Citation
The effect of government-sponsored health care programs upon Colorado hospital profitability

Material Information

Title:
The effect of government-sponsored health care programs upon Colorado hospital profitability
Creator:
Shroyer, A. Laurie W
Publication Date:
Language:
English
Physical Description:
xi, 122 leaves : illustrations, forms ; 28 cm

Subjects

Subjects / Keywords:
Hospitals -- Finance -- Colorado ( lcsh )
Hospitalization insurance -- Colorado ( lcsh )
Hospitalization insurance ( fast )
Hospitals -- Finance ( fast )
Colorado ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references.
General Note:
Submitted in partial fulfillment of the requirements for the degree, Doctor of Philosophy, Graduate School of Public Affairs.
General Note:
School of Public Affairs
Statement of Responsibility:
by A. Laurie W. Shroyer.

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:
25739159 ( OCLC )
ocm25739159

Full Text
THE EFFECT OF GOVERNMENT-SPONSORED HEALTH CARE PROGRAMS
I
| UPON COLORADO HOSPITAL PROFITABILITY
I '
i by
j A. Laurie W. Shroyer
I B.A., University of Colorado, 1980
I
I University of Colorado, 1984
i
I
i
i
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
I
j of the requirements for the degree of
1 Doctor of Philosophy
Graduate School of Public Affairs
1991


1991 by A. Laurie W. Shroyer
All rights reserved.


This thesis for the Doctor of Philosophy
degree by
A. Laurie W. Shroyer
has been approved for the
Graduate School of
Public Affairs
by
Patricia A. Butler
Linda M. deLeon
Richard W. Foster
L[
Date


ABSTRACT
Shroyer, A. Laurie W. (Ph.D., Public Administration)
i
The Effect of Government-Sponsored Health Care Programs
Upon Colorado Hospital Profitability
i
Thesis directed by Professor Peter deLeon
I 1
I
i ,
Government-sponsored health care programs represent a large and
growing portion of the hospital industry's revenues. According to
!
the literature, in 1960 24.7% of all health care expenditures were
paid by government programs. In 1988, government expenditures had
increased toj42.1% correspondingly. Government programs represented
$115.2 billion (54%) of hospital expenditures in 1988. A policy
concern exists that government programs should not adversely impact
hospitals' profitability for care rendered to government recipients.
The purpose of this dissertation is to investigate whether government
programs (Medicare, Medicaid, and the Colorado Indigent Care
Programs) have a differential impact upon Colorado hospitals' patient
service profitability.
To test this relationship, an explanatory model of hospital
profitability, was developed. A data base of 3,412 Colorado hospital
monthly financial records, for the period January 1986 to August
1990, was used for this study. The hospital data base was linked to
environmental!, structural, and government program data. A multiple
i
regression analysis with a simulated repeated measures design was
used to test the primary hypothesis.
This study found that hospital management variables (including
occupancy rate, charges per discharge, and average length of stay)
appear primary in importance in predicting hospital patient service


profitability. Government-sponsored health care program utilization
(payor mix')
was not found to be associated with profitability. The
only government program variable found to be statistically
i
significant in relationship to hospital profitability was the
I
Medicare payment rate. Other government program variables (e.g.,
government program payor mix, Medicaid and Medically Indigent
Programs' payment rate per discharge, disproportionate share status,
i
and government program payment methodology) were inferred to be
i
indirectly associated with profitability, based upon the associations
|
establishejd between government program variables and secondary
hypotheses1 tested.
!,
In general, this study did not find that government programs
l
reduced hojspital patient service profitability. The results indicate
that hospital management appears to have significant control over
patient sejrvice profitability. To maximize profitability, hospital
i . .
administration should focus on increasing occupancy levels, limiting
l
levels of jnon-emergent charity care provided, reducing bad debt
!
allowances by increasing collection activities, increasing non-
i.:
patient cajre service product lines with potential high return on
investment, and decreasing average length of stay.
The form hiid content of this abstract
roved. I recommend its
publication.
Signed
Peter deLeon
i
i
I
I
I
I
I
V
I


DEDICATION
Many individuals participated to make this endeavor successful.
A special thanks to:
|
i
l|ly committee members, for their extraordinary efforts;
' Peter deLeon, chairperson;
| Bill Braithwaite;
! Pat Butler;
'' Linda deLeon;
1 Rich Foster; and
Tom Granneman;
Bob Weaver, for his PC hardware and software expertise;
i
Kevin Reed, for his Revelation data base expertise;
Larry Mac Neill, for his SAS assistance;
i
Dennis Lazotte, for his statistical expertise;
David West, for his patience and support;
Jim Suver, for his wisdom and guidance; and
Jim Hart, for granting access to Colorado hospital data.
I
]
This thesis is dedicated with love and gratitude to my family-
especially to,my Mom, my Dad, Ken, and Robby.


CONTENTS
i
Figures
i
Tables j
i
CHAPTER |
1. INTRODUCTION . .
Purpose ........
Scope of the Study
1
1
3
Organization of the Thesis ........................... 4
i
2
REVIEW OF THE LITERATURE
5
The American Hospital Industry ......................
National Health Care Expenditure Trends .............
The Role of Government in the Health Care Market
j
Government's Role in the Provision of Hospital
j, Services to the Elderly and the Poor .............
Government Financing of Health Care Services . .
! Medicare: The Initiation of the
i Prospective Payment System .................
i
I
I Medicaid: A Changing Reimbursement
| Environment ................................
i
| The Colorado Indigent Care Programs:
1 An Allocation of State General Funds
I to Providers ..............................
5
6
7
8
10
11
15
16
The Colorado Health Policy Environment ............... 17
i
Colorado Hospital Financial Trends ................... 19
3. THEORETICAL FRAMEWORK ................................... 20
i
Profitability ....................................... 20
!
Background......................................20


I Defining an Adequate Profit Level ..............
| Special Considerations for Hospitals . .
i
Impact of the Government-Sponsored Programs
Upon Hospitals .......................................
|
Medicare Prospective Payment System ............
j Colorado's Medicaid Payment System .............
] Colorado's Indigent Care Programs ..............
j
Hospital Cost Behavior ...............................
Hospital Provision of Uncompensated Care .............
i
1 Background .....................................
I
| Recognized Need for Financing:
Disproportionate Share Payments .............
! Distribution of Uncompensated Care ...........
Factors Affecting Hospital Closure ...................
Financially Distressed Hospitals .....................
Proposed Model of Hospital Profitability .............
|
4. METHODOLOGY AND DATA SOURCES .............................
I
I
Primary Hypothesis ...................................
I
Design ...............................................
i
Profitability Function ...............................
! 1
Secondary Hypotheses .................................
primary Hypothesis: Dependent Variable ...............
!
Primary Hypothesis:
! Independent Variables under Study ..................
i
i
Secondary Hypotheses:
I Dependent and Independent Variables under Study
I '
Regression Analysis ..................................
i
Primary Data Source ..................................
i
i
Life Cycle Changes in Hospital Facilities ............
ojata Screening.......................................
22
23
24
24
26
27
28
29
29
31
32
32
34
34
37
37
37
38
38
39
40
41
44
46
47
48
viii


5
6
RESULTS AND DISCUSSION .............................. 52
j
Characteristics of Study Hospitals .................. 52
Primary Hypothesis
Secondary Hypotheses
53
55
Discussion ...........................................
I
I
! Influences on H,(2): Revenues Per Discharge,
j, H(3): Salary Per Hour, and
i H(4): Uncompensated Care Provided . .
!
I
i
CONCLUSION ...........................................
i 1
i
Summary ..............................................
i
Revised Model of Hospital Profitability ..............
60
63
75
75
77
Management Implications of the Results ............. 81
policy Implications of the Results ................. 84
Directing Future Colorado Health Policy
! Initiatives ............................ ..... 88
APPENDIX
I
1. ISSUES FOR FURTHER RESEARCH ............................ 103
I '
2 SENSITIVITY ANALYSIS: REPEATED MEASURES DESIGN . 105
i
3. SENSITIVITY ANALYSIS: COMBINED UTILIZATION
j OF ALL GOVERNMENT PROGRAMS........................108
4. AN EXAMINATION OF HOSPITAL CHARACTERISTICS
j FOR POLICY PURPOSES .....................
!
5. CHA DATA BANK MONTHLY REPORT ...........
I
6. GLOSSARY ...................................
i
I
REFERENCES J ..........................................
110
112
114
117


FIGURES
Figure
i
3.1 Proposed model of hospital profitability
I
6.1 Revised model of hospital profitability
i
I


I
I
I
I
! TABLES
Table
2.1
3.1
3.2
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
A. 1
A.2
A. 3
A.4
A. 5
Hospital Specific Base Rate Phase-in Schedule .............. 14
Longitudinal Comparison of Profit Margins .................. 22
i
Cross-Sectional Comparison of Similar Industries .... 23
H(l): |Univariate and Multiple Linear
Regression Analysis Results ............................ 54
i
H(2): I Univariate and Multiple Linear
Regression Analysis Results . . ..................... 56
i
H(3): jUnivariate and Multiple Linear
Regression Analysis Results ............................ 58
H(4): [univariate and Multiple Linear
Regression Analysis Results ............................ 60
Names, Definitions, and Descriptive Statistics for the
Dependent Variables Used in the Analysis................67
Names, [Definitions, and Descriptive Statistics for the
Independent Variables Used in the Analysis ............. 68
Cross-Sectional Analysis of Dependent
Variables of Study ..................
71
Longitudinal Analysis of Dependent Variables of Study .
i
Comparison of Equation Correlation Results (R-square)
Between Multiple Regression Models Both With and
Without Hospital Identification Dummy Variables . .
Comparison of Equation Correlation Results
(Standardized Estimates) When Independent Variables
(NOT. Statistically Significant) Are Deleted
from' the Model......................................
73
105
106
Compari|son of Equation Correlation Results (R-square)
Betw'een Multiple Regression Models.....................108
Key Hospital Performance Indicators by Location
and Bed Size........................................... 110
Key Hospital Performance Indicators by Ownership
and Bed Size.............................................Ill
i
i
I


CHAPTER 1
INTRODUCTION
I care about the future because that is
where I hope to spend the rest of my life.
Mark Twain, circa 1900.
Purpose
The financial condition of the Colorado hospital industry is
i
deteriorating! During 1989, six hospitals in Colorado closed
("Hospitals' jpata Watch," 1990). The Colorado Hospital Association
reported recently that Colorado hospitals' profit margins fell from
i
3.5% in 1988jto 3.2% in 1989 (Colorado Hospital Association, 1990b).
i
According to !the American Hospital Association (AHA) reports,
i
hospital profits nationwide declined from 6.3% in 1984 to 5.1% in
i
1989 (Burda, |1990). The worsening financial condition of the
i
Colorado hospital industry, reflecting to the national industry
I
trends, constitutes a statewide policy concern.
During jthe period from 1984 to 1990, the Colorado hospital
industry experienced dramatic changes in methods used by government-
sponsored programs to pay for health care services. The revisions to
the historicajl hospital cost-based reimbursement system were
initiated by jthe Medicare Program in 1984. As government-sponsored
health care programs have strengthened cost containment initiatives
I
in an attemptj to moderate the growth in hospital expenditures, the
Colorado Hospital Association asserts that profit margins at Colorado
hospitals havb dropped precipitously. The future financial viability
i
of the Colorado hospital industry is threatened. Thus, an
understanding! of the effect of government-sponsored health care


I
program reimbursement upon hospital profitability is of critical
importance. 1
i
The purpose of this study, therefore, is to build an
i
explanatory model of hospital profitability. This research is
l
exploratory,iwhere the hospital profitability model tested is based
I
on a comprehensive review of the literature to determine factors
which may predict hospital profitability. The factors explored fall
I
into four categories: environmental, structural, management, and
government program variables. Thus, the impact of government-
sponsored health care programs upon hospital profitability can be
I
identified (holding all other factors constant).
This dissertation is both unique and important for the
following reasons:
1. A conclusive analytical model of hospital profitability
tihat addresses government program reimbursement does not
!
currently exist.
2. The effect of government-sponsored health care programs
upon hospital profitability is not well understood.
I
3. A inter-program comparison of different government-
sponsored health care programs' (federal, federal/state,
and state) effect upon hospital profitability has not
i-
been conducted.
i
This is a study of Colorado hospital profitability for the purposes
i
of directing Icolorado health policy.
A recent study by Rizzo (1990), based on national hospital
data, indicates that characteristics of financially distressed
hospitals chajnged dramatically from early to middle 1980s. Prior to
the introductjion of Medicare's prospective payment system in 1984,
!
distressed hospitals had slightly higher expenses (prior to
I
adjustment) than comparable facilities. Rizzo found that
i
2


I
i
I
I
i
financially distressed hospitals after 1984 had critically poor
revenue positions and served a disproportionately higher proportion
of low income patients. This study will similarly attempt to examine
hospital revenue, cost, and disproportionate share variables from
Colorado hospital financial data to assess inherent differences that
must be accounted for in developing state health policy initiatives.
i
i Scope of the Study
This paper examines the relationship between hospital
profitability and a variety of factors believed to influence hospital
financial behavior including: environmental, structural, management,
and government program variables. More concretely, the focus of this
I
i
study is to determine the effect government-sponsored health care
programs (Medicare, Medicaid, and the Colorado Indigent Care Program)
have on Colorado hospital profitability. The Colorado Hospital
Association (CHA) Data Bank Monthly Financial Reports, a unique
l
proprietary data set of Colorado hospital financial data from January
1986 to August 1990 (56 months), makes such an effort possible. The
i
I
CHA Data Bank hospital financial data are matched to national
i
i
inflation indices, state demographic data, data commission reports,
hospital licejnsure and survey data, graduate medical education data,
i
and government program hospital-specific detailed reimbursement data,
j
It is the combination of the factors analyzed, in an attempt to build
an explanatory model of hospital profitability, that makes this study
unusual. j
i
The study population includes 65 Colorado general acute care
hospitals. Aj multiple regression analysis, using a simulated
repeated measures design holding hospital facility constant across
repeated measures, is used. This study expands on previous work by
comparing similarities and differences between government-sponsored
! 3
i


health care programs. In addition, this study attempts to determine
i"
the extent to which hospital profitability can be controlled by
I
hospital management and the degree to which hospital profitability is
influenced by external factors, such as government program
reimbursement.
I Organization of the Thesis
i'
i
This paper is organized in six major sections. The first
!
i
chapter, thejintroduction, defines the purpose and scope of this
study. A broad literature review is included in the second chapter,
to provide a I historical perspective of the hospital industry and the
i
i
role of government-sponsored health care programs in the provision of
health care services. The third chapter discusses previous work in
i
this research1 area. Based upon findings of previous studies, a
theoretical model of hospital profitability is developed. A
methodology section, chapter 4, describes the data sources and
statistical methods used to estimate the relationships between
hospital profitability and the four types of independent variables:
environmental!, structural, management, and government program
!
variables. ijhe detailed results of the analysis are presented in
chapter 5, contrasting the relative importance of the independent
i
variables in [predicting hospital profitability. The concluding
section, chapiter 6, summarizes the results, discusses the study's
findings in context of the literature reviewed, and sets forth the
state health policy options, which may be used to address concerns
r
regarding the) Colorado hospital industry's future financial
viability.
i
i
! i
i
I
i
I
i 4
i


CHAPTER 2
REVIEW OF THE LITERATURE
The American Hospital Industry
Hospitals are the core of the American health care delivery
i
system. Hospitals currently face many pressures: rising public
i
expectations ] for expanded service, increased capital requirements to
I
provide state-of-the-art technology, serious personnel shortages,
I
intensified public scrutiny, contihued growth of government
|
initiatives tip contain health care expenditures, increased
i
involvement of medical staff in hospital management activities,
diminished access to capital markets, burgeoning uncompensated care
demands, and Idecreased profitability of patient care services.
The demand for hospital services will continue to rise,
especially as the proportion of elderly in the general population is
projected to
increase. The elderly simply require greater amounts of
health and hpspital services. In 1989, the portion of the American
population oV|6r age 65 was approximately 12%. By 2050, current
i
population projections indicate that 23% of all Americans will be
over age 65 (jBurglass, 1989; Exter, 1990). Colorado vital statistics
data (Colorado Department of Health, 1987) indicates that the
i
population over age 65 years increased from 8.6% in 1980 to 9.1% m
1987. Thus, !the portion of elderly in Colorado appears to be
i
i
growing. !
I
The aging of the Colorado population contribute to a lower
birth rate and to increased mortality and morbidity associated with
!
higher rates of chronic and degenerative conditions (e.g., chronic
i
obstructive pulmonary disease). The health care needs of the elderly


I'
will require!more intensive, longer term treatment protocols for
i
I
ongoing medical conditions, such as diabetes, hypertension,
arthritis, and heart disease (Ostroff, 1989). Given continued
technological advancements in medicine, hospitals will continue to be
the focal point in the health care delivery system. The hospital
i
environment is unique, as extensive capital equipment and highly
]
specialized staff in a multi-disciplinary team can be effectively
organized tojoptimize patient care outcomes. From the hospital's
i
centralized network of services, complementary arrangements for
medical care!can be coordinated. In summary, the hospital industry
I
is under dynamic, conflicting, and multi-dimensional pressures. As
the center of the American health care delivery system, the future of
the hospital|industry is seemingly vulnerable and therefore, an issue
of public importance.
i
I
! National Health Care Expenditure Trends
Since the 1960s, the percentage of the gross national product
I
(GNP) attributable to health care expenditures has continued to grow.
I;
In 1965, health care expenditures represented 5.9% of the GNP. In
i.
1988, health jspending amounted to 11.1% of the GNP, up from 10.8% in
1987 and more' than twice the percentage for health spending in 1960
(U.S. Department of Commerce, 1991)
Hospital care expenditures in
1960 were $9.j3 billion (34.3% of all national health care
!
expenditures);; in 1988, hospital care represented $211.8 billion
j
(39.2% correspondingly) (Levit, Freeland, and Waldo, 1990). For
i
1991, hospital expenditures are projected to exceed $285 billion (43%
of all national health care expenditures) (U.S. Department of
Commerce, 1991). Richard Darman, President Bush's budget director,
warns by the year 2030 health spending will reach 37% of the gross
i
6


national product (Pear, 1991). Thus, hospital services represent the
single largest national health care expenditure and continue to grow.
In I960, government health expenditures (federal, state, and
local) were $6.7 billion (24.7% of all national health care
expenditures). In 1988, government spent $227.5 billion for health
care (42.1% correspondingly) (Levit, Freeland, and Waldo, 1990). The
role of government in reimbursing for all health care services has
increased substantially (170%) since 1960.
The majority of government expenditures for personal health
care services was spent upon hospital services. Government
! -
(Medicaid, Medicare, and other state and local government programs)
i
spent $115.2Jbillion on hospital care in 1988 (54% of hospital
expenditures) (U.S. Department of Commerce, 1991). Thus, government-
I
sponsored health programs have a significant and a growing direct
effect upon hospital provider's financial viability.
i
The Role of Government in the Health Care Market
I
The role of government in the health care market is multi-
dimensional. Keynesian economic theory suggests that a market
system's equilibrium can be stimulated and/or stabilized by direct
government intervention (Heilbroner and Thurow, 1982). In American
society, government directly intercedes in the health care market to
I
ensure: (1) ^minimum benefits for recipients (e.g., no cost prenatal
care services), (2) minimum quality standards (e.g., licensure
requirements), (3) access to essential services (e.g., emergency room
services for |all citizens, and services for emergent/urgent medical
conditions regardless of the patient's ability to pay for the
emergency care), and (4) a system of equity and justice (e.g., anti-
dumping laws)| : (Feldstein, 1979; Jacobs, 1987).
7


Health, care in America is viewed as a social commodity having
i
significant external benefits to society at large. Medical care
information asymmetries (where consumers receive incomplete or
inaccurate information regarding health care conditions and their
I
corresponding, medical treatment alternatives) are dissolving by
special governmental educational efforts (e.g., AIDS Education and
Training Act). Government is a both a significant supplier and
purchaser of hospital care. In short, the role of government in the
health care market is extensive and significant.
Many conclude that a pure competitive economic model (Adam
Smith's "invisible hand") applied to the health care system would not
I
adequately address social values other than efficiency (Feldstein,
1979; Jacobs,. 1987) Government is ultimately responsible for issues
of efficiency and effectiveness, in addition to addressing non-
economic issues such as equity.
Government's Role in the Provision of Hospital Services
| to the Elderly and the Poor
The provision of hospital services to the elderly and the poor
has long been,a part of the American culture. Prior to 1935, this
i '
assistance was primarily financed by state and local governments, as
i
well as philanthropy. The 1935 Social Security Act established
federal government payments for the medical care rendered to welfare
patients. As participation in this program was optional for states,
initially the provision of health care services to the poor remained
a minor component of the federal welfare assistance program.
In 1950, amendments to the Social Security Act provided federal
matching funds to meet state payments made to medical care providers
j
(vendor payments). Amendments during the next decade attempted to
encourage statie participation in the program. In 1960, the Kerr-
Mills Act (the forerunner of Medicaid) was enacted to assist the
8


elderly poor. The Kerr-Mills Act was designed as a federal/state
j
program, administered directly by the states, with program
i !
eligibility linked to an income-related standard. Kerr-Mills also
i
allowed patients to "spend down" their income to qualify for medical
eligibility.j
i
The 1965 Social Security Amendments initiated the Medicare
i
i
(Title XVIII) and Medicaid (Title XIX) Programs. The Medicare Part A
i
i
Program (Hospital Insurance for the Aged) was intended to provide
J
hospital reimbursement for services received by Medicare recipients.
The Medicareiprogram established uniform eligibility, benefits, and
|
reimbursement criteria at the national level. The Medicaid program
was appendedJlate in the legislative process, a "sleeper" that did
I
l i
not provoke significant debate, as a mechanism to distribute more
evenly the health services provided to the poor. The Medicaid
program, in contrast to Medicare's federal focus, was modeled after
I
the Kerr-Mills Act in a federal/state program design.
The purpose of the 1965 Social Security Amendments was to
I
provide equal access to mainstream health services (including
|
hospital care) for the elderly and the poor. These 1965 changes
consolidated (some of the existing state and local government-based
programs for Jthese populations, increased overall federal funding for
these efforts, mandated minimum eligibility and benefit levels, and
defined expanded program options. For the first time, the government
i
became a major funding source for hospital services.
i
The 1972 Social Security Amendments expanded Medicare
eligibility t!o include the disabled population, with specific
i
categorical benefit for end-stage renal disease patients. As a
I
result of thdse Amendments, only the elderly who do not meet the
I
social security income (SSI) qualifications (either 40 quarters
i
minimum participation or SSI pension exemption) are not covered by
9


I
Medicare. Thus, the Medicare program covers most of Colorado's
elderly and disabled.
The Meclicaid program in Colorado covered fewer than 60% of the
state's poorJin 1989 (Center for Health Ethics and Policy, 1989).
The Medicaidjprogram, although expanded in recent years, has remained
!
inadequate in providing a comprehensive health care safety net for
Colorado's poor. For purposes of this policy discussion, it is
i
I
assumed that;Colorado's medically indigent population can be defined
as the uninsured population with incomes under 200% of the federal
poverty level ($26,800 for a family of four in 1991). The size of
this population in Colorado has been roughly estimated to be between
313.000 persons (Center for Health Ethics and Policy, 1989) and
260.000 persons (Butler and Yondorf, 1990).
In a partial attempt to supplement the Medicare and Medicaid
funding for hospital services, Colorado initiated in 1974 Long Bill
i
Appropriation funding for various indigent care programs. These
programs were consolidated and expanded, by statutory authority, into
an integrated indigent care program in 1983. This program is known
;
I
as the "Colorado Indigent Care Programs," which is entirely funded by
I
Colorado stafe general fund revenues.
These three government-sponsored care programsMedicare,
i
Medicaid, and. the Colorado Indigent Care Programsrepresent a
combination of federal, federal/state, and state programs that pay
!
Colorado hospitals for care rendered to their respective clientele.
I
i
Government Financing of Health Care Services
I
For Colorado general acute care hospitals (for the period
i .
January 1986 |to August 1990), approximately 43% of all hospital
i
discharges were reimbursed by government programs. Medicare
represented Approximately 32% of the discharges. Medicaid
10
i


i
I
represented approximately 7% of the discharges. The Medically
j '
Indigent Program represented approximately 4% of the discharges. The
1989 CHA Report indicated that Medicare represented approximately
36.6% of the(gross revenues (billed charges) and Medicaid represented
approximately 7.5% of gross revenues correspondingly.
i
As a point of history, neither the Medicare program nor the
Medicaid program were ever expected to be a large burden to the
federal government. Between the years 1970 to 1982, federal outlays
i
for Medicare;and Medicaid had increased 600%. Average hospital costs
per stay rose from $316 in 1965 to $2,168 in 1981 (586% increase)
(U.S. President, 1983). The costs of inpatient care were thought to
i
be out of control, demanding a new focus on cost containment
i
strategies, i,
I '
Medicare: The Initiation of the Prospective Payment System
Prior tlo 1983, hospital providers were reimbursed primarily on
i
a reasonable(cost basis. This payment methodology meant that
|
Medicare paid hospitals only for reasonable costs incurred in
I
providing services to Medicare beneficiaries.
i
The rules defining "reasonable cost" were developed by the
Health Care Financing Administration (HCFA). These rules were
I
changed frequently prior to 1983. This type of payment system proved
!
to be inflationary, since the more costs that a hospital could
i ,
justify, the jgreater its Medicare reimbursement.
I
In 1982, the Tax Equity and Fiscal Responsibility Act (TEFRA),
PL 97-248, was passed. TEFRA directed the Secretary of the
Department of Health and Human Services (HHS) to develop and report
I
on a "Prospective Payment System" (PPS), which would alter incentives
!
to hospitals by restructuring the Medicare inpatient payment system.
HHS put together a proposal modeled after the New Jersey State
All-Payors syjstem. The concept of a prospective payment system was
11


not new. Several experimental model systems had been developed since
the early 1970s. The New Jersey system was just being implemented
I
statewide atjthe time of the HHS proposal (U.S. Congress, Senate,
Committee on!Finance, 1983).
Under the HHS proposal, Medicare payments for hospital
inpatient operating costs would be determined in advance. Payments
j
would be made based on a fixed amount per case type, identified by
the diagnosis1 related grouping (DRG) into which the case was
classified, jInitially, separate payment amounts were established for
467 DRGs. |
!
The DRG assignment was based on results of clinical
information, j Several factors would determine the DRG, including:
(1) principal'diagnosis (reason for admission), (2) whether surgery
|
was performed, (3) type of surgery performed, (4) presence of
complicating|surgical conditions, and (5) age of patient (U.S.
i
Congress, Senate, Committee on Finance, Committee on Health, 1983).
The HHS administrative proposal (December 31, 1982) delineated
j
the following features of the initial PPS:
1. Base Rate: All hospitals would be paid the same amount
I
i
for treating the same diagnosis, with adjustments made
i
for regional wage differences and teaching expenses. The
!
HHS Secretary would be required to determine a national
standard rate per discharge for each DRG. The rate,
derived from historical Medicare cost/charge data, would
I
be the product of: (1) the standard cost level per
discharge, and (2) the appropriate weighting factor for
i
each DRG. The national average rate per discharge would
i
be adjusted by region using established hospital wage
i
indices. Teaching adjustments would be included, based
|
on the hospital's ratio of interns and residents to beds.


i ,
I 1
2.
3.
4.
5.
6.
The national rate would be updated annually to reflect
increases in the medical care component of the consumer
i
price index
i
Outliers:
(CPI).
Atypical cases would receive additional
payment, as they may require additional resources to
treat. (Note: The original proposal only established
I
clay outlier payments. Cost outlier payments were
]
developed later.) Outliers are defined as cases
exceeding the average length of stay by 30 days. Outlier
i
payment is a per diem rate of 60% of the average DRG
l
daily rate.
Budget Neutrality: The PPS proposal was designed to be
I
budget neutral, in that the overall budget for Medicare
hospital expenditures is fixed. Amounts paid for outlier
I
days would reduce the DRG payment level across the board.
A proposal constraint was that the Medicare inpatient
l
expenditures would not increase over the current TEFRA
payment system.
j
Exceptions: Certain types of costs are excluded from the
DRG payment system, such as capital equipment costs,
direct medical education costs, and costs for outpatient
I
care. Certain types of facilities are excluded such as
psychiatric hospitals, rehabilitation hospitals, long-
I
term care facilities, and children's hospitals.
I
Sole community hospital providers may request
differential base rate payments.
Time Frame: Implementation was scheduled for hospital
fiscal years starting on or after October 1, 1983.
i
i
13


Many minor modifications were made to the HHS administrative
i
proposal during its evolution into law. These modifications
included: j
1. A phase-in period and change in effective dates.
2. Hospital specific base rates would be phased-in according
t:o the following schedule (see Table 2.1).
Table 2.1. Hospital Specific Base Rate Phase-in Schedule
1 1 Hospital Rate Federal Rate
1 Year 1 75% 25%
Year 2 50% 50%
Year 3 25% 75%
Year 4 0% 100%
3. Annual CPI inflation adjustments for rates. (Note: Due
!
to federal budget deficits, the update factor has been
j
amended annually based on an amount less than the CPI.)
1
4. Selection of an advisory commission (PROPAC) to assist
tjhe Secretary of HHS in making PPS decisions.
5. Recalibration of DRG weights every 4 years.
6. formulae for outlier payments defined (including cost-
j
outliers).
i
7. Guaranteed minimum payment for sole community hospitals.
8. increased payment for public hospitals that serve a
disproportionate number of the poor.
The majority
by the House
Continuation and approval for statewide demonstration
. j
projects.
of these modifications were based on an initial report
of Representatives (U.S. Congress, House of
Representatives, 98th Congress).
i ,
14


Hospitals had, by means of this new payment methodology, been
placed at risk financially to control escalating health care costs.
i
The new payment methodology was virtually untested, and the projected
effect was modelled to be highly uneven across the hospital industry.
I '
The effect on the provision of quality and quantity of care to
!
Medicare beneficiaries was unknown. The Social Security Amendments
of 1983 brought a new and innovative health care payment system into
j
being withinja period of 83 days. Quietly and quickly, Congress
initiated a transformation of the hospital industry.
I
Medicaid: A 1 Changing Reimbursement Environment
J
Prior to July 1988, participating Medicaid hospital providers
|
in Colorado Were paid based on the methodology established in a
i
Consent Decree issued by the U.S. District Court on December 13,
1977. The consent decree represented a settlement between the CHA
et al. and the State of Colorado. Department of Social Services
(CDSS). to define reasonable cost-based payment for hospital services
i
and to establish a per-diem rate setting methodology. Each hospital
I
per-diem wasjre-based upon the hospital's justified cost for the
I
hospital's fiscal year ending immediately preceding July 1, 1975.
Justified costs were approved annually in the Medicaid Audit
Settlement. 1
Rate modifiers, add-ons, and decreases in patient days were
I
used in the per-diem negotiation process. Add-ons could be requested
based upon: 1(1) additional or improved services, (2) capital cost
increases, and (3) case mix changes. Negotiations for increased
Medicaid rates, however, could only be justified based on major cost
i
increases. Each year, the prior year per-diem would automatically be
I
increased bylthe Consumer Price Index for wages (CPI-W). Only rate
modifiers and add-ons in excess of the CPI-W were valid increases
J
applied to the per-diem rates. i
i
i
15


The consent decree methodology is currently in use for several
|
types of facilities: Institutions of Mental Disorders, Specialty
!.
Hospitals, and Rehabilitation Hospitals. In July 1988, the Colorado
Medicaid Program initiated the Medicaid Prospective Payment System
(PPS) to pay
for hospital services on a DRG basis. The Medicaid PPS
methodology is based upon Medicare payment principles, with Medicaid-
!
specific values for DRG criteria: Base rates, relative weights, trim
I
points, and average length of stay values. The Colorado Medicaid PPS
I 1
does not have:a cost-outlier payment mechanism. From July 1988 to
June 1989, all psychiatric units within general acute care hospitals
I;
were paid on[a per-diem rate. As of July 1989, Medicaid psychiatric
services are
also paid based upon DRGs. Exempt rehabilitation care
units, within|general acute care facilities, continue to be paid a
per-diem rate. Hospital Medicaid outpatient charges are reimbursed
at 80% of the cost, as determined by Medicare reasonable cost
I
definitions.
The Colorado I Indigent Care Programs;
An Allocation of State General Funds to Providers
!
Two separate programs (with five separate line items)
constitute tie Colorado Indigent Care Programs (CICP). These
programs were,: designed to provide partial relief to designated health
care providers (primarily hospitals) for care rendered to Colorado's
I I
medically indigent residents. The funding priority is placed on
i
emergent or urgent care services.1
[;
1. Indigent Care Programs: Four line item appropriations
j ,
(the Denver Indigent Care Program, the Health Sciences
Center Indigent Care Program, the Specialty Indigent Care

1 Emergent care is defined in statute as medical care which
is, in the judgment of the physician, required within minutes to hours
to prevent loss of life or viable fetus or to prevent permanent
disability, j
i
i
16


Program, and the Out-State Indigent Care Program) are
reimbursed on a retrospective cost-based allocation. A
i
fixed level of funding is appropriated annually for each
line item. Within each line, provider facilities are
reimbursed a percentage of their costs based upon their
i
individual facility costs incurred as a portion of the
i
tlotal facility costs incurred for the program.
2. Community Maternity Program. This program reimburses
I
providers for obstetrical delivery services based upon
i
i
legislative "capitated" rates. Inpatient reimbursement
i
is based on type of delivery (normal vaginal or cesarean
j
section) and combines the hospital and physician payments
into one prospectively determined rate.2
Thus, CICP reimburses both on a cost-based methodology and a unique
|
packaged hospital/physician fee structure.
I i
In summary, these three government sponsored health care
i
programs (Medicare, Medicaid, and CICP) have different hospital
payment rates for the identical service provided to a given patient.
The respective payment schedules are independent of the hospital
costs incurred. The payment discrepancy is a function of each
I
government health care program's: (1) statutory authority, (2)
administrative interpretation of legislative intent, (3) funding
availability^ and (4) eligibility and benefit (covered service)
limitations, j
I
The Colorado Health Policy Environment
I
i
A recent survey conducted between December 1987 and April 1988
by Louis Harris for The Colorado Trust reported several findings: I
2 The Community Maternity Program is scheduled to end on
June 30, 1991..
17
I
i


1. The vast majority of Coloradans believe that access to
health care should be available to all Colorado
I
residents, regardless of ability to pay. Additionally,
the majority of residents are willing to support various
I
I 1
financing strategies to pay for this increased access
l
(e.g., cigarette and liquor taxes, premiums paid by
|
persons covered, and employer taxes).
2. According to the Harris survey, approximately 13% of
i
Colorado residents (312,000 persons) had unpaid medical
care bills.
i
The Colorado General Assembly historically has been
conservative on the issue of health policy. Only mandated Medicaid
expansions (for new populations) have been funded. No federal
i
optional Medicaid expansion features have been added. In the 1990
. i
session, however, a new public/private sector, budget-neutral program
\ i
for medically indigent children (The Child Health Plan) was funded.
Outside of this new program initiative, the aggregate funding for the
i
Colorado Indigent Care Programs (CICP) had remained nearly constant
(deteriorating in terms of "real" 1983 dollars).
The timing for a new health policy direction appears ripe: A
i
conflict exists currently between constituent values and legislative
tradition regarding access to quality health care services for
|
medically indigent citizens. The potential for a change is
exacerbated by the sunset provision of the CICP on June 30, 1991,
which would eliminate CICP. This program is the only assistance
available to many of Colorado's medically indigent residents. The
Colorado General Assembly must face these interrelated, complex, and
j
dynamic health policy issues in the 1991 session.
I
18


Colorado Hospital Financial Trends
The CHArecently released a comparative analysis on individual
hospital financial information for the calendar years 1988 and 1989
i
I
for 67 general-service, acute-care hospitals. Based on a comparison
of the gross
charges to the payments received by payor, the CHA
report asserts that underfunding of the "... Medicare and Medicaid
programs topped one-half billion dollars in 1989, increasing the
burden of payment to insured individuals and self-insured businesses"
|'j
(Colorado Hospital Association, 1990b).
CHA documents that profit margins are eroding statewide,
expressing a
concern for the future stability of the Colorado health
care delivery system. Total profit margins (from both patient
services and other sources) fell from 3.5% in 1988 to 3.2% in 1989.
The patient service profit margin fell from 1988 (unspecified in the ,
report) to a [level of 2.2% in 1989. The patient service margin
j
varied by hospital location, with urban hospitals at 2.4% as compared
to rural hospitals at 0.7%. As one might expect, differences in
profit margin varied substantially across individual facilities. The
fiscal situation for some Colorado hospitals appears precarious. The
question, then, is what action can hospitals take on an individual
basis to ward off the imminent fiscal threat and avoid potential
closure.
! i
19


CHAPTER 3
THEORETICAL FRAMEWORK
Profitability
Background I.:
A strong financial position is necessary for providers in the
health care industry, as in any other type of business, to assure the
continued ability to provide access to services. All hospitals have
i
financial requirements that must be met. These total financial
requirement s
(defined as the net costs for current and future
operations) include (Suver and Neumann, 1981):
1. Operating Expenses: Salaries and benefits and supplies.
2. Overhead Expenses: Administration, support service
activities, and facility operational expenses (e.g.,
utilities).
3. Teaching/Research Expenses: Education of
interns/residents and basic medical research.
i
4. Debt Service Expenses: Working capital requirements,
i
interest on debt, and repayment of debt.
I i
5. Capital Expenses: Replacement of existing equipment and
changes in technology, as well as depreciation expenses.
6. IjI6npatient-Related Activities: The net gain (revenues
minus expenses) from: (a) fund drives, (b) grant
proposals, and (c) investment management activities.3
I
7. Return on Equity (Profit Margin): Net of charity care,
bad debt, and contractual allowances.
3 Unlike other expenses, this factor is usually a positive
contributor used to offset other financial requirements.


Identifying these financial requirements is an important function for
i
hospital administrators and government program officials alike.
Hospital charges must be established adequate to meet the total
i
i
facility's financial requirements. Effective financial planning is
critical fori long-term hospital viability (Cleverley, 1985).
i
As discussed earlier, the hospital services required in the
I
future will be dramatically different than those provided today. The
i
growing number of elderly with multiple, chronic disorders will
require a coordinated, interdisciplinary hospital service delivery
system.
The
continued advancement of medical technology will require
I
both highly specialized professional health care teams and large
i
funds for capital investment. The financial requirements of health
I
care providers should accommodate these future changes to include
funds for: (!) service mix changes, (2) vertical integration
strategies (providing a continuum of care), (3) horizontal
integration strategies (new programs and technological advancements),
I 1
(4) financing working capital, and (5) existing capital equipment and
j
facility replacements. The financial performance of medical care
providers is
of importance, therefore, in assuring the timely
evolution and' future viability of the health care delivery system.
Financial performance can be measured in several ways. The
hospital industry has, in the last decade, adopted traditional
financial ratio analysis methods both to pinpoint individual facility
strengths andj weaknesses and to compare individual hospitals to
industrial averages (Suver and Neumann, 1981). A standard
performance ratio used is the operating margin ratio. This ratio
I
i
identifies the net operating profit as a percentage of gross
I
I
21


operating revenues (billed charges). Net operating profit is defined
as a function; of net operating revenues (payments received) minus
operating expenses. The operating margin ratio has been
i
demonstrated) as a single measure of financial performance, to be
reflective of, differences between high-performing and low-performing
i
hospitals (Cleverley, 1990b).
Defining an Adequate Profit Level
j
As with' any other business, it is important for hospitals to
earn an adeqiate profit margin. What constitutes an adequate profit
margin for hospitals, however, is subject to significant debate. A
comparison pgint might be the average profit demanded by the market.
Yet hospitals traditionally, as primarily a non-profit industry, have
i
earned significantly lower profit margins than other types of
businesses. (Millenson, 1990). Hospital industry profit margins over
the last 35 years are listed in Table 3.1.
i
Table 3.1. Longitudinal Comparison of Profit Margins
Year Hospital Profit Margin Service Industry Profit Margin
1955 1.5% n. a.
1965 3.4% n. a.
1984 6.3% n. a.
1985 5.8% 12.1%
1986 5.3% 9.2%
1987 5.1% 7.5%
1988 4.8% 8.2%
1989 5.1% 9.0%
i
Hospital profit margins in the period from 1984 to 1988 decreased
each year with a slight upward trend in 1988 to 1989 (Burda, 1990).
Other sources, outside of AHA, indicate that hospital profit margins
decreased from 1988 to 1989 (Kim, 1990). General service industry
gross profit
margins are listed for comparison ("Are the Good Times
22


Really Over," 1990). A comparison of similar industries holding
asset size constant, from the Leo Troy Almanac (1983 and 1990
editions), islisted in Table 3.2:
Table 3.2. Gross-Sectional Comparison of Similar Industries
Net Profit Before Taxes
Industry 1979 1987
Hospitals with $100-250K in Assets 7.8% 3.4%
Hotels!and Lodging with $100-250K in Assets 6.1% 46.2%
Social!Services with $100-250K in Assets 5.3% 3.6%
Hospital industry experts are projecting that hospital profit margins
will reach ajlow point of 2.0% in 1990 and increase to 2.6% in 1995
I ,
(Cleverly, 1989). The question of what an adequate and fair profit
margin should'be for the hospital industry, however, has remained
unanswered.
I
Special Considerations for Hospitals
Two additional circumstances must be taken into consideration
in any analysis of hospital profitability. First, the demand for
hospital services is to a large degree generated by physicians, not
by patients (Jacobs, 1987). Physician-induced demand causes
imperfections in the hospital market. To the degree that physicians
influence their patients' perceptions of their health status and the
i ,
effectiveness of medical care in improving their health, then the
fundamental hospital and physician economic interdependency is
enhanced. This interdependency is highest for elective procedures


I
with elastic
Remand curves, as demonstrated by the data showing that
a greater number of surgeons in a given area will result in a greater
number of elective surgeries being performed (Wennberg, 1984).
Second^ the majority of hospitals are organized as non-profit
or government;, facilities. Tax incentives and disincentives do not
have the same effect on hospital behavior as compared to other
i
business settings. The hospital industry is fundamentally different
in this regarcl than traditional industry. Hospitals have, by design,
j
multiple objectives. Non-profit hospitals may not be economic profit
maximizers, iut rather attempt to provide increased output (e.g.,
i
quantities ofJuncompensated care) or enriched services (e.g., more
attractive maternity suites) (Dranove, 1988). For many non-profit
hospitals, a | large separation exists between management and ownership
(e.g., a community hospital). Non-profit or governmental hospitals
do not need to respond to the forces of an organized equity market
(Glendon et al., 1987). Thus, the hospital industry is unique when
examining measures of financial performance.
I
i
Impact :of the Government-Sponsored Programs upon Hospitals
Medicare Prospective Payment System
i
A recent body of literature suggests that the Medicare PPS
methodology las directly affected the hospital industry. Initially
studies were j focused on changes in hospital behavior based on changes
in payment level. A concern existed that hospitals would serve only
i
i
patients with profitable DRG payments in relationship to their
'! '
individual facility costs (Broyles and Rosko, 1986). In contrast, it
appears that
dramatic product and productivity changes occurred in
hospitals "across the board" as a result of Medicare PPS (Long,
Chesney, andlFleming, 1989). Based on a calculated index of product
changes, theinumber of Medicare patients discharged to home or to
24


self care wasj reduced overall after PPS implementation. A
corresponding increase occurred in the number of transfers to another
I
short-term hospital, discharges to a skilled nursing facility, and
i
discharges to home health agency. Based on a calculated index of
productivity!changes, overall reductions in inputs used for all DRGs
(i.e., increased productivity) occurred. Long, Chesney and Fleming
concluded that hospitals did not appear to be selective in their
I
product or productivity changes within a set of DRGs. Their
conclusion is associated with the fact that hospitals have been
unable to identify their costs of providing patient care for a given
i
i
DRG for a Medicare recipient. Thus, specific identification of the
profitable and non-profitable DRGs for a given hospital has not been
possible from a technical, financial cost accounting perspective.
I
Since the implementation of the prospective payment system, a
|
health policy concern has existed that hospitals would make trade-
offs between;cost containment and quality of patient care (e.g.,
discharging patients with significant impairments at discharge with
increased frequency). An increase in readmission and mortality rates
was projected to occur. Early research by Des Harnais et al. (1987)
i
indicated that readmission rates and mortality rates for the Medicare
i
population after 1984 (the initiation of Medicare's prospective
payment system) were comparable with the rates for the 1980 to 1983
period. DesJHarnais et al. found no evidence that the quality of
care deteriorated after the revised payment methodology. However,
similar research later conducted by Kosecoff et al. (1990) determined
I
that, in fact, more recent mortality rates and readmission rates
supported the theory that hospitals are discharging patients "quicker
and sicker."i Thus, the current Medicare hospital reimbursement
system appears to impact adversely the quality of patient care.


I
As a result of the potential adverse patient care implications
i;
and trends of1 decreasing hospital profitability, a growing policy
i
concern is adequacy and fairness of Medicare PPS (Davis et al., 1990;
Guterman et al., 1990). A recent study demonstrates the negative
I
effect of the! Medicare PPS upon hospital Medicare operating ratios
(Gianfrancesco, 1990).4 A regression analysis was performed with
the following^ non-Medicare payment variables included in the
i
analysis: beds, occupancy rates, ratio of Medicare discharges to
j
total discharges, geographic locator code, and dummy variables for
i
profit status! (for-profit, voluntary, or government). The study
reported that1 the Medicare PPS methodology systematically under-
compensated, certain classes of hospitals (non-metropolitan). This
!'
study, consequently, would support the CHA contention that
i
government-sponsored health care programs differentially impact
i ,
hospital profitability.
Colorado's Medicaid Payment System
I
As described earlier, the Medicaid PPS methodology in Colorado
I .
,1
was initiated; m July 1988. There was substantial concurrence from
j |
the hospitaliindustry (the Medicaid DRG Advisory Committee and CHA)
on the methodology developed. No concerns were indicated in the
following areas of the Medicaid PPS methodology: (1) the relative
weights used
(based on four other state Medicaid PPS programs) for
each DRG, (2) the designation of three peer groups (urban, rural, and
i
rural referral) for hospital base rates, (3) definition of Medicare
allowable costs used in the Medicaid base rate calculation, and (4) a
i ;
resource consumption adjustment factor of 88% (based on an inter-
state study that demonstrated that Medicaid patients consume 12%
as (Medicare
by (Medicare
Operating ratio for the purposes of this study was defined
inpatient revenue minus Medicare inpatient cost) divided
inpatient revenue).
26


I I
fewer resources for a given DRG than Medicare patients). However, a
disagreement persisted between the hospital industry and CDSS as to
whether the Federal Boren Amendment requirements (that state Medicaid
agencies musi1 pay the costs of an economically and efficiently
operated hospital) were met. The disagreement was focused primarily
'I
on the departments use of a Medicaid budget adjustment factor of
i
54%, which was applied to the base rates established to ensure that
the Medicaid IPPS methodology was budget neutral. The budget
adjustment factor was calculated by the CDSS modelling the funds
|l
required to reimburse hospitals under the new Medicaid PPS
methodology,
as compared to the 1988 Colorado Long Bill Appropriation
Hospital Line Budget. This contention over the use of a budget
i
adjustment factor in the rate setting process resulted in the AMISUB
!
(PSL) et al.lv CDSS lawsuit. The U.S. 10th Circuit Court found in
favor of the
hospitals (No. 88-2482 filed July 11, 1989). Under the
Medicaid PPSjprogram, it was demonstrated that no Colorado hospital
i
(no matter how efficiently run) would be paid its costs for serving
Medicaid clients. The CDSS implemented a new cost-based PPS
!'
methodology on November 28, 1990, retroactive to December 15, 1989,
in accordance with the lawsuit settlement. This settlement resulted
in 17 rural hospitals (41.5%), 3 other urban hospitals (23%), and 5
i
Denver metropolitan hospitals (33%) receiving their hospital's cost
j-'
per discharge, as the basis for establishing their facility's
i
prospective payment reimbursement.
Colorado's Indigent Care Programs
i
Total provider facility costs reported in Colorado for fiscal
year 1989 (medically indigent patient care, medically indigent bad
debt, and administrative costs associated with CICP participation)
across all programs were $100,115,708. Reimbursement to these same
providers was $39,734,055. Overall, the CICP reimbursed on the
i
27


average 40% of costs. (The unreimbursed costs of CICP are accounted
i
for by hospitals as uncompensated care in the form of either charity
care or bad debt.) The CICP reimbursement rates varied dramatically
among providers, as non-hospital facilities generally received
significantly lower relative reimbursement rates compared to hospital
facilities (Colorado Indigent Care Program, 1990).
Hospital Cost Behavior
Hospitals are multi-product firms that provide large numbers of
separate product lines, such as emergency room services, transplant
!
services, and obstetrical services (Jacobs, 1987). In examining the
production function of hospital costs, all hospitals incur fixed
costs (costs that do not vary with output, such as depreciation
expenses) and, variable costs (costs which do vary with output,
including a portion of labor and supply expenses). In examining the
fundamental differences across hospital cost functions, several
underlying conditions may affect the production relationship: (1)
case mix (a measure of the complexity of the case), (2) severity of
illness, (3)
quality of care (input, process, and outcome measures),
(4) technology, (5) capital intensity, (6) efficiency of personnel,
i
j
and (7) incentive systems in place internally for cost containment.
For example, technological advancements can increase the average cost
per casee.g., increased equipment and personnel costs (more highly
trained operators)or decrease the average cost per case (new
laboratory equipment automating functions previously labor
intensive). (Additionally, hospitals may substitute different input
ratios to aclieve the same product (e.g., use of licensed practical
nurses for a!larger portion of the nursing care) and therefore affect
the hospital I cost function.
28


Extensile research on hospital cost behavior has been conducted
(Granneman and Brown, 1986; Thorpe, 1988). The findings of these
studies have
identified several important variables which may explain
hospital cost differences, including:
1. Environmental Variables: Competitiveness of the hospital
j
environment, geographic location, and per capita income
of population.
i
2. Structural Variables: Teaching, age of medical staff,
hospital ownership, and profit status.
3. Management Variables: Case mix, wages, ratio of
forecasted to actual admissions, emergency room activity,
inverse of occupancy rates, staffing patterns, scope and
J
mix of services, and use of technological advancements.
Several of tlese factors, hypothesized to influence hospital cost
i,
behavior, are included in theoretical model developed to explain
I
i
hospital profitability.
Hospital Provision of Uncompensated Care
Background
The number of uninsured or under-insured people living in
Colorado appears to be growing. Recent estimates for 1988 indicate
I
that approximately 522,000 persons in Colorado are uninsured.
I
Approximately 60% (313,000) are medically indigent, that is, they
]:
would be unable to pay for their medical care bills (Center for
Health Ethics! and Policy, 1989). As uncompensated care has
historically
been financed by implicit subsidies from other payor
sources (cross-subsidization), the growth of the medically indigent
population and the recent market pressures for increased competition
placed upon hospitals have resulted in diminished access to non-
emergency care (Feder and Hadley, 1985; Hadley and Feder, 1985).
29


Charity! care is defined as care provided to medically indigent
persons and others who are unable to pay. In Colorado, the CICP
i
residual charges (not covered by the CICP reimbursement, third party
payors or patient co-payments) for care rendered to medically
indigent patients can be accounted for as charity care. Bad debt is
differentiated by the presumption that the person had the ability to
I
pay, but chose not to do so. (For example, the uncollected CICP
j i
patient co-payments could be accounted for as bad debt.) The
distinction between bad debt and charity care is not well-defined and
varies across hospitals. Uncompensated care is usually defined as
!
the sum of both charity care and bad debt.
j
Hospitals with emergency rooms are required by federal law to
treat all urgent patients regardless of ability to pay (COBRA 1985,
PL 99-272).
Most non-profit hospitals in Colorado have had
historical obligations requiring a certain level of charity care
j 1
under Hill-Bdrton building program requirements. The Hospital Survey
j
and Construction Act of 1946 (i.e., Hill-Burton) provided funds for
non-profit hospital building programs. Hill-Burton facilities are
required to provide uncompensated care in an annual amount which is
. i
the lesser of: (1) 3% operating costs, or (2) 10% of all federal
assistance provided to the facility, adjusted for inflation. The
Hill-Burton obligation is for up to a 20-year period after completion
of the structure. The last Hill-Burton funding (PL 93-641) was
distributed in 1975. Thus, the Hill-Burton obligations for most
i
hospitals are: quickly coming to a close.
Colorado is one of the few states in the nation where
! '
"medically indigent persons are not entitled to receive medical
services . .j as a matter of right" (C.R.S. 26-15-103). In Denver
County, the Denver Health and Hospital system is designated under
City Charter
as having the legal responsibility to provide medical
30


care for the Denver poor. Outside of Denver County, local government
tax subsidies are available to hospitals in different parts of the
state. These1 counties, however, have no legal requirements for
subsidized hospitals to provide service to medically indigent county
residents. Additionally, as part of the standard Colorado Indigent
Care Program
contract any
contractual process, hospitals may exclude from their
j(or all) non-emergent care services. Thus, Colorado
hospitals have the capability to determine the non-emergent charity
care levels provided.
I
i
i 1
Recognized Need for Financing: Disproportionate Share Payments
Both Medicare and Medicaid pay hospitals at increased rates, if
the hospitalJqualifies as a disproportionate share provider. For the
Medicare Progiram, the disproportionate share providers either:
j
1. Serve a significantly disproportionate portion of low
! i
I i
income patients (e.g., greater than 15% low income
patient days for an urban hospital with 100 or more
lleds); or
I:
2. Are located in an urban area, have greater than 100 beds,
and have greater than 30% of revenues derived from state
and local government payments for indigent patients (not
covered by Medicare or by Medicaid).
For the Colorado Medicaid Program, the disproportionate share
j I
providers are1 defined by the following criteria:
1. Medicaid Utilization Rate: (Medicaid days)/(Total days)
i
is greater than 1 standard deviation above the mean
i1
ratio; or 2
2. Low Income Utilization Rate: The inpatient revenues
i
associated with charity care and unsponsored care
constitute at least 25% of the facility's total inpatient
revenues.
31


I
This higher payment by both Medicare and Medicaid is intended to
recognize anil to cross-subsidize a portion of the hospital's costs
for charity care.
i
Distribution I Of Uncompensated Care
i
i
A significant amount of research has been done on the
characteristics (location, size, teaching status, ownership and
profit status!) of hospitals with uncompensated care. The findings
reported are
not consistent. However, environmental and structural
factors appear to impact the provision of uncompensated cares
1
1. Teaching Status and Geographic Location: Public teaching
hospitals and inner-city hospitals provide relatively
i 1
larger percentages of charity care (based on bed size)
I
(Mulstein, 1984; Bazzoli, 1986).
I
2. Geographic Location: Small rural hospitals provide
i
significant amounts of charity care to their surrounding
i
communities (Saywell et al., 1989).
3. Ownership: Government facilities provided overall the
j
largest share of charity care (Ohsfeldt, 1985).
Investor-owned facilities provided overall the smallest
i :
share of charity care (Wilensky, 1984; Frank, Salkever,
i
apd Mullann, 1990).
This research: indicates that teaching status and ownership are the
i
most significant predictors of a hospital's charity care provided.
j Factors Affecting Hospital Closure
Closure; is the last symptom of financial decay (Solovy, 1989).
In order to understand the important variables influencing hospital
profitability and ongoing financial viability, the hospital closure
literature was reviewed. Hospitals with inadequate financial
performance are more often subject to closure (Cleverley, 1990c).
I
i
32


Recent studies have attempted to determine the factors which may
impact hospital closure (Longo and Chase, 1984; Mayer et al., 1987;
Mullner and Whiteis, 1988; Burda, 1989; Cleverly, 1990a)s
i
i ,
1. Environmental Variables: Region, SMSA status, physician
i
to population ratio, hospital-bed to population ratio,
number of other hospitals in the county, per capita
i
income, regulation, unemployment rate, and percent
population change.
2. Structural Variables: Ownership type, bed size, lack of
membership in a multi-hospital system, CEO's affiliation
with American College of Healthcare Administration
i
I .
(ACHA), teaching status, presence of a nursing home unit,
and diversification.
|
3. Management Variables: Occupancy rate, charity care and
i
bad debt provided, physician practice patterns, and
Quality.
4. Government Program Variables: Medicare and Medicaid
i
payments.
These studiesi demonstrate an inconsistency between the CHA (1989) and
Gianfrancesco (1990) assertions that government-sponsored health care
programs have1 differentially impacted hospital profitability. The
i
i
results of the closure research indicated that hospitals closed for
reasons unrelated to government-sponsored health care program payment
'j
levels (Burda, 1989; Cleverly, 1990a). The primary variables
affecting hospital closure appeared to include: excessive prices and
costs, lack of investment in plant, deteriorating capital equity,
!
eroding liquidity, excess capacity, and uncompensated care.
33


Financially Distressed Hospitals
Rizzo '(1990) studied the characteristics of financially-
distressed hospitals, both before and after PPS. The relationship
i '
between financial distress (defined as poor financial performance
|
sustained over a three year period) and factors which were
I
controllable!and uncontrollable by hospital management was analyzed.
j
The independent variables were classified into six categories:
(1) revenue-related factors, (2) cost-related factors, (3) payor mix,
I
(4) patient characteristics, (5) treatment patterns, and (6) patient
discharge status. Rizzo's study demonstrated that the
characteristics of distressed hospitals changed from the early to
middle 1980s1.! Distressed hospitals, when compared to solvent
hospitals, had: (1) provided more services to low income patients
j,,
(both self-pay and Medicaid patients), (2) provided more charity
care, (3) a critically poor revenue situation, and (4) overall higher
costs than comparable facilities. In summary, Rizzo's research lends
j
support to the CHA assertion that the profitability of the hospital
industry has
been affected by the Medicaid Program.
Proposed Model of Hospital Profitability
The li:prature search, as discussed above, revealed no
conclusive explanatory model of the relationship between hospital
profitability! and government-sponsored health care program
reimbursement,. Research examining the factors influencing hospital
profits, hospital cost behavior, levels of uncompensated care, and
I
hospital closure were reviewed. Based primarily upon this
I
literature, the comprehensive hospital profitability model has been
developed. ^('In the figure, the original model established by Longo
and Chase inj 11987 has been expanded to address the influence of
government prbgram variables.)
34


I ,
Figur|e 3.1. Proposed model of hospital profitability.
*Note: Longo-Chase model expanded to include this new dimension.
Environmental! variables will include:
1. Per capita income of population in hospital's service
i
region.
2. General inflation index (Medical Consumer Price Index).
3. Seasonal (fall, winter, spring, and summer) variation.
Structural variables will include:
1. A, hospital specific variable, which will be adjusted over
time. This variable inherently reflects the geographic
location, the ownership and the bed size for the
hospital.
|
2. Teaching status.
j'
Management variables will include:
1. General hospital statistics (occupancy rate, case mix
index, and average length of stay).
2. Revenue variables.
L!
3. Expense variable.
4. A measure of the proportion of patient care revenues
dedicated to uncompensated care.
Government program variables will include:
1. Percent government payor activity by payor.
35


2.
Reimbursement methodology (PPS/Non-PPS).
3. Reimbursement rate per discharge by payor.
4. Disproportionate share payment status by payor.
|
The independent variables outlined above incorporate factors
I
I"
shown to be important in studies of hospital financial performance
(Coyne, 1982; Coyne, 1986; Glendon et al., 1987; Cleverley, 1990b;
!
Vanselow, 1990). This study, however, addresses only one facet of a
complex, multi-dimensional, dynamic problem. (For other possible
research pursuits, see Appendix 1.) This proposed model of
profitability will be used to explore the effect of government-
sponsored health care programs upon Colorado hospital profitability.
|
i
!
i
j
i,
!
i
i
i
i
j
i
i
i
i
i
I
i
i
i
I
i
i
36


CHAPTER 4
METHODOLOGY AND DATA SOURCES
Primary Hypothesis
As discussed earlier, -the purpose of -this study is to
I :
investigate the association between government-sponsored program
activity and
hospital financial performance. Thus, the primary
hypothesis directly addresses this research question.
H(l): Government-sponsored health care programs (Medicare,
Medicaid, and the Colorado Indigent Care Program) are
negatively associated with Colorado hospital
profitability. Therefore, it is postulated that
increases in government program utilization and/or
decreases in government program reimbursement rates will
be associated with decreases in Colorado hospital
profitability.
Design
This study's design is passive correlational in nature
(Tabachnick and Fidell, 1989). As a result of this design, neither
M
direct causal relationships nor the direction of any relationship can
j
be established. The purpose of using multiple regression analysis,
with repeated measures design simulated, is to determine the degree
r
of association between hospital profitabilityi.e., the dependent
i !
variable (DV)--and a variety of independent variables (IVs). As this
study focuses primarily on profitability, it is by design one-
dimensional .
A multi-dimensional assessment of financial performance
(based on equity, debt, and asset data) could not be performed with
the data available. The profitability index permits comparison of
hospitals along a continuum of financial performance over time
(controlling
for hospital-specific characteristics). The model


separates endogenous and exogenous dimensions of hospital
profitability. This study is exploratory in nature with the purpose
of building an explanatory model of hospital profitability.
Profitability Function
In accordance with the model proposed, hospital profitability
i.e., (PSOM)-j--is proposed to be explained by the following linear
relationshipjto the independent variable categories:
PSOM =j f (environment variables)
I +f (structure variables)
| +f (management variables)
! +f (government program variables)
Secondary Hypotheses
As discussed earlier, profitability is defined as a function of
(net revenues minus expenses). Net revenues are a function of
[(gross revenues from both patient and non-patient service revenues)
minus (charity care plus bad debt plus contractual adjustments)].
Expenses areja summation of payroll, benefits, supplies,
depreciation/ interest, and all other expenses. (The costs of
. i
!
providing charity care are routinely included in general expenses,
such as payroll expenses, and they cannot be distinguished.)
I '
Hospitals, to counteract decreases in profitability, may
directly respond by: (1) raising charges, (2) lowering costs, and/or
(3) lowering
the amount of uncompensated care provided.5 The
following secondary hypotheses will provide insight into the findings
of the primary hypothesis, by further exploring the relationship
between government-sponsored health care programs and a hospital's
management choices:
5 Although other hospital reactions are possible (e.g.,
increasing marketing efforts to change payor mix), this study is
limited to the three management responses defined above.
38


H2(1) :
H3(1):
i
H4(l) :
i
Government-sponsored health care program activity
is negatively associated with hospital inpatient
gross revenues (charges) per discharge. Therefore,
it is postulated that increases in government
program utilization and/or decreases in government
program reimbursement rates will be associated with
increased gross revenues per discharge.
Government-sponsored health care program activity
is positively associated with hospital salary
expenses per hour. Therefore, it is postulated
that increases in government program utilization
rates and/or decreases in government program
reimbursement rates will be associated with
decreased payroll expenses per hour.
Government-sponsored health care program activity
is positively associated with the percent of gross
patient revenues dedicated to uncompensated care.
Therefore, it is postulated that increases in
government utilization rates and/or decreases in
government program reimbursement rates will be
associated with decreased levels of uncompensated
care.
Primary Hypothesis: Dependent Variable
The dependent variable within this study is hospital patient
!
service operating margin (PSOM) (Furst, 1981). This variable is
calculated as follows:6
1. (psOM) = (Net PS Revenues PS Expenses)/(PS Revenues)
' = ({[3] [4]> /[2])
2. (PS Revenues) = (Gross Inpatient Revenue)
+ (Gross Outpatient Revenue)
3. (Net PS Revenues) = (Gross Inpatient Revenue)
I + (Gross Outpatient Revenue)
j - (Bad Debts + Charity Care)
| - (Contractual Adjustments)
4. (PS Expenses) = (Operating Expenses)
Note: PS = Patient Service.
6 The CHA formula used to calculate patient service margin
is differentl Patient service margin =
i (Net PS Revenues PS Expense
_____4- Operating Revenue from Non-Patient Services)____
(PS Rev^nues+Operating Revenue from Non-Patient Services)
39


Primary Hypothesis; Independent Variables under Study
The independent variables have been chosen carefully (extracted
primarily from the research literature by Longo and Chase, 1984;
Wilensky, 1984; Granneman and Brown, 1986; Thorpe, 1988; Saywell
et al., 1989: Cleverley, 1990a; Gianfrancesco, 1990; and Rizzo,
1990), based ,on their relationship to the theoretical model
developed. No extraneous variables are included in this analysis.
For the purposes of this study, hospital profitability is proposed to
be a function of four types of independent variables:
1. The environment within which the hospital is located.
a, . Cumulative Consumer Price Index for Medical Care
j Goods, where January 1986 = 100% (xxx.xx%)
j (MCPIWCUM)
b. Regional average per capita income adjusted by the
j: medical CPI-W (CADJAPCI) ($xx,xxx)
cl. Seasonality dummy code: Spring (SPRING)
d. Seasonality dummy code: Summer (SUMMER)
e. Seasonality dummy code: Winter (WINTER)
f. No seasonality code: indicates Fall season
2. The structure of the hospital (not directly controlled by
hospital management), which characterizes the hospital's
mission and services provided.
a|. A total of 65 hospital specific identifiers, which
are used to correlate repeated measures for each
facility. (HOSP1 to HOSP64)
b. Teaching Status: (Dummy = 1 if funding UCHSC
residents + interns) (TEACH)
3. The management of the hospital, which describes the
efficiency of the operations internally and the
facility's commitment to charity care. (Note: The
provision of non-emergent charity care is a management
choice.)
a, . Descriptive statistics
| 1) Occupancy rate (based on total days/licensed
| bed days) (OCC) (xx.xxl)
2) Average length of stay (ALOS) (xxx.xx)
3) Hospital Case Mix Index (CMI) (xx.xxxx)
b. Revenues/unit
1) Gross Inpatient Revenues/Total Discharges
(IRV_TD) ($xx,xxx)
2) Non-patient Revenues/Total Gross Revenues
percent (NPT_PCT) (xx.xx%)
40


I
I
I
c. Expenses/unit7
| Salary and Benefit Expenses per Paid Hour (SAIi_HR)
! ($xx.xx)
! ,
d. Uncompensated care as a percent of Total Gross
! Inpatient and Outpatient Revenues (xx.xxl)
(UNCP_PCT) (Note: Uncompensated care includes bad
| debt and charity care.)
4. The government program activity and reimbursement.
a. Medicare (MC)
i 1) Percent discharges (MC_D) (xx.xx%)
j 2) Base PPS rate per discharge (MED_RATE)
i ($xx,xxx) (Note: This rate does not reflect
! disproportionate share, indirect medical
; education, direct medical education, or
' capital pass-through payments.)
3) Disproportionate share (MED_DSP) dummy6
j /
b. Medicaid (MK)
j 1) Percent discharges (MK_D) (xx.xx%)
; 2) PPS or Non-PPS dummy (MKD_PPS)
1 3) Disproportionate share (MKD_DPS) dummy
, 4) Base payment rate per discharge (MKD_RATE)
! ($xx,xxx) (Note: This rate does not reflect
! disproportionate share or major teaching
j hospital payments.)
c. Medically Indigent (MI)
I 1) Percent discharges (MI_D) (xx.xx%)
! 2) Payment rate per discharge (MI_RATE)
j ($xx,xxx)
I Secondary Hypotheses:
Dependent and Independent Variables under Study
For each of the secondary hypotheses, the form of each equation
j
of study is proposed:
H2(1): | 1 Government-sponsored health care program activity
! is negatively associated with inpatient hospital
! gross revenues per discharge.
Dependent variables of study:
Inpatient Revenues Per Discharge
adjusted by the CPI-W (ADJIRV_D) ($xxx,xxx)
i
7 Although it would have been preferable to include other
measures of expenses per unit in this study, visit data was not
captured by CHA Data Bank until 1988. Thus, it was not possible to
include other cost indicators in this study.
j
8 Medicare teaching payment information was not available.
41


Independent variables include:
1. The environment within which the hospital is located.
a. Regional average per capita income adjusted by the
medical CPI-W (CADJAPCI) ($xx,xxx)
b. Seasonality dummy code: Spring (SPRING)
c. Seasonality dummy code: Summer (SUMMER)
clj. Seasonality dummy code: Winter (WINTER)
e. No seasonality code: indicates Fall season
2. The structure of the hospital.
j
a. A total of 65 hospital specific identifiers, which
| are used to correlate repeated measures for each
; facility. (HOSP1 to HOSP64)
b. Teaching Status: (Dummy = 1 if funding UCHSC
residents + interns) (TEACH)
3. The management of the hospital.
!
a. Occupancy rate (based on total days/licensed bed
j days) (OCC) (xx.xxl)
b. Average length of stay (ALOS) (xxx.xx)
c. Hospital Case Mix Index (CMI) (xx.xxxx)
d. Non-patient Revenues/Total Gross Revenues percent
! (NPT_PCT) (xx.xx%)
4. The government program activity and reimbursement,
a. Medicare (MC)
1) Percent discharges (MC_D) (xx.xx%)
2) Adjusted base PPS rate per discharge
(ADJMCRAT) ($xx,xxx)
| 3) Disproportionate share (MED_DSP) dummy
!>;. Medicaid (MK)
1) Percent discharges (MK_D) (xx.xx%)
2) PPS or Non-PPS dummy (MKD_PPS)
3) Disproportionate share (MKD_DPS) dummy
4) Adjusted base payment rate per discharge
j (ADJMKRAT) ($xx,xxx)
j
c. Medically Indigent (MI)
1) Percent discharges (MI_D) (xx.xx%)
2) Adjusted payment rate per discharge
(ADJMIRAT) ($xx,xxx)
H3 (1)
Government-sponsored health care program activity
is positively associated with hospital salary
expenses per paid hour.
Dependent variable of study: Payroll expense per paid hour
adjusted by the CPI-W (ADJSALHR) ($xx.xx)
Independent variables include:
42


1.
The environment within which the hospital is located.
j
a. Regional average per capita income adjusted by the
J medical CPI-W (CADJAPCI) ($xx,xxx)
Is. Seasonality dummy code: Spring (SPRING)
c. Seasonality dummy code: Summer (SUMMER)
d. Seasonality dummy code: Winter (WINTER)
e. No seasonality code: indicates Fall season
j
2. The structure of the hospital.
!
a. A total of 65 hospital specific identifiers, which
i are used to correlate repeated measures for each
facility. (H0SP1 to HOSP64)
b. Teaching Status: (Dummy = 1 if funding UCHSC
' residents + interns) (TEACH)
3. The management of the hospital.
a. Occupancy rate (based on total days/licensed bed
t days) (OCC) (xx.xx%)
b. . Average length of stay (ALOS) (xxx.xx)
c. Hospital Case Mix Index (CMI) (xx.xxxx)
d. Non-patient Revenues/Total Gross Revenues percent
(NPT_PCT) (xx.xx%)
4. The government program activity and reimbursement.
a. Medicare (MC)
j 1) Percent discharges (MC_D) (xx.xx%)
2) Adjusted base PPS rate per discharge
1 (ADJMCRAT) ($xx,xxx)
! 3) Disproportionate share (MED_DSP) dummy
I
b. Medicaid (MK)
! 1) Percent discharges (MK_D) (xx.xx%)
I 2) PPS or Non-PPS dummy (MKD_PPS)
' 3) Disproportionate share (MKD_DPS) dummy
j 4) Adjusted base payment rate per discharge
| (ADJMKRAT) ($xx,xxx)
c. Medically Indigent (MI)
1) Percent discharges (MI_D) (xx.xx%)
| 2) Adjusted payment rate per discharge
| (ADJMIRAT) ($xx,xxx)
H4(l): Government-sponsored health care program activity
i is positively associated with the percentage of
| gross patient services revenues dedicated to
; uncompensated care.
!
Dependent variable of study: Uncompensated care revenues/Total
I
Patient Revenues (UNCP_PCT) (xx.xx%)
Independent variables include:
j
The environment within which the hospital is located.
1.


2.
3.
4.
a.
b.
i
I
c.
cl.
|..
The
Cumulative Consumer Price Index for Medical Care
Goods, where January 1986 = 100% (xxx.xx%)
(MCPIWCUM)
Regional average per capita income adjusted by the
medical CPI-W (CADJAPCI) ($xx,xxx)
Seasonality dummy code: Spring (SPRING)
Seasonality dummy code: Summer (SUMMER)
Seasonality dummy code: Winter (WINTER)
No seasonality code: indicates Fall season
structure of the hospital.
a. A total of 65 hospital specific identifiers, which
I are used to correlate repeated measures for each
| facility. (HOSP1 to HOSP64)
b. Teaching Status: (Dummy =1 if funding UCHSC
residents + interns) (TEACH)
The management of the hospital.
a. Occupancy rate (based on total days/licensed bed
| days) (OCC) (xx.xx%)
b. Average length of stay (ALOS) (xxx.xx)
c. Hospital Case Mix Index (CMI) (xx.xxxx)
d. Non-patient Revenues/Total Gross Revenues percent
| (NPT_PCT) (xxxx%)
The government program activity and reimbursement,
ai. Medicare (MC)
j 1) Percent discharges (MC_D) (xx.xx%)
I 2) Base PPS rate per discharge (MED_RATE)
! ($xx,xxx)
3) Disproportionate share (MED_DSP) dummy
i
b. Medicaid (MK)
1 1) Percent discharges (MK_D) (xx.xx%)
I 2) PPS or Non-PPS dummy (MKD_PPS)
i 3) Disproportionate share (MKD_DPS) dummy
4) Base payment rate per discharge (MKD_RATE)
($xx,xxx)
c. Medically Indigent (MI)
1) Percent discharges (MI_D) (xx.xx%)
2) Payment rate per discharge (MI_RATE)
($xx,xxx)
Regression Analysis
For the purposes of this study, the hospital financial monthly
I
data available are described by a multiple linear regression repeated
measures design (Tabachnick and Fidell, 1989). The multiple linear
I
regression repeated measures design is required, as the 3,412 records
are not independent. The 3,412 records used in this analysis
|
i
44


I
I
I
represent summary financial data for 65 hospitals over a 56-month
period. To simulate the repeated measures design, each record was
assigned a hospital identification dummy variable (H0SP1 to HOSP64)
i
to simultaneously account for differences between hospitals across
time (Cody and Smith, 1987). Using the hospital identification dummy
variables resulted in increased R-squared correlation coefficients
I
for all regression equations (e.g, PSOM analysis R-square increased
) 1
from 0.4087 tp 0.5220). It was not possible to simultaneously
include the hospital characteristics as separate independent
i
variables in!the regression analysis (e.g., hospital ownership,
hospital geographical location, bed size, nursing home bed size, and
i
swing bed availability), as these characteristics were directly
i
associated with the hospital identification variables. Thus, a
I
weakness inherent in this analysis is that cross-sectional variation
in the independent variables of study may be partially contained in
the hospital|identification dummy variable. (For a discussion of the
sensitivity of the regression analysis repeated measures design, see
Appendix 2 .)
In order to test each hypothesis [H(l), H(2), H(3), and H(4)],
four linear regression analyses were performed. The estimated linear
regression coefficients and significance levels for each variable
were calculated. An aggregate F statistic and R-square for the
equation was|calculated. The beta coefficient for each variable
i
measured the'change in the dependent variable based on a single
i
i
standard deviation change in the independent variables. A
i
standardizedj regression coefficient was calculated for each variable.
Assuming the|independent variables are uncorrelated, the importance
i
of an independent variable is related to the absolute value of its
standardized!regression coefficient. The probability associated with
i
each variable, if less than 0.05, indicates statistical significance
i
j
i , 45 I
I
I
1
I


at the 95% confidence level. The F statistic will take into
i
consideration changes in all regression coefficients simultaneously.
The R-square
measures the goodness of fit" of the model with the
data (Tabachnick and Fidell, 1989). An R-square of 0.30 was
considered acceptable for purposes of this analysis (30% of the
i
variation in! the dependent variable of study was explained by the
!
equation).
Several assumptions are key in this analysis:
!
1. No other variables (i.e., the variables excluded from the
i
analysis) are relevant.
I
2. The variables used in the regression are measured
accurately and uniformly across hospitals.
3. The independent variables are independent of each other.
(Note: Tests for multicollinearity effects were
performed and used to identify problems.)
Changes in the independent variables result in
I
instantaneous changes in the dependent variable,
delay factors were included in this analysis.
No time
Failure to find a significant effect of an independent
i
variable upon the dependent variable of study does not
I
indicate conclusively that an effect is not present.
Rather, it indicates that an association could not be
i ,
established given the defined research methodology.
i
i
) Primary Data Source
A data|base of Colorado hospital monthly financial data, for
the period January 1986 to August 1990, from the Colorado Hospital
!
Association Data Bank was used for this study. In total, 3,726
i
records wereireceived. This data base contains monthly financial
records for ail general acute care hospitals in the state, with the
46


exception of! the Veterans Administration Hospital, military
1 |
hospitals, Colorado State Hospital's general acute care unit, and
i
National Jewish Hospital. The population of Colorado general acute
care hospitals is nearly completely represented in this study.
!'
The Colorado Hospital Association data base was linked (based
on hospital and time period) to environmental, structural, and
|
government program data. All data were normalized to a standardized
basis to address unit of analysis discrepancies across information
I '
resources (e'.g., all government program reimbursement was calculated
per discharge). Several potential problems were encountered with the
data sources> due to inherent limitations both in data integrity and
i i
availability;. These problems may result in internal validity issues
i
such as: (1) different hospital accounting reporting methodologies,
i
i
(2) different government program definitions of payment and
utilization,|(3) lack of period-specific retroactive adjustments to
i
the CHA databank data, and (4) lack of recent updates for two
i
independent variables (case mix index and per capita income rates).
1
i Life Cycle Changes in Hospital Facilities
For the purposes of this study, data for all new or closed
hospitals for the 18-month period (either prior to closing or
subsequent to opening) were automatically excluded as outlier data.
I
It is assumed that hospital profitability data will be skewed during
these periods of instability. Additionally, data for combined
facilities were carefully aggregated. The facilities impacted by
these life cycle changes are:
I
ClosedlHospitals: Boulder Memorial Hospital, Greeley Memorial
i
I
Hospital, Walsh District Hospital, Washington County Hospital,
I
Doctor's Hospital in Colorado Springs, Rocky Mountain Hospital,
AMC Cancer Research Center, and Monte Vista Community Hospital.
i
I
47


New Hospitals: Avista, Porter-Littleton, and Vencor.
Combined Hospitals: Penrose/St.Francis, St. Thomas More/St.
Joseph's Hospital in Florence, St. Anthony's Central/Saint
!
Anthony's North/Saint Anthony's Pavillion, Presbyterian
I
Denver/Presbyterian Aurora/St. Luke's Hospital, and Humana
Aurora/Humana Mountain View.
Additionally1, data for Craig Hospital and Centennial Peaks Hospital
j
were eliminated, as their primary business is either rehabilitation
j
care or psychiatric care, in spite of their facility's general acute
care licensure.
Data Screening
I
As the:accuracy of the data used in the linear regression model
I
is of major importance, extra attention was paid to data screening
techniques. The raw hospital financial data obtained from the
Colorado Hospital Association were reviewed for completeness,
I
reasonableness, and uniformity. Of the 3,726 records originally
received, only 3,412 were used in the regression analysis (91.6%).
With 3,412 cases analyzed for 65 hospital facilities and 86
independent variables, the cases to IV ratio is 38:1, well above the
!
minimum requirements for regression. In addition to the data
exclusions for hospitals that had recently opened or closed (listed
above), all records containing data which did not meet minimum intra-
i
field (e.g.,inegative values reported for discharges, patient days,
j
expenses, uncompensated care amounts, or patient care revenues) or
j'
inter-field (e.g., no discharges with positive patient days, no
I
patient days;with positive discharges, or payroll expenses in excess
of total) logical tests were deleted. All records with missing data
i ,
I;
in any of the required fields were deleted. For each raw datum
element, the
mean and the standard deviation (Std) were calculated


I
!
i
i
I
i
If a value in any record exceeded 2 STD from the mean, then the
I
record was reviewed to determine if a data problem existed. For the
|
most part, this statistical review for univariate outliers resulted
i
in very few records being deleted. Most record deletions occurred
I
due to logical review for out-of-range criteria. Data were reviewed
j .
by use of scatterplots for abnormal and non-linear relationships.
I
Problems were detected for Medicaid disproportionate share status and
Medicaid prospective payment system dummy variables, as the Medicaid
Program initiated the prospective payment system methodology in 1988.
I
Thus, all records with dates prior to July 1988 were coded as zero
for these two variables.
I
Four raw data elements may represent internal validity
i
problems: the number of interns and residents at a facility, the
!
hospital's adjusted case mix index, the countys average per capita
I
income, and the hospital's licensed bed capacity reported.
1. Colorado adjusted hospital aggregate case mix index was
reported by the Colorado Health Data Commission (CHDC)
for only two six month periods during 1986 to 1990. For
records in 1986, the 1986 case mix index was used. For
records from 1987 to 1990, the 1987 case mix index was
i
used. For hospitals (primarily facilities under 50 beds)
that did not participate in the CHDC report, an estimated
adjusted case mix index of 0.5000 was assigned. (Note:
This estimate was based on the assumption that smaller,
j
rural hospitals would likely have an overall facility
case mix index tending towards the lower reported
i
values.) The assumption was made that hospital case mix
i
index remained relatively constant over the period from
i
1987 to 1990. Given the reported intra-hospital case mix
!
i
i
49


I I
changes that occurred between 1986 and 1987, this
assumption may represent a weakness in this analysis.
2. The Department of Local Affairs had not published data
i
for per capita income rates for periods after 1988.
Although the statewide average per capita income
increased from $16,463 in 1988 to $17,494 in 1989, the
changes in per capita income were likely to vary
substantially at the local level. Based on the
recommendation of the Department of Local Affairs staff,
the 1988 per capita income figures were used for 1989 and
1990 hospital records.
3. The majority of hospitals did not finance positions for
interns and residents across all time periods. Thus, the
data for this continuous value field were non-zero only
I
in 16% of all records (563 records out of 3412). This
I
type of distribution was not typical for most of the
continuous variables in the analysis. This variable was
transformed to a dichotomous variable as a result of
I !
scatterplot data analysis (Tabachnick and Fidell, 1989).
4. The licensed bed capacity, as reported by the State
Department of Health, for the majority of hospitals did
hot change substantially during the study period. The
average length of stay at hospitals, however, changed
dramatically. Hospitals, traditionally, have not chosen
to reduce the number of licensed beds. According to CHA
(1989), many hospitals have converted inpatient care
space to other uses. Licensed beds may have no
relationship to the number of staffed or maintained beds.
This may be due, in part, to a perception that the
previous licensed bed capacity may be difficult to
50


I
II
regain. The resul-bs of -this study may be weakened by
this reporting deficiency. This phenomenon represents a
i
licensed paper bed syndrome. Thus, occupancy rates m
this analysis may be understated.
i
I
I
I
51


CHAPTER 5
RESULTS AND DISCUSSION
i
Characteristics of Study Hospitals
The unit of analysis, for purposes of this study, is a
hospital's monthly financial report. The characteristics of the
hospital financial records used in the study, therefore, are
important inj analyzing the regression model results. The 3,412
I
records for hospital monthly financial data across the 65 hospitals
had the following general characteristics:
1. Ownership:
la!. For-profit 3.3%
Is. Non-profit 54.1%
c. Government 42.6%
I .
2. Medicaid Peer Group Status
a'. Denver 19.5%
. Other Urban 21.2%
'!cj. Rural 59.3%
3. Average Licensed Acute Care Beds 168 beds
i
I'
4. Average Licensed Nursing Home Beds 8 beds
5. Percentage of Records with Swing Beds 38%
6. Medicare Peer Group Status
ja. Large Urban 21.2%
b. Other Urban 16.3%
jc. Rural 27.8%
,d. Sole Community Hospital 33.0%
|e. Rural Referral Center 1.7%
7. Percentage of Government Payor Discharges
ja. Medicare 32.1%
b. Medicaid 7.3%
[c. Colorado Indigent Care Program 3.8%
d. Total Government (d=a+b+c) 43.2%


8.
Average Patient Service Operating Margin -10.1%9
The definitions, descriptive statistics, and sources used for
variables included in this study are listed in Tables 5.5 and 5.6.
i
The dependent variables included in this study are summarized cross-
i
sectionally and longitudinally in Tables 5.7 and 5.8, respectively.
j
! Primary Hypothesis
!
i
The following table (Table 5.1) presents the estimated linear
I
regression coefficient and significance levels for only the
significant 'independent variables (p < 0.05) associated with hospital
j
patient service operating margin (PSOM). Dummy variables for each
hospital were introduced to account simultaneously for the
I
differences between hospitals. Analysis was performed using SAS
Regression (PROC RR6), with assistance from SAS Univariate (PROC
UNIVARIATE) and Correlation (PROC CORR) Procedures for evaluation of
assumptions ^and plotting means. Although the univariate procedure
I
indicated sk'ewness for several variables, no transformations (other
i
than for teaching status) were used to reduce skewness or to induce
I !
normality. Transformations were not used because of the large sample
!
size collected for these analyses. No additional outlier analysis,
beyond the djata screening techniques used, was applied. No cases had
missing variables.
H(l); Government-sponsored health care programs (Medicare, Medicaid,
and th'e Colorado Indigent Care Program) are negatively
associated with Colorado hospital profitability. Therefore, it
is postulated that increases in government program utilization
and/or decreases in government program reimbursement rates will
be associated with decreases in Colorado hospital
profitability.
j
i
9 jpsOM is not calculated using the CHA formula. For
further explanation, see page 39.


Table 5.1. H( 1) : Univariate and Multiple Linear Regression Analysis
Results
H(l) : Multiple Regression Analysis on PSOM
! PSOM Mean -0.1014991
PSOM Std 0.3252746
Univariate Analysis Linear Regression
Variable Name Mean Std Beta Prob >JT[
MED RATE 2862.73 686.80 0.000129 0.0001
ALOS 4.52298 1.4884 -0.063764 0.0001
IRV TD 4237.49 2178.2 0.000084 0.0001
NPT PCT , 0.08663 0.1160 -0.460597 0.0001
SAL HR 11.9721 2.8967 -0.022166 0.0001
OCC 0.35370 0.1830 1.287525 0.0001
TEACH 0.16501 0.3712 -0.068933 0.0401
MCPIWCUM 1.16633 0.1097 -0.551925 0.0001
UNCP PCT 0.04384 0.0402 -1.301472 0.0001
WINTER 0.24736 0.4315 0.030535 0.0084
Intercept 0.0177
R-squared 0.5220
Adjusted R-sduared 0.5096
F Value 42.221
Probability > F 0.0000
Records 3412
Standardized
Variabl e Estimate
1. OCC 0.73856517
2. IRV TD 0.57434783
3. ALOS -0.29747961
4. MED RATE 0.27747909
5. SAL HR -0.20126171
6. MCPIWCUM -0.18959007
7. NPT PCT -0.16741093
8. UNCP PCT -0.16395002
9. TEACH -0.08021356
10. WINTER | 1 1 0.04130288
This regress ion analysis would indicate that the listed independent
variables arje statistically significant in predicting patient service
operating mJrgin. Thus, patient service operating margin would be
i 1
projected toj decrease as: 1
1. Occupancy rates decrease;
2. jlnpatient charges per discharge decrease;
3. 'Average length of stay increases;
4. 'Medicare payment rate decreases;
5. jsalary expenses per hour increase;
54


6.
9.
10.
!
The consumer price
inflationary index)
index for medical goods (an
increases;
Percentage of total revenues associated with non-patient
care activities increases;
Percentage of patient service revenues dedicated to
uncompensated care increases;
A facility participates in a teaching program; and
The seasons change, as increased PSOM is more likely
during the winter.
The only government program variable determined to be statistically
significantly associated with hospital profitability was the Medicare
payment rate (base payment per discharge). Thus, an association
between Medicare payment level and hospital patient service operating
margin is evident. An association between other government
independent yariables and PSOM was not found.10 Other non-
government program variables appear, however, to be important. Given
I
the absolute! relative influence (larger standardized estimate)
associated with the MED_RATE variable, other independent variables
(such as OCC!, IRV_TD, and ALOS) appear to have a larger impact upon
changes in PSOM. An association between PSOM and other independent
variables (SjAL_HR, MCPIWCUM, NPT_PCT, UNCP_PCT, TEACH, and WINTER)
was also demonstrated.
Secondary Hypotheses
The following tables (Tables 5.2-5.4) present the estimated
linear regression coefficient and significance levels for only the
significant independent variables (p < 0.05) associated with the
listed dependent variables (ADJIRV_D, ADJSALHR, and UNCP_PCT). Once
I
again, dummy variables for each hospital were introduced to account
simultaneous
ly for the differences between hospitals.
10 The failure to find a significant effect of other
government program variables upon PSOM does not indicate conclusively
that an effect is not present. Rather, it indicates that an
association jcould not be established given the defined research
methodology.!
55


H2(l): {Government-sponsored health care program activity is
negatively associated with hospital inpatient gross
revenues (charges) per discharge. Therefore, it is
postulated that increases in government program
utilization and/or decreases in government program
reimbursement rates will be associated with increased
^ross revenues per discharge.
Table 5.2. H(2): Univariate and Multiple Linear Regression Analysis
! Results
H(2): Multiple Regression Analysis on ADJIRV_D
[Inpatient Revenue (Gross Charges) Per Discharge Adjusted by CPI-W]
j ADJIRV_D Mean 3600.96
ADJIRV_D Std 1730.48
i
Univariate Analysis Linear Regression
Variable Nam e Mean Std Beta Prob >|T
ADJMKRAT 1301.44 1280.45 -0.084749 0.0001
MKD PPS 0.45281 0.49784 173.9478 0.0001
ADJMCRAT 2463.72 592.767 -0.857986 0.0001
MI D 0.03762 0.09963 1192.129 0.0003
ALOS 4.52297 1.48840 377.1626 0.0001
NPT PCT 0.08663 0.11596 -903.4867 0.0001
TEACH 0.16501 0.37124 246.8024 0.0032
occ 0.35370 0.18301 -1182.403 0.0001
MC D 0.32109 0.14642 997.0463 0.0001
CADJAPCl 13147.2 3381.51 -0.153479 0.0001
SPRING 0.26905 0.44353 105.4814 0.0002
WINTER 0.24736 0.43154 130.0754 0.0001
Intercept | 9843
R- squared i 0.8987
Adjusted R-squared 0.8962
F Value ! 360.22
Probability |> F 0.0000
Records j 3412
i | Standardized
Variable Estimate
1. ALOS 0.32440037
2. CADJAPCI -0.29991098
3. adjmcra!t -0.29389838
4. OCC -0.12504511
5 . MC D i 0.08436045
6 . MI D ! 0.06863579
7 . ADJMKRAT -0.06270925
8. NPT PCT -0.06054136
9. TEACH 1 0.05294654
10 . MKD PPS 0.05004297
11 . WINTER 0.03243782
12 . SPRING ' 0.02703542
56


j
This secondary regression analysis for H(2) would indicate that the
!
listed independent variables are statistically significant in
predicting inpatient gross revenue per discharge (adjusted for the
medical CPI-^). Thus, adjusted inpatient revenue (gross charges) per
discharge would be projected to increase as:
1. Average length of stay increases;
2. The average per capita income in the area decreases;
3. Medicare payment rate, adjusted for the medical CPI-W,
decreases;
4. Occupancy rate decreases;
5. Percent of medicare patient discharges increases;
6. Percent of medically indigent discharges increases;
7. Medicaid payment rate, adjusted for the medical CPI-W,
Idecreases;
8. .The percentage of revenues generated by non-patient
services decreases;
9. !a facility participates in a teaching program;
10. Medicaid payment methodology switched from a per diem
{payment to a prospective payment methodology; and
11. iThe seasons change, increased inpatient revenues per
discharge are more likely for spring and winter.
Several government program variables were determined to be
statistically significantly associated with inpatient revenues per
discharge, lit appears likely that hospitals did raise their charges
i
in response |to government program increased utilization and reduced
I
payment levels. In other words, a direct negative association
between government-sponsored health care program activity and
hospital inpatient gross revenues per discharge is evident. Again,
I
other non-government program variables are also clearly important.
i
i
H3(1):
Government-sponsored health care program activity is
positively associated with hospital salary expenses per
hour. Therefore, it is postulated that increases in
government program utilization rates and/or decreases in
government program reimbursement rates will be associated
'with decreased payroll expenses per hour.
i
i
!
i
i
i
i
57
I


Table 5.3. H(3)s Univariate and Multiple Linear Regression Analysis
! Results
H(3): Multiple Regression Analysis on ADJSALHR
(Salary Per Hour Adjusted by CPI-W)
i 1
I ADJSALHR Mean 10.2572
ADJSALHR Std 2.2538
I
Univariate Analysis Linear Regression
Variable Name Mean Std Beta Prob >|T
MKD DPS > ' 0.14713 0.35429 0.458180 0.0001
MKD PPS | ' 0.45281 0.49784 0.402404 0.0001
ADJMCRAT j , 2463.72 592.767 0.000373 0.0050
MED DSP i 0.08910 0.28493 -0.836537 0.0001
ADJMIRAT | 556.845 968.178 0.000077 0.0111
MC D | 0.32109 0.14642 0.524200 0.0333
CADJAPCI ! 13147.3 3381.51 0.000159 0.0001
SPRING i. 0.26905 0.44353 0.264808 0.0001
WINTER 1 0.24736 0.43154 0.202427 0.0002
Intercept | 7.9235
R- squared . 0.7812
Adjusted R-squared 0.7758
F Value | 144.94
Probability > F 0.0000
Records i 3412
1 Standardized
Variable Estimate
1. CADJAPCI 0.23821463
2. MED DSP -0.10575456
3. ADJMCRAT 0.09822169
4. MKD PPS: 0.08888645
5. MKD DPS| 0.07202295
6. SPRING 0.05211203
7 . WINTER 0.03875919
8. MC D , j 0.03405410
9. ADJMIRA T 0.03330011
i
This secondary regression analysis for H(3) would indicate that the
listed independent variables are statistically significant in
predicting salary expense per hour (adjusted for the medical CPI-W).
Thus, adjusted salary expense per hour would be projected to increase
as s i
1
2
3
The average per capita income in the county increases;
Medicare disproportionate share decreases;
Medicare payment rate, adjusted for the medical CPI-W,
increases;
11
i!
58


I
i
4. Medicaid payment methodology switched from a per diem
payment to a prospective payment methodology;
5. Medicaid disproportionate share increases;
6. The seasons change, increased salary expense per hour are
more likely for spring and winter;
7. Percent of medicare discharges increases; and
8. Medically indigent payment per discharge, adjusted for
|the medical CPI-W, increases.
i
Several government program variables were determined to be
statistically significantly associated with salary expense per hour.
The impact of increases in government program utilization upon salary
i
expenses per! hour appears to be mixed.11 This may be partially due
to the shortj time frame of study, where salary expenses may not
adjust simultaneous to substantial changes in volume. It appears
i
likely that {hospitals did, in fact, lower their cost in response to
reduced payment levels by government programs. In other words, a
direct positive association between government-sponsored health care
program activity and hospital salary expense per hour is evident.
Again, otherj non-government program variables are also clearly
important, j
H4(l): [Government-sponsored health care program activity is
positively associated with the percent of gross patient
{revenues dedicated to uncompensated care. Therefore, it
jis postulated that increases in government utilization
Irates and/or decreases in government program
'reimbursement rates will be associated with decreased
levels of uncompensated care.
i;
11 jThe relationship between Medicare disproportionate share
status and Medicaid disproportionate share status was investigated
further to attempt to explain the seemingly conflicting results. For
the period from January 1986 to July 1988, the Medicaid Program
reimbursed hospitals based on a per diem. Thus, Medicaid
disproportionate share was initiated during the middle of the study.
Additionally several providers were grandfathered into the Medicare
disproportionate share system due to historical Medicare agreements.
Finally, thejcriteria to qualify as a disproportionate share facility
for Medicare 1 and Medicaid are substantively different.
!; 59
i1
[


Table 5.4.
H(4): Univariate and Multiple Linear Regression Analysis
! Results
H H4): Multiple Regression Analysis on UNCP_PCT
l UNCP PCT Mean 0.043839
i' UNCP_PCT Std 0.040189
1 I: i Univariate Analysis Linear Regression
11 Variable Name Mean Std Beta Prob >|T|
MKD DPS 1 1 0.147127 0.35429 0.005616 0.0394
TEACH 0.165006 0.37124 -0.009680 0.0494
CADJAPCI j; 13147 .27 3381.51 -0.000003 0.0184
Intercept 0.0980
R-squared ' 0.3498
Adjusted R-squared 0.3336
F Value I , 21.573
Probability j> F 0.0001
Records ' 3412
1 . Standardized
Variable Estimate
1. CADJAPCI -0.26774262
2. TEACH 1 , -0.08941395
3. MKD DPS i 0.04950717
1 .
I 1
This secondary regression analysis for H(4) would indicate that
uncompensated care levels would be projected to increase as:
I
1.
2.
3.
The county's adjusted per capita income decreases;
A facility is not associated with a teaching program; and
Medicaid disproportionate share increases.
A negative rather than positive association between government-
sponsored health care programs and the provision of uncompensated
' i
hospital car^ 1S evident. In other words, the more low-income
I '
government-sponsored patients that a facility serves, the facility
will likely
I
provide greater levels of uncompensated care.
Discussion
I '
i i
j |
All of | the results of this study are applicable only to
!'!
Colorado hospitals for fiscal reporting periods from January 1986 to
II
I
August 1990 .J As discussed earlier, this study controls for I
I .
i
i 60
I


individual hospital characteristics and focuses on differences
between hospital facilities over time. Not surprisingly, many of the
| '
relationships (p < 0.05) established by this analysis are consistent
!1
with general| health care management theory.
The CHA assertions, that the future financial viability of the
Colorado hospital industry is at risk, have been confirmed by this
study but nojt necessarily for the reasons CHA cited. The hospital
industry's pjrimary business of providing patient care appears
precarious, j ;However, there is hope for the future. Rather than
being largely dependent on exogenous government programs, the patient
service operating margin appears, to a large degree, to be
controllable^ 'by choices made by hospital management. Focused changes
in the management variables (which decrease average length of stay,
increase inpatient charges per discharge, decrease salary expense per
hour, decreajse the proportion of non-emergent uncompensated care
provided, decrease funding for interns and residents, and increase
occupancy ra!te) would be predicted by this model to increase the
patient service operating margin. The control of management over
these factors, however, may be compromised to a large degree by the
i
hospital's mission, the marketplace, the political climate, and the
regulatory environment.
i [
Several findings, however, were not expected. The percentage
of total revenues associated with non-patient care activities
I
I
(NPT PCT) appears to be a means to cross-subsidize a hospital's
II
decreasing patient service profitability. This finding would suggest
|m
that non-patient service activities (e.g., physician office buildings
i
and sick chilld day care services) may be financially complementary to
i '
the patient jservices provided. Additionally, non-patient service
revenue includes tax revenues from cities, counties, or special
hospital districts, which may be used to directly subsidize the
I
61
i


hospital's operation. In further examination of the detailed data on
|
NPT_PCT (regressing the independent variable set against NPT_PCT as
i
i
the dependent variable), the percentage of total revenues associated
with non-pat|ient care activities appeared to increase after the
i
Medicaid program initiated the prospective payment system
i
methodology. Conversely, the detailed analysis of hospital data
appears also^ to indicate that as hospital patient profitability
increases, tjhe hospital is partially redirecting this profit toward
investments jnot directly related to patient care services. These
|
investments may include vertical integration strategies (e.g.,
I
developing a[ durable medical equipment supply purchasing consortium),
which are complementary to the patient care services provided. Thus,
!
i
the traditional economic theory of a non-profit hospital's desire to
i
maximize quantity of care provided (by increasing the level of
j
charity carej provided) may not be entirely accurate in today's
competitive jmarketplace. Augmenting traditional patient care
services mayj be, in fact, important for hospital management
consideration.
j,
Secondly, the findings of the linear regression model support
I
the concept |that hospitals, as any other business, are subject to
changes in tjhe general market condition and seasonal variations in
the business cycle. Not surprisingly, the consumer price index for
j
medical goods (an inflationary index) was demonstrated to have an
i
inverse relationship to patient service operating margin.
The only government-sponsored health care program variable
found to be {associated with patient service profitability was the
I
Medicare payment rate per discharge. In contrast with current
industry perceptions, the utilization of government-sponsored payors


i I
(payor mix) yias not found in this research study to be an important
factor in determining patient service profitability.12
Influences on H(2): Revenues Per Discharge, H(3): Salary Per Hour,
and H(4): Uncompensated Care Provided
i1
The findings of the secondary hypotheses are also fairly
I 1
consistent with general health care management theory:
H(2): jcross-Subsidization of Care: Charges per discharge
H(2)
H(2)
H(2)
appear to be associated with hospitals explicitly cross-
j
'subsidizing government patient care. This cross-
I
i
jsubsidization appears to be linked to both decreases in
jgovernment payment levels and increased levels of
utilization by government patients.
Increased Charges Associated with Empty Beds: As
occupancy rates drop, the charges per discharge appear to
be increased. This is likely due to the need for
i
hospital management to allocate the fixed costs of
[operation over a smaller base of activity.
Changes in Practice Patterns: Charges per discharge are
;(by definition) directly related to length of stay, due
to additional daily charges accumulating. Thus, as
[average length of stay declines, the charges per
[discharge would be projected to decrease correspondingly.
'Teaching Affiliation: A facility's choice to fund
interns and residents is associated with increased
i 1
[charges per discharge. This may be due to the fact that
[interns and residents are normally part of an educational
i ,
12 For comparison purposes, all government program
utilization iwas combined to determine if, in aggregate, government
programs reduced hospital profitability. This supplementary analysis
again did not show an association between government program payor
mix and hospital profitability. For further discussion, see
Appendix 3. 1
63
I
i
i


jejnvironment, where researching a patient's condition
'comprehensively (with additional tests and procedures) is
jaicceptable medical practice.
H(2): Financial Risk Shifts to Provider: As Medicaid payment
I
methodology switched from a per diem payment to a
prospective payment methodology, hospital charges per
discharge appear to have increased.
I
H(2): {Stabilization of Hospital Profits: As charges per
discharge decrease, the percentage of revenues generated
by non-patient services appears to increase. This may be
{due, in part, to local governmental subsidies to offset
Ipatient care service losses. Conversely, as patient
jservice profits increase, this margin appears to be
partially redirected toward non-patient care activities.
I
Although a causal relationship cannot be established,
jhospital charges for patient care appear to be inversely
jrelated to the non-patient care revenues available.
H(3): Cost Containment Initiatives: As government payment
i i
H(3)
!levels decrease, salary expenses per hour decrease,
i (This may be partially explained by reduced staff to
{patient ratios and changes in the mix of staff used.)
The Nursing Shortage and Changes in Nursing Staffing
patterns: The Medicaid payment methodology and the
!
medicaid disproportionate share designation appear to be
! '
'linked to the changes in salary per hour. However, the
^change to a Medicaid prospective payment system and
|initiation of disproportionate share payments to Medicaid
jproviders occurred in July 1988, concurrent with the
Colorado nursing shortage and conscious changes by many
I 1
'providers to move towards an all registered nurse


H(4);
H(4):
staffing model. Thus, it is speculated that the nursing
shortage and changes in the registered nurse/licensed
practical nurse mix, not the Medicaid payment method, are
:
likely associated with increased salary expenses per hour
(University of Colorado Health Sciences Center Task Force
on Nursing, 1989).
Medicaid Disproportionate Share Designation: Designation
jas a Medicaid disproportionate share facility is
jassociated with increased levels of uncompensated care.
This may be due to the fact that, by definition, the
facility serves a disproportionate percentage of low
income patients.
Teaching Affiliation: In contrast to the general
findings in the literature, teaching status appears to be
associated with lower levels of uncompensated care. In
Colorado, this may be partially explained by the
existence of the Colorado Indigent Care Programs (CICP).
, i
CICP reimburses the teaching hospitals (e.g., Denver
General and University Hospital) at significantly higher
rates (approximately 2.5 times higher on average over the
period of study) than other hospital facilities.
Additionally, teaching facilities participating in CICP
have approximately 6.8 times the medically indigent
patient volume, on average, than non-teaching facilities.
Thus, teaching facilities may report lower levels of bad
debt and charity care due to the offsetting CICP direct
governmental subsidy. Additionally, offsetting Medicare
payments for indirect and direct medical education costs
may be partially responsible for this unusual result.
65


H(2,3,&4): Local Economic Conditions: As the average per capita
income in the area increases, it appears that hospital
charges per discharge increase, salary expenses per hour
increase, and the level of uncompensated care decreases.
For example, this finding is consistent with the
I
literature which has demonstrated that income level and
'insurance status are directly related. Thus, the
i '
hospital industry is clearly subject to variations in
local economic conditions.
In summary, these findings indicate that hospital management
has signific
maximize pro
focusing on
ant control over patient service profitability. To
fitability, hospital administration should consider
increasing occupancy levels (for example, using creative
strategies such as regional physician recruitment and inter-hospital
cooperative
Arrangements), limiting the level of non-emergent
uncompensated care provided, decreasing funding for teaching
activities, increasing non-patient care service product lines with
potential high return on investment, and decreasing average length of
stay (for example, in psychiatric care developing outplacement
alternatives
or using new technological advances to discharge
patients soonler). As discussed earlier, the control of management
over these factors may be significantly constrained by the hospital's
!
mission, the jmarketplace, the political climate, and the regulatory
environment.
The following two tables attempt to further clarify the
definitions of the dependent and independent variables of study
(Tables 5.5 and 5.6 respectively). The final two tables are included
to demonstrate cross-sectional and longitudinal trends in the
dependent variables of study (Tables 5.7 and 5.8 respectively)
66


Table 5.5
Names, Definitions, and Descriptive Statistics for the
Dependent Variables Used in the Analysis
Name Definition Mean ! Std Source
PSOM Patient Service Operating Margin 0.1014991 0.3252746 CHA-CALC
ADJIRV_D Inpatient Revenue Per Discharge Adjusted by the Medical CPI-W 3600.96 1730.48 CHA+DOL-CALC
ADJSALHR Salary Expense Per Paid Hour Adjusted by the Medical CPI-W 10.2572179 2.2538129 CHA+DOL-CALC
UNCP_PCT Percentage of Patient Service Revenues Dedicated to Uncompensated Care 0.0438391 0.0401894 CHA-CALC
Records = 3412
Source Kev . i : CHA Colorado Hospital Association, Data Bank DOL Regional Office, Department of Labor CALC Calculated variable, based on raw data from source(s) listed
67
i


Table 5.6. Names, Definitions, and Descriptive Statistics for the
Independent Variables Used in the Analysis
Name Definition Mean Std Source
MCPIWCUM Cumulative CPI-W for .Medical. Care 11663335 0.1095898 DOL-CALC
Goods, Where January 1986 =100%
CADJAPCI County Average Per Capita Income Adjusted by the Medical CPI-W 13147.27 3381.51 LA+DOL-CALC
SPRING Dummy Variable = 1 for March, April, and May 0.2690504 0.4435312 CHA-CALC
SUMMER Dummy Variable = 1 for June, July, and August 0.2675850 0.4427649 CHA-CALC
WINTER Dummy Variable = 1 for December, January, and February 0.2473623 0.4315423 CHA-CALC
TEACH Teaching Status Dummy =1 If Hospital Funds Interns and Residents 0.1650060 0.3712400 HSC
occ Occupancy Rate 0.3537010 0.1830071 CHA+DOH-CALC
ALOS Average Length of Stay 4.5229765 1.4883998 CHA-CALC
CMI Colorado Adjusted Hospital Case Mix Index 0.7589143 0.2556960 CHDC
IRVJED Inpatient Revenues Per Discharge 4237.49 2178.20 CHA-CALC
NPT_PCT Non-patient Revenues as a Percent of Total Gross Revenues 0.0866299 0.1159571 CHA-CALC


Table 5.6. (contd.)
Name Definition
Mean
Std
Source
SAL_HR Salary and Benefit Expenses Per Paid Hour 11.9720799 2.8966831 CHA-CALC
UNCP_PCT Uncompensated Care Revenues as a Percent of Gross Patient Revenues (Note: Includes Bad Debt and Charity Care) 0.0438391 0.0401894 CHA-CALC
MC_D Medicare Discharges as a Percent of Total Discharges 0.3210949 0.1464166 CHA-CALC
MED_RATE Medicare Base Payment Rate Per Discharge (for a DRG = 1.0000) 2862.73 686.81 FA
MED_DSP Medicare Disproportionate Share Dummy Variable (Dummy = 1 If the Hospital Receives Disproportionate Share Payments) 0.0890973 0.2849259 FA
MK_D Medicaid Discharges as a Percent of Total 0.0732306 0.0617867 CHA-CALC
MKDPPS Medicaid Payment Method Dummy Variable Prospective Payment = 1 0.4528136 0.4978414 DSS
MKD_DPS Medicaid Disproportionate Share Dummy Variable (Dummy = 1 If the Hospital Receives Disproportionate Share Payments) 0.1471278 0.3542852 DSS
MKD_RATE Medicaid Base Payment Rate Per Discharge (for a DRG = 1.0000 After PPS) 1515.44 1421.01 DSS+CHA-CALC


Table 5.6. (contd.)
Name Definition
Mean
Std
Source
MI_D Medically Indigent Discharges as a Percent of Total Discharges 0.0376236 0.0996309 HSC+CHA-CALC
MI_RATE Medically Indigent Payment Per Discharge Rate 647.32 1102.64 HSC-CALC
ADJMCRAT Medicare Base Payment Rate Discharge Adjusted for the Medical CPI-W Per 2463.72 592.77 FA+DOL-CALC
ADJMKRAT CALC Medicaid Base Payment Rate Discharge Adjusted for the Medical CPI-W Per 1301.44 1280.45 DSS+CHA+DOL-
ADJMIRAT Medically Indigent Payment Per Discharge Adjusted for Rate the 556.85 968.18 HSC+DOL-CALC
Medical CPX-W
Records = 3412
Source Key
LA Colorado Department of Local Affairs
DOL Regional Office, Department of Labor
CHA Colorado Hospital Association, Data Bank
HSC University of Colorado, Health Sciences Center
DOH Colorado Department of Health
CHDC Colorado Health Data Commission
FA Fiscal Agent, Medicare Program
DSS Colorado Department of Social Services
CALC Calculated variable, based on raw data from source(s) listed


0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
i
5:7. Cross-Sectional Analysis of Dependent Variables
I of Study
PSOM ADJIRV D ADJSALHR UNCP PCT
0.11 7191.18 12.90 0.03
-0.04 3402.58 8.75 0.05
-0.11 4079.10 13.04 0.04
0.05 3690.60 11.80 0.03
-0.57 2568.06 6.66 0.01
-0.31 3016.71 10.76 0.04
0.02 3440.13 10.80 0.03
-0.26 2152.76 7.64 0.05
-0.05 2777.15 9.17 0.05
-0.13 5443.66 13.29 0.16
-0.03 2977.67 9.32 0.03
-0.21 1942.50 9.06 0.03
0.00 3014.60 9.29 0.05
-0.04 2326.93 10.83 0.05
-0.22 2319.79 6.75 0.03
-0.22 2857.77 6.70 0.06
0.17 5153.75 11.99 0.04
-0.21 1768.75 7.24 0.04
-0.60 2484.88 10.64 0.03
-0.20 3015.09 11.34 0.04
-0.05 1740.33 6.22 0.04
0.06 3624.92 11.19 0.05
0.03 4145.93 12.42 0.02
0.03 3153.97 10.55 0.03
-0.19 2571.51 7.68 0.03
-0.06 6128.44 11.08 0.07
-0.05 7066.52 11.85 0.03
-0.03 3516.86 10.70 0.05
-0.01 3272.60 9.87 0.04
-0.03 3723.83 7.48 0.05
0.04 3930.31 11.53 0.04
0.01 5347.05 9.85 0.03
0.01 5398.02 11.93 0.05
-0.79 2617.08 8.93 0.07
0.02 3733.70 10.99 0.05
0.03 5954.98 12.61 0.02
0.07 3710.40 12.74 0.02
-0.07 2091.88 8.85 0.06
-0.77 2367.54 9.84 0.04
-0.00 4854.54 13.23 0.02
0.07 2799.77 10.40 0.04
-0.03 2121.39 8.84 0.05
0.05 2079.77 9.26 0.06
-0.18 1995.87 8.09 0.05
-0.17 1773.03 6.27 0.03
-0.03 3078.47 8.88 0.06
-0.02 6006.32 12.80 0.04
0.16 4989.50 11.70 0.06
-0.22 1795.11 8.79 0.10
0.02 3826.62 13.22 0.02
71


Table 5.7. (contd.)
!
HOSPNO
PSOM ADJIRV D ADJSALHR UNCP PCT I
50 0.02 4231.34 11.06 0.05
51 0.02 5983.46 10.66 0.03
52 | -0.06 3253.68 7.97 0.05
53 ! -0.05 3126.82 9.47 0.05
54 i 0.01 3795.53 9.72 0.03
55 0.02 4830.12 12.74 0.03
56 | -0.02 9538.86 12.65 0.02
57 -0.16 2162.32 10.47 0.05
58 -0.03 6561.69 13.22 0.13
59 1 -0.05 2859.51 11.78 0.04
60 | 0.06 3498.16 13.55 0.04
61 i -0.28 1987.66 9.07 0.04
62 i -0.89 2133.18 8.09 0.02
63 -0.11 1860.51 8.51 0.03
64 ; -0.34 2340.35 8.40 0.03
ALL RECORDS = 3412 ! 0 r-H O 1 3600.96 10.26 0.04
I
72
i
I


Table 5.8.' Longitudinal Analysis of Dependent Variables of Study
PERIOD PSOM ADJIRV_D ADJSALHR UNCP_PCT
0 -0.06 3324.23 10.62 0.04
1 -0.06 3232.30 10.36 0.04
2 -0.09 3077.52 10.47 0.04
3 -0.10 3130.42 10.39 0.04
4 i -0.17 3182.42 10.30 0.04
5 ! -0.13 3097.48 10.08 0.04
6 -0.08 3225.04 9.92 0.04
7 -0.13 3022.08 10.04 0.04
8 -0.12 3224.89 10.08 0.04
9 , -0.10 3119.68 10.08 0.04
10 i i -0.15 3218.45 9.98 0.04
11 > -0.10 3234.66 10.04 0.04
12 1 1 1 -0.01 3391.40 10.05 0.04
13 -0.06 3318.17 10.02 0.04
14 -0.08 3436.32 9.89 0.05
15 -0.10 3315.67 10.11 0.04
16 1 -0.13 3173.79 9.92 0.04
17 1 1 i -0.11 3135.32 9.79 0.04
18 -0.08 3273.93 9.79 0.05
19 i -0.13 3209.27 9.73 0.04
20 -0.14 3432.28 9.73 0.04
21 -0.12 3360.00 9.79 0.04
22 1 -0.18 3459.52 9.68 0.04
23 -0.15 3381.24 9.80 0.04
24 -0.04 3806.68 10.74 0.04
25 -0.04 3669.43 10.81 0.04
26 -0.07 3538.39 10.71 0.04
27 -0.18 3539.01 10.54 0.04
28 -0.16 3548.74 10.46 0.04
29 , -0.13 3646.72 10.40 0.04
30 1 -0.11 3513.29 10.43 0.04
31 -0.14 3535.08 10.28 0.04
32 -0.13 3728.47 10.42 0.05
33 -0.16 3621.88 10.42 0.05
34 1 -0.11 3775.66 10.28 0.04
35 -0.11 3559.70 10.51 0.05
36 1 -0.05 4007.61 10.42 0.05
37 -0.06 3861.22 10.58 0.04
38 , -0.13 3844.14 10.75 0.05
39 -0.13 3895.99 10.77 0.04
40 1 ! -0.06 3937.64 10.41 0.05
41 j -0.12 3816.87 10.51 0.04
42 -0.06 3731.53 10.41 0.05
43 -0.07 3980.28 10.43 0.05
44 -0.06 3950.98 10.55 0.04
45 -0.09 3994.35 10.43 0.04
46 , -0.12 3937.69 10.38 0.05
47 -0.05 4017.43 10.81 0.06
48 -0.08 4288.08 10.23 0.04
49 -0.04 4384.45 10.18 0.05
73
i '


Table 5.8. (contd.)
i
PERIOD 1 \ 1 PSOM ADJIRV_D ADJSALHR UNCP_PCT
1 50 -0.04 4042.41 10.16 0.05
51 1 -0.05 4261.05 10.28 0.05
52 -0.05 4115.96 10.23 0.05
53 -0.13 4033.81 10.13 0.05
54 { -0.05 4166.60 10.00 0.05
55 j -0.02 4119.26 10.04 0.04
ALL RECORDS = 3412 i -0.10 3600.96 10.26 0.04


CHAPTER 6
CONCLUSION
Summary
All of!the results of this study are applicable only to
Colorado hospitals for fiscal reporting periods from January 1986 to
!
August 1990. | The results of this analysis suggest the following
j
associations I may be drawn:
I
1. Government-sponsored health care programs (Medicare,
Medicaid and the Colorado Indigent Care Programs) do not
appear to be a primary force associated with Colorado
hospitals' patient service profitability. No association
between government-sponsored health care program
utilization and hospital profitability (given the
research methodology defined) was documented.
Environmental factors (e.g., the rate of inflation for
medical care goods and seasonal fluctuations in business
i
cycle) and managerial factors (e.g., level of
uncompensated care provided, charges per discharge,
j
occupancy rate, salary expenses per hour, non-patient
care activities and average length of stay) represent
statistically significant factors associated with
hospital profitability. The Medicare payment rate was
tjhe only government program variable found to be
associated with profitability. The strongest predictor
I
alf hospital profitability was occupancy rate.
j
2. Government-sponsored health care programs (Medicare,
Medicaid and the Colorado Indigent Care Programs) are


I
negatively associated with inpatient hospital gross
revenues per discharge. Associations were demonstrated
which suggest that hospitals have raised their charges in
i
response to decreasing government program payment levels
i
and to increasing government patient utilization. Based
I
on the associations demonstrated in the primary
i
hypothesis, the hospitals' rate structure changes (to
cross-subsidize care to government patients) appear to
!
have indirectly influenced hospital patient service
i
profitability. Environmental factors (e.g., the local
per capita income and seasonal fluctuations in the
l
business cycle), managerial factors (e.g., average length
l
of stay, occupancy rate, and non-patient care
activities), and structural factors (e.g., teaching
status) are also statistically significant factors which
influence adjusted average gross charges per discharge.
Government-sponsored health care programs (Medicare,
Medicaid, and the Colorado Indigent Care Program) appear
I
to be positively associated with hospital salary expenses
per hour. Associations were demonstrated that
i
interrelate government program payment with salary
!
expenses per hour. It appears that as the payment levels
i
for Medicare and Medically Indigent programs (adjusted
for medical inflation) have decreased, the salary expense
!
paid per hour has decreased. The strongest predictor of
salary expenses per hour was the economic environment
(e.g., local per capita income).
Government-sponsored health care programs (Medicare,
Medicaid, and the Colorado Indigent Care Programs) do not
appear to be primarily associated with the percentage of
76


'gross patient revenues dedicated to uncompensated care.
I
The strongest predictor of a hospital's provision of
uncompensated care (holding constant the hospital's
pnique characteristics) is the local economic
environment. Teaching status and Medicaid
i
disproportionate share status also appeared to be
statistically significant.
i
In spxjte of a dismal industry-wide financial situation, these
results indicate that hospital management has significant control
i
over patientj service profitability. These conclusions are in general
agreement with the findings of Cleverley (1990a) and Burda (1989)
that high prices, excessive costs, and uncontrolled uncompensated
care levels are significantly related to a hospital's financial
performance.! However, other factors (including average length of
!
stay, occupancy rate, general changes in the economic environment,
seasonal variation in hospital business cycles, non-patient care
activities, teaching status, and the Medicare Program payment level)
also were demonstrated as statistically significant factors. To
I.
maximize profitability, hospital administration should worry less
about the financial effect of government programs and focus more on
management variables, such as limiting levels of non-emergent charity
i
care provided, reducing bad debt allowances by increasing collection
r
activities, increasing occupancy levels, increasing non-patient care
service product lines with potential high return on investment, and
decreasing average length of stay.
Revised Model of Hospital Profitability
An explanatory model of Colorado hospital profitability, based
on these results, would include:
;
1. Managerial Factors (Controllable):
i
i
i
i 77
.i
i
i


a. Occupancy Rate
b. Inpatient Charges Per Discharge
c. Average Length of Stay
cl. Salary Expenses Per Hour
e. Non-patient Service Revenues (as a percent of
| total)
f. Level of Uncompensated Care (non-emergent) Provided
2. Government Factor (Non-Controllable):
i
a. Medicare Payment Level
3. Economic Environment (Non-Controllable):
a. Medical Inflation Rate
b. Seasonal Fluctuations in the Business Cycle
i
4. Structural Factor (Controllable):
i
a. Teaching Status
j
The reportedimodel of Colorado hospital profitability would be
configured as shown in Figure 6.1.
Indirect relationship
Direct relationship
Figure 6.1. Revised model of hospital profitability.
i
j
This model, a refinement of the Longo-Chase model, is designed
to display explicitly the relative influence of the four factors
studied. Direct relationships documented are shown with a solid
line. Indirect relationships documented (based on direct
relationships between the secondary hypotheses significant
I
independent variables) to the management factors of study are shown
!
78


with a dotted line [e.g., government program payment rate and
utilization |was shown to be strongly associated with charges per
j
discharge in' H(2)]. Charges per discharge were shown to be related
I
to hospital profitability in H(l). Thus, an indirect relationship
between government factors and profitability is inferred. From this
!
model, it is| apparent that: management factors directly influence
|
hospital profitability, government-sponsored health care programs are
both directly and indirectly associated with profitability, the
i
general economic environment both directly and indirectly impacts
hospital profitability (similar to all other businesses), and
I
i
structural factors are both directly and indirectly associated with
i
hospital financial performance. Both the economic environment and
I
the regulatory environment indirectly impact the hospital management
dependent variables studied.
I
Because the hospital identification variable was held constant
in this analysis (to account for the repeated measures design), many
structural characteristics (e.g., bed size, ownership, and geographic
1
location) could not be included in the analysis. Obviously, a
'i
hospital's profitability may potentially be related to hospital
I
specific characteristics. Outside of teaching status, the importance
of these variables in predicting hospital profitability could not be
i
determined. !To this end, a further refinement of the hospital
I
profitability model (Figure 6.1) would include a consideration of
such structural variables.
The major health care management and policy implications
findings of this study, although based exclusively on Colorado
I
hospital data, do not strictly represent local phenomena. The health
care finance!literature indicates that:
i
1. Occupancy rate is critical: Empty beds are a major
factor in escalating hospital costs ("Hospitals' Inside
79


Track," 1989). Government programs (both Medicare and
|
Medicaid) have initiated payment reductions for capital
I
expenses. Minimum hospital occupancy level standards
have been extensively debated. It has been proposed that
i
hospitals with occupancy rates under the minimum standard
i
be required to go through a state review process
i
(Torkarski, 1989) Lower occupancy rates were also found
I
jto be related to hospital closure (Longo and Chase, 1984;
Cleverly, 1990a).
|
2. Cost-shifting is social policy: Hospital cost-shifting
praising prices to non-government patients in response to
reductions in government payments to hospitals) has been
well documented (Sloan and Becker, 1984; Colorado
Hospital Association, 1988; Dranove, 1988) However, it
I
is anticipated that businesses will not continue to
j
pccept cost-shifting as an implicit social policy, as the
hospital market becomes more competitive (Burda and
|
Torkarski, 1990) .
3. Local economic environment is critical: Geographic
region has been determined to be directly related to
hospital costs (Grannemann and Brown, 1986), levels of
uncompensated care (Saywell et al., 1989) and hospital
I
failure (Longo and Chase, 1984) .
!
4. Payor mix is unrelated: According to research by
Cleverley (1990a), government payor mix did not appear to
contribute to hospital closure.
j
In short, the major findings of this study are congruent with
I
other research findings. Thus, the proposed revised model of
Colorado hospital profitability may have a more generalized
applicability.
80


Management Implications of the Results
The results of this study indicate that hope for the future of
Colorado's hjospital industry exists, as hospital management appears
to be instrumental in influencing the patient service profitability.
Rather than being largely dependent on exogenous government programs,
the patient ^service operating margin appears to be primarily
controlled by management choices. It is important to consider,
however, the environmental constraints placed upon hospital
management functions. The ability of hospital management to initiate
changes affecting the facility's patient service activities is often
a function Ojf the hospital's mission, the marketplace, the political
climate, andj the regulatory environment. Additionally, hospital
administrators and boards are faced with balancing social goals
(e.g., serving the community's residents regardless of ability to
'!
pay) and institutional goals (e.g., profit maximization).
I
In very real terms, hospital administrators must become expert
I
in analyzing the costs and benefits of their management decisions.
High prices,
excessive costs, and uncontrolled uncompensated care
levels have been demonstrated, in both this study as well as the
health policy literature, as critical factors in determining patient
service profitability (Cleverly, 1990a). This study suggests that
i
management attention should also be placed upon critically low
occupancy rates, long lengths of stay, poor billing and collection
i
I
activity (e.g., bad debt), and vertical/horizontal product
i
integration strategies (e.g., non-patient care revenue sources).
j
Many management approaches to the problem of providing a
quality product at a competitive price in a productive manner exist
i
(Shorten ancl Kaluzny, 1983) Deming (1982 and 1986) developed a
i
series of quality management principles that focus on the customer
i
(e.g., the hospital patient). Deming incorporates the concepts of
81


goal setting, senior management commitment, work systems improvement,
data-based decision making, and continuous communication. These
principles, jalthough applicable to many business settings, may be
particularly! useful to hospital management:
!
1. |6oal setting can be used to establish service level
projections and to make changes to the existing hospital
I
services provided. For example, non-patient service
product lines with potentially high return on investment
|(e.g., home health care services) should be considered to
]
augment patient care services.
2. Senior management commitment is essential to facilitating
^hange. Cooperative efforts (i.e., team work) are
essential in a complex, dynamic health care environment.
Implementation strategies with senior management
!
assignments are the first step in developing a work plan.
This quality management principle may be useful in
addressing length of stay problems (e.g., patient
I
discharge planning).
3. Work systems improvement is clearly applicable to billing
!
and collection activities within a hospital. Using
i
management engineering techniques to address productivity
l
and timeliness problems within hospital billing offices
may result in reductions to bad debt levels and, thus,
significant improvement in the operating margin.
4. Data-based decision making, using statistical techniques,
has applicability to hospital staffing problems. Given
assumptions of patient service workload, the hospital's
i
staffing patterns (for both management, support service,
i.
and patient care staff) can be designed at minimum cost.
Additionally, data-based decision making is applicable to
82


occupancy decisions. Before initiating attempts to fill
{existing empty beds, hospital management should determine
[the marginal costs associated with increasing patient
(activity. The marginal costs of increased activity
i
should be compared to the marginal revenues to be gained
|
by increasing occupancy. To the degree that the marginal
(costs are covered and a contribution toward overhead
1
costs is achieved, hospital management should be willing
(to negotiate volume discounts with third party insurers.
Lastly, continuous communication is an important
management principle for hospital management to use in
^addressing patient service profitability issues. For
(example, a hospital may choose to limit the inappropriate
vise of its emergency room by the local medically indigent
Toward this goal, the hospital may choose to
!'
population
design an incentive system, whereby medically indigent
persons using emergency room services inappropriately
I
(will be fully responsible for payment of their bill.
Implementation of this change will be critically affected
by the hospital management's ability to communicate this
new payment policy to: the hospital board, the
j
{community, the medically indigent population, the
|
emergency room staff, the hospital's medical staff, and
jthe billing office.
Quality management principles may be used as a management tactic to
address the patient service profitability problem. In spite of the
{
Colorado hospital industry's precarious financial situation, the hope
for the future resides primarily in Colorado hospital leaders.
83


Policy Implications of the Results
The Colorado Hospital Association has warned that the financial
I
condition of; the Colorado hospital industry is progressively
!
deteriorating. Hospital services are the core of the existing
Colorado health care delivery system. Government-sponsored health
care programs represent a large and growing portion of the hospital
I
industry's revenues. Thus, a widespread policy concern exists that
!
government programs adversely impact the financial position of the
Colorado hospital industry, thereby limiting citizen access to
i
hospital services. One policy objective of the Colorado State
legislature,| therefore, should be to assure that minimum access
levels to hojspital services are maintained in the future.
I
Recently, the CHA asserted that the underfunding of government-
i
sponsored cajre programs has led to a "hidden tax" (Colorado Hospital
Association,! 1990) Government programs for hospitals (Medicare and
;
Medicaid) do not pay the full charges associated with care rendered
to beneficiaries. CHA claims that this payment shortfall is the
i
direct cause1 of inflated charges, forcing private business, insurers,
i
and patientsj to cross-subsidize the care rendered to government-
sponsored patients. In these cases, government-sponsored health care
programs are perceived to be threatening the Colorado hospital
industry's profitability, due to this "hidden tax to non-government
payors. The| results of this study appear to support the CHA
assertion that government programs have negatively affected hospital
i
gross charges. This study demonstrates that this association
(between government payment levels and gross charges per discharge)
!
appears to influence indirectly hospital patient service
profitability, but the relative influence of government programs on
PSOM appears to be less than CHA asserts.


The future ability of hospitals to continue to cost-shift to
non-government program payors, in an increasingly competitive market,
is dubious. A growing trend by all third party payors to
disassociate payments from charges is evident. Future attempts by
hospitals to cross-subsidize care by raising charges will likely have
only a residual influence on profitability.
The Medicare payment rate (Medicare is the largest of the
government-sponsored health care payors) is the only government
program variable which was found to be statistically significant in
directly predicting profitability. The relatively low beta
coefficient of this factor indicates that it is not primary in
importance. Other management factors were more influential in
predicting hospital patient service profitability.
The Medicare Program, a federal program, has initiated major
payment reforms in an attempt to moderate the growth in hospital
expenditures. The effect of these payment reforms is directly
evidenced by changes in hospital daily operations. The Medicare
prospective payment system provides incentives to: (1) ensure that
|
all admissions are medically necessary, (2) minimize the cost of
resources used to provide patient care, (3) minimize the length of
stay, and (4|) ensure that discharge planning (perhaps to alternative
care settings) is expedited. These Medicare payment reform
incentives liave been demonstrated to impact indirectly profitability,
by means of changes to inpatient charges per discharge and salary
expenses per; hour. Thus, these results indicate that Colorado
i.
hospitals hajve responded to a new reimbursement environment (in an
attempt to stabilize profitability) by raising charges and reducing
salary and benefit costs.
The Medicare Program has established a precedent for other
government programs, as well as third party payors. The results of
85


this study do not demonstrate a relationship between either the
!
Colorado Medicaid Program or the Colorado Indigent Care Programs with
i
hospital patient service operating margin. It may be that this
research study's methodology was not sensitive enough to detect the
impacts of either of these two programs because the current
proportion Cf hospital discharges and billed charges attributed to
I
these two programs is relatively small. Therefore, the relationship
between these program payments and hospital profitability may be
i
difficult to discern in aggregate. On an individual basis, a
i
particular hospital's profitability may be directly impacted by these
programs. In considering industry-wide financial performance, the
Medicare Program's hospital reimbursement policies have been
demonstrated to be primary in importance. The importance of the
j
Medicaid and the Colorado Indigent Care Programs (given the AMI
lawsuit settlement and medically indigent hospital refinancing plans
which have been recently implemented) may grow in the future. Thus,
it is not anticipated that the results of this study (placing primary
importance c>n the Medicare program payment rate) represent an
i
enduring condition, as the Medicaid and Colorado Indigent Care
j
Programs continue to evolve toward becoming a larger portion of
i
Colorado hospitals' net revenue base.
The political, economic, and regulatory environments establish
the framework for hospital management choices. Given that the vast
majority of ^Coloradans believe that access to health care should be
available to all Colorado residents, regardless of ability to pay
(based on the 1988 Colorado Trust Survey), hospitals must be
sensitive to their local economic and political environment. A
hospital (as part of its mission) may choose to respond to its
community's Ineed for charity care, in spite of the implications for
the hospital's financial condition.
I
i I
I
i
1
86


categorized
The results of this study indicate that the proportion of gross
patient service revenues dedicated to uncompensated care is a factor
in determining hospital profitability. In context of the belief that
access to care is a fundamental right, hospital care may be
as a public good. To this end, it becomes the
responsibility of government to ensure access to at least minimum
levels of hospital care. The role of government-sponsored health
care programs is to set a precedent. In developing eligibility,
benefit, and reimbursement policies, government-sponsored programs
must be extremely sensitive to the basic public concern for at least
minimum access to hospital services.
ic debate exists as to whether excess capacity (unfilled
direct relationship to rising health care costs
1989) Low Colorado hospital occupancy rates (on average
A publ
beds) has a
(Torkarski,
35%) indicate that patient beds were vacant, but still minimally
staffed and
maintained. Assuming that the licensed bed capacity
reported reflects a hospital's true inpatient capacity, the results
of this study indicate that occupancy levels are a very influential
i
factor in determining hospital profitability. One current question
faced by govjernment-sponsored payors is whether the costs of unfilled
I '
beds should be a government program responsibility. This study
indicates that unfilled beds (given the extremely low reported
occupancy levels) present a management challenge, not a government-
sponsored hejalth care program payment responsibility. Given the
declining financial position of individual hospitals throughout
Colorado, it appears that additional hospital closures may be
imminent. I|f hospital closures were to be distributed on a
geographically equitable basisa critical assumptionthen access to
care might b!e maintained and overall occupancy levels at remaining
facilities increased. Future hospital closures may be, in fact,
I
87


critical to jassuring the future financial viability of a smaller,
more efficient, hospital market in Colorado.
i
Directing Future Colorado Health Policy Initiatives
From tjhe viewpoint of Colorado citizens, access to hospital
j
services is |Considered an essential public good. Hospital care is
I
not rivalrous by nature; given current low hospital occupancy levels,
!
use by one individual does not preclude use by others. The marginal
social cost |of a hospital bed day is very small; (for the unit of
analysis used in this study, a hospital's monthly financial report) a
significant jportion of hospital costs are fixed costs, such as
I
minimal staff and supplies needed for existing bed capacity, plant
and equipment. Over a longer time frame, management has greater
control and flexibility to reduce expenses. Several management
i
strategies to reduce costs might include: (1) changing the mix of
services provided, (2) changing the staff-to-patient ratios, (3)
replacing higher cost patient care providers (e.g., registered
nurses) witli less expensive substitutes (e.g., licensed practical
i
nurses), and (4) reducing management staff. (See pp. 81-83 on
"Management Implications of the Model.")
The public's need to ensure geographical access to hospital
i
services, regardless of the patient's ability to pay for medical
services, in the face of a financially unstable industry, indicates
i
the presence of traditional market failure. The role of Colorado
government in addressing the hospital market failure should be to
ensure that;at least minimum access to hospital services exists
I
throughout the state. The geographical access problem to hospital
services injColorado may be a symptom of a larger problem, the lack
of a comprehensive healtii care safety net for Colorado residents
i
(Enright, 1987). The gaps in the health care safety net are
88
i


perceived to exist due a variety of factors, including (but not
limited to)
the lack of universal insurance coverage for all Colorado
citizens and the shortage of physician services in rural Colorado.
The choices
complex and
faced by Colorado to address these issues are both
dynamic, as well as not entirely within the control of
the legislature and the state's bureaucracy. The public policy
j
questions thlat must be debated include: (1) at what level is the
Colorado hospital geographical access policy problem best addressed
!
(national, state, or local level), and (2) what policy problem should
be addressed (geographical hospital provider survival or individual
i
access to calre)?
The purpose of this discussion is not to provide a
comprehensive analysis of the national, state and local policy
options that can be used to address Colorado's hospital geographical
access and hjealth care safety net problems. As reported earlier,
this study did not find a primary association between government-
sponsored health care programs and hospital profitability. This
discussion,
therefore, represents an attempt to determine how the
other results of this study may be interpreted in directing future
i
health policy. It is assumed under the new-federalism philosophy
that: (1) state level problems are best resolved by state-initiated
and implemented solutions, (2) less government and bureaucratic
involvement |will lead to more efficient (market-driven) results, and
i
(3) a combination of strategies targeted at both the hospital
industry and the safety net issue will be required.13
In order to focus the Colorado health policy discussion on the
key problem
areas, a search for patterns of hospital performance by
13 This policy discussion framework was chosen based on the
new federalism philosophy, which shifts responsibility for social
change to states, places a greater reliance on market mechanisms, and
targets solutions to policy problems directly (Palmer and Sawhill,
1982). j