
Nursing Home Services
Age Year 1990 Year 2005 Year 2020
Number* % Population Number % Population Number % Population
65-74 .404 18 .390 12 .659 17
75-84 .913 40 1.216 37 1.295 32
85 & over .968 42 1.666 51 2.067 51
Total 2.285 100 3.272 100 4.021 100
Home Care Services
Age Year 1990 Year 2005 Year 2020
Number* % Population Number % Population Number % Population
65-74 1.235 31 1.306 24 1.840 29
75-84 1.715 43 2.295 43 2.230 37
85 & over 1.059 26 1.803 33 2.190 34
Total 4.009 100 5.404 100 6.359 100
* Numbers in millions
Source: Rivlin and Weiner, 1988
8


Utilization of formal LTC services is expected to approximately double in 30 years.
Financing this increase in LTC utilization is expected to be an even greater real and relative
burden, as the LTC inflation rate will probably continue to outstrip the consumer price index (CPI)
by about 2 percentage points (Rivlin & Weiner, 1988).
Financially, the elderly are a heterogenous population. The already wide gap between the
median income of the "young" elderly (under 75) and the "old" elderly (over 85) is expected to
grow dramatically over the next 30 years:
Table 2.2. Income of the Elderly
1987 2016*
Median Family Income of "Young" Elderly $10,806 $20,303
Median Family Income of "Old" Elderly $ 6,837 $ 7,999
Percent "Old" to "Young" Median Family Income 63% 40%
* in constant 1987 dollars
Source: Rivlin and Weiner, 1988
It is the "old" elderly who are at the highest risk for needing LTC services and who have
the most limited income. Therefore, they are the population most likely to continue to rely on
public monies for paid LTC services (Rivlin & Weiner, 1988).
Acute-hospital care cost-containment practices blossomed in the 1980s in response to
spiraling acute-care costs. The principal acute-care cost-containment mechanism is a diagnosis-
related payment scheme that pays a set fee to the hospital based on a patients primary diagnosis.
This prospective reimbursement scheme is widely known as diagnosis-related groupings (DRGs).
DRGs were designed to create a financial incentive for hospitals to increase efficiency, decrease
utilization, and decrease length of stay of the Medicare population in acute-care hospitals. DRGs
9


have been associated with increased utilization of home and community and nursing home post-
hospital care. Harlow and Wilson (1985) found a 196% increase for in-home skilled nursing and a
63 % increase in personal-care services in the year following the implementation of DRGs. The
expenditures associated with post-hospital care for the elderly are felt, predominantly, by the
Medicare program although an effect on private payers has been demonstrated (Morrisey, Sloan &
Valvona, 1988). It is unlikely that acute-care cost-containment measures will weaken in the near
future. Shortened hospital stays for the elderly may ultimately lead to increased demand for LTC.
The bulk of unpaid LTC services are provided by family members, usually spouses,
daughters, or daughters-in-law. At present, approximately 75% of LTC service is provided by this
informal network. It is estimated that 27 million days of unpaid care are provided each week at
untold emotional, physical, and financial hardship to the caregivers. These costs are considered the
"hidden" costs of LTC in our current system. Changes in the size of families, the structure of
families, geographic mobility of family members, working patterns of female family members, and
the general health of adult children with "old" elderly parents may significantly reduce the ability
of this network to provide assistance. The diminished ability of the informal caregiver system to
provide unpaid LTC services for the elderly will fundamentally alter financial projections of
required public monies to pay for Medicaid LTC recipients.
In comparing the relative mix of LTC services paid for with Medicaid funds (i.e., home
care versus community care versus institutional care), it is important to note that both the American
elderly and the younger population have expressed a preference for receiving LTC in their home as
opposed to a nursing home (National Long-Term Care Survey, 1982 for the attitude of the elderly;
Harris Poll, 1988 for the non-elderly). Despite this strong preference, only 25% of the disabled
elderly receive any funded in-home services (DHHS, LTC Survey, 1982).
10


The role of private-sector initiatives in financing LTC for the elderly, particularly private-
sector LTC insurance, is at present highly speculative. Currently, less than 1 % of LTC services
are paid with private insurance. The Health Insurance Association of America (1991) reported
that, as of December 1989, more than 1.5 million long-term care insurance policies had been
purchased. This figure represented an increase of 36% over 1988. Moreover, the prospect of
rapid saturation of private LTC insurance policies has been considered slim until such issues as
price, policy restrictions, and screening mechanisms are resolved (Rivlin & Weiner, 1988).
However, as Zedlewski, Raffaty and McBride (1991) state: "Ultimately, the viability of private
financing for long-term care will hinge on politics that reduce the cost of long-term care insurance
and encourage voluntary purchase of insurance among the elderly." The recent GOP
Congressional budget proposal (November, 1995) aimed at reducing the projected growth in
Medicaid spending by 18 to 20% over the next seven years includes a tax reform amendment
allowing individuals to deduct premiums paid for private-sector long-term care insurance from their
taxable income. This tax reform measure could catapult the private-sector LTC insurance industry
into the 21st century regardless of current flaws in the product.
History of Medicaid Financing of Long-Term Care
It is important to clarify the financing role of Medicare and Medicaid in the current LTC
system in the United States.
Medicare
Medicare is a federally funded and administered health-insurance program enacted into law
in 1965. It provides acute-medical care primarily to persons over the age of 65. Medicare Part A
covers hospitalization, skilled nursing home care for upwards of 100 days, and intermittent home-
11


care services. Skilled nursing home care must be considered medical follow-up to an acute-care
episode and be ordered by a physician to be covered by Medicare (Pierce & Hansen, 1987). In
1993 4% of Medicare dollars were spent on skilled nursing home care (HCFA, Office of the
Actuary, 1995).
Medicare Part B is a voluntary, supplemental program that covers physician services, lab
services, and outpatient services. All services have limits and deductibles. No long-term or
chronic-care service provisions exist within Medicare.
The recent Republican Congressional proposal to balance the federal budget in seven years
calls for a reduction in the projected growth in Medicare spending by $270 billion. This proposal
would encourage the elderly to enroll in private health-care plans, such as health-maintenance
organizations or new medical savings accounts. The plan would also have Medicare beneficiaries
pay 31.5% of their optional Part B premium instead of having their share drop to 25% under
current law, and the proposal would tighten payments to doctors, hospitals, laboratories, and
nursing homes to keep spending within limits (Zuckerman, 1995).
Medicaid
Medicaid is a federal-state welfare program enacted into law in 1966. It provides a wide
range of hospital, physician, and long-term care services for low-income recipients (Pierce &
Hansen, 1987). Medicaid was built upon the historic role state and local governments played in
providing health care to the poor (McDevitt & Buczko, 1985). From Medicaids conception in
1966, states were given great latitude in structuring their programs within broad federal guidelines
on mandated eligibility and benefits.
States receive federal financial participation for both mandatory and optional services. By
the early 1980s, economic and political forces were placing serious fiscal constraints on Medicaid.
12


These forces led to the passage of the Omnibus Budget Reconciliation Act (OBRA) of 1981, and
the Tax Equity and Fiscal Responsibility Act (TEFRA) of 1982 (Harrington et al., 1985). One
purpose of these acts was to control program expenditures by reducing the level of federal financial
participation in Medicaid. Expenditures for health-care services continued to spiral upward even as
federal contribution declined and state revenues were reduced by the economic slowdown of the
early 1980s (Buchanan, 1987). OBRA and TEFRA encouraged states to control Medicaid costs by
offering them even greater discretion in determining eligibility criteria, benefit packages, and
method of provider reimbursement (McDevitt & Buczko, 1985; GAO, April, 1987). The Federal
Government currently matches state funds by 50% to 78% depending, primarily, on the states per
capita income (Jazwiecki, 1989).
Section 2176 of OBRA allowed for new home and community-based waiver programs,
which call for experimental spending on home and community services for Medicaid recipients who
qualified for nursing home services. These services are designed as alternatives to nursing home
care. In 1982, six states participated in the 2176 Waiver Program. By 1989, 46 states were
participating in the program (Miller, 1990).
Medicaid Eligibility
There are three types of eligibility for medical assistance under Medicaid. One is
mandated by federal law, while two are state options. The federally mandated and primary way to
qualify for Medicaid is to be "categorically needy." People who meet income, resource, and asset
standards for any one of several income maintenance programs, i.e., SSI, State Supplemental
Payment (SSP), and Aid to Families with Dependent Children (AFDC), automatically qualify for
Medicaid. Typically, low-income aged, blind, and disabled persons are eligible for Medicaid under
13


SSI and/or SSP, while low-income families with children and deprived of the support of at least
one parent are eligible under AFDC (Harrington et al., 1985).
Each state also has the option of providing Medicaid coverage to a number of other
groups. Two of the major optional coverage groups are "medically needy individuals, (i.e., those
who do not qualify by income, but have medical bills that would reduce their income below a
medically needy maximum), and individuals who would be income eligible for public assistance if
they applied.
Medicaid Benefits
The benefits covered by state Medicaid programs are also categorized as either mandated
or optional. Federal guidelines mandate that certain basic medical services be covered. They
include physician, inpatient and outpatient hospital services, laboratory and X-ray services, skilled-
nursing facilities services for those 21 years of age or older, home health care, family planning,
rural health clinics, and early and periodic screening (Harrington et al., 1985).
Additionally, states have the discretion of providing any one of 32 optional services.
Optional services include chiropractic, dental, eyeglasses, podiatry, and prescription drugs.
Elective long-term care services also include intermediate-care facilities, adult-day care, and case
management.
The Problems Associated with Medicaid LTC Financing
There are at least five major problems associated with the use of Medicaid as a major
LTC funding source. It is imperative that these issues be specifically addressed in any
comprehensive evaluation of Medicaid spending for LTC.
14


As of 1988, a full half of the Medicaid recipients of nursing home services were private
pay upon admission into the facility. The costs of long-term care reduced these people to poverty
level. Most researchers and politicians agree that long-term, care-induced poverty is undesirable.
Second, Medicaid financing of LTC perpetuates inequities in the medical-care delivery
system. A two-class system of chronic health care for the elderly has developed around the
patients ability to pay. Medicaid-dependent persons are severely restricted in nursing home
selection as a direct result of Medicaid reimbursement policies, and are generally considered to be
receiving lower quality of care (Rivlin & Weiner, 1988).
Third, access to even low-quality nursing facilities is difficult for Medicaid-dependent
elders. Lengthy waiting lists for nursing home placement are common, with an average occupancy
rate of 92% in 1985 (Rivlin & Weiner, 1988). The less financing an elder carries, or the greater
the need for services,'the less likely that placement will occur.
Fourth, Medicaid financing of LTC, as it is currently structured, creates a bias toward
institutionalization even when other options are still viable, simply because financing is more
widely available for nursing home care than preferred home or community care.
Fifth, with one-third of the national Medicaid budget spent on LTC for the elderly,
Reutzel (1984) suggests that the Medicaid program has become a supplement to the Medicare
program in providing health services to the aged. This LTC financing mechanism may undermine
the essential public purpose of the Medicaid program, namely, to provide an assurance of access to
mainstream medicine to the nations poor (Holahan & Cohen, 1987).
Finally, since Medicaid budgets are constrained (severely so in some states), the disabled
elderly are competing with poor families for limited Medicaid dollars. The issue of inter-
generational equity becomes a major concern when reviewing the allocation of Medicaid funds.
15


The recent Republican proposal to eliminate the federal deficit in seven years would
reform Medicaid via "block grants" to the states. The program has been dubbed "MediGrant by
the GOP. The proposal would end the entitlement of the poor to health care but would require
states to spend "set aside" funds for health care to serve pregnant women, poor children under the
age of 13, and the disabled. The proposal currently preserves most federal standards for nursing
homes and allows citizens to deduct premiums for private-sector long-term care insurance from
their taxable income. The proposal is projected to save approximately $163 billion from future
growth in Medicaid spending over the next seven years (Zuckerman, 1995).
While states response to "MediGrant" is expected to vary widely should the program be
enacted, the general consensus among LTC analysts is that fewer services would be available for
the disabled who require long-term care.
While MediGrant may not be the answer, it is imperative that Medicaid LTC budgets be
systematically evaluated in terms of efficacy, equity, and efficiency before the coupling of
increased demand and reduced informal caregiving bankrupt an already-strained system.
16


CHAPTER 3
LITERATURE REVIEW
The literature review is presented in two sections, namely, theory and empirical work.
The first section, theory, discusses fundamental theoretical bases for modeling state variations in
long-term care spending. It is conceptual in nature.
The second section, empirical work, presents a review of the methodologies and findings
of research specific to variations in Medicaid and/or long-term care services. It positions the
dissertation relative to the field.
Theory
A comprehensive review of the LTC literature supports demographic, socio-economic, and
political models for predicting utilization and expenditure levels for LTC services. Any serious
attempt to explain state Medicaid spending on LTC for the elderly represents an integration of the
"purist" models presented separately in this section of the literature review for reasons of clarity.
Seven specific models related to public expenditures on LTC will be reviewed. They are: the
medical model, the cultural model, the economic model, the structural model, the political model,
the public-goods model, and finally, the incremental budgeting model.
The Medical Model
The normative goal of health care in the medical model is to cure disease. As such, the
medical model predicts that spending on LTC services for the elderly depends on the level of
"disease" in a given population.
17


Disease, in the acute-care medical model, is defined as any dysfunction of an organ, organ
system, or structure, while health is defined as the absence of disease (Caplan, 1988). Talcott
Parson, in his discussion on the cultural and social dimensions of disease, defined the "sick" role
as one where four conditions are met: namely, that a person is incapable of curing himself; a
person is exempt from usual roles and duties; a person wants to get better; and a person wants to
seek help from experts and receive treatment (Parsons, 1981).
Characteristics of the medical model, as it applies to LTC, include a prerequisite physician
order for initiation of LTC services, and a plan of treatment for LTC services to be supervised by
medical personnel while payment is made directly to providers. LTC services are perceived as a
"health-care benefit" as recipients of such services assume the patient or "sick" role (Zola, 1988).
Clearly, chronic care and disability do not fit well into the medical model of disease,
although it is certainly the model upon which long-term care has been built (Caplan, 1988).
According to Evashwick (1988), the ideal system of care for those who have multiple,
multifaceted, and chronic illness is one that provides comprehensive, integrated care on an ongoing
basis and offers various levels of intensity of care that change as a clients needs change. The goal
is to provide the medical and support services required to enable the person to maximize functional
independence. This contrasts with the goal of acute care, which is to "cure" the patient of illness.
To reiterate, the medical model predicts that Medicaid spending on LTC services for the
elderly is a function of the level of "disease" in the Medicaid population. Researchers have
operationally defined disease in chronic care in one of two ways, namely, by the presence of
pathophysiology (illness) or by limitations in function.
Buchanan (1987) hypothesized that states with high rates of incidence of arteriosclerosis,
heart disease, and cerebrovascular disease would have higher utilization levels of nursing home
care. Buchanan logically assumed that the death rates published for each of these disease
18


conditions were reliable indices of the prevalence of each disease in the living population of the
state. Buchanan concluded that differences among states in the incidence of these diseases
explained little, if any, of the variance in the utilization of skilled nursing home beds by Medicaid
patients.1
Conversely, differences in these disease incidence rates explained upwards of 20% of the
variance in the utilization of intermediate nursing home beds by Medicaid patients.
Many more researchers, e.g., Henry, Morreale, Dunlop, Chiswick and Scanlon, have
focused on the functional limitations of the LTC recipient, using age as a proxy variable for
intensity of disability (Scanlon, 1980). Certainly, illness and functional limitation are related, as
both increase with age, but they are not conceptually identical. Manton and Soldo (1985) estimated
that 2.3% of persons under the age of 19 are limited in performing activities of daily living due to
chronic illness, while 22% of persons 55-64 years of age are limited in ADL to some degree, and
36% of persons 74 years of age and older are limited in ADL while another 3.5% in this age
group are limited in minor or instrumental activities of daily living. The current estimate of
persons requiring continuing care due to functional limitations is more than 8 million, with
approximately two-thirds of these people over the age of 65. This number represents
approximately 3.5% of the general population of the Unites States (Weissert, 1983).
This age/disability factor is reflective of the "need" for LTC services. Scanlon defines the
"need" for LTC services as a value judgment made by someone other than the LTC consumer
1 Skilled nursing home beds are in facilities with an organized, professional staff who
provide medical, continuous nursing, and various other health and social services to patients who
are not in an acute phase of illness, but who require primarily rehabilitation or skilled nursing care
on an inpatient basis. Intermediate care nursing home beds are provided to individuals who require
health related care and services of a more custodial nature (Pierce & Hansen, 1987).
19


about the benefits different individuals will receive from care, or their ability to function without it.
As such, the "needs" of chronically disabled persons have traditionally been assessed and addressed
by the medical community via an acute-care medical model. But a recent trend in medical care for
chronic disease toward the enhancement of health-related quality of life (HQOL), rather than the
cure of disease or increased survival, has emerged (Revicki, 1989).
This acute-care model leads one to posit that the number of elderly in the general
population will positively relate to aggregate LTC spending on the elderly, while the number of
persons 85 or older will positively relate to LTC spending on nursing home services for the
elderly.
The Cultural Model
The social, racial, and ethnic mix of a community may differentially affect Medicaid LTC
spending. The majority of the LTC Medicaid budget is used for nursing home care. Researchers
studying the probability of institutionalization for disabled elderly have identified living alone as an
important factor increasing the use of musing home care. Research further indicates that higher
income, white, native-born elderly are more likely to live alone, whereas non-white females and
foreign-bom males are less likely to live alone across all levels of income and disability (Bishop,
1988; Macken, 1986).
Therefore, white, native-born elderly have been identified as more susceptible to nursing
home placement once disability occurs due to their propensity for living alone, or, on the other
hand because of their more distant family ties.
Conversely, Scanlon (1980) has determined that when one takes into account differences
in the availability of beds between areas with high and low concentrations of black elderly, blacks
were likely to demand more nursing home care than whites. This finding is consistent with the
20


greater incidence of functional impairments and chronic illness among the black elderly. Scanlon
attributed black persons lower utilization of nursing home beds to their concentrations in states
unwilling to fund greater utilization of nursing homes. Accordingly, some of the unsatisfied
demand for nursing home services represents variations in access of elderly persons to nursing
home care because of interstate differences in the ability and willingness of state governments to
finance this care.
Economic, cultural, and political state influences are interwoven in Scanlons (1980)
discussion of nursing home utilization. He contends that cultural norms (e.g., living arrangements
with extended family) influence the demand for formal LTC for the elderly, while economic and
political factors (e.g., tax base and political support for redistribution of funds toward the poor and
elderly) influence the supply of state subsidized LTC services for the elderly.
The cultural model leads one to conclude that: the number of white elderly and single
family households in the elderly population will positively relate to LTC spending, especially
nursing home spending; and that the number of persons per household may inversely relate to
number of LTC elderly recipients of nursing home and/or home and community services as
informal care substitutes for formal (paid) care; and, ultimately, that the number of persons per
household may negatively relate to expenditures on nursing home and/or home and community
LTC services if recipients are served by the informal, as opposed to the formal, caregiving
network.
Scanlons contention that demand is constrained by a states ability and willingness to fund
LTC for the elderly suggest that state economic indicators of ability to pay i.e., average state
income and state employment, and indicators of state political will to fund LTC for the elderly will
each relate positively to LTC spending.
21


The Economic Model
The relatively pure economic model of Medicaid expenditures for LTC, as expressed by
Paul Feldstein (1988a), is a function of demand and supply. Demand for LTC services is related
to the level of disability in the population as well as to economic factors. Once again, the level of
disability in the Medicaid population is considered indicative of the "need" for LTC services. The
economic factors that affect the demand for LTC services include out-of-pocket prices to the
consumer, income of the elderly, out-of-pocket prices for substitute services, and caregiving by the
informal network.
The out-of-pocket price to the consumer influences the demand for LTC services as the
demand for nursing home care has been shown to be price elastic, with price elasticities ranging
from -1 to -2.3 (Rivlin & Weiner, 1988). The price elasticity for home care is also considered to
be highly elastic, although no specific estimate could be found. Medicaid has dramatically lowered
the out-of-pocket price of nursing home care, while leaving the price of alternatives unaffected.
The income of the elderly strongly affects the demand for paid LTC services since LTC is
considered a "normal economic good in that consumption has been demonstrated to increase with
disposable income (Feldstein, 1988a). The indigent disabled elderly, with significantly less
disposable income, are more vulnerable to institutionalization as institutional care is subsidized to a
much greater extent than alternative care.
The supply of informal LTC services (unpaid and typically rendered by family) is
considered a substitute for formal (paid) LTC services. The higher the opportunity cost of
informal care to the provider of such services, the greater will be the demand for paid and
government subsidized services (Chiswick in Feldstein, 1988a).
22


Utilization of formal LTC services is influenced by both demand and supply. The supply
of "paid" LTC services depends upon the price received for services, the cost of inputs, and
government regulations.
In a competitive market, one would expect the supply of paid LTC services to increase as
the price received for such services increases, all other factors held constant. The predominance of
government as "fixed" payor for institutional LTC services immediately links the effects of supply
received for nursing home services to government regulation on the price of LTC services.
Feldstein (1988a) discusses the effects of the governments monopsonistic power over rate setting
and its effects on private pay and Medicaid residents. Feldstein states that aggregate nursing home
capacity (supply of beds) and rates for Medicaid patients are heavily regulated by the government.
Nursing homes charge a higher price to private-pay patients as compared to both Medicaid patients
and private-pay patients in an unregulated nursing home environment because the demand for both
private pay and government-subsidized nursing home beds continues to outstrip supply.
The supply of nursing home beds is tightly regulated by state governments, especially in
states with active certificate-of-need (CON) legislation. This surcharge on private-pay residents
causes a redistribution of income from private-pay persons toward the nursing home, resulting in
economic profits accruing to the nursing home investors. This resulting economic profit, which
accrues because of government-constrained supply, allows inefficient nursing homes to survive.
This differential in price between private pay and Medicaid patients also causes access to nursing
home care to be reduced for Medicaid patients, especially heavy-care Medicaid patients. The
"unmet need" of heavy-care Medicaid patients, arguably the most "needy," is enlarged while the
informal caregiving network is further burdened.
Feder and Scanlon (1981) contend that Medicaid programs attempt to control spending for
nursing home care by limiting the supply of beds. Higher per diem rates were consistently linked
23


to lower utilization of nursing home services by Medicaid-eligible persons. This is an unlikely
outcome if providers were the true obstacles to access to Medicaid LTC beds. One would expect
unregulated (i.e.,"rational") economic providers to supply more beds as the payment increased in
an attempt to maximize profits. Harrington and Swan (1987) found nursing home bed supply to be
the dominant factor affecting the number of Medicaid recipients and Medicaid expenditures per
total aged population. They contend that lower nursing home bed supply will lower utilization of
Medicaid nursing home services and overall Medicaid costs. They also point out that lowering
nursing home bed supply without providing alternative LTC services may only increase problems
of access and quality of care.
The supply of LTC services may vary geographically with variations in the cost of inputs,
e.g., land, labor, local inflation, and supplies. Individual state government regulations regarding
standards of care, personnel requirements, and safety requirements differentially affect the cost of
inputs in the LTC industry (Paringer, 1983).
It is important to note that the utilization of LTC services, both private pay and
government subsidized, depends upon the decisions of consumers, providers, and government. In
reference to the LTC market, Feldstein (1988a) considers that state government regulation is
represented by the number of nursing home beds, the standards of care, and the Medicaid
reimbursement policies.
In summary, the economic model concludes that: median income of the elderly will
positively relate to private expenditures on LTC; the supply of nursing home beds will positively
relate to aggregate LTC spending and to nursing home spending; and the cost of inputs (labor, cost
of capital, etc.) will positively relate to private and public expenditures on LTC across all services
areas.
24


The Structural Model
The structural model states that Medicaid spending on LTC for the elderly is a direct
result of the structure of the Medicaid program. Once again, the models display a degree of
overlap and integration. The economic model, which was discussed immediately above, contains
elements of both the medical and structural models for explaining variation in LTC spending for
the elderly.
The structures of state Medicaid programs differ regarding eligibility criteria, benefits,
reimbursement rates and methods, and methods of limiting utilization. The various state
reimbursement policies are rather complex with many dimensions. State methods differ in the cost
components allowed, ceilings on cost components, and in consideration of such factors as
respective property costs, inflation rates, profit rates, and state licensing standards (Harrington &
Swan, 1984). States with reimbursement systems other than purely retrospective have been
correlated with lower nursing home rates, and lower expenditures per person, although causation
has not been determined (Harrington & Swan, 1984).
Scanlon (1980) states that there is a three-part objective in designing a Medicaid
reimbursement system for nursing home services. Reimbursement should secure
access to the service for eligible persons, encourage high quality, and minimize costs. Tradeoffs
among the three are required since a system that promotes one of these goals is likely to reduce the
likelihood of achieving the others, especially affecting the third.
States may also modify service-delivery patterns and costs through incentives and
disincentives involving such factors as occupancy rates, quality of care, and case mix (Harrington
& Swan, 1984). McDevitt and Buczko (1985) determined that states with the optional medically
needy eligibility program directly influenced Medicaid spending on LTC by increasing the number
of eligible users.
25


Harrington and Swan (1987) found that state policy had major effects on Medicaid nursing
home expenditures and utilization. They determined that the reimbursement rate to nursing homes
providing custodial care (called intermediate-care facilities or ICFs) was the strongest predictor of
expenditures per recipient for all Medicaid recipients.
It is interesting to note that researchers consider the effects of state Medicaid program
controls on the utilization of health services and expenditures to be complex. State controls of the
Medicaid program may affect eligible groups differentially, and the net effect of the program
control may be the opposite of that intended (Mauskopf, Rodgers, & Dobson, 1985). Reutzel
states that manipulation of some Medicaid program decision variables have counter-intuitive results
because of complex interactions among individual behavioral and managerial variables. Physician
reimbursement rates are a good example. Policy makers may decrease physician Medicaid-
reimbursement rates in an attempt to reduce Medicaid expenditures. The unintended consequence
may be that physicians with excess time or decreased income increase services in an attempt to
reach a personal-target income while Medicaid recipients seek out more expensive alternatives
(hospital emergency room rather than physician office visits), thus increasing overall state Medicaid
expenditures. Decreasing the number of optional Medicaid services may also lead to counter-
intuitive results. While it is expected that reducing optional services will decrease Medicaid
expenditures, recipients may substitute more expensive covered services for now lost and less
expensive optional services, thus driving up overall Medicaid expenditures. Soumerai, Ross-
Degnan, Avom, McLaughlin and Choodnovskiyet (1991) found that limiting Medicaid
reimbursement for effective drugs put frail, low-income elderly patients at increased risk for more
costly institutionalization in nursing homes, thus increasing the overall Medicaid expenditures.
The structure imposed by the federal government on state Medicaid programs may be
another important structural determinant of state LTC spending. Federal policies regarding
26


eligibility and coverage are clearly important determinants of state Medicaid spending. Home and
community service waivers (the 2176 Medicaid program) must be approved by the HCFA, a
federal body, in accordance with federal guidelines. The level of federal financial participation
(i.e., the matching rate) state Medicaid programs is generally held to have an effect on state
generosity. The U.S. General Accounting Office Report entitled "Medicaid: Interstate Variations
in Benefits and Expenditures" (1987) found the matching rate to be an important determinant of
Medicaid expenditures. The report further concluded that increasing the matching rate was not
sufficient incentive to encourage very restrictive states to broaden their Medicaid programs. This
may be due to the fact that Federal Medicaid Matching Rates have proven to be insufficiently
weighted toward low-income states to overcome the effects of interstate differences in fiscal
capacity and poverty rates (Grannemann, 1979). Howell et al. (1988) state that the federal
matching rate may have less impact on state Medicaid programs than state political philosophies
regarding health care for the poor.
The structural model suggests several relationships. They are: that retrospective
reimbursement positively correlates to average spending per nursing home recipient and to total
nursing home spending; that Medicaid generosity regarding eligibility criteria will be positively
related to aggregate LTC spending and number of recipients across all services; that Medicaid
generosity regarding optional benefits may behave in a counter-intuitive manner and relate
negatively to nursing home and aggregate LTC spending; that nursing home reimbursement rates
positively relate to average expenditure per recipient and aggregate spending across all service
areas; and, finally, that the Federal Medicaid Matching Rate may positively relate to spending
across all LTC service areas.
27


The Political Model
The political model states that a states Medicaid expenditure on LTC for the elderly is a
function of that states political climate.
The political climate of a state can be defined by that states position on the
conservative/liberal spectrum (Buchanan, 1987). Conservative and liberal indices were developed
by Buchanan for each state using annual ratings of each states delegation to the U.S. House of
Representatives by various political organizations. The liberal index was the statewide average of
the ratings given by the Americans for Democratic Action (ADA) and the AFL-CIOs Committee
on Political Action (COPE). The conservative index was an average of the ratings given by the
Americans for Constitutional Action (ACA), the National Association of Businessmen (NAB), and
the Chamber of Commerce (COC). The final state score was an average of the liberal and
conservative indices. Barrilleaux and Miller (1988) simplified this methodology by using the
statewide average of the ratings given only by the ADA as indicative of the liberal index of the
states political ideology.
Buchanan found clear and consistent patterns when the political climate of a state was
correlated with the Medicaid per diem payment rate for nursing homes and the state-utilization rate
for Medicaid-certified nursing home beds. A more conservative political climate was linked to
lower Medicaid per diem rates, while a more liberal political climate was linked to higher
Medicaid per diem rates. This correlation may be spurious in that more liberal states
(Massachusetts and Connecticut ranked first and second in 1985) may have a higher cost of living
necessitating a higher Medicaid reimbursement rate. Buchanan did find that differences in the cost
of living among states statistically explained upwards of 60% of the variance in Medicaid payments
for skilled and intermediate nursing home care (an economic index). The political indices
explained upwards of 20% of the variance in the Medicaid-certified nursing home beds per 1,000
28


elderly ratio for skilled care. Barrilleaux and Miller (1988) developed a market-based model for
explaining variations in states Medicaid policy decisions. They consider measures of supply,
realized demand, and policy to be endogenous factors in the model, while measures of states
social (demographic), economic, and political characteristics are exogenous variables within the
system. The authors claim that socio-economic and political characteristics of states are
complementary in their influence on policy outcome. It is worth noting that the authors argue that
state political characteristics reflect economic development and that economic development is a
catalyst for political and policy development. Carrell (1990) also found political liberalism to
affect positively state expenditure per recipient of LTC and total Medicaid expenditure on LTC.
The political model leads one to observe that a more liberal state political climate will
relate positively to per diem nursing home rates, average expenditures per recipient across all
services areas, number of LTC recipients across all service areas, and total spending across all
service areas.
The Public Goods Model
The median voter theory of political decision-making states that politicians court the
median voter, and once elected, allocate resources to meet the needs of the median voter. Godwin
and Shepard (1976) argue that the political process translates median-voter desires into public
policy, frequently financing services as public goods. Orr (1976) applied this theory to the AFDC
program and Grannemann (1979) extended it to Medicaid, viewing
...voter characteristics, preferences, and environment as the ultimate determinants
of public expenditures; while the political system established in a state is only a
[voter selected] means of attaining desired policies.
29


The public goods model of interstate variation in Medicaid programs, as specified by
Grannemann, explains the number of Medicaid recipients and average benefits per recipient. States
that contain relatively few poor persons, e.g., Minnesota and Wisconsin, offered high benefit
packages; states with many poor persons and a limited tax base, e.g., Mississippi, offered low
benefit packages; while states with many poor persons and a good tax base, e.g., New York and
California, offered generous benefits with less-restrictive eligibility requirements. Demographic
and economic conditions were reflected in the median voters preferences that translated into public
expenditure decisions.
If one assumes that politicians are rational actors and are acting to maximize their political
support and chances for re-election in accord with the median voter theory, and one assumes
democratic participation via voting on the part of those with strong attitudes toward public spending
on LTC for the elderly (i.e., the elderly who represent a large and growing segment of the
population and potential voting bloc) one would expect congruence between the attitudes and values
of the community and U.S. House of Representatives voting records (Feldstein, 1988b).
It is empirically unclear to what extent the values of the community have actually been
incorporated into Medicaid program characteristics and into Medicaid spending.
According to Emmett Keeler and Robert Kane:
Our treatment of our elderly is a mirror of our society. Values lie at the very
heart of long-term care. Our willingness to transfer income from one generation
to another, our valuing of prolongation of dependent life often filled with pain and
loneliness, our preferences for encouraging maximal independence at the risk of
untoward consequencesall these decisions and others like them may tell us more
about ourselves than about the elderly. It is a lesson we may not be eager to
learn, but a responsibility we cannot shirk (Keeler & Kane, 1982).
To the extent that the public goods model holds true, median voter attitudes, their
concomitant values, and aggregate preferences are reflected in LTC expenditure measures within
30


the constraints of economic realities, inclusive of federal subsidies. This is consistent with the
findings of Orr (1976), Grannemann (1979), and Buchanan, Cappelleri, and Ohsfeldt (1991) who
found political variables to be statistically insignificant regarding Medicaid expenditures once
underlying state economic and demographic factors relevant to the voting population were
controlled for in the analysis.
The public goods model leads one to suspect that economic conditions, demographic
characteristics of the voting population, and the effective price of subsidized LTC services to
taxpayers (net of federal matching dollars) are the ultimate determinants of LTC spending, number
of recipients, and average expenditures per recipient across all service areas. Other factors, i.e.,
political philosophies, may relate positively to LTC expenditures, but are statistically insignificant
when economic, demographic, and federal policy constraints are controlled for in the analysis.
The Incremental Budgeting Model
The incremental budgeting model suggests that current year Medicaid expenditure levels is
dependent upon previous years Medicaid expenditure level (Wildavsky, 1988).
Schneider (1988) reported that in 1987 the Office of the Budget, Ohio Department of
Human Services, surveyed state Medicaid officials across the nation to identify how Medicaid
policy decisions were made. She states that of those who responded to the survey, 90 % indicated
that previous years expenditure level played an extremely important role in program decision
making."
Both Schneider (1988) and Buchanan (1991) incorporated previous years Medicaid
expenditure level into their model for explaining current years Medicaid expenditure level. Each
found previous years Medicaid spending to be statistically significant in explaining current years
Medicaid spending. This finding is consistent with theoretical work on incremental policy making
31


as presented by Charles Lindblom, in "The Science Of Muddling Through" (1959), where he
states:
It is a matter of common observation that in Western democracies public
administrators and policy analysts in general do largely limit their analyses to
incremental or marginal differences in policy that are chosen to differ only
incrementally. They do not do so, however, solely because they desperately need
some way to simplify the problem; they also do so in order to be relevant.
Democracies change their policies almost entirely through incremental adjustment.
Policy does not move in leaps and bounds, (page 81)
Aaron Wildavsky (1988) argues that policy makers lack control over Medicaid
expenditures because of the open-ended, entitlement nature of the Medicaid program. Thus, it
follows that incrementalism per se as a hypothesis about budgetary behavior can be neither
accepted nor rejected by examining total Medicaid expenditures (Lurie and Wyckoff, 1989).
The incremental budgeting model posits that previous years Medicaid expenditure level is
strongly associated with current years Medicaid expenditure level. Interpretation of this
correlation is fraught with problems as it is difficult to isolate the influence of previous years
Medicaid expenditure that is directly attributable to incremental behavior. Therefore, previous
years Medicaid expenditure level is not incorporated into the present analysis of models.
Empirical Work
This section of the literature review sets the context of the study within the literature on
determinants of geographic variation in spending and/or utilization of medical services. It
encompasses three specific areas of empirical work, namely,
1. Geographic variation in overall Medicaid expenditure or utilization patterns,
2. Geographic variation in Medicaid and non-Medicaid nursing home expenditure and
utilization, and
32


3. Geographic variation in Medicare home health care utilization.
It is worth noting that no empirical evidence exists on the determinants of geographic
variation in Medicaid home health-care spending, or the Medicaid 2176 Waiver Program funding
of LTC for the elderly. Geographic variation has been well documented, but the variance has not
been carefully explained (Pierce & Hansen, 1987; GAO, 1987; Miller, 1990).
It is also noteworthy that no estimates, let alone explanatory models, of private spending
on nursing home care at the state level currently exist in the literature. Data limitations have
previously not allowed for estimates of state-level private spending on nursing home care.
Determinants of Geographic Variation in Medicaid Expenditures and Utilization Patterns
Stuart and Bair (1971) represent one of the earliest attempts to explain interstate variations
in Medicaid spending. The researchers used FY 1968 state Medicaid expenditure data to test the
hypothesis that states with greater income inequality would be slower to adopt new Medicaid
programs and would have smaller benefit packages. Stuart and Bair suggested that wealthy persons
in political control would affect this result.
The researchers regressed Medicaid benefits (expressed as Medicaid expenditures divided
by the state population) and the number of months each state had particular optional Medicaid
programs against the measure of state income inequality (a function of the state income
distribution). The analysis demonstrated a significant and negative relationship between income
inequality and Medicaid benefit level and speed of adoption of optional Medicaid programs.
This study has been criticized as simplistic and, ultimately, misleading. Critical
explanatory factors, i.e., demographic, economic, and supply variables, were omitted from the
model, casting doubt on the true predictive powers of the independent variables.
33


In a follow-up effort, Stuart and Bair (1972) developed a model relating state Medicaid
benefits (again expressed as expenditure divided by the state population) to demographic variables
(percent state population 65 or older, and percent state population white), economic variables (per
capita income), and local health-care system variables (nurses, hospital beds, and physicians per
1,000 state population). The researchers found Medicaid benefits significantly and positively
related to the number of MDs per 1,000 population.
The methodology employed by Stuart and Bair in this second study represented a major
step forward. The utilization of multi-factorial variables for explaining geographic variation in
medical care expenditures and utilization of services has since become the standard within the field.
Both studies by Stuart and Bair have been criticized for equating Medicaid expenditures
with benefits in their operational definitions. Reutzel (1984) argues that Medicaid program
decisions regarding reimbursement systems and rates may allow some states to receive greater real
benefits at similar expenditure levels, i.e., as "prudent" buyers. Differences in cross-state price
levels would also argue against the equivalence of expenditure level with benefit (Buchanan et al.,
1991).
Grannemann (1979) utilized a median voter model for explaining interstate variation in
Medicaid spending. He utilized a mixed cross-section and time-series sample of state Medicaid
expenditure data from FY 1973 to 1977. The dependent variables were total Medicaid
expenditure, Medicaid expenditures for various Medicaid eligible groups (AFDC adults, AFDC
children, and SSI recipients), Medicaid expenditures for inpatient hospital services and physician
services, number of Medicaid enrollees per general population, number of optional Medicaid
services offered, and the share of state taxpayer income devoted to Medicaid.
Grannemanns explanatory variables included economic variables (an indicator of state
medical care costs, number of taxpayers, effective price of Medicaid to taxpayers, adjusted gross
34


income of average-taxpayer federal-income tax deflated by CPI, and an index of income inequality
within the state); demographic variables of the poor (percent black who are poor, percent poor
over 65, percent poor under 16, and percent state population below the federal poverty level);
demographics of the general population (percent population in metropolitan areas, percent
population completed high school, and region of the state within the United States); and medical
care market characteristics (physicians per capita and hospital beds per capita).
Utilizing ordinary least square and reduced form regression analysis, Grannemann found
Medicaid benefits significantly and negatively related to the effective price of Medicaid to the state
taxpayer, and positively related to the needs of potential recipients (aged and poor), the income of
the taxpayer, and the supply of health-care services.
Grannemann and subsequently the GAO (1987), argue strongly that the Federal Medicaid
Matching Rate, which directly affects the effective price of the Medicaid program to state
taxpayers, is not sufficiently weighted toward poorer states and regions of the country. Poorer
states economic disadvantage is so great that even with the highest federal matching rate their
Medicaid programs remain, essentially, underfunded. Inadequate federal contribution to poorer
states Medicaid program results in a relative redistribution of monies to wealthier states (as state
taxpayers contribute to federal revenues) and in more restricted Medicaid eligibility and benefit
packages in poorer states. Grannemanns work is most important to the field because of its policy
implications regarding the federal-Medicaid matching rate.
Grannemanns work, however, has been criticized for omitting structural characteristics of
state Medicaid programs and political influences as possible explanatory factors. Grannemann
considers both variables endogenous explanatory factors, in that they are driven by the fundamental
explanatory variables of demography, economy, and federal financial participation rates. Reutzel,
once again, refutes the equivalent use of expenditure data with benefit level. The GAO contends
35


that Medicaid expenditure level can be used as a surrogate measure of benefits. GAO analysts
regressed state-level Medicaid expenditure data (expressed as Medicaid spending per person below
the federal poverty level) against two independent variables. These independent variables were
Medicaid eligibility (expressed as the number of Medicaid enrollees as a percent of persons under
the federal poverty level) and the scope of Medicaid services provided (expressed as a percent of
the possible optional Medicaid services provided by the state). Taken together, the eligibility and
coverage variables accounted for 69% of the state variance in Medicaid expenditure levels.
Reutzels doctoral dissertation (1984) studied the impact of underlying state actuarial
characteristics (demographic, income, and health-care system variables) and state Medicaid-policy
decisions on interstate variations in Medicaid expenditure per elderly recipient of SSI (Medicaid-
aged eligible). His thesis utilized five dependent variables, the primary dependent variable of
Medicaid expenditure per SSI recipient, plus four measures of services utilization. These services
areas consisted of inpatient hospital days, nursing home days, outpatient MD visits, and units of
miscellaneous services, all divided by the number of elderly SSI recipients per state.
Reutzels independent variables consisted of actuarial factors (expressed as Medicare
dollars spent per Medicare enrollee in 1979 as a reflection of all pertinent actuarial factors,
excluding those relevant to nursing home utilization); economic factors (state cost of living in
1979); demographic factors (percent of population residing in metropolitan areas); health-care
system variables (the number of nursing home beds per 1,000 elderly); and Medicaid program
characteristics (number of optional Medicaid services provided, limits present on the number of
visits or dollars spent on ambulatory MD visits, presence of limits on inpatients hospital days,
Medicare/Medicaid fee ratios for physician specialists, and presence of mandatory rate-setting
hospitals).
36


Utilizing regression analysis, Reutzel investigated state Medicaid expenditure and
utilization data for 1979. Reutzels analysis produced a R2 of .46 when he regressed actuarial
variables alone against total Medicaid expenditure level per SSI recipient, and a R2 of .61 when he
regressed the actuarial variables together with state Medicaid-program characteristic variables.
Reutzel concluded that state Medicaid-expenditure levels are influenced by state Medicaid-program
characteristics. Critics argue that these program characteristics may reflect fundamental socio-
economic, demographic, or political variables that have not been represented in the model.
Reutzel contends that states choose between liberal-Medicaid benefit packages for
recipients and liberal-reimbursement levels to providers. He suggests that political influences may
be important determinants of state Medicaid policy. Once again, the nature of political influences
on state Medicaid expenditure levels surfaces.
McDevitt and Buczko (1985) utilized 1980 state data to examine the relationship between
Medicaid program characteristics and program expenditure level. Two dependent variables were
calculated, namely, Medicaid recipients per 1,000 poor (as a measure of state generosity regarding
Medicaid eligibility) and Medicaid expenditure per recipient (as a measure of state generosity
regarding Medicaid services). Taken together with the number of poor people in each state, these
variables allow for estimation of states total Medicaid expenditure. The dependent variables
(number of Medicaid recipients, expenditure per recipient, and total expenditure) were estimated
for total Medicaid population, per SSI recipient, per AFDC child, and per AFDC adult. The
researchers regressed these dependent variables against both Medicaid program characteristics and
state demographic and health-care system variables.
McDevitt and Buczkos findings indicate that states exercise considerable control over the
number of Medicaid recipients eligible via AFDC standards and the optional medically needy
37


program, but exert relatively little control over the number of Medicaid recipients eligible via SSI
eligibility.
They also found that state Medicaid-program characteristics exerted the greatest control
over physician expenditures and the least control over long-term care expenditures. The presence
of the optional medically needy eligibility category was the only Medicaid program characteristic
that demonstrated a significant effect on both the number of state LTC recipients and the level of
LTC Medicaid spending (significant at the .01 level).
McDevitt and Buczko (1985) considered aspects of the relationship between Medicaid-
policy variables and either program participation or expenditure to be dynamic in nature, and
thereby best represented by a time-series model. The results of their study demonstrate the need
to: (1) explicate the determinants of Medicaid expenditure/ utilization specific to LTC, and, (2)
use actual Medicaid expenditure and recipient data, as opposed to surrogate measures.
Mauskopf et al. (1985) examined the effects of specific state Medicaid-program controls
on health care utilization patterns of individual Medicaid recipients. The researchers utilized linear
regression analysis to evaluate 1980 household-level utilization data from the National Medical
Care Utilization and Expenditure Survey. They tested the hypothesis that states that offered fewer
optional Medicaid services to their enrollees and imposed more limits on services would decrease
the overall utilization of health-care services by the Medicaid population.
The dependent variables used in the regression equation were the probability of using
specific health-care services in 1980, and the level (intensity) of use in 1980 for specific health
services. These utilization measures were analyzed for specific Medicaid eligible groups, i.e.,
AFDC adults, AFDC children, and SSI recipients (aged, blind, and disabled SSI recipients were
not delineated). The independent factors included price variables, income variables, health status
variables, socio-demographic variables, supply variables, and a regional background variable.
38


These variables were included in the equation to adjust for variations due to non-Medicaid
programmatic factors, i.e., economic and demographic variables. The Medicaid program
characteristics in the equation were classified as outpatient utilization controls, inpatient utilization
controls, dental service controls, reimbursement controls, and reimbursement method.
The findings of Mauskopf et al. indicate that utilization of health-care services is
responsive to some Medicaid program controls, but that these controls affect eligible groups
differently. Less generous reimbursement of physicians was seen to decrease total-Medicaid
physician expenditures. This variable (reduced reimbursement to MDs) decreased physician visits
for the SSI population, had no affect on AFDC adults, and increased visits for AFDC children.
Alternative-Medicaid reimbursement systems (away from a customary cost method) was seen to
reduce hospital LOS for AFDC children, reduced the probability of a hospitalization for AFDC
adults, but had no affect on hospitalizations or LOS on the SSI population. For the most part,
controls on the utilization of health-care services had a greater impact on AFDC children with
chronic conditions (arguably the most in need) than on AFDC children without chronic conditions.
Among AFDC adults, greater impact of controls was observed for those without chronic
conditions.
The researchers state that the relationships among Medicaid program controls are surely
complex. The net result of program controls are frequently not the anticipated result. Medicaid-
eligible groups are affected differently by program controls, indicating a heterogenous patient
population.
Barrilleaux and Miller (1988) offer empirical support for the independent influence of state
political factors on Medicaid expenditure levels. They regressed the dependent variable of
"Medicaid effort" (expressed as the proportion of total state personal income devoted to Medicaid
spending) against seven independent variables. These independent variables were supply (number
39


of MDs), demand (number of Medicaid recipients), price (average expense per inpatient day), state
political ideology (Americans for Democratic Action rating of that states congressional delegation
voting record), interest group influence (index of interest group density), bureaucratic density
(spending on Medicaid administration) and urban population (percent of population residing in
urban areas).
Models were estimated cross-sectionally using 1982 state data. Regression results indicate
that Medicaid spending is most sensitive to supply variables, although the political system variables
(political ideology, bureaucratic and interest group density) each were significant at the .05 level.
Medicaid spending effort was found to increase .60% with each percent increase in Medicaid
administrative funding, .17% with each percent increase in political liberalism, and .35% with each
percent increase in special-interest group density.
Unfortunately, scant indicators of alternative demographic, economic, and supply variables
are included in this model. To the extent that political measures simply reflect any of these
omitted factors, the results may be spurious. This dissertation attempts to incorporate more
thoroughly those variables.
Schneider (1988, 1989) employed a mixed cross-sectional, time-series analysis to examine
the role of federal, state, and local government on state Medicaid-expenditure levels. The
dependent variable was simply the total nominal dollar amount of Medicaid spending for each state
for years 1975 to 1985.
Schneider divided the independent variables into two sets, one set corresponding to
national-level influences on Medicaid spending, and another set corresponding to state-level
influences on Medicaid spending. The two national-level variables were the percentage of the U.S.
population identified by the federal government as "in poverty" in a given year, and, second, the
medical CPI for a given year. The three state-level independent variables were the number of
40


Medicaid recipients in a given year, the level of government within the state that was responsible
for administration of the Medicaid program (state versus local), and finally, the previous years
state Medicaid-expenditure level.
Schneider estimated the model for two time periods, namely, 1975-1980 (the "pre-Reagan
period") and 1981-1985 ("the Reagan period"). The author later recanted, stating the actual time
periods utilized were 1975-1981 versus 1982-1985. The regression analysis for the 1975-1981 time
period generated an adjusted R2 of .74, with local Medicaid administration and previous years
Medicaid-expenditure level having a statistically significant and positive impact on current-year
Medicaid-expenditure level. The regression analysis for the 1982-1985 time period generated an
adjusted R2 of .68, but presented with a marked contrast to the earlier time period. The impact of
the previous years spending dropped to zero, while the impact of the number of Medicaid
recipients and local administration increased dramatically.
Schneider drew three conclusions from her research. First, it is necessary to take
administrative variables into account when explaining Medicaid policy as local administration of the
Medicaid program was associated with higher Medicaid-expenditure levels. Second, national,
state, and local-level factors all have important influences on Medicaid program development.
Third, the time dimension must be included in empirical analysis of Medicaid program
development as the policy output process is dynamic in nature.
Lurie and Wyckoff (1989) have criticized Schneiders work on three scores: one, her
coefficient estimates depended on adjustments made to HCFA data that could not be replicated;
second, that the model omitted variables which affected the interpretation of included variables; and
third, she drew conclusions from coefficients that were statistically not significant.
41


Despite these serious criticisms, Schneiders work does question the influence of local
administration of the Medicaid program and incremental budgeting on state Medicaid-expenditure
levels.
Buchanan et al. (1991) also found previous years Medicaid spending and local
administration of the Medicaid program to affect significantly current Medicaid-spending levels.
The researchers classified their independent variables into three categorieseconomic factors,
political factors, and implementation factors. The economic factors measured were personal
income per capita and number of Medicaid recipients. The political factors included in the model
were an index of general liberal ideology of the states U.S. Representatives, an index of the
states inter-party competition, and the ratio of patient-care physicians per 1,000 population as a
proxy measure for the political strength of the medical industry. The implementation factors were
the Federal Medicaid Matching Rate, local versus state administration of the Medicaid program,
and the previous years Medicaid expenditure level.
The researchers used a generalized two-stage least squares regression analysis to analyze
cross-sectional data for years 1977-1987. Mixed cross-sectional, time-series analysis explained
63 % of the state variation in Medicaid-expenditure levels. Relationships between the current level
of Medicaid spending and the log of the previous years Medicaid spending, the log of the number
of Medicaid recipients, personal income per capita, number of patient-care physicians, and local
Medicaid administration were significant at the p=.05 level.
Determinants of Geographic Variation in Expenditure and/
or Utilization Patterns Specific to Nursing Home Care
This section does not pretend to do total justice to the body of literature concerning the
determinants of nursing home utilization at the individual or family unit level. Many researchers,
e.g., Wan and Weissert (1981), Branch and Jette (1982), Capitman (1985), and Grannemann and
42


Grossman (1986) have employed multiple regression techniques to explain variations in nursing
home utilization and predict institutionalization at the individual level (Grannemann, 1986). These
studies unfortunately all report a relatively low R2 statistic. Among the factors found to have been
associated with nursing home usage are lack of informal support at home, adequate financial
resources or Medicaid coverage to pay for nursing home care, the recent use of a hospital or
nursing home service, favorable attitudes toward institutionalization, and greater impairment in
ADL and IADL as compared to elders living in the community. To the extent possible, the present
dissertation incorporates many of these factors at the aggregate state level, i.e., average persons
per household as a proxy for informal support.
Scanlon (1980) reviewed five multivariate studies of nursing home utilization. Each
researcher, i.e., Henry (1970), Morreale (1975), Dunlop (1976), Chiswick (1976), and Scanlon
(1978), employed regression analysis to explain nursing home utilization (expressed as the
percentage of the elderly population in nursing homes at the state or SMS A level). The
independent variables included Medicaid generosity (in eligibility and/or benefits), the age structure
of the elderly population, the availability of family support, urban versus rural residence of the
elderly population, financial resources of the elderly, out-of-pocket price of nursing home care, and
the availability of alternative sources of care. Analyses produced R2 estimates ranging from .32 to
.81 with out-of-pocket price, financial resources of the elderly, and age structure of the elderly
population most often statistically significant.
The focus of the literature concerning variation in nursing home utilization in the late
1970s and early 1980s was on the demographics and economics of the elderly in conjunction with
price variables. Scanlon also focused on the importance of the supply variable as availability of
Medicaid-certified beds is constrained in many states, impacting directly on nursing home
43


utilization and expenditure. Most empirical work on nursing home utilization and expenditure
throughout the 1980s incorporated health-care system supply variables in the equation.
Ray et al. (1987) studied interstate differences in the characteristics of Medicaid nursing
home residents and in their utilization of medical-care services. The researchers used Medicaid
claims and enrollment data for calendar year 1981 for three statesMichigan, California, and New
York. State resident characteristics included the proportion of Medicaid-covered nursing home
residents relative to the overall state nursing home population, demographic characteristics of the
Medicaid nursing home population, and the diagnostic case mix of each states Medicaid nursing
home population.
The second dependent variable (medical care utilization) encompasses measures of
turnover among nursing home residents, expenditures for nursing home care, and inpatient-hospital
utilization patterns of the nursing home elderly.
Medicaid-resident characteristics for each state were very similar, approximately three-
fourths of the Medicaid nursing home residents were women, over 45 % were over the age of 85,
and over 95 % were SSI-aged Medicaid eligible. There was little interstate variation in these rates.
Diagnostic case mix of Medicaid nursing home residents across the three states was also consistent
reflecting the well-known causes of morbidity among the elderly.
Medical-care utilization patterns of Medicaid nursing home residents for 1981 showed
marked interstate differences. The turnover among nursing home residents (expressed as the
proportion of elderly Medicaid nursing home residents who entered the nursing home in 1981) was
17% for New York, 25% for Michigan, and 32% for California.
Among the 52,306 Medicaid recipients who entered nursing homes in the three states in
1981, 43% had a Medicaid-covered hospitalization within 30 days of entry in California, in
contrast to 24% in New York, and 20% for Michigan. There were also pronounced differences in
44


the average pre-nursing home hospital LOS. The average pre-nursing home hospitalization LOS
was 60 days in New York, 18 days in Michigan, and 12 days in California. This variability was
not explained by any differences in residents diagnosis among states.
Ray and colleagues employed descriptive analysis only; multivariate statistical analysis has
not been applied to their findings. This limitation of their study prohibits generalization of the
results. The study also has serious sample-size limitations. The research does, nevertheless,
demonstrate important relationships between LTC utilization patterns and health-care system
variables such as numbers of hospital beds, average length of inpatient stays, and availability of
nursing home beds. These relationships hold serious implications for both future Medicare and
Medicaid levels of expenditures.
Harrington and Swan (1987) utilized a cross-sectional time-series regression analysis to
investigate the impact of state Medicaid nursing home policies on nursing home utilization and
expenditure. The data set for this study was 42 states over a six-year period, 1978-1983.
Their three dependent variables were number of Medicaid nursing home recipients per
1,000 aged (utilization), Medicaid nursing home expenditure per recipient, and total Medicaid
nursing home expenditure per aged state population.
Three sets of independent variables were used in the model: (1) state Medicaid nursing
home policies on utilization, eligibility, and reimbursement; (2) provider supply (nursing home
beds, hospital beds, hospital-occupancy rate, and number of physicians); and (3) macrocontextual
variables (percent state unemployment, percent state population over 65, percent state population
residing in metropolitan areas, and pay per nursing employee). Harrington and Swan (1987)
discovered the supply of nursing home beds to be the strongest predictor of Medicaid nursing home
utilization and of overall Medicaid nursing home expenditures. The presence of a medically needy
45


eligibility program was the only significant Medicaid policy predictor of nursing home utilization.
Reimbursement rates had little to no affect on access of Medicaid recipients to nursing home beds.
The intermediate-care facility (ICF) reimbursement rate was the strongest predictor of
nursing home expenditure per recipient. Average nurses pay was the strongest macrocontextual
predictor of nursing home expenditure per recipient. Another statistically significant predictor of
expenditure per nursing home recipient was the percent of the population living in metropolitan
areas.
Harrington and Swan (1987) found that hospital-occupancy rate had a strong positive effect
on expenditure per nursing home recipient and oh overall Medicaid nursing home expenditures.
The researchers hypothesized that hospitals in states with higher occupancy rates release patients
earlier leading to longer nursing home stays and higher Medicaid expenditures. Both the presence
of a medically needy program and ICF per diem rates were significant predictors of Medicaid
nursing home expenditures per aged population. Eligibility affects Medicaid expenditure by
increasing number of recipients (utilization) and reimbursement rates affect expenditure per
recipient and, therefore, overall Medicaid spending.
The present dissertation enhances Harrington and Swans research design by using actual
Medicaid expenditure and recipient data specific to both categories of spending (nursing home,
home care, and 2176 spending) and target population, i.e., the elderly. This thesis also
incorporates characteristics of the general state population, expands the Medicaid program
characteristic variable, and incorporates federal funding and state political ideology into the
equation.
Carrell (1990) attempted to locate long-term care on Theodore Lowis distributive-
redistributive continuum of public policy. The researcher used education (distributive) and welfare
(redistributive) policies to provide polar benchmarks to render a judgment regarding the placement
46


of LTC on the distributive-redistributive continuum. Carrell states that distributive policies will be
driven by economic factors (ability to pay), while redistributive policies will be driven by political
factors, especially inter-party competition. Carrell examined state-level expenditure data for the
three policy areas, long-term care, education, and welfare for 1977-1986.
The LTC dependent variable in Carrells study is total Medicaid expenditure on LTC
services for physically and cognitively impaired persons over 65 years of age and per recipient
expenditure on LTC services for the elderly. The explanatory factors were economic (median
personal income and state fiscal capacity) and political (political liberalism, inter-party competition,
and intergenerational competition).
Carrell employed explanatory variables concerning the aged state population in the
expenditure-per-recipient model and total-Medicaid-expenditure model, and number of LTC
recipients in the total expenditure model.
The explained variation of the model for expenditure per LTC recipient was 59% with
political liberalism, fiscal capacity, and aged population statistically significant. The explained
variation of total Medicaid LTC expenditure was 96% with the number of LTC recipients
(standardized coefficient of +.903) and political liberalism statistically significant at the p<.05
level.
Carrell places LTC somewhere in the middle of the distributive-redistributive continuum
regarding public policy, with economic capacity, political liberalism, and interparty competition
important explanatory variables. He states that any movement in LTCs placement in the
continuum between 1977 and 1986 has been toward the redistributive end, with political liberalism
emerging as a more powerful explanatory variable in later years.
Swan (1990) used sets of supply, demand, and policy factors to explain interstate variation
and temporal variation in the proportion of state Medicaid monies expended on nursing home care.
47


The supply variable was nursing home bed stock. The demand factors were the proportion of
persons over 65 in the state, the percentage of women in the state labor force, the percent state
unemployment, heating degree days, personal income per capita, hospital beds per capita, hospital
occupancy, office MDs per capita, and Medicaid spend-down level.
The policy variables were Medicaid intermediate-care facility expenditure minus that
portion of intermediate-care facility expenditure used for the mentally retarded population (ICF-
MR) per capita and Medicaid-hospital expenditure per capita. Swan also included average nursing
pay per employee in a given state, as staffing constitutes a major portion of nursing home costs and
is expected to influence positively the percent of Medicaid expended on nursing homes.
Analysis included cross-sectional regression for each year of the period from 1979-1984,
and mixed cross-sectional time-series analysis of the pooled data for the entire period. The
adjusted R2 for each individual year of the study ranged from .45 (1983) to .61 (1981).
Swans analysis revealed that nursing-bed stock had the strongest positive influence on the
percent of Medicaid monies going to nursing home care. Each one-percent population aged was
associated with over one-percent share of Medicaid expenditure spent on nursing home care.
Furthermore, higher hospital occupancy was positively related to increased percent of
Medicaid monies being spent on nursing home care. This was interpreted as an early affect of
Medicares hospital-prospective payment system (PPS) as hospitals attempted to reduce the average
LOS, in part, by discharge to nursing homes. Medicaid expenditure per capita for ICF-MR and
acute-hospital care negatively affected the share of the Medicaid budget going to nursing homes.
Controlling for all factors in the model, the percentage of Medicaid spent for nursing
home care declined over time. Swan suggests that this decline in percent of the Medicaid budget
dedicated to nursing home care may reflect omitted state-policy factors aimed at containing
Medicaid nursing home utilization and expenditures.
48


This dissertation attempts to control for such policy variables.
Determinants of Geographic Variation in
Medicare Home Health Care Utilization
Swan and Benjamin (1990) examined Medicare home health utilization at the state level as
a function of state nursing home market factors. The authors proposed that home health utilization
(Medicare home health visits per 100,000 aged state population) is a function of demand and
supply factors.
The variables thought to affect the demand for Medicare home-care services and therefore
incorporated into the model were nursing home bed stock, Medicaid nursing home days, percent of
state population 85 or older, spend-down level for Medicaid eligibility, income per capita, percent
of women in the labor force, and the number of general hospital beds per capita. Total nursing
home bed stock is "used in the model. The authors acknowledge that Medicare-certified SNF beds,
hospital-based nursing home beds, hospital swing-beds, and rehabilitation-hospital beds may be the
critical substitutes for Medicare home health utilization.
The variables thought to affect the supply of Medicare home-care services were home
health-care agencies per 100,000 state population, percent of the state population in metropolitan
areas, and the percent of the work force that is unionized.
Regression analysis were performed for individual years in the 1978-1984 period, and
mixed cross-sectional time-series were used for the pooled data. The statistically significant
variables for explaining the number of Medicare home health visits per 100,000 aged state
population were the percent of the state population 85 or older (+), nursing home beds per 1,000
aged population (-), Medicaid nursing home days per 100 nursing home days (+), Medicaid spend-
down level (+), Medicare-certified home health agencies per 100,000 (+), and percent women in
the work force (-). Adjusted R2 was approximately .30 for individual years.
49


Swan and Benjamin suggest that a major factor in the home health-care market is what is
going on in the nursing home market. Greater demand for Medicare home health-care is generated
by a relative scarcity of nursing home beds, and perhaps, by greater access of Medicaid recipients
to those beds. The authors suggest that PPS may have strengthened those relationships.
This study suggests that a substantial portion of the unexplained variance in one health-
care service may be accounted for elsewhere"across the boundaries between Medicare and
Medicaid, and between acute and long-term care.
Summary of Empirical Work
Theoretical underpinning from multiple disciplines along with previous empirical work
indicate that Medicaid spending on LTC for the elderly is a function of endogenous state socio-
economic and demographic factors in conjunction with state-selected variables related to structural
components of the Medicaid program, the medical marketplace, and political ideology.
Socio-economic variables found to correlate with state Medicaid spending, or more
specifically, state Medicaid spending on nursing home care are:
effective price of Medicaid to taxpayers (-), (Grannemann, 1979)
income of taxpayers (+), (Grannemann, 1979)
state level per capita income (+), (Buchanan et al., 1991)
state cost of living (+), (Reutzel, 1984)
average nurses pay at the state level (+), (Harrington & Swan, 1987)
financial resources of the elderly (+), (Scanlon, 1980)
fiscal capacity of the state (+), (Carrell, 1990).
50


State demographic variables associated with either overall Medicaid spending or Medicaid
spending on nursing home care are:
number of aged persons in the state population (+), (Grannemann, 1979; Carrell,
1990; Swan, 1990)
age structure of the elderly population (+), (Scanlon, 1980)
percent state population living in metropolitan areas (+), (Harrington & Swan, 1987).
State-selected policy variables regarding either the structure of the state Medicaid program
or the state-medical marketplace associated with overall Medicaid spending or Medicaid spending
on nursing home care are:
physicians per capita (+), (Grannemann, 1979; Barrilleaux & Miller, 1988; Buchanan
etal., 1991)
hospital beds per capita (+), (Grannemann, 1979; Ray, 1987descriptive study only)
number of nursing home beds per 1,000 elderly (+), (Scanlon, 1980; Reutzel, 1984;
Harrington, 1987; Ray, 1987descriptive study only; Swan, 1990)
number of optional Medicaid services (+), (Reutzel, 1984)
presence of the optional "medically needy" program (+), (McDevitt & Buczko, 1985;
Harrington & Swan, 1987)
Medicaid per diem nursing home reimbursement rate (+), (Harrington & Swan, 1987)
reduced reimbursement rates to physicians (+), (Mauskopf et al., 1985)
limits on number or dollars on ambulatory physician visits (+), (Reutzel, 1984)
average LOS of elderly in acute-care hospitals (+) (Ray, 1987descriptive study only)
hospital occupancy rates (+), (Harrington & Swan, 1987; Swan, 1990)
level of administration of the Medicaid program (local +), (Schneider, 1988;
Buchanan et al., 1991).
51


State political variables found to influence either overall Medicaid spending or Medicaid
spending on nursing home care are:
political ideology (liberalism, +), (Barrilleaux & Miller, 1988; Carrell, 1990)
bureaucratic density (spending on Medicaid administration, +), (Barrilleaux & Miller,
1988)
interest group density (+), (Barrilleaux & Miller, 1988).
Swan and Benjamin (1990) examined Medicare home-care utilization at the state level as a
function of nursing home market factors. They found Medicare home health-care visits per
100,000 state population to be significantly correlated with:
state population over 85 (+)
percent women in the workforce (-)
Medicaid nursing home days/100 nursing home days (+)
Medicaid spend-down level (+)
number of Medicare-certified home health agencies per 100,000 state population (+)
number of nursing home beds per 100 state-aged population (-)
The present dissertation is consistent with previous empirical and theoretical work on
interstate variation in Medicaid and Medicare spending in that it incorporates many of these socio-
economic, demographic, political and policy variables in the explanatory model. This study
contributes to our understanding of the issue by expanding usual state Medicaid policy, economic,
political, and state-taxpayer variables in the model. Furthermore, this study delineates the
components of Medicaid spending on LTC for the elderly into both the components of spending
i.e., average expenditure per recipient and number of recipients of LTC in a manner consistent
with Harrington and Swan (1987), and service area within LTC, i.e., nursing home and home and
community care. The explained variation across states in Medicaid spending on LTC differentiated
52


by the components of spending and service area may show very different results as states have
been shown to spend their Medicaid dollars very differently (Grannemann, 1979; Rivlin & Weiner,
1987). For instance, the state of New York was responsible for 80% of the national Medicaid
expenditure on home health care in 1986 (Swan, 1990).
To the best of present knowledge, this dissertation represents the first attempt to model
state variation in Medicaid spending on both home care and the 2176 Waiver Program. This study
also represents an initial attempt to calculate and explain interstate variation in private spending on
nursing home care for the elderly. Data limitations have previously precluded such an attempt.
53


CHAPTER 4
METHODOLOGY
The purpose of this study is to explain interstate variations in long-term care spending for
the elderly. The central hypothesis is that endogenous socio-economic and demographic state
variables and state-selected policy and political variables differentially affect access, spending, and
utilization patterns of the elderly to Medicaid long-term care services.
Public expenditures on long-term care for the elderly (Medicaid) is analyzed by service
area (nursing home care, home care, and Medicaid 2176 Waiver Program) and, when possible, by
the components of spending (average expenditure per recipient, number of recipients, and total
expenditure).
Private spending on long-term care for the elderly is analyzed for total expenditure on
nursing home care only. Socio-economic, demographic, policy, and political variables are used in
the analysis to explain variations in long-term care spending across states. These independent
variables are consistent with the six theoretical models of public spending on long-term care
discussed in the preceding chapter. To reiterate, these six models are: the medical model, the
cultural model, the economic model, the structural model, the political model, and the public-goods
model.
The objectives of this study are:
1. To determine which socio-economic, demographic, policy, and political variables drive
public long-term care spending for the elderly;
2. To offer historical data for forecasting state spending on long-term care for the elderly;
54


3. To determine whether political/policy variables differentially affect long-term care
spending on the elderly by service area, or by the components of spending (average expenditure
amounts or recipient data across service areas);
4. To suggest long-term care cost-containment strategies to state policy makers;
5. To determine whether Medicaid home care spending and utilization act as a
compliment or substitute for Medicaid nursing home care; and
6. To determine which socio-economic, demographic, policy, and political variables
drive private spending on nursing home care for the elderly across states.
Model
The general model of this study is expressed in functional notation by:
Y = f (E, L, D, F, M, H, P)
where:
Y = State level LTC use or expenditure level
E = Economic factors
L = Demographic characteristics of the elderly population
D = Demographic characteristics of the voting population
F = Federal Medicaid policy constraints
M = Medicaid program characteristics
H = Health-care system characteristics
- P = Political climate of the state
These explanatory variables can be categorized as either socio-economic or demographic
variables that are relatively external to state manipulation, or state-selected political or policy
variables that influence the state long-term care system, but may be driven by state socio-economic
55


conditions or demographics. For example, health-care system characteristics, such as the supply of
nursing home beds, may reflect underlying demand for services that is a function of economic and
demographic factors.
Hypotheses
Once again, the central hypothesis of the study is that state socio-economic and
demographic variables, as well as state-selected policy and political variables, differentially affect
access, expenditure, and utilization patterns of the state elderly population to LTC services. As
such, state variables regarding health-care policy, the medical market place, and political ideology
do not merely reflect states macrocontextual economic and demographic status, rather they exert a
separate and distinct influence on LTC services used by the elderly.
The reduced model utilized in the study includes only the socio-economic and demographic
variables and can be expressed in linear form as: (note that specific variables are now presented in
the model as opposed to categories of variables)
Y = aWGNH + bOLD + cELDVT + dAVINC + eCOST + fUEMP
+ gSMSA + hHOUS + iTEMP + jYEAR
While the expanded model can be expressed in linear form as:
Y = aWGNH + bOLD + cELDVT + dAVINC + eCOST + fUEMP + gSMSA
+ hHOUS + iTEMP + jYEAR + kEMN + 1MOB + mRATE + nMAD
+ oBEDS + pLOS + qLQ
56


Independent Variables
1. Socio-Economic. Demographic Variables
WGNH = average weekly wage in the nursing home industry across states, 1982 dollars
OLD = number of persons >85 per 1,000 elderly
ELDVT = number of persons >65 per 1,000 persons >18
AVINC = average income of four-person family, 1982 dollars
COST = effective unit cost of LTC to state taxpayers, 1982 dollars, calculated via
([1-CORTAX] x [1-FMMP] x WGNH)
where:
CORTAX = corporate net income to government as a percentage of state tax collection
FMMP = Federal Medicaid Matching percent
WGNH = average weekly wage in nursing home industry
UEMP annual state unemployment rate
SMSA = percent state elderly population living in standard metropolitan statistical area;
1980 data of percent elderly in SMSA projected forward for 1981-1988 based
on percent elderly population in SMSA/percent general population in SMSA
from base year 1980; percent general population in SMSA available for years
1980 through 1988
HOUS = average population per household in state
2. State-Selected Variables
EMN = inclusion of elderly in Medicaids optional medically needy eligibility program
MOB = number of optional benefits afforded the categorically needy in state Medicaid
program (optional benefits specific to < 21 population excluded in count)
57


RATE = weighted average of SNF and ICF Medicaid per diem rates, 1982 dollars,
calculated:
{SNF rate X [SNF days/SNF + ICF (non-MR) days of care]}
+ {ICF rate X [ICF (non-MR) days of care/ICF (non-MR) + SNF days of care]},
(personal communication with HCFA, Medicaid division)
MAD = level of administration of Medicaid program, state versus local
BEDS = number of nursing home beds (freestanding and hospital based) per 1,000 elderly
LOS = average length of stay in an acute hospital for state Medicare population
LQ = liberal quotient of the state political climate
3. Control Variables
TEMP average annual state temperature
YEAR = year, 1981-1988
Dependent Variables
There are three sets of dependent variables:
1. state-level Medicaid spending on LTC for the elderly, all states as data allow, 1981 to
1988;
2. state-level Medicaid spending on LTC for the elderly, excluding New York as New
York is the clear outlier state, particularly on home care spending and utilization as previously
noted, 1981 to 1988; and
3. private spending on nursing home care for the elderly, all states as data allow, 1987
and 1988.
58


Whenever possible, expenditure measures have been delineated into expenditure per 1,000
elderly, expenditure per recipient, and number of recipients per 1,000 elderly; and service area,
i.e., spending on nursing home care versus home and community care.
There are a total of ten dependent variables: nine specific to Medicaid expenditures on
LTC for the elderly; and one on private spending on nursing home care for the elderly.
Medicaid Dependent Variables (1981 to 19881
1. Total state Medicaid spending on LTC for the elderly per 1,000 elderly population.
2. State Medicaid spending on nursing home care for the elderly per 1,000 elderly
population.
3. Number of elderly recipients of Medicaid nursing home dollars per 1,000 elderly
population.
4. Average medicaid expenditure per elderly Medicaid nursing home recipient.
5. State Medicaid spending on home care for the elderly per 1,000 elderly population.
6. Number of elderly recipients of home care spending per 1,000 elderly population.
7. Average Medicaid expenditure per elderly Medicaid home-care recipient.
8. State Medicaid spending on the 2176 waiver program for the elderly per 1,000 elderly
population.
9. State Medicaid spending on institutional care (nursing home) as a percent of total state
Medicaid spending on LTC for the elderly.
Private Spending Dependent Variable (1987 and 1988)
10. Total private spending on long-term nursing home care per 1,000 elderly population.
Table 4.1 summarizes the hypothesized relationships among the independent and dependent
variables.
59


as

Table 4.1. Summary of Hypothesized Relationships
Total LTC Exp. N.H. Exp. tt N.H. Rec. Avg. N.H. Exp. Per Rec. Home Care Exp. tt Home Care Rec. Avg Home Care Exp. Per Rec. 2176 Waiver Exp. N.H. Exp. as % of LTC Exp. Priv. N.H. Exp.
Wages in the nursing home industry + + 0 + + 0 + + 0 +
Old elderly (a 85)/elderly (> 65) population + + + + + + 0 + + +
Elderly voting pop. (> 65)/voting (> 18) pop. + + + + + + 0 + 0 0
Average state income + + + + + + + + 0 +
Unit cost of long-term care to state taxpayers - - - - - - 0 0 0 0
Percent state unemployment - - - - - - - 0 -
Percent state elderly population living in SMSA 0 + 0 + 0 + - 0 0 0
Average persons per household - - - 0 - - 0 - 0 -
Average state temperature - - - 0 0 0 - 0 - -
Year + + 0 + + + + + - 0
Elderly included in "medically needy" Medicaid eligibility category + + + 0 0 + 0 0 + 0
Number of optional benefits to elderly in state Medicaid Program 0 0 - + 0 0 0 0 - 0
Medicaid per diem nursing home rcimb. rate + + 0 + + 0 + 0 0 0
State/local administration of Medicaid Program + + + 0 0 0 0 0 0 0
Number of nursing home beds/1,000 eld. population + + + 0 - - 0 - + +
Medicare acute care hospital length of stay - - - - - - - - 0 -
Liberal quotient of Congressional voting record + + + + + + + + - 0


Total Medicaid spending for LTC of the elderly is disaggregated into nursing home, home
care, and 2176 Waiver spending to investigate conventional wisdom that state LTC services have
mirrored the evolution of the LTC literature toward the provision of services in an environment of
least restriction. The components of aggregate spending, namely, number of recipients and
average expenditure per recipient, are investigated as each may be affected differently by Medicaid
policy. Each component of LTC spending carries implications regarding access to services (e.g.,
number of recipients) and intensity of services (e.g., average expenditure per recipient).
Data on the components of Medicaid spending, both services areas and recipient/-
expenditure data, provide a longitudinal view of expenditures and gross utilization patterns for LTC
services across states.
Design
A mixed cross-sectional, time series design is used in the research. The nature of the
data, i.e., a state-level program, mandates a cross-sectional approach. A time-series is utilized to
capture dynamic changes in the coefficients across years. It is generally understood that mixed
cross-sectional, time series design may complicate the estimation process, but there is little
alternative in this case given the breadth of the analysis.
Stimson argues that data which are "cross-sectional dominant" is appropriate for least
squares analysis when dummy variables marking time frame are incorporated into the design
(Schneider, 1988). Cross-sectional dominant refers to a design where there are more data sets in
the cross-section than in the time series. This mixed cross-sectional, time series approach is
consistent with the majority of researchers who have investigated interstate variations in Medicaid
or long-term care. These researchers include Grannemann (1979); Harrington and Swan (1987);
61


Schneider (1988); Harrington, Swan, and Grant (1988); Carrell (1990); and Buchanan et al.
(1991).
A pooled series for eight years (1981 to 1988) is used for Medicaid expenditures on
nursing home, home care, and total LTC expenditures for the elderly. A seven year pooled time-
series, 1982-1988, is employed for Medicaid 2176 spending, because the 2176 waiver program did
not exist prior to 1982.
Regression analyses utilizing the expanded model does NOT include year 1982 as the
number of optional Medicaid benefits offered to the categorically needy (MOB) were not available.
State LOS measures were estimated for years 1987 and 1988, as state measures were not available.
Actual state 1986 LOS data were projected forward using percent change in national 1987 and
1988 figures provided by the Department of the Census. Dramatic changes occurred in LOS
between 1981-1986, for which actual state data exist. It is generally held that the dramatic changes
in acute-care LOS for the Medicare population had stabilized by the late 1980s.
Private spending on long-term nursing home care is estimated for 1987 and 1988, using
nursing home cost reports as a basis. Private spending is captured as private expenditure on
nursing home care per 1,000 elderly only. No recipient or average expenditure per recipient data
are estimated.
Unit of Analysis
To model interstate variation of Medicaid spending on LTC for the elderly, Medicaid data
from the 49 states with Medicaid programs plus Washington, D.C. were collected. Long-term care
was not covered in the Arizona Health Care Cost Containment System (AHCCS) during the
observation period.
62


Buchanan et al. (1991) omitted Alaska and Hawaii from their analysis of the factors
influencing the level of state Medicaid spending because these states have been found to have
disproportionate influence in quantitative analyses using state-level data. They excluded
Washington D.C., because it lacks a state government. Alaska, Hawaii, and Washington D.C. are
included in this analysis as data allow. No nursing home wage data were available for Alaska,
eliminating Alaska from the study. Nursing home wage data were available for four years only for
Washington D.C., and no liberal quotient was available for D.C., since D.C. lacks a congressional
voting record. Washington D.C. is, therefore, omitted from all expanded models and half (4/8) of
the reduced models. Hawaii is included in all models and time series.
Data Collection
Data for the independent variables are collected from published secondary sources
accounting for a high degree of reliability in the data. The economic variables are collected from
the Bureau of the Census and the Bureau of Labor Statistics. The demographic data on state
elderly and general populations are collected from various branches of the Bureau of the Census
and the Internal Revenue Service. The Federal Medicaid Matching Rate and Medicaid program
characteristics are collected from the Office of Medicaid Management and the HCFA. Health care
system variables are obtained from the Statistical Abstract of the United States and previously
published empirical work. The political quotient was computed from rankings given Congressional
members by the ADA. Measures for each independent variable have a high degree of internal
validity in that they quantify what they are intended to represent, with the possible exception of the
ADAs ranking of Congressional voting records as a surrogate for state political ideology. All data
collection occurred at the state level. The appendix lists specific measurements and data sources.
63


Data for annual Medicaid expenditures on total LTC services for the elderly, nursing
home expenditures for the elderly, and home care expenditures for the elderly are collected from
the HCFAs Division of Medicaid Statistics. The Division of Medicaid Statistics also provided
data on the components of spending for each of the service areas, i.e., number of recipients and
average expenditure per recipient. No attempt was made to verify this data which are assumed to
be both valid and highly reliable.
Aggregate expenditure data after 1982 for 2176 waiver spending on the elderly by each
state were obtained from the HCFAs Office of Research and Demonstrations.
Private expenditures on nursing home services across all states were estimated in general
accord with the methods used by the HCFAs Office of the Actuary in determining annual national
private spending on nursing home care (personal communication and HCFA, 1990). HCFA uses
financial data from the 1977 National Nursing Home Survey to estimate total nursing home
spending, then extrapolates spending forward based, primarily, on Bureau of Labor Statistics data
on employment and work hours associated with nursing and personal care facilities. HCFA then
extracts public spending on nursing home care, (Medicare, Medicaid, ICF/MR funds, VA funds,
etc.) which leaves a residual called private spending.
Total (public plus private) state-level nursing home patient revenue is estimated for this
study via nursing home cost reports collated by Health Care Investment Analysts (HCIA) of
Baltimore, Maryland. HCIA has cost reports from nursing homes from each state approximating
40% of the total number of nursing home facilities in the nation for 1987 and 1988. This sample
allows an extrapolation to total state level nursing home revenues via a simple multiplier of total
musing home beds in the state divided by total nursing home beds in the sample. This revenue
projection was adjusted for the elderly population via the percent elderly nursing home population
64


statistic published by the National Center for Health Statistics for 1986 (U.S. Department of Health
and Human Services, 1988).
Medicaid and Medicare associated expenditures on nursing home care for the elderly were
extracted from these amounts. Medicare covered charges" for SNF services are subtracted from
the revenue calculation as opposed to Medicare "reimbursements." Medicare covered charges
include both Medicare reimbursements and co-payments, either out-of-pocket expenses or private-
insurance supplements. This study targets long-term care for the elderly, while Medicare
associated charges represent short-term, skilled nursing home services following an acute-care
episode. State level VA and ICF/MR expenditures are not subtracted from the calculation as long-
term VA and ICF/MR beds are not included in the state projections of nursing home beds
(personal communication with researcher, Grant, 1991). State-only monies for nursing home care
will be subtracted for Delaware and Massachusetts, as these are the only states using state-specific
funds on chronic nursing home care. The vast majority of state-only programs target home and
community services, with few state dollars, if any, being spent on nursing home care. A state-
level private-pay residual for long-term nursing home care for the elderly results. It is assumed
that nursing home occupancy rates are equal in the state samples and the general state population.
It is further assumed that state samples represent an equivalent ratio of private pay to publicly
subsidized nursing home beds as exists in the general state population. HCIA states that there is
nothing inherent in their sampling technique to disallow either assumption (HCIA, personal
communication, 1991). This assurance by HCIA regarding their sampling technique in conjunction
with the utilization of a generally accepted methodology used by HCFA enhances the degree to
which the results of the regression analysis specific to private spending on nursing home care can
be generalized to the universe beyond the sample (external validity).
65


For all other aspects of the study the entire universe of appropriate data are examined as
data allow.
Data Analysis
The data for this study are analyzed in two ways:
1. Descriptive analysis using simple means and standard deviations of state spending and
type expenditure;
2. Regression analysis of dependent variables as a function of independent explanatory
variables, using both reduced and expanded models to more precisely isolate the impact of state-
selected variables.
Multiple regression analysis is the cornerstone statistical analysis of this study since the
goal of the research is to develop a model that maximizes the explained variance in the state
variations in expenditures on LTC for the elderly. Ordinary least square regression is used for the
reduced and expanded models.
66


CHAPTER 5
RESULTS
The results of this study are presented by descriptive and regression analysis. Descriptive
analysis examines both dependent and independent variables for individual years 1981 through 1988
to note national trends. Descriptive analysis also examines aggregate mean values for dependent
and independent variables, 1981 through 1988, with high and low outlier states noted.
Regression analysis results are organized around dependent variables. Each dependent
variable has four equations, i.e., reduced and expanded models run inclusive and exclusive of the
state of New York.
Descriptive Analysis
Descriptive analysis examines, first, the dependent variables in the study (Medicaid
spending and utilization data), and second, the independent variables in the study (state socio-
economic, demographic, policy, and political variables) by and across years, 1981 through 1988.
Third, correlation coefficients between dependent and independent variables are presented and
discussed.
Dependent Variables
Table 5.1 documents a slow increase in total Medicaid dollars spent on LTC for the
elderly and total Medicaid dollars spent on nursing home care for the elderly, on a per 1,000
elderly population basis, from 1981 to 1988, all in 1982 dollars. Expenditure figures were
adjusted by the CPI, not a medical or LTC CPI. It is generally assumed that LTC throughout the
1980s outstripped the general CPI by about 2 percentage points (Rivlin & Weiner, 1988).
67


Table 5.1. Descriptive Data of Dependent Variables, By Year, 1981 1988, in 1982 Dollars
Mean (Standard Deviation)
1981 1982 1983 1984 1985 1986 1987 1988
Total Medicaid spending on long-term care for the elderly per 1,000 elderly population 313,307 (159,537) 319,937 (172,015) 316,141 (146,033) 327,103 (158,934) 338,770 (158,551) 345,561 (170,107) 336,907 (158,509) 362,985 (183,023)
Medicaid spending on nursing home care for the elderly per 1,000 elderly population 308,370 (153,573) 313,839 (165,314) 307,677 (136,003) 313,549 (146,977) 322,074 (142,481) 323,176 (149,287) 313,883 (142,360) 333,183 (161,289)
Average Medicaid expenditure per elder recipient of nursing home care 6,689 (1,863) 7,019 (2,153) 7,113 (2,835) 7,543 (3,275) 7,524 (3,058) 7,649 (3,613) 7,734 (3,360) 7,847 (3,567)
Elderly Medicaid recipients of nursing home care per 1,000 elderly population 45.37 (14.91) 44.16 (13.44) 43.96 (12.74) 42.25 (12.22) 43.63 (13.16) 43.56 (12.58) 41.49 (11.13) 43.56 (11.55)
Medicaid spending on home care for the elderly per 1,000 elderly population 4,933 (16,539) 5,815 (18,716) 6,414 (20,681) 7,900 (22,544) 11,025 (28,076) 15,154 (31,640) 16,691 (26,180) 20,472 (38,962)
Average Medicaid expenditure per elder recipient of home care 660 (492) 824 (642) 894 (652) 1,051 (648) 1,223 (774) 1,515 (949) 1,578 (997) 1,745 (1,316)
Elderly Medicaid recipients of home care per 1,000 elderly population 4.84 (8.25) 4.63 (8.36) 4.98 (8.28) 5.44 (7.70) 6.33 (7.91) 7.64 (8.48) 8.13 (7.52) 9.23 (7.98)
Medicaid spending on 2176 Waiver Program for the elderly per 1,000 elderly population 211 (1,048) 1,996 (6,136) 3,958 (8,605) 5,352 (11,781) 6,858 (10,818) 7,954 (11,477) 9,310 (11,073)
Private spending on nursing home care for the elderly per 1,000 elderly population 422,897 (175,221) 448,510 (194,540)
Percent of long-term care Medicaid spending on nursing home care 98.86 (2.54) 98.55 (2.87) 97.73 (4.01) 96.58 (4.93) 95.63 (6.04) 93.79 (6.24) 93.30 (6.03) 91.65 (7.17)


Adjusting Medicaid expenditure amounts by a LTC component of the CPI is likely to have
demonstrated no change across years in real dollar spending on nursing home care, or overall LTC
for the elderly on a per 1,000 state elderly population basis. Unfortunately, yearly LTC CPI
statistics are currently not published.
National average of the number of elderly Medicaid recipients of nursing home care has
slipped from 45.4 in 1981 to 43.6 in 1988, each on a per 1,000 elderly state population basis. Real
average Medicaid expenditure per nursing home resident increased by 17% from 1981 to 1988.
Dramatic change occurred in non-institutional Medicaid spending and utilization patterns
from 1981 to 1988. Real Medicaid spending on home health services per 1,000 elderly population
more than quadrupled; the number of Medicaid recipients of home-care services per 1,000 state
elderly population approximately doubled; and real average expenditure per Medicaid home health
care recipient approximately tripled. Medicaid spending on the 2176 Waiver Program per 1,000
state elderly population grew from $211 in 1982 to $9,310 in 1988, in 1982 dollars. Concurrently,
the percent of the Medicaid LTC budget for the elderly devoted to nursing home care, i.e.,
institutional versus non-institutional care, shrank from 98.9% in 1981 to 91.7% in 1988.
Within these national averages lie great variation (Table 5.2). This study certainly
supports the long-held assertion that wide variations exist across states in Medicaid expenditure and
utilization patterns. This assertion holds true even within one services area (LTC) for one target
group, the elderly.
69


Table 5.2. Summary of Descriptive Data of Dependent Variables, Per 1,000 State
Elderly Population, By State, 1981 1988
High Low Mean
Total Medicaid spending on long-term 893,313 133,650 321,194
care New York Florida
Medicaid spending on nursing home 716,725 125,930 313,005
care New York Florida
Medicaid recipients of nursing home 81.7 18.9 43.9
care Minnesota Florida
Average Medicaid spending per nursing 15,779 4,625 7,382
home recipient New York Iowa
Medicaid expenditure on home care 173,111 210 10,192
New York Wyoming
Medicaid recipients of home care 45.6 .3 6.2
New York Wyoming
Average Medicaid spending per home 4,009 162 1,115
care recipient New York Mississippi
Medicaid spending on 2176 Waiver 44,080 -0- 5,079
Program Oregon 9 States
Percent Medicaid long term care 100.00 79.3 96.2
spending on nursing home care Wyoming Michigan Oregon
Private spending on nursing home care 929,603 155,124 425,204
Nebraska Alabama
70


Independent Variables
Table 5.3 lists national, yearly averages for the independent variables utilized in the study.
Demographic and economic trends are obvious from the data. The aging of the population is
evident from the increase in the number of persons over 85 per 1,000 persons over 65; the number
of persons over 65 per 1,000 persons over 18; and the decline in the average number of persons
per household. The percent elderly living in standard metropolitan statistical areas was remarkably
stable, with a slight rise from 1981 to 1988.
Economic indicators, i.e., average state income and wages in the nursing home industry
dipped in the early 1980s and rose in the later 1980s, conversely, the unemployment rate rose in
the early 1980s and declined in the late 1980s. This is consistent with the economic recession of
the early 1980s. The unit cost of LTC to state taxpayers rose by approximately 2.5%, from 1981
to 1988, in real dollars.
Policy variables indicate heightened generosity in Medicaid programs toward the elderly in
the later 1980s with states increasing the number of optional Medicaid benefits to the categorically
needy and more states including the elderly in their optional medically needy program. This
finding is consistent with findings of Chang and Holahan (1990), who documented increased
generosity in state Medicaid programs across many service areas and eligibility groups from 1984
through 1987, inclusive of LTC. This increased generosity is also reflected in Medicaid
reimbursement rates as the real average Medicaid per diem rate for nursing home care rose more
than 50%, from $36.9 in 1981 to $56.6 in 1988. The nursing home bed stock has diminished
slightly, on a per 1,000 state elderly population basis, from 56.00 in 1981 to 55.61 in 1988, with a
low of 55.24 in 1986. The average LOS for the Medicare population in an acute-care hospital
declined from 10.08 in 1981 to 8.09 in 1985 to approximately 8.30 in 1988. This trend
demonstrates the effects of Medicares prospective payment system that was implemented in
71


Table 5.3. Descriptive Data of Independent Variables, By Year, 1981 1988, in 1982 Dollars
Mean (Standard Deviation)
1981 1982 1983 1984 1985 1986 1987 1988
Wages in the nursing home industry 157.6 154.2 155.2 157.1 156.3 158.4 158.5 163.5
(21.10) (18.3) (20.0) (22.9) (22.9) (25.2) (25.6) (32.7)
Old elderly (S 85)/elderly (> 65) 90.8 92.5 93.5 94.6 95.3 95.9 96.7 97.5
population (13.4) (13.9) (14.1) (14.4) (15.0) (15.0) (15.3) (15.4)
Elderly voting population 154.1 155.4 156.8 158.3 159.8 161.9 163.9 165.3
(> 65)/voting (> 18) population (28.0) (28.2) (28.0) (27.6) (27.0) (27.1) (26.5) (26.2)
Average state income 19,034 18,869 18,390 19,025 19,233 19,700 20,171 20,877
(2,121) (2,028) (2,093) (2,240) (2,344) (2,567) (2,686) (3,037)
Unit cost of long-term care to state 59.9 58.7 60.0 60.2 59.3 60.5 59.8 61.2
taxpayers (18.2) (16.5) (17.4) (19.9) (19.0) (20.1) (20.1) (23.8)
Percent state unemployment 7.3 9.2 9.3 7.3 7.1 7.0 6.3 5.5
(1.9) (2.3) (2.6) (2.2) (2.0) (2.3) (2.2) (2-0)
Percent state elderly population living 58.3 58.3 58.2 58.3 58.4 58.5 58.7 58.8
in SMSA (24.4) (24.4) (24.4) (24.3) (24.3) (24.2) (24.1) (24.1)
Average persons per household 2.73 2.72 2.72 2.69 2.67 2.66 2.64 2.61
(.13) (.13) (14) (.14) (.14) (.13) (.14) (.12)
Average state temperature 52.9 51.8 52.3 52.4 51.8 53.3 52.6 52.4
(7.9) (9.1) (7.9) (8.5) (9.0) (8.7) 8.5) (8.1)
Year
Elderly included in "medically needy" .60 .60 .58 .60 .58' .64 .66 .68
Medicaid eligibility category (.49) (.49) (.50) (.49) (.50) (.48) (.48) (.47)
Number of optional benefits for elderly 17.4 18.0 18.3 18.6 18.8 19.4 19,8
in state Medicaid Program (6.3) (5.8) (5.6) (5.5) (5.4) (5.2) (5-3)
Medicaid per diem nursing home 35.9 39.0 42.9 46.3 48.9 50.4 53.2 56.6
reimbursement rate (9.4) (10.7) (16.9) (19.0) (20.0) (20.6) (25.9) (26.2)


Table 5.3. (Cont.) Descriptive Data of Independent Variables, By Year, 1981 1988, in 1982 Dollars
Mean (Standard Deviation)
1981 1982 1983 1984 1985 1986 1987 1988
State/local administration of Medicaid .12 .12 .12 .12 .12 .12 .12 .12
Program (.33) (.33) (.33) (.33) (.33) (.33) (.33) (.33)
Number of nursing home beds/1,000 56.00 55.70 55.50 . 55.36 55.25 55.24 55.48 55.61
elderly population (17.68) (17.45) (17.08) (17.20) (17.46) (16.97) (16.70) (16.56)
Medicare acute care hospital length of 10.08 9.88 9.56 8.65 8.09 8.20 [ 8 30 1 [ 8 30 ]
stay (1.78) (1.68) (1.54) (1.49) (1.45) (1.25) 1 (125) || r (125) 1
Liberal quotient of Congressional 37.5 41.4 45.6 46.2 41.2 42.2 49.9 50.2
voting record (20.1) (22.6) (19.8) (21.2) (20.1) (20.8) (20.3) (20.6)


October of 1983. The liberal climate of the state, as measured by Congressional voting records,
actually increased from 1981 to 1988.
Wide variations in these variables exist across states. Table 5.4 summarizes the high and
low outliers for each independent variable. Clearly, the United States is a vast country with wide
and divergent state preferences and capabilities, indicating a greater federalism than one might
otherwise expect.
Correlation Coefficients
Economic state indicators, i.e., wages in the nursing home industry, average income, unit
cost of long-term care to state taxpayers, and percent unemployment are strongly correlated in
expected direction (Table 5.5). Economic indicators are also strongly, positively correlated with
Medicaid nursing home per diem rate, and surprisingly, with average LOS of the state Medicare
population in acute-care hospitals.
Rational, economic thought would suggest that suppliers of nursing home care, 80% of
whom are "for profit concerns, would increase their bed stock in areas with higher reimbursement
rates (Scanlon and Feldstein, 1988a). Surprisingly, states with higher per diem Medicaid nursing
home reimbursement rates have fewer nursing home beds per 1,000 state elderly population (-.30).
Regulatory agencies in states with high per diem rates may have greater incentive to contain
Medicaid costs by limiting nursing home bed stock. Limiting nursing home bed stock may delay
discharges from acute-care hospitals for the elderly who are seeking nursing home care, thereby
increasing the average LOS for the Medicare population in acute-care facilities. Surprisingly, the
correlation coefficient between acute-care hospital LOS for the Medicare population and nursing
home bed stock is -. 13 only. A stronger, positive correlation may be found between state
74


L
Table 5.4. Summary of Descriptive Data of Independent Variables, By State, 1981 1988
High Low Mean
Wages in the nursing home industry (1982 243.40 130.40 157.0
dollars) New York Arkansas
Old elderly (> 85)/elderly (> 65) 125.3 57.1 94.0
population Nebraska Nevada
Elderly voting population (> 65)/voting 227.6 48.0 162.8
(> 18) population Florida Alaska
Average state income (1982 dollars) 24,848 14,825 19,379
Connecticut South Dakota
Unit cost of long-term care to state 110.60 28.33 59.95
taxpayers (1982 dollars) New York Mississippi
Percent state unemployment 12.89 4.21 7.3
West Virginia New Hampshire
Percent state elderly population living in 100 15.2 56.6
SMSA D.C. Vermont
Average persons per household 3.20 2.36 2.69
Utah D.C.
Average state temperature 77.5 37.5 52.5
Hawaii Alaska
Year
Elderly included in "medically needy 1.00 (yes) .00 (no) .62
Medicaid eligibility category 29 States 21 States
Number of optional benefits to elderly in 28.3 6.9 18.6
state Medicaid Program Minnesota Wyoming
Massachusetts
Medicaid per diem nursing home 87.83 31.66 45.74
reimbursement rate (1982 dollars) D.C. Louisiana
State/local administration of Medicaid 1.00 (Local) .00 (State) .12
Program 6 States 44 States
Number of nursing home beds/1,000 86.1 24.1 55.6
elderly population Wisconsin Florida
Medicare acute care hospital length of stay 13.5 6.7 8.96
New York Idaho
Liberal quotient of Congressional voting 84.9 6.88 44.0
record Massachusetts Utah
Wyoming
75


Table 5.5.
Correlation Matrix
Wages in NH Industry Old/Elderly Elderly Elderly Voters/ Voting Population Average State Income Unit Cost of LTC to State Taxpayers Un- Employ- ment Rate Elderly in SMSA # Persons Per Household Ave. Temp.
Wages in the nursing home industry 1.00 -.08 -.04 .63 .84 -.33 .16 .03 -.09
Old elderly (S 85)/elderly (> 65) pop. -.08 1.00 .06 -.16 .02 -.32 -.36 -.41 -.51
Eld. voting (a 65)/voting (a 18) pop. -.04 .06 1.00 -.08 -.09 .21 -.07 -.12 -.02
Average state income .63 -.16 -.08 1.00 .73 -.37 .56 -.13 -.05
Unit cost of LT care to state taxpayers .84 .02 -.09 .73 1.00 -.31 .63 -.12 -.12
Percent state unemployment -.33 -.32 .21 -.37 -.31 1.00 -.05 .12 .17
Percent state elderly population in SMSA .16 -.36 -.07 .56 .63 -.05 1.00 .03 .23
Average persons per household .03 -.41 -.12 -.13 -.12 .19 .03 1.00 .27
Average state temperature -.09 -.51 -.02 -.05 -.12 .17 .23 .27 1.00
Year .02 .15 .02 .21 -.01 -.23 -.03 -.24 .01
Elderly included in "medically needy" Medicaid eligibility category .24 .10 .07 .17 .18 -.17 .02 -.06 -.23
Number of optional benefits to elderly in state Medicaid Program .32 .22 .02 .30 .32 -.16 .12 -.19 -.37
Medicaid per diem nursing home reimbursement rate .51 -.01 -.02 .34 .54 -.39 .19 .06 -.05
State/local administration of Medicaid .27 .10 -.02 .03 .21 -.07 -.19 -.10 -.18
No. of nursing home beds/1,000 eld. pop. -.28 .63 -.07 -.13 -.04 -.15 -.15 -.26 -.38
Medicare acute care hospital length of stay .54 -.11 -.02 .43 .47 -.10 .24 -.04 -.03
Liberal quotient of Congressl vote record .38 .21 .13 .24 .38 -.10 .23 -.32 -.22


Table 5.5. (Cont.) Correlation Matrix
Year Elderly in "Medically Needy" Optional Medicaid Benefits to Elderly Medicaid NH Per Diem Rate State/Local Administra- tion of Medicaid tt of Nursing Home Beds /1.000 Elderly Medicare Acute Hospital LOS Liberal Quotient of Congress
Wages in the nursing home industry .02 >, -24 .32 .51 .27 -.28 .54 .38
Old elderly (> 85)/elderly (> 65) pop. .15 .10 .22 -.01 .10 .63 -.11 .21
Eld. voting pop. (a 65)/voting (a 18) pop. .02 .07 .02 -.02 -.02 -.07 -.02 .13
Average state income .21 .17 .30 .34 .03 -.13 .43 .24
Unit cost of long-term care to state taxpayers -.00 .18 .32 .54 .21 -.04 .47 .38
Percent state unemployment -.23 -.17 -.16 -.39 -.07 -.15 -.10 -.10
Percent state elderly population in SMSA -.03 .02 .12 .19 -.19 -.15 .24 .23
Average persons per household -.24 -.06 -.19 .06 -.10 -.26 -.04 -.32
Average state temperature .01 -.23 -.37 -.05 -.18 -.38 -.03 -.22
Year 1.00 .09 .07 .42 -.06 -.03 -.39 .12
Elderly included in "medically needy" Medicaid eligibility category .09 1.00 .46 .27 .15 .07 .13 .33
Number of optional benefits to elderly in state Medicaid Program .07 .46 1.00 .23 .27 .21 .14 .40
Medicaid per diem nursing home reimbursement rate .42 .27 .23 1.00 .10 -.30 .03 .39
State/local administration of Medicaid Program -.06 .15 .27 .10 1.00 .04 .20 .13
Number of nursing home beds/1,000 eld. pop. -.03 .07 .21 -.30 .04 1.00 -.13 .07
Medicare acute care hospital length of stay -.39 .13 .14 .03 .20 -.13 1.00 .35
Liberal quotient of Congressl vote record .12 .33 .40 .39 .13 .07 .35 1.00


Medicare population LOS in acute-care hospitals and nursing home occupancy rates. Nursing
home occupancy rates were not included in this study.
States with a higher average income have a greater concentration of their elderly
population living in urban areas (+.56). This positive relationship is expected to hold true for the
general state population.
States with a greater concentration of "old" elderly (age 85 or older) in their elderly
concentration have less unemployment, (-.32); have a smaller percentage of their elderly population
in urban areas, (-.36); have fewer people per household, (-.41); have more nursing home beds per
1.000 elderly population, (.63); and, have lower average temperatures, (-.51). A strong, positive
correlation between nursing home bed stock and the number of "old" elderly was expected as the
"old elderly" are at greater risk for physical and cognitive disabilities predisposing them to be in
institutional (nursing home) care (Rivlin & Weiner, 1988; Liu, Coughlin, and McBride, 1989).
State planning bodies reportedly consider the demographics of the state population and the
stratification of the elderly state population in approving additional nursing home bed stock (HCIA,
1991).
States with a greater number of optional Medicaid benefits for the elderly are more apt to
include the elderly in their optional "medically needy" eligibility category (+.46), and are more
likely to be high wage, high income states, +.32 and +.30 respectively. Colder states have fewer
optional Medicaid benefits for the elderly (-.37) and have a smaller nursing home bed stock per
1.000 elderly population (-.38).
Liberal states tend to be more "expensive" states with higher wages (+.38), higher
effective cost of long-term care to state taxpayers (+.38), and higher Medicaid per diem rates
(+.39). Liberal states appear to be more generous in terms of Medicaid eligibility for the elderly
via the "medically needy program (+.33), and optional Medicaid benefits for the elderly (+.40).
78


Liberal states also tend to have longer lengths of stay in acute-care hospitals for their Medicare
population (+.35); and fewer persons per household (-.32).
Regression Analysis
The results of the regression analysis are presented by dependent variable. As previously
stated, each dependent variable was examined via four equations, reduced and expanded models
run inclusive and exclusive of New York. New York is a clear outlier state regarding Medicaid
spending on LTC for the elderly, particularly concerning home care. The state of New York
accounted for 80% of the national Medicaid expenditure on home care in 1986 (Swan, 1990).
Dependent Variable
The dependent variables in this study are expenditure amounts for long-term care services
for the elderly per 1,000 state elderly population, percent of Medicaid spending on LTC for the
elderly devoted to nursing home care, elderly recipients of state long-term care services per 1,000
state elderly population, or average expenditure per elderly recipient of state long-term care
services. Each category is clearly delineated.
Model
Two models, reduced and expanded, are used to explain interstate variations in long-term
care spending for the elderly in this study. The reduced model includes state socio-economic,
demographic, and temperature/year measures as independent variables. The expanded model adds
variables specific to a states Medicaid program, health-care system, and political climate. Models
were also run exclusive of New York.
79


Total Medicaid Expenditure on Long-Term Care for the Elderly
and Medicaid Expenditure on Nursing Home Care for the Elderly
Total Medicaid spending on LTC for the elderly per 1,000 elderly population and
Medicaid spending on nursing home care for the elderly per 1,000 elderly population are discussed
together as regression analyses produced nearly identical results for each dependent variable. This
is not surprising as nursing home expenditures account for well over 90% of total Medicaid
spending on LTC for the elderly.
Previous theoretical and empirical work suggest that state Medicaid spending on nursing
home care for the elderly is associated with state economic variables (wages in the nursing home
industry [+], average state income [+], state unemployment [-], and unit cost of LTC to state
taxpayers [-]), demographic variables ("older" voting and elderly population [+] and household
size [-]), structure of the Medicaid program (elderly in the "medically needy" program [+],
Medicaid per diem nursing home reimbursement rate [+], and local administration of the Medicaid
program [+]), medical care market place variables (nursing home bed stock [+] and average
length-of-stay of Medicare population in acute-care hospitals [-]), and finally, a political variable
(liberal state ideology [+]). These findings form the hypotheses relevant to "total Medicaid
spending on LTC for the elderly" and "Medicaid spending on nursing home care for the elderly"
found summarized in Table 4.1.
The explained interstate variation in total Medicaid (nursing home, home care, and waiver
2176) spending per 1,000 state elderly population for years 1981-1988, utilizing the reduced model
is 59%, while the expanded model explains 78% of the variation across states (Table 5.6).
The explained interstate variation in Medicaid spending on nursing home care per 1,000
elderly population for years 1981-1988 utilizing the reduced model is 56%, while the expanded
model explains 79% (Table 5.7).
80


Table 5.6. State Medicaid Spending on Long-Term Care for the Elderly Per 1,000 Elderly Population, 1981 1988
Including New York Excluding New York
B T B T B T B T
Wages in the nursing home industry 5,974.4 13.8+ 6,934.7 15.1* 4,430.9 9.7* 4,707.5 10.3+
Old elderly (> 85)/elderly (> 65) population 5,193.0' 12.0* 902.7 1.7 4,636.6 11.2+ 518.1 1.1
Eld. voting pop. (> 65)/voting (> 18) pop. -220.2 -1.6 -356.5 -3.2+ -204.6 -1.6 -293.9 -3.1 +
Average state income -5.1 -1.3 -5.9 -1.5 -4.1 -1.1 4.2 1.2
Unit cost of long-term care to state taxpayers -2,671.2 -4.2* -3,290.1 -6.2* -2,169.3 -3.6+ -4,481.1 -8.4*
Percent state unemployment 1107.8 .4 7,055.2 2.8* -2291.0 -.90 2,806.2 1.3
Percent state elderly population living in SMSA 758.7 2.2+ 61.1 .20 826.9 2.5+ 283.9 1.0
Average persons per household 46,703.0 1.0 -56,642.9 -1.3 56,648.7 1.3 -34,339.4 -.90
Average state temperature -223.3 -1.7 -159.3 . -1.6 -155.8 -1.3 -90.0 -1.0
Year 537.0 .2 1,290.2 .4 66.7 .0 -12,401.8 -4.1*
Elderly included in "medically needy" Medicaid eligibility category 28,634.7 2.8+ 18,167.3 2.0*
Number of optional benefits to elderly in state Medicaid Program -2,825.6 -2.8+ -2,614.0 -3.0+
Medicaid per diem nursing home reimb. rate 2,150.4 3.2+ 5,066.6 7.8*
State/local administration of Medicaid Program 93,496.8 6.3+ 46,099.9 3.4+
No. of nursing home beds/1,000 eld. pop. 4,161.9 9.9+ 4,624.9 12.6+
Medicare acute care hospital length of stay 8,510.2 2.0+ -39,770.0 -1.0
Liberal quotient of Congressional voting record 547.8 1.9* 1,233.9 4.7*
Adjusted R2 .59 .78 .44 .73
Statistically significant at p = £.05


Table 5.7. State Medicaid Expenditure on Nursing Home Care for the Elderly Per 1,000 Elderly Population, 1981 1988
Including New York Excluding New York
B T B T B T B T
Wages in the nursing home industry 4,972.5 12.6* 5,919.8 15.2* 4,306.7 9.7* 4,651.9 10.9*
Old elderly (> 85)/elderly (> 65) pop. 4,900.9 12.3* 1,034.3 2.3* 4,660.1 11.6* 796.7 1.9*
Elderly voting pop. (> 65)/voting (> 18) pop. -178.1 -1.4 -289.4 -3.1* -171.9 -1.4 -252.6 -2.8*
Average state income -4.8 -1.3 -1.5 -.4 -4.2 -1.2 4.7 1.4
Unit cost of long-term care to state taxpayers -2,213.1 -3.8* -4,050.4 -7.8* -1,993.4 1 u> *4* * -4,496.5 -9.0*
Percent state unemployment -386.0 -.16 6,507.4 3.0* -1,827.5 -.76 4,052.0 1.9*
Percent state elderly population in SMSA 751.1 2.4* 164.6 .6 768.2 2.5* 284.4 1.1
Average persons per household 76,956.4 1.9 -14,406.7 -.4 81,281.6 2.0* -3,082.9 -.10
Average state temperature -195.1 -1.6 -126.9 -1.5 -166.4 -1.4 -86.7 -1.1
Year -2338.1 -1.0 -6,804.7 -2.6* -2,452.2 -1.0 -14,874.5 -5.2*
Elderly included in "medically needy" Medicaid eligibility category 29,806.8 3.4* 23,494.1 2.8*
Number of optional benefits to elderly in state Medicaid Pgm. -3,197.1 -3.7* -3,108.9 -3.8*
Medicaid per diem nursing home reimb. rate 3,460.7 6.2* 5,195.9 8.6*
State/local administration of Medicaid Pgm. 75,761.4 6.2* 48,768.9 3.9*
No. of nursing home beds/1,000 eld. pop. 4,432.9 12.4* 4,720.2 13.8*
Medicare acute care hospital length of stay 5,976.2 1.6 -1,717.1 -.50
Liberal quotient of Congressl voting record 691.9 2.8* 1,096.4 4.5*
Adjusted RJ .56 .79 .44 .75
Statistically significant at p = £ .05


Clearly macrocontextual variables of socio-economy and demography play a fundamental
role in driving nursing home and aggregate Medicaid expenditures on LTC for the elderly, while
policy/political variables (such as number of nursing home beds per 1,000 elderly and Medicaid
per diem nursing home reimbursement rate) exert a decidedly smaller, but distinct influence.
The statistically significant economic variables in the expanded model are nursing home
wages (+), the unit cost of LTC to state taxpayers (-), and percent state unemployment (+).
This study is consistent with Swan (1990), who also found nursing wages to be positively
related to Medicaid spending on nursing home care, but it runs contrary to the findings of
Harrington and Swan (1987) who found nursing wages NOT to be statistically related to Medicaid
expenditures on the elderly when Medicaid per diem nursing home rates were included in the
model (1987). This study found both nursing wages and Medicaid nursing home per diem rate to
be positively related to Medicaid spending on nursing home care for the elderly. The impact of the
nursing home wage variable reflects two important concepts, namely, the labor intensive nature of
the LTC industry (labor accounts for approximately 75-80% of the input costs of nursing home
care) and geographic cost of living (Swan, Harrington & Grant, 1990; HCIA 1990). Buchanan
(1987) found that differences in cost of living among states explained upwards of 60% of the
variance in Medicaid payments for skilled and intermediate nursing home care. Buchanan et al.
(1991) then developed a cross-state price index to deflate expenditure and per capita income for
their model explaining state Medicaid spending. This approach aligns expenditures more closely
with benefits. This dissertation adjusted nominal dollars into real 1982 dollars, but did not adjust
for regional variations in the cost of living. Expenditure levels should, therefore not be equated
with LTC service level.
83


The variable "unit cost of LTC to state taxpayers" is heavily influenced by regional
differences in cost of living as state wages in the nursing home industry is utilized in the
calculation. The correlation between wages in the nursing home industry and unit cost of LTC to
state taxpayers is +.84.
A negative relationship was hypothesized between unit cost of LTC to state taxpayers and
Medicaid nursing home recipient and average expenditure measures as rational economic theory
suggests that state taxpayers would be willing to "buy more" LTC, either by increasing
expenditures per recipient or increasing number of recipients as unit cost declined.
The unit cost of LTC to state taxpayers, as calculated in this study, is affected not only by
state nursing wages in the nursing home industry, but also by the percent state income derived
from corporate taxes and the FMMP applied to state Medicaid programs. As such, the FMMP
directly affects the unit cost of LTC to state taxpayers, which in turn relates inversely to Medicaid
spending on LTC for the elderly, i.e., the higher the FMMP to the state, the lower the unit cost of
LTC to state taxpayers and the higher the state Medicaid expenditure on LTC for the elderly. The
federal governments FMMP formula uses state per capita personal income to determine the
Federal Medicaid Matching Rate paid to states to subsidize their Medicaid programs. The formula
has been criticized on several counts. Compare Grannemann (1979), who argues that the FMMP
is inadequately weighted toward poor states, with Blumberg etc. (1993), who argue that the FMMP
fails to capture differences in the cost of living across states thus overstating the buying power of
high cost states. Once again, the important distinction between expenditure level and service level
is underscored, while the affect of the FMMP on states Medicaid LTC expenditure level appears
quite real.
Percent state unemployment is statistically significant with a positive coefficient in
explaining Medicaid spending on nursing home care for the elderly and total Medicaid spending for
84


the elderly when New York is included in the analysis. When New York is excluded from the
analysis, percent state unemployment remains statistically significant and positive relative to
Medicaid nursing home care spending for the elderly, but is no longer statistically significant
relative to total LTC spending for the elderly. A negative relationship between percent state
unemployment and Medicaid spending on total LTC and nursing home care for the elderly was
hypothesized. Percent unemployment was included in the model as a measure of the fiscal capacity
of the state, i.e., states with less unemployment would have more revenue to pay for LTC for the
elderly. It appears that percent unemployment is acting as a demand variable in this model,
whereby states with higher unemployment have a greater demand for Medicaid nursing home
services for the aged. State employment conditions may diminish the possibility of some elderly
finding work and may also diminish financial support from family members, thus increasing
demand for publicly subsidized nursing home care. Percent state unemployment is also statistically
significant, in a positive direction, regarding number of Medicaid nursing home care recipients.
Previous empirical work does support the notion that higher state unemployment increases demand
for Medicaid services in the general state population due to enhanced eligibility (Harrington &
Swan, 1987).
Three demographic variables are statistically significant in total Medicaid spending on
LTC for the elderly per 1,000 state elderly population and Medicaid nursing home spending for the
elderly, namely, the percent of the elderly population residing in metropolitan areas (+), the
number of eligible elderly voters (age 65 or over) per 100 eligible general voting population (age
18 or over) (-), and the number of old elderly (age 85 or over) per 100 elderly population (age 65
or. older) (+).
Harrington and Swan (1987) found states with a higher percentage of the population
residing in urban areas have higher Medicaid nursing home expenditures. This study found a
85


statistically significant relationship between percent elderly residing in urban areas and Medicaid
expenditure on nursing home and total LTC spending for the elderly in reduced models only.
When health care and political variables were added to the model, the percent elderly population
living in urban areas was no longer found to be statistically significant. It is of interest to note that
the percent elderly population residing in urban areas is statistically significant with a positive
coefficient relative to the number of elderly Medicaid recipients of nursing home care, and
statistically significant with a negative coefficient relative to average Medicaid expenditure per
nursing home care recipient in the expanded models. The net affect regarding Medicaid
expenditure on nursing home care appears to be neutral in the expended model.
The ratio of number of persons age 65 or older per 100 persons age 18 or older was
intended as a surrogate measure of a potential voting bloc for the elderly. As such, higher
Medicaid expenditures on nursing home care and total LTC was expected in states with a higher
ratio of potential aged voters to the general eligible voter population. A negative relationship was
not expected. This variable may in fact be operating as an economic indicator, whereby states with
a lower aged to working population ratio may have a greater fiscal capacity to support nursing
home and LTC services for the elderly.
The number of old persons (age 85 or older) per 100 state elderly population (65 or older)
is statistically significant in explaining state variation in nursing home care and total Medicaid
spending per 1,000 state elderly population in the reduced models, but remains statistically
significant in the expanded model for Medicaid spending on nursing home care for the elderly
only, with a greatly reduced T statistic. The explanatory power of the variable appears to be
diluted in the expanded model when the number of nursing home beds per 1,000 elderly population
is included in the analysis. The correlation between the number of old persons per 100 elderly
persons and the number of nursing home beds per 1,000 elderly is +.63.
86


Given the assumption that old persons do not migrate to states with higher nursing home
bed stock, it is presumed that the indigenous number of old persons is driving the number of
nursing home beds. There are separate but intersecting paths whereby the number of old persons
will drive nursing home bed stock. First, the nursing home industry is heavily weighted toward
the for-profit sector, with approximately 75% of nursing homes proprietary organizations
(Feldstein, 1988a). Conventional business strategy dictates that, within the confines of government
regulation, for-profit nursing home beds are positioned relative to potential markets, i.e., supply of
nursing home beds is market driven. The market for institutional LTC services is predominantly
the over 85 population. Market researchers for proprietary nursing home organizations attend to
demographics that relate to demand.
A second explanation for the strong relationship between the number of old persons in a
state and the number of nursing home beds involves government regulation. Many states with
certificate-of-need legislation intact consider allowing an increase in nursing home bed stock based,
in part, on the current and projected ratio of nursing home beds per 1,000 state population over the
age of 65. Thirty-seven states had some variant of CON regarding the addition of LTC beds as of
December, 1988 according to The Guide To The Nursing Home Industry published by Health Care
Investment Analysts, Inc. and Arthur Anderson (1990).
On the surface, the supply of nursing home beds as an explanatory variable of total
Medicaid LTC expenditures and nursing home Medicaid expenditures (and of the number of
recipients of nursing home care, discussed later) would seem to support Roemers Law. Roemers
Law states that supply of health-care services creates demand for those services (Feldstein, 1988a).
Nyman (1985) argues that the quality of care provided by nursing homes is lower in
markets where there is excess Medicaid demand than in areas where excess demand does not exist.
He postulates that high occupancy rates due to excess demand constrain market competition and
87


patient care expenditures and eliminate information important to residents concerning the quality of
care at an institution. He postulates that providers of nursing home care in under-bedded areas,
can reduce patient-care expenditures and quality of care without census repercussions because there
is no competition for Medicaid nursing home residents.
Davis and Freeman (1994) suggest that reducing excess demand may not always lead to
increases in patient-care expenditures or quality of care.
This study supports the long-held assertion that increasing nursing home bed stock
increases Medicaid nursing home expenditures. This relationship has not been lost on policy
makers, who have been charged with restraining nursing home bed supply in an attempt to control
Medicaid LTC costs (Harrington & Swan, 1988; Feder & Scanlon, 1981; Feldstein, 1988a).
The structure and administration of the Medicaid program is statistically significant at the
.05 level in explaining state variation in Medicaid spending on nursing home care and total
Medicaid spending on LTC for the elderly on a 1,000 elderly population basis. Including the
elderly in the optional Medicaid medically needy program increases the demand for nursing home
care services. This finding is further addressed in the nursing home recipient data section. A
statistically significant, negative relationship is found between the number of optional Medicaid
benefits offered to the aged categorically needy and Medicaid expenditure on nursing home care
and aggregate Medicaid spending on LTC for the elderly on a per 1,000 elderly population basis.
This relationship is mediated through a negative relationship between optional Medicaid benefits
and the number of elderly Medicaid recipients of nursing home care, as such it is discussed in that
section.
The level of administration of the Medicaid program, state versus local, is positive and
statistically significant in explaining total Medicaid expenditure on LTC for the elderly and
Medicaid expenditure on nursing home care for the elderly. This finding is consistent with
88


Schneider (1988) and Buchanan (1991), who found local Medicaid administration to be positively
related to overall Medicaid spending. The principal mechanism for the positive relationship
between local administration of the Medicaid program and Medicaid nursing home expenditure
levels is via local administrative effect on the number of Medicaid recipients of nursing home care
that drives overall expenditure on nursing home care, which in turn drives total Medicaid spending
on LTC for the elderly. This is further discussed in nursing home recipient section.
The per diem Medicaid nursing home reimbursement rate is positively associated with both
Medicaid nursing home care spending for the elderly and total LTC spending for the elderly.
Conventional wisdom and previous empirical evidence (Harrington & Swan, 1987; Davis &
Freeman, 1994) suggest that higher Medicaid per diem rates lead to higher average expenditures
per nursing home recipient and higher overall nursing home and total Medicaid LTC expenditures,
unless number of recipients decline to offset the increase. This study suggests that, when all other
factors are held constant (particularly nursing home bed stock), higher Medicaid nursing home per
diem rates increase both average Medicaid expenditure per nursing home care recipient and number
of Medicaid nursing home care recipients per 1,000 elderly population.
A positive, statistically significant relationship was established between the average LOS
for the state Medicare population in acute-care hospitals and total state Medicaid spending on LTC
for the elderly when New York was included in the analysis only. It appears that New Yorks
spending on home care is driving the finding, as no statistically significant relationship was
established between average LOS of the state Medicare population in acute-care hospitals and state
Medicaid spending on nursing home care for the elderly, with New York either included or
excluded from the analysis. A negative relationship was hypothesized as studies by Shaughnessy
and Kramer (1990) and Liebig (1988) have demonstrated an increase in the number of LTC
recipients and services since Medicares PPS was implemented as PPS decreased the average LOS
89


Full Text

PAGE 1

INTERSTATE VARIATIONS IN LONG-TERM CARE SPENDING FOR THE ELDERLY by Patricia Thomas B.S., Northeastern University, 1978 M.S., Boston University, 1981 M.B.A., University of Vermont, 1988 A thesis submitted to the Faculty of the Graduate School of the University of Colorado at Denver in partial fulfillment of the requirement for the degree of Doctor of Philosophy Public Administration 1996

PAGE 2

1996 by Patricia Thomas All rights reserved.

PAGE 3

' TJiis Thesis for the Doctor of Philosophy degree by Patricia A. Thomas has been approved for the Graduate School of Public Affairs Peter deLeon '= William Atkinson Richard Foster Date

PAGE 4

Thomas, Patricia (Ph.D., Public Administration) Interstate Variations in Long-Term Care Spending for the Elderly Thesis directed by Professor Peter deLeon ABSTRACT In 1987, Medicaid expenditure on long-term care (LTC) differed among states, on a per elderly basis by more than a factor of ten. This variation in state spending on LTC for the elderly holds serious policy implications regarding horizontal, inter-generational, and taxpayer equity. The purpose of this study was to first, determine the factors driving this varied resource allocation and, second, clarify the role of federal, state, and local government policy and administration on state spending on LTC for the elderly. The basic theoretical underpinning for the study is Feldstein and Scanlon's contention that LTC utilization is a function of both demand and supply, each of which is constrained by a state's ability and willingness to fund LTC services. Medicaid expenditure on LTC for the elderly was analyzed, on a per 1,000 state elderly population basis, for years 1981-1988, by service area (nursing home, home care, and the 2176 Waiver Program [1982-1988]) and by components of spending (total expenditure, average expenditure, and number of elderly recipients) Private spending on nursing home care for the elderly was analyzed across states for years 1987-1988. These dependent variables were regressed against state socio-economic and demographic variables alone (reduced model) and, additionally, with state-selected policy (Medicaid program and health system characteristics) and political variables (expanded model). Findings suggest that state socio-economic and demographic variables have a stronger relative effect on average Medicaid expenditure per elderly LTC recipient, while state-selected policy variables have a stronger relative effect on the number of elderly recipients of Medicaid LTC services. Results of this study further indicate that "poor" states increase access to Medicaid LTC services and restrict the benefit package while "rich" states enhance the Medicaid LTC benefit package, not access. Reform strategies that reduce Medicaid per diem nursing home reimbursement rates, optional Medicaid eligibility categories, or target optional Medicaid service may decrease Medicaid iv

PAGE 5

LTC utilization. Reform strategies that further reduce average length of stay for the Medicare population in acute care hospitals may increase Medicaid LTC utilization. Nursing home bed stock remains a powerful state policy variable regarding both public and private spending on nursing home care. This abstract accurately represents the content of the candidate's thesis. I recommend its publication. Signed-----------------Peter deLeon v

PAGE 6

ACKNOWLEDGEMENTS With gratitude and appreciation, I must acknowledge the contributions and support of the following persons: Carol Jacobson and Roger Carver for logistical computer support with statistical analysis; Ellen Fryxell for editorial comments and manuscript preparation; Dr. Thomas Grannemann for generous guidance with conceptual and design development; Dr. William Atkinson, Linda deLeon, Richard Foster, and Sam Overman, members of my committee, for intellectual insights and good will throughout the defense; A very special thank you to Professor Peter deLeon, my chair, mentor, and spiritual guide, without whom this study would not have come to fruition; A final note of thanks to the Thomas family for epitomizing the word. --. .: I am deeply indebted and grateful to each of you. Thank you. Patricia Thomas June, 1996 Andover, Massachusetts

PAGE 7

CONTENTS CHAPTER 1. INTRODUCTION ............................................ Problem Statement . . . . . . . . . . . . . . . . . . . . Dissertation Organization . . . . . . . . . . . . . . . . . . 2. BACKGROUND ............................................ Definition of Long-Term Care ................................. Key Factors Influencing Demand for LongTerm Care . . . . . . . . . History of Medicaid Financing of LongTerm Care . . . . . . . . . . Medicare ......................................... Medicaid ......................................... Medicaid Eligibility . . . . . . . . . . . . . . . . . Medicaid Benefits . . . . . . . . . ; . . . . . . . . The Problems Associated with Medicaid LTC Financing ................. Page 1 2 6 7 7 7 11 11 12 13 14 14 3. LITERATURE REVIEW ....................................... 17 17 The Medical Model . . . . . . . . . . . . . . . . . 17 Theory ................................................. The Cultural Model . . . . . . . . . . . . . . . . . 20 The Economic Model . . . . . . . . . . . . . . . . . The Structural Model . . . . . . . . . . . . . . . . . The Political Model . . . . . . . . . . . . . . . . . The Public Goods Model . . . . . . . . . . . . . . . . The Incremental Budgeting Model . . . . . . . . . . . . . Empirical Work .......................................... Determinants of Geographic Variation in Medicaid 22 25 28 29 31 32 Expenditures and Utilization Patterns . . . . . . . . . . . . . . . 33 Determinants of Geographic Variation in Expenditure and/ or Utilization Patterns Specific to Nursing Home Care . . . . . . . 42 Determinants of Geographic Variation in Medicare Home Health Care Utilization . . . . . . . . . . . . . . . . 49 Summary of Empirical Work . . . . . . . . . . . . . . 50 4. METHODOLOGY ............................................. 54 Model ................................................ Hypotheses . . . . . . . . . . . . . . . . . . . . . . . Independent Variables . . . . . . . . . . . . . . . . . Dependent Variables . . . . . . . . . . . . . . . . . Medicaid Dependent Variables (1981 to 1988) .................. Private Spending Dependent Variable (1987 and 1988) ............. vii 55 56 57 58 59 59

PAGE 8

Page Design . . . . . . . . . . . . . . . . . . . . . . . . 61 Unit of Analysis . . . . . . . . . . . . . . . . . . . . . 62 Data Collection . . . . . . . . . . . . . . . . . . . . . 63 Data Analysis . . . . . . . . . . . . . . . . . . . . . . 66 5. RESULTS . . . . . . . . . . . . . . . . . . . . . . . . 67 Descriptive Analysis . . . . . . . . . . . . . . . . . . . . 67 Dependent Variables . . . . . . . . . . . . . . . . . 67 Independent Variables . . . . . . . . . . . . . . . . . 71 Correlation Coefficients . . . . . . . . . . . . . . . . 74 Regression Analysis . . . . . . . . . . . . . . . . . . . . 79 Dependent Variable . . . . . . . . . . . . . . . . . 79 Model . . . . . . . . . . . . . . . . . . . . . 79 Total Medicaid Expenditure on LongTerm Care for the Elderly and Medicaid Expenditure on Nursing Home Care for the Elderly . . . 80 Elder Medicaid Recipients of Nursing Home Care . . . . . . . . 91 Average Medicaid Expenditure per Elder Recipient of Nursing Home Care . . . . . . . . . . . . . 96 Percent of Medicaid Expenditure on LTC for the Elderly Spent on Nursing Home Care .................... 101 Medicaid Expenditure for Home Health Care for the Elderly .......... 106 Elderly Medicaid Recipients of Home Health Care Services . . . . . 113 Average Medicaid Expenditure per Elder Recipient of Medicaid Home Health Care Services . . . . . . . . 118 Medicaid Expenditure on the 2176 Waiver Program for the Elderly ...... 123 Medicaid Nursing Home Expenditure/Recipient Equations Inclusive of Non-Institutional Medicaid Expenditure/Recipient Data . . . 128 Private Expenditure on Nursing Home Care for the Elderly, Per 1,000 State Elderly Population, 1987 and 1988 ................ 131 Summary Of Findings ....................................... 134 6. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . 143 Policy Considerations . . . . . . . . . . . . . . . . . . . 148 The Future of Medicaid and LongTerm Care . . . . . . . . . . . . 154 Implications for Further Research ............................... 160 APPENDIX ............................................ . .... 163 A. DATA SOURCES Independent Variables . . . . . . . . . . . . . . . . . 163 Dependent Variables . . . . . . . . . . . . . . . . . 164 BffiLIOGRAPHY ............................................... 165 viii

PAGE 9

FIGURES Figure 1.1 Page 0 0 0 0 0 0 0 0 ............. 4 ix

PAGE 10

TABLES Table Page 1.1. Interstate Variation in LongTerm Care . . . . . . . . . . . . . . 3 2.1. Forecast of LTC Utilization by the Elderly ........................... 8 2.2. Income of the Elderly . . . . . . . . . . . . . . . . . . . . 9 4.1. Summary of Hypothesized Relationships ............................. 60 5.1. Descriptive Data of Dependent Variables, By Year, 1981 1988, in 1982 Dollars . . . . . . . . . . . . . . . . . . . . . . 68 5.2. Summary of Descriptive Data of Dependent Variables, Per 1,000 State Elderly Population, By State, 1981 -1988 ............................ 70 5.3. Descriptive Data of Independent Variables, By Year, 1981 1988, in 1982 Dollars . . . . . . . . . . . . . . . . . . . . . . 72 5.4. Summary of Descriptive Data of Independent Variables, By State, 19811988" ... ........................................... 75 5.5. Correlation Matrix .......................................... 76 5.6. State Medicaid Spending on Long-Term Care for the Elderly Per 1,000 Elderly Population, 1981 -1988 ........................... 81 5.7. State Medicaid Expenditure on Nursing Home Care for the Elderly Per 1,000 Elderly Population, 1981 -1988 .......................... 82 5.8. Number of Elderly Medicaid Recipients of Nursing Home Care Per 1,000 Elderly Population, 1981 -1988 ........................... 92 5.9. Average Medicaid Expenditure on Nursing Home Care for the Elderly Per 1,000 Elderly Population, 1981 -1988 .......................... 97 5.10. Percent of Medicaid Long-Term Spending for the Elderly or Institutional (Nursing Home) Care Per 1,000 Elderly Population, 1981 -1988 ............ 102 5.11. Medicaid Expenditure on Home Care for the Elderly Per 1,000 State Elderly Population, 1981 -1988 ................................. 108 5.12. Number of Elderly Medicaid Recipients of Home Care Per 1,000 State Elderly Population, 1981 -1988 ... . . . . . . . . . . . . . . 114 X

PAGE 11

5.13. Average Medicaid Expenditure Per Elderly Recipient of Home Care Per 1,000 State Elderly Population, 1981 1988 ....................... 119 5.14. Medicaid Expenditure on the Medicaid 2176 Waiver Program Per 1,000 State Elderly Population, 1981 -1988 . . . . . . . . . . . 124 5.15. Medicaid Expenditure on Nursing Home Care for the Elderly, Per 1,000 State Elderly Population Basis, 1981 1988, Including Medicaid Expenditure on Home Care and 2176 Waiver Program in the Model . . . . . . . . . . 129 5.16. Elderly Medicaid Recipients of Nursing Home Care, Per 1,000 State Elderly Population, 1981 1988, Including the number of Elderly Medicaid Recipients of Home Care in the Model . . . . . . . . . . . . . . . . . . 130 5.17 Private Expenditure on Nursing Home Care for the Elderly, State Level, 1987 and 1988 . . . . . . . . . . . . . . . . . . . . . . 132 5.18. Summary of Statistically Significant Findings, Inclusive of New York .......... 136 5.19. Summary of Statistically Significant Findings, Exclusive of New York ......... 137 XI

PAGE 12

CHAPTER 1 INTRODUCTION Long-term care for the elderly consumes approximately one-third of all federal and state Medicaid dollars (Meltzer, 1988). Financial pressure is expected to intensify dramatically over the next three decades, especially in light of recent Republican budget reduction proposals. There are important demographic, economic, and psycho-social factors setting the contextual framework for this rise in long-term care (LTC) spending. The essential forecasts are that demand for formal (i.e., paid) LTC services will double in the next 30 years as informal (i.e., unpaid) caregiving is reduced because of geographic mobility of family members and working patterns of female family members. The informal caregiver network currently provides about 75% of all LTC services in this country (Rivlin & Weiner, 1988; Shearer, 1989). Annual expenditure for the remaining 25% of LTC services, i.e., paid services, is estimated at $45 billion in 1985 and $108 billion in 1993 (United States General Accounting Office [GAO], 1988; Burner, Waldo, & McKirsich, 1992). LTC services for the elderly are financed, principally, as either personal out-of-pocket expenses by the aged or their families or by state welfare programs, i.e., Medicaid. Medicaid spending for LTC within and across states varies widely, not only in absolute dollars, but also in the service mix provided (Harrington, Newcomer, & Estes, 1985). The primary purpose of this study is to explain interstate variation in state Medicaid spending on LTC for the elderly. A secondary purpose of this study is to approximate, then explain, interstate variation in private spending on nursing home care for the elderly. 1

PAGE 13

As such, the results of this study are important because they enable policy makers to: 1. Better understand the determinants of long-term care spending and utilization; 2. Identify those variables that are amenable to policy manipulation; 3. Evaluate long-term care spending and utilization patterns relative to state and federal objectives; and, finally 4. More accurately predict long-term care spending and utilization patterns under various policy scenarios. The ability to more accurately predict the consequences of restructuring the LTC financing and/or delivery system is especially relevant as a public policy issue given the recent Congressional proposal to reduce the projected growth in Medicaid spending by approximately 20% over the next seven years (Toner, 1995). Problem Statement In 1988, long-term care consumed about 47% of the national Medicaid budget, with LTC for the elderly consuming approximately 31% of the national Medicaid budget (Meltzer, 1988). These expenditures are particularly impressive when one considers that the elderly composed only 12% of the population in 1988 while the national and global trend is toward an increasingly elderly population (Statistical Abstract of the United States, 1988; Rivlin & Weiner, 1988). Within these national figures lie wide variations in state spending. Given the great latitude states have in setting specific eligibility criteria, benefits, reimbursement policies, utilization controls, and regulation of the local health care environment, variation in Medicaid spending on LTC for the elderly is not surprising. Nevertheless, the degree of variation is impressive. The following chart (Table 1.1) provides information on the kind of interstate Medicaid variation on LTC that is commonplace: 2

PAGE 14

Table 1.1. Interstate Variation in LongTerm Care Elderly as Public LTC Public LTC Spending #Elderly %of general Spending on the on the elderly as in millions population elderly in millions % of State 1985 State (1980) (1980) (1984) Medicaid Budget Alaska .012 3.0% $ 12 40% California 2.400 10.2% $ 588 29% Colorado .250 8.6% $ 88 45% D.C. .075 11.6& $ 38 20% Minnesota .480 11.8% $ 382 69% Source: Pierce and Hansen, 1987 Although this data are comparatively old, similar disparities continue today (Burner et al., 1992). Howell, Baugh and Pine (1988) studied patterns of Medicaid utilization and expenditures in selected states from 1980 through 1984. Average 1981 Medicaid long-term care expenditures per Supplemental Security Income (SSI) enrollee across the five states studied varied widely, from $1,428 per aged enrollee in California to $4,937 in New York, a threefold difference. For disabled enrollees, expenditures ranged from $835 per enrollee in Tennessee to $2,157 in Michigan. Figure 1.1 was compiled using Medicaid data published by the Health Care Financing Administration (HCFA) for fiscal year 1987. It demonstrates that on a per-elderly-person basis, Medicaid payments for both nursing home and home health differ among states by more than a factor of ten. New York state alone accounted for 80% of the national Medicaid expenditure on home care in 1986 (Swan, 1990). These interstate variations in Medicaid spending for LTC for the elderly present the central question, why? What factors explain this great variation, and what are their relative importance? These variations in state spending on long-term care for the elderly hold serious policy implications regarding horizontal, inter-generational, and taxpayer equity. For example, 3

PAGE 15

Figure 1.1 MEDICAID NURSING HOME PAYMENTS PER ELDERLY PERSON 1987 12 lii 8 I.J...' 0 5 6 CD :i 4 2 1 00 200 .300 400 500 600 700 800 900 1 000 PAYMENTS(DOLLARS) MEDICAID HOME HEALTH PAYMENTS PER ELDERLY PERSON 1987 12 1-;! Vl 8 I.J... 0 ffi 6 CD :i 4 2 40 80 120 160 200 240 280 .320 .360 400 PAYMENTS (DOLLARS) 4

PAGE 16

Scanlon and Feder (1984) consider state variation in nursing home spending to be economic discrimination that poses a severe access problem for the disabled elderly most in need of care. Their research indicates that where supply of nursing home beds is most limited, smaller proportions of the most impaired population actually reside in nursing homes, while in states with the highest nursing home bed-to-elderly population ratios, more than 90% of persons most in need (unmarried persons 75 or older, needing assistance in all activities of daily living) were in nursing homes. Real interstate variations in Medicaid spending hold serious access, equity, and efficiency implications for both recipients and taxpayers. Tiiis study will offer empirical evidence regarding the factors driving the varied resource allocation of public and private dollars across states on one service area (LTC) for one group (the elderly). Commerce Committee Chairman, Thomas Bliley, Jr. a Republican Representative from Virginia states (1995), "Medicaid is broken and we all know it. It's a complex, bureaucratic, and uncontrolled entitlement riddled with waste and inefficiency." As such, Medicaid is frequently cited as a state and federal "budget buster" as budgetary controls are considered of limited success because of the open-ended entitlement nature of the program (Wildavsky, 1988). This study is designed to help clarify the role of state-selected policy and political variables on Medicaid spending on LTC for the elderly. Differentiating the influence of various state-selected variables on access, expenditures, and service mix of the state elderly population to taxpayersubsidized LTC services is important to state policy makers, Medicaid administrators, health planners and regulators, and providers. Determining which state-selected variables affect various aspects of LTC spending enables policy makers to align incentives with state and federal objectives regarding long-term care. 5

PAGE 17

Dissertation Organization This dissertation is organized into five remaining chapters, namely, the background of LTC, the literature review, the methodology, the findings of the study, and finally, the conclusions of the study. The next chapter defines then discusses key financing aspects of LTC in the United States. It also summarizes the major problems associated with Medicaid funding of LTC. This background information is critical to the context of LTC. The review of the literature, chapter 3, organizes relevant LTC and Medicaid work around both theoretical and empirical bases. It reflects an inter-disciplinary approach to analyzing state variation in Medicaid spending. The essential assumptions and theoretical arguments upon which this study is based are offered in this section. Chapter 4 presents the method of inquiry used for this study. Critical issues of design, modeling, variable selection, and hypothesized relationships are discussed. The results of the study are outlined in chapter 5. The findings are organized by both descriptive and quantitative analysis. The results of the regression analysis, which is intended to explain interstate variation in elderly Medicaid recipients of LTC and Medicaid expenditure levels, is presented inclusive and exclusive of state-selected variables. The sixth chapter concludes this study with a discussion of the policy implications of the study, the possible future of Medicaid funding of LTC, and the implications of the study for future research 6

PAGE 18

CHAPTER2 BACKGROUND A critical assessment of the results of this study, namely determining the factors that explain state variation in spending on LTC for the elderly, requires an understanding of four inter related background areas: a working definition of LTC; the key factors influencing demand for LTC; the history of Medicaid financing of LTC; and the main problems associated with Medicaid financing of LTC. This background information will be drawn upon extensively in discussing both the results of this study and its implications for further research. Definition of Long-Term Care Long-term care is defined as the range of health, social, emotional, and personal care services needed by functionally disabled individuals who require assistance in performing necessary daily activities, i.e., eating, bathing, toiletry, dressing, or mobility. These activities are appropriately called activities of daily living (ADLs). Long-term care services also encompass instrumental activities of daily living (IADLs) such as housekeeping, shopping, meal preparation, laundry and household chores (Pierce & Hansen, 1987). Key Factors Influencing Demand for Long-Term Care Many psycho-social, economic, demographic and epidemiological factors influence the demand for LTC services. Demographic projections indicate that the number of elderly (age 65 or older) in the United States will grow from 31.3 million persons in 1986 to 50.3 million in 2016, an increase of 7

PAGE 19

61%. The over-85 population is expected to grow from 3.5 million in 1986 to 5.9 million in 2000, to 7.2 million in 2020. It is the over-85 age group that is most important to the LTC issue, since 22% of this population currently reside in nursing homes and 35% of the remaining noninstitutionalized group require personal care assistance (Rivlin & Weiner, 1988). Coupling these demographic projections with current LTC utilization, epidemiological, and medical effectiveness patterns provides the following forecast of LTC usage, drawing a distinction between nursing home and home-care services (Table 2.1) Table 2.1. Forecast of LTC Utilization by the Elderly Nursing Home Services Year 1990 Year 2005 Year 2020 Age Number* % Population Number % Population Number % Population 65-74 404 18 390 12 .659 17 75-84 913 40 1.216 37 1.295 32 85 & over 968 42 1.666 51 2.067 51 Total 2.285 100 3.272 100 4 021 100 Home Care Services Year 1990 Year 2005 Year 2020 Age Number* % Population Number % Population Number % Population 65-74 1.235 31 1.306 24 1.840 29 75-84 1.715 43 2.295 43 2.230 37 85 & over 1.059 26 1.803 33 2.190 34 Total 4.009 100 5.404 100 6.359 100 Numbers in millions Source: Rivlin and Weiner, 1988 8

PAGE 20

Utilization of formal LTC services is expected to approximately double in 30 years. Financing this increase in LTC utilization is expected to be an even greater real and relative burden, as the LTC inflation rate will probably continue to outstrip the consumer price index (CPI) by about 2 percentage points (Rivlin & Weiner, 1988). Financially, the elderly are a heterogenous population. The already wide gap between the median income of the "young" elderly (under 75) and the "old" elderly (over 85) is expected to grow dramatically over the next 30 years: Table 2.2. Income of the Elderly 1987 2016* Median Family Income of "Young" Elderly $10,806 $20,303 Median Family Income of "Old" Elderly $6,837 $7,999 Percent "Old" to "Young" Median Family Income 63% 40% in constant 1987 dollars Source: Rivlin and Weiner, 1988 It is the "old" elderly who are at the highest risk for needing LTC services and who have the most limited income. Therefore, they are the population most likely to continue to rely on public monies for paid LTC services (Rivlin & Weiner, 1988). Acute-hospital care cost-containment practices blossomed in the 1980s in response to spiraling acute-care costs. The principal acute-care cost-containment mechanism is a diagnosisrelated payment scheme that pays a set fee to the hospital based on a patient's primary diagnosis. This prospective reimbursement scheme is widely known as diagnosis-related groupings (DRGs). DRGs were designed to create a financial incentive for hospitals to increase efficiency, decrease utilization, and decrease length of stay of the Medicare population in acute-care hospitals. DRGs 9

PAGE 21

have been associated with increased utilization of home and community and nursing home post hospital care. Harlow and Wilson (1985) found a 196% increase for in-home skilled nursing and a 63% increase in personal-care services in the year following the implementation of DRGs. The expenditures associated with post-hospital care for the elderly are felt, predominantly, by the Medicare program although an effect on private payers has been demonstrated (Morrisey, Sloan & Valvona, 1988). It is unlikely that acute-care cost-contaimnent measures will weaken in the near future. Shortened hospital stays for the elderly may ultimately lead to increased demand for LTC. The bulk of unpaid LTC services are provided by family members, usually spouses, daughters, or daughters-in-law. At present, approximately 75% of LTC service is provided by this informal network. It is estimated that 27 million days of unpaid care are provided each week at untold emotional, physical, and financial hardship to the caregivers. These costs are considered the "hidden" costs of LTC in our current system. Changes in the size of families, the structure of families, geographic mobility of family members, working patterns of female family members, and the general health of adult children with "old" elderly parents may significantly reduce the ability of this network to provide assistance. The diminished ability of the informal caregiver system to provide unpaid LTC services for the elderly will fundamentally alter financial projections of required public monies to pay for Medicaid LTC recipients. In comparing the relative mix of LTC services paid for with Medicaid funds (i.e., home care versus community care versus institutional care), it is important to note that both the American elderly and the younger population have expressed a preference for receiving LTC in their home as opposed to a nursing home (National Long-Term Care Survey, 1982 for the attitude of the elderly; Harris Poll, 1988 for the non-elderly). Despite this strong preference, only 25% of the disabled elderly receive any funded in-home services (DHHS, LTC Survey, 1982). 10

PAGE 22

The role of private-sector initiatives in financing LTC for the elderly, particularly private sector LTC insurance, is at present highly speculative. Currently, less than 1% of LTC services are paid with private insurance. The Health Insurance Association of America (1991) reported that, as of December 1989, more than 1.5 million long-term care insurance policies had been purchased. This figure represented an increase of 36% over 1988. Moreover, the prospect of rapid saturation of private LTC insurance policies has been considered slim until such issues as price, policy restrictions, and screening mechanisms are resolved (Rivlin & Weiner, 1988). However, as Zedlewski, Raffaty and McBride (1991) state: "Ultimately, the viability of private financing for long-term care will hinge on politics that reduce the cost of long-term care insurance and encourage voluntary purchase of insurance among the elderly." The recent GOP Congressional budget proposal (November, 1995) aimed at reducing the projected growth in Medicaid spending-b:fi8 to 20% over the next seven years includes a tax reform amendment allowing individuals to deduct premiums paid for private-sector long-term care insurance from their taxable income. This tax reform measure could catapult the private-sector LTC insurance industry into the 21st century regardless of current flaws in the product. History of Medicaid Financing of Long-Term Care It is important to clarify the financing role of Medicare and Medicaid in the current LTC system in the United States. Medicare Medicare is a federally funded and administered health-insurance program enacted into law in 1965. It provides acute-medical care primarily to persons over the age of 65. Medicare Part A covers hospitalization, skilled nursing home care for upwards of 100 days, and intermittent home-11

PAGE 23

care services. Skilled nursing home care must be considered medical follow-up to an acute-care episode and be ordered by a physician to be covered by Medicare (Pierce & Hansen, 1987). In 1993, 4% of Medicare dollars were spent on skilled nursing home care (HCFA, Office of the Actuary, 1995). Medicare Part B is a voluntary, supplemental program that covers physician services, lab services, and outpatient services. All services have limits and deductibles. No long-term or chronic-care service provisions exist within Medicare. The recent Republican Congressional proposal to balance the federal budget in seven years calls for a reduction in the projected growth in Medicare spending by $270 billion. This proposal would encourage the elderly to enroll in private health-care plans, such as health-maintenance organizations or new medical savings accounts. The plan would also have Medicare beneficiaries pay 31.5% of their optional Part B premium instead of having their share drop to 25% under current law, and the proposal would tighten payments to doctors, hospitals, laboratories, and nursing homes to keep spending within limits (Zuckerman, 1995). Medicaid Medicaid is a federal-state welfare program enacted into law in 1966. It provides a wide range of hospital, physician, and long-term care services for low-income recipients (Pierce & Hansen, 1987). Medicaid was built upon the historic role state and local governments played in providing health care to the poor (McDevitt & Buczko, 1985). From Medicaid's conception in 1966, states were given great latitude in structuring their programs within broad federal guidelines on mandated eligibility and benefits. States receive federal financial participation for both mandatory and optional services. By the early 1980s, economic and political forces were placing serious fiscal constraints on Medicaid. 12

PAGE 24

These forces led to the passage of the Omnibus Budget Reconciliation Act (OBRA) of 1981, and the Tax Equity and Fiscal Responsibility Act (TEFRA) of 1982 (Harrington et al., 1985). One purpose of these acts was to control program expenditures by reducing the level of federal financial participation in Medicaid. Expenditures for health-care services continued to spiral upward even as federal contribution declined and state revenues were reduced by the economic slowdown of the early 1980s (Buchanan, 1987). OBRA and TEFRA encouraged states to control Medicaid costs by offering them even greater discretion in determining eligibility criteria, benefit packages, and method of provider reimbursement (McDevitt & Buczko, 1985; GAO, April, 1987). The Federal Government currently matches state funds by 50% to 78% depending, primarily, on the state's per capita income (Jazwiecki, 1989). Section 2176 of OBRA allowed for new home and community-based waiver programs, which call for experimental spending on home and community services for Medicaid recipients who qualified for nursing home services. These services are designed as alternatives to nursing home care. In 1982, six states participated in the 2176 Waiver Program. By 1989, 46 states were participating in the program (Miller, 1990). Medicaid Eligibilitv There are three types of eligibility for medical assistance under Medicaid. One is mandated by federal law, while two are state options. The federally mandated and primary way to qualify for Medicaid is to be "categorically needy." People who meet income, resource, and asset standards for any one of several income maintenance programs, i.e., SSI, State Supplemental Payment (SSP), and Aid to Families with Dependent Children (AFDC), automatically qualify for Medicaid. Typically, low-income aged, blind, and disabled persons are eligible for Medicaid under 13

PAGE 25

SSI and/or SSP, while low-income families with children and deprived of the support of at least one parent are eligible under AFDC (Harrington et al., 1985). Each state also has the option of providing Medicaid coverage to a number of other groups. Two of the major optional coverage groups are "medically needy" individuals, (i.e., those who do not qualify by income, but have medical bills that would reduce their income below a medicaliy needy maximum), and individuals who would be income eligible for public assistance if they applied. Medicaid Benefits The benefits covered by state Medicaid programs are also categorized as either mandated or optional. Federal guidelines mandate that certain basic medical services be covered. They include physician, inpatient and outpatient hospital services, laboratory and X-ray services, skilled nursing facilities services for those 21 years of age or older, home health care, family planning, rural health clinics, and early and periodic screening (Harrington et al., 1985). Additionally, states have the discretion of providing any one of 32 optional services. Optional services include chiropractic, dental, eyeglasses, podiatry, and prescription drugs. Elective long-term care services also include intermediate-care facilities, adult-day care, and case management. The Problems Associated with Medicaid LTC Financing There are at least five major problems associated with the use of Medicaid as a major LTC funding source. It is imperative that these issues be specifically addressed in any comprehensive evaluation of Medicaid spending for LTC. 14

PAGE 26

As of 1988, a full half of the Medicaid recipients of nursing home services were private pay upon admission into the facility. The costs of long-term care reduced these people to poverty level. Most researchers and politicians agree that long-term, care-induced poverty is undesirable. Second, Medicaid financing of LTC perpetuates inequities in the medical-care delivery system. A two-class system of chronic health care for the elderly has developed around the patient's ability to pay. Medicaid-dependent persons are severely restricted in nursing home selection as a direct result of Medicaid reimbursement policies, and are generally considered to be receiving lower quality of care (Rivlin & Weiner, 1988). Third, access to even low-quality nursing facilities is difficult for Medicaid-dependent elders. Lengthy waiting lists for nursing home placement are common, with an average occupancy rate of 92% in 1985 (Rivlin & Weiner, 1988). The less financing an elder carries, or the greater the need for serviCes, "the less likely that placement will occur. Fourth, Medicaid financing of LTC, as it is currently structured, creates a bias toward institutionalization even when other options are still viable, simply because financing is more widely available for nursing home care than preferred home or community care. Fifth, with one-third of the national Medicaid budget spent on LTC for the elderly, Reutzel (1984) suggests that the Medicaid program has become a supplement to the Medicare program in providing health services to the aged. This LTC financing mechanism may undermine the essential public purpose of the Medicaid program, namely, to provide an assurance of access to mainstream medicine to the nation's poor (Holahan & Cohen, 1987). Finally, since Medicaid budgets are constrained (severely so in some states), the disabled elderly are competing with poor families for limited Medicaid dollars. The issue of inter generational equity becomes a major concern when reviewing the allocation of Medicaid funds. 15

PAGE 27

The recent Republican proposal to eliminate the federal deficit in seven years would reform Medicaid via "block grants" to the states. The program has been dubbed "MediGrant" by the GOP. The proposal would end the entitlement of the poor to health care but would require states to spend "set aside" funds for health care to serve pregnant women, poor children under the age of 13, and the disabled. The proposal currently preserves most federal standards for nursing homes and allows citizens to deduct premiums for private-sector long-term care insurance from their taxable income. The proposal is projected to save approximately $163 billion from future growth in Medicaid spending over the next seven years (Zuckerman, 1995). While states' response to "MediGrant" is expected to vary widely should the program be enacted, the general consensus among LTC analysts is that fewer services would be available for the disabled who require long-term care. While MediGrant may not be the answer, it is imperative that Medicaid LTC budgets be systematically evaluated in terms of efficacy, equity, and efficiency before the coupling of increased demand and reduced informal caregiving bankrupt an already-strained system. 16

PAGE 28

CHAPTER 3 LITERATURE REVIEW The literature review is presented in two sections, namely, theory and empirical work. The first section, theory, discusses fundamental theoretical bases for modeling state variations in long-term care spending. It is conceptual in nature. The second section, empirical work, presents a review of the methodologies and findings of research specific to variations in Medicaid and/or long-term care services. It positions the dissertation relative to the field. Theory A comprehensive review of the LTC literature supports demographic, socio-economic, and political models for predicting utilization and expenditure levels for LTC services. Any serious attempt to explain state Medicaid spending on LTC for the elderly represents an integration of the "purist" models presented separately in this section of the literature review for reasons of clarity. Seven specific models related to public expenditures on LTC will be reviewed. They are: the medical model, the cultural model, the economic model, the structural model, the political model, the public-goods model, and finally, the incremental budgeting model. The Medical Model The normative goal of health care in the medical model is to cure disease. As such, the medical model predicts that spending on LTC services for the elderly depends on the level of "disease" in a given population. 17

PAGE 29

Disease, in the acute-care medical model, is defined as any dysfunction of an organ, organ system, or structure, while health is defined as the absence of disease (Caplan, 1988). Talcott Parson, in his discussion on the cultural and social dimensions of disease, defined the "sick" role as one where four conditions are met: namely, that a person is incapable of curing himself; a person is exempt from usual roles and duties; a person wants to get better; and a person wants to seek help from experts and receive treatment (Parsons, 1981). Characteristics of the medical model, as it applies to LTC, include a prerequisite physician order for initiation of LTC services, and a plan of treatment for LTC services to be supervised by medical personnel while payment is made directly to providers. LTC services are perceived as a "health-care benefit" as recipients of such services assume the patient or "sick" role (Zola, 1988). Clearly, chronic care and disability do not fit well into the medical model of disease, although it is certainly the model upon which long-term care has been built (Caplan, 1988). According to Evashwick (1988), the ideal system of care for those who have multiple, multifaceted, and chronic illness is one that provides comprehensive, integrated care on an ongoing basis and offers various levels of intensity of care that change as a client's needs change. The goal is to provide the medical and support services required to enable the person to maximize functional independence. This contrasts with the goal of acute care, which is to "cure" the patient of illness. To reiterate, the medical model predicts that Medicaid spending on LTC services for the elderly is a function of the level of "disease" in the Medicaid population. Researchers have operationally defined disease in chronic care in one of two ways, namely, by the presence of pathophysiology (illness) or by limitations in function. Buchanan (1987) hypothesized that states with high rates of incidence of arteriosclerosis, heart disease, and cerebrovascular disease would have higher utilization levels of nursing home care. Buchanan logically assumed that the death rates published for each of these disease 18

PAGE 30

conditions were reliable indices of the prevalence of each disease in the living population of the state. Buchanan concluded that differences among states in the incidence of these diseases explained little, if any, of the variance in the utilization of skilled nursing home beds by Medicaid patients.1 Conversely, differences in these disease incidence rates explained upwards of 20% of the variance in the utilization of intermediate nursing home beds by Medicaid patients. Many more researchers, e.g., Henry, Morreale, Dunlop, Chiswick and Scanlon, have focused on the functional limitations of the LTC recipient, using age as a proxy variable for intensity of disability (Scanlon, 1980). Certainly, illness and functional limitation are related, as both increase with age, but they are not conceptually identical. Manton and Soldo (1985) estimated that 2 3% of persons under the age of 19 are limited in performing activities of daily living due to chronic illness, wliile 22% of persons 55-64 years of age are limited in ADL to some degree, and 36% of persons 74 years of age and older are limited in ADL while another 3.5% in this age group are limited in minor or instrumental activities of daily living. The current estimate of persons requiring continuing care due to functional limitations is more than 8 million, with approximately of these people over the age of 65. This number represents approximately 3.5% of the general population of the Unites States (Weissert, 1983). This age/disability factor is reflective of the "need" for LTC services. Scanlon defines the "need" for LTC services as a value judgment made by someone other than the LTC consumer 1 Skilled nursing home beds are in facilities with an organized, professional staff who provide medical, continuous nursing, and various other health and social services to patients who are not in an acute phase of illness, but who require primarily rehabilitation or skilled nursing care on an inpatient basis. Intermediate care nursing home beds are provided to individuals who require health related care and services of a more custodial nature (Pierce & Hansen, 1987). 19

PAGE 31

about the benefits different individuals will receive from care, or their ability to function without it. As such, the "needs of chronically disabled persons have traditionally been assessed and addressed by the medical community via an acute-care medical model. But a recent trend in medical care for chronic disease toward the enhancement of health-related quality of life (HQOL), rather than the cure of disease or increased survival, has emerged (Revicki 1989). This acute-care model leads one to posit that the number of elderly in the general population will positively relate to aggregate LTC spending on the elderly, while the number of persons 85 or older will positively relate to LTC spending on nursing home services for the elderly. The Cultural Model The social, racial, and ethnic mix of a community may differentially affect Medicaid LTC spending. The majority of the LTC Medicaid budget is used for nursing home care Researchers studying the probability of institutionalization for disabled elderly have identified living alone as an important factor increasing the use of nursing home care. Research further indicates that higher income, white, native-born elderly are more likely to live alone whereas non-white females and foreign-born males are less likely to live alone across all levels of income and disability (Bishop, 1988; Macken, 1986). Therefore, white, native-born elderly have been identified as more susceptible to nursing home placement once disability occurs due to their propensity for living alone, or, on the other hand because of their more distant family ties. Conversely Scanlon (1980) has determined that when one takes into account differences in the availability of beds between areas with high and low concentrations of black elderly, blacks were likely to demand more nursing home care than whites. This finding is consistent with the 20

PAGE 32

greater incidence of functional impainnents and chronic illness among the black elderly. Scanlon attributed black persons' lower utilization of nursing home beds to their concentrations in states unwilling to fund greater utilization of nursing homes. Accordingly, some of the unsatisfied demand for nursing home services represents variations in access of elderly persons to nursing home care because of interstate differences in the ability and willingness of state governments to finance this care. Economic, cultural and political state influences are interwoven in Scanlon's (1980) discussion of nursing home utilization. He contends that cultural norms (e.g., living arrangements with extended family) influence the demand for formal LTC for the elderly, while economic and political factors (e.g., tax base and political support for redistribution of funds toward the poor and elderly) influence the supply of state subsidized LTC services for the elderly. The cultural model leads one to conclude that: the number of white elderly and single family households in the elderly population will positively relate to LTC spending, especially nursing home spending; and that the number of persons per household may inversely relate to number of LTC elderly recipients of nursing home and/or home and community services as infonnal care substitutes for formal (paid) care; and, ultimately, that the number of persons per household may negatively relate to expenditures on nursing home and/or home and community LTC services if recipients are served by the infonnal, as opposed to the formal, caregiving network Scanlon's contention that demand is constrained by a state's ability and willingness to fund LTC for the elderly suggest that state economic indicators of ability to pay i.e., average state income and state employment, and indicators of state political will to fund LTC for the elderly will each relate positively to LTC spending. 21

PAGE 33

The Economic Model The relatively pure economic model of Medicaid expenditures for LTC, as expressed by Paul Feldstein (1988a), is a function of demand and supply. Demand for LTC services is related to the level of disability in the population as well as to economic factors. Once again, the level of disability in the Medicaid population is considered indicative of the "need" for LTC services. The economic factors that affect the demand for LTC services include out-of-pocket prices to the consumer, income of the elderly, out-of-pocket prices for substitute services, and caregiving by the informal network. The out-of-pocket price to the consumer influences the demand for LTC services as the demand for nursing home care has been shown to be price elastic, with price elasticities ranging from -1 to -2.3 (Rivlin & Weiner, 1988). The price elasticity for home care is also considered to be highly elastic, although no specific estimate could be found. Medicaid has dramatically lowered the out-of-pocket price of nursing home care, while leaving the price of alternatives unaffected. The income of the elderly strongly affects the demand for paid LTC services since LTC is considered a "normal" economic good in that consumption has been demonstrated to increase with disposable income (Feldstein, 1988a). The indigent disabled elderly, with significantly less disposable income, are more vulnerable to institutionalization as institutional care is subsidized to a much greater extent than alternative care. The supply of informal LTC services (unpaid and typically rendered by family) is considered a substitute for formal (paid) LTC services. The higher the opportunity cost of informal care to the provider of such services, the greater will be the demand for paid and government subsidized services (Chiswick in Feldstein, 1988a). 22

PAGE 34

Utilization of formal LTC services is influenced by both demand and supply. The supply of "paid" LTC services depends upon the price received for services, the cost of inputs, and government regulations. In a competitive market, one would expect the supply of paid LTC services to increase as the price received for such services increases, all other factors held constant. The predominance of government as "fixed" payor for institutional LTC services immediately links the effects of supply received for nursing home services to government regulation on the price of LTC services. Feldstein (1988a) discusses the effects of the government's monopsonistic power over rate setting and its effects on private pay and Medicaid residents. Feldstein states that aggregate nursing home capacity (supply of beds) and rates for Medicaid patients are heavily regulated by the government. Nursing homes charge a higher price to private-pay patients as compared to both Medicaid patients and private-pay patients iii an unregulated nursing home environment because the demand for both private pay and government-subsidized nursing home beds continues to outstrip supply. The supply of nursing home beds is tightly regulated by state governments, especially in states with active certificate-of-need (CON) legislation. This surcharge on private-pay residents -causes a redistribution of income from private-pay persons toward the nursing home, resulting in economic profits accruing to the nursing home investors. This resulting economic profit, which accrues because of government-constrained supply, allows inefficient nursing homes to survive. This differential in. price between private pay and Medicaid patients also causes access to nursing home care to be reduced for Medicaid patients, especially heavy-care Medicaid patients. The "unmet need" of heavy-care Medicaid patients, arguably the most "needy," is enlarged while the informal caregiving network is further burdened. Feder and Scanlon (1981) contend that Medicaid programs attempt to control spending for nursing home care by limiting the supply of beds. Higher per diem rates were consistently linked 23

PAGE 35

to lower utilization of nursing home services by Medicaid-eligible persons. This is an unlikely outcome if providers were the true obstacles to access to Medicaid LTC beds. One would expect unregulated (i.e.,"rational") economic providers to supply more beds as the payment increased in an attempt to maximize profits Harrington and Swan (1987) found nursing home bed supply to be the dominant factor affecting the number of Medicaid recipients and Medicaid expenditures per total aged population They contend that lower nursing home bed supply will lower utilization of Medicaid nursing home services and overall Medicaid costs. They also point out that lowering nursing home bed supply without providing alternative LTC services may only increase problems of access and quality of care The supply of LTC services may vary geographically with variations in the cost of inputs, e.g., land, labor, local inflation, and supplies. Individual state government regulations regarding standards of care, personnel requirements and safety requirements differentially affect the cost of inputs in the LTC industry (Paringer, 1983) It is important to note that the utilization of LTC services, both private pay and government subsidized, depends upon the decisions of consumers, providers, and government. In reference to the LTC market, Feldstein (1988a) considers that state government regulation is represented by the number of nursing home beds, the standards of care, and the Medicaid reimbursement policies. In summary, the economic model concludes that: median income of the elderly will positively relate to private expenditures on LTC; the supply of nursing home beds will positively relate to aggregate LTC spending and to nursing home spending; and the cost of inputs (labor, cost of capital, etc.) will positively relate to private and public expenditures on LTC across all services areas. 24

PAGE 36

The Structural Model The structural model states that Medicaid spending on LTC for the elderly is a direct result of the structure of the Medicaid program. Once again, the models display a degree of overlap and integration. The economic model, which was discussed immediately above, contains elements of both the medical and structural models for explaining variation in LTC spending for the elderly. The structures of state Medicaid programs differ regarding eligibility criteria, benefits, reimbursement rates and methods, and methods of limiting utilization. The various state reimbursement policies are rather complex with many dimensions. State methods differ in the cost components allowed, ceilings on cost components, and in consideration of such factors as respective property costs, inflation rates, profit rates, and state licensing standards (Harrington & Swan, 1984). States with reimbursement systems other than purely retrospective have been correlated with lower nursing home rates, and lower expenditures per person, although causation has not been determined (Harrington & Swan, 1984). Scanlon (1980) states that there is a three-part objective in designing a Medicaid reimbursement system for nursing home services. Reimbursement should secure access to the service for eligible persons, encourage high quality, and minimize costs. Tradeoffs among the three are required since a system that promotes one of these goals is likely to reduce the likelihood of achieving the others, especially affecting the third. States may also modify service-delivery patterns and costs through incentives and disincentives involving such factors as occupancy rates, quality of care, and case mix (Harrington & Swan, 1984). McDevitt and Buczko (1985) determined that states with the optional medically needy eligibility program directly influenced Medicaid spending on LTC by increasing the number of eligible users. 25

PAGE 37

Harrington and Swan (1987) found that state policy had major effects on Medicaid nursing home expenditures and utilization. They determined that the reimbursement rate to nursing homes providing custodial care (called intermediate-care facilities or ICFs) was the strongest predictor of expenditures per recipient for all Medicaid recipients. It is interesting to note that researchers consider the effects of state Medicaid program controls on the utilization of health services and expenditures to be complex. State controls of the Medicaid program may affect eligible groups differentially, and the net effect of the program control may be the opposite of that intended (Mauskopf, Rodgers, & Dobson, 1985). Reutzel states that manipulation of some Medicaid program decision variables have counter-intuitive results because of complex interactions among individual behavioral and managerial variables. Physician reimbursement rates are a good example. Policy makers may decrease physician Medicaid reimbursement rates in an attempt to reduce Medicaid expenditures. The unintended consequence may be that physicians with excess time or decreased income increase services in an attempt to reach a personal-target income while Medicaid recipients seek out more expensive alternatives (hospital emergency room rather than physician office visits), thus increasing overall state Medicaid expenditures. Decreasing the number of optional Medicaid services may also lead to counter intuitive results. While it is expected that reducing optional services will decrease Medicaid expenditures, recipients may substitute more expensive covered services for now lost and less expensive optional services, thus driving up overall Medicaid expenditures. Soumerai, Ross Degnan, Avom, McLaughlin and Choodnovskiyet (1991) found that limiting Medicaid reimbursement for effective drugs put frail, low-income elderly patients at increased risk for more costly institutionalization in nursing homes, thus increasing the overall Medicaid expenditures. The structure imposed by the federal government on state Medicaid programs may be another important structural determinant of state LTC spending. Federal policies regarding 26

PAGE 38

eligibility and coverage are clearly important determinants of state Medicaid spending. Home and community service waivers (the 2176 Medicaid program) must be approved by the HCFA, a federal body, in accordance with federal guidelines. The level of federal financial participation (i.e., the matching rate) state Medicaid programs is generally held to have an effect on state generosity. The U.S. General Accounting Office Report entitled "Medicaid: Interstate Variations in Benefits and Expenditures" (1987) found the matching rate to be an important determinant of Medicaid expenditures. The report further concluded that increasing the matching rate was not sufficient incentive to encourage very restrictive states to broaden their Medicaid programs. This may be due to the fact that Federal Medicaid Matching Rates have proven to be insufficiently weighted toward low-income states to overcome the effects of interstate differences in fiscal capacity and poverty rates (Grannemann, 1979). Howell et al. (1988) state that the federal matching rate may-have less impact on state Medicaid programs than state political philosophies regarding health care for the poor. The structural model suggests several relationships. They are: that retrospective reimbursement positively correlates to average spending per nursing home recipient and to total nursing home spending; that Medicaid generosity regarding eligibility criteria will be positively related to aggregate LTC spending and number of recipients across all services; that Medicaid generosity regarding optional benefits may behave in a counter-intuitive manner and relate negatively to nursing home and aggregate LTC spending; that nursing home reimbursement rates positively relate to average expenditure per recipient and aggregate spending across all service areas; and, finally, that the Federal Medicaid Matching Rate may positively relate to spending across all LTC service areas. 27

PAGE 39

The Political Model The political model states that a state's Medicaid expenditure on LTC for the elderly is a function of that state's political climate. The political climate of a state can be defined by that state's position on the conservative/liberal spectrum (Buchanan, 1987). Conservative and liberal indices were developed by Buchanan for each state using annual ratings of each state's delegation to the U.S. House of Representatives by various political organizations. The liberal index was the statewide average of the ratings given by the Americans for Democratic Action (ADA) and the AFL-CIO's Committee on Political Action (COPE). The conservative index was an average of the ratings given by the Americans for Constitutional Action (ACA), the National Association of Businessmen (NAB), and the Chamber of Commerce (COC). The final state score was an average of the liberal and conservative indices. Barrilleaux and Miller (1988) simplified this methodology by using the statewide average of the ratings given only by the ADA as indicative of the liberal index of the state's political ideology. Buchanan found clear and consistent patterns when the political climate of a state was correlated with the Medicaid per diem payment rate for nursing homes and the state-utilization rate for Medicaid-certified nursing home beds. A more conservative political climate was linked to lower Medicaid per diem rates, while a more liberal political climate was linked to higher Medicaid per diem rates. This correlation may be spurious in that more liberal states (Massachusetts and Connecticut ranked first and second in 1985) may have a higher cost of living necessitating a higher Medicaid reimbursement rate. Buchanan did find that differences in the cost of living among states statistically explained upwards of 60% of the variance in Medicaid payments for skilled and intermediate nursing home care (an economic index). The political indices explained upwards of 20% of the variance in the Medicaid-certified nursing home beds per 1,000 28

PAGE 40

elderly ratio for skilled care. Barrilleaux and Miller (1988) developed a market-based model for explaining variations in states' Medicaid policy decisions. They consider measures of supply realized demand, and policy to be endogenous factors in the model, while measures of states' social (demographic), economic, and political characteristics are exogenous variables within the system. The authors claim that socio-economic and political characteristics of states are complementary in their influence on policy outcome. It is worth noting that the authors argue that state political characteristics reflect economic development and that economic development is a catalyst for political and policy development. Carrell (1990) also found political liberalism to affect positively state expenditure per recipient of LTC and total Medicaid expenditure on LTC. The political model leads one to observe that a more liberal state political climate will relate positively to per diem nursing home rates, average expenditures per recipient across ail services areas, number of LTC recipients across all service areas, and total spending across all service areas. The Public Goods Model The median voter theory of political decision-making states that politicians court the median voter, and once elected, allocate resources to meet the needs of the median voter. Godwin and Shepard (1976) argue that the political process translates median-voter desires into public policy, frequently financing services as public goods. Orr (1976) applied this theory to the AFDC program and Grannemann (1979) extended it to Medicaid, viewing ... voter characteristics, preferences, and environment as the ultimate determinants of public expenditures; while the political system established in a state is only a [voter selected] means of attaining desired policies. 29

PAGE 41

The public goods model of interstate variation in Medicaid programs, as specified by Grannemann, explains the number of Medicaid recipients and average benefits per recipient. States that contain relatively few poor persons, e.g., Minnesota and Wisconsin, offered high benefit packages; states with many poor persons and a limited tax base, e.g., Mississippi, offered low benefit packages; while states with many poor persons and a good tax base, e.g., New York and California offered generous benefits with less-restrictive eligibility requirements. Demographic and economic conditions were reflected in the median voter's preferences that translated into public expenditure decisions If one assumes that politicians are ratiorui.l actors and are acting to maximize their political support and chances for re-election in accord with the median voter theory, and one assumes democratic participation via voting on the part of those with strong attitudes toward public spending on LTC for the elderly (i.e., the elderly who represent a large and growing segment of the population and potential voting bloc) one would expect congruence between the attitudes and values of the community and U.S. House of Representatives voting records (Feldstein, 1988b). It is empirically unclear to what extent the values of the community have actually been incorporated into Medicaid program characteristics and into Medicaid spending. According to Emmett Keeler and Robert Kane: Our treatment of our elderly is a mirror of our society Values lie at the very heart of long-term care. Our willingness to transfer income from one generation to another, our valuing of prolongation of dependent life often filled with pain and loneliness, our preferences for encouraging maximal independence at the risk of untoward consequences-all these decisions and others like them may tell us more about ourselves than about the elderly. It is a lesson we may not be eager to learn but a responsibility we cannot shirk (Keeler & Kane, 1982). To the extent that the public goods model holds true, median voter attitudes, their concomitant values, and aggregate preferences are reflected in LTC expenditure measures within 30

PAGE 42

the constraints of economic realities, inclusive of federal subsidies. This is consistent with the findings of Orr (1976), Grannemann (1979), and Buchanan, Cappelleri, and Ohsfeldt (1991) who found political variables to be statistically insignificant regarding Medicaid expenditures once underlying state economic and demographic factors relevant to the voting population were controlled for in the analysis. The public goods model leads one to suspect that economic conditions, demographic characteristics of the voting population, and the effective price of subsidized LTC services to taxpayers (net of federal matching dollars) are the ultimate determinants of LTC spending, number of recipients, and average expenditures per recipient across all service areas. Other factors, i.e., political philosophies, may relate positively to LTC expenditures, but are statistically insignificant when economic, demographic, and federal policy constraints are controlled for in the analysis. The Incremental Budgeting Model The incremental budgeting model suggests that current year Medicaid expenditure levels is dependent upon previous year's Medicaid expenditure level (Wildavsky, 1988). (1988) reported that in 1987 the Office of the Budget, Ohio Department of Human Services, surveyed state Medicaid officials across the nation to identify how Medicaid policy decisions were made. She states that of those who responded to the survey, 90% indicated that previous year's expenditure level played "an extremely important role in program decision making." Both Schneider (1988) and Buchanan (1991) incorporated previous year's Medicaid expenditure level into their model for explaining current year's Medicaid expenditure level. Each found previous year's Medicaid spending to be statistically significant in explaining current year's Medicaid spending. This finding is consistent with theoretical work on incremental policy making 31

PAGE 43

as presented by Charles Lindblom, in "The Science Of Muddling Through" (1959), where he states: It is a matter of common observation that in Western democracies public administrators and policy analysts in general do largely limit their analyses to incremental or marginal differences in policy that are chosen to differ only incrementally. They do not do so, however, solely because they desperately need some way to simplify the problem; they also do so in order to be relevant. Democracies change their policies almost entirely through incremental adjustment. Policy does not move in leaps and bounds. (page 81) Aaron Wildavsky (1988) argues that policy makers lack control over Medicaid expenditures because of the open-ended, entitlement nature of the Medicaid program. Thus, it follows that incrementalism per se as a hypothesis about budgetary behavior can be neither accepted nor rejected by examining total Medicaid expenditures (Lurie and Wyckoff, 1989). The incremental budgeting model posits that previous year's Medicaid expenditure level is strongly associated with current year's Medicaid expenditure level. Interpretation of this correlation is fraught with problems as it is difficult to isolate the influence of previous year's Medicaid expenditure that is directly attributable to incremental behavior. Therefore, previous year's Medicaid expenditure level is not incorporated into the present analysis of models. Empirical Work This section of the literature review sets the context of the study within the literature on determinants of geographic variation in spending and/or utilization of medical services. It encompasses three specific areas of empirical work, namely, 1. Geographic variation in overall Medicaid expenditure or utilization patterns, 2. Geographic variation in Medicaid and non-Medicaid nursing home expenditure and utilization, and 32

PAGE 44

3. Geographic variation in Medicare home health care utilization. It is worth noting that no empirical evidence exists on the determinants of geographic variation in Medicaid home health-care spending, or the Medicaid 2176 Waiver Program funding of LTC for the elderly. Geographic variation has been well documented, but the variance has not been carefully explained (Pierce & Hansen, 1987; GAO, 1987; Miller, 1990). It is also noteworthy that no estimates, let alone explanatory models, of private spending on nursing home care at the state level currently exist in the literature. Data limitations have previously not allowed for estimates of state-level private spending on nursing home care. Determinants of Geographic Variation in Medicaid Expenditures and Utilization Patterns Stuart and Bair (1971) represent one of the earliest attempts to explain. interstate variations in Medicaid spending. The researchers used FY 1968 state Medicaid expenditure data to test the hypothesis that states with greater income inequality would be slower to adopt new Medicaid programs and would have smaller benefit packages. Stuart and Bair suggested that wealthy persons in political control would affect this result. The researchers regressed Medicaid benefits (expressed as Medicaid expenditures divided by the state population) and the number of months each state had particular optional Medicaid programs against the measure of state income inequality (a function of the state income distribution). The analysis demonstrated a significant and negative relationship between income inequality and Medicaid benefit level and speed of adoption of optional Medicaid programs. This study has been criticized as simplistic and, ultimately, misleading. Critical explanatory factors, i.e., demographic, economic, and supply variables, were omitted from the model, casting doubt on the true predictive powers of the independent variables. 33

PAGE 45

In a follow-up effort, Stuart and Bair (1972) developed a model relating state Medicaid benefits (again expressed as expenditure divided by the state population) to demographic variables (percent state population 65 or older, and percent state population white), economic variables (per capita income), and local health-care system variables (nurses, hospital beds, and physicians per 1,000 state population). The researchers found Medicaid benefits significantly and positively related to the number of MDs per l ,000 population. The methodology employed by Stuart and Bair in this second study represented a major step forward. The utilization of multi-factorial variables for explaining geographic variation in medical care expenditures and utilization of services has since become the standard within the field. Both studies by Stuart and Bair have been criticized for equating Medicaid expenditures with benefits in their operational definitions. Reutzel (1984) argues that Medicaid program decisions regarding reimbursement systems and rates may allow some states to receive greater real benefits at similar expenditure levels, i.e., as "prudent" buyers. Differences in cross-state price levels would also argue against the equivalence of expenditure level with benefit (Buchanan et al., 1991). Grannemann (1979) utilized a median voter model for explaining interstate variation in Medicaid spending. He utilized a mixed cross-section and time-series sample of state Medicaid expenditure data from FY 1973 to 1977. The dependent variables were total Medicaid expenditure, Medicaid expenditures for various Medicaid eligible groups (AFDC adults, AFDC children, and SSI recipients), Medicaid expenditures for inpatient hospital services and physician services, number of Medicaid enrollees per general population, number of optional Medicaid services offered, and the share of state taxpayer income devoted to Medicaid. Grannemann's explanatory variables included economic variables (an indicator ofstate medical care costs, number of taxpayers, effective price of Medicaid to taxpayers, adjusted gross 34

PAGE 46

income of average-taxpayer federal-income tax deflated by CPI, and an index of income inequality within the state); demographic variables of the poor (percent black who are poor, percent poor over 65, percent poor under 16, and percent state population below the federal poverty level); demographics of the general population (percent population in metropolitan areas, percent population completed high school, and region of the state within the United States); and medical care market characteristics (physicians per capita and hospital beds per capita). Utilizing ordinary least square and reduced form regression analysis, Grannemann found Medicaid benefits significantly and negatively related to the effective price of Medicaid to the state taxpayer, and positively related to the needs of potential recipients (aged and poor), the income of the taxpayer, and the supply of health-care services. Grannemann and subsequently the GAO (1987), argue strongly that the Federal Medicaid Matching Rate, which directly affects the effective price of the Medicaid program to state taxpayers, is not sufficiently weighted toward poorer states and regions of the country. Poorer states' economic disadvantage is so great that even with the highest federal matching rate their Medicaid programs remain, essentially, underfunded. Inadequate federal contribution to poorer states' Medicaid program results in a relative redistribution of monies to wealthier states (as state taxpayers contribute to federal revenues) and in more restricted Medicaid eligibility and benefit packages in poorer states. Grannemann's work is most important to the field because of its policy implications regarding the federal-Medicaid matching rate. Grannemann's work, however, has been criticized for omitting structural characteristics of state Medicaid programs and political influences as possible explanatory factors. Grannemann considers both variables endogenous explanatory factors, in that they are driven by the fundamental explanatory variables of demography, economy, and federal financial participation rates. Reutzel, once again, refutes the equivalent use of expenditure data with benefit level. The GAO contends 35

PAGE 47

that Medicaid expenditure level can be used as a surrogate measure of benefits. GAO analysts regressed state-level Medicaid expenditure data (expressed as Medicaid spending per person below the federal poverty level) against two independent variables. These independent variables were Medicaid eligibility (expressed as the number of Medicaid enrollees as a percent of persons under the federal poverty level) and the scope of Medicaid services provided (expressed as a percent of the possible optional Medicaid services provided by the state). Taken together, the eligibility and coverage variables accounted for 69% of the state variance in Medicaid expenditure levels Reutzel's doctoral dissertation (1984) studied the impact of underlying state actuarial characteristics (demographic, income, and health-care system variables) and state Medicaid-policy decisions on interstate variations in Medicaid expenditure per elderly recipient of SSI (Medicaid aged eligible). His thesis utilized five dependent variables, the primary dependent variable of Medicaid expenditure per SSI recipient, plus four measures of services utilization. These services areas consisted of inpatient hospital days, nursing home days, outpatient MD visits, and units of miscellaneous services, all divided by the number of elderly SSI recipients per state. Reutzel's independent variables consisted of actuarial factors (expressed as Medicare dollars spent per Medicare enrollee in 1979 as a reflection of all pertinent actuarial factors, excluding those relevant to nursing home utilization); economic factors (state cost of living in 1979) ; demographic factors (percent of population residing in metropolitan areas); health-care system variables (the number of nursing home beds per 1,000 elderly); and Medicaid program characteristics (number of optional Medicaid services provided, limits present on the number of visits or dollars spent on ambulatory MD visits, presence of limits on inpatients hospital days, Medicare/Medicaid fee ratios for physician specialists, and presence of mandatory rate-setting hospitals) 36

PAGE 48

Utilizing regression analysis, Reutzel investigated state Medicaid expenditure and utilization data for 1979. Reutzel's analysis produced a R2 of .46 when he regressed actuarial variables alone against total Medicaid expenditure level per SSI recipient, and a R2 of .61 when he regressed the actuarial variables together with state Medicaid-program characteristic variables. Reutzel concluded that state Medicaid-expenditure levels are influenced by state Medicaid-program characteristics. Critics argue that these program characteristics may reflect fundamental socio economic, demographic, or political variables that have not been represented in the model. Reutzel contends that states choose between liberal-Medicaid benefit packages for recipients and liberal-reimbursement levels to providers. He suggests that political influences may be important determinants of state Medicaid policy. Once again, the nature of political influences on state Medicaid expenditure levels surfaces. McDevitt and Buczko (1985) utilized 1980 state data to examine the relationship between Medicaid program characteristics and program expenditure level. Two dependent variables were calculated, namely, Medicaid recipients per 1,000 poor (as a measure of state generosity regarding Medicaid eligibility) and Medicaid expenditure per recipient (as a measure of state generosity regarding Medicaid services). Taken together with the number of poor people in each state, these variables allow for estimation of states' total Medicaid expenditure. The dependent variables (number of Medicaid recipients, expenditure per recipient and total expenditure) were estimated for total Medicaid population, per SSI recipient, per AFDC child, and per AFDC adult. The researchers regressed these dependent variables against both Medicaid program characteristics and state demographic and health-care system variables. McDevitt and Buczko's findings indicate that states exercise considerable control over the number of Medicaid recipients eligible via AFDC standards and the optional medically needy 37

PAGE 49

program, but exert relatively little control over the number of Medicaid recipients eligible via SSI eligibility. They also found that state Medicaid-program characteristics exerted the greatest control over physician expenditures and the least control over long-term care expenditures. The presence of the optional medically needy eligibility category was the only Medicaid program characteristic that demonstrated a significant effect on both the number of state LTC recipients and the level of LTC Medicaid spending (significant at the .01 level). McDevitt and Buczko (1985) considered aspects of the relationship between Medicaidpolicy variables and either program participation or expenditure to be dynamic in nature, and thereby best represented by a time-series model. The results of their study demonstrate the need to: (1) explicate the determinants of Medicaid expenditure/ utilization specific to LTC, and, (2) use actual Medicaid expenditure and recipient data, as opposed to surrogate measures. Mauskopf et al. (1985) examined the effects of specific state Medicaid-program controls on health care utilization patterns of individual Medicaid recipients. The researchers utilized linear regression analysis to evaluate 1980 household-level utilization data from the National Medical Care Utilization and Expenditure Survey. They tested the hypothesis that states that offered fewer optional Medicaid services to their enrollees and imposed more limits on services would decrease the overall utilization of health-care services by the Medicaid population. The dependent variables used in the regression equation were the probability of using . specific health-care services in 1980, and the level (intensity) of use in 1980 for specific health services. These utilization measures were analyzed for specific Medicaid eligible groups, i.e., AFDC adults, AFDC children, and SSI recipients (aged, blind, and disabled SSI recipients were not delineated). The independent factors included price variables, income variables, health status variables, socio-demographic variables, supply variables, and a regional background variable. 38

PAGE 50

These variables were included in the equation to adjust for variations due to non-Medicaid programmatic factors, i.e., economic and demographic variables. The Medicaid program characteristics in the equation were classified as outpatient utilization controls, inpatient utilization controls, dental service controls, reimbursement controls, and reimbursement method. The findings of Mauskopf et al. indicate that utilization of health-care services is responsive to some Medicaid program controls, but that these controls affect eligible groups differently. Less generous reimbursement of physicians was seen to decrease total-Medicaid physician expenditures. This variable (reduced reimbursement to MDs) decreased physician visits for the SSI population, had no affect on AFDC adults, and increased visits for AFDC children. Alternative-Medicaid reimbursement systems (away from a customary cost method) was seen to reduce hospital LOS for AFDC children, reduced the probability of a hospitalization for AFDC adults, but had no affect on hospitalizations or LOS on the SSI population. For the most part, controls on the utilization of health-care services had a greater impact on AFDC children with chronic conditions (arguably the most in need) than on AFDC children without chronic conditions. Among AFDC adults, greater impact of controls was observed for those without chronic conditions. The researchers state that the relationships among Medicaid program controls are surely complex. The net result of program controls are frequently not the anticipated result. Medicaid eligible groups are affected differently by program controls, indicating a heterogenous patient population. Barrilleaux and Miller (1988) offer empirical support for the independent influence of state political factors on Medicaid expenditure levels. They regressed the dependent variable of "Medicaid effort" (expressed as the proportion of total state personal income devoted to Medicaid spending) against seven independent variables. These independent variables were supply (number 39

PAGE 51

of MDs), demand (number of Medicaid recipients), price (average expense per inpatient day), state political ideology (Americans for Democratic Action rating of that state's congressional delegation voting record), interest group influence (index of interest group density), bureaucratic density (spending on Medicaid administration) and urban population (percent of population residing in urban areas). Models were estimated cross-sectionally using 1982 state data. Regression results indicate that Medicaid spending is most sensitive to supply variables, although the political system variables (political ideology, bureaucratic and interest group density) each were significant at the .05 level. Medicaid spending effort was found to increase .60% with each percent increase in Medicaid administrative funding, .17% with each percent increase in political liberalism, and .35% with each percent increase in special-interest group density. Unfortunately, scant indicators of alternative demographic, economic, and supply variables are included in this model. To the extent that political measures simply reflect any of these omitted factors, the results may be spurious. This dissertation attempts to incorporate more thoroughly those variables. Schneider (1988, 1989) employed a mixed cross-sectional, time-series analysis to examine the role of federal, state, and local government on state Medicaid-expenditure levels. The dependent variable was simply the total nominal dollar amount of Medicaid spending for each state for years 1975 to 1985. Schneider divided the independent variables into two sets, one set corresponding to national-level influences on Medicaid spending, and another set corresponding to state-level influences on Medicaid spending. The two national-level variables were the percentage of the U.S. population identified by the federal government as "in poverty" in a given year, and, second, the medical CPI for a given year. The three state-level independent variables were the number of 40

PAGE 52

Medicaid recipients in a given year, the level of government within the state that was responsible for administration of the Medicaid program (state versus local), and finally, the previous year's state Medicaid-expenditUre level. Schneider estimated the model for two time periods, namely, 1975-1980 (the "pre-Reagan period") and 1981-1985 ("the Reagan period"). The author later recanted, stating the actual time periods utilized were 1975-1981 versus 1982-1985. The regression analysis for the 1975-1981 time period generated an adjusted R2 of .74, with local Medicaid administration and previous years' Medicaid-expenditure level having a statistically significant and positive impact on current-year Medicaid-expenditure level. The regression analysis for the 1982-1985 time period generated an adjusted R2 of .68, but presented with a marked contrast to the earlier time period. The impact of the previous years' spending dropped to zero, while the impact of the number of Medicaid recipients and locil administration increased dramatically. Schneider drew three conclusions from her research. First, it is necessary to take administrative variables into account when explaining Medicaid policy as local administration of the Medicaid program was associated with higher Medicaid-expenditure levels. Second, national, -state. and local-level factors all have important influences on Medicaid program development. Third, the time dimension must be included in empirical analysis of Medicaid program development as the policy output process is dynamic in nature. Lurie and Wyckoff (1989) have criticized Schneider's work on three scores: one, her coefQ.cient estimates depended on adjustments made to HCFA data that could not be replicated; I second, that the model omitted variables which affected the interpretation of included variables; and third, she drew conclusions from coefficients that were statistically not significant. 41

PAGE 53

Despite these serious criticisms, Schneider's work does question the influence of local administration of the Medicaid program and incremental budgeting on state Medicaid-expenditure levels. Buchanan et al. (1991) also found previous year's Medicaid spending and local administration of the Medicaid program to affect significantly current Medicaid-spending levels. The researchers classified their independent variables into three categories-economic factors, political factors, and implementation factors. The economic factors measured were personal income per capita and number of Medicaid recipients. The political factors included in the model were an index of general liberal ideology of the state's U.S. Representatives, an index of the state's inter-party competition, and the ratio of patient-care physicians per 1,000 population as a proxy measure for the political strength of the medical industry. The implementation factors were the Federal Medicaid Matching Rate, local versus state administration of the Medicaid program, and the previous year's Medicaid expenditure level. The researchers used a generalized two-stage least squares regression analysis to analyze cross-sectional data for years 1977-1987. Mixed cross-sectional, time-series analysis explained 63 % of the state variation in Medicaid-expenditure levels. Relationships between the current level of Medicaid spending and the log of the previous year's Medicaid spending, the log of the number of Medicaid recipients, personal income per capita, number of patient-care physicians, and local Medicaid administration were significant at the p = 05 level. Determinants of Geographic Variation in Expenditure and/ or Utilization Patterns Specific to Nursing Home Care This section does not pretend to do total justice to the body of literature concerning the determinants of nursing home utilization at the individual or family unit level. Many researchers, e.g., Wan and Weissert (1981), Branch and Jette (1982), Capitman (1985), and Grannemann and 42

PAGE 54

Grossman (1986) have employed multiple regression techniques to explain variations in nursing home utilization and predict institutionalization at the individual level (Grannemann, 1986). These studies unfortunately all report a relatively low R2 statistic. Among the factors found to have been associated with nursing home usage are lack of informal support at home, adequate financial resources or Medicaid coverage to pay for nursing home care, the recent use of a hospital or nursing home service, favorable attitudes toward institutionalization, and greater impairment in ADL and IADL as compared to elders living in the community. To the extent possible, the present dissertation incorporates many of these factors at the aggregate state level, i.e., average persons per household as a proxy for informal support. Scanlon (1980) reviewed five multivariate studies of nursing home utilization. Each researcher, i.e., Henry (1970), Morreale (1975), Dunlop (1976), Chiswick (1976) and Scanlon (1978), employed regression analysis to explain nursing home utilization (expressed as the percentage of the elderly population in nursing homes at the state or SMSA level). The independent variables included Medicaid generosity (in eligibility and/or benefits), the age structure of the elderly population, the availability of family support, urban versus rural residence of the elderly population, financial resources of the elderly out-of-pocket price of nursing home care, and the availability of alternative sources of care. Analyses produced R2 estimates ranging from 32 to .81 with out-of-pocket price, financial resources of the elderly, and age structure of the elderly population most often statistically significant. The focus of the literature concerning variation in nursing home utilization in the late 1970s and early 1980s was on the demographics and economics of the elderly in conjunction with price variables. Scanlon also focused on the importance of the supply variable as availability of Medicaid-certified beds is constrained in many states, impacting directly on nursing home 43

PAGE 55

utilization and expenditure. Most empirical work on nursing home utilization and expenditure throughout the 1980s incorporated health-care system supply variables in the equation. Ray et al. (1987) studied interstate differences in the characteristics of Medicaid nursing home residents and in their utilization of medical-care services. The researchers used Medicaid claims and enrollment data for calendar year 1981 for three states-Michigan, California, and New York. State resident characteristics included the proportion of Medicaid-covered nursing home residents relative to the overall state nursing home population, demographic characteristics of the Medicaid nursing home population, and the diagnostic case mix of each state's Medicaid nursing home population The second dependent variable (medical care utilization) encompasses measures of turnover among nursing home residents, expenditures for nursing home care, and inpatient-hospital utilization patterns of the nursing home elderly. Medicaid-resident characteristics for each state were very similar, approximately three fourths of the Medicaid nursing home residents were women, over 45% were over the age of 85, and over 95% were SSI-aged Medicaid eligible. There was little interstate variation in these rates. Diagnostic case mix of Medicaid nursing home residents across the three states was also consistent reflecting the well-known causes of morbidity among the elderly. Medical-care utilization patterns of Medicaid nursing home residents for 1981 showed marked interstate differences. The turnover among nursing home residents (expressed as the proportion of elderly Medicaid nursing home residents who entered the nursing home in 1981) was 17% for New York, 25% for Michigan, and 32% for California. Among the 52,306 Medicaid recipients who entered nursing homes in the three states in 1981, 43% had a Medicaid-covered hospitalization within 30 days of entry in California, in contrast to 24% in New York, and 20% for Michigan. There were also pronounced differences in 44

PAGE 56

the average pre-nursing home hospital LOS. The average pre-nursing home hospitalization LOS was 60 days in New York, 18 days in Michigan, and 12 days in California. This variability was not explained by any differences in residents' diagnosis among states. Ray and colleagues employed descriptive analysis only; multivariate statistical analysis has not been applied to their findings. This limitation of their study prohibits generalization of the results. The study also has serious sample-size limitations. The research does; nevertheless, demonstrate important relationships between LTC utilization patterns and health-care system variables such as numbers of hospital beds, average length of inpatient stays, and availability of nursing home beds. These relationships hold serious implications for both future Medicare and Medicaid levels of expenditures. Harrington and Swan (1987) utilized a cross-sectional time-series regression analysis to investigate the impact state Medicaid nursing home policies on nursing home utilization and expenditure. The data set for this study was 42 states over a six-year period, 1978-1983. Their three dependent variables were number of Medicaid nursing home recipients per 1,000 aged (utilization), Medicaid nursing home expenditure per recipient, and total Medicaid nursing home expenditure per aged state population. Three sets of independent variables were used in the model: (1) state Medicaid nursing home policies on utilization, eligibility, and reimbursement; (2) provider supply (nursing home beds, hospital beds, hospital-occupancy rate, and number of physicians); and (3) macrocontextual variables (percent state unemployment, percent state population over 65, percent state population residing in metropolitan areas, and pay per nursing employee). Harrington and Swan (1987) discovered the supply of nursing home beds to be the strongest predictor of Medicaid nursing home utilization and of overall Medicaid nursing home expenditures. The presence of a medically needy 45

PAGE 57

eligibility program was the only significant Medicaid policy predictor of nursing home utilization. Reimbursement rates had little to no affect on access of Medicaid recipients to nursing home beds. The intermediate-care facility (ICF) reimbursement rate was the strongest predictor of nursing home expenditure per recipient. Average nurses' pay was the strongest macrocontextual predictor of nursing home expenditure per recipient. Another statistically significant predictor of expenditure per nursing home recipient was the percent of the population living in metropolitan areas. Harrington and Swan (1987) found that hospital-occupancy rate had a strong positive effect on expenditure per nursing home recipient and oil overall Medicaid nursing home expenditures. The researchers hypothesized that hospitals in states with higher occupancy rates release patients earlier leading to longer nursing home stays and higher Medicaid expenditures. Both the presence of a medically needy program and ICF per diem rates were significant predictors of Medicaid nursing home expenditures per aged population. Eligibility affects Medicaid expenditure by increasing number of recipients (utilization) and reimbursement rates affect expenditure per recipient and, therefore, overall Medicaid spending. The present dissertation enhances Harrington and Swan's research design by using actual Medicaid expenditure and recipient data specific to both categories of spending (nursing home, home care, and 2176 spending) and target population, i.e., the elderly. This thesis also incorporates characteristics of the general state population, expands the Medicaid program characteristic variable, and incorporates federal funding and state political ideology into the equation. Carrell (1990) attempted to locate long-term care on Theodore Lowi's distributive redistributive continuum of public policy. The researcher used education (distributive) and welfare (redistributive) policies to provide polar benchmarks to render a judgment regarding the placement 46

PAGE 58

of LTC on the distributive-redistributive continuum. Carrell states that distributive policies will be driven by economic factors (ability to pay), while redistributive policies will be driven by political factors, especially inter-party competition. Carrell examined state-level expenditure data for the three policy areas, long-term care, education, and welfare for 1977-1986. The LTC dependent variable in Carrell's study is total Medicaid expenditure on LTC services for physically and cognitively impaired persons over 65 years of age and per recipient expenditure on LTC services for the elderly. The explanatory factors were economic (median personal income and state fiscal capacity) and political (political liberalism, inter-party competition, and intergenerational competition). Carrell employed explanatory variables concerning the aged state population in the expenditure-per-recipient model and total-Medicaid-expenditure model, and number of LTC recipients in the total expenditure model. The explained variation of the model for expenditure per LTC recipient was 59% with political liberalism, fiscal capacity, and aged population statistically significant. The explained variation of total Medicaid LTC expenditure was 96% with the number of LTC recipients (standardized coefficient of +. 903) and political liberalism statistically significant at the p < .05 level. Carrell places LTC somewhere in the middle of the distributive-redistributive continuum regarding public policy, with economic capacity, political liberalism, and interparty competition important explanatory variables. He states that any movement in LTC's placement in the continuum between 1977 and 1986 has been toward the redistributive end, with political liberalism emerging as a more powerful explanatory variable in later years. Swan (1990) used sets of supply, demand, and policy factors to explain interstate variation and temporal variation in the proportion of state Medicaid monies expended on nursing home care. 47

PAGE 59

The supply variable was nursing home bed stock. The demand factors were the proportion of persons over 65 in the state, the percentage of women in the state labor force, the percent state unemployment, heating degree days, personal income per capita, hospital beds per capita, hospital occupancy, office MDs per capita, and Medicaid spend-down level. The policy variables were Medicaid intermediate-care facility expenditure minus that portion of intermediate-care facility expenditure used for the mentally retarded population (ICF MR) per capita and Medicaid-hospital expenditure per capita. Swan also included average nursing pay per employee in a given state, as staffing constitutes a major portion of nursing home costs and is expected to influence positively the percent of Medicaid expended on nursing homes. Analysis included cross-sectional regression for each year of the period from 1979-1984, and mixed cross-sectional time-series analysis of the pooled data for the entire period. The adjusted R2 for each individual year of the study ranged from .45 (1983) to .61 (1981). Swan's analysis revealed that nursing-bed stock had the strongest positive influence on the percent of Medicaid monies going to nursing home care. Each one-percent population aged was associated with over one-percent share of Medicaid expenditure spent on nursing home care. Furthermore, higher hospital occupancy was positively related to increased percent of Medicaid monies being spent on nursing home care. This was interpreted as an early affect of Medicare's hospital-prospective payment system (PPS) as hospitals attempted to reduce the average LOS, in part, by discharge to nursing homes. Medicaid expenditure per capita for ICF-MR and acute-hospital care negatively affected the share of the Medicaid budget going to nursing homes. Controlling for all factors in the model, the percentage of Medicaid spent for nursing home care declined over time. Swan suggests that this decline in percent of the Medicaid budget dedicated to nursing home care may reflect omitted state-policy factors aimed at containing Medicaid nursing home utilization and expenditures. 48

PAGE 60

This dissertation attempts to control for such policy variables. Determinants of Geographic Variation in Medicare Home Health Care Utilization Swan and Benjamin (1990) examined Medicare home health utilization at the state level as a function of state nursing home market factors. The authors proposed that home health utilization (Medicare home health visits per 100,000 aged state population) is a function of demand and supply factors. The variables thought to affect the demand for Medicare home-care services and therefore incorporated into the model were nursing home bed stock, Medicaid nursing home days, percent of state population 85 or older, spend-down level for Medicaid eligibility, income per capita, percent of women in the labor force, and the number of general hospital beds per capita. Total nursing home bed stock is used in the model. The authors acknowledge that Medicare-certified SNF beds, hospital-based nursing home beds, hospital swing-beds, and rehabilitation-hospital beds may be the critical substitutes for Medicare home health utilization. The variables thought to affect the supply of Medicare home-care services were home health-care agencies per 100,000 state population, percent of the state population in metropolitan areas, and the percent of the work force that is unionized. Regression analysis were performed for individual years in the 1978-1984 period, and mixed cross-sectional time-series were used for the pooled data. The statistically significant variables for explaining the number of Medicare home health visits per 100,000 aged state population were the percent of the state population 85 or older ( + ), nursing home beds per 1,000 aged population(-), Medicaid nursing home days per 100 nursing home days ( + ), Medicaid spenddown level ( + ), Medicare-certified home health agencies per 100,000 ( + ), and percent women in the work force(-). Adjusted R2 was approximately .30 for individual years. 49

PAGE 61

Swan and Benjamin suggest that a major factor in the home health-care market is what is going on in the nursing home market. Greater demand for Medicare home health-care is generated by a relative scarcity of nursing home beds, and perhaps, by greater access of Medicaid recipients to those beds. The authors suggest that PPS may have strengthened those relationships. This study suggests that a substantial portion of the unexplained variance in one health care service may be accounted for elsewhere-" across the boundaries between Medicare and Medicaid, and between acute and long-term care." Summarv of Empirical Work Theoretical underpinning from multiple disciplines along with previous empirical work indicate that Medicaid spending on LTC for the elderly is a function of endogenous state socio economic and demographic factors in conjunction with state-selected variables related to structural components of the Medicaid program, the marketplace, and political ideology. Socio-economic variables found to correlate with state Medicaid spending, or more specifically, state Medicaid spending on nursing home care are: effective price of Medicaid to taxpayers(-), (Grannemann, 1979) income of taxpayers ( + ), (Grannemann, 1979) state level per capita income ( +), (Buchanan et al., 1991) state cost of living ( + ), (Reutzel, 1984) average nurses' pay at the state level ( + ), (Harrington & Swan, 1987) financial resources of the elderly ( + ), (Scanlon, 1980) fiscal capacity of the state ( + ), (Carrell, 1990). 50

PAGE 62

State demographic variables associated with either overall Medicaid spending or Medicaid spending on nursing home care are: number of aged persons in the state population ( + ), (Grannemann, 1979; Carrell, 1990; Swan, 1990) age structure of the elderly population ( + ), (Scanlon, 1980) percent state population living in metropolitan areas ( + ), (Harrington & Swan, 1987). State-selected policy variables regarding either the structure of the state Medicaid program or the state-medical marketplace associated with overall Medicaid spending or Medicaid spending on nursing home care are: physicians per capita ( + ), (Grannemann, 1979; Barrilleaux & Miller, 1988; Buchanan et al., 1991) hospital beds per capita ( + ), (Grannemann, 1979; Ray, 1987-descriptive study only) number of nursing home beds per 1,000 elderly ( + ), (Scanlon, 1980; Reutzel, 1984; Harrington, 1987; Ray, 1987-descriptive study only; Swan, 1990) number of optional Medicaid services ( + ), (Reutzel, 1984) presence of the optional "medically needy" program ( + ), (McDevitt & Buczko, 1985; Harrington & Swan, 1987) Medicaid per diem nursing home reimbursement rate ( +), (Harrington & Swan, 1987) reduced reimbursement rates to physicians ( +), (Mauskopf et al., 1985) limits on number or dollars on ambulatory physician visits ( + ), (Reutzel, 1984) average LOS of elderly in acute-care hospitals ( +) (Ray, 1987-descriptive study only) hospital occupancy rates ( +), (Harrington & Swan, 1987; Swan, 1990) level of administration of the Medicaid program (local + ), (Schneider, 1988; Buchanan et al., 1991). 51

PAGE 63

State political variables found to influence either overall Medicaid spending or Medicaid spending on nursing home care are: political ideology (liberalism, +),(Barrilleaux & Miller 1988; Carrell, 1990) bureaucratic density (spending on Medicaid administration, + ), (Barrilleaux & Miller, 1988) interest group density ( + ), (Barrilleaux & Miller, 1988). Swan and Benjamin (1990) examined Medicare home-care utilization at the state level as a function of nursing home market factors. They found Medicare home health-care visits per 100,000 state population to be significantly correlated with: state population over 85 ( +) percent women in the workforce (-) Medicaid nursing home days/100 nursing home days ( +) Medicaid spend-down level ( +) number of Medicare-certified home health agencies per 100,000 state population ( +) number of nursing home beds per 100 state aged population(-) The present dissertation is consistent with previous empirical and theoretical work on interstate variation in Medicaid and Medicare spending in that it incorporates many of these socio economic, demographic, political and policy variables in the explanatory model. This study contributes to our understanding of the issue by expanding usual state Medicaid policy, economic, political, and state-taxpayer variables in the model. Furthermore, this study delineates the components of Medicaid spending on LTC for the elderly into both the components of spending i.e., average expenditure per recipient and number of recipients of LTC in a manner consistent with Harrington and Swan (1987), and service area within LTC, i.e., nursing home and home and community care. The explained variation across states in Medicaid spending on LTC differentiated 52

PAGE 64

by the components of spending and service area may show very different results as states have been shown to spend their Medicaid dollars very differently (Grannemann, 1979; Rivlin & Weiner, 1987). For instance, the state of New York was responsible for 80% of the national Medicaid expenditure on home health care in 1986 (Swan, 1990). To the best of present knowledge, this dissertation represents the first attempt to model state variation in Medicaid spending on both home care and the 2176 Waiver Program. This study also represents an initial attempt to calculate and explain interstate variation in private spending on nursing home care for the elderly Data limitations have previously precluded such an attempt. 53

PAGE 65

CHAPTER 4 METHODOLOGY The purpose of this study is to explain interstate variations in long-term care spending for the elderly. The central hypothesis is that endogenous socio-economic and demographic state variables and state-selected policy and political variables differentially affect access, spending and utilization patterns of the elderly to Medicaid long-term care services. Public expenditures on long-term care for the elderly (Medicaid) is analyzed by service area (nursing home care, home care, and Medicaid 2176 Waiver Program) and, when possible, by the components of spending (average expenditure per recipient, number of recipients, and total expenditure). Private spending on long-term care for the elderly is analyzed for total expenditure on nursing home care only. Socio-economic, demographic, policy, and political variables are used in the analysis to explain variations in long-term care spending across states. These independent variables are consistent with the six theoretical models of public spending on long-term care discussed in the preceding chapter. To reiterate, these six models are: the medical model, the cultural model, the economic model, the structural model, the political model, and the public-goods model. The objectives of this study are: 1. To determine which socio-economic, demographic, policy, and political variables drive public long-term care spending for the elderly; 2. To offer historical data for forecasting state spending on long-term care for the elderly; 54

PAGE 66

3. To determine whether political/policy variables differentially affect long-term care spending on the elderly by service area, or by the components of spending (average expenditure amounts or recipient data across service areas); 4. To suggest long-term care cost-containment strategies to state policy makers; 5. To determine whether Medicaid home care spending and utilization act as a compliment or substitute for Medicaid nursing home care; and 6. To determine which socio-economic, demographic, policy, and political variables drive private spending on nursing home care for the elderly across states. Model The general model of this study is expressed in functional notation by: Y = f (E, L, D, F, M, H, P) where: Y = State level LTC use or expenditure level E = Economic factors L = Demographic characteristics of the elderly population D = Demographic characteristics of the voting population F = Federal Medicaid policy constraints M = Medicaid program characteristics H = Health-care system characteristics P = Political climate of the state These explanatory variables can be categorized as either socio-economic or demographic variables that are relatively external to state manipulation, or state-selected political or policy variables that influence the state long-term care system, but may be driven by state socio-economic 55

PAGE 67

conditions or demographics. For example, health-care system characteristics, such as the supply of nursing home beds, may reflect underlying demand for services that is a function of economic and demographic factors. Hypotheses Once again, the central hypothesis of the study is that state socio-economic and demographic variables, as well as state-selected policy and political variables, differentially affect access, expenditure, and utilization patterns of the state elderly population to LTC services. As such, state variables regarding health-care policy, the medical market place, and political ideology do not merely reflect states' macrocontextual economic and demographic status, rather they exert a separate and distinct influence on LTC services used by the elderly. The reduced model utilized in the study includes only the socio-economic and demographic variables and can be expressed in linear form as: (note that specific variables are now presented in the model as opposed to categories of variables) Y = aWGNH + bOLD + cELDVT + dA VINC + eCOST + tUEMP + gSMSA + hHOUS + iTEMP + jYEAR While the expanded model can be expressed in linear form as: Y = aWGNH +bOLD+ cELDVT + dAVINC + eCOST + tUEMP + gSMSA + hHOUS + iTEMP + jYEAR + kEMN + lMOB + mRATE + nMAD + oBEDS + pLOS + qLQ 56

PAGE 68

Independent Variables 1. Socio-Economic. Demographic Variables WGNH = average weekly wage in the nursing home industry across states, 1982 dollars OLD = number of persons > 85 per 1,000 elderly ELD VT = number of persons > 65 per 1 ,000 persons > 18 A VINC = average income offour-person family, 1982 dollars COST = effective unit cost of LTC to state taxpayers, 1982 dollars, calculated via ([1-CORTAX] X [1-FMMP] X WGNH) where: CORTAX = corporate net income to government as a percentage of state tax collection FMMP = Federal Medicaid Matching percent WGNH ,.; average weekly wage in nursing home industry UEMP = annual state unemployment rate SMSA = percent state elderly population living in standard metropolitan statistical area; 1980 data of percent elderly in SMSA projected forward for 1981-1988 based on percent elderly population in SMSA/percent general population in SMSA from base year 1980; percent general population in SMSA available for years 1980 through 1988 HOUS = average population per household in state 2. State-Selected Variables EMN = inclusion of elderly in Medicaid's optional medically needy eligibility program MOB = number of optional benefits afforded the categorically needy in state Medicaid program (optional benefits specific to <21 population excluded in count) 57

PAGE 69

RATE = weighted average of SNF and ICF Medicaid per diem rates, 1982 dollars, calculated: {SNF rate X [SNF days/SNF + ICF (non-MR) days of care]} + {ICF rate X [ICF (non-MR) days of care/ICF (non-MR) + SNF days of care]}, (personal communication with HCFA, Medicaid division) MAD = level of administration of Medicaid program, state versus local BEDS = number of nursing home beds (freestanding and hospital based) per 1,000 elderly LOS = average length of stay in an acute hospital for state Medicare population LQ = liberal quotient of the state political climate 3. Control Variables TEMP = average annual state temperature YEAR = year, 1981-1988 Dependent Variables There are three sets of dependent variables: 1. state-level Medicaid spending on LTC for the elderly, all states as data allow, 1981 to 1988; 2. state-level Medicaid spending on LTC for the elderly, excluding New York as New York is the clear outlier state, particularly on home care spending and utilization as previously noted, 1981 to 1988; and 3. private spending on nursing home care for the elderly, all states as data allow, 1987 and 1988. 58

PAGE 70

Whenever possible, expenditure measures have been delineated into expenditure per 1,000 elderly, expenditure per recipient, and number of recipients per 1,000 elderly; and service area, i.e., spending on nursing home care versus home and community care. There are a total of ten dependent variables: nine specific to Medicaid expenditures on LTC for the elderly; and one on private spending on nursing home care for the elderly. Medicaid Dependent Variables (1981 to 1988) 1. Total state Medicaid spending on LTC for the elderly per 1,000 elderly population. 2. State Medicaid spending on nursing home care for the elderly per 1,000 elderly population. 3. Number of elderly recipients of Medicaid nursing home dollars per 1,000 elderly population. 4. Average medicaid expenditure per elderly Medicaid nursing home recipient. 5. State Medicaid spending on home care for the elderly per 1,000 elderly population. 6. Number of elderly recipients of home care spending per 1,000 elderly population. 7. Average Medicaid expenditure per elderly Medicaid home-care recipient. 8. State Medicaid spending on the 2176 waiver program for the elderly per 1,000 elderly population. 9. State Medicaid spending on institutional care (nursing home) as a percent of total state Medicaid spending on LTC for the elderly. Private Spending Dependent Variable (1987 and 1988) 10. Total private spending on long-term nursing home care per 1,000 elderly population. Table 4.1 summarizes the hypothesized relationships among the independent and dependent variables 59

PAGE 71

0'1 0 Table 4.1. Summary of Hypothesized Relationships -----------Total # LTC N.H. N.H Exp. Exp. Rec. Wages in the nursing home industry + + 0 Old elderly 85)/elderly 65) population + + + Elderly voting pop. 65)/voting 18) pop. + + + Average state income + + + Unit cost of long-term care to state taxpayers --Percent state unemployment --Percent state elderly population living in SMSA 0 + 0 Average persons per household --Average state temperature -Year + + 0 Elderly included in "medically needy" Medicaid + + + eligibility category Number of optional benefits to elderly in state 0 0 -Medicaid Program Medicaid per diem nursing home reimb. rate + + 0 State/local administration of Medicaid Program + + + Number of nursing home beds/1,000 eld + + + population Medicare acute care hospital length of stay ---Liberal quotient of Congressional voting record + + + -Avg N.H Exp. Per Rec. + + + + -+ 0 0 + 0 + + 0 0 -+ -------Home #Home Avg Home 2176 N H. Exp Priv. Care Care Care Exp. Waiver as % of N H. Exp. Rec. Per Rec Exp. LTC Exp Exp. + 0 + + 0 + + + 0 + + + + + 0 + 0 0 + + + + 0 + -0 0 0 0 -0 0 + 0 0 0 0 0 0 0 0 -+ + + + 0 0 + 0 0 + 0 0 0 0 0 0 + 0 + 0 0 0 0 0 0 0 0 0 -0 -+ + -0 -+ + + + 0

PAGE 72

Total Medicaid spending for LTC of the elderly is disaggregated into nursing home, home care, and 2176 Waiver spending to investigate conventional wisdom that state LTC services have mirrored the evolution of the LTC literature toward the provision of services in an environment of least restriction. The components of aggregate spending, namely, number of recipients and average per recipient, are investigated as each may be affected differently by Medicaid policy. Each component of LTC spending carries implications regarding access to services (e.g., number of recipients) and intensity of services (e.g., average expenditure per recipient). Data on the components of Medicaid spending, both services areas and recipient/ expenditure data, provide a longitudinal view of expenditures and gross utilization patterns for LTC services across states. Design A mixed cross-sectional, time series design is used in the research. The nature of the data, i.e., a state-level program, mandates a cross-sectional approach. A time-series is utilized to capture dynamic changes in the coefficients across years. It is generally understood that mixed cross-sectional, tin!e series design may complicate the estimation process, but there is little alternative in this case given the breadth of the analysis. Stimson argues that data which are "cross-sectional dominant" is appropriate for least squares analysis when dummy variables marking time frame are incorporated into the design (Schneider, 1988). Cross-sectional dominant refers to a design where there are more data sets in the cross-section than in the time series. This mixed cross-sectional, time series approach is consistent with the majority of researchers who have investigated interstate variations in Medicaid or long-term care. These researchers include Grannemann (1979); Harrington and Swan (1987); 61

PAGE 73

Schneider (1988); Harrington, Swan, and Grant (1988); Carrell (1990); and Buchanan et al. (1991). A pooled series for eight years (1981 to 1988) is used for Medicaid expenditures on nursing home, home care, and total LTC expenditures for the elderly. A seven year pooled time series, 1982-1988, is employed for Medicaid 2176 spending, because the 2176 waiver program did not exist prior to 1982. Regression analyses utilizing the expanded model does NOT include year 1982 as the number of optional Medicaid benefits offered to the categorically needy (MOB) were not available. State LOS measures were estimated for years 1987 and 1988, as state measures were not available. Actual state 1986 LOS data were projected forward using percent change in national 1987 and 1988 figures provided by the Department of the Census. Dramatic changes occurred in LOS between 1981-1986, for which actual state data exist. It is generally held that the dramatic changes in acute-care LOS for the Medicare population had stabilized by the late 1980s. Private spending on long-term nursing home care is estimated for 1987 and 1988, using nursing home cost reports as a basis. Private spending is captured as private expenditure on nursing home care per 1,000 elderly only. No recipient or average expenditure per recipient data are estimated. Unit of Analysis To model interstate variation of Medicaid spending on LTC for the elderly, Medicaid data from the 49 states with Medicaid programs plus Washington, D.C. were collected. Long-term care was not covered in the Arizona Health Care Cost Containment System (AHCCS) during the observation period. 62

PAGE 74

Buchanan et al. (1991) omitted Alaska and Hawaii from their analysis of the factors influencing the level of state Medicaid spending because these states have been found to have disproportionate influence in quantitative analyses using state-level data. They excluded Washington D.C., because it lacks a state government. Alaska, Hawaii, and Washington D.C. are included in this as data allow. No nursing home wage data were available for Alaska, eliminating Alaska from the study. Nursing home wage data were available for four years only for Washington D.C and no liberal quotient was available for D.C., since D.C lacks a congressional voting record. Washington D.C. is, therefore, omitted from all expanded models and half (4/8) of the reduced models. Hawaii is included in all niodels and time series. Data Collection Data for the independent variables are collected from published secondary sources accounting for a high degree of reliability in the data. The economic variables are collected from the Bureau of the Census and the Bureau of Labor Statistics. The demographic data on state elderly and general populations are collected from various branches of the Bureau of the Census and the Internal Revenue Service. The Federal Medicaid Matching Rate and Medicaid program characteristics are collected from the Office of Medicaid Management and the HCFA. Health care system variables are obtained from the Statistical Abstract of the States and previously published empirical work. The political quotient was computed from rankings given Congressional members by the ADA. Measures for each independent variable have a high degree of internal validity in that they quantify what they are intended to represent with the possible exception of the ADA's ranking of Congressional voting records as a surrogate for state political ideology. All data collection occurred at the state level. The appendix lists specific measurements and data sources. 63

PAGE 75

Data for annual Medicaid expenditures on total LTC services for the elderly, nursing home expenditures for the elderly, and home care expenditures for the elderly are collected from the HCFA's Division of Medicaid Statistics. The Division of Medicaid Statistics also provided data on the components of spending for each of the service areas, i.e., number of recipients and average expenditure per recipient. No attempt was made to verify this data which are assumed to be both valid and highly reliable. Aggregate expenditure data after 1982 for 2176 waiver spending on the elderly by each state were obtained from the HCFA's Office of Research and Demonstrations. Private expenditures on nursing home services across all states were estimated in general accord with the methods used by the HCFA's Office of the Actuary in determining annual national private spending on nursing home care (personal communication and HCFA, 1990). HCFA uses financial data from the 1977 National Nursing Home Survey to estimate total nursing home spending, then extrapolates spending forward based, primarily, on Bureau of Labor Statistics data on employment and work hours associated with nursing and personal care facilities. HCFA then extracts public spending on nursing home care, (Medicare, Medicaid, ICF/MR funds, VA funds, etc.) which leaves a residual called private spending. Total (public plus private) state-level nursing home patient revenue is estimated for this study via nursing home cost reports collated by Health Care Investment Analysts (HCIA) of Baltimore, Maryland HCIA has cost reports from nursing homes from each state approximating 40% of the total number of nursing home facilities in the nation for 1987 and 1988. This sample allows an extrapolation to total state level nursing home revenues via a simple multiplier of total nursing home beds in the state divided by total nursing home beds in the sample. This revenue projection was adjusted for the elderly population via the percent elderly nursing home population 64

PAGE 76

statistic published by the National Center for Health Statistics for 1986 (U.S. Deparonent of Health and Human Services, 1988). Medicaid and Medicare associated expenditures on nursing home care for the elderly were extracted from these amounts. Medicare "covered charges" for SNF services are subtracted from the revenue calculation as opposed to Medicare "reimbursements." Medicare covered charges include both Medicare reimbursements and co-payments, either out-of-pocket expenses or private insurance supplements. This study targets long-term care for the elderly, while Medicare associated charges represent short-term, skilled nursing home services following an acute-care episode. State level VA and ICF/MR expenditures are not subtracted from the calculation as longterm VA and ICF/MR beds are not included in the state projections of nursing home beds (personai communication with researcher, Grant, 1991). State-only monies for nursing home care will be subtracted foi Delaware and Massachusetts, as these are the only states using state-specific funds on chronic nursing home care. The vast majority of state-only programs target home and community services, with few state dollars, if any, being spent on nursing home care. A statelevel private-pay residual for long-term nursing home care for the elderly results. It is assumed that nursing home occupancy rates are equal in the state samples and the general state population. It is further assumed that state samples represent an equivalent ratio of private pay to publicly subsidized nursing home beds as exists in the general state population. HCIA states that there is nothing inherent in their sampling technique to disallow either assumption (HCIA, personal communication, 1991). This assurance by HCIA regarding their sampling technique in conjunction with the utilization of a generally accepted methodology used by HCFA enhances the degree to which the results of the regression analysis specific to private spending on nursing home care can be generalized to the universe beyond the sample (external validity). 65

PAGE 77

For all other aspects of the study the entire universe of appropriate data are examined as data allow. Data Analysis The data for this study are analyzed in two ways: 1. Descriptive analysis using simple means and standard deviations of state spending and type expenditure; 2. Regression analysis of dependent variables as a function of independent explanatory variables, using both reduced and expanded models to more precisely isolate the impact of state selected variables. Multiple regression analysis is the cornerstone statistical analysis of this study since the goal of the research is to develop a model that maximizes the explained variance in the state variations in expenditures on LTC for the elderly. Ordinary least square regression is used for the reduced and expanded models. 66

PAGE 78

CHAPTER 5 RESULTS The results of this study are presented by descriptive and regression analysis. Descriptive analysis examines both dependent and independent variables for individual years 1981 through 1988 to note national trends. Descriptive analysis also examines aggregate mean values for dependent and independent variables, 1981 through 1988, with high and low outlier states noted. Regression analysis results are organized around dependent variables. Each dependent variable has four equations, i.e., reduced and expanded models run inclusive and exclusive of the state of New York. Descriptive Analysis Descriptive analysis examines, first, the dependent variables in the study (Medicaid spending and utilization data), and second, the independent variables in the study (state socio economic, demographic, policy, and political variables) by and across years, 1981 through 1988. Third, correlation coefficients between dependent and independent variables are presented and discussed. Dependent Variables Table 5.1 documents a slow increase in total Medicaid dollars spent on LTC for the elderly and total Medicaid dollars spent on nursing home care for the elderly, on a per 1,000 elderly population basis, from 1981 to 1988, all in 1982 dollars. Expenditure figures were adjusted by the CPI, not a medical or LTC CPl. It is generally assumed that LTC throughout the 1980s outstripped the general CPI by about 2 percentage points (Rivlin & Weiner, 1988). 67

PAGE 79

0\ 00 Table 5.1 Descriptive Data of Dependent Variables, By Year, 1981 1988, in 1982 Dollars I Mean (Standard Deviation) I I 1981 I 1982 I 1983 I 1984 I 1985 I Total Medicaid spending on long-term care 313,307 319 937 316,141 327,103 338,770 for the elderly per 1,000 elderly population (159,537) (172,015) (146,033) (158,934) (158,551) Medicaid spending on nursing home care for 308,370 313 839 307,677 313,549 322,074 the elderly per 1,000 elderly population (153,573) (165,314) (136,003) (146,977) (142,481) Average Medicaid expenditure per elder 6,689 7,019 7,113 7,543 7 524 recipient of nursing home care (1,863) (2,153) (2,835) (3,275) (3,058) Elderly Medicaid recipients of nursing home 45 37 44.16 43 96 42 25 43 63 care per 1,000 elderly population (14 91) (13.44) (12.74) (12.22) (13.16) Medicaid spending on home care for the 4,933 5,815 6,414 7,900 11,025 elderly per 1,000 elderly population (16,539) (18,716) (20,681) (22,544) (28,076) Average Medicaid expenditure per elder 660 824 894 1,051 1,223 recipient cif home care (492) (642) (652) (648) (774) Elderly Medicaid recipients of home care per 4.84 4 63 4.98 5.44 6.33 1,000 elderly population (8.25) (8 36) (8.28) (7 70) (7.91) Medicaid spending on 2176 Waiver Program 211 1,996 3,958 5,352 for the elderly per I ,000 elderly population (1, 048) (6,136) (8,605) (11,781) Private spending on nursing home care for the elderly per 1,000 elderly population Percent of long-term care Medicaid spending 98.86 98 55 97 73 96 58 95 63 on nursing home care (2.54) (2.87) (4.01) (4 93) (6.04) I 1986 I 1987 I 1988 I 345,561 336,907 362,985 (170,107) (158,509) (183,023) 323,176 313,883 333,183 (149,287) (142,360) (161,289) 7 649 7,734 7,847 (3,613) (3,360) (3,567) 43.56 41.49 43 56 (12.58) (11.13) p1.55) 15,154 16,691 20,472 (31,640) (26,180) (38,962) 1,515 1,578 1,745 (949) (997) (1,316) 7 64 8.13 9.23 (8.48) (7.52) (7 98) 6,858 7,954 9,310 (10,818) (11,477) (11 ,073) 422,897 448,510 (175,221) (194,540) 93.79 93.30 91.65 (6 24) (6.03) (7.17)

PAGE 80

Adjusting Medicaid expenditure amounts by a LTC component of the CPI is likely to have demonstrated no change across years in real dollar spending on nursing home care, or overall LTC for the elderly on a per state elderly population basis. Unfortunately, yearly LTC CPI statistics are currently not published. National average of the number of elderly Medicaid recipients of nursing home care has slipped from 45.4 in 1981 to 43.6 in 1988, each on a per 1,000 elderly state population basis. Real average Medicaid expenditure per nursing home resident increased by 17% from 1981 to 1988. Dramatic change occurred in non-institutional Medicaid spending and utilization patterns from 1981 to 1988. Real Medicaid spending on home health services per 1,000 elderly population more than quadrupled; the number of Medicaid recipients of home-care services per 1,000 state elderly population approximately doubled; and real average expenditure per Medicaid home health care recipient appioX:iinately tripled. Medicaid spending on the 2176 Waiver Program per 1,000 state elderly population grew from $211 in 1982 to $9,310 in 1988, in 1982 dollars. Concurrently, the percent of the Medicaid LTC budget for the elderly devoted to nursing home care, i.e institutional versus non-institutional care, shrank from 98.9% in 1981 to 91.7% in 1988. Within these national averages lie great variation (Table 5.2). This study certainly supports the long-held assertion that wide variations exist across states in Medicaid expenditure and utilization patterns. This assertion holds true even within one services area (LTC) for one target group, the elderly. 69

PAGE 81

Table 5.2. Summary of Descriptive Data of Dependent Variables Per 1 000 State Elderly Population, By State, 1981 1988 I I High I Low I Mean Total Medicaid spending on long-term 893,313 133,650 321,194 care New York Florida Medicaid spending on nursing home 716,725 125 930 313,005 care New York Florida Medicaid recipients of nursing home 81.7 18 9 43. 9 care Minnesota Florida Average Medicaid spending per nursing 15,779 4,625 7,382 home recipient New York Iowa Medicaid expenditure on home care 173,111 210 10,192 New York Wyoming Medicaid recipients of home care 45.6 .3 6.2 New York Wyoming Average Medicaid spending per home 4,009 162 1,ll5 care recipient New York Mississippi Medicaid spending on 2176 Waiver 44,080 -05,079 Program Oregon 9 States Percent Medicaid long term care 100.00 79.3 96 2 spending on nursing home care Wyoming Oregon Michigan Private spending on nursing home care 929,603 155,124 425,204 Nebraska Alabama 70 I

PAGE 82

Independent Variables Table 5.3 lists national, yearly averages for the independent variables utilized in the study. Demographic and economic trends are obvious from the data. The aging of the population is evident from the increase in the number of persons over 85 per 1,000 persons over 65; the number of persons over 65 per 1, 000 persons over 18; and the decline in the average number of persons per household. The percent elderly living in standard metropolitan statistical areas was remarkably stable, with a slight rise from 1981 to 1988. Economic indicators, i.e., average state income and wages in the nursing home industry dipped in the early 1980s and rose in the later 1980s, conversely, the unemployment rate rose in the early 1980s and declined in the late 1980s. This is consistent with the economic recession of the early 1980s. The unit cost of LTC to state taxpayers rose by approximately 2.5%, from 1981 to 1988, in real dollars Policy variables indicate heightened generosity in Medicaid programs toward the elderly in the later 1980s with states increasing the number of optional Medicaid benefits to the categorically needy and more states including the elderly in their optional medically needy program. This finding is consistent with findings of Chang and Holahan (1990), who documented increased generosity in state Medicaid programs across many service areas and eligibility groups from 1984 through 1987, inclusive of LTC. This increased generosity is also reflected in Medicaid reimbursement rates as the real average Medicaid per diem rate for nursing home care rose more than 50% from $36.9 in 1981 to $56.6 in 1988. The nursing home bed stock has diminished slightly, on a per 1 000 state elderly population basis, from 56.00 in 1981 to 55 .61 in 1988, with a low of 55.24 in 1986. The average LOS for the Medicare population in an acute-care hospital declined from 10.08 in 1981 to 8.09 in 1985 to approximately 8.30 in 1988. This trend demonstrates the effects of Medicare s prospective payment system that was implemented in 71

PAGE 83

-...) N Table 5 3. Descriptive Data of Independent Variables By Year, 1981 1988, in 1982 Dollars Mean (Standard Deviation) 1981 1982 1983 1984 1985 Wages in the nursing home industry 157.6 154 2 155.2 157 1 156 3 (21.10) (18 .3) (20.0) (22.9) (22 .9) Old 85)/elderly 65) 90.8 92. 5 93. 5 94.6 95. 3 populati o n (13.4) (13.9) (14.1) (14.4) (15.0) Elderly voting population 154 1 155.4 156 8 158 3 159.8 65)/voting 18) populat i on (28.0) (28.2) (28.0) (27.6) (27 0) Average state income 19,034 18, 869 18,390 19, 025 19, 233 (2,121) (2,028) (2,093) (2 240) (2,344) Unit cost of long-term care to state 59.9 58 7 60.0 60.2 59.3 taxpayers (18.2) (16.5) (17.4) (19.9) (19.0) Percent state unemployment 7.3 9.2 9.3 7.3 7 1 (1.9) (2.3) (2.6) (2.2) (2.0) Percent state elderly population living 58.3 58.3 58.2 58.3 58.4 inSMSA (24.4 ) (24 4) (24.4) (24.3) (24.3) Average persons per household 2 .73 2.72 2.72 2 .69 2 .67 ( 13) (.13) (.14) (.14) (.14) Average state temperature 52.9 51.8 52.3 52.4 51.8 (7.9) (9 .1) (7.9) (8. 5) (9. 0) Year Elderly included in "medically needy" .60 .60 .58 .60 .58 Medicaid eligibility category ( 49) (.49) (.50) (.49) ( .SO) Number of optional benefits for elderly 17.4 18.0 18.3 18. 6 in state Medicaid Program (6. 3) ( 5 8) (5.6) (5.5) Medicaid per diem nursing home 35. 9 39 0 42.9 46.3 48 9 reimbursement rate (9.4) (10 7) (16 9) (19 0) (20.0) --1986 1987 1988 158.4 158 5 163.5 (25 2) (25 .6) (32 7) 95. 9 96 7 97.5 (15.0) (15.3) (15.4) 161.9 163 9 165 3 (27 1) (26 .5) (26.2) 19,700 20,171 20,877 (2,567) (2,686) (3, 037) 60.5 59.8 61.2 (20 .1) (20.1) (23.8) 7 0 6.3 s.s (2.3) (2.2) (2.0) 58.5 58.7 58.8 (24.2) (24 1) (24.1) 2.66 2.64 2.61 ( 13) (.14) ( 12) 53.3 52.6 52.4 (8.7) 8.5) (8.1) .64 .66 .68 (.48) (.48) (.47) 18. 8 19.4 19.8 (5.4) (5 2) (5.3) 50.4 53.2 56.6 (20.6) (25 .9) (26 2)

PAGE 84

-....I VJ Table 5.3. (Cont.) Descriptive Data of Independent Variables, By Year, 1981 1988, in 1982 Dollars Mean (Standard Deviation) 1981 1982 1983 1984 1985 1986 State/local administration of Medicaid .12 .12 .12 .12 .12 .12 Program (.33) (.33) (.33) (.33) ( 33) ( 33) Number of nursing home beds/! ,000 56 00 55. 70 55.50 55. 36 55.25 55. 24 elderly population (17. 68) (17.45) (17. 08) (17.20) (17.46) (16.97) Medicare acute care hospital length of 10. 08 9 88 9 56 8 .65 8 09 8.20 stay (1.78) (1.68) (1.54) (1.49) (1.45) (1.25) Liberal quotient of Congressional 37 5 41.4 45. 6 46 2 41.2 42.2 voting record (20.1) (22 .6) (19.8) (21.2) (20.1) (20.8) 1987 1988 .12 .12 (.33) ( 33) 55.48 55.61 (16. 70) (16.56) 49 9 50.2 I (20 .3) (20.6)

PAGE 85

October of 1983. The liberal climate of the state, as measured by Congressional voting records, actually increased from 1981 to 1988. Wide variations in these variables exist across states. Table 5.4 summarizes the high and low outliers for each independent variable. Clearly, the United States is a vast country with wide and divergent state preferences and capabi.lities, indicating a greater federalism than one might otherwise expect. Correlation Coefficients Economic state indicators, i.e., wages in the nursing home industry, average income, unit cost of long-term care to state taxpayers, and percent unemployment are strongly correlated in expected direction (Table 5.5). Economic indicators are also strongly, positively correlated with Medicaid nursing home per diem rate, and surprisingly, with average LOS of the state Medicare population in acute-care hospitals. Rational, economic thought would suggest that suppliers of nursing home care, 80% of whom are "for profit" concerns, would increase their bed stock in areas with higher reimbursement rates (Scanlon and Feldstein, 1988a). Surprisingly, states with higher per diem Medicaid nursing home reimbursement rates have fewer nursing home beds per 1,000 state elderly population (-.30). Regulatory agencies in states with high per diem rates may have greater incentive to contain Medicaid costs by limiting nursing home bed stock. Limiting nursing home bed stock may delay discharges from acute-care hospitals for the elderly who are seeking nursing home care, thereby increasing the average LOS for the Medicare population in acute-care facilities. Surprisingly, the correlation coefficient between acute-care hospital LOS fo. r the Medicare population and nursing home bed stock is -.13 only. A stronger, positive correlation may be found between state 74

PAGE 86

t Table 5.4. Summary of Descriptive Data of Independent Variables, By State, 1981 -1988 I I High I Low I Mean I Wages in the nursing home industry (1982 243.40 130 40 157 0 dollars) New York Arkansas Old elderly ( 85)/elderly ( 65) 125.3 57 1 94.0 population Nebraska Nevada Elderly voting 65)/voting 227.6 48.0 162.8 18) population Florida Alaska Average state income (1982 dollars) 24,848 14,825 19,379 Connecticut South Dakota Unit cost of long-term care to state 110.60 28.33 59 .95 taxpayers (1982 dollars) New York Mississippi Percent state unemployment 12.89 4.21 7.3 West Virginia New Hampshire Percent state elderly populat ion living in 100 15.2 56 6 SMSA D.C. Vermont Average persons per household 3 20 2.36 2.69 Utah D.C. Average state temperature 77.5 37.5 52.5 Hawaii Alaska Year Elderly included in "medically needy" 1.00 (yes) 00 (no) .62 Medicaid eligibility category 29 States 21 States Number of optional benefits to elderly in 28 3 6 9 18 6 state Medicaid Program Minnesota Wyoming Massachusetts Medicaid per diem nursing home 87.83 31.66 45.74 reimbursement rate (1982 dollars) D C. Louisiana State/local administration of Medicaid 1. 00 (Local) .00 (State) .12 Program 6 States 44 States Number of nursing home beds/1,000 86.1 24.1 55.6 elderly population Wisconsin Florida Medicare acute care hospital length of stay 13.5 6.7 8.96 New York Idaho Liberal quotient of Congressional voting 84.9 6 88 44 0 record Massachusetts Utah Wyoming 75

PAGE 87

-.1 0\ Table 5.5. Correlation Matrix -Wages in the nursing home industry Old 85)/elderly 65) pop. Eld 65)/voting 18) pop. Average state income Unit cost of LT care to state taxpayers Percent state unemployment Percent state elderly population in SMSA Average persons per household Average state temperature Year Elderly included in "medically needy" Medicaid eligibility category Number of optional benefits to elderly in state Medicaid Program Medicaid per diem nursing home reimbursement rate State/local administration of Medicaid No of nursing home beds/1,000 eld. pop. Medicare acute care hospital length of stay Liberal quotient of Congress'! vote record ---Wages in NH Industry 1.00 -.08 -.04 .63 .84 -.33 .16 .03 -.09 .02 .24 .32 .51 .27 -.28 .54 .38 Elderly Voters/ Old/Elderly Voting Elderly Population -.08 -.04 1.00 06 .06 1.00 -.16 08 02 -.09 -.32 .21 -.36 07 -.41 .12 -.51 02 .15 .02 .10 07 .22 .02 -.01 -.02 .10 -.02 .63 -.07 -.1 I -.02 .21 .13 -----Unit Cost Average of LTC to Un#Persons State State EmployElderly in Per Ave. Income Taxpayers ment Rate SMSA Household Temp. .63 .84 -.33 .16 .03 .09 -.16 02 -.32 -.36 -.41 -.51 -.08 -.09 .21 -.07 -.12 -.02 1.00 73 -.37 .56 -.13 -.05 .73 1.00 -.31 .63 12 -.12 -.37 -.31 1.00 -.05 .12 .17 .56 .63 -.05 1.00 .03 .23 -.13 -.12 .19 03 1.00 .27 .05 .12 .17 .23 27 1.00 .21 -.01 -.23 -.03 -.24 .01 .17 .18 .17 .02 .0 6 -.23 .30 .32 -.16 .12 -.19 -.37 .34 .54 39 .19 06 .05 .03 .21 -.07 -.19 .10 .18 -.13 -.04 -.15 -.15 .2 6 -.38 .43 .47 -.10 .24 -.04 .03 .24 .38 -.10 .23 -.32 -.22

PAGE 88

Table 5.5. (Cont.) Correlation Matrix Optional Medicaid State/Local #of Nursing Medicare Elderly in Medicaid NHPer AdministraHome Beds Acute Liberal "Medically Benefits Diem tion of 11,000 Hospital Quotient of Year Needy" to Elderly Rate Medicaid Elderly LOS Congress Wages in the nursing home industry .02 .24 .32 .51 .27 -.28 .54 .38 Old 85)/elderly 65) pop. .15 .10 .22 .01 .10 .63 .11 .21 Eld. voting pop. 65)/voting 18) pop .02 .07 .02 -.02 -.02 -.07 02 .13 Average state income .21 .17 .30 .34 .03 -.13 .43 .24 Unit cost of long-term care to state taxpayers -.00 .18 .32 .54 .21 -.04 47 .38 Percent state unemployment .23 -.17 -.16 -.39 -.07 -.15 .10 -.10 Percent state elderly population in SMSA -.03 02 .12 .19 .19 .15 .24 .23 -..l Average persons per household 24 -.06 -.19 .06 .10 -.26 -.04 32 -..l Average state temperature .01 -.23 .37 .05 -.18 -.38 03 22 Year 1.00 09 07 .42 -.06 -.03 39 .12 Elderly included in "medically needy 09 1.00 .46 27 .15 .07 .13 .33 Medicaid eligibility category Number of optional benefits to elderly in state .07 .46 1.00 .23 27 .21 .14 .40 Medicaid Program Medicaid per diem nursing home .42 .27 .23 1.00 .10 -.30 .03 .39 reimbursement rate State/local administration of Medicaid Program 06 .15 .27 .10 1.00 .04 20 .13 Number of nursing home beds/1 ,000 eld. pop. -.03 .07 .21 .30 .04 1.00 .13 .07 Medicare acute care hospital length of stay -.39 .13 .14 .03 .20 -.13 1.00 .35 Liberal quotient of Congress'! vote record .12 .33 .40 39 .13 .07 .35 1.00

PAGE 89

Medicare population LOS in acute-care hospitals and nursing home occupancy rates. Nursing home occupancy rates were not included in this study. States with a higher average income have a greater concentration of their elderly population living in urban areas (+.56). This positive relationship is expected to hold true for the general state population. States with a greater concentration of "old" elderly (age 85 or older) in their elderly concentration have less unemployment, (-.32); have a smaller percentage of their elderly population in urban areas, (-.36); have fewer people per household, (-.41); have more nursing home beds per 1,000 elderly population, (.63); and, have lower average temperatures, (-.51). A strong, positive correlation between nursing home bed stock and the number of "old" elderly was expected as the "old elderly" are at greater risk for physical and cognitive disabilities predisposing them to be in institutional (nursing home) care (Rivlin & Weiner, 1988; Liu, Coughlin and McBride, 1989). State planning bodies reportedly consider the demographics of the state population and the stratification of the elderly state population in approving additional nursing home bed stock (HCIA, 1991). States with a greater number of optional Medicaid benefits for the elderly are more apt to include the elderly in their optional "medically needy" eligibility category ( + .46), and are more likely to be high wage, high income states, +.32 and +.30 respectively. Colder states have fewer optional Medicaid benefits for the elderly (. 37) and have a smaller nursing home bed stock per 1,000 elderly population (-.38). Liberal states tend to be more "expensive" states with higher wages ( + .38), higher effective cost of long-term care to state taxpayers ( + 3 8), and higher Medicaid per diem rates ( + .39). Liberal states appear to be more generous in terms of Medicaid eligibility for the elderly via the "medically needy" program ( + .33), and optional Medicaid benefits for the elderly ( + .40). 78

PAGE 90

Liberal states also tend to have longer lengths of stay in acute-care hospitals for their Medicare population ( + .35); and fewer persons per household (-.32). Regression Analysis The results of the regression analysis are presented by dependent variable. As previously stated, each dependent variable was examined via four equations, reduced and expanded models run inclusive and exclusive of New York. New York is a clear outlier state regarding Medicaid spending on LTC for the elderly, particularly concerning home care. The state of New York accounted for 80% of the national Medicaid expenditure on home care in 1986 (Swan, 1990). Dependent Variable The dependent variables in this study are expenditure amounts for long-term care services for the elderly per 1,000 state elderly population, percent of Medicaid spending on LTC for the elderly devoted to nursing home care, elderly recipients of state long-term care services per 1,000 state elderly population, or average expenditure per elderly recipient of state long-term care services. Each category is clearly delineated. Two models, reduced and expanded, are used to explain interstate variations in long-term care spending for the elderly in this study. The reduced model includes state socio-economic, demographic, and temperature/year measures as independent variables. The expanded model adds variables specific to a state's Medicaid program, health-care system, and political climate. Models were also run exclusive of New York. 79

PAGE 91

Total Medicaid Expenditure on LongTerm Care for the Elderly and Medicaid Expenditure on Nursing Home Care for the Elderly Total Medicaid spending on LTC for the elderly per 1,000 elderly population and Medicaid spending on nursing home care for the elderly per 1,000 elderly population are discussed together as regression analyses produced nearly identical results for each dependent variable. This is not surprising as nursing home expenditures account for well over 90% of total Medicaid spending on LTC for the elderly. Previous theoretical and empirical work suggest that state Medicaid spending on nursing home care for the elderly is associated with state economic variables (wages in the nursing home industry [ + ], average state income [ +], state unemployment [-), and unit cost of LTC to state taxpayers [-]), demographic variables ("older" voting and elderly population [ +] and household size [-]), structure of the Medicaid program (elderly in the "medically needy" program [ +], Medicaid per diem nursing home reimbursement rate [ +], and local administration of the Medicaid program [ + ]), medical care market place variables (nursing home bed stock [ +] and average length-of-stay of Medicare population in acute-care hospitals [-]), and finally, a political variable (liberal state ideology [ + ]). These findings form the hypotheses relevant to "total Medicaid spending on LTC for the elderly" and "Medicaid spending on nursing home care for the elderly" found summarized in Table 4.1. The explained interstate variation in total Medicaid (nursing home, home care, and waiver 2176) spending per 1,000 state elderly population for years 1981-1988, utilizing the reduced model is 59%, while the expanded model explains 78% of the variation across states (Table 5.6). The explained interstate variation in Medicaid spending on nursing home care per 1, 000 elderly population for years 1981-1988 utilizing the reduced model is 56%, while the expanded model explains 79% (Table 5.7). 80

PAGE 92

00 -Table 5.6. State Medicaid Spending on Long-Term Care for the Elderly Per 1,000 Elderly Population, 1981 1988 -------I I Including New York I Excluding New York B T B T B T B Wages in the nursing home industry 5,974.4 13.8 6,934 7 15.1 4,430 9 9.7 4,701.5 Old 85)/elderly 65) population 5,193.0' 12.0 902.7 1.7 4,636.6 11.2 518.1 Eld. voting pop. 65)/voting 18) pop. -220.2 -1.6 -356.s' -3.2 -204.6 -1.6 -293.9 Average state income -5.1 -1.3 -5.9 -1.5 -4.1 -1.1 4.2 Unit cost of long-term care to state taxpayers -2,671.2 -4.2 -3,290 1 -6 .2 -2,169 3 -3.6+ -4,481.1 Percent state unemployment 1107.8 .4 7,055 2 2 .8+ -2291.0 -.90 2,806.2 Percent state elderly population living in SMSA 758.7 2.2+ 61.1 .20 826.9 2.5 283.9 Average persons per household 46,703.0 1.0 -56,642.9 -1.3 56,648.7 1.3 -34,339.4 Average state temperature -223.3 -1.7 -159.3 -1.6 -155.8 -1.3 -90.0 Year 537 0 .2 1,290.2 .4 66.7 .0 -12,401.8 Elderly included in "medically needy" 28,634.7 2.8+ 18,167.3 Medicaid eligibility category Number of optional benefits to elderly in state -2,825.6 -2.8 -2,614.0 Medicaid Program Medicaid per diem nursing home reimb rate 2 150.4 3 .2+ 5,066 6 State/local administration of Medicaid Program 93,496.8 6 .3+ 46,099.9 No. of nursing home beds/1 ,000 eld. pop. 4,161.9 9.9+ 4,624.9 Medicare acute care hospital length of stay 8,510.2 2 .0+ -39,770.0 Liberal quotient of Congressional voting record 547 8 1.9+ 1,233.9 Adjusted R2 .59 78 .44 + Statistically significant at p = !0: .OS I T 10.3 1.1 -3.1 1.2 -8.4+ 1.3 1.0 -.90 -1.0 -4 1 + 2.0 -3.0+ 7 .8+ 3.4+ 12.6+ -1.0 4.7+ 73

PAGE 93

00 N Table 5.7. State Medicaid Expenditure on Nursing Home Care for the Elderly Per 1,000 Elderly Population, 1981 1988 I I Including New York I Excluding New York I B T B T B T B T Wages in the nursing home industry 4,972.5 12.6"' 5,919.8 15.2"' 4,306.7 9.7"' 4 651.9 10.9"' Old 85)/elderly 65) pop. 4,900.9 12.3"' 1,034.3 2.3"' 4,660.1 11.6"' 796.7 1.9"' Elderly voting pop. 65)/voting 18) pop. -178.1 -1.4 -289.4 -3.1"' -171.9 -1.4 -252.6 -2.8"' Average state income -4.8 -1.3 1.5 -.4 -4.2 -1.2 4.7 1.4 -Unit cost of long-term care to state taxpayers -2,213.1 -3.8"' -4,050.4 -7.8"' -1, 993.4 -3.4"' -4,496.5 -9.0"' Percent state unemployment -386.0 -.16 6,507.4 3.0"' -1, 827.5 -.76 4 052.0 1.9"' Percent state elderly population in SMSA 751.1 2.4"' 164.6 .6 768.2 2.5"' 284.4 1.1 Average persons per household 76,956.4 1.9 -14,406 7 -.4 81, 281.6 2 .0"' -3, 082.9 .10 Average state temperature -195.1 -1.6 -126.9 -1.5 -166.4 -1.4 -86.7 -1.1 Year -2338.1 -1.0 -6,804.7 -2.6"' -2 452.2 -1.0 -14,874.5 -5.2"' Elderly included in "medically needy" 29 806 8 3.4"' 23,494 1 2 .8"' Medicaid eligibility category Number of optional benefits to elderly in state -3,197.1 -3.7"' -3, 108 9 -3.8"' Medicaid Pgm. Medicaid per diem nursing home reimb rate 3,460.7 6.2"' 5,195.9 8.6"' State/local administration of Medicaid Pgm 75,761.4 6.2"' 48 768.9 3.9"' No. of nursing home b e ds/1,000 eld. pop. 4,432.9 12.4"' 4,720.2 13.8 Medicare acute care hospital length of stay 5,976.2 1.6 1 717.1 -.50 Liberal quotient of Congre s s'! voting record 691.9 2.8 1,096.4 4.5 Adjus ted R2 .56 79 .44 .75 "' Statistically significant a t p = .05

PAGE 94

Clearly macrocontextual variables of socio-economy and demography play a fundamental role in driving nursing home and aggregate Medicaid expenditures on LTC for the elderly, while policy/political variables (such as number of nursing home beds per 1,000 elderly and Medicaid per diem nursing home reimbursement rate) exert a decidedly smaller, but distinct influence. The statistically significant economic variables in the expanded model are nursing home wages ( + ), the unit cost of LTC to state taxpayers ( -), and percent state unemployment ( + ). This study is consistent with Swan (1990), who also found nursing wages to be positively related to Medicaid spending on nursing home care, but it runs contrary to the findings of Harrington and Swan (1987) who found nursing wages NOT to be statistically related to Medicaid expenditures on the elderly when Medicaid per diem nursing home rates were included in the model (1987). Tills study found both nursing wages and Medicaid nursing home per diem rate to be positively related to Medicaid spending on nursing home care for the elderly. The impact of the nursing home wage variable reflects two important concepts, namely, the labor intensive nature of the LTC industry (labor accounts for approximately 75-80% of the input costs of nursing home care) and geographic cost of living (Swan, Harrington & Grant, 1990; HCIA 1990). Buchanan (1987) found that differences in cost of living among states explained upwards of 60% of the variance in Medicaid payments for skilled and intermediate nursing home care. Buchanan et al. (1991) then developed a cross-state price index to deflate expenditure and per capita income for their model explaining state Medicaid spending. This approach aligns expenditures more closely with benefits. This dissertation adjusted nominal dollars into real 1982 dollars, but did not adjust for regional variations in the cost of living. Expenditure levels should, therefore not be equated with LTC service level. 83

PAGE 95

The variable ".unit cost of LTC to state taxpayers" is heavily influenced by regional differences in cost of living as state wages in the nursing home industry is utilized in the calculation. The correlation between wages in the nursing home industry and unit cost of LTC to state taxpayers is + .84. A negative relationship was hypothesized between unit cost of LTC to state taxpayers and Medicaid nursing home recipient and average expenditure measures as rational economic theory suggests that state taxpayers would be willing to "buy more" LTC, either by increasing expenditures per recipient or increasing number of recipients as unit cost declined. The unit cost of LTC to state taxpayers, as calculated in this study, is affected not only by state nursing wages in the nursing home industry, but also by the percent state income derived from corporate taxes and the FMMP applied to state Medicaid programs. As such, the FMMP directly affects the unit cost of LTC to state taxpayers, which in turn relates inversely to Medicaid spending on LTC for the elderly, i.e., the higher the FMMP to the state, the lower the unit cost of LTC to state taxpayers and the higher the state Medicaid expenditure on LTC for the elderly. The federal government's FMMP formula uses state per capita personal income to determine the Federal Medicaid Matching Rate paid to states to subsidize their Medicaid programs. The formula has been criticized on several counts. Compare Grannemann (1979), who argues that the FMMP is inadequately weighted toward poor states, with Blumberg etc. (1993), who argue that the FMMP fails to capture differences in the cost of living across states thus overstating the buying power of high cost states. Once again, the important distinction between expenditure level and service level is underscored, while the affect of the FMMP on states' Medicaid LTC expenditure level appears quite real. Percent state unemployment is statistically significant with a positive coefficient in explaining Medicaid spending on nursing home care for the elderly and total Medicaid spending for 84

PAGE 96

the elderly when New York is included in the analysis. When New York is excluded from the analysis, percent state unemployment remains statistically significant and positive relative to Medicaid nursing home care spending for the elderly, but is no longer statistically significant relative to total LTC spending for the elderly. A negative relationship between percent state unemployment and Medicaid spending on total LTC and nursing home care for the elderly was hypothesized. Percent unemployment was included in the model as a measure of the fiscal capacity of the state, i.e., states with less unemployment would have more revenue to pay for LTC for the elderly. It appears that percent unemployment is acting as a demand variable in this model, whereby states with higher unemployment have a greater demand for Medicaid nursing home services for the aged. State employment conditions may diminish the possibility of some elderly finding work and may also diminish financial support from family members, thus increasing demand for publicly Slibsidized nursing home care. Percent state unemployment is also statistically significant, in a positive direction, regarding number of Medicaid nursing home care recipients. Previous empirical work does support the notion that higher state unemployment increases demand for Medicaid services in the general state population due to enhanced eligibility (Harrington & Swan, 1987). Three demographic variables are statistically significant in total Medicaid spending on LTC for the elderly per 1,000 state elderly population and Medicaid nursing home spending for tlie elderly, namely, the percent of the elderly population residing in metropolitan areas ( + ), the number of eligible elderly voters (age 65 or over) per 100 eligible general voting population (age 18 or over) (-), and the number of old elderly (age 85 or over) per 100 elderly population (age 65 or_ older) ( + ). Harrington and Swan (1987) found states with a higher percentage of the population residing in urban areas have higher Medicaid nursing home expenditures. This study found a 85

PAGE 97

statistically significant relationship between percent elderly residing in urban areas and Medicaid expenditure on nursing home and total LTC spending for the elderly in reduced models only. When health care and political variables were added to the model, the percent elderly population living in urban areas was no longer found to be statistically significant. It is of interest to note that the percent elderly population residing in urban areas is statistically significant with a positive coefficient relative to the number of elderly Medicaid recipients of nursing home care, and statistically significant with a negative coefficient relative to average Medicaid expenditure per nursing home care recipient in the expanded models. The net affect regarding Medicaid expenditure on nursing home care appears to be neutral in the expended model. The ratio of number of persons age 65 or older per 100 persons age 18 or older was intended as a surrogate measure of a potential voting bloc for the elderly. As such, higher Medicaid expenditures on nursing home care and total LTC was expected in states with a higher ratio of potential aged voters to the general eligible voter population. A negative relationship was not expected. This variable may in fact be operating as an economic indicator, whereby states with a lower aged to working population ratio may have a greater fiscal capacity to support nursing home and LTC services for the elderly. The number of old persons (age 85 or older) per 100 state elderly population (65 or older) is statistically significant in explaining state variation in nursing home care and total Medicaid spending per 1,000 state elderly population in the reduced models, but remains statistically significant in the expanded model for Medicaid spending on nursing home care for the elderly only, with a greatly reduced T statistic. The explanatory power of the variable appears to be diluted in the expanded model when the number of nursing home beds per 1,000 elderly population is included in the analysis. The correlation between the number of old persons per 100 elderly persons and the number of nursing home beds per 1,000 elderly is + .63. 86

PAGE 98

Given the assumption that old persons do not migrate to states with higher home bed stock, it is presumed that the indigenous number of old persons is driving the number of nursing home beds. There are separate but intersecting paths whereby the number of old persons will drive nursing home bed stock. First, the nursing home industry is heavily weighted toward the for-profit sector, with approximately 75% of nursing homes proprietary organizations (Feldstein, 1988a). Conventional business strategy dictates that, within the confines of government regulation, for-profit nursing home beds are positioned relative to potential markets, i.e., supply of nursing home beds is market driven. The market for institutional LTC services is predominantly the over 85 population. Market researchers for proprietary nursing home organizations attend to demographics that relate to demand. A second explanation for the strong relationship between the number of old persons in a state and the number of nursing home beds involves government regulation. Many states with certificate-of-need legislation intact consider allowing an increase in nursing home bed stock based, in part, on the current and projected ratio of nursing home beds per 1,000 state population over the age of 65. Thirty-seven states had some variant of CON regarding the addition of LTC beds as of December, 1988 according to The Guide To The Nursing Home Industry published by Health Care Investment Analysts, Inc. and Arthur Anderson (1990). On the surface, the supply of nursing home beds as an explanatory variable of total Medicaid LTC expenditures and nilrsing home Medicaid expenditures (and of the number of recipients of nursing home care, discussed later) would seem to support Roemer's Law. Roemer's Law states that supply of health-care services creates demand for those services (Feldstein, 1988a). Nyman (1985) argues that the quality of care provided by nursing homes is lower in markets where there is excess Medicaid demand than in areas where excess demand does not exist. He postulates that high occupancy rates due to excess demand constrain market competition and 87

PAGE 99

patient care expenditures and eliminate information important to residents concerning the quality of care at an institution. He postulates that providers of nursing home care in under-bedded areas, can reduce patient-care expenditures and quality of care without census repercussions because there is no competition for Medicaid nursing home residents. Davis and Freeman (1994) suggest that reducing excess demand may not always lead to increases in patient-care expenditures or quality of care. This study supports the long-held assertion that increasing nursing home bed stock increases Medicaid nursing home expenditures. This relationship has not been lost on policy makers, who have been charged with restraining nursing home bed supply in an attempt to control Medicaid LTC costs (Harrington & Swan, 1988; Feder & Scanlon, 1981; Feldstein, 1988a). The structure and administration of the Medicaid program is statistically significant at the .05 level in explaining state variation in Medicaid spending on nursing home care and total Medicaid spending on LTC for the elderly on a 1,000 elderly population basis. Including the elderly in the optional Medicaid medically needy program increases the demand for nursing home care services. This finding is further addressed in the nursing home recipient data section. A statistically significant, negative relationship is found between the number of optional Medicaid benefits offered to the aged categorically needy and Medicaid expenditure on nursing home care and aggregate Medicaid spending on LTC for the elderly on a per 1,000 elderly population basis. This relationship is mediated through a negative relationship between optional Medicaid benefits and the number of elderly Medicaid recipients of nursing home care, as such it is discussed in that section. The level of administration of the Medicaid program, state versus local, is positive and statistically significant in explaining total Medicaid expenditure on LTC for the elderly and Medicaid expenditure on nursing home care for the elderly. This finding is consistent with 88

PAGE 100

Schneider (1988) and Buchanan (1991), who found local Medicaid administration to be positively related to overall Medicaid spending. The principal mechanism for the positive relationship between local administration of the Medicaid program and Medicaid nursing home expenditure levels is via local administrative effect on the number of Medicaid recipients of nursing home care that drives overall expenditure on nursing home care, which in turn drives total Medicaid spending on LTC for the elderly. This is further discussed in nursing home recipient section. The per diem Medicaid nursing home reimbursement rate is positively associated with both Medicaid nursing home care spending for the elderly and total LTC spending for the elderly. Conventional wisdom and previous empirical evidence (Harrington & Swan, 1987; Davis & Freeman, 1994) suggest that higher Medicaid per diem rates lead to higher average expenditures per nursing home recipient and higher overall nursing home and total Medicaid LTC expenditures, unless number of -recipients decline to offset the increase. This study suggests that, when all other factors are held constant (particularly nursing home bed stock), higher Medicaid nursing home per diem rates increase both average Medicaid expenditure per nursing home care recipient and number of Medicaid nursing home care recipients per 1,000 elderly population. A statistically significant relationship was established between the average LOS for the state Medicare population in acute-care hospitals and total state Medicaid spending on LTC for the elderly when New York was included in the analysis only. It appears that New York's spending on home care is driving the finding, as no statistically significant relationship was established between average LOS of the state Medicare population in acute-care hospitals and state Medicaid spending on nursing home care for the elderly, with New York either included or excluded from the analysis. A negative relationship was hypothesized as studies by Shaughnessy and Kramer (1990) and Liebig (1988) have demonstrated an increase in the number of LTC recipients and services since Medicare's PPS was implemented as PPS decreased the average LOS 89

PAGE 101

of the Medicare population in acute-care hospitals. Interestingly, a statistically significant, negative relationship was found between average state Medicare LOS in acute-care hospitals and number of elderly Medicaid recipients of nursing home care, while a statistically significant, but positive relationship was found between average LOS for the state Medicare population in acute-care hospitals and average Medicaid expenditure per nursing home care recipient. It appears that average LOS of the state Medicare population in acute-care facilities has important but opposing interactions relative to nursing home recipient and average expenditure data. These findings are discussed in forthcoming sections. The liberal quotient of the state, as measured by scoring Congressional voting records by the ADA is positive and statistically significant relative to both total Medicaid expenditure on LTC for the elderly and Medicaid expenditure on nursing home care. The political climate of a state does appear to exert an independent influence on LTC expenditures for the elderly. This observation supports the findings of Carrell (1990) and Barrilleaux and Miller (1988) and refute the findings of Buchanan et al. (1991). Critics may argue, however, that the liberal voting record of state representatives may correlate with socio-economic or demographic variables of the Medicaid or general state population that are simply omitted from the model. The finding that the liberal quotient of a state was statistically significant may support Carrell's contention that LTC has moved slightly toward the redistribution area of public policy. A statistically significant, negative relationship was found between year and Medicaid expenditure on nursing home care per 1,000 elderly population. This indicates that when other variables are controlled for in the analysis, states spent more dollars (adjusted to 1982 dollars) on nursing home care for the elderly in the early 1980s as opposed to the later 1980s on a 1,000 elderly population basis. This was mediated by fewer elderly Medicaid nursing home care recipients on a 1,000 elderly population basis in later years 90

PAGE 102

Elder Medicaid Recipients of Nursing Home Care The explained variation in the number of elderly Medicaid recipients of nursing home care per 1,000 elderly population for years 1981-1988, via the reduced model is 23%, while the expanded model explains 66% of the variation across states (Table 5 8). This is, by far, the greatest discrepancy in explained variation between the reduced and expanded models for any dependent variable. Clearly, policy /political variables exert a strong and distinct influence on the number of recipients of nursing home care. All seven policy/political variables are statistically significant at the .05 level in explaining the number of nursing home care recipients per 1,000 elderly population when New York is included or excluded from the analysis. It was hypothesized that the number of elderly Medicaid recipients of nursing home care would be positively associated with relatively "older" voting and elderly populations, average state income, nursing lioriie bed stock, inclusion of the elderly in the "medically needy" category, local administration of the Medicaid program, and political liberalism of the state. Furthermore, it was hypothesized that number of elderly Medicaid recipients of nursing home care would be negatively correlated with unit cost of LTC to state taxpayers, percent state unemployment, average persons -per household, average state temperature, number of optional Medicaid benefits to the elderly and average LOS of the Medicare population in acute care hospitals. Policy and political variables will be discussed first as they represent the driving force behind the number of Medicaid nursing home recipients on a 1,000 elderly state population basis. The T -value associated with nursing home bed stock is double that of any other variable in the model. Simply put, the number of nursing home beds allows demand to translate into utilization for the elderly Medicaid population seeking nursing home care (Scanlon, 1980). 91

PAGE 103

Table 5.8. Number of Elderly Medicaid Recipients of Nursing Home Care Per 1,000 Elderly Population, 1981-1988 ---I I Includin.!l New York I Excludin.!l New York I B T B T B T B T Wages in the nursing home industry 07 1.4 37 7.9"' 09 1.5 .39 7.0"' Old 85)/elderly 65) pop .52 10.0"' .01 .11 .53 9.9"' .01 .17 Elderly voting pop. 65)/voting 18) pop. 02 1.5 -.04 -3.1"' 02 1.5 04 -3.1"' Average state income -.00 1.4 .00 -.2 00 -1.3 00 3 Unit cost of long-term care to state taxpayers -.10 -1.4 -.47 -7.5"' -.11 -1.4 -.47 -7.3"' Percent state unemployment .02 .07 .61 2.3"' .08 .24 .64 2.3"' Percent state elderly population living in SMSA .13 3 .1"' .11 3.4"' .13 3.1"' .11 3.3"' Average persons per household 12. 7 2 .3"' 5 5 1.1 12. 6 2 .2"' 5.2 1.1 te Average state temperature 02 1.2 00 -.5 02 -1.2 -.01 5 Year .33 -1.1 1.1 -3.3"' -. 32 1.0 .97 -2 .6 Elderly included in "medically needy" 4.0 3.8"' 4 0 3 .8"' Medicaid eligibility category Number of optional benefits to elderly in state -.38 -3.6"' -.4 3.7"' Medicaid Program Medicaid per diem nursing home reimb rate .19 2.8"' .17 2.2"' State/local administration of Medicaid Pgm. 4.8 3.2"' 5 2 3.2"' Number of nursing home beds/1 ,000 eld. pop. .70 16.4"' .70 16.0"' Medicare acute care hospital length of stay .99 -2.2"' -1.0 -1.9"' ) Liberal quotient of Congress vote record .12 4.1"' .12 3 .8"' Adjusted R2 .23 66 .23 .66 Statistically significant at p = 05

PAGE 104

The inclusion of the elderly in the Medicaid optional medically needy category increases the pool of Medicaid eligible who are frequently in need of intense LTC services. These findings are consistent with previous empirical work (Swan, 1990; Harrington & Swan, 1987). Local administration of the Medicaid program appears to increase access of elderly persons to Medicaid subsidized nursing home care services. This is consistent with the notion that local administrators and bureaucrats are more vested in serving the local patronage. Decentralized management may enhance the responsiveness of bureaucrats to local need (Rosenbloom, 1988). This finding is consistent with those of Schneider and Buchanan (1990) who found a significant, positive relationship between overall Medicaid spending and local administration of the Medicaid program, although their studies did not elucidate the mechanism. A statistically significant, negative relationship is generated between the number of optional Medicaid benefits offered to the aged categorically needy and the number of elderly Medicaid recipients of nursing home care. This supports the finding of Soumerai et al. (1991) who found that limiting reimbursement for effective drugs (an optional Medicaid benefit) put frail, low income elderly patients at increased risk of institutionalization in nursing homes and may have increased Medicaid expenditures. Targeting certain optional benefits or eligibility groups may decrease the number of nursing home care recipients and help contain Medicaid expenditures on LTC. This cost containment strategy may not be a given as many variables have behaved in counterintuitive ways. It has not yet been determined which optional Medicaid benefits act as substitutes for institutional care and which act as complements to institutional care. Results discussed thus far are consistent with the hypothesis. Medicaid per diem nursing home care rate was found to be statistically significant in explaining number of Medicaid nursing home care recipients with a positive coefficient. 93

PAGE 105

Harrington and Swan (1987) did not find these variables to be related in a statistically significant fashion. Rational economic theory would suggest that providers would locate in areas with higher reimbursement rates, so higher rates would entice nursing home investors to increase nursing home bed stock within the confines of government regulation. Ironically, the correlation between Medicaid nursing home per diem rate and nursing home bed stock was -.30. Scanlon and Feder (1980) have argued that it is the supply of nursing home beds, and not Medicaid reimbursement rate that is the essential barrier to access of the Medicaid population to nursing home care This study implies that both Medicaid nursing home per diem rate and nursing home bed supply influence access of the Medicaid population to nursing home care. Higher Medicaid per diem nursing home reimbursement rates may enable Medicaid recipients to better compete with Medicare and private-pay nursing home care recipients for a restricted bed supply. A more liberal state political climate appears to increase access of the Medicaid elderly to nursing home care This is consistent with Carrell's finding that there is at least some movement of LTC toward the redistributive end of the policy continuum The average LOS for the state Medicare population in acute-care hospitals is statistically, negatively related to number of Medicaid nursing home care recipients. A negative finding was hypothesized as Medicare's PPS, introduced in October 1983, has created incentives to reduce the LOS for the Medicare population in acute-care hospitals, which in turn has created an increased demand for post-acute care health services (Miller and Saunders, 1993). Post-acute hospital care is delivered at nursing facilities, rehab hospitals, and at home via home health care. Studies have shown an increase in the number of recipients of LTC services and intensity (average expenditure per recipient) of LTC services since the introduction of Medicare s prospective payment system (Shaughnessy & Kramer, 1990; Liebig, 1988). 94

PAGE 106

The relationship of each of the policy variables found to be significantly related to the number of Medicaid recipients of nursing home care is consistent as hypothesized. The socio-econori:rlc and demographic variables statistically significant regarding the number of Medicaid recipients of nursing home care in the expanded model are: wages in the nursing home industry ( + ); ratio of elderly (age 65 or older) to adult state population 18 or older) (-); unit cost of LTC to state taxpayers (-); percent state unemployment ( + ); and percent elderly population residing in urban areas ( + ). States with higher wages in the nursing home industry more likely to have a greater number of elderly Medicaid recipients of nursing home care per 1,000 elderly state population when all other variables in the analysis are kept constant. Issues of multicollinearity may arise as high-wage states are also associated with higher state income ( + .64), higher Medicaid per diem nursing home rates -(+ :51)," and higher liberal quotient of Congressional voting ( + .38). Ironically, states with higher unit cost of LTC to state taxpayers have fewer elderly Medicaid recipients of nursing home care despite a + .84 correlation between state nursing home care wages and unit cost of nursing home care to state taxpayers -States with a lower ratio of elderly persons (65 and older) to general adult population (18 and older) have a greater number of elderly Medicaid recipients of nursing home care. Once again, this variable appears to be reflecting the fiscal capacity of the state, i.e., states with a relatively greater working to non-working population can generate greater revenues with which to subsidize nursing home care for the elderly. This runs contrary to the hypothesis whereby a greater concentration of voting elders was expected to enhance Medicaid generosity toward the state elderly population. Percent state unemployment rate is acting as a demand variable with higher percent unemployment correlating with greater numbers of elderly Medicaid recipients of nursing home 95

PAGE 107

care per 1,000 elderly population. Once again, this runs contrary to the hypothesized relationship between percent unemployment and elderly Medicaid nursing home recipients that was expected to be negative, with percent unemployment acting as a surrogate measure for the fiscal capacity of the state, not as a demand variable. States with a greater percentage of their elderly in urban areas as opposed to rural areas have more elderly Medicaid recipients of nursing home care when all else is held constant in the analysis. Dubay (1991) found Medicare covered admissions to nursing homes (short term, skilled nursing facility [SNF] benefit admissions) to be 17% lower in rural areas. The 17% lower Medicare SNF benefit admissions to nursing hooies in rural areas is more striking when one considers that rural areas have a higher nursing home bed stock (Dubay, 1991). A portion of the recipients of short term, Medicare covered nursing home days of care continue to receive nursing home care when their Medicare Part A benefit expires, either as private pay or Medicaid subsidized residents. The positive relationship between the percent elderly residing in urban areas and Medicaid recipients of nursing home care per 1,000 elderly population is consistent with the hypothesis. The final statistically significant variable in the expanded model is year, with a negative coefficient When all other independent variables are held constant, states had more elderly Medicaid recipients of nursing home care on a 1 000 elderly population basis in the earlier years of the 1980s. Average Medicaid Expenditure per Elder Recipient of Nursing Home Care The explained variation in average Medicaid expenditure per recipient of Medicaid nursing home care services for years 1981-1988 via the reduced model is 52%. while the expanded model explains 69% of the variation across states (Table 5.9). Excluding New York from the 96

PAGE 108

Table 5.9. Average Medicaid Expenditure on Nursing Home Care for the Elderly Per 1,000 Elderly Population, 1981 1988 I ---I ExcludingNewYork I B T B T B T B T Wages in the nursing home industry 87. 75 13.8* 46.2 5.6* 77.21 11.2* 24.1 2.7 Old 85)/elderly 65) population 19.6 3.0 2.5 .3 17.75 2.7 4.4 .5 Elderly voting 65)/voting 18) pop. .01 .01 -.15 -.2 .17 -.2 09 .1 Average state income .09 1.6 .13 1.9* .02 .43 .13 2.0 Unit cost of long-term care to state taxpayers -22.1 -2.5* -6.1 -.6 -20.8 -2.4 -12.7 -1.3 Percent state unemployment -16.0 -.44 -25.1 -.6 -58.5 -1.6 -74.7 -1.8 Percent state elderly population Jiving in SMSA 12.5 .34 -10.8 -1.9 -3.8 -.76 -16.2 -3.17* Average persons per household 588.41 .40 256.1 .4 940.7 1.5 545.2 .8 :s Averagestatetemperature -1.06 -.10 -11.9 -1.1 -7.7 -.76 -12.2 -1.2 Year -5.15 -.14 -3.8 -.1 6.8 .18 -122.7 -2.1* Elderly included in "medically needy" -150.3 -.9 -210.4 -1.3 Medicaid eligibility category Number of optional benefits io elderiy in stite -10.3 -.6 -6.7 -.4 Medicaid Program Medicaid per diem nursing home reimb. rate 60.0 5.1* 89.9 7.2* State/local administration of Medicaid Program 615.7 2.6* 216.3 .9 Number of nursing home beds/1,000 eld. pop. -11.1 -1.6 -8.0 -1.2 Medicareacutecarehospitallengthofstay 303.1 4.3 -176.9 2.45* Liberal quotient ofCongress'l voting record 7.2 1.5 10.0 2 12* Adjusted R2 .52 .69 .49 .64 .. Statistically significant at p =

PAGE 109

analysis reduces the explained variation via the reduced model to 49% and the expanded model to 64%. Excluding New York also changes the statistical significance of three independent variables, i.e., year(-) and liberal quotient ( +) become statistically significant while local administration of the Medicaid program loses statistical significance. Macrocontextual variables hypothesized to be related to average Medicaid expenditure per elderly nursing home care recipient were economic (wages in the nursing home industry [ + ], average state income [ +], average state unemployment [-], and unit cost of LTC to state taxpayers [-]), demographic (greater concentration of "old" persons in the voting and elderly populations [ +] and percent elderly in metropolitan areas [ +]) and year. Economic variables found to be statistically significant relative to average expenditure per Medicaid recipient of nursing home care are wages in the nursing home industry ( +) across all models and average state income ( +) in expanded models only. Each finding is consistent with the hypothesis. Average state income and nursing home wages each reflect regional cost of living. Average state income also reflects the relative wealth of a state, i.e., its ability to spend more per nursing home resident, while state nursing home wages also gauge input costs of providing nursing home care. This finding is consistent with the finding of Harrington and Swan (1987). Unit cost of LTC to state taxpayers and state unemployment were not found to be statistically significant relative to average expenditure per Medicaid recipient of nursing home care as predicted. The percent of the elderly residing in urban areas is negatively related to average expenditures per Medicaid recipient of nursing home care, i.e., states with a greater percentage of their elderly population residing in rural areas have greater average Medicaid expenditures on nursing home care. This finding runs contrary to the hypothesis. A positive relationship between average Medicaid expenditure per nursing home care recipient and the percent elderly living in metropolitan areas is difficult to explain by level of disability as Dor et al. (1990) found a lower 98

PAGE 110

level of debility in the rural nursing home population, even though non-institutionalized rural Medicare enrollees have been shown to be more debilitated than urban Medicare enrollees (Agency for Health Care Policy Research, 1990). Policy variables that are consistently statistically when New York is excluded or included in the analysis are Medicaid per diem nursing home reimbursement rate ( + ), and average LOS of the state Medicare population in acute-care hospitals ( + ). A positive relationship between Medicaid per diem reimbursement rate and average Medicaid expenditure per elderly Medicaid recipient of nursing home care services was predicted based on the work of Harrington and Swan (1987) who found the Medicaid ICF reimbursement rate to be the strongest predictor of average expenditure per Medicaid nursing home care recipient in years 1978-1983. The hypofuesl.zed -direction of the relationship between average Medicare LOS in an acute care hospital and average Medicaid expenditure on nursing home care was negative. Empirical evidence suggests that the level of disability of the nursing home population has risen with time, most dramatically in the post-DRG era (Carroll & Lewin, 1987; Swan, de la Torre & Steinhart, 1990; Shaughnessy & Kramer, 1990; Rantz, 1991). This increase in disability was expected to increase intensity of services, necessitating an increase in average expenditure per recipient. A positive relationship was found in this study between LOS and average Medicaid expenditure per nursing home recipient. This finding is difficult to reconcile in the post DRG era. States with a disproportionate number of acute-care "outliers" may drive high average post-hospital expenditures. It is also noteworthy that states with a high cost of living as reflected in wages in the nursing home industry and average state income would be expected to have higher average Medicaid expenditures per nursing home care recipient. The correlation between LOS and wages in the nursing home industry and average state income is +.54 and + .43, respectively. It is also possible that states 99

PAGE 111

with high average expenditures on Medicaid nursing home recipients would restrict their nursing home bed supply in an attempt to contain costs. This strategy could cause a greater LOS of the Medicare population in acute-care hospitals awaiting placement in nursing homes. Jim Knickman has done extensive work regarding regional variations in hospital utilization. Knickman (1984, 1985) has identified New York as having a high average LOS and an unusual number of very long hospital stays (over 51 days) with a backlog of patients in acute-care hospitals awaiting placement in long-term care facilities. New York had the highest LOS, 13.5 days across 1981-1988, with the all-state average 8.5 to 9.0 days. Kenney and Holahan (1990) found that patients experience more discharge delays, which increases LOS, at hospitals located in areas with few nursing home beds. Surprisingly, the correlation between average LOS of the Medicare population in acute-care hospitals and nursing home bed supply per 1,000 elderly state population is -.13 only. The state of New York skews the remaining two statistically significant policy/political variables. Local administration of the Medicaid program is found to be positively related to average Medicaid expenditure on nursing home care, but this relationship is no longer statistically significant when New York is excluded from the analysis. The state of New York has the highest state average Medicaid expenditure per elderly recipient of nursing home care. (See Table 5.2). New York is one of six states that administers its Medicaid program at the local level. Clearly, New York is driving the finding of statistical significance between local administration and average Medicaid expenditure per nursing home recipient. New York is also masking the statistically significant, positive relationship between politically liberal states and high-average Medicaid expenditures on nursing home care recipients. The high correlation of + .38 should be pointed out between the liberal quotient of the state and wages in the nursing home industry. High cost of living states tend to be more liberal, and high 100

PAGE 112

cost of living states have higher average expenditures per elder Medicaid recipient of nursing home care. Excluding New York from the analysis also produces a negative statistically significant finding between average Medicaid expenditure on nursing home care residents and year, implying that states spent greater real dollars per Medicaid nursing home care recipient in the early 1980s as opposed to the later 1980s when all other factors are held constant in the model. It is difficult to explain in terms of multicollinearity as the variable with the strongest correlation with year is Medicaid per diem nursing home reimbursement rate, with a positive .42 (Table 5.5). A positive relationship between year and average Medicaid expenditure per elderly nursing home care recipient was hypothesized as the level of disability in the nursing home population has increased over time, particularly since 1984 with the implementation of Medicare's prospective payment system (DRGs or diagnosis-related groupings) which created incentives to discharge the elderly more quickly from acute-care hospitals (Shaughnessy & Kramer, 1990; Rantz, 1991). A more disabled nursing home population was expected to increase average nursing home expenditure per recipient. Percent of Medicaid Expenditure on LTC for the Elderly Spent on Nursing Home Care The hypothesized model does a poor job in explaining interstate variations in the percent of Medicaid LTC expenditures spent on nursing home care. The reduced model explained 25% of the variation while the expanded model explains 40% of the variation across states (Table 5.10). The explained variation is consistent when New York is excluded from the analysis, but the statistical significance of several vanables change, i.e., Medicare average LOS in acute-care hospitals(+), and liberal quotient(-) become statistically significant, while unit cost of LTC to state taxpayers (-), and average state income ( +) lose statistical significance. 101

PAGE 113

.... s Table 5.10. Percent of Medicaid Long-Term Spending for the Elderly or Institutional (Nursing Home) Care Per 1,000 Elderly Population, 1981 1988 I I Includin& New York I Excludin& New York I B T B T B T B T Wages in the nursing home industry -.06 -2.07 -.05 -1.6 .03 1.5 .06 1.8 Old 85)/elderly 65) population 04 1.7 .08 2 2 .. .07 3 .5 .01 2 .9 Elderly voting pop. 65)/voting 18) pop .01 1.4 .01 1.6 .01 1.4 .01 1.3 Average state income -. 00 1.1 .00 2 8 .. .00 82 .00 .87 Unit cost of long-term care to state taxpayers 04 1.2 1.0 -2.3 .01 .27 .06 1.5 Percent state unemployment .01 08 .33 1.9 .20 1.6 .54 3 .3 Percent state elderly population living in SMSA -.02 -1.3 .01 -.6 -.03 -1.7 -.02 -1.2 Average persons per household 10. 2 4.3 12. 5 4 2 .. 9 6 4.4 11.3 4.0 Average state temperature .00 .05 .00 .39 00 -.63 -.00 .12 Year 1.0 -7 .1 -1.5 -6.9 -.98 -7.7 -.78 -3.5 Elderly included in "medicaily needy" 2.1 2.9 .. 2.6 3.9 Medicaid eligibility category No of optional benefits to elderly in state 20 -2 .8 -.20 -3.0 Medicaid Pgm. Medicaid per diem nursing home reimb rate .24 5.2 .09 1.9 State/local administration of Medicaid Pgm. -1.0 -1.0 1.4 1.4 Number of nursing home beds/1,000 eld. pop. .12 4.3 .10 3 .7 Medicare acute care hospital length of stay .37 1.3 1.0 3 .5 Liberal quotient of Congress'[ vote record .01 -.26 -.04 -2.1 Adjusted R2 25 .40 .28 .39 .. Stat i stically significant at p = .05

PAGE 114

The macrocontextual variables predicted to be significantly related to the percent of Medicaid LTC spending devoted to nursing home care were the number of "old" persons in the elderly population ( + ), average state temperature (-), and year(-). The results of the regression analysis indicate that percent unemployment is positively related to percent Medicaid LTC spending devoted to nursing home care whether New York is excluded or included in the analysis. Once again percent unemployment appears to be acting as a strong demand variable for institutional LTC services. New York appears to be masking an interesting interplay between state unemployment and Medicaid home care spending, in that percent unemployment is statistically significant and negative for all aspects of Medicaid home care spending when New York is excluded from the analysis. States with higher unemployment may have greater demand and expenditures for Medicaid nursing home care, but appear to have fewer recipients and less expenditUre on Medicaid home care. (See the following sections). These contrary relationships account for the positive finding between percent Medicaid spending on nursing home care and percent unemployment. Two demographic variables are statistically significant in explaining state variations in percent Medicaid LTC budget devoted to nursing home care. They are: the number of old (age 85 or older) persons per 100 elderly (age 65 or older) persons in the state ( + ); and the average number of person per household in the state ( + ). Clearly the density of old people in the elderly populations is a strong demand variable in that the risk for institutionalization rises with disability that rises with age (Rivlin & Weiner, 1988). This finding is consistent with the hypothesis. The average number of persons per household is a statistically significant explanatory variable of percent Medicaid LTC budget for the elderly devoted to nursing home care, with a positive coefficient. This suggests that states with more persons per household spend a greater 103

PAGE 115

percentage of their LTC dollars on institutional care It may be that the number of persons per household is a better proxy for informal LTC substitutes for Medicaid home and community care than nursing home care. Number of persons per household was found to have a negative relationship with Medicaid utilization of home care and community services for the elderly. More persons in a household may lower the individual cost of informal care in sustaining an elder in the community. When the cost of informal care rises with intensity of need to some critical level, institutional services are utilized. If number of persons in a household substitutes differentially for home and community care versus institutional care, the percent of the Medicaid budget devoted to nursing home care would rise in states with larger households. A statistically significant, negative relationship was found between the variable year and percent Medicaid LTC budget for the elderly spent on nursing home care. This finding reflects the evolution of the LTC field throughout the 1980s toward care within an environment of least restriction and, whenever possible, to allow the elderly to "age in place." This result was hypothesized. Table 5.1 shows national percent Medicaid LTC budgets for the elderly spent on nursing home care dropping from 98.9% in 1981 to 91.7% in 1988. The state-selected variables hypothesized to be related to percent Medicaid spending on LTC for the elderly devoted to nursing home care are the inclusion of the elderly in the optional "medically needy" category ( + ), number of optional Medicaid benefits ( -), number of nursing home beds ( + ), and political liberalism of the state ( + ). The inclusion of the elderly in state Medicaid optional medically needy programs surfaces as a positive, statistically significant variable in explaining the percent Medicaid LTC budget spent on nursing home care for the elderly. Including the elderly in the medically needy program increases demand for high intensity LTC services, which states appear to satisfy, to a larger degree, with institutional care as opposed to home and community care. 104

PAGE 116

States that offer fewer optional Medicaid benefits to their elderly spend a greater percentage of their Medicaid LTC budget on institutional care, when all other factors in the equation are held constant. This appears mediated by both the previously discussed negative relationship between optional Medicaid benefits and number of elderly Medicaid recipients of nursing home care and the positive relationship between optional benefits to the elderly and average expenditure per Medicaid home-care recipient and Medicaid 2176 spending. These findings are discussed in forthcoming sections. Medicaid per diem nursing home reimbursement rate is poSitively associated with a higher percentage of the Medicaid LTC budget for the elderly devoted to institutional care. This is a reflection of the strong positive association between Medicaid per diem nursing home reimbursement rate and number of recipients and average expenditure on nursing home care, in addition to the strongly negative association between Medicaid per diem nursing home reimbursement rate and Medicaid home-care utilization. States with high per diem nursing home reimbursement rates may be limiting their home-care programs in an attempt to contain Medicaid expenditures. Home care may represent greater discretionary spending than nursing home care. The number of nursing home beds per 1,000 elderly population is statistically significant ( +) regarding percent Medicaid LTC spending on the elderly devoted to nursing home care. Conventional wisdom and the previous results of this study suggest that increased nursing home beds leads to increased numbers of nursing home care recipients, which leads to increased expenditures on nursing home care. A positive relationship was hypothesized. New York appears to disguise a statistically significant, positive relationship between average LOS of the state Medicare population in acute-care hospitals and percent LTC Medicaid spending on institutional care. This may represent an issue of multicollinearity between the variables "LOS" and "year" (-.39). In the early 1980s, when the average LOS for the Medicare 105

PAGE 117

population was high, scant state Medicaid expenditures were utilized for home or community long term care. New York, with the highest LOS in the nation, also has the highest utilization and expenditures on non-institutional Medicaid programs in LTC. The state of New York also masks the statistically significant, negative relationship between political liberalism and percent Medicaid LTC spending on the elderly devoted to nursing home care. When all other factors are held constant, states that are politically more conservative spend a greater percentage of their Medicaid LTC budgets on institutional services. This finding is consistent with the positive relationships found between politically liberal states and more recipients and higher expenditures on home health care, although positive relationships were also found between Medicaid recipients and spending on nursing home care and political liberalism. The relationships between the political liberalism of the state and components of Medicaid home-care spending are discussed in forthcoming sections. The majority (60%) of the variation in the percent of state LTC Medicaid budgets spent on nursing home care remains unexplained by the model utilized in this study. Decisions regarding institutional versus non-institutional resource allocation may lie within the state Medicaid bureaucracy, which is not accounted for in the model. Medicaid Expenditure for Home Health Care for the Elderly The macrocontextual state variables utilized in the model expected to relate to Medicaid expenditure on home care for the elderly on a 1,000 state elderly population basis were economic (wages in the nursing home industry [ + ], average state income [ + ], unit cost of LTC to state taxpayers [-], and percent state unemployment [-]), demographic ("old" persons in the voting and elderly population [ +] and average persons per household [-]). It was also hypothesized that states 106

PAGE 118

would spend greater real dollars on home care for the elderly in the later 1980s as opposed to the early 1980s. Furthermore, the state of New York was expected to greatly influence home health care results as New York alone accounted for approximately 80% of the national Medicaid home health care budget in 1986 (Swan, 1990). Excluding the state of New York from the analysis reduces the explained variation in Medicaid spending for home health care for the elderly on a 1,000 elderly population basis from 39% to 30% via the reduced model, and 64% to 30% via the expanded model (Table 5.11). Excluding New York, socio-economic, demographic, and control variables explain the same amount of variation in home health-care spending for the elderly across states as when policy and political variables are added to the model with adjusted R2 of .30 (Table 5.11). Two macrocontexfual variables are statistically significant in explaining interstate variation in Medicaid spending on home health care for the elderly in the expanded model when New York is either excluded or included in the analysis, namely, year ( +) and percent state elderly population living in metropolitan areas (-). The coefficient for "year" is positive indicating that states spent more Medicaid dollars on non-institutional LTC services for the elderly on a per 1,000 elderly basis, in the late 1980s as opposed to the early 1980s. This finding is consistent with the evolution of the LTC field toward care in the environment of least restriction and the concept of "aging in place." The positive, statistical significant finding of "year" with Medicaid spending on home health care for the elderly appears mediated by both a positive relationship between "year" and both average expenditure on home care per elderly recipient and the number of elderly Medicaid recipients of home health care services per 1,000 elderly population. 107

PAGE 119

..... 0 00 Table 5.11. Medicaid Expenditure on Home Care for the Elderly Per 1,000 State Elderly Population, 1981 1988 ---I I Including New York I Excluding New York B T B T B T B Wages in the nursing home industry 1,010.1 10.4+ 953.2 8.4+ 138 5 3 .3+ 42.8 Old elderly 85)/elderly 65) population 370.5 3.8+ 110.3 .85 62.6 1.7 -41.2 Elderly voting pop. 65)/voting 18) pop. -34.1 -1.1 -42.4 -1.6 -25.4 -2.2 -19 22 Average state income .55 .61 -2.8 -2 .8+ .93 2.7+ .65 Unit cost of long-term care to state taxpayers -476.2 -3.4+ 202.4 1.3 -182 3 -3.4+ -13 5 Percent state unemployment 1,619.9 2.8+ 774.4 1.2 -272.1 -1.2 -866.6 Percent state elderly population living in SMSA -41.0 54 -165.5 -2.1+ 6 9 .24 -67.4 Average persons per household -18,134.0 -1.8 -10,830. -1.0 -12 311. -3 .2+ -1,070.0 8 4 Average state temperature -45.4 -1.5 -38.5 -1.5 -7 8 70 -11.4 Year 1,539.0 2 .6+ 7,452.3 9.7+ 1,240.6 -5.6+ 2,240 8 Elderly included in "medically needy" 3,280.0 1.3 -527.9 Medicaid eligibility category Number of optional benefits to elderly in state -60.1 24 81.05 Medicaid Program Medicaid per diem nursing home reimb. rate -1,398.9 -8 72+ -272 7 State/local administration of Medicaid Program 17,740.3 5 .o+ -852.8 Number of nursing home beds/1 ,000 eld pop -200.5 -2 .0+ -28 5 Medicare acute care hospital length of stay 5,333.3 5 .o+ 1,022.6 Liberal quotient of Congress'! vote record -191.9 -2.7+ 62.8 Adjusted R1 39 .64 .30 .. Statistically significant at p = .05 I T .70 -.68 -1.5 1.4 .19 -2.9+ -1.9 -.21 -1.0 5 :5+ -.45 .70 -3.2 -.48 60 1.9+ 1.9+ .30

PAGE 120

The percent of the elderly population residing in metropolitan areas is negatively related to Medicaid spending on home care for the elderly. This finding is tempered through higher average Medicaid expenditures per elderly home-care recipient in states with a greater concentration of rural elderly. Macrocontextual variables found to be statistically significant in the expanded model only when New York is included in the analysis are wages in the nursing home industry ( + ), and average state income(-). Wages in the nursing home industry were intended to reflect input costs across all LTC services. As such, wages in the nursing home industry and Medicaid expenditures on home health care and average Medicaid expenditures per home-care recipient were expected to yield strong positive relationships, even when New York was excluded from the analysis. This finding did not occur Furthermore, no statistically significant relationship was found between wages in the nursing home industry and Medicaid spending on the 2176 waiver program (community services provided to those who qualify for Medicaid nursing home placement). All LTC services are labor intensive. Nursing homes may have a higher professional-to-non-professional staffing ratio than non-institutional LTC services; thus, the nursing home wage variable may reflect input costs more specific to institutional (nursing home care) than non-institutional (home and community care) care. Home and community services in the 1980s tended to be more custodial in nature stressing assistance with instrumental activities of daily living (IADLs such as transportation and shopping, etc.) as opposed to skilled services (IV's, ventilators, etc.). As such, a weaker but still positive finding was hypothesized between wages in the nursing home industry and Medicaid spending on home and community care. This finding occurred only in home care and only when New York was included in the analysis. New York has the highest wages in the nursing home industry, the 109

PAGE 121

highest average Medicaid expenditure per home-care recipient, and the most Medicaid recipients of home-care services per 1,000 state elderly population by far. (See Tables 5.2 and 5.4). A positive relationship was hypothesized between average state income and all aspects of home-care spending for the elderly. Higher average state income was expected to increase state revenues that would be used, in part, to support state programs such as Medicaid. Higher average state income is associated with fewer elderly Medicaid recipients of home care when New York is included in the analysis only, but higher average expenditures per recipient of Medicaid home-care services when New York is either included or excluded from the analysis. The net result is a negative relationship between state income and Medicaid expenditures for home care for the elderly only when New York is included in the analysis. Excluding New York also results in a statistically significant, negative finding between Medicaid spending on home health care for the elderly and percent state unemployment. This finding is consistent with that hypothesized. It was expected that percent unemployment would act as an economic indicator of a states' fiscal capacity. This finding is buffered by negative relationships between average state unemployment rate and average Medicaid expenditure per home health-care recipient and number of elderly recipients of Medicaid home health-care services per 1,000 elderly population. Percent state unemployment operates differently regarding Medicaid home-care spending and nursing home care spending when New York is excluded from the analysis. Percent unemployment acts as an economic measure of a state's fiscal capacity, i.e., ability to pay regarding home care services. Conversely, percent unemployment acts as a demand variable regarding number of nursing home care recipients. The state-selected variables hypothesized to be related to Medicaid expenditure on home care spending for the elderly were Medicaid nursing home per diem rate ( + ), nursing home bed stock(-), Medicare acute-care hospital LOS (-), and political liberalism of the state ( + ). 110

PAGE 122

The results of the regression analysis indicate that two health-care variables are statistically significant inclusive or exclusive of New York, namely, the Medicaid per diem nursing home reimbursement rate (-), and the average LOS of the state Medicare population in acute-care hospitals ( + ). Each finding runs contrary to the hypothesis. It was expected that higher Medicaid nursing home per diem rates would be associated with higher per visit home care rates, increasing the average expenditure per home care recipient and, therefore, increasing overall home care expenditures. The results of the analysis indicate that states with high Medicaid nursing home per diem rates are actually associated with fewer recipients of Medicaid home-care services translating into lower overall Medicaid home-care expenditures for the elderly. The relationship is discussed in the next section. The positive relationship between Medicaid spending on home care for the elderly and higher average acute hospital LOS for the Medicare population appears to be mediated by both higher average expenditures per elderly Medicaid home-care recipient and number of home-care recipients per 1,000 elderly population. A negative relationship was hypothesized between LOS and Medicaid spending on home care for the elderly, average Medicaid spending per elderly recipient of home care, and number of elderly home-care recipients. A negative coefficient was anticipated as shorter hospital stays (particularly in a DRG environment) were expected to increase both the demand and intensity of post-hospital services. Studies have shown an increase in both number of recipients of LTC services and intensity of services since the implementation of DRGs (Shaughnessy & Kramer, 1990; Lieberg, 1988). Although the affects of shorter acute-care LOS stay was expected to be felt initially in the Medicare program, a trickle-down affect to the state Medicaid program was anticipated. 111

PAGE 123

New York has a nursing home occupancy rate approaching 98% and continues to aggressively limit nursing home bed stock (HCIA, 1990). New York also has a backlog of elderly in acute-care hospitals awaiting nursing home placement which is contributing to New York's high average LOS (Knickman, 1985). New York, in addition to other states, may be attempting to contain the costs of an extremely expensive Medicaid program by substituting home-care services for nursing home care, while continuing to limit nursing home bed supply. As such, higher rates of Medicaid home-care utilization is associated with states with higher lengths of stay of their Medicare population in acute-care hospitals. The liberal quotient of the state is positively related to Medicaid spending on home health care for the elderly when New York is excluded from the analysis, but negatively related when New York is included in the analysis. These findings are mediated by a positive relationship between a state's liberal quotient and the number of Medicaid home-care recipients when New York is excluded from the analysis, and a negative relationship between a state's liberal quotient and average Medicaid expenditure per home-care recipient inclusive of New York. It is difficult to reconcile the negative relationship between political liberalism and Medicaid spending on home care that appears to be driven by the state of New York The New York Medicaid program spends 17 times more than the average state on home health care for the elderly on a per 1,000 elderly population basis, while the average liberal quotient of New York's congressional voting record from 1981-1988 is 58.5. The average liberal quotient across all states for the same time period is 44 (Tables 5.2 and 5.3). Two health-policy variables are statistically significant only when New York is included in the analysis, namely, local administration of the state Medicaid program ( + ), and number of nursing home beds per 1,000 state elderly population(-). 112

PAGE 124

New York is one of six states with local administration of their Medicaid program. Local administration is associated with higher average expenditures per Medicaid home-care recipient and greater numbers of Medicaid recipients of home-care services when New York is included in the analysis, although neither is statistically significant. Central administration of the state Medicaid program is statistically significant relative to number of Medicaid home-care recipients when New York is excluded from the analysis. Number of nursing home beds per 1,000 state elderly population is statistically significant ( with a negative relationship with number of elderly Medicaid recipients of Medicaid home-care services, inclusive or exclusive of New York. However, it is statistically significant relevant to overall Medicaid expenditures on home care for the elderly only when New York is included in the analysis. Elderly Medicaid Recipients of Home Health Care Services Excluding New York from the analysis reduces the explained variation in the number of Medicaid recipients of home health care per 1,000 elderly population from 31% to 20% in the reduced model, and 48% to 23% in the expanded model (Table 5.12). The macrocontextual variables expected to be correlated with the number of elderly Medicaid recipients of home-care services per 1,000 elderly state population were economic (average state income [+],unit cost of LTC to state taxpayers [-], and percent state unemployment [-]), demographic (percent "old" persons in the voting and elderly population [ +], percent elderly in urban areas [ +], and average persons per household [-]). It was also expected that there would be relatively more elderly Medicaid recipients of home care in the later 1980s as opposed to the early 1980s. 113

PAGE 125

...... ...... Table 5.12. Nwnber of Elderly Medicaid Recipients of Home Care Per 1,000 State Elderly Population, 1981 1988 I I Including New York I Excluding New York B T B T B T B Wage s in the nursing home industry .31 9.9"' .24 6 .2 .12 4.4"' .06 Old elderly 85)/elderly 65) population .09 2.8 .00 04 .02 .69 .03 Elderly voting pop 65)/voting 18) population -.01 -1.1 -.02 -1.8 .01 -1.1 -.01 Average state income 00 -1.4 -.00 -4.1"' 00 -.64 -.00 Unit cost of long-term care to state taxpayers .23 -5.1 .02 .37 -.17 -4.9 06 Percent state unemployment .13 .70 -.24 -1.1 -.28 -2.0"' -.59 Percent state elderly population living in SMSA .03 1.4 -.03 -1.1 .03 1.8 .01 Average persons per household -15.4 -4.7"' 10.1 2.7"' -13.9 -5.6 -8.4 Average state temperature -.01 -.65 -.01 .93 .00 .31 -.00 Year .29 1.5 2.1 7 .9 .33 2.3"' .90 Elderly included in "medically needy" Medicaid eligibility 1.4 1.6 .40 category No. of optional benefits to elderly in state Medicaid Pgm .03 .33 .01 Medicaid per diem nursing home reimbursement rate -.44 -8.o -.18 State/local administration of Medicaid Program 1.3 1.1 -2.6 Number of nursing home beds/1 ,000 elderly population -1.0 -3.0"' -.05 Medicare acute care hospital length of stay 1.7 4.5"' .40 Liberal quotient of Congress'! vote record .00 .06 06 Adjusted R2 .31 .48 .20 "' Statistically significant at p = 05 I T 1.6 -.8 -1.4 -1.3 -1.3 -3.2"' -.56 -2.6"' -.31 3.6"' .54 .15 -3.4"' -2.3"' -1.9"' 1.2 3.0"' 23

PAGE 126

The state-selected variables hypothesized to be associated with the number of elderly Medicaid recipients of home-care services are the inclusion of the elderly in the "medically needy" program ( +), nursing home. bed stock (-), LOS of the state Medicare population in acute-care hospitals(-), and the political liberal climate of the state ( + ). The variables actually found to be consistently statistically significant by regression analysis when New York is excluded or included in the analysis are year ( + ), average persons per household(-), Medicaid per diem nursing home reimbursement rate(-), and number of nursing home beds per 1,000 state elderly population(-). The independent variable "year" is consistently, positively related to all aspects of home care spending for the elderly regardless of the inclusion or exclusion of New York. This finding is consistent with the hypothesis. The LTC field has evolved markedly throughout the 1980s toward providing services in afi environment of least restriction. This evolution has been fostered by both ideology and anticipated cost containment. A negative, statistically significant relationship emerged between the average number of persons per household and the number of Medicaid recipients of home health care services per 1,000 elderly population. This finding is consistent with the hypothesis that more persons in a household would increase the likelihood of informal LTC, reducing the need for formal (paid) LTC services. This study suggests that a greater number of persons per household substitutes for home and community services, but not for institutional services. A negative relationship was also found between Medicaid nursing home per diem rate and the number of elderly Medicaid recipients of home health-care services per 1,000 elderly state population. This relationship was not predicted. When all else is held constant, states with lower Medicaid per diem nursing home rates may offer alternative services to elderly LTC recipients, while states with high per diem Medicaid nursing home rates may be hesitant to add any 115

PAGE 127

discretionary services in an attempt to contain overall Medicaid costs. One suspects the effects of Medicaid nursing home per diem rate on the number of Medicaid recipients of home care are contained at some point by nursing home bed supply and nursing home occupancy rates. A negative relationship was also found between nursing home bed stock per 1,000 elderly and number of elderly Medicaid home-care recipients. This is consistent with both the hypothesis and Swan and Benjamin's (1990) finding that Medicare home health-care spending was negatively related to nursing home bed supply. Although many researchers have indicated that a national nursing home bed shortage exists (Harrington et al., 1988; Richardson, 1990), there may still be sufficient slack in nursing home bed stock within many states (average nursing home occupancy rate of 92% in the mid-1980s) that home and community services have yet to act as substitutes for nursing home care. The influence generated by the state of New York appears to drive the statistical significant relationship found between the number of Medicaid recipients of home-care services per 1,000 elderly population and wages in the nursing home industry ( + ), average state income(-), and average LOS of the Medicare population in acute-care hospitals ( + ). New York is the clear outlier regarding Medicaid home-care utilization with an average of 45.6 elderly recipients per 1,000 elderly population from 1981-1988, while the national average for the same time period was 6.2 (Table 5.2). New York also has the highest wages in the nursing home industry and the highest 9 year average for LOS at 13.5 days while the national average was 9.4 (Table 5.3). New York had a nursing home occupancy rate of 99% in 1988 (HCIA 1990), with a backlog of patients awaiting post-hospital care. It was hypothesized that a negative relationship would exist between LOS and number of elderly Medicaid recipients of home health care as a reduction in LOS would increase the demand for post-hospital services (Manton, 1985). The effects of heightened demand for post-hospital care 116

PAGE 128

for the elderly would initially be felt in the Medicare program, particularly Medicare Part A, but a trickle-down affect to the Medicaid home-care program was expected. This finding did not occur. New York may be expanding Medicaid home-care services for the elderly in lieu of nursing home care. This would explain the positive correlation between LOS and elderly Medicaid recipients of home care. New York also has the fifth highest average state income for years 1981 to 1988, at $22,800. New York ranked behind Connecticut, New Jersey, Alaska, and Maryland. It is difficult to understand the mechanism for New York intensifying a negative relationship between average state income and number of elderly Medicaid recipients of home care. A positive relationship was hypothesized as states with higher income were expected to have a greater fiscal capacity with which to fund greater access to non-institutional LTC services. Conversely, New York appears to mask a negative, statistically significant relationship between unemployment and the number of Medicaid recipients of home health care. A lower state unemployment rate is associated with more Medicaid recipients of home health care, and greater Medicaid expenditure on home health care when New York is excluded from the analysis. In this instance, percent unemployment is acting as an economic indicator of fiscal capacity with greater fiscal capacity associated with increased access of elders to Medicaid home-care services. This relationship is consistent with the hypothesis. Similarly, New York appears to also disguise a statistically significant relationship between central administration of the state Medicaid program and the number of elderly Medicaid recipients of home health care. A central state organization may be more efficient at fostering the development and statewide implementation of less traditional LTC services both at home and in the community. 117

PAGE 129

A positive, statistically significant relationship is found between the liberal quotient of a state and the number of elderly Medicaid recipients of home health care, per 1,000 elderly population when New York is excluded from the analysis. Political liberalism has been associated in a statistically significant manner with all aspects of Medicaid spending on LTC for the elderly (with the exception of average expenditure per elderly home-care recipient) when New York is excluded froi;Il the analysis. This is consistent with the hypothesis that politically liberal states would be more supportive of state Medicaid programs for the elderly. Average Medicaid Expenditure per Elder Recipient of Medicaid Home Health Care Services Over half of the variation across states in average expenditure per elder Medicaid recipient of home care remains unexplained by the model utilized in this study (Table 5.13). Excluding New York from the analysis produces similar statistically significant results, but the explained variation falls from 48 to 39 percent (Table 5.13). It was hypothesized that average Medicaid expenditure per elderly home-care recipient would be driven by macrocontextual state variables of economy (wages in the nursing home industry [ + ], average state income [ + ], and percent state unemployment [-]) and demography (percent elderly in metropolitan areas [-]). It was also predicteq that average Medicaid expenditure per elderly Medicaid recipient of home-care services would be negatively related to average state temperature and positively related to "year" with greater average home-care expenditure per recipient occurring in the later s as compared to the early 1980s. J Average income of the state population ( + ), percent of the elderly population residing in metropolitan areas(-), average state temperature(-), and year ( +) are the macrocontextual independent variables that demonstrate consistent statistically significant relationships with average 118

PAGE 130

...... ...... \0 Table 5.13. Average Medicaid Expenditure Per Elderly Recipient of Home Care Per 1,000 State Elderly Population, 1981 1988 I I Including New York I Excluding New York I B T B T B T B T Wages in the nursing home industry 18.5 6.0* 10. 6 2.4* 6 2 2 .0 -4 9 -1.0 Old 85)/elderly 65) population I 2.8 .89 -1.2 -.24 -1.4 -.03 -3 8 8 Elderly voting pop. ( 65)/voting ( 18) population -1.7 -1.8 -1.4 -1.3 -1.6 -1.8 -1.0 -1.0 Average state income .12 4 0* .07 1.9 .11 4.3 -.11 3 0* Unit cost of long-term care to state taxpayers -7.4 -1.6 7.3 1.2 -2 8 68 4.9 .89 Percent state unemployment -6.4 .35 -23.4 1.0 -32.1 -1.8 -50.4 -2.2* Percent state elderly population living in SMSA -4.4 1.8 -9 7 -3.2* -3. 3 -1.5 -7.9 -2.8* Average persons per household 46.5 .14 314.2 73 122 3 .4 481.2 1.2 Average state temperature -2 8 -3.0* -2 9 -2.9* -2 3 -2.7 -2.4 -2 7* Year 120 8 6.4* 208 8 6.8* 114 9 6.7 126.5 4 0* Elderly included in medically needy" Medicaid eligibility 154.9 -1.6 211.8 -2 3* category Number of optional benefits to elderly in state Medicaid 20 8 2.1 24.8 2.7 Pgm. Medicaid per diem nursing home reimbursement rate 8.4 -1.3 9.4 1.4 Statellocal administration of Medicaid Program 173. 8 1.2 138 8 1.0 Number of nursing home beds/1 000 elderly population -4.4 -1.1 2 0 -.53 Medicare acute care hospital length of stay 165. 1 3.9* 106.7 2 .6 Liberal quotient of Congress'! vote record -5.5 -2.0 -1.7 62 Adjusted R2 .44 .48 37 .39 .. Statistically significant at p = 05

PAGE 131

Medicaid expenditure per home health-care recipient inclusive and exclusive of New York via regression analysis. Average state income is acting as an economic indicator of "ability to pay" regarding the average Medicaid expenditure on home health services per elder recipient. Recipients of home health care in poorer states receive, on average, a lower-priced package of home health-care services, while recipients in richer states receive a higher-priced package of services. The GAO (1987) has argued that "higher priced package" is strongly associated with greater real benefit and does not simply reflect regional cost of living, while Buchanan et al. (1991) suggest that cross-state price levels argue against the equivalent use of expenditure with benefit. A positive relationship was hypothesized between average state income and average expenditure per Medicaid home-care recipient. States with a relatively greater percentage of their elderly population residing in rural areas spend more per elderly Medicaid home care recipient. The Agency for Health Care Policy Research (1990) found higher levels of debility in the rural, as compared to the urban, Medicare population. Higher average Medicaid expenditures for rural elderly home care recipients may relate to greater intensity of need secondary to greater debility, as well as increased input costs with fuel, travel time, and decreased staff productivity. A positive association between "year" and average Medicaid expenditure per elderly home-care recipient indicates that intensity of home health-care services have increased throughout the 1980s. This is consistent with findings of increased functional impairment in home-care settings postulated by Shaughnessy and Kramer (1990). Colder temperature is associated with higher per recipient spending on home care. Colder temperatures are expected to increase input costs of home care, i.e., fuel and lost productivity with travel delays of staff, etc., thus increasing per recipient expenditure data. 120

PAGE 132

Wages in the nursing home industry is positively associated with average expenditure per elderly Medicaid recipient of home care only when New York is included in the analysis. A negative association is found when New York is excluded from the analysis, although it is not statistically significant. A positive relationship was hypothesized as increased nursing home wages were expected to reflect labor costs throughout institutional and home and community LTC services. The macrocontextual variable that become statistically significant when New York is excluded from the analysis is state average percent unemployment(-), which appears to act as an economic indicator of the fiscal capacity of the state to support home-care services. A negative relationship was initially hypothesized. The state-selected variables hypothesized to be related to average Medicaid expenditure per elderly home-care recipient were Medicaid nursing home per diem rate ( +), LOS of the state Medicare population in acute-care hospitals(-) and the political liberal climate of the state ( + ) The results of the regression analysis indicate that two policy variables are statistically significant relative to average Medicaid expenditure per elderly recipient of home-care services when New York is excluded or included in the analysis. These policy variables are the number of optional benefits to the elderly in the Medicaid program ( + ), and the average LOS of the state Medicare population in acute-care hospitals ( +). States that are more generous regarding optional Medicaid benefit for the elderly tend also to be more generous regarding average expenditures on home-care recipients and community spending on the 2176 Waiver Program. States do appear to have an institutional versus non institutional care bias regarding their optional Medicaid Program choices. Longer LOS of the state Medicare population in acute-care hospitals is associated with higher average Medicaid expenditures per elder home-care recipient. A negative relationship was 121

PAGE 133

hypothesized as shorter hospital stays were expected to increase both the demand and intensity of post-hospital services. As previously reported in this chapter, studies have shown an increase in both number of recipients of LTC services and intensity of services since the implementation of DRGs (Shaughnessy & Kramer, 1990; Lieberg, 1988). Again, the affects of shorter acute-care LOS was expected to be initially felt in the Medicare program with a trickle-down effect to state Medicaid programs. It seems possible that states with a high LOS may have high nursing home occupancy rates with a backlog of patients awaiting nursing home placement, as is the case with New York. States may be addressing these backlogs by serving some of these patients in the community with intense (high expenditure) service packages. States that allow "medically needy" elders to be eligible for the Medicaid program tend to have lower average expenditures on home care per elder recipient of Medicaid home-care services. This relationship becomes statistically significant when New York is excluded from the analysis. These same states tend to have more recipients of nursing home care, spend relatively more of their LTC budget on institutional care, and spend less on the Medicaid 2176 waiver program. Once again, states appear to have an institutional (nursing home) versus non-institutional (home and community) preference when providing LTC to the elderly. The political liberal quotient of the state is negatively associated with average Medicaid expenditure per elderly recipient of home-care services. A positive correlation was hypothesized. The negative relationship between political liberalism and average Medicaid expenditure per elderly home-care recipient is statistically significant only when New York is included in the analysis. It appears that political liberalism is loosely associated with more generous eligibility, but a lower priced benefit package regarding Medicaid home care for the elderly, when all else is held constant in the analysis. 122

PAGE 134

Medicaid Expenditure on the 2176 Waiver Program for the Elderly The Medicaid 2176 Waiver Program allows Medicaid monies to be spent sustaining persons in the community who qualify for Medicaid-subsidized nursing home placement. The cross-state variation in Medicaid expenditure on the 2176 Waiver Program for the elderly per 1,000 state elderly population, 1981-1988, is explained poorly by this model (Table 5.14). The reduced model explains 14% of the variation across states, while the expanded model explains 31%. Virtually identical results are demonstrated when New York is excluded from the analysis. The hypothesis states that Medicaid expenditure on the 2176 Waiver Program is related to economic variables (wages in the nursing home industry [ +], average state income [ +], percent state unemployment [-]), demographic variables (stratification by age of the voting and elderly population [ +] and average" persons per household [-]), and state-selected variables (nursing home bed stock [-], LOS of the state Medicare population in acute-care hospitals [-], and political liberalism of the state [ + D-It was also predicted that states would spend progressively more on the 2176 Waiver Program in the later 1980s as compared to the mid-1980s. Average state income acts in a counterintuitive fashion regarding Medicaid spending on the elderly via the 2176 Waiver Program. A positive relationship was hypothesized. It was expected that states with higher income would have greater revenues with which to support discretionary, non-institutional spending; conversely, poorer states were expected to restrict the expansion of services. Administrators of state Medicaid programs in poorer states (states with lower average state income) may view the 2176 Waiver Program as a lower-cost substitute of community services for nursing home care services. Although cost containment has not yet been demonstrated between experimental Medicaid spending to support the elderly who qualify for nursing home care placementin the community, it is considered one of the primary goals of the 123 I -I I I

PAGE 135

Table 5.14. Medicaid Expenditure on the Medicaid 2176 Waiver Program Per 1,000 State Elderly Population, 1981 1988 I I Including New York I Excluding New York I B T B T B T B T Wages in the nursing home industry 27.9 -. 62 33. 1 .55 -27.6 .54 -23.4 -.34 Old elderly ( 85)/elderly ( 65) population -92 0 -2.0 -160.0 -2.02 -92.0 -2.0 -170 7 -2.4 Elderly voting pop. ( 65)/voting ( 18) pop -14.6 -1.1 -23 6 1.7 -14.5 -1.1 -21.9 -1.6 Average state income -1.4 -3 .2 -1.6 -3.0 -1.3 -3.1 -1.1 -2.0 Unit cost of long-term care to state taxpayers 70.9 1.1 124 8 1.5 69.4 1.1 86.3 1.0 Percent state unemployment -3.7 -.01 -229 8 -.7 1.4 .01 -324 9 -1.0 Percent state elderly population living in SMSA 66.4 1.8 83. 6 2 .0 64.2 1.8 87.4 2 1 .. Average persons per household -15,074.6 -3.2 -29,032.7 -s.1 -15,054 9 -3.1* -28,572.7 -s.o ,_ Average state temperature 15.1 1.2 8.3 .7 15.2 1.1 10 5 .83 Year 1,839.0 s.s 815.8 1.6 1,855.4 5.4 342 8 .62 Elderly included in medically needy" Medicaid -4 081.5 3 0* -4,450.6 -3 3* eligibility category Number of optional benefits to elderly in state 469.3 3 .4 457.3 3 .3 Medicaid Program Medicaid per diem nursing home reimb. rate -46.3 -.54 57.8 .6 State/local administration of Medicaid Program -2,290.2 -1.2 -3,761 0 -1.8 Number of nursing home beds/1,000 eld. pop. -159.3 -2.9 -135.3 -2.4 Medicare acute care hospital length of stay -3,364 5 -5.4* -3,974.8 -6.0* Liberal quotient of Congress'! vote record 73.1 1.9* 98.5 2 .5 Adjusted R2 .14 .33 .14 .35 Statistically significant at p = .05

PAGE 136

2176 Waiver Program. As such, the incentive to substitute community services for institutional 1 services may be stronger in "poorer" states. The demographics of the state elderly population do appear to influence state spending on the Medicaid 2176 Waiver Program. A greater concentration of "younger" elderly (age 65 to 84) in the state elderly population and a greater concentration of the state elderly population residing in urban areas correlate with greater state Medicaid spending on the 2176 Waiver Program per 1,000 state elderly population. The "young elderly have, on average less disability with fewer limitations in activities of daily living. As such, the young elderly" may be sustained in the community by the Medicaid 2176 Waiver Program, while the more disabled "old elderly (age 85 and older) require institutional care. There is a statistically significant, positive relationship between the number of elderly Medicaid recipients of nursing home care per 1,000 elderly state population and the concentration of "old elderly" in the state elderly population in the reduced model. In the expanded model, the concentration of old elderly" becomes statistically insignificant when the number of nursing home beds per 1,000 state elderly becomes statistically significant. States with a relatively greater urban elderly population appear to spend more Medicaid dollars sustaining the disabled elderly in the community via the 2176 Waiver Program, while states with a relatively greater rural elderly population spend less. This is in contrast to Medicaid homecare spending for the elderly, where states with a relatively greater rural elderly population had both higher average expenditures per home care recipient and higher overall Medicaid expenditures on home care for the elderly on a 1,000 elderly population basis. It may be logistically simpler, in an urban setting, to provide the intensity of services needed to sustain an elder in the community who qualifies for nursing home care placement. It is also possible that states choose between non-institutional programs for the elderly. 125

PAGE 137

Coordination and delivery of services and productivity of staff may be more efficient in urban areas. Medicaid administrators may see community care as a "cheaper" alternative to nursing home care in urban as opposed to rural settings if the mechanism for higher Medicaid expenditures on the 2176 Waiver Program is via increased number of recipients, not increased average expenditure per 2176 Waiver recipient. Unfortunately, those data were not available. Larger average state households are associated with fewer Medicaid dollars spent on community care for the elderly under the 2176 Waiver Program. Once again, larger households may lower the individual cost of providing informal care for the elderly in the community. This may encourage the substitution of informal for formal (paid) Medicaid services. This is consistent with the negative relationship found between the number of elderly Medicaid recipients of home care per 1,000 elderly population and the average number of persons per state household. Larger households may offer informal care as a substitute for publicly subsidized home and community services, but do not appear to substitute for nursing home care for the elderly when compared to smaller households. States appear to make trade-offs in the structure of their Medicaid programs. States that spend more Medicaid dollars on the experimental 2176 Waiver Program to sustain elders who qualify for nursing home placement in the community are more apt to have a greater number of optional Medicaid benefits available to the elderly. These same states have a relatively smaller nursing home bed supply and disallow Medicaid eligibility based on the optional Medicaid "medically needy" criteria. Both of these variables have been strongly associated with increased numbers of elderly Medicaid recipients of nursing home care. It is possible that state Medicaid administrators and policy makers have an inherent institutional versus non-institutional bias in the distribution of their resources. It is more likely that state policy makers are expanding community services for the disabled elderly while restricting Medicaid access to nursing home care in an 126

PAGE 138

attempt to contain costs. To date, home and community services for the disabled elderly have been shown to compliment, not substitute for nursing home care. State policy makers may still be pursuing a cost-containment strategy if they anticipate one-for-one substitution of community-care recipients for nursing home care recipients when the nursing home bed stock is saturated, if that is politically feasible. A statistically significant negative relationship was found between Medicaid expenditure on the 2176 Waiver Program per 1,000 state elderly population and average LOS of the state Medicare population in acute-care hospitals. This is consistent with the hypothesis. Shorter LOS has been associated with an increase in post-hospital services with the Medicare Program feeling the initial effects and Medicaid the later effects. This finding is in contrast to home care where a positive association was found between average LOS of the state Medicare population in acute-care hospitals and numoet 6f recipients and average expenditure per Medicaid home-care recipient. Surprisingly, a statistically significant positive relationship was found between the variable "year" and Medicaid spending on the 2176 Waiver Program in reduced models only. A strong positive relationship was expected between "year" and all aspects of non-institutional LTC spending. The negative of .39 between LOS and "year" may raise the issue of multicollinearity between these variables Once again, states with a higher liberal quotient were associated with greater relative spending on the elderly via the 2176 Waiver Program. Conversely, more conservative states were associated with relatively less spending on the 2176 Waiver Program. This finding is consistent with the hypothesis that liberal states would tend to redistribute more funds toward the indigent elderly. 127

PAGE 139

Medicaid Nursing Home Expenditure/Recipient Equations Inclusive of Non-Institutional Medicaid Expenditure/Recipient Data The expanded model explains 81 % of the variation in Medicaid expenditure on nursing home care when Medicaid expenditure on non-institutional (home health care plus 2176 Waiver Program) spending on LTC for the elderly is incorporated into the analysis (Table 5.15). In the expanded model, Medicaid spending on non-institutional care (MENI) increases the explained variation in Medicaid spending on nursing home care from 79% to 81%, while all other statistically significant relationships remain intact. The relationship between Medicaid expenditures on non-institutional and institutional LTC services for the elderly is positive. The expanded model for explaining Medicaid recipients of nursing home care that includes the number of Medicaid recipients of home health-care generated an adjusted R2 of .68 (Table 5 .16). The number of elder! y Medicaid recipients of home care per 1, 000 elder! y population is positively related to the number of elderly Medicaid recipients of aursing home care per 1,000 elderly population. States with greater access (recipients) of Medicaid elders to home health-care services appear to also have greater access to nursing home care services, across years 1981-1988. States with relatively higher Medicaid expenditure on home health care appear to also have a relatively higher Medicaid expenditure on nursing home care services for the elderly. States do not appear to have substituted home care for nursing home care services regarding LTC for the elderly from years 1981 to 1988. Furthermore, there is no empirical evidence from this study that home and community services contain nursing home care expenditures. This finding is consistent with findings that home and community services represent new costs to the state that are not offset by reductions in nursing home spending (Price & 128

PAGE 140

Table 5.15. Medicaid Expenditure on Nursing Home Care for the Elderly, Per 1,000 State Elderly Population Basis, 1981 1988, Including Medicaid Expenditure on Home Care and 2176 Waiver Program in the Model B T Wages in the nursing home industry 4,641.4 10.6 Old elderly ( 85)/elderly ( 65) population 748.2 1.9 Elderly voting pop. 65)/voting 18) pop. -250.3 -2.6 Average state income 4.9 1.5 Unit cost of long-term care to state taxpayers -4,475.2 -8.6 Percent state unemployment 4,050.0 1.9 Percent state elderly population living in SMSA 280.2 1.0 Average persons per household -3,117.1 -.1 Average state temperature -174.2 -1.5 Year -14,726.1 -4.7 Elderly included in "medically needy" Medicaid 23,341.0 2.4 eligibility category Number of optional benefits to elderly in state -3,004.3 -3.2 Medicaid Program Medicaid per diem nursing home reimbursement 5,291.4 7.9 rate State/local administration of Medicaid Program 48,762.3 5.5 Number of nursing home beds/1,000 elderly 4,720.1 12.8 population Medicare acute care hospital length of stay 1,631.4 1.6 Liberal quotient of Congress'! vote record 1,047.2 3.1 Medicaid Spending on Non-Institutional Long2,478.3 4.2 Term Care Services for the Elderly (Home Care and 2176 Waiver Program) Adjusted R2 .81 Statistically significant at p = :::;; .05 129

PAGE 141

Table 5.16. Elderly Medicaid Recipients of Nursing Home Care, Per 1,000 State Elderly Population, 1981 1988, Including the number of Elderly Medicaid Recipients of Home Care in the Model B T Wages in the nursing home industry .35 7.4 Old elderly ( 85)/elderly ( 65) population .01 .09 Elderly voting pop. ( 65)/voting ( 18) pop. -.06 -3.0 Average state income .00 -.3 Unit cost of long-term care to state taxpayers -.47 -7.2 Percent state unemployment .62 2.2 Percent state elderly population living in SMSA .11 3.2 Average persons per household 5.3 1.0 Average state temperature -.01 -.5 Year -1.0 -3.0 Elderly included in "medically needy" Medicaid 4.0 3.8 eligibility category Number of optional benefits to elderly in state -.37 -3.4 Medicaid Program Medicaid per diem nursing home reimb. rate .18 2.4 State/local administration of Medicaid Program 5.0 3.1 Number of nursing home beds/1,000 elderly .70 16.1 population Medicare acute care hospital length of stay -.96 -2.3 Liberal quotient of Congress'l vote record .12 4.0 Medicaid Recipients of Home Care .68 3.2 Adjusted R2 .68 Statistically significant at p = :5 .05 130

PAGE 142

O'Shaughnessy, 1988). Some researchers have suggested that home health services do, in fact, substitute for nursing home care services and have lowered the projected growth in demand for nursing home beds (Harrington et al., 1988). The most common outcomes measured in studies of community-based care relate to cost and use of hospital and nursing home services (Hawes, 1988). Community-care based programs should not be viewed merely as alternatives to nursing home care. There are a full range of issues and situations surrounding home and community based services, i.e., increase in client and caregiver well being, physical functioning, mental/cognitive status, satisfaction with services, etc., that also measure the effectiveness of community care. Weissert warns against the "costeffectiveness" trap in evaluating home and community services (Hawes, 1988; Kemper, Applebaum & Harrigan, 1987). Private Expenditure on Nursing Home Care for the Elderly, Per 1.000 State Elderly Population, 1987 and 1988 The hypothesis states that private spending on nursing home care at the state level is associated with state economic variables (wages in the nursing home industry [ + ], average state income [ +], and percent state unemployment [-]) and demographic variables (percent "old" elderly in the elderly population [ +] and average persons per household [ ]). Two state-selected variables were expected to significantly effect private spending on nursing home care, namely, the number of nursing home beds per 1,000 elderly state population ( +) and the average LOS of the Medicare population in acute-care hospitals (-). Furthermore, it was anticipated that states in colder climates would have greater private spending on nursing home care. The results of the regression analysis indicate that private expenditure on nursing home care for the elderly is driven, primarily, by state economic and demographic factors (Table 5.17). 131

PAGE 143

,_. l.JJ N Table 5.17 Private Expenditure on Nursing Home Care for the Elderly, State Level, 1987 and 1988 -------I I Including New York I Excluding New York B T B T B T B Wages in the nursing home industry -1,917.4 -1.6 365.7 .22 1,298.0 -1.0 1,034.6 Old elderly (;;;: 85)/elderly (;;;: 65) population 5,936.5 4 .7 3,616.4 2.0 6 ,2 08 3 4.9 3 765.8 Elderly voting pop (";;;: 65)/voting (";;;: 18) pop 1,378 5 1.4 1,248.8 1.2 1,293.4 1.3 1,208.6 Average state income 30.0 2.6 32 1 2.1 27.2 2 .3 27 3 Unit cost of long term care to state taxpayers 1,547 3 .78 -1,999.2 .79 1 ,543. 0 .77 -1,786.8 Percent state unemployment -15,511.6 -1.8 -16,495.1 -1.8 -14,201.2 -1.6 -14,263.4 Percent state elderly population living in SMSA -1,230.1 -1.2 -782.2 -.69 -1,174.1 -1.1 -701.6 Average persons per household 76,293 9 .47 -136,237.4 .71 81,657.1 .50 -150,098.7 Average state temperature -6,199.8 -3.1 -6,421.9 -2.6 -5,969.1 -3.0 -7,009.6 Year 3,774.4 .14 -6,197.5 -.21 850 78 .03 -3,724 5 Elderly included in "medically needy" 28,672.1 .83 31,805.9 Medicaid eligibility category Number of optional benefits to elderly in state -2, 907.7 -.79 -3,062 8 Medicaid Program Medicaid per diem nursing home reimb. rate 421.1 .18 -483.7 State/local administration of Medicaid Program -54,819 7 -1.0 -29 227 2 Number of nursing home beds/ I ,000 eld. pop 4,006.7 3.0 3,763.4 Medicare acute care hospital length of stay 5 657 7 .31 11,289 1 Liberal quotient of Congress'! vote record -1,200.9 -1.2 -1,584 3 Adjusted R2 .60 .62 .61 .63 Statistically significant at p = 05 I T .59 2.1 1.1 1.9 -.70 -1.5 -.62 -.78 -2.8 -.13 92 .83 -.19 -.50 2.8 .59 -1.5

PAGE 144

The reduced model for 1987 and 1988 explains 60% of the variation in private spending on nursing home care for the elderly, while the expanded model explains 62% of the variation in private spending on nursing home care for the elderly. The economic factor found to be statistically significant in explaining state variation in private spending on nursing home care for the elderly per 1,000 state elderly population was the state average income ( +). This .finding is consistent with the hypothesis as average state income reflects the ability of an individual or family to bear out-of-pocket expenses for nursing home care. Conspicuously missing as a significant economic influence on private spending on nursing home care for the elderly is wages in the nursing home industry. It was hypothesized to reflect the input costs in the nursing home industry and the state cost of living. Private-pay pricing strategies appear driven by the market's ability and willingness to pay, and appear more removed from the cost function. This may be true only in a market with constrained supply. Once again, the possibility of multicollinearity arises as the correlation between wages in the nursing home industry and average state income is high for 1987 (.72) and 1988 (.84). The concentration of "old" persons in the elderly state population is a strong demand factor for private and public expenditures on nursing home care as disability has been demonstrated to rise with age. This demand for nursing home care translates into actual utilization of Medicaid nursing home care services as the supply of nursing home beds allows. This finding is consistent with the hypothesis. The number of nursing home beds per 1,000 state elderly population is the only state selected variable that has a statistically significant correlation with private expenditure on nursing home care per 1,000 state elderly population. Scanlon and others (1980) have argued that the private-pay sector of the nursing home industry may be the only segment in equilibrium as price fluctuates to equilibrate supply and demand. If the supply of nursing home beds increases the 133

PAGE 145

number of private-pay nursing home care recipients, without a concomitant change in price, Roemer's Law may be acting in this segment of the nursing home industry. It is difficult to say if increased nursing home bed stock (supply) is creating private-pay nursing home demand or allowing unrealized demand to translate into utilization (realized demand) as in the publicly subsidized segment of the nursing home industry. It appears that when all else is held constant, including nursing home bed stock, an increase in the demand for nursing home care as evidenced by relatively more II old II persons in a state's elderly population, increases private expenditures on nursing home care. This could be mediated by higher private-pay per diem nursing home rates or greater number of private-pay recipients of nursing home care, possibly in preference to Medicaid nursing home recipients. Temperature is the remaining statistically significant explanatory variable of private expenditure on nursing home care. The direction of the coefficient is negative as hypothesized. Colder climates may increase the number of private-pay nursing home recipients by erecting further barriers to independence, i.e., transportation in winter months, climate related health risks, trauma, and heating bills (Harrington & Swan, 1988). Average temperature may also increase average expenditure per recipient of private-pay nursing home care via increased costs, i.e., utilities, maintenance, and accident liabilities. It is not clear how directly input costs influence pricing strategies in the private-pay segment of the nursing home industry. Summazy Of Findings When all else is held constant in the analysis: 1. Higher wages in the nursing home industry are associated with all aspects of nursing home utilization (number of recipients and average and total expenditures), and with all aspects of 134

PAGE 146

home-care utilization when New York is included in the analysis (Tables 5.18 and 5.19). Wages in the nursing home industry reflect both input cost of labor and grossly, area cost of living. 2. States with a higher average income consistently spend more per elderly nursing home and home-care recipient, spend less on the 2176 Waiver Program, and spend more on private expenditures on nursing home care for the elderly. States with a higher average income have a greater ability to finance both a higher-priced Medicaid benefit package to nursing home and home care recipients, as well as private expenditures for nursing home care for the elderly. "Richer" states appear to increase the Medicaid LTC benefit package for the elderly, not eligibility. 3. States with a lower unit cost of LTC borne by state taxpayers have relatively more elderly Medicaid nursing home care recipients, which increases nursing home and total state Medicaid LTC expenditures. States with lower unit costs of LTC to state taxpayers (i.e. low nursing home wage states, low-average state income states with corresponding high FMMP, or states with high corporate income tax base) opt to increase access/eligibility, as opposed to altering intensity of service. "Poor" states appear to increase access while restricting the Medicaid LTC benefit package. 4 States With higher unemployment have more elderly Medicaid recipients of nursing home care, which drives Medicaid nursing home expenditures, total Medicaid LTC expenditures, and a higher percent of the state Medicaid LTC budget devoted to nursing home care as opposed to home/community care. State unemployment acts as a demand variable regarding Medicaid nursing home care for the elderly. 5. Excluding New York, states with lower unemployment have relatively more elderly Medicaid home-care recipients, spend more per elderly Medicaid home-care recipient, and, 135

PAGE 147

-w 0\ Table 5.18. Summary of Statistically Significant Findings, Inclusive of New York Total # Avg Home LTC N H N.H N H Exp. Care Exp Exp. Rec Per Rec. Exp Wages in the nur s ing home industry + + + + + Old elderly 85)/elderly 65) population + Elderly voting pop. 65)/voting 18) pop ---Average state income + Unit cost of long-term care to state taxpayers --Percent state unemployment + + + Percent state elderly population living in SMSA + --Aver a ge persons per household Average state temperature Year --+ Elderly included in medically needy Medicaid + + + eligibility category No. of optional benefits to elderly in state ---Medica i d Program Medicaid per diem nursing home re i mb. rate + + + + -State/l o cal administration of Medicaid Program + + + + + Number ofN. H beds/1 000 eld population + + + -Medicare acute care hospital length of stay + -+ + Liberal quotient of Congressional voting record + + + -Adjusted R2 78 79 66 69 64 ---#Home Avg Home 2176 N H. Exp. Priv Care Care Exp Waiver as % of N.H. Rec. Per Rec. Exp. LTC Exp. Exp + + -+ + -+ -+ + -+ -+ --+ --+ + -+ + + -+ --+ + + + --+ 48 48 33 40 62

PAGE 148

...... w ...:I Table 5.19. Summary of Statistically Significant Findings, Exclusive of New York Total # Avg Home #Home LTC N .H. N.H. N .H. Exp. Care Car e Exp Exp Rec. Per Rec. Exp Rec Wages in the nursing home industry + + + + Old 85)/elderly 65) population + Elderly voting pop. 65)/voting 18) pop. ---Average state income + Unit cost of long-term care to state taxpayers --Percent state unemployment + + -Percent state elderly population living in SMSA + --Average persons per hous ehold -Average state temperature Year ----+ + Elderly included in "medically needy" Medicaid + + + eligibility category Number of optional benefits to elderly in state ---Medicaid Program Medicaid per diem nursing home reimb rate + + + + --State/local administration of Medicaid Program + + + -Number of nursing home beds/ I ,000 eld + + + -population Medicare acute care hospital length of stay -+ + Liberal quotient of Congressional voting record + + + + + + Adjusted R2 73 75 66 64 30 23 Avg Home 2176 N .H. Exp Priv. Care Exp. Waiver as % of N.H. Per Rec. Exp. LTC Exp. Exp -+ + + -+ -+ -+ -+ --+ -+ + + -+ -+ + + -+ + -39 35 39 63

PAGE 149

therefore spend more overall on Medicaid home-care services for the elderly. State unemployment acts as an economic indicator of II ability to pay II regarding Medicaid home-care services for the elderly. 6. States with fewer persons age 65 or older relative to the state population age 18 or older, had relatively more elderly Medicaid nursing home care recipients that influenced higher Medicaid nursing home care and total Medicaid LTC spending for the elderly. The opposite was hypothesized. It was expected that a higher concentration of elderly voters would increase political support for Medicaid services for the elderly resulting in increased access and Medicaid spending on LTC for the elderly. It now appears that a higher concentration of potential voters (age 18 and older) to elderly voters (age 65 and older) may, more importantly, reflect a higher working to non working state population with which to financially support publicly subsidized LTC services for the elderly. 7. States with a greater concentration of "old" elderly (age 85 and older) in their elderly population spend relatively more on nursing home care (both Medicaid and private pay), spend less on the Medicaid 2176 Waiver Program, and consequently, spend a greater portion of their Medicaid LTC budget for the elderly on institutional LTC services The "old" elderly have, in general, greater disability which may require institutional care, while the "young" elderly can more often be sustained in the community. 8. States with a greater percentage of their elderly living in metropolitan areas have relatively more elderly Medicaid recipients of nursing home care and spend relatively more on the Medicaid 2176 Waiver Program. It is not known if the mechanism for greater spending on the 2176 Waiver Program is due to a greater number of recipients, or average expenditure per recipient. States with a relatively greater rural elderly population spend more per elderly Medicaid recipient of both nursing home and home-care services, and spend more, overall, on Medicaid 138

PAGE 150

home care. LTC input costs may be higher in rural settings, or states with a greater rural elderly population may, in fact, offer a greater home care and nursing home benefit package, while offering services to fewer elderly recipients. 9. States with more persons per household have fewer elderly Medicaid home-care recipients, spend less on the 2176 Waiver Program, and spend a greater portion of their Medicaid LTC budget for the elderly on nursing home care. More persons in a household appears to substitute informal for formal home and community care, but not for nursing home care for the state Medicaid elderly population. 10. Colder states are associated with higher-average Medicaid expenditures per home-care recipients and higher private-pay expenditures on nursing home care. Once again, the mechanism for "private pay" is not known, i.e., more private-pay nursing home care residents versus higher average private-pay expenditures per resident in colder states. Colder temperatures may increase input costs of LTC, i.e., fuel bills, and erect further barriers to independent community living for the elderly, thus increasing the number of recipients of LTC. Temperature may also be associated with regional variation in LTC utilization which may reflect cultural or ideological differences in the nation that are -not accounted for in the analysis. 11. In the early 1980s, states spent relatively more on institutional (nursing home) LTC services for the elderly than in later years. This parallels the evolution of the LTC field toward providing care in a setting of least restriction to allow for "aging in place" when possible. 12. Including the elderly in the optional Medicaid eligibility "medically needy" category increases the number of elderly Medicaid recipients of nursing home care which, in turn, is associated with higher Medicaid nursing home care expenditures, overall Medicaid LTC expenditures, and percent Medicaid LTC budget spent on nursing home care. This variable is negatively associated with non-institutional Medicaid 2176 Waiver spending. States that allow their 139

PAGE 151

elderly to qualify for Medicaid because of their medical expenses spend relatively more on institutional LTC and relatively less on non-institutional LTC for the elderly. 13. Conversely, states that have a greater number of optional Medicaid services offered to the elderly spend relatively more on non-institutional LTC services for the elderly (the 2176 Waiver Program) and relatively less on all aspects of nursing home care for the elderly. Targeting optional Medicaid LTC services has been offered as a strategy for reducing the increase in nursing home-care utilization. 14. Higher Medicaid per diem nursing home reimbursement rate is positively associated with all aspects of Medicaid nursing home care utilization, overall Medicaid LTC expenditures, and higher percent Medicaid spending on institutional care. Conversely, higher Medicaid per diem nursing home reimbursement rate is associated with lower Medicaid home-care expenditure with fewer elderly Medicaid recipients of home-care services. It was hypothesized that higher Medicaid per diem nursing home reimbursement rates would be associated with higher average expenditures per home-care recipient and thus, higher overall home-care expenditures. States with high per diem nursing home reimbursement rates may be hesitant to expand LTC services for home and community care as home and community care has not yet been shown to reduce overall Medicaid LTC expenditures. Elderly Medicaid recipients in states with a higher Medicaid nursing home per diem rate appear to have greater access to nursing home care but less access to home care. States with a higher Medicaid per diem nursing home reimbursement rate spend relatively more on institutional care and less on non-institutional LTC for the elderly. 15. Local administration of the Medicaid program is associated with more elderly Medicaid recipients of nursing home care, higher Medicaid expenditures on nursing home care, and higher overall Medicaid expenditures on LTC for the elderly. Local administration of the Medicaid program is also associated with higher Medicaid home-care expenditures for the elderly 140

PAGE 152

when New York is included in the analysis, while state administration of the Medicaid program is associated with more elderly Medicaid recipients of home care when New York is excluded from the analysis. Local administration of the Medicaid program may increase access of the elderly Medicaid population to nursing home care, while state administration of the Medicaid program may increase access of the elderly Medicaid population to home care. 16. States with a relatively greater nursing home bed supply have more elderly Medicaid nursing home care recipients, spend relatively more on private and public monies on nursing home care for the elderly, have relatively fewer elderly Medicaid recipients of home care, and spend relatively less on the Medicaid 2176 Waiver Program. Greater nursing home bed stock is associated with more institutional long-term care of the elderly and less non-institutional long-term care for the elderly. 17. Generally, states with shorter LOS of their Medicare population in acute-care hospitals are associated with relatively more elderly Medicaid recipients of nursing home care and relatively greater Medicaid spending on the 2176 Waiver Program. Conversely, states with longer LOS of their Medicare population in acute-care hospitals are associated with higher average Medicaid expenditures per elderly recipient of both home care and nursing home care services and higher overall home-care expenditures. There is a strong positive correlation of .54 between wages in the nursing home industry and average state LOS that may confound the relationship between LOS and average expenditure per elderly recipient of Medicaid LTC services (Table 5.5). 18. Finally, states with a more liberal Congressional voting record having greater Medicaid nursing home utilization, (greater number of elderly Medicaid recipients and higher average expenditures per recipient), spend relatively more on the 2176 Waiver Program, and spend relatively more, overall, on Medicaid LTC services for the elderly. When New York is excluded from the analysis, liberal states have relatively more elderly Medicaid recipients of home-care 141

PAGE 153

services, have relatively higher Medicaid expenditures on home-care services for the elderly, and spend a smaller portion of their Medicaid LTC budget for the elderly on institutional care. In general, political liberalism is associated with greater Medicaid spending on both institutional and home and community long-term care for the elderly. 142

PAGE 154

CHAPTER 6 CONCLUSIONS The general hypothesis of this study states that both endogenous socio-economic and demographic state variables and state-selected policy and political variables profoundly effect LTC spending and utilization patterns for the elderly. To that end, reduced and expanded models were utilized to explain interstate variation in LTC service mix (nursing home, home care, and the 2176 Waiver Program) and components of spending (average expenditure per recipient, number of recipients, and total spending). The reduced model contained macrocontextual economic and demographic variables endogenous to the state while the expanded model added state-selected variables regarding the structure of the Medicaid program, medical-care marketplace, and political ideology. The results of this study indicate that endogenous state characteristics have a stronger relative influence on average expenditure per elder Medicaid recipient of LTC services while structural of a state's Medicaid program and medical marketplace have a stronger relative effect on access of the state's elderly to Medicaid LTC services. Access of the elderly to publicly subsidized LTC services is largely a function of the choices states make regarding optional Medicaid eligibility and services, reimbursement rates, nursing home bed supply, administrative level of the Medicaid program, and political ideology. In short, there is little evidence from this study to characterize state Medicaid LTC services for the elderly as a "runaway" entitlement. This study lends empirical support to each of the theoretical models discussed in the literature review relative to specific aspects of LTC spending for the elderly. The theoretical 143

PAGE 155

models incorporated into this study are the cultural model, the medical model, the structural model, the public-goods model, the political model, and the economic model. The cultural model suggests that the social, racial, and ethnic mix of a community may differentially affect Medicaid LTC spending or LTC utilization patterns of the elderly. Of particular importance are the living arrangements of elderly persons, i.e., living alone has been identified as a risk factor for nursing home placement once disability occurs, and the availability and willingness of family members to supply informal (unpaid) support. The results of this study indicate that larger households substitute informal for formal community LTC for the elderly (number of elderly Medicaid recipients of home-care services and Medicaid 2176 Waiver Program expenditures) but do not substitute for institutional (nursing home) care. Much more work is needed to understand the attitudes, abilities, and needs of the informal caregiving network in order to support this network with appropriate public policy. The medical model predicts that LTC spending and utilization will reflect the level of disability in the state population. Since disability with corresponding functional impairment has been found to increase with age, the over 85-year-old population would have the greatest "need" for LTC services, particularly nursing home care. The results of this study indicate that states with a greater concentration of "old" elderly in their elderly population do spend relatively more public and private dollars on nursing home care for their elderly residents. Concentration of "old" persons in a state's elderly population is statistically significant regarding overall Medicaid expenditure on nursing home care, but is statistically significant regarding the components of Medicaid nursing home care spending (average Medicaid expenditure per elderly recipient and number of elderly Medicaid nursing home care recipients) in the reduced models only. In the expanded model, the Medicaid per diem nursing home reimbursement rate becomes a strong indicator of average Medicaid expenditure per elderly Medicaid recipient of nursing home care 144

PAGE 156

services, while the nursing home bed stock and the inclusion of the elderly in the optional "medically needy" eligibility category become strong explanatory variables regarding number of elderly Medicaid recipients of nursing home care. While there are strong correlations between state Medicaid and medical-care marketplace variables with macrocontextual state variables, i.e., per diem nursing home reimbursement rate with wages in the nursing home care industry (+.51) and number of "old" elderly in the state elderly population with nursing home bed stock ( + .63), it appears that the structure of a state's Medicaid program and health-care system exert a distinct and powerful influence on LTC utilization. Furthermore, adding state policy and political variables to the equation increased the explained interstate variation in number of elderly Medicaid nursing home care recipients from 23% to 66%. The structural model, particularly as it relates to access of the state elderly population to Medicaid subsidized nursing home care, is reinforced by this study. Previous research (Schneider, 1988; Buchanan et al., 1991) indicates that local, as opposed to state, administration of the Medicaid program is positively related to overall state Medicaid spending. This study suggests that states with local administration of their Medicaid program also have relatively greater nursing home and overall LTC expenditures for their elderly. More importantly, the mechanism has been elucidated. Local administration of the Medicaid program enhances access (number of elderly Medicaid recipients of nursing home care) of the elderly to institutional LTC services. Local administrators of the Medicaid program appear more sensitive to the "needs" of local constituents regarding intense institutional care. Local bureaucrats may actually be responding more to the needs of the informal caregiver network, typically family members, who frequently apply for Medicaid nursing home placement for an elderly family member State administration of the Medicaid program is associated with greater access of the elderly state population to Medicaid home-care services when New York is excluded 145

PAGE 157

from the analysis. Central administration may allow for a speedier development and diffusion of newer programs into the community Clearly, access of the elderly to state Medicaid LTC services is a function of the three levels of government. The federal government has prescribed mandates regarding eligibility and benefits and affects the cost of the Medicaid program to state taxpayers via the FMMP. The state government selects optional Medicaid eligibility, services, reimbursement rates, and nursing home bed stock, each of which has been demonstrated to influence access of the elderly to LTC. Local administration of the Medicaid program appears to enhance access of the elderly to nursing home care but may retard the development of newer home and community services. The results of this study are inconclusive regarding the public goods and political models as they relate to Medicaid spending on LTC for the elderly. The public goods model suggests that services underwritten by government as a "public good" represent the values, preferences, and economics of the voting population, more specifically, the median voter. Furthermore, if the totality of median voter demand was represented in the equation, the political process becomes a mere conduit for the expression of that demand and does not exert a separate and distinct influence. There is some evidence from this study that publicly subsidized LTC for the elderly does reflect state taxpayer status if not median voter status, i.e., lower unit cost of LTC to state taxpayers is associated with greater access and relative spending on LTC services for the elderly. States' liberal quotientis also associated with greater access and relative spending on LTC for the state elderly population. These results suggest that either the political process is not merely a conduit for the expression of taxpayer and median voter demands, or that the variable "liberal quotient of the state" reflects characteristics of the taxpaying and voter population which are not accounted for in the model. It is also possible that the liberal quotient of a state, as measured by the voting record of Congress, is not a sufficient measure of a state's political environment. Legislative 146

PAGE 158

action on general issues may offer little influence on policy outcomes concerning Medicaid or LTC. Other aspects of a state's political process may bear influence on Medicaid expenditures on LTC for the elderly i.e.,-5pecial interest concerns, inter-party competition, implementation issues, and bureaucratic decision making within the Medicaid organization. Private spending for nursing home care for the elderly reflects both the "willingness and ability" of persons to pay out-of-pocket expenses for care (average state income) and the state "need" for services (concentration of "old" elderly in the elderly state population). State-selected policy and political variables do not influence private spending on nursing home care, with the important exception of nursing home bed stock. In his discussion of the application of the economic model to the nursing home care industry, Scanlon (1980) states that demand for LTC services refers to the number of persons who wish to consume a LTC product at a certain price, while aggregate deinaild depends on consumers' preferences, tastes, financial resources, prices of other goods and services, and the availability of informal caregivers. Utilization (the actual purchase or consumption of nursing home care) depend on the aggregate decisions of potential consumers, providers, and government policy. Normally, market mechanisms allow for the equilibration of demand and utilization via the mechanics of supply and price. Private spending on nursing home care approaches the economic model in this study in that state Medicaid policy and political variables were not found to influence private spending on nursing home care in any statistically significant way. The "private pay" sector of the nursing home industry still cannot be construed as a true free, economic market as the vital "supply" variable (nursing home bed stock) is heavily regulated by state government. In conclusion, socio-economic and demographic state variables, in conjunction with state selected policy and political variables, explain close to 80% of the interstate variation in Medicaid long-term care spending for the elderly. State policy variables are particularly influential regarding 147

PAGE 159

access of the elderly population to Medicaid LTC services. Again, Medicaid spending on LTC for the elderly is characterized inappropriately as a "runaway" entitlement program. Rather, state spending on LTC for the elderly is a function, largely, of choices made by federal, state, and local government over the past 30 years. Policy Considerations The results of this study are important to policy makers involved in the LTC issue for four specific reasons: first, policy makers will better understand the determinants of LTC spending; second, policy makers are better able to identify the determinants of LTC spending that are amenable to policy manipulation; third, policy makers are better able to evaluate LTC spending and utilization patterns relative to federal and state objectives; and finally, policy makers are better able to predict future LTC expenditures under different scenarios. This study offers empirical support that average Medicaid expenditures per recipient of LTC services are driven, primarily, by socio-economic state variables, specifically, wages in the nursing home industry and average state income. Public policy can do little to influence directly the input costs of LTC, short of wage or price freezes. Indirectly, states and/or the federal government can alter reimbursement to providers creating incentives to either increase or decrease inputs into the delivery of long-term care. For instance, the Weld administration of Massachusetts recently (November, 1995) proposed a new Medicaid nursing home reimbursement system that changes (reduces) the way Massachusetts nursing homes are reimbursed for money they spend on nursing care, food, and for capital used to build nursing homes. The Weld administration argues that the proposed reductions in Medicaid per diem reimbursement rates are an attempt to promote efficiency in "one of the most expensive nursing home systems in the country" (Stein, 1995). Susan Bailis, president of the Massachusetts Extended Care Federation, counters that, "These are 148

PAGE 160

the most regressive regulations we have ever seen. They threaten to destroy the quality of care." Bailis also states that nursing homes would be forced to lay off staff, slash wages, and cut the quality of the food they offer residents. The proposal is also intended to create a disincentive to adding "more and more nursing home beds." The Weld proposal to reduce Medicaid nursing home per diem reimbursement rate was later abandoned in favor of a 2.5% increase for 1996 in return for a commitment from the Massachusetts Extended Care Federation to develop a managed care program for Medicaid LTC recipients (1995, December 19, Boston Globe). Clearly, federal and state policy regarding reimbursement can have a profound affect on inputs, access, and delivery of Medicaid LTC services. This study suggests that the elderly in states with higher Medicaid per diem reimbursement rates have greater access to nursing home care. It is likely that the Medicaid elderly in states with higher Medicaid per diem reimbursement rates compete more successfully with "private pay" clients for a limited nursing home bed stock. This study also demonstrated that "private pay" spending on nursing home care for the elderly reflects "ability and willingness" of persons to pay out-of-pocket for services. Increased demand (greater number of "old" elderly in the state population) coupled with greater ability to pay (indicated by higher average state income and lower unemployment) and greater nursing home bed stock correlated with greater private spending on nursing home care. This bodes well for the demand for "private pay" nursing home care, which is expected to triple by the year 2030, as the elderly are the fastest growing segment of the population and the financial profile of the "young" elderly has improved markedly (Price & O'Shaughnessy, 1988). Proprietary nursing homes in states that allow both price discrimination between "private pay" and Medicaid nursing home care recipients and a generous nursing home bed stock appear well positioned regarding demand and profitability for the next three decades. It seems logical that administrators of nursing homes in states with wide discrepancies between "private pay" and 149

PAGE 161

Medicaid per diem nursing home rates will preferentially serve the "private pay" market. Administrators may even hold empty beds until "private pay" residents are available. States could, of course, act to require nursing homes to "set aside" a certain portion of their beds to serve the Medicaid population. Horizontal inequity regarding the distribution of Medicaid benefits across states has long been acknowledged in the Medicaid system. This concept is clouded by the fact that no one accepted operational definition of "equity" exists in medicine or Medicaid. Stuart, Reutzel, and Reutzel (1985) argue, correctly, that broad definitions like "uniform treatment of persons in like circumstances" presents the question of what constitutes "treatment" and "like circumstances." By convention, issues of eligibility, benefits package, and service accessibility are addressed as sources of inequity in the Medicaid program. Horizontal equity is defined, for this discussion, as equal real LTC Medicaid expenditures per elderly population across all states, i.e., adjusted for both temporal and regional medical CPl. Real LTC Medicaid expenditures have two components-number of recipients and expenditure per recipient. This study confirms large disparities in both areas, further begging the question of horizontal inequity in the Medicaid program (Table 5.2). Interstate variations in nominal Medicaid expenditure amounts and in Medicaid spending levels adjusted for yearly CPI only are not equivalent with horizontal inequities in the Medicaid program. This study suggests that a significant portion of the state variation in Medicaid expenditures per elderly state population is related to the cost of inputs (labor) at the state level. Stuart, Reutzel and Reutzel (1985) found that half or more of the interstate differences in Medicaid spending for elderly recipients of Social Security Income (categorically eligible for Medicaid services) were not due to inequity issues, i.e., actual differences in service levels, but rather to actuarial or efficiency of state administration factors. 150

PAGE 162

Assuming that upwards of half of the state variation in Medicaid spending on LTC represents true horizontal inequity in access of the elderly to various intensity of LTC service, the federal government may intervene in several ways. First, the federal government can affect disparities in eligibility, benefits, and supply of services across states by federal mandate. Federal mandate may equalize eligibility standards and benefit packages or ease regulatory mechanisms now restraining the supply of nursing home beds. Second, the FMMP could be enhanced to poor states, creating a more powerful incentive to equalize real Medicaid LTC expenditures across states. Grannemann (1979) found that poor persons in poor states are confronted with more stringent Medicaid eligibility requirements with fewer benefits than persons in wealthier states. Horizontal equity (or inequity) has been further advanced by arbitrary benefit limitation imposed as cost saving strategies. For instance, Medicaid programs have set limits on hospital days, physician visits, and provider reilnbutsement rates. Such limits can reduce benefits without regard to recipient need. Case management has been introduced by some states as a means to better match need with services in resource-restrained environments. The degree to which one views state variation in access and LTC service level for the elderly Medicaid population as problematic, in that it may reflect horizontal inequities, is a matter of personal and political philosophy. Clearly, federal efforts to equalize real Medicaid benefits to the elderly across states was a greater priority with the Clinton administration as evidenced by the long-term care component of his "Universal Access" Healthcare Reform bill of 1994, but is not a priority of the GOP Budget Reform Proposal of November, 1995 which proposes to reform Medicaid via block grants to the states. This Medicaid restructuring initiative is discussed in a later section. Federal mandates, particularly unfunded federal mandates, currently assure that a portion of the cost of the Medicaid program is borne by state taxpayers. This study demonstrates that the 151

PAGE 163

price of LTC to state taxpayers negatively influences the number of Medicaid recipients of nursing home care. Federal policy directly affects the price of LTC to state taxpayers via the FMMP. The FMMP affects not only horizontal equity, but also taxpayer equity, as the matching rate fonnula is not designed to equalize tax burdens of the Medicaid program across states. Taxpayer equity is defined as the same fraction of taxpayer income being devoted to state Medicaid expenditures in all states. Grannemann (1979) found a redistribution of federal Medicaid dollars toward wealthier states, as many low-income states have found it too expensive to make Medicaid benefits widely available despite favorable Medicaid matching rates. Taxpayer equity affects access and service mix availability of the elderly to Medicaid long-term care. Many states appear to substitute intensity of services (benefit packages) for access (number of recipients). States decide on optional benefits, but must deliver federally mandated benefits. States have greater discretion in the number of recipients of LTC services because states set the income level for SSI and AFDC, the two categorically needy Medicaid eligibility programs. This study indicates that state-selected variables (number of nursing home beds, including the elderly in the optional medically needy category, and offering optional Medicaid benefits) strongly influence the number of state elderly Medicaid nursing home care recipients. Constraining the state nursing home bed stock and excluding the elderly from the optional medically needy program effectively restrict the number of nursing home care recipients, and thus contain Medicaid expenditures on nursing home care. It has long been proposed that states do in fact restrict bed stock as a cost containment strategy. It has been strongly argued that this cost-containment strategy transfers the cost of LTC to a potentially less solvent informal caregiving network and results in an unmet need of the most frail elderly (Harrington et al., 1988; Scanlon & Feder, 1980). Policy makers must contend with the ethical dilemma of reducing public expenditures on LTC for the elderly at the expense of those in need of LTC and their families. 152

PAGE 164

This study further finds that providing more optional Medicaid benefits may actually reduce the number of nursing home care recipients. Targeting optional benefits toward specific groups at an otherwise high risk for institutionalization may reduce nursing home utilization. The use of optional benefits as a cost-containment strategy must be viewed skeptically, though, as the aggregate number of optional Medicaid benefits was negatively associated with number of elderly Medicaid recipients of nursing home care, but was associated with neither reduced expenditures per nursing home care recipient nor overall Medicaid spending on nursing home care. The overall, long-term trend in Medicaid spending has been toward greater portions of the Medicaid budget being allocated to LTC, especially for the elderly. Evidence indicates that, in recent years, a decreasing proportion of Medicaid dollars has been spent on nursing home care (Swan, 1990). Intergenerational equity issues arise as poor and near-poor children and non-elderly adults are denied access to medical care, while the elderly are supported by both the Medicare program and, to a large extent, the Medicaid program. Currently, about 70% of the Massachusetts nursing home care population is supported by Medicaid dollars (Stein, 1995). Nationally, about two thirds of nursing home residents are supported by Medicaid dollars (Toner, 1995). Age restrictions on many LTC programs restrict access of the non-elderly who require LTC services even though approximately one third of the LTC population is non-elderly (Harrington, Cassel, Estes, et al., 1991). Grannemann (1979) postulates that states use Medicaid funds to "buy into" the Medicare program when providing care for the elderly. States use Medicaid funds to pay the Medicare Part B premium for the elderly who are unable to afford the premium. Thus, state funds are matched twice by the federal government, once via the Federal Medicaid Matching Percent and once again via Part B of Medicare. This system redistributes funding for services toward the elderly via a cross-generational subsidy as the monies come from general federal revenues 153

PAGE 165

(Samuelson, 1995). The federal government needs to examine the incentives it provides to rational economic state actors in both the Medicaid and Medicare systems relative to the redistribution of resources toward the elderly. Policy makers must look beyond the LTC system when allocating Medicaid dollars for the elderly. Significant cost shifting occurs between acute and chronic care, institutional and non institutional care, formal and informal care, medical and social services, medical and psychiatric services, and elderly and non-elderly services. Policies that increase equity, access, or efficiency in one system or for one recipient group may well decrease those values in other systems or for other groups. Useful policy analysis must present alternative plans of action and predict the outcomes of those plans under various scenarios. The microsimulation models traditionally used to forecast LTC spending and utilization have been driven by demographic factors under various economic scenarios (Brookings ICF Model in Rivlin & Weiner, 1988). This research augments these demographic/economic models by incorporating indirectly the behavioral responses of the providers to state policy and the state to federal policy. Integrating these behavioral responses allows analysts to predict better the long-term care expenditure levels and utilizations patterns and plan accordingly. The Future of Medicaid and Long-Term Care Many private and public sector initiatives have been proposed for reforming the LTC financing and delivery systems. Private-sector strategies include home equity conversion, private long-term care insurance, continuing-care retirement communities, and individual medical accounts. Public-sector strategies include altering the tax code to shelter private LTC funds, consolidation of 154

PAGE 166

federal long-term care dollars into block grants to the state, and liberalizing Medicaid and/or Medicare (Rivlin & Weiner, 1988; Davis & Rowland, 1986). The most recent; serious attempt to restructure Medicaid is the Republican plan to reduce future Medicaid spending by $170 billion over the next seven years via federal block grants to the states. This reduction represents an 18-20% cut in projected Medicaid spending (PT Bulletin, 9/29/95; PT Bulletin, 11/10/95; Toner, 1995). The House of Representatives bill would require states to use MediGrant block funds for health care only. The bill eliminates federal mandates to health benefits for people who are poor, aged, or disabled. It would require states to set aside a percentage of the MediGrant fund on behalf of poor people with disabilities, women and children, nursing home residents, and the aged who cannot afford to pay Medicare premiums, but states would have flexibility in h<;>w the "set-aside" money is spent. The bill, as i(exists currently (11120/95), would allow every state a 7.24% increase in its federal subsidy for the first year. Subsequently, increases would vary from 2% to 9% per year, with a projected average of 4.9% annually instead of the current 10% (Advance, 10/23/95). The initial Republican bill eliminated federal nursing home standards leaving nursing home care quality to the -states without any federal oversight. The final bill approved by the House preserves most federal standards for nursing homes (1995, November 18, Zuckerman, Boston Globe). Republican proponents of the bill argue that states would find greater flexibility and efficiencies once freed from "onerous federal requirements and regulations" to deal more innovatively with issues like long-term care. Gail Wilensky, former President Bush's health advisor and HCFA administrator, predicted that the possible effects of Medicaid cuts would be blunted by state initiatives to extend health insurance to some poor and near-poor citizens, 155

PAGE 167

combined with state subsidies of "last resort" health care providers dedicated to care of t!Ie poor (1995, November 9, Knox, Boston Globe). Opponents of the bill argue that spending limits imposed by MediGrant would result in approximately 2 million people losing coverage for long-term care, either nursing home or home care, by the year 2002 if the bill becomes law unless states raise taxes to offset the limits (1995, November 12, AARP in New York Times). As stated in a recent report on the Future of Medicaid by the Urban Institute (1995), "Large numbers of persons needing Medicaid assistance to pay for long-term care may no longer secure it. Quality of care is likely to be at risk, and greater burdens will fall on informal care givers." The results of this study could assist policy makers in predicting states' responses to the change in federal incentives. First, this study indicates that as unit cost of long-term care rises to state taxpayers, fewer elderly receive Medicaid-subsidized nursing home care. It is likely, once states have spent their "set aside" MediGrant nursing home care monies, that few will further increase the tax burden to state residents to enhance access of the elderly to nursing home care. States that are more likely to raise state taxes in an effort to offset federal MediGrant limits have either a larger fiscal capacity via a greater concentration of working to non-working (elderly) population, and/or states with a decidedly politically liberal bend. The typical state response is expected to be limiting access of the elderly to Medicaid nursing home (and probably) home care, thus creating an increased burden on the informal caregiver network. Some states may decide to offer tax incentives (e.g., home care tax deductions) to informal caregivers in lieu of directly supplying LTC services. If states choose to limit access of the elderly Medicaid population to nursing home care, this study indicates the most direct route would be to limit the number of nursing home beds. This cost-containment strategy may actually no longer be necessary in an "entitlement free" world, i.e., 156

PAGE 168

no longer will the indigent elderly who "qualify" for nursing home care automatically be entitled to that care by the Medicaid program, as is the current situation. States may choose to deregulate the industry to allow the proprietary nursing home industry to supply nursing home beds at will in hopes that increased competition will drive down the cost of nursing home care, or states could specify that proprietary nursing homes "set aside" a portion of their beds for the indigent elderly at a greatly reduced rate. As such, the "private pay" segment of the nursing home care market (which is projected to be robust for the next 30 years) will subsidize the indigent elderly. The proprietary nursing home care industry may support such a strategy in exchange for the freedom to stock beds. If the nursing home bed stock mushrooms because proprietary nursing home investors anticipate huge demand from the "private pay" market segment, access of the indigent elderly to nursing home care could actually be maintained or improved upon depending on the negotiations between the state and the nursing home industry. States may be in a stronger position to negotiate as "buyers" in an over-bedded as opposed to an under-bedded market. If states choose to continue to limit the nursing home bed stock, access of the Medicaid elderly to nursing home care could be further constrained if states also choose to reduce the Medicaid per diem reimbursement rate as a cost-containment strategy. This study indicates that states with higher Medicaid per diem reimbursement rates have more elderly Medicaid recipients of nursing home care and fewer elderly Medicaid home care recipients. States with expensive Medicaid programs with corresponding high per diem nursing home reimbursement rates have chosen to restrict their home care services as home and community services have, in the past, been shown to be an expansion, not a substitution, for institutional long-term care. If the Medicaid elderly are progressively squeezed out of nursing homes and nursing home care becomes a "private pay" commodity, i.e., a true economic good, Medicaid home and community services may become, by default, a real substitution for institutional care. 157

PAGE 169

In these cases local, community-based, cheaper alternatives like congregate homes may also become a viable Medicaid alternative with only the most severely handicapped elderly persons gaining access to institutional care via the Medicaid program. This scenario also allows for a heightened standard of care in nursing homes, in that persons with private funds would simply go elsewhere (if beds are available) if care did not meet their or their families' expectations. This private-sector, consumer approach to quality could catapult the industry forward in a way not yet accomplished by state and federal regulation. It is worth noting that this scenario is simplistic. Currently, the average age of the nursing home care population is 85 and about 60% have some form of dementia (Toner, 1995). The degree to which home and community services can safely, effectively, or efficiently be substituted for institutional care for the functionally impaired, demented elderly is certainly uncertain. If access of the Medicaid elderly to nursing home care is restrained by MediGrant (which is likely if relative nursing home bed stock and/or reimbursement rates are reduced), and if Medicaid home and community services do not mushroom as a substitute, hospitals can expect a greater number of "administratively necessary days" in the Medicare program as the elderly in acute-care hospitals await placement. This cost shifting onto the Medicare program may be disallowed depending on the future restructuring of the Medicare program. Also, the Medicare program is currently used as a "buy in" for the Medicaid elderly into nursing homes. Elderly Medicaid recipients usually leave acute-care hospitals with as many as 100 Medicare Part A SNF days. These "Med A" days carry a very attractive reimbursement rate (up to $400 per day in some instances). If the Medicare program is restructured to reduce or eliminate these Medicare Part A SNF days, the Medicaid elderly will find their access to nursing home care further reduced. Currently, once nursing home care residents convert from Medicare to Medicaid payment, it is illegal in many states for nursing homes to require the resident to leave. 158

PAGE 170

It appears likely that states will respond very differently to the changes proposed by the MediGrant program, should it become law. The across-state variation in the Medicaid program in eligibility and services has been well documented historically and frequently referred to as inequitable. The MediGrant bill is expected to heighten across-state variation in eligibility and intensity of services for all groups. In a recent article in the New York Times, "Critics Say Republican Budget Will Create Shortage of Nursing Home Beds for Elderly" (Toner, 1995), Gov. Howard Dean, a Democrat from Vermont, said The people who will be outraged by this are patients and their families. This will cause us to go back to an era when patients were cared for at home. In some cases this could be a positive development, to the extent it encourages more disabled elderly to stay in their homes, with the help of home health aides or other services. It could also give states the flexibility to care for people in less institutional; and less expensive places, like assisted-care residences. We're going to have to restructure long-term care. Some of the things will be good, but the transition will be painful. MediGrant, and other long-term care restructuring initiatives, offer both a great threat and a great opportunitr to meet the needs of the functionally impaired. To the extent that MediGrant! or any initiative, alters federal subsidies to the state and the state taxpayer, redesigns Medicaid or Medicare eligibility criteria/benefit packages/or reimbursement systems, or influences the local health care environment, it will affect access, utilization, and cost of LTC services. The results of this study allow analysts to predict more accurately the direction and magnitude of the effect of any restructuring initiative on long-term care for the elderly. Predicted outcomes must then be examined for congruence with policy objectives and norms and values of local, state, and national communities. 159

PAGE 171

Implications for Further Research While the issues discussed in this dissertation have raised and answered a number of critical questions, surely additional work is needed to address the complex issues surrounding long term care. This study raises important questions of methodology, interpretation, and social relevance. The factors included in this model were reasonably independent of one another, with the exception of economic indicators of state wealth/cost of living, i.e., wages in the nursing home industry, average state income, unit cost of LTC to state taxpayers, and Medicaid per diem reimbursement rate (Table 5.5). Standard errors for the average number of persons per household and average state temperature were relatively large, creating confidence problems in interpretation. It is possible that these variables correlate with important characteristics of the elderly, general, or voting population of a state, or with state political or health care variables which were omitted from the model. Number of persons per household may be a poor proxy for the informal caregiver network. Other researchers have utilized state female workforce statistics as a reflection of the available informal caregiver network (Swan & Benjamin, 1990). Potentially important state-selected variables that have been omitted from this model include an index of the special interest concentrations in the state for both providers and consumers of LTC and general Medicaid services. Some measure of the ability or propensity of the state Medicaid bureaucracy to innovate regarding new programs may help explain the distribution of Medicaid dollars toward non-institutional LTC. The breakdown of Medicaid LTC spending into its components of recipients and average expenditure per recipient is a major advantage of this study. Aggregating recipient and average expenditure data in one expenditure measure provides little information regarding access to services or intensity of services. Intensity of service (average expenditure per recipient) is expressed as a 160

PAGE 172

dollar figure only, as a regional cost of living deflator was not employed. This is a limitation of the study. Expenditure amounts are equivalent to intensity of service only when expenditure amounts are deflated by both temporal and regional CPl. This study suggests several directional relationships that require further quantification. The price of LTC to the average taxpayer is negatively correlated with number of recipients of nursing home services. What is the magnitude of the change in number of recipients of nursin home versus home care recipients when the price of care changes to the taxpayer? Is access to Medicaid home health care more elastic than nursing home care relative to the price of LTC to the taxpayer? Transforming the model into log form would further quantify these relationships. Temporal influences need to be clarified regarding adjustment lags between independent and dependent variables. Some independent variables may have immediate affect on Medicaid expenditures, i.e., including the elderly in the medically needy program would almost immediately increase the number of eligible recipients of nursing home care, and the number of nursing home beds (IF available to the Medicaid population) would almost immediately affect the number of recipients of nursing home care. Other independent variables may lag in their influence on Medicaid expenditure on LTC for the elderly, i.e., the general liberal political environment may take several years to influence Medicaid policy and expenditures, and the average state income may dwindle for several years before the Medicaid recipient population increases. Differentiating lag periods for independent variables allows policy makers to predict when changes in independent variables will have their greatest affect on Medicaid spending and plan accordingly. Some states no longer allow price discrimination in the nursing home market between Medicaid and private-pay residents. Investigation is needed to determine the affects of disallowing cross-subsidization of Medicaid LTC recipients on Medicaid expenditures and utilization patterns (Schlenker, 1991). 161

PAGE 173

Medicaid spending on the elderly exists within a context of larger medical, social, political, legal, and economic systems. Therefore, the scope of the investigation into Medicaid spending on long-term care for the elderly must be broadened to encompass the relevant interactions among these spheres of influence. Setting the problem in a broader context will certainly complicate the analysis but will also hopefully minimize the untoward consequences of well intended, but misguided, policy 162

PAGE 174

1. 2. 3. 4. 5. 6 7. 8. 9. 10. 11. 12. 13. 14. 15 16. APPENDIX A. DATA SOURCES Independent Variables Explanatory Variable Notation Average weekly wage in the nursing home WGNH and personal care industry SIC 8051, 8052, and 8059 85 or over per 1 000 elderly population OLD 65 or over per 1,000 persons 18 or over EL_DVT Average income of taxpaying population AVINC Effective cost of LTC to the state taxpayers COST Annual state unemployment rate UEMP Average P9Pulation per household HOUS Average annual temperature TEMP Percent elderly population in SMSA SMSA Elderly in Medicaid's medically needy EMN program Number of optional Medicaid benefits for MOB over 21 aged population We i ghted average Medicaid per diem RATE nursing home rate Level of administration of Medicaid MAD program state versus local Number of nursing home beds in state per BEDS 1,000 elderly Average length of stay in an acute-care LOS hospital for state Medicaid population Liberal state political climate LQ 163 Data Source Bureau of Labor Statistics, "Employment & Wages, Annual Averages" U.S. Bureau of Census Census of Population U S Bureau of Census Census of Population Internal Revenue Service Calculated; see Methodology Bureau of Labor Statistics U.S. Bureau of the Census Household Division National Climatic Data Center, North Carolina U.S. Bureau of the Census Medicaid State Data Tables, Table 6 HCFA, Office of Intergovernmentaf Affairs Calculated from Medicaid State Data Tables ; Per diem rates for SNF and ICF; (Harrington 1991) HCFA, Office of Medicaid Management Harrington, Preston, Grant, & Swan; "Trends In Nursing Home Bed Capacity In The States"; 1990 HCFA, Office of Medicare Management Calculated from ADA scores for Congress

PAGE 175

Variable 1. Total Medicaid spending on LTC for the elderly 2. to 5. Nursing home Medicaid spending on LTC for the elderly 6. to 8. Home care Medicaid spending on LTC for the elderly 9. Medicaid waiver spending on LTC for the elderly 10. Private state spending on nursing home care "Data Collection" DATA SOURCES (Cont.) Dependent Variables Measure a. Aggregate expenditure per 1,000 elderly population a. Aggregate expenditure per 1,000 elderly population b. #recipients per 1,000 elderly c. average expenditure per recipient d. Medicaid spending on nursing home care as a percent of total LTC spending a. Aggregate expenditure per 1,000 elderly population b. #recipients per 1,000 elderly c. average expenditure per recipient a. aggregate expenditure per 1,000 elderly a. aggregate expenditure per 1,000 elderly 164 Data Source Computed from nursing home, home care, and 2176 data HCFA's Division of Medicaid Statistics, Compiled from State Medicaid Form 2082 HCFA's Division of Medicaid Statistics HCFA's Medicaid Bureau, Office of Research & Demonstrations, Annual State Waiver Cost Reports See Methodology, "Data Collection"

PAGE 176

BffiLIOGRAPHY Advance. (1995, October 23). Medicaid Bill Imperils Elderly. p 4. Barrilleaux, C & Miller, M. (1988). The Political Economy of State Medicaid Policy. American Political Science Review, 82(4), pp. 1089-1107. Bishop, C. (1984). Living Arrangement Choices of Elderly Singles: Effects of Income and Disability. Health Care Financing Review, 1(3) pp Bishop, C. (1988). Competition in the Market for Nursing Home Care. Journal of Health Politics. Policy. and Law, 13 (2), pp. 341-361. Blumberg, L., Holahan, J. & Moon, M. (1993). Options for Reforming the Medicaid Matching Formula. The Urban Institute-Final Report. Buchanan, R. (1984). Medicaid Utilization of Long-Term Care: The Skilled and Intermediate Case Mix. The Journal of Long Term Care Administration. Buchanan, R. (1987). Medicaid Cost Containment. Associated University Press. Buchanan, R., Cappelleri, J. & Ohsfeldt, R. (1991). The Social Environment and Medicaid Expenditures: Factors Influencing the Level of State Medicaid Spending. Public Administration Review, 51(1), pp. 67-73. Burner, S., Waldo, D. & McKirsich, D. (1992). National Health Care Expenditures Through 2030. Health Care Finance Review, 14(1) p. 1-29. Caplan, A. (1988). Is Medical Care the Right Prescription for Chronic Care? In Sean Sullivan & Marion Ein Lewin (Eds.), The Economics of Long-Term Care and Disability. American Enterprise Institute for Public Policy Research. Washington D.C., pp. 73-89. Carrell, D. (1990). The Changing political Economy of Long-Term Care: The Politics of Reassessment in an Era Of Fiscal Stress. 1977-1986 Dissertation, University of Washington. Carroll, N. & Erwin, W. (1987). Patient Shifting as a Response To Medicare Prospective Payment. Medical Care, 25(12), pp. 1161-1167. Chang, D. & Holahan, J. (1990). Medicaid Spending in the 1980s: The Access-Cost Containment Trade-Off Revisited. The Urban Institute Report, p. 90-02. Davis, K. & Rowland, D. (1986). Medicare Policy: New Directions for Health And Long-Term Care. Baltimore, Maryland: The Johns Hopkins University Press. pp. 165

PAGE 177

Davis, M. & Freeman, J. (1994). Excess Demand and Cost Relationships Among Kentucky Nursing Homes. Health Care Financing Review, 15(4). Dubay, L. (1991). Use of Medicare's Skilled Nursing Benefit in Urban and Rural Areas: Is It Different? The Urban Institute, p. 3971-4007. Evashwick, C. (1988). The Continuum of Long-Term Care. InS. Williams & P. Torrens, (Eds.), third edition. Introduction To Health Services, Wiley Series in Health Services. John Wiley and Sons, pp. 212-254. Feder, J. & Scanlon, W. (1981). Medicare & Medicaid Patients Access to Skilled Nursing Facilities. Health Care Financing Review. Feldstein, P. (1988a). Health Care Economics. Wiley Medical . Feldstein, P. (1988b). The Politics of Health Care Legislation: An Economic Perspective. Health Administration Press Perspectives. Godwin, R. & Shepard, W. (1976). Political Processes and Public Expenditures: AReexamination Based on Theories of Representative Government. American Political Science Review; 70, pp. 1127-1135. Grannemann, T. (1979). The Demand for Publicly Financed Medical Care: The Role of Interdependent Preferences. Center for Health Services and Policy Research, Northwestern University. Harlow, K. & Wilson, L. (1985). DRGs and the Community-Based Long-Term Care System. The Southwest Long-Term Care Gerontology Center, Dallas, Texas, p 5. Harrington, C. & Swan, J. (1987). The Impact of State Medicaid Nursing Home Policies on Utilization and Expenditures. Irumi!y, 24, pp. 157-172. Harrington, C., Newcomer, R. & Estes, C. (1985). Long-Term Care of the Elderly: Public Policy Issues. Sage Library of Social Research. Harrington, C., Cassel, C., Estes, C., Woolhandler, S, & Himmelstein, D. (1991). A National Long-Term Care Program for the United States. Journal of the American Medical Association, 226(21), pp. 3023-3029. Harrington, C., Preston, S. Grant, L., & Swan, J. (1991). Trends in Nursing Home Bed Capacity in the States. Working paper, pp. 1-15. Harrington, C. & Swan, J. (1994). Medicaid Nursing Home Reimbursement Policies, Rates, and Expenditures. Health Care Financing Review, Q., 1. Harrington, C., Swan, J. & Grant, L. (1988). Nursing Home Bed Capacity in the United States, 1978-1986. Health Care Financing Review, .2(4), pp. 81-97. 166

PAGE 178

Haws, C., Kane, R., Powers, L. & Reinardy, J. (1988). The Case for A Continuum of Long Term Care Services: Lessons From The Community Based Care Demonstrations. The Public Policy Institute. American Association of Retired Persons, pp. 7-21. Health Care Investment Analysts, Inc. & Arthur Andersen. (1990). The Guide To The Nursing Home Industry. Holahan, J. & Cohen, J. (1987). Nursing Home Reimbursement: Implications for Cost Containment, Access, and Quality. The Milbank Quarterly 65(1), pp. 112-147. Howell, E., Baugh, D. & Pine, P. (1988). Patterns of Medicaid Utilization and Expenditures in Selected States: 1980-84. Health Care Financing Review, Winter. Jazwiecki, T. (1989). Long-Term Care for the Elderly in the United States. In Teresa Schwab (Ed.), Caring for an Aging World: International Models for Long-Term Care. Financing. and Delivery. McGraw-Hill Book Company, pp. 288-338. Jazwiecki, T. (1988). How States Pay for Long-Term Care Facility Services Under Medicaid in LongTerm Care, Healthcare Financial Management Association. Kane, R. (1989). Testimony before The United States Bipartisan Commission on Comprehensive Health Care. Keeler, E. & Kane, R. (1982). What Is Special About Long-Term Care? In R. Kane & R. Kane (Eds.), Values and Long-Term Care, D.C. Heath & Company, pp. 85-99. Kemper, P., Applebaum, R. Robert & Harrigan, M. (1987). Community Care Demonstrations: What Have We Learned? Health Care Financing Review, .8_(4), pp. 87-100. Kenney, G. & HoJahan, J. (1990). The Nursing Home Market and Hospital Discharge Delays._ Inquiry, 271, pp. 73-85. Knickman, J. & Foltz, A. (1984). Regional Differences in Hospital Utilization. How Much Can Be Traced To Population Differences? Medical Care, 22(11), pp. 971-86. Knickman, J. & Foltz, A. (1985). A Statistical Analysis of Reasons for East-West Differences in Hospital Use. Inquiry, 22(1), pp. 45-58. Knox, R. (1995, November 9) Congress' Plan to Curtail Medicaid Could Swell Ranks of Uninsured. The Boston Globe, p. 13. Liebig, P. (1988). The Issue of High Technology for Health Care at Home: Issues and Implications Medical Instrumentation, 22(5). Lindblom, C. (1959) The Science of Muddling Through. Public Administration Review, 19, pp. 79-88. 167

PAGE 179

Liu, K., Coughlin T. & McBride, T. (1989). Nursing Home Utilization: Risks and Length of Stay. The Urban Institute, pp. 3876-3902. Luehrs, J. (1983). State Medicaid Budget Process. The Intergovernmental Health Policy Project, pp. 1-14. Lurie, I. & Wyckoff, J. (1989). Intergovernmental Influences on Medicaid Program Expenditures: Comments and Additional Evidence: Public Administration Review, pp. 402-404. Macken, C. (1986). A Profile of Functionally Impaired Elderly Persons Living in the Community Health Care Financing Review, 1(4). Manton, K. & Soldo, B. (1985). Dynamics of Health Changes in The Oldest Old: New Perspectives and Evidence, Milbank Memorial Fund Quarterly, 63. Marmor, T. (1983). Political Analysis and American Medical Care. Cambridge University Press. McDevitt, R. & Buczko W. (1985). Medicaid Program Characteristics and their Consequences for Program Spending. Health Care Financing Review, 1(2). Meltzer, J. (1988). Financing Long-Term Care: A Major Obstacle to Reform. InS. Sullivan & M. Ein Lewin The Economic and Ethics of LongTerm Care and Disability. American Enterprise Institute for Public Policy Research. Washington D.C. pp. 56-72. Miller, N. (1990). Medicaid 2176 Home and Community-Based Waivers: Program Trends, Research Directions Health Care Financing Administration, Office of Research and Demonstrations. Miller, N. & Saunders W. (1993). Overview. Health Care Finance Review, 14(4). Morrisey, M., Sloan F. & Valvona J. (1988). Medicare Prospective Payment and Post Hospital Transfers to Subacute Care. Medical Care, 26(7), pp. 685-698. Nyman, J. (1985). Prospective and Cost-Plus Medicaid Reimbursement, Excess Medicaid Demand, and the Quality of Nursing Home Care. Journal of Health Economics, 1;, pp. 237-259. Orr, L. (1976). Income Transfers as a Public Good: An Application to AFDC. American Economic Review, .Q. Paringer, L. (1983). Economic Incentives in the Provision of Long Term Care. In J. Meyer, (Ed.), Reform in Health Care: Current Issues. New Directions. and Strategic Decisions, American Enterprise Institute. 168

PAGE 180

Parsons, T. (1981). Definitions of Health and Illness in the Light of American Values and Social Structures. In Arthur Caplan et al., (Eds.), Concepts Of Health And Disease, Reading MA: Addison-Wesley, pp. 545-560. Pierce, R. & Hansen, K. (1987). Long-Term Care for the Elderly: A Legislator's Guide. National Conference of State Legislators, Washington D.C. PT Bulletin. (1995, September 25). Capital News Individual States Will Decide Health Care for Poor With Disabilities, p. 14, 15. PT Bulletin. (1995, November 10). Capital News Experts Say Medicaid Block Grants Could Deny Care to Millions of People, p. 5,13. Rantz, M. (1991). Inadequate Reimbursement for Long-Term Care: The Impact Since DRG's. Nursing and Health Care, .11, pp. 470-472. Reutzel, T. (1984). State Medicaid Program Policy Decisions & Medical Service Expenditures for Elderly Cash Grant Eligibles: An Empirical Analysis. A Thesis in Community Planning and Development, The Pennsylvania State University, 1984. Revicki, D. (1989). Health-Related Quality of Life in the Evaluation of Medical Therapy for Chronic Illness. The Journal of Family Practice, 29(4). Richardson, H. (1990). Long Term Care. In Health Care Delivery in the United States, New York: Spring Publishing. Rivlin, A. & Weiner J. (1988). Caring for The Elderly: Who Will Pay?. Brooking Institute. Rohrer, J. (1987). Access to Nursing Home Care: A Need-Based Approach. Medical Care, 25(8), pp. 796-800. Rosenbloom, D. (1988). Public Administration. Random House, pp. 338-340. Samuelson, R. (1995, September 18). Getting Serious: At last, Americans are really confronting costly entitlement for the elderly. As the first skirmish in the generational war begins, young and old must get ready to get less. Newsweek, pp. 40-44. Scanlon, W. (1980). Nursing Home Utilization Patterns: Implications for Policy. Journal of Health Politics. Policy. and Law, 1:(4). Scanlon, W. & Feder, J. (1988). The Long-Term Care Marketplace: An Overview. In Long Term Care, Healthcare Financial Management Association. Schlenker, R. (1991). Nursing Home Costs, Medicaid Rates, and Profits under Alternative Medicaid Payment Systems. Health Services Research, 26(5), pp. 623-649. 169

PAGE 181

Schneider, S. (1988). Intergovernmental Influences on Medicaid Program Expenditures. Public Administration Review, 7/8, pp. 756-763. Schneider, S. (1989). Intergovernmental Influences on Medicaid: A Rejoiner. Public Administration Review, 7/8, pp. 404-405. Shaughnessy, P. & Kramer, A. (1990). The Increased Needs of Patients in Nursing Homes and Patients Receiving Home Health Care. The New England Journal of Medicine, 322(1), "pp. 21-27. Shearer, G. (1989). Long-Term Care: Analysis of Public Policy Options, Consumers' Union. Soumerai, S., Ross-Degnan, D:. Avom, J., McLaughlin T. & Choodnovskiy, I. (1991). Effects of Medicaid Drug-Payment Limits on Admission to Hospitals and Nursing Homes. The New England Journal of Medicine, 325(10) pp. 1072-1077. Statistical Abstract of the United States. (1988). 108th Edition, U. S. Department of Commerce, Bureau of the Census. Current Population Reports, series P-25, No. 13. Total Population by Age and Sex: 1960 to 1986. Stein, C. (1995, November 18) Nursing Homes Face Cuts in Funds, Group Fights Weld Plan on Medicaid. The Boston Globe, p. 1,9. Stuart, B., Reutzel E. & Reutzel T. (1985). Medicaid Reform: Programming Solutions to the Equity Problem. Journal of Health Politics. Policy. and Law, 10 (1), pp. 93-118. Swan, J. (1990). The Share of Medicaid for Nursing Home Care. Journal of Health and Social Policy, 1(3), pp. 35-53. Swan, J. & Benjamin, A. (1990). Medicare Home Health Utilization as a Function of Nursing Home Market Factors. Health Services Research, 25(3), pp. 479-500. Swan, J., de Ia Torre, A. & Steinhart, R. (1990). Ripple Effects of PPS on Nursing Homes: Swimming or Drowning in the Funding Stream? The Gerontologist, 30(3), pp. 323-331. Swan, J. & Harrington, C. (1990). Certificate of Need and Nursing Home Bed Capacity. Journal of Health and Social Policy, 6(2), pp. 87-101. Swan, J., Harrington, C. & Grant L. (1988). State Medicaid Reimbursement for Nursing Homes, 1978-86. Health Care Financing Review, .2(3), pp. 35-49. Swan, J., Harrington, C., Grant, L., Luehrs, J. & Preston S. (1991). Recent Trends in Medicaid Nursing Home Reimbursement. Working Paper. Toner, R. (1995, November 12). Critics Say Republican Budget Will Create Shortage of Nursing Home Beds for Elderly. The New York Times, p. 30 170

PAGE 182

US Government, General Accounting Office. (1988). Long-Term Care for the Elderly: Issues of Need, Access. and Cost. US Government, GeneralAccounting Office. (1987). Medicaid: Interstate Variations in Benefits and Expenditures. US Government, General Accounting Office. (1987). Medicare and Medicaid: Stronger Enforcement of Nursing Home Requirements Needed. US Government, General Accounting Office. (1987). Medicare and Medicaid: Effects of Recent Legislation on Program and Beneficiary Cost. Weissert, W. & Scanlon, W. (1983). Estimating the Long-Term Care Population: National Prevalence Rates and Selected Characteristics. Project to Analyze Existing Long Term Care Data, Final Report, ll, U.S. Department of Health and Human Services. Zedlewski, J., Raffaty S. & McBride, T. (1991). The Changing Profile of the Elderly: Effects on Future LTC Needs and Financing. The Urban Institute, 2. Zola, I. (1988). Policies and Programs Concerning Aging and Disability: Toward a Unifying Agenda. In S. Sullivan & M. Bin Lewin, The Economics and Ethics of Long-Term Care and Disability. Washington, D.C.: American Enterprise Institute for Public Research, pp. 90-130. Zuckerman, J. (1995, November 18). Bid to Balance the Budget in 7 Years Clears House, Senate The Boston Globe, p. 1,9. 171