Citation
The increasing role of private psychiatric hospitals in Colorado

Material Information

Title:
The increasing role of private psychiatric hospitals in Colorado consequences and policy implications
Creator:
Dickson, Ellen O'Connor
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
vi, 177 leaves : ; 29 cm.

Subjects

Subjects / Keywords:
Psychiatric hospitals -- Colorado ( lcsh )
Hospitals, Proprietary -- Colorado ( lcsh )
Hospitals, Proprietary ( fast )
Psychiatric hospitals ( fast )
Colorado ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Colorado at Denver, 1993. Public administration
Bibliography:
Includes bibliographical references (leaves [158]-177).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Doctor of Philosophy, Public Administration.
General Note:
School of Public Affairs
Statement of Responsibility:
by Ellen O'Connor Dickson.

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:
30656531 ( OCLC )
ocm30656531

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

1HE INCREASING ROLE OF PRIVATE PSYCHIA1RIC HOSPITALS IN COLORADO: CONSEQUENCES AND POLICY IMPLICATIONS by Ellen O'Connor Dickson B.S., University of Minnesota M.S., Texas Woman's University A thesis submitted to the Faculty of the Graduate School of the University of Colorado at Denver in partial fulfillment of the requirements for the degree of Doctor of Philosophy Public Administration 1993

PAGE 2

Dickson, Ellen O'Connor (Ph.D., Public Administration) The Increasing Role of Private Psychiatric Hospitals in Colorado: Consequences and Policy Implications Thesis directed by Professor Franklin James In 1987, Colorado deregulated hospital construction by eliminating the certificate of need review process. Subsequently, the number of private psychiatric hospitals in the state doubled between 1987 and 1989. The availability of a comprehensive, patient-level database in Colorado provided a unique opportunity to examine the effects of a rapid growth in private psychiatric hospital capacity on the mental health care delivery system. The following hypotheses, based on findings in the literature, were tested: Changes in utilization are a function of changes in capacity (Roemer's Law); hospital type is a significant predictor of length of stay; and private psychiatric hospitals will draw privately insured patients from other hospital types. Hypothesis testing was accomplished through application of methods widely used in the social sciences, including interrupted time-series and regression analysis. Findings indicate that utilization was affected by changes in capacity, that hospital type was an important determinant of length of stay,

PAGE 3

CONTENTS CHAPTER 1. INTRODUCTION 1 Issues Raised by Privatization . . . . . . . . . . . . . . 3 Cream Skimming . . . . . . . . . . . . . . . . 5 Curtailed Community Services . . . . . . . . . . . 7 Inappropriate Adolescent Admissions . . . . . . . 8 Supplier-Induced Demand . . . . . . . . . . . . . 10 Differences in Length of Stay . . . . . . . . . . . 12 Purpose of the Study . . . . . . . . . . . . . . . . . 13 Contribution of the Study . . . . . . . . . . . . .. . . . 14 2. PROBLEM CONTEXT . . . . . . . . . . . . . . . . . 17 Mental Health Policy . . . . . . . . . . . . . . 17 The Emerging Federal Role . . . . . . . . . . . . 18 Deinstitutionalization . . . . . . . . . . . . . . . 19 The Privatization Process . . . . . . . . . . . . . . . . 22 Shifts in locus of Care . . . . . . . . . . . . . . 22 Changes in Payer Mix . . . . . . . . . . . . . . 24 Medicare and Medicaid . . . . . . . . . . . 25 Private Insurers . . . . . . . . . . . . . . 27 Deregulating Hospital Construction . . . . . . . . . . . . 28 Consequences of Deregulation . . . . . . . . . . . 29 The Colorado Experience . . . . . . . . . . . . . 31 3. LITERATURE REVIEW .................................. 35 The Evolving Structure . . . . . . . . . . . . . . . . 35 For-Profit Ownership . . . . . . . . . . . . . . . . . 38 Proponents' Perspective . . . . . . . . . . . . . . 38 Opponents' Perspective . . . . . . . . . . . . . . 40 Effects of Ownership and Specialization . . . . . . . . . . 42 Ownership and General Hospitals . . . . . . . . . . 43 Diagnostic Payer Mix Differences . . . . . . . . 46 Inappropriate Admissions . . . . . . . . . . . . . 49 Supplier-Induced Demand ............................... 53 Alternative Models ............................... 54 Roemer's Law . . . . . . . . . . . . . . . . . 56 CON Review .............. . . . . . . . . . . . 58 GAO's Study of Roemer's Law . . . . . . . . . . . 59

PAGE 4

Psychiatric LOS . . . . . . . . . . . . . . . . . . . 62 Hospital Type . . . . . . . . . . . . . . . . . 63 D' IagnOSIS ............................... ..... Payer Groups .................................. Age and Gender ................................ 65 67 69 4. METIIODOLOGY . . . . . . . . . . . . . . . . . . . 71 Study Design . . . . . . . . . . . . . . . . . . . . 73 Data Sources . . . . . . . . . . . . . . . . . . . . 76 CHDC Discharge Data Set . . . . . . . . . . . . 77 Supplementary Data Sources . . . . . . . . . . . . 79 Data Analyses . . . . . . . . . . . . . . . . . . . . 80 Changes in Utilization . . . . . . . . . . . . . . 80 Variables Affecting LOS . . . . . . . . . . . . . 81 Method Used . . . . . . . . . . . . . . . 84 LOS and Independent Variables ................ 84 Limitations . . . . . . . . . . . . . . . . 87 Cream Skimming . . . . . . . . . . . . . . . . 87 5. FINDINGS . . . . . . . . . . . . . . . . . . . . . . 88 Testing Roemer's Law ........... . . . . . . . . . . . 89 Increase in Discharges . . . . . . . . . . . . . . 90 Changes in Patient Days . . . . . . . . . . . . . 94 Decreases in Length of Stay . . . . . . . . . . . . 97 Average Daily Charges ............................ 99 Capacity Changes, Hospital Type, and LOS . . . . . . . . 100 Effects of Hospital Type on WS . . . . . . . . . . 101 The Roemer Effect and LOS . . . . . . . . . . . 102 Patient Characteristics and LOS . . . . . . . . . . 108 Source of Payment . . . . . . . . . . . . 108 Age Groups, Gender, and LOS . . . . . . . . 110 DRGs and LOS . . . . . . . . . . . . . 112 Cream Skimming . . . . . . . . . . . . . . . . . . 114 Payer Mix by Hospital Type . . . . . . . . . . . . 115 Patient Characteristics . . . . . . . . . . . . . . 119 Age .................................... 119 Diagnostic Groups . . . . . . . . . . . . 123 11

PAGE 5

6. CONCLUSIONS : . . . . . . . . . . . . . . . . . . . 126 The Roemer Effect and Psychiatric Treatment . . . . . . . 127 Application of Roemer's Law . . . . . . . . . . . 129 Discharges . . . . . . . . . . . . . . . 130 Length of Stay . . . . . . . . . . . . . . 131 Patient Days . . . . . . . . . . . . . . . 131 Limitations of Roemer's Law . . . . . . . . . . . 132 Physician's Role . . . . . . . . . . . . . 133 Marketing Psychiatric Services . . . . . . . . 134 Utilization in Public Psychiatric Hospitals . . . . 135 Hospital Type and LOS ................. . . . . . . . 137 Implications of LOS Variation by Hospital Type . . . . 137 Cost Impacts . . . . . . . . . . . . . . . . . 139 Implications for Colorado . . . . . . . . . . . . 140 Cream Skimming . . . . . . . . . . . . . . . . . . 141 The Changing Role of Private Psychiatric Hospitals . . . 144 Policy Recommendations .............. . . . . . . 145 APPENDIX A ............................................ 148 Recoded Variables . . . . . . . . . . . . . . . . . 148 APPENDIX B . . . . . . . . . . . . . . . . . . . . . . 150 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . 158 iii

PAGE 6

FIGURES Figure 2.1 Beds Per 100,000 Population, Private Psychiatric and Non-Federal General Hospitals: United States, Selected Years 19701988 . . . . . . . . . . . . 23 Figure 4.1 Hospital Types Defined . . . . . . . . . . . . . . 73 iv

PAGE 7

TABLES Table 1.1 Private Psychiatric Hospitals, U.S. and Colorado . . . . 2 Table 2.1 Beds Available for Psychiatric Treatment by Hospital Type in Colorado, 1987 and 1989 . . . . . . . . . . 33 Table 3.1 Median Days of Stay, Psychiatric Discharges: By Gender, Age, Selected Principal Diagnosis, and Hospital Type, United States, 1986 . . . . . . . . . . . . . . . 64 Table 4.1 Variables Included in Regression Analyses . . . . . . . 83 Table 4.2 Hypothesized Direction of Relationship Between Dependent and Independent Variables . . . . . . . . 86 Table 5.1 Number of Discharges by Hospital Type . . . . . . . 91 Table 5.2 Percentage Psychiatric Discharges by Hospital Type . . . 93 Table 5.3 Number of Patient Days by Hospital Type . . . . . . . 95 Table 5.4 Percentage of Total Patient Days by Hospital Type . . . 96 Table 5.5 Mean Length of Stay by Hospital Type ................ 98 Table 5.6 Average Daily Charges by Hospital Type, 1987 and 1989 . . . . . . . . . . . . . . . . . 100 Table 5.7 Effects of Hospital Type on LOS: All But Public Psychiatric Hospitals . . . . . . . . . . . . . . 102 Table 5.8 Length of Stay for Psychiatric Discharges: Regression Coefficients Compared Across Hospital Types . . . . . 104 Table 5.9 Average Length of Stay for Psychiatric Discharges: By Independent Variable and Hospital Type . . . . . 106 v

PAGE 8

Table 5.10 Colorado Psychiatric Discharges: By Source of Payment and Hospital Type: 1987 and 1989 . . . . . . . . . 118 Table 5.11 Colorado Psychiatric Discharges: By Age Group and Hospital Type: 1987 and 1989. . . . . . . . . . . 122 Table 5.12 Colorado Psychiatric Discharges: By Diagnosis and Hospital Type: 1987 and 1989 . . . . . . . . . . . 125 Table B.1 Regression on LOS, Psychiatric Discharges: All But Public Psychiatric Hospitals . . . . . . . . . . . . . . 150 Table B;2 Regression on LOS, Psychiatric Discharges: Public General Hospitals . . . . . . . . . . . . . . . 152 Table B.3 Regression on LOS, Psychiatric Discharges: Private NonProfit General Hospitals . . . . . . . . . . . . . 153 Table B.4 Regression on LOS, Psychiatric Discharges: Private ForProfit General Hospitals . . . . . . . . . . . . . 154 Table B.5 Regression on LOS, Psychiatric Discharges: Private NonProfit Psychiatric Hospitals . . . . . . . . . . . . 155 Table B.6 Regression on LOS, Psychiatric Discharges: .Private ForProfit Psychiatric Hospitals . . . . . . . . . . . . 156 Table B.7 Regression on LOS, Psychiatric Discharges: Public Psychiatric Hospitals . . . . . . . . . . . . . . 157 VI

PAGE 9

CHAPTER 1 INTRODUCTION In 1987, Colorado deregulated hospital construction by eliminating the requirement that the need for new hospital facilities or major renovations be documented and approved through a certificate-of-need (CON) review process. The elimination of CON review had one major and largely unexamined consequence in Colorado: The number of private psychiatric hospitals more than doubled, increasing from five facilities in 1986 to 11 in 1989 (Rehak, 1990). The ownership structure of the state's private psychiatric hospitals also changed: In 1987, only two of the five private psychiatric hospitals were for profit; by eight of the 11 were investor-owned. The post-deregulation growth of private, predominantly for-profit psychiatric hospitals was a national as well as a Colorado phenomenon. The federal CON review requirement was eliminated in 1983 and federal financing for state CON review agencies was phased out between 1983 and 1986. This was followed by a nationwide doubling in the number of private psychiatric hospitals between 1984 and 1988; nearly all of the new hospitals were for profit facilities (Center for Mental Health Services & National Institute of Mental Health [CMHS & NIMH], 1992). Most of the growth was concentrated in southern or western states, such as Colorado, with little or no state regulation of hospital construction (Dorwart, Schlesinger, Davidson,

PAGE 10

Epstein & Hoover, 1991). By 1991, however, the trend began to reverse: The number of private psychiatric hospitals decreased slightly nationally as well as in Colorado. Table 1.1 illustrates the growth and decline in the number of private psychiatric facilities that occurred. between 1984 and 1991. Table 1.1 Private Psychiatric Hospitals, U.S. and Colorado 1984 1986 1988 1989 1991 United States 220* 314* 444* N/A 420** Colorado*** 4 5 9 11 10 ** *** from Center for Mental Health Services & National Institute of Mental Health, (1992). Mental Health, United States, 1992. from American Hospital Association (1993). Hospital Statistics, 1992-1993. from Health Facilities Division, Colorado Department of Health, (1991). During this period, two psychiatric hospitals closed but one general hospital converted to an alcohol rehabilitation hospital. Although the federal deregulation of hospital construction was a precipitating factor, the growth of private psychiatric hospitals was the result of disparate demographic, social, and public policy changes that gradually shifted the locus of care from public to private providers. The privatization of inpatient psychiatric care began more than 25 years ago with the trend to deinstitionalize the chronically mentally ill (Pardes, 1990). Other factors such 2

PAGE 11

as demographic changes, wider recognition of the need for mental health services, greater public acceptance of the value of treatment, and broader insurance coverage for inpatient mental health services all contributed to the shift to private providers of care. In the last few decades, the mental health care delivery system has evolved from a public monopoly to a diversified mix of public and private providers. The initial shift in the treatment of mental disorders was from long-term care in public psychiatric hospitals to brief periods of hospitalization in specialized psychiatric units of community-based general hospitals. General hospitals became the primary providers of inpatient psychiatric treatment. Only the most seriously and chronically mentally ill continued to be institutionalized in public psychiatric facilities. Increased private psychiatric hospital capacity recently produced a shift in locus of care to private and specialized psychiatric facilities. This study will address the implications of the changing role played by private psychiatric hospitals and the impacts of this change on the overall mental health care delivery system. Issues Raised by Privatization Throughout the 1980s, observers raised concerns about the potential consequences of privatization, although little research was done to determine 3

PAGE 12

what impacts this phenomenon was having on the mental health care delivery system (Levenson, 1983; Eisenberg, 1986; Brodie & Banner, 1987; Dorwart, Horgan, Schlesinger & Davidson, 1989). This research gap continued because the shift to private providers involved complex interactions between ownership types and degree of specialization as the locus of care moved from public psychiatric hospitals to primarily non-profit general hospitals, and more recently, to private psychiatric hospitals dominated by investor-ownership (Dorwart et al., 1989). The shift from predominantly non-profit general hospital psychiatric units to private psychiatric hospitals dominated by investor-ownership raised concerns about the effects of for-profit ownership on the delivery of psychiatric services. Eisenberg (1986), for example, argued that investor-owned facilities will "cream skim" to insure profitability; that is, they will provide only profitable services to well-insured patients, leaving the general hospitals with a large proportion of Medicaid and uninsured patients. Droste (1988), Pruitt and Weiner (1989), and Nazario (1992) reported evidence of inappropriate admissions to investor-owned psychiatric hospitals, particularly among adolescents, to generate revenues. Finally, Lyles and Young (1987) found that for-profit psychiatric hospitals are less likely to provide important but unprofitable community serVices (Lyles & Young, 1987). 4

PAGE 13

Utilization and costs of inpatient treatment for mental disorders may also be affected. Increases in private psychiatric hospital capacity may lead to overutilization of inpatient services and thereby raise costs. Moreover, a shift from general hospital psychiatric units, with low average lengths of stay (LOS), to private psychiatric hospitals with characteristically higher average lengths of hospital stay, could also result in higher mental health care costs because LOS is a critical determinant of total charges (Lyles & Young, 1987). Cream Skimming Cream skimming creates a competitive advantage for for-profit hospitals and erodes the general hospitals' ability to finance care through cross subsidization from private payers. As the market share of the private psychiatric hospitals increases, general community hospitals may be left with a disproportionate share of Medicaid and other low-income patients whose treatment costs are not fully reimbursed. This threatens the financial viability of psychiatric and substance abuse treatment in general hospitals and could eventually impair access to care for low-income patients, especially the chronically mentally ill (Marmor & Gill, 1990). Steven Scharfstein, a former director of the National Institute of Mental Health, suggested that two major and paradoxical problems may have occurred as a result of growth in private psychiatric hospitals: Access to necessary 5

PAGE 14

services may have decreased for the seriously and chronically mentally ill while overutilization of services by well-insured patients may have increased costs (Kanaan, 1991). Because investor-owned psychiatric hospitals primarily serve the well-insured patient population, the additional capacity to provide inpatient psychiatric care in these facilities may have increased utilization of services among the well-insured and thereby increased mental health care costs. Further, the financial viability of the general hospital sector may be weakened by a shift of well-insured patients to private psychiatric hospitals, leaving general hospitals with a disproportionate share of poor, uninsured, and Medicaid patients. This, in turn, could threaten access to care for this vulnerable population. There is evidence that private psychiatric hospitals do selectively admit patients based on ability to pay. Differences in payer mix between general hospital psychiatric units and private psychiatric hospitals have been documented. Private insurers pay approximately 40 percent of the overall costs of mental health treatment in the United States (NIMH, 1990a) but account for almost 70 percent of the revenues received by private psychiatric hospitals (National Association of Private Psychiatric Hospitals, 1990) and only 41 percent of the revenues received by general hospitals (NIMH, 1990a). Moreover, Hickman and.Dokecki (1989) have identified several strategies used by for-profit hospitals to admit patients based on ability to pay: 6

PAGE 15

(a) eliminating low-frequency and unprofitable (but apparently necessary) services; (b) recruiting medical staff who have only "desirable" practices and patients; (c) locating in affluent areas where there are fewer or no indigent, uninsured, or Medicaid patients; and, (d) denying all but emergency services to indigent patients. Curtailed Community Services Emergency services and outpatient clinics are "portals of entry" for the general patient population (Lyles & Young, 1987). Lyles and Young (1987) found that a disproportionately small percentage of for-profit psychiatric hospitals offered psychiatric emergency services or outpatient psychiatric clinics; the lack of these services provides such facilities the opportunity for greater selectivity in patient admissions. Private ownership has also been linked to other important aspects of institutional performance. For-profit hospitals, for example, are less likely to provide unprofitable community services or participate in the education of health professionals (Schlesinger & Dorwart, 1984; Frank & Salkever, 1991). The provision of graduate medical education in teaching hospitals results in increased costs that are not fully reimbursed by third.:party payers (Lyles & Young, 1987). In a study of California general and psychiatric hospital 7

PAGE 16

facilities, Lyles and Young (1987) found that no for-profit hospital-general or psychiatric-offered residency training. The authors noted that: Nonparticipation in physician training confers a competitive, cost reducing advantage on for profit hospitals, despite the clear dependence of these hospitals on physicians trained in not for profit and public hospitals. This transfer of talent represents an indirect subsidy of for profit hospitals by society at large (p. 55). Thus, for-profit hospitals have distinct competitive adva:ntages: A large proportion of well-insured patients and reduced costs resulting from not providing poorly-reimbursed but valuable community services and not participating in training programs for health professionals. Inappropriate Adolescent Admissions Private psychiatric hospitals have been successful at increasing access to care for children and adolescents, a previously underserved population (Dorwart et al., 1991). According to Greer and Greenbaum (1992), there was a 400 percent increase in adolescent admissions between 1982 and 1987. However, the dramatic expansion of care for this age group raised questions about potentially unnecessary hospitalization (Droste, 1988; Pruitt & Weiner, 1989; Dorwart et al., 1991). 8

PAGE 17

Compared to other types of hospitals, private psychiatric hospitals do serve a disproportionately large share of adolescents hospitalized for psychiatric or substance abuse treatment. In 1986, for example, patients 18 and under accounted for six percent of overall inpatient psychiatric admissions in all types of hospitals combined; six percent in general hospitals; and five percent in public psychiatric hospitals (NIMH, 1990a). However, this age group accounted for 20 percent of the admissions to private psychiatric hospitals in 1986 (Redick, Stroup, Witkin, Atay & Manderscheid, 1989). Frank, Salkever, and Scharfstein (1991) found that the use of inpatient care for substance abuse treatment and adolescent psychiatric treatment grew at .. rates well above the rate of growth for treatment of other conditions .. and thus were the major contributors to the escalating costs of mental health treatment benefits. Moreover, the authors concluded that 11extraordinary rates of increase in inpatient treatment of adolescents reported here and elsewhere require special scrutiny .. because they appear to be driven, in part, by the growth of private psychiatric hospitals (Frank et al., 1991, p. 122). Recently, state and federal agencies in Texas, New Mexico, and Florida investigated a large national psychiatric hospital chain and found substantial evidence of inappropriate and even fraudulent psychiatric hospitalization of children and adolescents (Nazario, 1992). 9

PAGE 18

Supplier-Induced Demand The traditional model of health facilities construction assumes that health care providers assess community needs, then build hospital facilities to fill those needs. The rapid growth in private psychiatric hospitals within a short period of time suggests that careful needs assessment was not a major factor in the decision to build these new facilities. Additionally, most of the new private psychiatric hospitals were built in states with little or no requirement that the need for new facilities be documented. Another model-called Roemer's Law-assumes that increases in the ability to provide hospital services leads to increases in the rate of utilization of services (United States General Accounting Office [GAO], 1991). Roemer's Law hypothesizes that additions to hospital bed capacity generate demand for hospital services; specifically, both the volume of admissions and the average length of hospital stay increase as a result of capacity increases. Although Roemer's Law has not previously been applied to private psychiatric hospital capacity changes, there is some evidence of a "Roemer effect" in the utilization of psychiatric and substance abuse treatment. According to Dorwart et al. (1991), the number of patients admitted for inpatient psychiatric treatment did increase dramatically in states where the number of private psychiatric hospitals increased. Other studies have also found evidence of a possible Roemer effect in inpatient psychiatric treatment. 10

PAGE 19

Mechanic (1989b) noted that, in areas where general hospitals increased the number of specialized psychiatric units, the per-capita utilization of psychiatric services increased. Lave (1989) pointed out that inpatient psychiatric services continue to grow despite decreasing overall hospital admissions and noted that, in recent years, psychiatric patients-but not medical patients-have been found to fill excess hospital beds. Schulz, Greenley, and Peterson (1984) found that average length of stay in psychiatric units of general hospitals increased as available bed supply increased. Other factors may limit the Roemer effect, however. Growing admission rates, particularly among privately insured patients, have led to disproportionate increases in the cost of mental health benefits. By the end of the 1980s, the costs of psychiatric and substance abuse treatment were the fastest growing portion of health care expenditures in the United States (CMHS & NIMH, 1992). In 1991, Kanaan predicted that rising mental health treatment costs would lead to a "managed care counterrevolution as private payers struggle to contain costs". Managed care organizations seek to reduce the utilization of mental health services and thereby reduce costs through a variety of strategies such as pre-admission screening and utilization review (CMHS & NIMH, 1992). Payer constraints such as managed care may provide countervailing pressures that limit the Roemer effect in inpatient psychiatric and substance abuse services just as changes in methods of paymet;1t led to 11

PAGE 20

decreasing utilization of inpatient care for medical and surgical treatment. Additionally, changes in the treatment of mental illness may continue to shift the locus of care to outpatient rather than inpatient services (Cooper, 1993). Differences in Length of Stay Roemer's Law predicts that average length of stay (LOS) as well as the number of admissions will increase as hospital capacity increases. LOS is affected, not only by changes in capacity, but also by patient and payer characteristics as well as by type of hospital in which services are provided. LOS, an important component of total hospital charges, varies significantly by hospital type. In 1986, comparisons among hospital types showed that the median length of hospital stay was 15 days: Public psychiatric hospitals had the longest median stay (28 days); private psychiatric hospitals followed with a 24 day median; and general hospitals had the shortest stay at 10 days (NIMH, 1990a). LOS also varied by age and diagnosis within each hospital type. While the longer length of hospital stay for those treated in public psychiatric hospitals can be explained by the severity of mental illness found in patients treated in these facilities (NIMH, 1990a), reasons for the wide variations in length of stay between general hospitals and private psychiatric hospitals are not yet clear (NIMH, a). As private psychiatric facilities provide a growing proportion of inpatient treatment for mental disorders, the 12

PAGE 21

longer length of stay typical of these facilities could lead to significant increases in the costs of providing care. Pumose of the Study The purpose of this study is to examine how Colorado's mental health care delivery system was modified when private psychiatric hospitals became major providers of inpatient care. Specifically, this study will examine the effects of increased private psychiatric hospital capacity on the utilization of inpatient mental health treatment and will analyze changes in patterns of patient allocation and payment among the diverse types of hospitals providing treatment. Colorado offers a unique opportunity to examine the changes in the overall inpatient mental health care delivery system arising from the rapid increase in the number of private psychiatric hospitals-an increase that paralleled the national experience. The Colorado Health Data Commission (CHDC) has collected the data necessary to evaluate the impact of these changes on all sectors of the inpatient mental health delivery system. The Commission's discharge data base includes detailed information on patients discharged from Colorado hospitals, including age, gender, diagnosis, length of stay, and source of payment. Data are collected from general, public psychiatric, and private psychiatric hospitals. 13

PAGE 22

Comprehensive hospital discharge data from all hospital types providing inpatient psychiatric or substance abuse treatment are available in only a few states. Twenty states, including Colorado, have a publicly mandated health data commission requiring hospitals to submit discharge data; however, 12 of those states exempt psychiatric hospitals from reporting requirements (Hughes, 1990). Of the remaining states that do collect discharge data from public and private psychiatric hospitals, Colorado is the only state that has also deregulated hospital construction (Hughes, 1990). Contribution of the Study Inpatient treatment for mental illness is now provided by a blend of public and private institutions. Although important differences have been identified among the diverse hospital types providing psychiatric and substance abuse treatment, very little research has been done to analyze the effects of those differences on the system as a whole (Goldman & Skinner, 1989). Moreover, during the past decade, mental health researchers have repeatedly identified the need for research in this important area (Levenson, 1982; Schlesinger & Dorwart, 1984; Brodie & Banner, 1987; Goldman & Skinner, 1989). It is clear that important changes are occurring; however, not enough is yet known to formulate public policy responses to the issues raised by the 14

PAGE 23

increasing role of private psychiatric hospitals in the provision of inpatient treatment for mental illness. It is hoped that the results of the study will provide insight into the changes that have occurred in Colorado's and other states' mental health care delivery systems as a result of the greater role now played by private psychiatric hospitals. The study findings will enable policymakers to understand better how Colorado's mental health care delivery system was modified by the emergence of private for-profit psychiatric hospitals. An understanding of the changing roles played by the diverse hospital types providing inpatient treatment for mental illness will enable national as well as state policymakers to craft more effective mental health policies for the 1990s. The growth in private investor-owned psychiatric hospitals was concentrated in southern and western states, including Colorado, with little or no regulation of hospital construction. It is expected that study results will be generalizable to those states that experienced a rapid growth in the number of private psychiatric hospitals within a relatively short period of time. However, study findings may not be generalizable to states that showed little or no increase in private psychiatric hospitals. Additionally, this study focuses on the years 1987 through 1989, when the number of private psychiatric hospitals in Colorado changed most significantly. Subsequently, the number of such facilities began to drop, both within the state and nationally. Finally, payer 15

PAGE 24

policies and methods of treatment for mental illness play a central role in the utilization of services; their continuing evolution may further constrain the generalizability of study results. 16

PAGE 25

CHAPTER 2 PROBLEM CONTEXT The United States has a long tradition of government involvement in the provision of treatment for the mentally ill. Public policies have determined the location and type of treatment provided. In the past few decades, profound changes in mental health policies have led to the deinstitutionalization of the chronically mentally ill and broadened the definition of mental illness, thereby creating a growing demand for community based psychiatric and substance abuse treatment. Privatization of mental health services began with the shift in locus of care from public psychiatric hospitals to predominately non-profit general hospitals. During this time, broader insurance coverage for mental illness expanded demand for services. More recently, deregulation of hospital construction permitted the rapid growth in the number of private investor owned psychiatric hospitals. To understand fully changes in the mental health care delivery system resulting from the privatization of services, it is necessary to examine the historical context in which they occurred. Mental Health Policy Changes Hospital care in this country has been primarily provided by the private sector. In contrast, inpatient psychiatric treatment was largely a state

PAGE 26

government responsibility for over 100 years (Scharfstein, 1988). In colonial times, towns and counties assumed financial responsibility for the care of the mentally ill in local almshouses and institutions. By the 1850s, care of the insane shifted to state-financed and administered institutions. This continued throughout the 1950s, when state hospitals provided over 90 percent of all treatment for mental disorders and financed virtually all of the costs. The states' primacy in the provision and financing of treatment for the mentally ill continued until the 1960s when federal policymakers began to play a more active role (Pardes, 1990). The Emerging Federal Role The policymaking role of the federal government in the treatment of mental illness gradually emerged after World War IT. During the war, the widespread incidence of psychiatric impairment among potential inductees, as well as in those who actually served, created a new awareness of the extent of mental disorders in the general population (English, Kritzler & Scher!, 1984). In 1946, the National Mental Health Act was passed, leading to a new federal role in mental health policy through the establishment of the National Institute of Mental Health (NIMH) in 1949. NIMH's mandate, however, was restricted to programs related to research and research training. 18

PAGE 27

In the 1980s, NIMH, in effect, broadened demand by expanding the definition of mental illness through studies documenting the prevalence of mental disorders among the general population (Lee, 1987). In 1984, for example, NIMH published a landmark epidemiological study that identified 29.4 million Americans as having mental disorders and further noted that only 20 percent of those so identified seek medical treatment (Custer, 1990). Later NIMH publications estimated that almost one-third of all American adults will have a mental or substance abuse disorder requiring treatment at some time during their lifetime (NIMH, 1988). By identifying significant latent demand for mental health treatment, these studies were a major impetus for the increase in community hospital capacity for the treatment of mental disorders (Kim, 1990a). Deinstitutionalization One of the most far-reaching policy changes was the trend to deinstitutionalize the chronically mentally ill. During the 1960s, both federal and state mental health policymakers began to emphasize community rather than institutional care (Pardes, 1990). Deinstitutionalization of the severely mentally ill became a viable option with the introduction of psychotropic drugs that could keep the symptoms under control and which soon came to be viewed as a less expensive and more humane option than institutional care. 19

PAGE 28

Deinstitutionalization was also stimulated by increased federal social welfare spending during the 1960s. The passage of the federal Medicaid legislation in 1964 and the expansion of disability benefits during the 1970s enabled the states to shift some of the costs of caring for the chronically mentally ill in community settings to the federal budget (Mechanic, 1989b ). The promise of federal revenues gave the states additional impetus to deinstitutionalize these patients. In 1980, the passage of the Mental Health Systems Act by Congress promised to further increase the federal role in the financing and delivery of mental health services. However, later cost containment strategies effectively jettisoned the Act, although at the same time, the number of persons at risk for serious mental illness increased (Mechanic, 1989b; Pardes, 1990). During the same period, community general hospitals serving the chronically mentally ill were confronted with an additional challenge. In 1981, Talbott warned that the mental health care delivery system would face a serious crisis as the baby boom generation entered the ages (25 to 44) when chronic mental illness becomes manifest. Sixty-three million babies were born between 1946 and 1961; as a result, the absolute number of persons at risk for developing schizophrenia and other serious mental disorders during this decade was substantial. Serious mental illness tends to be chronic, so the cumulative demand for services continues to increase with the accumulation of 20

PAGE 29

each successive birth cohort (Talbott, 1981). The increasing number of young, chronically mentally ill patients added to the number of deinstitutionalized older patients placed a tremendous burden on community mental health providers during the same decade that public resources became more constrained (Ghiselli & Frances, 1985; Mechanic, 1989b). Despite the growing number of persons at risk to develop chronic and severe mental illness, the trend to deinstitutionalize the seriously and chronically mentally ill-and therefore reduce public psychiatric hospital bed capacity-continued throughout the 1980s. Recently, Lamb (1992) documented the magnitude of the decrease in public psychiatric hospital beds: We have gone from 559,000 state hospital beds for a population of 165 million in 1955 to 103,000 state hospital beds for a population of more than 248 million today. Thus, state hospital beds per 100,000 have dropped from 339 to 41 (p. 669). The author noted that 53 long-stay psychiatric beds per 100,000 population is recommended and concluded that the trend to decrease public psychiatric hospital bed capacity must be stopped (Lamb, 1992). 21

PAGE 30

The Privatization Process In the last two decades, the locus of care for inpatient psychiatric treatment has gradually shifted from the public to the private sector. Concurrently, reimbursement for such treatment has shifted from state to federal programs. Costs also shifted to private insurers as mental health care coverage was added to employee health insurance benefits. Shifts in Locus of Care Deinstitutionalization initiated the shift of inpatient treatment of mental illness from almost entirely public providers of service (state and county mental hospitals) to private providers. The locus of care gradually shifted-first, to psychiatric units in general hospitals and, most recently, to private psychiatric hospitals. By 1990, approximately half of all psychiatric beds nationally were in the public sector; the remainder were evenly distributed between the specialty psychiatric hospitals and psychiatric units in general hospitals (Dorwart et al., 1991). Increases in the number of beds available to treat mental illness in the private sector are illustrated in Figure 2.1. 22

PAGE 31

Figure 2.1 Beds Per 100,000 Population, Private Psychiatric and Non-Federal General Hospitals: United States, Selected Years 1970;.. 1988* I !HWPrivate Psychiatric Non-Federal General I 19.8 19.8 ------1-9;1-----------------------------------------------------------20 15 r-------------------fas----------,3:7---------10 5 0 1 ......... .,,w., 1970 1976 1980 1984 1986 1988 Source: Center for Mental Health Services & National Institute of Mental Health, Mental Health, United States, 1992.

PAGE 32

The number of general hospitals providing separate inpatient psychiatric services grew slowly but steadily between 1970 and 1980, increased substantially between 1981 and 1984, then remained fairly stable between 1984 and 1988 (CMHS & NIMH, 1992). After the growth of psychiatric units in general hospitals, the deregulation of hospital construction between 1983 and 1986 permitted the rapid increase in the number of private psychiatric hospitals. Nationally, the number of private psychiatric hospitals had increased slowly between 1970 and 1980, then nearly doubled between 1984 and 1988 (CMHS & NIMH, 1992). Changes in Payer Mix Shifts in the locus of care for the treatment of mental illness were accompanied by significant changes in the pattern of reimbursement for these services (NIMH, 1990a). Historically, when mental illness was treated almost exclusively in state mental hospitals, there was an almost total reliance on state funds. By 1986, however, the states' proportion of expenditures for inpatient psychiatric and substance abuse treatment had diminished to just over 33 percent of total reimbursement for mental health services, including the states' share of Medicaid financing (NIMH, 1990a). The federal Medicare and Medicaid programs reimburs,ed 23 percent of total expenditures and the 24

PAGE 33

largest proportion--44 percent-was paid by private insurance companies and patient fees (NIMH, 1990a). Medicare and Medicaid. Medicare and Medicaid are the two major public insurance programs financing the treatment of mental illness. The Medicare program provides health insurance for those 65 and over as well as for disabled workers and is entirely federally financed. The Medicaid program provides medical assistance to the needy and is financed through federal and state matching funds. The combined federal and state Medicaid budgets for mental health treatment make Medicaid the largest single source of payment for mental health services (Marmor & Gill, 1990). In 1965, federal Medicaid dollars became available to cover the costs of inpatient psychiatric care in general hospitals but not in state psychiatric hospitals. This created a powerful incentive to deinstitutionalize the chronically mentally ill. Torrey (1990) concluded that, while deinstitutionalization was publicly promoted as a more humane and less restrictive treatment option, this was merely a rationalization of the states' desire to shift some of the costs of caring for this population to the federal government. In contrast, the Medicare program was. designed to provide only limited reimbursement for mental health treatment to reduce the states' ability to shift the costs of care to Medicare (Lave & Goldman, 1990). This strategy appears 25

PAGE 34

to have been successful: In 1989, only three percent of the Medicare budget was spent on inental health, suggesting that Medicare beneficiaries may underuse mental health benefits (Lave & Goldman, 1990). Total Medicare and Medicaid costs for inpatient psychiatric treatment are likely to increase in response to the growing number of chronically mentally ill now covered under these programs. Between 1986 and 1991, the number of chronically mentally ill receiving disability benefits through either the Social Security Disability Program (SSDI) or the Supplemental Security Income Program (SSI)_ increased by over 50 percent (CMHS & NIMH, 1992). Currently, the largest proportion of beneficiaries receiving disability payments from either program are classified as disabled as a result of mental illness (CMHS & NIMH, 1992). SSDI recipients receive health insurance through the Medicare program while SSI recipients are covered under the Medicaid program. The costs of treatment for beneficiaries disabled by chronic mental illness will be borne by the Medicare and Medicaid programs. In the last decade, Congress and state legislatures have reacted to the burgeoning costs of the Medicare and Medicaid programs by steadily reducing reimbursement to providers of health services. As a result, Medicare and Medicaid payments to hospitals often do not cover the full cost of providing inpatient care. Consequently, hospitals must shift these unreimbursed costs to privately insured patients through higher charges. Typically, Medicare 26

PAGE 35

reimbursement covers a higher percentage of per-patient costs than does the Medicaid program. According to some estimates, Medicaid pays only 50 percent of the fully allocated costs of inpatient care (Moore & Coddington, 1991). As a result of inadequate levels of reimbursement, hospitals serving large proportions of Medicare and Medicaid patients are threatened by financial insolvency. Private Insurers. Private insurance coverage of the costs of hospitalization for psychiatric and substance abuse treatment was gradually added to other health care benefits as treatment shifted from state to community hospitals (Astrachan & Astrachan, 1989). The expansion of health insurance benefits was not always voluntary, however. In the past decade, most states, including Colorado, mandated some form of coverage for mental health treatment for those covered by health insurance programs (Custer, 1990). As private insurance coverage increased and the social stigma previously associated with the treatment of mental illness decreased, the use of inpatient psychiatric and substance treatment expanded. Employers became more aware of the costs of substance abuse in the work place and were more willing to refer troubled employees to treatment programs. Parents became more likely to hospitalize problem_ adolescents. Juvenile court judges 27

PAGE 36

increasingly referred youthful offenders to drug abuse treatment programs rather than to juvenile detention centers. In the past few years, however, mental health benefits for the privately insured have been threatened by the growing costs of providing these benefits (Frank et al., 1991). According to the Foster Higgins Annual Benefits Survey, private insurance expenditures for mental health care benefits increased 34 percent in 1987, 27 percent in 1988, and 18 percent in 1989 (Ferguson, 1990). Escalating mental health care benefit costs are currently of serious concern to employers and insurers because these benefits are disproportionately high contributors to rising health insurance premiums (Frank et al., 1991). Private insurers have responded to the rising costs of mental health benefits by adopting a number of strategies aimed at decreasing use of services and thereby reducing costs. These "managed care" strategies include utilization review, selective contracting with providers offering discounted services, and capitated rather than fee-for-service reimbursement. Deregulating Hospital Construction The most recent development in the privatization of psychiatric and substance abuse treatment has been the doubling of private psychiatric hospitals that occurred in the United States between 1984 and 1988. The rapid expansion of private psychiatric hospitals was a major and unintended 28

PAGE 37

consequence of the federal deregulation of hospital construction in 1983. In 1974, federal concern over rising health care costs led to the regulation of hospital capital spending through the federally mandated and financed certificate of need review process administered by the states. The CON review process was based on the assumption that the containment of hospital bed expansion would help control the growth in hospital costs (Goldsmith, 1984). Underlying this assumption was the theory that additions to hospital bed capacity generate demand for hospital services (Ginsburg & Koretz, 1983). This approach to cost-containment, however, gave way to the national emphasis on deregulation during the 1980s. As a result, the federal CON review requirement was eliminated and federal funding for CON review was phased out between 1983 and 1986, although many states still retain some form of CON review. Consequences of Deregulation While the elimination of CON review was followed by a dramatic increase in the number of private psychiatric hospitals, the number of general hospitals remained stable. Prior to deregulation, general hospitals were faced with significant excess capacity. In the early 1980s, both public and private payers of hospital services placed restrictions on reimbursement for services through reforms such as prospective payment and managed care (Dorwart & 29

PAGE 38

Schlesinger, 1988). Payment reforms led to lower average lengths of hospital stay for medical/surgical patients and a growing trend to provide medical and surgical services in outpatient, rather than inpatient, settings; the result was significant excess general hospital capacity (Dorwart & Schlesinger, ). In contrast, inpatient psychiatric care remained profitable during this period: Medicare's reimbursement limitations did not usually include psychiatric diagnoses; private insurance coverage for mental illness was stable or expanding; and demand for services was growing (Dorwart & Schlesinger, 1988). Psychiatric and substance abuse treatment programs were seen as major revenue producers in an era of stiff competition, excess hospital capacity, and constrained revenues. Thus, declining occupancy rates and high profit margins for inpatient psychiatric care led general hospitals to convert medical beds to psychiatric beds (Dorwart & Schlesinger, 1988). Excess capacity to provide inpatient medical and surgical treatment also encouraged the investor-owned multi-hospital chains to "shift their impressive purchasing power into the market for psychiatric services" (Dorwart, Epstein & Davidson, 1988). When hospital construction was deregulated between 1983 and 1986, investor-owned chains had access to the capital necessary to expand rapidly. Consequently, when the number of private psychiatric hospitals in the United 30

PAGE 39

States doubled between 1984 and 1988, over 70 percent of these hospitals were investor-owned (National Association of Private Psychiatric Hospitals, 1990). The Colorado Experience After Colorado deregulated hospital construction in 1987, the number of private psychiatric hospitals in the state doubled by 1989. This growth, which paralleled the national experience, was not in response to a shortage of private psychiatric hospital beds in the state. In 1986, the national average number of private psychiatric hospital beds was 12.6 beds per 100,000 population; in that year, Colorado had 12 beds per 100,000 (NIMH, 1990a). The national per-capita private psychiatric hospital rate peaked at 17.4 beds per 100,000 in 1988 (CMHS & NIMH, 1992). Colorado, by comparison, had 32 private psychiatric beds per 100,000 population in 1990, after the number of private psychiatric hospitals had doubled (Rehak, 1990). Inpatient psychiatric and substance abuse treatment is provided, not only in specialized public and private psychiatric hospitals, but also in general hospitals providing medical and surgical as well as psychiatric treatment. Capacity is measured by the number of hospitals providing treatment and the number of beds available to treat patients within each facility. Accordingly, an accurate assessment of the capacity available for the treatment of mental disorders must include general as well as specialized psychiatric hospitals and 31

PAGE 40

also must consider available bed capacity as well as the number of facilities providing treatment. By both measures, the capacity to provide inpatient treatment for mental illness increased in Colorado between 1986 and 1990. The doubling of private psychiatric hospitals was not offset by a reduction in number of facilities providing services in either the public psychiatric or the general hospital sector. The number of public psychiatric hospitals and general hospitals with specialized psychiatric units remained stable during the study period (Health Facilities Division, Colorado Department of Health, 1991). When the number of beds available to provide treatment ("staffed beds") is used to measure changes in the capacity during this period, capacity increased dramatically in the private psychiatric hospitals and increased slightly in both the public psychiatric and general hospitals. Table 2.1 illustrates the number of beds available to treat mental illness by type of hospital during this time period. 32

PAGE 41

Table 2.1 Beds Available for Psychiatric Treatment by Hospital Type in Colorado, 1987 and 1989* % Hospital Type 1987 1989 Change Public General 100 113 13% Private Non-Profit General 484 540 11.6% Private For-Profit General 94 118 25.5% ** from the American Hospital Association, Annual Survey of Hospitals for the years 1987 and 1989 as reported to the 'Health Facilities Division, Colorado Department of Health. the number of staffed beds available in the private psychiatric hospitals was estimated because only five out of the eleven facilities reported the number of staffed beds to the AHA In the five reporting facilities, staffed beds were an average of 87 percent of total beds licensed by the state for each facility. The general hospitals' ability to use beds in medical/surgical units-called .. scatter beds .. -for psychiatric and substance abuse treatment can make it difficult to determine accurately the total bed capacity available to treat mental illness in general hospitals. However, only seven percent of the general hospital psychiatric discharges reported to the Colorado Health Data Commission in the years 1987 and 1989 were from general hospitals without a 33

PAGE 42

distinct psychiatric unit. Thus, despite the general hospitals, ability to use "scatter beds", most inpatient psychiatric or substance abuse treatment provided by Colorado general hospitals was in specialized psychiatric units. The growth in Colorado,s capacity to provide inpatient treatment for psychiatric and substance abuse has the potential to alter the state,s mental health care delivery system in ways that are not yet clear. During the study period, the private psychiatric hospital sector evolved from a minor toa major provider of inpatient treatment for mental disorders. Additionally, investor ownership became the dominant ownership form for this hospital type. This study addresses issues arising from the restructuring of Colorado,s mental health care delivery which occurred as a result of the emergence of private psychiatric hospitals as major providers of service. 34

PAGE 43

CHAPTER 3 LITERATURE REVIEW In the past few decades, the institutional structure of the mental health care delivery system has been transformed by public policies that have encouraged a shift from public to private providers of services. The system's restructuring began with the deinstitutionalization of the chronically mentally ill. Demand for community-based psychiatric or substance abuse treatment also grew as the stigma associated with mental illness diminished and treatment became more socially acceptable. Additionally, changes in financing, such as broader insurance coverage for the treatment of mental disorders and availability of federal Medicaid funds, encouraged a shift in locus of care to community settings. The Evolving Structure The movement to deinstitutionalize the chronically mentally ill was based on the assumption that appropriate care, when needed, would be available in community general hospitals. Proponents of deinstitutionalization argued that long hospital stays in state mental hospitals "reinforced disability and isolation" and that brief hospital stays in community-based general hospitals would enable the mentally ill to receive needed services while learning to become functioning members of society (Starr, 1982, p. 365).

PAGE 44

Additionally, federal financing for community, but not institutional, care created a powerful incentive for the states to deinstitutionalize the mentally ill (Torrey, 1990). Initially, the planned shift from public psychiatric hospitals to general hospitals in community settings raised concerns that mental health care for the poor, the uninsured, and the seriously mentally ill would simply shift from one public setting (state psychiatric hospitals) to another (public general hospitals). Early studies suggested that this had occurred. Recently, however, Fisher et al. (1992, p. 1117) found that public and private general hospitals "currently share, but have not fully absorbed, the role of the state hospital". General hospitals are now the centerpiece of the mental health care delivery system. The acceptance of this role is consonant with the mission and goals of most general hospitals, which have traditionally been expected to respond to community needs. General hospitals were predominantly private, non-profit entities established to serve local communities and financed by philanthropic organizations or public initiatives (Fisher et al., 1992). This created a social contract between general hospitals and the communities they served. Non-profit general hospitals additionally receive tax-exempt status in exchange for community service. The recent emergence of private psychiatric hospitals dominated by investor-ownership with a focus on profit maximization has raised concerns 36

PAGE 45

about the effects of this development on the evolving mental health care delivery system. The chief goal of investor-owned facilities is profit maximization. As economist Burton Weisbrod (1987) has noted, hospitals motivated primarily by financial incentives are less likely to respond to community needs. Moreover, for-profit hospitals have "an undisguised preference for privately-insured patients" (Starr, 1982, p. 436). Weisbrod (1987) raised the possibility that non-profit hospitals will be driven out of business by the exploitive practices of for-profit hospitals such as providing lucrative services to well-insured patients while avoiding less remunerative services and patients. Three potential problems resulting from the growing role of private psychiatric hospitals have been identified by observers and researchers. First, the growth in investor-owned psychiatric hospitals could result in the diversion of privately insured patients away from general hospitals and thereby financially weaken the general hospital sector. Overutilization by well-insured patients or their families induced by the private psychiatric hospital sector's increased capacity to provide inpatient treatment was another concern. More specifically, the need to insure profitability might encourage investor-owned psychiatric hospitals to inappropriately admit targeted patient groups such adolescents or substance abusers. Finally, a shift from general hospitals, with low average lengths of stay, to private psychiatric hospitals, characterized by 37

PAGE 46

longer average lengths of stay, might result in increasing lengths of hospital stay and thereby increase costs. For-Profit Ownership When the private psychiatric hospital sector began its dramatic growth in the rnid-1980s, concerns about investor-ownership heightened. Proponents and opponents debated the merits of for-profit ownership; both agreed that the inpatient mental health care system would be changed significantly by the increasing number of private for-profit psychiatric facilities. Proponents' Perspective Proponents of for-profit ownership argued that investor-owned hospitals provided a positive .benefit to society because for-profit facilities increased competition, thereby increasing incentives for all hospitals to make economically sound decisions (Levenson, 1982). Buck (1987) argued that free enterprise and the profit motive could improve the field of psychiatry by delivering better treatment at lower costs while at the same time providing a reasonable return to investors. Rather than condemning the emergence of for profit psychiatric hospitals, Buck recommended that the marketplace be allowed to determine whether these facilities deliver care that satisfies patients. He further argued that quality assurance mechanisms such as 38

PAGE 47

accreditation and the utilization review programs of third-party payers were already in place to monitor effectiveness. Buck's assertion notwithstanding, there is little evidence to support the claim that for-profit hospitals are less costly or more efficient than non-profits. Studies comparing costs of services by type of ownership have found no evidence that for-profit hospitals are less costly than non-profit hospitals (Reiman, 1983; Gray & McNerney, 1986). As Starr (1982, p. 433) noted, the argument that for-profits are more efficient assumes that the "incentives facing hospitals reward efficiency". Under traditional methods of reimbursement, however, hospitals-both investor-owned and non-profit-are rewarded for providing additional services and admitting more patients. Costs for psychiatric treatment in specialized psychiatric hospitals are generally higher than general hospital costs because average length of stay is longer in private psychiatric hospitals. Private psychiatric hospitals report average lengths of stay two to three times longer than those reported in general hospitals (NIMH, 1990a). Accordingly, a shift to private psychiatric hospitals would likely increase the overall costs of inpatient treatment of mental illness. Brodie and Banner (1987) theorized that the marketing of psychiatric services by the investor-owned psychiatric hospitals performed "a public service by making people aware of where they can get help for their problems, and 39

PAGE 48

perversely, by lending to psychiatric treatment the American societal stamp of approval that is advertising". The authors noted that approximately half of the people using soup kitchens needed psychiatric treatment but were unaware of how to get it; they suggested that public hospitals could prevent considerable mental anguish by adopting marketing strategies to reach the indigent mentally ill. The marketing of psychiatric services has potential disadvantages as well as advantages, however. Dorwart et al. (1988) conceded that the marketing of inpatient psychiatric services could increase the public's lrnowledge of available services and decrease the stigma of using those services. The authors also noted that marketing had the potential to induce demand for minimally necessary services and thereby drive up mental health care costs. This was confirmed in a study of the effects of advertising of adolescent psychiatric services on the parents of teenagers. Greer and Greenbaum (1992) found that parents experiencing no serious problems with their adolescent children were more likely to respond to the advertisements than were parents of troubled adolescents. Opponents' Perspective Critics of for-profit ownership argued that the purpose of investor owned hospitals-profit maximization-is diametrically opposed to sound public 40

PAGE 49

health policy (Eisenberg, 1986). According to Eisenberg, the for-profit psychiatric hospital sector is characterized by: .. .its ability to charge and collect for services, pass through capital costs and skim off profitable patients-and at the same time, shun its proportionate responsibility for the medically indigent, for the costs of medical education and research and for meeting community needs (Eisenberg, 1984, p. 1015). To insure profitability, critics argued, investor-owned psychiatric hospitals would provide only profitable services to well-insured patients by eliminating necessary but unprofitable community services such as psychiatric emergency room care and by using several strategies to admit primarily well insured patients (Lyles & Young, 1987). Moreover, Bickman and Dokecki (1989) identified several strategies used by investor-owned hospitals to select patients based on ability to pay. These included locating hospitals in affluent areas with few indigent or uninsured patients, eliminating unprofitable but necessary community services (for example, emergency services and telephone hot lines), and recruiting medical staff with the desirable mix of patients. Bickman and Dokecki (1989) further noted that for-profit hospitals "functionally skim" when they do not participate in teaching or research activities. 41

PAGE 50

Dorwart et al. (1991) and Lyles and Young (1987) found that for-profit psychiatric hospitals were much less likely to offer important but unprofitable community services such as psychiatric emergency services, medical education, case management for chronically mentally ill patients, and telephone hot lines. Such services were offered most frequently by public and non-profit general hospitals. Among private psychiatric hospitals, non-profits were more likely to offer these services than for-profits. Additionally, critics feared that the primarily non-profit general hospitals would be left with a disproportionately large volume of Medicaid and other low income patients whose treatment costs are not fully reimbursed. Competitive pressures could then force general hospitals to decrease or eliminate unprofitable psychiatric services. Access to needed services would be threatened for the poor, the uninsured, and the Medicaid patients needing treatment for mental illness (Dorwart et al., 1989). Effects of Ownership and Specialization Several research studies have identified critical differences among the diverse types of hospitals providing inpatient treatment for mental illness. Patterns of patient characteristics such as age, diagnostic mix, and source of payment for services differ by ownership type (public, private non-profit, and private for-profit) and degree of specialization (general or psychiatric hospital). 42

PAGE 51

The growth of private psychiatric hospitals may alter existing patterns and thereby modify the overall mental health care delivery system. Ownership and General Hospitals Schulz et al. (1984), Goplerud (1986), and Fisher et al. (1992) studied the effects of ownership type on general hospitals providing psychiatric services. Schulz et al. (1984) found that public general hospitals were more expensive than private non-profit general hospitals; they concluded that excess capacity in public general hospitals led to higher average lengths of stay and, therefore, higher costs. Goplerud (1986) found that for-profit management of general hospital psychiatric units resulted in higher occupancy rates, greater profitability, and better quality of care. Fisher et al. (1992) found that general hospitals now provide a significant proportion of inpatient for the seriously and chronically mentally ill, a patient population previously served by state mental hospitals. Schulz et al. (1984) compared the costs of psychiatric care in public general hospitals and private non-profit general hospitals by surveying a representative sample of both ownership types located in Wisconsin. The authors found that average costs per admission were 59 percent higher in public general hospitals than in private non-profit general hospitals. Moreover, cost differences were not found to be significantly influenced by 43

PAGE 52

demographic characteristics, diagnostic mix, or severity of illness of the patient populations; average daily census was the most significant variable affecting cost (Schulz et al., 1984). The authors concluded that the lower occupancy rates combined with a method of reimbursement based on per-diem rates multiplied by the number of patient days led to higher average lengths of stay in the public general hospitals and, therefore, to higher costs. Although private general hospitals shared the same method of reimbursement and had the same incentives to expand length of hospital stay, higher average occupancy rates led to a lower average length of stay in private general hospitals. To evaluate the effects of for-profit management on general hospital psychiatric units, Goplerud (1986) compared the performance of 13 general hospitals before and after a for-profit firm was hired to manage the hospitals' psychiatric units. No significant change in either length of hospital stay or diagnostic mix of patients occurred as a result of for-profit management. However, for-profit management did result in higher occupancy rates and higher collections-to-billings ratios, and thereby increased the units' profitability. Higher occupancy rates were attributed to aggressive marketing efforts that broadened the patient referral base. The payer mix changed slightly: The proportion of government payers increased and the proportion of privately insured patients decreased. 44

PAGE 53

Goplerud (1986) also found that quality of care, as measured by staff to-patient ratios, staff inservice training, and structured treatment time provided to patients daily, showed improvements under for-profit management. The author noted that the hospitals included in the study were not indicative of a random sample; rather, these hospitals contracted with a for-profit management firm precisely because their psychiatric units were in trouble. Nonetheless, Goplerud concluded that for-profit management was not incompatible with high-quality inpatient psychiatric care. Fisher et al. (1992) surveyed public and private general hospitals to determine whether there were significant differences in diagnostic and payer mix based on ownership. The purpose of the study was to evaluate the extent to which private general hospitals had increased their role in the care of the poor and chronically mentally ill who were once treated primarily in state hospitals. Previous studies had found that, in the 1970s, public general hospitals were the primary caregivers to deinstitutionalized patients. Data for the study were obtained from the National Mental Health Facilities Study and consisted of a questionnaire mailed to a random sample of public and private general hospitals with specialized psychiatric units. Private general hospitals were not further differentiated by for-profit or non-profit ownership. The degree to which patients treated in general hospitals 45

PAGE 54

resembled the population historically treated in state psychiatric hospitals was determined by analyzing the diagnostic and payer mix of patients. Fisher et al. (1992) found that general hospitals, both public and private, served a significant proportion of patients once treated in state psychiatric institutions. These findings suggest that private general hospitals are playing an increasingly important role in the treatment of the chronically mentally ill, although patients with serious mental illness and Medicaid patients are still more prevalent in public general hospitals than private general hospitals. Diagnostic and Payer Mix Differences Dorwart et al. (1991) analyzed how the privatization of inpatient psychiatric and substance abuse treatment affected access to inpatient mental health services. Differences in diagnostic and payer mix among hospital types were used to assess changes in access to care. Dorwart et al. (1991) surveyed a national random sample of non-federal general and psychiatric hospitals; a total of 915 hospitals responded. The survey response rate varied from 78 percent from public psychiatric hospitals to only 38 percent from private for-profit psychiatric hospitals. To reduce possible sample bias in the for-profit psychiatric hospital category, results were compared to data collected by the National Institute of Mental Health and the 46

PAGE 55

National Association of Private Psychiatric Hospitals (NAPPH). Study results were consistent with those found by NIMH and NAPPH. Dorwart et al. (1991) compared patients admitted to all hospital types and found key differences in patient characteristics such as age, diagnostic mix, and length of stay. Additionally, the payer mix varied across hospital types. The authors found that public psychiatric hospitals had a higher proportion of schizophrenic patients and a considerably longer average length of hospital stay than other hospital types. This finding was supported by earlier analysis of the patient mix in public psychiatric hospitals (NIMH, 1990a). Private general and psychiatric hospitals had a higher proportion of patients with affective disorder or chemical dependency and shorter average lengths of stay. The diagnostic mix of patients admitted to either private general hospitals or private psychiatric hospitals was similar; both appeared to be specializing in disorders that respond to short-term treatment. However, private for-profit psychiatric hospitals admitted a significantly greater proportion of patients for disorders of childhood and adolescence than any other hospital type. Additionally, source of revenue varied considerably by hospital type (Dorwart et al., 1991). Private insurance paid for more than 40 percent of patient care overall, but accounted for more than two-thirds of the revenues received by for-profit psychiatric hospitals. Health maintenance organizations and preferred provider organizations accounted for a small proportion (five 47

PAGE 56

percent or less) of total revenues, although ten percent of the American population was enrolled in these organizations. Surprisingly, over one-third of the public psychiatric hospitals' revenues came from payers such as Medicare, Medicaid, or commercial insurers, rather than from state funding as was previously the case. The authors also noted that the proportion of Medicaid payments for psychiatric care was higher among all hospital types in their study as compared to earlier studies, indicating that overall Medicaid spending for inpatient psychiatric care may have increased. Dorwart et al. (1991) found that private non-profit general and psychiatric hospitals received over one third of their revenues from public programs such as Medicare and Medicaid, while for-profit psychiatric hospitals received one-quarter of their revenues from these public programs. This study also confirmed Lyles and Young's (1987) earlier finding that for-profit psychiatric hospitals were much less likely to offer important but unprofitable community services such as psychiatric emergency services, medical education, case management for chronically mentally ill patients, and telephone hot lines (Dorwart et al., 1991). Such services were offered most frequently by public and non-profit general hospitals. Among specialty psychiatric hospital for-profit hospitals were least likely to offer these services. 48

PAGE 57

Dorwart et al. (1991) concluded that the increases in private psychiatric hospitals and psychiatric units in general hospitals had increased geographic accessibility for well-insured patients but that methods of screening out unprofitable patients had reduced accessibility for low-income patients. Moreover, private facilities-both general and psychiatric-appeared to specialize in disorders requiring short-term treatment, thereby limiting access to services for the seriously and chronically mentally ill. Finally, the expanded role of for-profit psychiatric hospitals: .. .increased access to care for children and adolescents, a long underserved population. But because this population is also at risk for excessive institutionalization, this expansion also raises concerns about potentially unnecessary hospitalization, particularly since these facilities conduct relatively limited review of the appropriateness of admissions (Dorwart et al., 1991, p. 210). Inappropriate Admissions Other observers have also raised questions about the appropriateness of inpatient psychiatric care for the growing number of children and adolescents admitted for such treatment (Pruitt & Weiner, 1987; Droste, 1988; Nazario, 1992). Approximately a quarter of a million adolescents were admitted to 49

PAGE 58

psychiatric hospitals in 1987; this was a 400 percent increase in use between 1982 and 1987 (Greer & Greenbaum, 1992). Most were admitted to for-profit psychiatric hospitals. Treatment costs were estimated to be $30 billion and were paid primarily by private insurers (Greer & Greenbaum, 1992). Frank et al. (1991) found that the increasing use of inpatient mental health treatment by children and adolescents was the major factor in the rising costs of mental health care benefits paid by private insurers. According to the Foster Higgins Annual Benefits Survey, private insurance expenditures for mental health care benefits increased 34 percent in 1987, 27 percent in 1988, and 18 percent in 1989. As a result, escalating mental health care benefits have been of increasing concern to employers and insurers. In response to this concern, Frank et al. (1991) examined changes in expenditures from 1986 to 1989 to determine what factors contributed to the growth in costs for psychiatric and substance abuse treatment among private insurers. The authors noted that benefits for psychiatric and substance abuse treatment are viewed as disproportionate contributors to rising health insurance premiums, so they are vulnerable to cutbacks and restrictions (Frank et al., 1991). Frank et al. (1991) used three separate data bases to describe overall trends in changes in use and costs of mental health treatment, then further compared differences by state. The largest data source was a national 50

PAGE 59

insurance claims data base consisting of claims for employees of major U.S. corporations and their dependents. The national claims data base (MEDSTA T) provided information on changes in use and costs of mental health treatment for a well-insured, nationally representative population of up to 1.3 million persons (Frank et al., 1991). Employers were distributed across the U.S. and composed of diverse industries. The other two data bases were hospital discharge data sets collected by public agencies in the states of Maryland and Washington. An analysis of the national data base revealed a number of important findings (Frank et al., 1991). According to the authors, charges for psychiatric and substance abuse treatment rose significantly above the rate for health care overall between 1986 and 1989: Psychiatric care costs increased 20.1 percent and substance abuse treatment increased 32.4 percent while general medical care costs rose 13.0 percent during this time period. Most of the increase in costs for psychiatric care (72 percent) was due to an increase in utilization among persons 18 and under. Also contributing to the cost rise was a marked increase in treatment for substance abuse for both adults and juveniles. Separate analyses of hospital discharge data for privately insured patients who received inpatient mental health treatment between 1986 and 1988 in the states of Maryland and Washington, however, revealed different trends (Frank et al., 1991). In Maryland, use of inpatient mental health 51

PAGE 60

services increased slightly as a result of increased admissions for substance abuse treatment. No significant increases in use by juveniles occurred. The discharge data for Washington revealed similar patterns. Frank et al. (1991) concluded that the differences revealed in a comparison of findings between the analysis of the national claims data base and the analyses of the two state discharge data bases were due largely to key differences in the hospital sectors of Maryland and Washington. Neither state had many private psychiatric hospitals and there was no significant growth in the number of these facilities in either state. Both states regulate the hospital industry more intensively, including regulation of hospital construction through CON, than most other states (Frank et al., 1991). The authors concluded that the rapid growth in inpatient use by juveniles nationally was driven by the huge increase in private for-profit psychiatric hospitals. Frank et al. (1991) also suggested that the aggressive marketing campaigns specifically targeting the adolescent market contributed to the growth in psychiatric hospital admissions among this age cohort. The authors noted that rising mental health care expenditures "require targeted interventions aimed at containing costs in areas that have experienced unusual rates of growth", specifically psychiatric treatment for adolescents and substance abuse treatment for patients in all age categories (Frank et al., 1991). 52

PAGE 61

Aggressive marketing by for-profit hospitals has often been cited as an important factor in. the exponential growth in use of inpatient psychiatric services by adolescents (Frank et al., 1991; Kanaan, 1991). To determine the effectiveness of marketing psychiatric services to adolescents, Greer and Greenbaum (1992) studied the relationship between advertising and the likelihood that parents would seek psychiatric care for their adolescents. The study was limited to 88 mothers of high school students; most were college educated and virtually all had insurance coverage. The results suggested that fear-based advertising, in particular, encouraged referrals to hospital psychiatric services by parents. Interestingly, parents who reported no serious problems with their children were more likely to be frightened by the advertising, implying that such ads may arouse more fear among those who least need the services (Greer & Greenbaum, 1992). Supplier-Induced Demand The growth in the volume of inpatient admissions to private general and psychiatric hospitals for the treatment of mental disorders has raised concerns about the rising costs and potential overutilization of these services (Kanaan, 1991). According to Lave (1989), the increasing use of inpatient psychiatry is also puzzling: 53

PAGE 62

Why is inpatient psychiatry growing so rapidly? Why is it growing in the face of overall decreasing hospital admissions, increasing substitution of outpatient for inpatient medical services, and a research literature that suggests that, for most psychiatric services, inpatient services are no more effective than outpatient or a combination of outpatient and day hospitalization? (Lave, 1989, p. 157). Many researchers have suggested that increases in the use of inpatient mental health treatment may have been driven largely by the dramatic increases in the capacity to provide such care in the private sector (Mechanic, 1989b; Dorwart et al., 1991; Frank et al., 1991). There is research supporting the hypothesis that the supply of hospital beds influences the demand for inpatient medical and surgical services. However, there is a lack of definitive research on the effects of bed supply on the utilization of inpatient psychiatric services because the growth in the capacity to provide inpatient mental health treatment in the private sector is a relatively recent phenomenon. Alternative Models Under the traditional model of health facilities planning, hospitals are built when the need for a new facility is identified by health care providers within the community (Reeves, Bergwall & Woodside, 1984). Needs 54

PAGE 63

assessment, using information such as the health status and demographic characteristics of the community, establishes the basis for determining what type of services are needed. The providers' decision to construct new facilities is then based on the level of need demonstrated by the availaple evidence. The traditional model assumes that the supply of hospital beds is based on the demand for hospital services. This model also assumes the existence of a reasonably consistent set of criteria defining appropriate levels of hospital bed supply within a given community. Feldstein (1981}, however, observed that It has become a truism among those concerned with the practical problems of health-sector planning that when additional hospital beds are constructed, they are soon fully occupied. Administrators have learned that a technologically necessary level of hospital use is chimerical (p. 67). An alternative hypothesis was initially proposed by Milton Roemer in 1961 and has come to be known as Roemer's Law. Roemer theorized that the number of hospital beds available within a community determine the demand for hospital services. To test his hypothesis, Roemer compared one community's hospital utilization rates before and after a rapid increase in general hospital bed capacity. During this time, other factors, such as the total 55

PAGE 64

population, prevalence of illness, and supply of health care personnel, remained stable (Roemer, 1961). Roemer found that hospital utilization, as measured by number of admissions, average length of hospital. stay, and total patient days, grew substantially after capacity increased. The author was most surprised by the increases in average length of stay because average durations of hospital stay had been decreasing nationwide for many years. Roemer recognized that 11the need for hospital beds .. .is full of social and subjective determinants, and the level at which need is recognized is heavily influenced by the supply of beds available for its satisfaction .. (Roemer, 1961, p. 37). Roemer later (1976) concluded that, under conditions of widespread economic support of hospitalization through health insurance, effective control over hospital utilization rates, and therefore, hospital costs, would depend upon some sort of control over the supply of hospital beds. Roemer's Law To understand how Roemer's Law operates, it is necessary to discern what differentiates the demand for hospital services from the demand for most goods and services. The demand for hospital care differs from the demand for goods and services because it is the physician, acting as agent for the patient, who usually decides whether a patient is to be hospitalized. According to 56

PAGE 65

Feldstein (1981), the consumer of health services-the patient-lacks the knowledge necessary to make the decisions about appropriate treatment for his/her symptoms or illness; instead, he/she delegates this authority to the physician who, acting as the patient's agent, makes the treatment decisions. However, this agency relationship is flawed: The physician's decision also reflects "his own self-interest, the pressures from professional colleagues, a sense of medical ethics, and a concern to make good use of hospital resources" (Feldstein, 1981, p. 67). Feldstein (1981, p. 68) noted that a nonprice rationing of hospital services by physicians occurs when hospital occupancy rates are high and when the number of hospitalized patients with "less serious conditions and postponable treatments" is reduced. He concluded that physicians do not only act as agents of the patient, but also consider availability of resources and peer pressure when deciding whether to hospitalize patients. Physicians, however, perform a less prominent role in the decision to admit patients to private psychiatric hospitals. According to Dorwart et al. (1989), most patients admitted to general or public psychiatric hospitals are referred by physicians; in contrast, 60 percent of private psychiatric hospital patients admit themselves. Private psychiatric hospitals market their services directly to the potential consumer and, in a competitive environment, have a strong incentive to admit well-insured patients, regardless of medical necessity. 57

PAGE 66

Thus, increased private psychiatric hospital capacity is likely to be followed by increased utilization, particularly among the well-insured. CON Review Roemer's Law later became the major theoretical rationale for the regulation of hospital construction through the certificate-of-need (CON) review process (Ginsburg & Koretz, 1983). In 1974, federal concern over rising health care costs led Congress to enact the Health Planning Resource and Development Act that regulated hospital capital spending through a federally mandated and financed (CON) review process administered by the states. During the 1980s, federal deregulation policies led to the elimination of the CON review requirement. Federal funding for CON review was phased out between 1983 and 1986. Many of the states that eliminated CON review also experienced a rapid and dramatic growth in private, investor-owned psychiatric hospitals (NIMH, 1990a). Utilization of inpatient psychiatric services also jumped dramatically in states where the number of private psychiatric hospitals increased (Dorwart et al., 1991), suggesting a "Roemer effect" in the use of inpatient mental health services. Growing utilization rates among privately insured patients also led to disproportionate increases in the costs of mental health benefits (Frank et al., 1991). 58

PAGE 67

GAO's Study of Roemer's Law The relationship between capacity and utilization has not escaped the notice of policymakers who have recently reconsidered the federal decision to deregulate hospital construction: If Roemer's law is correct and the provision of some services is discretionary (that is, based at least partially on available capacity), that suggests a number of policy options for controlling costs. Foremost among these is a strategy based on the assumption that limiting or restricting hospitals' capacity will decrease volume and thus achieve cost savings without affecting needed access to care. However, if decisions to provide services are based primarily on the occurrence and prevalence of illness, limiting capacity might well result in lower volume, but at the sacrifice of reducing necessary access to care (U.S. General Accounting Office, 1991, p. 1). To determine if studies of health services delivery support Roemer's law, the U.S. General Accounting Office (GAO) analyzed the results of 29 studies that met their criteria of empiricism, timeliness, and relevance (GAO, 1991). A statistical method (meta-analysis) was used to synthesize the results of the studies. A relationship between capacity and volume was found: 59

PAGE 68

Capacity did contribute somewhat to the volume of services provided, but other factors were much more important. Nonetheless, even a weak relationship can be important from a budgetary point of view. For example, in 1983, a reduction of one bed per thousand population would have reduced admissions by over 4 percent (equivalent to about $5 billion in costs) or reduced length of stay by almost 3 percent (equivalent to over $3 billion in costs) (GAO, 1991, p. 4). The GAO report (1991) also identified gaps in the research that need to be filled in order more clearly to understand the. relationship between hospital capacity and the utilization of services. The report noted that the studies of Roemer's law were done before third-party payment reforms such as Medicare's prospective payment system were adopted; changes in reimbursement may have significant impact on utilization. Recent changes in medical practice patterns, such as the shift to outpatient treatment of many medical conditions, may also affect utilization. Further, the completed studies of Roemer's law did not adequately eliminate alternative explanations for increases in utilization such as potential differences among the health or demographic status of the populations studied (GAO, 1991). The GAO report 60

PAGE 69

and the studies reviewed by GAO did not deal specifically with the effects of capacity changes on the utilization of mental health services. Although the GAO study did report a weak but statistically significant relationship between increased capacity and longer average LOS, the studies reviewed in the GAO meta-analysis presented conflicting results (none of the studies included in the GAO meta-analysis analyzed psychiatric length of stay). In 1961, Roemer was surprised to find that LOS for medical and surgical care increased as a result of increases in hospital bed capacity because average LOS had been decreasing annually since the mid-1950s (Roemer, 1961). However, Gianfrancesco (1980) later found that the volume of medical and surgical admissions increased as capacity increased, but that average length of stay did not. Gianfrancesco (1980) concluded that increasing LOS was not widely used to generate additional demand. Robinson and Luft (1985) found that the average LOS for medical and surgical treatment in hospitals in very competitive markets (defined as five or more facilities within five miles) was affected by increases in capacity, but hospitals in more monopolistic markets did not increase LOS. Freund, Schactman, Ruffin, and Quade (1985) found that many factors, including demographic characteristics of patients (age, gender, diagnosis) and type of hospital ownership-as well as capacity changes-affected average LOS. Sloan and Valvona (1986) found that average lengths of hospital stay continued to decline as a result of changes in physician 61

PAGE 70

practice patterns, despite increases in hospital bed capacity. As a result of these conflicting findings, Freund et al. (1985) concluded that average length of stay for medical or surgical admissions was determined by a variety of variables and that the relationship between capacity and LOS is less clear than the relationship between capacity and volume of admissions. Psychiatric LOS Length of hospital stay for psychiatric or substance abuse treatment has been studied extensively; however, little consensus exists about which combination of variables best predicts LOS (Chang, Brenner & Bryant, 1991). Most researchers have concluded that variations in LOS were best explained by a combination of variables, including hospital type, diagnosis, type of reimbursement for services, age, and gender (Taube, Lee & Forthofer, 1984; Frank & Lave, 1985; Grazier & McGuire, 1987). The amount of variation accounted for by these variables is moderate, however. Although researchers have been attempting to predict length of psychiatric hospital stay for several years, no satisfactory model has yet been developed. Dorwart (1987) has suggested that additional variation in LOS may be explained by factors such as individual physician practice style, hospital-specific treatment philosophies, or other difficult-to-quantify variables. 62

PAGE 71

Hospital Type In an analysis of the 1986 National Client/Patient Sample Survey of Inpatient, Outpatient, and Partial Care Programs, the Survey and Reports Branch of the National Institute of Mental Health documented differences in LOS among hospital types, age groups, and diagnostic groups ( CMHS & NIMH, 1992). The fmdings showed that patients in public psychiatric hospitals had the longest median LOS (27 days), those in private psychiatric hospitals had a 21 day median WS, and patients in non-federal general hospitals had the lowest median LOS (12 days). Median LOS also varied by patient characteristics such as age, gender, and diagnosis. Differences in median LOS by hospital type and patient characteristics are illustrated in Table 3.1. 63

PAGE 72

Table 3.1 Median Days of Stay, Psychiatric Discharges: By Gender, Age, Selected Principal Diagnosis, and Hospital Type, United States, 1986* Public Private Non-Federal Patient Characteristics General Gender Male 27 25 9 Female 30 24 12 Age 18 and Under 43 41 12 19-24 27 24 7 25-44 28 20 10 45-64 24 21 12 65 and Over 43 18 15 Selected Principal Diagnoses Alcohol-Related Disorders 8 13 22 Drug-Related Disorders 27 27 8 Affective Disorders 33 25 13 Schizophrenia 38 20 13 Personality Disorders 9 18 7 Adjustment Disorders 9 18 7 Source: 1986 Client/Patient Sample Survey, Survey and Reports Branch, NIMH; Published in Mental Health, 1990. 64

PAGE 73

McGuire, Dickey, Shively, and Strumwasser (1987) also found significant differences in LOS among hospital types. Data used in this study were from insurance claims for inpatient psychiatric treatment submitted to a major Blue Cross/Blue Shield plan in 1981 and 1982. The authors found that LOS differed significantly by hospital type: General hospitals reported the lowest LOS while public and private psychiatric hospitals reported higher average LOS. Diagnosis The effects of diagnostic groups on psychiatric LOS have received much attention because diagnostic groups appear to explain a smaller proportion of the variance in LOS in the treatment of mental illness than they do in medical or surgical treatment. This has become an important issue because Medicare's prospective payment system (PPS) reform, is based on the average payment for a particular DRG, or diagnosis-related group. Researchers have found that psychiatric and alcohol/ drug abuse diagnostic groups account for approximately seven percent of the variation in LOS while diagnostic groups account for more than 30 percent of the variation in medical and surgical LOS (Taube et al., 1984). Taube et al. (1984) surveyed a national sample of patients admitted for treatment of mental disorders to public psychiatric hospitals, private psychiatric 65

PAGE 74

hospitals, and non-federal general hospitals to analyze the effects of diagnosis on length of stay. The authors found that seven percent of the variation in LOS was explained by diagnostic groups. Taube et al. (1984) also found that some diagnostic groups, such as childhood mental disorders, had a wide variation in the range of LOS. The authors concluded that the Medicare system of payment based on DRGs was unworkable for psychiatric diagnoses. Taube et al. (1984) noted that difficult-to-quantify variables (such as physician practice style, the individual hospital's treatment philosophy, patient resistance to treatment, and availability of psychiatric beds) may cause greater variation in LOS than clinical or demographic characteristics. Frank and Lave (1985) analyzed a national sample of Medicaid patients admitted to a variety of hospital types for psychiatric treatment. The authors confirmed Taube et al.'s early finding that DRGs account for approximately seven percent of the variation in psychiatric LOS. Frank and Lave (1985) also noted that LOS was higher for those diagnosed with schizophrenia or childhood mental disorders and lower for the alcohol or substance abuse diagnoses. According to Frank and Lave (1985), the state's Medicaid benefit structure also had an effect on LOS; states with restrictions on inpatient days, for example, had lower average LOS. In 1990, Lave and Frank extended their earlier study by including another set of variables in the analysis. The authors included region of the country as another set of independent variables. They 66

PAGE 75

reported that patients treated in the west had lower average LOS than any other region of the country. In a national study using a random sample of psychiatric discharge records from general hospitals, Wallen (1987) found that diagnosis was the most significant variable explaining variation in LOS in psychiatric patients. The author found that a diagnosis of psychosis was positively correlated with LOS while diagnoses of substance abuse, left against medical advice; alcohol or drug abuse treatment; neuroses except depression; adjustment reaction; and personality disorders were inversely related to LOS. Payer Groups In some studies, type of payment for services was a significant predictor of variation in LOS. Grazier and McGuire (1987) found that patients with any type of insurance, public or private, had higher average LOS than self-pay patients, although Medicaid patients also had lower-than-average LOS. Rupp, Steinwachs, and Salkever (1984) found that the adoption of per-case rather than per-diem reimbursement resulted in lower LOS. Freiman, Ellis, and McGuire (1989) found that the Medicare's adoption of a prospective payment system reduced average LOS by 25 percent. Grazier and McGuire (1987) compared the effects of payment source on the LOS of patients admitted to the psychiatric, medical, and obstetric 67

PAGE 76

services of one large urban hospital. Independent variables included patient age, gender, diagnosis, and source of payment. The total variation explained by this combination of variables-24 percent-was significantly higher than that reported by other studies of psychiatric LOS. One possible explanation is that all patients were treated in the same institution, thereby removing the effects of difficult-to-measure variables such as practice style or treatment philosophy. Regression results showed that payer group had a statistically significant effect on psychiatric LOS: Self-pay patients had the lowest average LOS; Medicaid patients also had lower-than-average LOS; patients with Blue Cross/Blue Shield insurance had. the longest LOS; followed by Medicare and commercially insured patients (Grazier & McGuire, 1987). Rupp et al. (1984) analyzed the effects of state rate-setting on LOS. The authors used a random sample of all hospitals providing treatment for mental illness in the state of Maryland from 1977 through 1980; the data included the period before and during the time when the state was initiating rate-setting. Rupp et al. (1984) concluded that rate-setting by using a per-case reimbursement system did reduce LOS; however, they also found that there was a 20 percent increase in admissions, possibly due to rapid readmission. The authors also found that patients in two age groups-18 and under, and 65 and over-had statistically significant longer LOS (Rupp et al., 1984). 68

PAGE 77

Freiman, Ellis, and McGuire (1989) used Medicare hospital discharge data for the years 1984 and 1985 to evaluate the effects of prospective payment on psychiatric LOS. The authors found that LOS declined by 25 percent in the two years following the adoption of prospective payment. Freiman et al. (1989) also found that 14 percent of the Medicare population was under 65 and receiving Medicare benefits through the Social Security Disability Insurance Program. Patients under 65 years of age had significantly lower LOS than those 65 and over (Freiman, Goldman & Taube, 1990). Additionally, patients with a diagnosis of psychosis or organic disorders had significantly longer LOS than patients in other diagnostic groups. Age and Gender Although demographic variables such as age and gender were included in virtually all the regression analyses, no study specifically focused on these variables. One consistent finding was that patients age 18 and under had significantly higher LOS than patients in other age groups (Rupp et al., 1984; Taube et al., 1984; Grazier & McGuire, 1987; NIMH, 1990a). NIMH (1990a) reported higher median LOS for females than for males. Despite some evidence that ownership type (public, private non-profit or private for-profit) also affects LOS (Schulz et al., 1984; Lyles & Young, 1987), the impact of a shift to investor-owned psychiatric hospitals -is not yet 69

PAGE 78

known. Public psychiatric hospitals continue to serve the most seriously and chronically mentally ill (NIMH, 1990a); therefore, the longer LOS found in this hospital type may be explained by the severity of illness of patients treated in these facilities. However, the reasons for the wide variations in LOS between general and private psychiatric hospitals are not yet clear. As private for-profit psychiatric facilities provide a larger proportion of inpatient treatment for mental illness, the longer length of stay typical of these facilities could lead to significant increases in length of stay and thereby increase mental health care costs. 70

PAGE 79

CHAPTER 4 METHODOLOGY This study investigates three principal hypotheses related to the rapid expansion of private for-profit psychiatric hospitals in Colorado following the deregulation of hospital construction in 1987. The period studied includes 1987 and 1989 with selected supplementary data from 1991. The hypotheses are summarized as follows: Hl: Roemer's Law predicts that utilization of hospital services will increase as the capacity to provide such services increases. It is hypothesized that the utilization of inpatient mental health services-as measured by number of discharges, length of stay, and total patient days-will increase as a function of an increase in hospital capacity. H2: Length of hospital stay for psychiatric and substance abuse treatment will vary by hospital type, controlling for differences in patient characteristics (age group, gender, and diagnosis), source of payment for services, and year in which service was provided. It is further hypothesized that LOS will be longest in private for-profit psychiatric hospitals, followed (in declining order of LOS) by private non-profit psychiatric hospitals, public general, private for-profit general, and private non-profit general hospitals.

PAGE 80

H3: An increase in the number of private psychiatric hospitals will result in "cream skimming"; that is, a diversion of privately insured patients from other types of hospitals providing inpatient treatment for mental illness to the private for-profit psychiatric hospitals. The study includes non-federal general hospitals and specialty psychiatric hospitals providing inpatient treatment of mental illness in Colorado. Hospital type is determined in two ways: By degree of specialization and by form of ownership. In this study, six hospital types are examined. Three are general hospitals: Public, private non-profit, and private for-profit. The other three are specialty psychiatric hospitals: Public, private non-profit, and private for-profit. 72

PAGE 81

Figure 4.1 General Hospital: Psychiatric Public: Non-Profit: For-Profit: Hospital Types Defined A hospital that admits patients for medical and surgical as well as psychiatric treatment. A psychiatric unit is an organization or administrative entity within a general hospital that provides services to patients with known or suspected psychiatric diagnoses. In a specialized unit, beds are set up and staffed specifically for psychiatric patients in a te ward or unit. A hospital that primarily provides care for mental illness. A general or psychiatric hospital controlled by state, county, and/or city governments, or by district/regional authorities. A general or psychiatric hospital owned by a foundation, church, or other non-profit group. A general or psychiatric hospital owned by a corporation, partnership, or individual and operated on a for-profit basis. Study Design Hypothesis testing was completed in three stages. An interrupted time series design was used to test the impact of changes in private psychiatric hospital capacity on the utilization of inpatient psychiatric and substance abuse treatment during the study period. Key measures of hospital utilization-the number of patient discharges, length of hospital stay, and total patient days-were described by hospital type. Regression analysis was used to measure time trends in length of stay and to evaluate the relative difference in 73

PAGE 82

LOS among hospital types, holding constant patient characteristics including age group, gender, diagnosis, and source of payment. Payer mix was compared across hospital types for the years 1987 and 1989 to determine whether the growth in the private for-profit hospital sector led to cream skimming; that is, the diversion of privately insured patients from other hospitals providing psychiatric and substance abuse treatment to private investor-owned hospitals. The purpose of an interrupted time-series design is to evaluate whether an intervention had an impact on the problem being addressed (Cook & Campbell, 1979). This technique was used to assess the impact of changes in hospital capacity on the utilization of psychiatric and substance abuse treatment during the study period. Cook and Campbell (1979) have identified a number of problems associated with the interrupted time-series technique that raises potential threats to the validity of that type of research design. These problems were summarized by the authors as follows: The response to the intervention may be delayed; the effects of the intervention may have unpredictable time delays that may vary among populations; most studies have a limited number of observations, although up to 50 observations are recommended; and data may be missing or incomplete. Whenever possible, threats to validity were addressed in the study design. The problem of a delayed response to capacity change was addressed 74

PAGE 83

by including the number of discharges for an additional year, 1991, in the analysis. This also provided an additional test of the hypothesis that hospital bed capacity would induce demand for hospital services because the number of private psychiatric hospitals increased between 1987 and 1989, then declined slightly in 1991. The problem of unpredictable time delays is not relevant to this study. Data were only available for the years 1987, 1989, and 1991. Seasonal variations in the use of inpatient mental health services precluded the use of quarterly rather than annual time intervals. As in any research study of this kind, the possibility that the changes observed during the study period may have been caused by factors other than those studied is a major threat to study validity. In the analyses of the effects of hospital type on LOS, multiple regression analysis was used to control for differences in patient characteristics among hospital types. Whenever possible, researchers, administrators, and clinicians were interviewed to provide additional insight on specific questions raised by this study. Difficult-to quantify factors such as changes in third'-party payer policies during the study period and changes in methods of treatment for mental illness could also have had significant impacts on the utilization of inpatient psychiatric and substance abuse treatment during the study period. The potential effects of these changes are discussed in later sections of this report. 75

PAGE 84

Data Sources Three distinct data sources were used in these analyses. The primary data source was the Uniform Hospital Discharge Data Sets for 1987 and 1989 which was obtained from the Colorado Health Data Commission ( CHDC). The Hospital Discharge Data Sets included patient-level information for patients discharged with a psychiatric or substance abuse diagnosis from Colorado general, public psychiatric, and private psychiatric hospitals in 1987 and 1989. The total number of psychiatric and substance abuse discharges reported to CHDC in 1991 by individual hospitals was also obtained from CHDC. The discharge data sets were supplemented by aggregate, hospital level data from the American Hospital Association's Annual Survey of Hospitals for 1987 and 1989 and aggregate, private psychiatric hospital data from the Medicare Cost Reports for 1987, 1989, and 1991. The Directory of Colorado Health Facilities. 1990, which provided a complete listing of licensed hospitals in Colorado, was used to construct a master list of hospitals. Hospitals in all data sets were classified into one of six hospital types according to degree of specialization and type of ownership based on information obtained from the Directocy. Psychiatric and/or substance abuse discharges from federal hospitals (Veteran's Administration hospitals) and hospitals specializing in the care of children, rehabilitation, or chronic lung disease were excluded from the analysis because psychiatric 76

PAGE 85

patients served by these hospital types often require medical, surgical, or rehabilitation treatment in addition to psychiatric treatment. It should be noted that most researchers in the field of mental health services delivery exclude patients from these specialty hospitals when analyzing changes in the delivery or financing of inpatient psychiatric or substance abuse treatment. CHDC Discharge Data Set Patient-level data used to test hypotheses two and three were only available from the CHDC Hospital Discharge Data Sets for 1987 and 1989; the 1988 data set was incomplete and was therefore not used in the analyses. The CHDC data sets included records for patients hospitalized for psychiatric or substance abuse treatment in general hospitals with 50 or more beds, public psychiatric hospitals, and private psychiatric hospitals. Each discharge record contained detailed information on patient characteristics such as age, gender, diagnosis, expected source of payment for services, length of stay, charges, and hospital in which service was provided. The information obtained from the CHDC discharge data sets provided an exceptionally detailed, relatively thorough description of patients admitted for psychiatric or substance abuse treatment in Colorado in 1987 and 1989. The data sets did have a number of limitations, however. An analysis of the 1988 data revealed a number of serious weaknesses resulting from 77

PAGE 86

administrative problems at the Data Commission in 1988. Accordingly, the 1988 data were not used in the analyses. The major limitation of the CHDC Discharge Data Sets was that only 12 percent of the 1987 discharges from private psychiatric hospitals and 29 percent of the 1989 discharges from this hospital type are included in the data sets. To check on possible sample bias, discharge data from private for-profit psychiatric hospitals included in the discharge data set were compared to data available from the Medicare Cost Reports as well as from national studies (Redick et al., 1989; NIMH, 1990a; Dorwart et al., 1991 ). The demographic characteristics and payer mix of the patients included in the private for-profit psychiatric hospital discharge records in the CHDC data set are similar to those reported in the national studies for this hospital type. Further, Medicare Cost Reports indicate that the payer mix of the private for-profit psychiatric hospitals that did report to CHDC is nearly identical to that of the non reporting private psychiatric hospitals. Additionally, some of the reporting hospitals did not provide complete information on source of payment for services. Public psychiatric hospitals were not required to report payment information and did not do so. A few private for-profit general hospitals and private non-profit psychiatric hospitals reported inaccurate source of payment information in 1987, resulting in a large number of discharges in the "unknown" payer category in 1987; however, this 78

PAGE 87

problem was corrected in 1989. Other sources of data were used to supplement the CHDC data sets whenever possible. Supplementacy Data Sources The American Hospital Association's Annual Survey of Hospitals for 1987 and 1989 and individual private psychiatric hospital Medicare Cost Reports for 1987, 1989, and 1991 were used to supplement the Hospital Discharge Data Sets. These sources reported information, such as the aggregate number of psychiatric or substance abuse discharges and total number of patient days, by individual hospitals. Information on the number of beds available to treat mental illness (staffed beds) in each hospital providing such services was also obtained from the Annual Survey of Hospitals. Detailed financial statements were submitted to Medicare by each psychiatric hospital and selected hospitals reported estimated payer mix. However, no detailed, patient-level information was available from either source. Over half of the private psychiatric hospitals did not comply with the mandated CHDC data reporting requirements; however, ten out of the state's 11 private psychiatric hospitals did file Medicare Cost Reports for 1987, 1989, and 1991. Medicare Cost Reports were available through the Freedom of Information Act from the Provider Audit Department of Blue Cross/Blue Shield of Colorado (Medicare's fiscal intermediary in Colorado). 79

PAGE 88

Data Analyses Data analysis was completed in three stages. First, changes in utilization were described by changes in the number of discharges, the number of total patient days, and the average length of hospital stay for all hospitals types combined, then further described by hospital type. Second, a series of regression analyses were performed to measure the relative importance of hospital type and year of service on the dependent variable, length of hospital stay, controlling for patient characteristics such as age, gender, diagnosis, and source of payment. Third, changes in payer mix and patterns of patient allocation were compared across hospital types for the years 1987 and 1989. Changes in Utilization Three measures of utilization-volume of discharges, average length of hospital stay, and total patient days-were used to describe changes occurring during the study period. Changes were also described and compared by hospital type. Three distinct data sources, the CHDC Discharge Data Sets, AHA Annual Survey of Hospitals, and Medicare Cost Reports submitted by private psychiatric hospitals, were used in the analysis to provide a comprehensive picture of utilization during the study period. Data describing the volume of discharges for the year 1991 were obtained from the Colorado Health Data Commission and the Medicare Cost Reports for 1991. Data for 80

PAGE 89

1991 were included in the analysis of change in volume of discharges because private psychiatric hospitals decreased slightly in that year. The inclusion of an additional year allowed a more thorough assessment of the capacity changes on volume of services provided. Frequency distributions were constructed for all hospitals within the CHDC data sets for 1987, 1989, and 1991. The volume of discharges, average length of stay, and total patient days for all hospitals missing from the CHDC data set were calculated using aggregate data obtained from the AHA Annual Survey reports and Medicare Cost Reports, and then added to the totals obtained from the CHDC data sets. As a result, the analysis of changes in utilization inducted all general hospitals, all public psychiatric hospitals, and ten out of 11 private hospitals providing inpatient psychiatric and substance abuse treatment in 1987 and 1989. Variables Affecting LOS A series of regression analyses were used to measure changes in length of hospital stay during the study period and to determine whether length of stay differs among hospital types, controlling for differences in patient characteristics among the hospital types. The CHDC data sets are the only source of patient-level data and are therefore, the only data sources used in the regression analyses. 81

PAGE 90

Seven regression analyses were included in the study. The first regression included discharges from public, private non-profit, and private for profit general hospitals and private nonand for-profit psychiatric hospitals. Hospital type was an additional independent variable in this analysis. Public psychiatric hospitals were excluded from the first regression (which included all other hospital types) because public psychiatric hospitals are not required to report source of payment information, an important variable affecting LOS in all other hospital types. Separate regression analyses were done for each hospital type, including public psychiatric hospitals. Length of stay was the dependent variable. Six classes of independent variables were used in the analyses and are described in Table 4.1. 82

PAGE 91

Table 4.1 Variables Included in Regression Analyses Age: Reference Group = 25-44 0-18 Years 19-24 Years 45-64 Years 65 and over Year: Reference Group = 1987 1989 Gender: Reference Group = Female Male Expected Source of Payment: Reference Group = Commercial Insurance Self-Pay Worker's Compensation Medicare Medicaid Other Government Blue Shield Diagnosis-Related Groups: Reference Group = 430 = Psychosis 424: Mental Disorder, Operating Room Procedure 425: Adjustment Reaction 426: Depression 427: Neuroses Except Depression 428: Personality Disorders, Impulse Control 429: Organic Disturbance or Mental Retardation 431: Childhood Mental Disorders 432: Other Mental Disorders 433: Alcohol/Drug Abuse Treatment/Left Against Medical Orders 434: Abuse Treatment with Detoxification Hospital Type: Reference Group = Private Non-Profit General Hospitals Public General Hospitals Private For-Profit General Private Non-Profit Psychiatric Private For-Profit Included only in regression analysis that includes both general hospitals and private psychiatric hospitals (Table 5. 7). 83

PAGE 92

Method Used. Both direct and stepwise (forward) methods were used. Results of stepwise analyses are presented in this report. Entry criteria (FIN and POUT) used in the forward method were changed to FIN= .005 and FOUT=.OOl. The independent variables used in the regressions were transformed into a series of dummy variables; a listing of the recoded dummy variables is appended (Appendix A). Within each class of variables, the variable with the largest proportion of discharges was selected as the reference group. LOS and Independent Variables. The hypothesized relationships between LOS, the dependent variable, and the independent variables are based on a review of literature on length of hospital stay for psychiatric and substance abuse diagnoses. In three classes of variables (age group, payer group, and diagnostic group) the expected direction of relationship between the dependent variable and the independent variables differs within the class. For example, among the age categories, patients 18 and under consistently exhibited significantly longer LOS than other age groups (Rupp et al., 1984; Taube et al., 1984; NIMH, 1990a). Rupp et al. (1984) and Freiman et al. (1987a) also report longer-than-average LOS for patients 65 and over. Among the payer groups, self-pay, Medicaid, and Worker's Compensation have shorter-than-average LOS, while Blue Cross/Blue Shield, Medicare and other 84

PAGE 93

government payer groups have longer-than-average LOS (Grazier & McGuire, 1987). LOS also varies according to diagnosis. More serious psychiatric diagnoses such as psychosis, childhood mental disorders, and organic disturbance/mental retardation generally require longer periods of hospitalization (Taube et al., 1984). The combined psychiatric/medical diagnosis "mental disorder, operating room procedure" also results in longer-than-average LOS (Robert Cowan, M.D., personal communication, 1993). Diagnoses such as depression and adjustment reaction are generally characterized by short LOS (Taube et al., 1984). Almost by definition, patients with a diagnosis of "alcohol/drug abuse treatment/left against medical advice" could be expected to have shorter-than average LOS. The independent variable "1989" is a proxy variable for the change in capacity that occurred in that year and is, therefore, expected to have a direct relationship to LOS (Roemer, 1961). The hypothesized directions (direct or inverse) of the dependent variable, LOS, to the independent variables (relative to the reference groups) are summarized in Table 4.2. 85

PAGE 94

Table 4.2 Hypothesized Direction of Relationship Between Dependent and Variables Age: Reference Group = 25-44 0-18 Years 19-24 Years 45-64 Years 65 and over Year: Reference Group = 1987 1989 Gender: Reference Group = Female Male Expected Source of Payment: Reference Group = Commercial Insurance Self-Pay Worker's Compensation Medicare Medicaid Other Government Blue Cross/Blue Shield Diagnosis-Related Groups: Reference Group = 430 = Psychosis 424: Mental Disorder, Operating Room Procedure 425: Adjustment Reaction 426: Depression 427: Neuroses Except Depression 428: Personality Disorders, Impulse Control 429: Organic Disturbance or Mental Retardation 431: Childhood Mental Disorders 432: Other Mental Disorders 433: Alcohol Abuse/Left Against Med. Advice 434: Detoxification Hospital Type: Reference Group = Private Non-Profit General Public General Hospitals Private For-Profit General Hospitals Private Non-Profit Psychiatric Hospitals Private For-Profit "tals Direct Inverse Inverse Direct Direct Inverse Inverse Inverse Direct Inverse Direct Direct Direct Inverse Inverse Inverse Inverse Direct Direct Inverse Inverse Inverse Direct Direct Direct Direct Included only in regression analysis with both general hospitals and private psychiatric hospitals. 86

PAGE 95

Limitations. As previously discussed, the CHDC Discharge Data Set bas a major limitation: The majority of private for-profit psychiatric hospitals did not rep rt data. Information from the Medicare Cost Reports also indicates th t average length of hospital stay was slightly longer in the nonreporting p vate psychiatric hospitals than in the private psychiatric hospitals included in the regression analyses. Cream Ski In t e third analysis, changes in payer mix and patterns of patient allocation ere compared among hospital types to determine whether privately insured pat ents were diverted to for-profit psychiatric hospitals. The CHDC Discharge ata Sets for 1987 and 1989 were used in the analysis. There are seven paye groups included in the data set: Self-pay; Worker's Compensat on; Medicare, Medicaid, other government; Blue Cross/Blue Shield, and private insurance. Cha ges in patient mix are described by gender, age category, and diagnostic roup. Patient age was recoded into one of five categories: 18 and under; 19-14; 25-44; 45-64; and 65 and over. These categories reflect age breakdownk commonly used in the literature (NIMH, 1990a). A description of diagnostic oups and corresponding DRG numbers are appended (Appendix A). 87

PAGE 96

CHAPTER 5 FINDINGS This tudy analyzes the impacts of the doubling of the number of private psyc iatric hospitals in Colorado between 1987 and 1989 on the state's inpatient me tal health care system. Three major hypotheses were tested. The first hyp thesis, based on Roemer's Law, predicts that the increased capacity to p ovide inpatient psychiatric and substance abuse treatment will result in incr ased utilization of such services. Second, it is hypothesized that average len of stay will be longer in private psychiatric hospitals than in general hosp als, holding patient characteristics and payer source constant. Finally, it is ypothesized that private for-profit psychiatric hospitals cream skim; that is, ivert privately insured patients from general hospitals and thereby threa en the financial viability of general hospital psychiatric services. Study esults indicate that the utilization and costs of psychiatric and substance ab se treatment were significantly affected by the growth in private investor-owne psychiatric hospitals in Colorado during the study period. The number of pa ients discharged after treatment for mental illness increased as private psychittric hospital capacity doubled between 1987 and 1989, then declined whe the number of private psychiatric hospitals decreased slightly in 1991. Averag LOS, however, decreased in all hospital types but public general hospit Is in 1989, suggesting that LOS was not used to generate

PAGE 97

additional emand. Reduction in LOS varied considerably by hospital type with public psychiatric hospitals reporting the largest decrease. A shift in locus of care fro general hospitals, characterized by short average LOS, to private psychiatric ospitals, with longer average LOS, raised the costs of mental health care in the state during the study period. The financial base of non profit gene al hospitals, the major providers of inpatient mental health treatment, as weakened by the diversion of privately insured patients to private psy hiatric hospitals. Testing Roemer's Law A m jor study hypothesis, based on Roemer's law, predicts that the inpatient ut ization of psychiatric or substance abuse treatment will increase as the capa 'ty to provide such services increases. To test this hypothesis, changes in tilization were measured in three ways: The number of hospital discharges, he mean length of hospital stay, and the number of days of hospital car provided to patients with psychiatric diagnoses (termed "patient days"). Thr e distinct data sources were used to provide a comprehensive picture of c anges in the utilization of inpatient psychiatric services during the study perio I othesized, the number of patients discharged with a psychiatric or substanc abuse diagnosis increased in 1989 when the capacity to provide 89

PAGE 98

services in reased; 1989 discharges increased 13 percent over 1987 (Table 5.1). Moreover, hen the number of private psychiatric hospitals decreased in 1991, the numbe of discharges declined by six percent. However, average length of hospital st declined between 1987 and 1989, despite capacity increases (Table 5.5) There was no apparent Roemer effect on length of stay. Total patient day the product of the number of discharges multiplied by the average le gth of stay, also declined in 1989. These changes are discussed in more deta' in the following sections and are summarized in Tables 5.1 Increase in Dischar es The e was a 13 percent increase in the total number of hospital discharges f psychiatric patients in Colorado between 1987 and 1989. This was follow d by a six percent decline in 1991.. Changes in discharges were preceded b increases in psychiatric bed capacity in 1989, then a decline in 1991. Duri g this same time period, the state's population increased by sixtenths of o e percent (Demographic Section, Colorado Division of Local Governme t, 1991). The number of discharges reported by hospital type were as follows: 90

PAGE 99

Table 5.1 Number of Discharges by Hospital Type % 1987 1989 Change 1987-1989 10.3% 6.7% 1991 % Change 1989-1991 not available from one hospital included in the 1987-1989 summary. discharged approximately 200 psychiatric patients per year. "ority of the increase in the volume of discharges between 1987 the number of patients discharged from private psychiatric hospitals more than doubled, 3,150 to 7,030. Discharges from private psychiatric hospitals 26 percent, with non-profit psychiatric hospitals reporting a blic psychiatric hospitals and public and private non-profit general there was a small increase in discharges from for-profit general hospitals. In 91

PAGE 100

1991, the number of discharges from public psychiatric hospitals increased 36 percent over 1989. Public general hospitals showed a decrease in discharges in 1989. Private general hospitals (both for-profit and non-profit) reported a slight increase in discharges in 1991. The changes in the percentage of discharges from each hospital type are illustrated in Table 5.2. 92

PAGE 101

Table 5.2 Percentage Psychiatric Discharges by Hospital Type Hospital Type Percent of Total Public General Hospitals 1987 1.::= << '== = .:: =>=:::::,: I 13.6% 1989 10.8% 1991 10.3% ,1' - Private Non-Profit General Hospitals 1987 I+=,:=:::::::::::::-:=' .. ==== ,=:== .. =::::::::>:=== : .. :::::.,::::,::<:' .. .:>=::::::=:::, : ...... ,. <=::::<, => =:::::.:::'==:==:<=====<=::i::<45,7%1 1989 37.8% 1991 40.9% Private For-Profit General Hospitals 1987 I ,,;/=. .. I 10.8% 1989 lmtmrrttttt:::t:tt:J 10.2% 1991 11.4% Public Psychiatric Hospitals 1987 L:=: : => ,,:, ,,: .. = :1 13.1% 1989 ntwn=:s:rwwm::l 8.0% 1991 11.5% Private Non-Profit Psychiatric Hospitals 1987 I =>>=:::=,::::: ::, ,:: I 1989 1991 Private For-Profit Psychiatric Hospitals 1987 1,=:::=::::::::= =::::. :I 8.8% 12.5% 6.6% 1989 t:ww:wnm:::::::tq;nwtt.n;mm:::::;wrnwn;wm::::::l 1991 93 8.0% 20.7% 19.4%

PAGE 102

Changes in Patient Days Hospital utilization is also measured by the total number of hospital days for all patients receiving inpatient psychiatric treatment in a specific period of time; that is, patient days. Total patient days are the product of the number of discharges multiplied by average length of stay. Comparison of patient days is particularly important when analyzing utilization if, for example, the number of discharges increases while the average length of stay decreases as occurred between 1987 and 1989. Data on the number of patient days and LOS for general and public psychiatric hospitals were not available for 1991. Between 1987 and 1989, total patient days declined from 454,503 to 403,298 days, a decrease of 11.3 percent. During the study period, long-stay public psychiatric hospitals reported a 44 percent drop in patient days. There was a marked shift in locus of care from general hospitals with low average lengths of stay, to private psychiatric hospitals with characteristically higher lengths of stay. The number of patient days in private for-profit psychiatric hospitals increased 176 percent during the study. Variations in the number of patient days by hospital type are shown in Table 5.3. 94

PAGE 103

Table 5.3 Number of Patient Days by Hospital Type Hospital Type Patient Days 1987 Patient Days 1989 % Changes Public General 29,253 26,799 -08.4% Private Non-Profit General 107,119 85,504 -20.2% Private For-Profit General 36.182 31.579 -12.7% :::;:::=:::::::::::::::::::;:;::;:<;::: ..:::;:;::::::::::::-..---:::-...,. ; :-:-:-:-:-:-::;:-:;:;:::;:;:;:;:;:;:;::;:;:;:;: .-.. .... -----................. .-.. .. .--:-:-;:-:-:--:-:-.-:;:_';: -:;.-:_-:;:;:-:; ..-.-..... ::. -...,..:::;:;:; .. :;:-... :. ,;,:=::=.=/_ .::: .:-i,[ :: .. ..Public : : = <:, : .: .. :2o2J)92 < .-= :,::: .:.: :::::_::: :::::::::. :_ .,, :;"=:i-Private NonProfit Psychiatric Private For-Profit Psychiatric 43,276 35.681 46,790 98.552 08.1% 176.2% When patient days are used to compare the relative market share of each type of facility providing inpatient psychiatric treatment, each major sector accounted for approximately one-third of the total market by 1989; private psychiatric hospitals (including both for-and non-profit oWnership types) accounted for 36 percent of the total; general hospitals (including public, private non-profit and private for-profit ownership types) accounted for 35.7 percent and public psychiatric hospitals accounted for 28.3 percent. Using patient days as a measure, the shift in locus of treatment to the private psychiatric hospitals is illustrated in Table 5.4. 95

PAGE 104

Table 5.4 Percentage of Total Patient Days by Hospital Type Hospital Type Percent of Total Public General Hospitals 1987 I = <:. =:::I 6.4% 1989 IUHMMtittl 6.7% Private Non-Profit General Hospitals 1987 ,.:.::::::.<;::: ::.= ): Private For-Profit General Hospitals 1987 I = =: .::.:::=:::=:! 1989 NM&tmttlHH Public Psychiatric Hospitals 1987 1989 8.0% 7.8% Private Non-Profit Psychiatric Hospitals 1987 I =::::.===:=:::.=.:1 1989 [WtMMMJM#.lttd&MI 9.5% 11.6% Private For-Profit Psychiatric Hospitals 1987 1:::=:=.:?: \ >:::1 198 9 96 23.6% 21.2% 7.9% 24.4%

PAGE 105

It should be noted that capacity discharges from public psychiatric hospitals are largely driven by the institutions' ability to place patients in appropriate community settings. The declining number of discharges in 1989 led to greater efforts to outplace patients in 1990 and 1991, which increased discharges during those years (Richard Ellis, Researcher, Colorado Department of Institutions, personal communication, 1993). Additionally, total public psychiatric hospital patient days may be understated because this measure does not include patient days of long-stay patients (patients with a length of stay greater than one year) who were not discharged in either study year. In both 1987 and 1989, less than four-tenths of one percent of all patients discharged from all hospital types had a length of hospital stay greater than one year: Public psychiatric hospital patients accounted for virtually all of these outliers. For both years, public psychiatric hospitals reported lengths of stay of greater than one year for three percent of the total discharges: The largest single length of stay reported by this hospital type was 8,051 days in 1987. Decreases in Length of Stay Despite the increase in capacity between 1987 and 1989, mean length of stay for psychiatric treatment declined overall, from 24.32 days in 1987 to 19.09 days in 1989, a decrease of 5.23 days. LOS declined in all hospital types 97

PAGE 106

except public general hospitals. During the study period, changes in average length of stay varied by hospital type as shown in Table 5.5. Table 5.5 Mean Length of Stay by Hospital Type LOS, 1987 LOS, 1989 The most dramatic reduction in mean length of stay occurred in the public psychiatric hospitals where average length of stay decreased by 15.11 days. One or more large outliers in the 1987 data may account for a significant percentage of this reduction in average length of stay: The maximum length of stay for this hospital type was 8,059 days in 1987 but 1,824 days in 1989. However, when public psychiatric hospitals are excluded from the analysis, average length of stay decreased only .59 days between 1987 and 1989. This modest overall reduction reflects a shift in volume of discharges 98

PAGE 107

from general hospitals with low average lengths of stay to private psychiatric hospitals with significantly higher average lengths of stay. Had this shift not occurred, the overall length of stay decline would no doubt have been greater. Type of hospital ownership (public or private, for-profit or non-profit) as well as degree of specialization (general or psychiatric) were shown by the study data to be important determinants of average length of hospital stay. Well over half of all psychiatric discharges reported during the study period were from general hospitals (including public, private non-profit, and private for-profit ownership types). These, in total, showed an average 1.56 day decline in length of stay: Public general hospitals reported a nominal .26 day increase; private general non-profit hospitals reported a reduction of 1.8 days; and private general for-profit hospitals reported a reduction of 3.23 days. Private psychiatric hospitals also reported differences in the decrease in average length of stay based on type of ownership; non-profit psychiatric hospitals showed a decrease of 8.5 days in average length of stay while for profits showed a decrease of 1.4 days. Average Daily Charges Length of hospital stay is an important determinant of the costs of inpatient treatment for mental illness. As previously discussed, average LOS varies by type of hospital providing service. Average daily charges, as reported 99

PAGE 108

in the CHDC hospital discharge data sets for 1987 and 1989, illustrate the variations in the costs of providing treatment resulting from differences in WS among the hospital types. Average daily charges are based on fully allocated costs, not on revenues actually collected by hospitals. While this measure of hospital costs has limitations, it does provide a rough estimate of cost impacts caused by variation in LOS among hospital types. Public psychiatric hospitals do not submit comparable cost data, so average daily charges are not available for this hospital type. Table 5.6 Average Daily Charges by Hospital Type, 1987 and 1989. All Except Public Psychiatric $476. $637 33.8% Public General $431 $635 47.3% Private Non-Profit Generals $489 $657 34.4% Private For-Profit Generals $489 $637 30.3% Private Non-Profit Psychiatric $445 $556 24.9% Private For-Profit Psychiatric $502 $699 39.2% Capacity Changes. Hospital Type. and WS Additional analyses were done to determine whether a Roemer effect occurred in LOS, but was masked by changes in the characteristics of patients treated for mental illness during the study period. Using the CHDC Hospital Discharge Data Sets for 1987 and 1989, a series of regression analyses were done to determine the relative importance of hospital type and changes in 100

PAGE 109

capacity on LOS, controlling for other variables such as age, gender, source of payment, and diagnosis. Seven separate regression analyses were done. The first analysis included all hospital types but public psychiatric hospitals which were excluded from this analysis because source of payment infonnation was not available for this hospital type. Hospital type was used as a separate class of independent variables only in the aggregate analysis. Using private non-profit general hospitals as a reference group, the effects of hospital type on LOS are summarized in Table 5.7. Individual regression analyses were also done for each hospital type, including public psychiatric hospitals. Results are compared across hospital types in Table 5.8. Effects of Hospital T)l)e on LOS As hypothesized, hospital type was found to be an important predictor of LOS, controlling for year of service, payer type, and patient characteristics. It was further hypothesized that the reference group, non-profit general hospitals, would have shorter average LOS than the other hospital types included in the analysis. This hypothesis was also confinned. The LOS in private for-profit psychiatric hospitals was, on average, 11.3 days longer than the reference group LOS. Also, because the number of discharges from private for-profit psychiatric hospitals increased by 190 percent 101

PAGE 110

between 1987 and 1989 (Table 5.1), the longer average LOS in this hospital type resulted in significant cost increases for inpatient treatment of mental illness in Colorado during the study period. LOS in private non-profit psychiatric hospitals and private for-profit general hospitals was also significantly longer than in the reference group, while LOS in public general hospitals was only slightly longer than LOS in private non-profit general hospitals. Results are summarized in Table 5.7. Table 5.7 Effects of Hospital Type on LOS: All But Public Psychiatric Hospitals Private For-Profit Psychiatric Hospital Private For-Profit General Public General Hospital Ps .0001 ** Ps .005 The Roemer Effect and LOS Regression Coefficient 11.34* 6.23* 3.98* 1.10** Study findings do not support the hypothesis, based on Roemer's Law, that LOS would increase as a function of increases in capacity. The effects of capacity changes on LOS were analyzed by using the year 1989 as a proxy variable for capacity increases. Holding source of payment and patient 102

PAGE 111

characteristics constant, LOS declined in all hospital types except public general hospitals in 1989 (Table 5.8). Any possible Roemer effect was masked by other, more significant factors that influenced LOS. The finding that LOS decreased in 1989 is consistent with national trends. Nationally, average LOS for psychiatric and substance abuse treatment has been declining annually for the past several years (Cooper, 1993). Cooper (1993) noted that more restrictive payer policies and changes in methods of treatment have driven these reductions. As the costs of providing benefits for psychiatric and substance abuse treatment grew, payers increasingly turned to utilization review and other managed care strategies to contain costs (Lutz, 1992). Moreover, other studies have raised questions about the value of long hospital stays for the treatment of mental illness (Lave, 1989). Alternative and less expensive forms of treatment, such as partial hospitalization or outpatient treatment, are now used more frequently (Lutz, 1992). When used in conjunction with inpatient care, these new treatment methods permit earlier discharge and thereby decrease LOS. Findings are summarized in Table 5.8. Table 5.9 describes the average LOS for each independent variable by hospital type. Individual regression analyses are appended in Appendix B. 103

PAGE 112

Table 5.8 Length of Stay for Psychiatric Discharges: Regression Coefficients Compared Across Hospital Types Except Private Private Private Private Independent Variable Non-For-Non-ForPsych. Public Profit Profit Public Profit Profit Gender: Ref. Group Female Male I N/A I 0.26 I -0.21 I 0.42 I 10.65 I -1.63 I 1.28 Age: Ref. Group 25-44 0-18 Years 11.06* 1.18 8.64* 14.76* 36.48** 21.55* 15.60* 11 19-24 Years 1.05** 0.88 0.63 1.01 5.36 2.93 1..59 45-64 Years 0.34 7.06 -0.51 -0.72 65 + Years -0.62 -2.62** -1.71 -1.88 4.38** 10.16*** Year: Ref. Group 1987 1989 I -1.96* I 0.12 I -1.66* I -3.16* I -17.13** 1 -2.15** I -7.93*** Source of Payment: Ref. Grp: Private Insurance Self-Pay -2.66* 3.21* -5.50* N/A -2.10 -11.61 ** Worker's Comp. 2.45*** -2.10 .;1.82 2.65 22.57* 12.08 Medicare 1.40** 4.12* 0.97* *"' .58 N/A -1.75 Medicaid 0.23 4.68* -1.12* 5.76* -3.37 2.99 Other Government 0.91 1.81 0.75 6.04*** 0.75 N/A Bl Cross/Bl Shield 2.64* -1.72 0.70 2.84*** 8.07* 5.81***

PAGE 113

...... 0 VI Table 5.8 (cont.) Length of Stay for Psychiatric Discharges: Regression Coefficients Compared Across Hospital Types Except Private Private Private Private Independent Variable Public NonForNonForPsych. Public Profit Profit Public Profit Profit Diagnostic Groups: Ref. Group: Psychosis Ment. Disorder/OR 1.21 12.34* 0.41 -3.54 106.70** N/A N/A Adj. Reaction -6.35* -6.30* -7.65* -9.31 45.46 0.92 N/A Depression -4.40* -5.12* -5.29* -2.92** -63.17*** -4.43** N/A Neuroses Exc Depression -7.15* -3.51 *** -7.47* -3.14 -56.85** -10.37* -0.78 II Personality Disorder -1.12 -5.24* -2.49** -2.24 32.48*** 4.01 7.56 Organic Disturb /MR -3.63* -3.97* -7.04* -23.46 -1.09 4.51 Child. Ment. Disord. 0.31 -3.58** -5.05* 7.17** -2.09 -2.38 13.59* Other Ment. Disord. 10.83* -8.61 *** 11.56* Alcohol/Drug Abuse 11.33* 83.88 -13.35 10.39 /left A.M.A. I -8.21 I -9.42* I -7.37* I -12.00* I -55.96 I N/A I -9.97 Alcohol/Drug Abuse with Detox. I -2.07* I -5.20* I -1.93* I -2.06 I -26.27 I 1.62 I -4.68 Ps .0001 ** Ps .005 *** Ps .05

PAGE 114

Table 5.9 Average Length of Stay for Psychiatric Discharges: By Independent Variable and Hospital Type Except I Private Private I Private Private Independent Variable I Public NonForNonFor. Public Profit Profit Public Profit Profit Gender: Ref. Group Female 13.87 11.47 11.80 16.40 67.78 17.58 23.62 Male 13.74 11.96 11.77 15.29 82.39 19.78 26.86 Age: Ref. Group 25-44 11.57 11.77 10.32 13.23 67.02 13.10 16.62 0-18 Years 22.83 12.42 18.01 28.72 105.98 32.15 36.71 19-24 Years 12.39 12.21 13.89 69.11 15.73 19.44 .,_. 11 45-64 Years 11.95 11.96 11.06 13.95 76.63 13.03 16.48 65 +Years 11.34 9.59 10.01 13.23 66.06 17.39 26.05 Year: Ref. Group 1987 13.82 11.57 12.68 17.62 82.45 19.10 34.08 1989 13.79 11.93 10.80 14.70 67.34 17.86 23. 94 Source of Payment: Ref. Grp: Private Insurance 15.42 9.01 13.15 15.95 N/A 17.98 24.52 Self-Pay 9.40 12.11 6.55 12.34 11.90 11.64 Worker's Comp. 13.40 6.25 9.43 12.85 35.46 27.11 Medicare 12.14 11.61 11.36 13.14 15.43 21.46 Medicaid 14.51 14.06 13.21 22.73 19.00 41.36 Other Government 16.37 9.62 13.96 25.34 18.48 27.53 Bl Cross/Bl Shield 18.09 7.23 14.02 18.65 25.90 29.74

PAGE 115

Table 5.9 Average Length of Stay for Psychiatric Discharges: By Independent Variable and Hospital Type (cont.) Except Private Private Private Private Independent Variable Public NonForNonForPsych. Public Profit Profit Public Profit Profit Diagnostic Groups: Ref. Group: Psychosis 14.67 15.98 12.26 16.06 80.84 17.49 22.63 Ment. Disorder/OR 12.75 25.15 12.11 15.43 189.06 -0-0Adj. Reaction 7.87 6.69 5.37 7.81 32.31 17.72 18.85 Depression 13.13 8.37 9.69 17.55 18.66 20.64 27.53 II Neuroses Bxc Depression 10.43 10.20 7.45 14.86 26.08 11.69 23.67 0 Personality Disorder 14.28 8.20 11.20 15.54 53.66 23.23 34.88 -.....1 Organic Disturb /MR 9.68 9.72 8.46 8.85 50.63 17.22 25.93 Child. Ment. Disord. 25.55 15.10 15.29 36.73 114.19 32.55 46.40 Other Ment. Disord. 26.38 4.50 26.39 28.73 169.05 16.60 29.50 Alcohol/Drug Abuse /left A.M.A. Alcohol/Drug Abuse I 5.58 I 4.64 I 6.44 I 3.85 I 26.00 I 17.00 I 5.73 with Detox. I 10.14 I 8.58 I 10.10 I 13.34 I 48.83 I 11.50 I 11.00

PAGE 116

Patient Characteristics and LOS The additional independent variables included in the analyses, payer type and patient characteristics, also affected WS. The impact of individual patient characteristics on LOS varied by hospital type. The effects of payer type, age group, gender, and diagnostic group are compared across hospital types. Source of Payment. Previous studies have also found that type of payment for services is a significant determinant of psychiatric length of hospital stay (Rupp et al., 1984; Grazier & McGuire, 1987; & Freiman et al., 1989). This study found unexpected variations within payer groups as well as among hospital types. Because source of payment information was not available from public psychiatric hospitals, this hospital type is not included in the following analysis. Previous studies indicated that the LOS of patients in three payer groups-self-pay, Worker's Compensation, and Medicaid-would be shorter, on average, than the LOS of the reference group, commercially insured patients (Rupp et al., 1984; Grazier & McGuire, 1987; & Freiman et al., 1989). Hospitals have a financial incentive to discharge self-pay patients as soon as possible because approximately 90 percent of self-pay accounts are uncollectible (Frank & Salkever, 1991). As expected, self-pay patients had shorter average LOS in private non-profit general hospitals as well as in 108

PAGE 117

private for-profit psychiatric hospitals. However, self-pay and Medicaid patients had longer LOS than commercially insured patients in public general hospitals, suggesting that source of payment does not affect decisions about LOS in public general hospitals. As a function of their role as the provider of hospital services for .the poor and uninsured, public general hospitals serve a large proportion of self-pay and Medicaid patients; so the longer LOS of these patients places an additional financial burden on this hospital type. The Medicaid program has traditionally reimbursed hospitals at a lower rate than have other payer groups. One unexpected study finding was that Medicaid patients discharged from public and for-profit general hospitals had longer LOS, on average, than commercially insured patients. Patients covered by Workmen's Compensation had longer average LOS than the reference group in non-profit psychiatric hospitals. Because less than one percent of the patients treated for mental illness in either 1987 or 1989 were covered by Workmen's Compensation, it is possible that the unexpectedly long LOS in this type was due to a few outliers. No other explanation for this disparity was found. In an earlier study, Grazier and McGuire (1987) found that patients insured by Medicare, Blue Cross/Blue Shield, and other government payers had longer average LOS than the reference group. Study findings did not always support the earlier findings by Grazier and McGuire. In this study, 109

PAGE 118

however, Medicare patients had longer LOS than the reference group in only three hospital types; public general, non-profit general, and non-profit psychiatric hospitals. Patients insured by Blue Cross/Blue Shield had longer LOS than privately insured patients in for-profit general hospitals. Because Grazier and McGuire (1987) did consider the effects of hospital type on LOS, discrepancies between previously reported results and study findings may be explained by differences among hospital types. Age Groups. Gender. and LOS. Age group and gender are two patient characteristics that are routinely included in studies of variation in psychiatric LOS. Average LOS in one age group, patients age 18 and under, is consistently higher than the average LOS of patients in other age groups (Rupp et al., 1984; Taube et al., 1984; NIMH, 1990a; CMHS & NIMH, 1992). Additionally, some studies have found that patients 65 and older have longer than-average LOS (Rupp et al., 1984; Freiman et al., 1990). NIMH (1990a) reported that females have slightly longer LOS than males. Accordingly, this study hypothesized that patients in the age groups 18 and under, 65 and over, and females would have longer-than-average LOS. This study confirmed earlier studies indicating that patients 18 and younger would have longer LOS than the reference group, patients 25-44. This finding was statistically significant (P s .005) in all hospital types but public general hospitals. The finding that patients age 18 and under have 110

PAGE 119

longer average LOS is particularly relevant in Colorado because study findings indicate that the state's rate of inpatient psychiatric treatment for adolescents is unusually high (Table 5.11). Two previous studies of LOS found that patients 65 and over had longer LOS than average (Rupp et al., 1984; Freiman et al., 1990). However, in this study, the direction of the relationship between patients in this age group and the reference group varied considerably by hospital type. Patients age 65 and over had shorter LOS than the reference group in public and non profit general hospitals but had longer-than-average LOS in private forand non-profit psychiatric hospitals. Because most patients 65 and over are insured under the Medicare program, Medicare reimbursement policies are a likely explanation for differences in LOS by hospital type for this age group. During the study period, Medicare reimbursement for psychiatric care in general hospitals was on a per-case basis while the treatment of Medicare patients in private psychiatric hospitals was reimbursed on a per-diem basis (Cowan, personal communication, 1993). Thus, general hospitals had a financial incentive to discharge patients as soon as possible while, conversely, private psychiatric hospitals were financially rewarded for longer LOS. No statistically significant relationship (P s .05) was found between gender and LOS. 111

PAGE 120

DRGs and LOS. Several studies of psychiatric LOS have reported that diagnostic groups explain a significant proportion of the variation in LOS (Taube et al., 1984; Frank & Lave, 1985; Lave & Frank, 1990). While diagnosis was a significant predictor of variation in LOS in this study, the summary of relationships between DRGs and LOS did not always follow the expected directions. This may be because previous studies of psychiatric LOS did not usmilly include hospital type as an independent variable. Discrepancies between study findings and previous studies may be due largely to the inclusion of hospital type as an additional class of explanatory variables. Psychosis was the most commonly diagnosed psychiatric disorder and was therefore used as the reference group. Patients with this diagnosis are among the most seriously mentally ill and often have longer-than-average LOS. Accordingly, it was eXpected that patients in only three diagnostic groups-mental disorder/operating room procedure, organic disturbance/ mental retardation, and childhood mental disorders-would have longer average LOS than the reference group. Study findings indicated LOS in these diagnostic groups varied by hospital type. For example, patients with the diagnosis "childhood mental disorder" bad longer LOS only in for-profit hospitals (both general and psychiatric). In public and non-profit general hospitals, patients with this diagnosis bad shorter average LOS than the reference group. As expected, 112

PAGE 121

patients with the combined psychiatric/medical diagnosis of "mental disorder/ operating room procedure" had higher average LOS than the reference group, although this finding was statistically significant (P s .005) only in public general and public psychiatric hospitals. Patients with a diagnosis of "organic disturbance/mental retardation" had shorter average LOS than the reference group in private non-profit and for-profit general hospitals. With two exceptions, diagnoses with an. expected shorter-than-average LOS followed the expected pattern. The two exceptions were the DRGs "personality disorder" and "other mental disorders". The average LOS for patients with the DRG "other mental disorders" was shorter than the reference group in public general hospitals but longer than the reference group in private nonand for-profit general hospitals. The LOS of patients with a diagnosis of "personality disorder" had an inverse relationship to the reference group in all general hospital types. However, average LOS for this DRG was higher than LOS for patients diagnosed psychotic (the reference group) in public psychiatric hospitals. Because only those patients with the most severe personality disorders are transferred to public psychiatric hospitals, severity of illness is the likely explanation for this finding. The inter-hospital variations in LOS for the same diagnosis may be explained by differences in physician practice style and treatment philosophy among the diverse hospital types providing psychiatric treatment. Because no 113

PAGE 122

other apparent reasons for these discrepancies were found in the literature, further research is needed to determine why LOS for the same DRG varied among hospital types. Cream Skimming The third issue addressed by this study focuses on changes in payer mix among the types of hospitals providing inpatient treatment for mental illness during the study period. The emergence of private for-profit psychiatric hospitals as major providers of inpatient mental health services in Colorado raised concerns that such hospitals would cream skim; that is, divert privately insured patients from general hospitals providing inpatient treatment for mental illness to investor-owned psychiatric hospitals. Findings indicate that private psychiatric hospitals did cater to privately insured patients; over two thirds of patients discharged from these hospital types were privately insured. Private insurers, including commercial insurance companies and Blue Cross/ Blue Shield, typically reimburse most, if not all, of the costs of treatment. In contrast, public and private non-profit general hospitals served a large proportion of Medicaid, underinsured, and uninsured psychiatric patients whose treatment costs were only partially reimbursed or not reimbursed at all. The hypothesis that private investor-owned psychiatric hospitals would cream skim was tested by comparing changes in payer mix among hospital 114

PAGE 123

types in 1987 and 1989. Data are from the CHDC Hospital Discharge Data Sets for 1987 and 1989. Source of payment data does not include public psychiatric hospitals because this hospital type is not required to report payment sources. Moreover, the 1987 payer picture is clouded by errors in reporting payer data by two hospital types, private for-profit general hospitals and private non-profit psychiatric hospitals. In 1987, private for-profit general hospitals placed 32 percent of discharges in the "other /unknown" category and private non-profit psychiatric hospitals placed 49 percent of total discharges in this category. These reporting errors were corrected in 1989. Despite these limitations, the available payer data provide insight into changes that occurred in the three hospital types accounting for more than three-quarters of all discharges during the study period. Payer Mix by Hospital T)l)e Private for-profit psychiatric hospitals were the fastest-growing sector of the inpatient mental health care delivery system during the study period: In 1987, eight percent of total psychiatric discharges were from this hospital type; in 1989 and 1991, approximately 20 percent were from private for-profit psychiatric hospitals. In 1987, 70 percent of patients discharged from this hospital type were privately insured; in 1989, over 80 percent were privately insured. Medicare patients accounted for another eight to nine percent of 115

PAGE 124

total discharges during the study period. Less than 12 percent of patients discharged from this hospital type were Medicaid, self-pay, or no-charge patients. The payer data from private non-profit psychiatric hospitals were also clouded by reporting errors in 1987. The 1989 payer mix, however, is similar to that reported by for-profit psychiatric hospitals, although non-profit psychiatric hospitals served a larger proportion of Medicare patients and very few Medicaid patients during the study period. When measured by percentage of discharges, private non-profit general hospitals were the major providers of inpatient mental health services, accounting for approximately 40 percent of all discharges in 1987, 1989, and 1991. The proportion of privately insured patients (including commercial insurance and Blue Cross/Blue Shield) discharged from this hospital type decreased from 40 percent in 1987 to 32 percent in 1989. This loss is significant because operating margins in private non-profit general hospitals are slim, averaging approximately five percent a year (Cowan, personal communication, 1993). The loss of privately insured patients supports the hypothesis that private investor-owned psychiatric hospitals would cream skim privately insured patients from non-profit general hospitals. A corresponding increase in Medicaid, self-pay, and no-charge patients is further evidence of a weakening payer mix for non-profit general hospitals in 1989. 116

PAGE 125

Public general hospitals discharged approximately 11 percent of the total number of psychiatric patients treated in 1987, 1989, and 1991. During the study period, this hospital type served the largest proportion of uninsured or poorly insured patients. In 1987, 46 percent of the psychiatric patients treated in public general hospitals were self-pay patients; it is estimated that over 90 percent of self-pay accounts are uncollectible (Frank & Salkever, 1991). During the same year, 16 percent of patients discharged from this hospital type were covered by Medicaid, which typically reimburses about half of the fully allocated costs of hospital treatment (Moore & Coddington, 1991). One surprising finding was that the payer nrix in public general hospitals improved slightly in 1989: The percentage of self-pay patients decreased to 37 percent, Medicaid patients increased to 19 percent, and the number of privately insured patients increased slightly. Private for-profit general hospitals reported approximately 11 percent of psychiatric discharges in 1987, 1989, and 1991. Although 1987 payer data from this hospital type is biased by a large number of reporting errors, 1989 data indicates that approximately 60 percent of psychiatric patients were privately insured, 20 percent were Medicare patients, and approximately 16 percent were either self-pay or Medicaid patients. Based on 1989 data, it can be concluded that private for-profit general hospitals had a strong payer nrix during the study period. Findings are summarized in Table 5.10. 117

PAGE 126

Table 5.10 Blue Cross/ Blue Shield Commercial 1987 13.6 34.4 25.8 31.6 57.3 Insurance 1989 15.1 28.3 57.7 71.7 68.5 ::::=: :::::::::::::::::::::: :;:::::::::::::::::: :::::::. 1987 14.1 19.7 15.1 6.6 9.3 ,_. 1989 = u;,:::::===:::':J8>4::::n;;:m::::-= 18.7 22.2 19.6 11.0 8.1 ,_. 00 1987 .. ::::-:;:j_i=;":! 11 Medicaid 15.6 17.0 7.4 0.1 7.1 19.0 17.9 7.9 0.0 3.4 Other** 1.0 3.0 1.3 5.5 2.2 Government 1.2 3.3 1.1 5.2 3.3 Self-Pay 1987 1 45.9 13.2 10.8 1.9 8.2 37.1 14.6 8.7 2.3 3.2 No Charge/ 1987 : ... .. 7.5 6.6 32.3 48.8 3.0 Unknown 1989 : : ::m::::n:::::=: s:s;:,::;::::::::=t 6.4 9.3 0.0 1.7 0.8 Total I 100% 100% 100% 100% 100% Sources 100% 100% 100% 100% 100% Includes Worker's Compensation

PAGE 127

Patient Characteristics An analysis of patient mix provided additional insight into patterns of patient allocation by hospital types during the study period. Differences in age group and diagnostic group were analyzed. Age. In 1986, seven percent of the patients hospitalized for psychiatric and substance abuse treatment in the U.S. were age 18 or younger (CMHS & NIMH, 1992). In contrast, according to the CHDC data, 18 percent of patients admitted for treatment of mental disorders in Colorado in 1987 were 18 or younger; and by the proportion had increased to 20 percent. Although the Colorado figures are significantly higher than the most recent national average, they nonetheless are understated because the data did not include most of the new private for-profit psychiatric hospitals which typically serve a large proportion of patients in the 18 and under age group. Private psychiatric hospitals discharged significantly higher percentages of patients 18 and under than did other hospital types. In 1987, 47 percent of total discharges from the private for-profit psychiatric hospitals were in this age group and in 1989, 39 percent were 18 and under. It should be noted, however, that the absolute number of patients 18 and under treated by this hospital type increased in 1989 because the total number of patients discharged from for-profit psychiatric hospitals increased dramatically between 119

PAGE 128

1987 and 1989. Patients in this age group accounted for 26 percent of total discharges from private non-profit psychiatric hospitals during both years. For all hospital types combined, the age group 25 to 44 years accounted for the largest proportion of total discharges---44 percent in both 1987 and 1989. Public hospitals, both general and psychiatric, reported a slightly higher percentage of patients in this age group than did the private hospitals. Private for-profit psychiatric hospitals reported significantly fewer discharges in this age group in both years. Patients 45 to 64 years of age accounted for approximately 16 percent of total discharges from all hospital types combined for both years. Private for-profit psychiatric hospitals, however, discharged significantly lower proportions of patients in this age group, five percent in 1987 and 11 percent in 1989. Patients 65 and over represented approximately 11 percent of total discharges for all hospital types combined in 1987 and 1989. Private general hospitals reported the highest percentage of patients in this age group in both years. Private non-profit psychiatric hospitals had slightly fewer discharges in this age group while patients 65 and over accounted for less than five percent of all discharges from private for-profit psychiatric hospitals. The smallest proportion-less than ten percent-of total discharges were in the 19 to 24 year age group in both years. This age group accounted for 120

PAGE 129

only seven percent of the discharges from the private non-profit psychiatric hospitals. Results are summarized in Table 5.11. 121

PAGE 130

Table 5.11 Colorado Psychiatric Discharges: By Age Group and Hospital Type: 1987 and 1989 Age Group 0-18 Yrs 1987 1989 19-24 Yrs 1987 10.0 8.7 9.0 11.4 6.6 I 10.4 II 1989 9.6 8.9 9.7 9.7 6.9 8.6 .. 25 44 Yrs I 47.5 42.5 43.7 I 49.0 I 44.1 I 33.0 46.8 42.5 44.4 49.2 46.0 37.8 45-64 Yrs ::: 20.0 17.3 16.7 14.4 I 13.8 I 4.9 19.7 16.3 16.1 16.7 13.4 10.6 65 & 8.9 14.4 15.0 5.1 I 9.2 I 4.9 8.8 14.3 13.2 5.6 8.1 4.4 ::: 1::::: I 100% I 100% I 100% I 100% I 100% 100% 100% 100% 100% 100%

PAGE 131

Diagnostic Groups. Three diagnostic groups-psychosis, alcohol and drug abuse treatment, and depression-accounted for over 75 percent of all psychiatric diagnoses during the study period. The most frequently occurring diagnosis was psychosis, which increased from 42 percent in 1987 to 45 percent in 1989. The proportion of discharges with a diagnosis of alcohol or drug abuse treatment decreased from 24 percent in 1987 to 20 percent in 1989. Depression accounted for 12 percent of all diagnoses in 1987 and 13 percent in 1989. Patients diagnosed psychotic are among the most seriously mentally ill. Accordingly, the proportion of patients in this DRG can be used to measure the severity of illness by hospital type. As might be expected, patients diagnosed as psychotic made up the largest percentage of discharges reported by the public psychiatric hospitals, increasing from 62 percent in 1987 to 69 percent in 1989. Private psychiatric hospitals, both forand non-profit, also reported higher-than-average percentages of discharges with a diagnosis of psychosis. All general hospital types reported lower-than-average percentages of discharges in this diagnostic group, with for-profit hospitals showing the lowest percentages (27 percent in 1987 and 35 percent 1989). Alcohol and drug abuse treatment was the second largest diagnostic group; accounting for 24 percent of total discharges in 1987 and 20 percent in 1989. The decline in the proportion of patients treated for alcohol and drug 123

PAGE 132

abuse is likely due to studies published in 1988 and 1989 that indicated that patients with this diagnosis could be treated successfully on an outpatient basis (Lave, 1989). Because outpatient treatment is considerably less expensive than hospitalization, insurers became less willing to reimburse inpatient care (Cowan, personal communication, 1993). Private for-profit general hospitals reported the highest percentage of alcohol/drug abuse treatment: 45 percent in 1987 and 40 percent in 1989. All psychiatric hospitals reported relatively low proportions of discharges in this diagnostic group. The diagnosis of alcohol/drug abuse treatment was particularly low in the public psychiatric hospitals, accounting for only two percent of total years in both years. Depression accounted for approximately 12 percent of total diagnoses for both years in all hospital types combined. Private non-profit psychiatric hospitals reported the highest percentage (20 percent) of discharges in this diagnostic group in both years. Overall, no other single diagnosis accounted for more than six percent of total discharges reported from all hospital types combined. The results are summarized in Table 5.12. 124

PAGE 133

Table 5.12 Psychosis 1987 12 I 39.2 43.0 Alcohol/ 1987 ::=:::::.;::?.3.:&:.:-;_ = ::'; 30.7 26.5 44.7 2.0 5.6 Abuse 1989 ::J:W?19.6'Cit 29.1 20.3 40.0 2.4 4.3 Depression 1987 7.6 15.1 12.2 4.8 19.5 .......... I! 1989 8.4 14.6 8.7 7.6 20.2 I Personality 1987 6.5 2.4 2.6 --I 10.7 I 6.3 I 12.1 Disorders 1989 5.7 2.5 3.1 3.1 3.3 2.1 ::.:::::::: :::.:.: ...... ....... .. 3.4 4.5 1.7 4.7 6.0 8.8 2.9 5.8 2.9 3.2 7.9 9.1 3.8 4.9 4.8 1.1 3.1 1.6 2.9 4.5 3.5 1.2 5.1 2.4 Child. Ment.t987 3.8 1.7 1.3 11.4 6.3 13.7 Disorders 1989 3.7 2.2 1.2 9.5 5.2 8.6 All Other 1987 :. __ .... _.=<' :;=-::,., 5.0 7.0 5.3 3.6 1.7 2.3 4.3 10.2 5.3 3.9 1.5 2.6 Diagnosis 100% 100% 100% 100% 100% Total 100% 100% 100% 100% 100%

PAGE 134

CHAPTER 6 CONCLUSIONS The treatment of mental illness is an integral part of the nation's health care delivery system and the total costs of psychiatric and substance abuse treatment are significant. In 1990 alone, the treatment of mental illness accounted for ten percent of U.S. health care spending-$67 billion (Castro, 1993). It is therefore important that, in current discussions about health care reform, policymakers not overlook the financing and delivery of mental health care. Over the last few decades, the locus of inpatient treatment for mental disorders has shifted from public psychiatric hospitals to general hospitals in community settings, and, most recently, to private psychiatric hospitals. Little research has been done to analyze the ramifications of the increasing privatization of the system, yet the implications of this process are of major importance to policymakers at both the state and federal levels. This study has examined issues raised by the emergence of private for profit psychiatric hospitals as major providers of inpatient treatment for mental illness in Colorado between 1987 and 1991. Findings indicate that the growth in private investor-owned psychiatric hospitals increased the use and costs of psychiatric treatment. At the same time, access to needed services for the poor, the uninsured, and the chronically mentally ill may be impaired by the

PAGE 135

shift in locus of care of privately insured patients to private for-profit psychiatric hospitals, thereby weakening the payer mix of non-profit general hospitals. The Roemer Effect and Psychiatric Treatment This study tested the hypothesis that utilization of inpatient psychiatric or substance abuse treatment-as measured by discharges; length of stay, and patient days-would change as a function of changes in capacity. This hypothesis, based on Roemer's Law, argues that additional hospital capacity creates the demand for services and further assumes that at least some of the additional services provided are discretionary; that is, in response to available capacity, rather than actual need (GAO, 1991). The alternate model, a traditional approach to hospital facilities planning, assumes that the supply of hospital beds increases where demand is high. According to this theory, subsequent increases in utilization are a response to previously unmet needs (GAO, 1991). In Colorado, however, the traditional demand-driven model, which assumes that health care providers base the decision to build new hospital facilities solely on community needs, does not adequately explain the relationship between the capacity and utilization of inpatient mental health services during the study period. 127

PAGE 136

According to the traditional model, an assessment of the need for additional hospital beds is measured, in large part, by comparisons of national and local per-capita hospital bed rates (Reeves et al., 1984). By this measure, Colorado did not need the additional private psychiatric hospitals built between 1987 and 1989. In 1986, Colorado's per-capita rate of private psychiatric hospital beds was close to the national average of 13 beds per 100,000 population (NIMH, 1990a). By 1990, the state's per-capita private psychiatric hospital rate was 32 beds per 100,000 population; nearly twice as high as the national average rate for this hospital type which had peaked at 17 beds per 100,000 in 1988. When compared to the national average, Colorado's high per-capita private psychiatric hospital bed rate strongly suggests that changes in the state's private psychiatric hospital capacity occurred in response to factors other than unmet needs. In contrast, the Roemer hypothesis assumes that increases in hospital capacity are not based on an objective set of criteria defining appropriate levels of hospital bed supply. According to Roemer's Law, health care providers adjust the volume of services provided in response to available capacity (GAO, 1991). Thus, Roemer's Law provides a more appropriate theoretical framework with which to explain the impacts of the rapid and substantial increase in private psychiatric treatment hospital capacity in Colorado during the study period. Additionally, a Roemer effect would have 128

PAGE 137

important policy implications because, according to this theory, increases in capacity lead, in part, to increases in the number of marginal or unnecessary hospital admissions. Application of Roemer's Law The results of this study indicate that there was evidence of a Roemer effect on the volume of patients hospitalized for the treatment of mental disorders in Colorado during the study period. Specifically, the number of discharges increased after the capacity to provide such treatment increased between 1987 and 1989 and declined slightly as capacity decreased in 1991. However, a more detailed analysis of the effects of capacity changes on utilization suggests that application of the Roemer hypothesis has serious limitations. There was no apparent Roemer effect on length of hospital stay. LOS decreased in all hospital types but public general hospitals in 1989 despite capacity increases. Largely as a result of a significant decline in public psychiatric hospital utilization, total patient days, the product of the number of discharges and length of stay, also declined during the study period. Finally, in public psychiatric hospitals, changes in utilization were found to be unrelated to changes in capacity. 129

PAGE 138

Discharges. As hypothesized, the number of persons discharged after treatment for mental disorders increased by 13 percent from 1987 to 1989, then decreased by 6 percent in 1991. These effects varied considerably by hospital type, suggesting that factors other than, or in addition to, increases in capacity may explain changes in discharges during the study period. The Roemer hypothesis is most strongly supported by changes in capacity and volume of discharges among private for-profit psychiatric hospitals. Between 1987 and 1989, bed capacity in this hospital type more than doubled; during the same period, the number of discharges from private for-profit psychiatric hospitals nearly tripled. Moreover, when capacity declined in 1991, discharges from this hospital type also decreased. Private for-profit general hospitals also showed some evidence of a Roemer effect: Capacity in specialized psychiatric units for this hospital type increased by 25 percent in 1989; subsequently, discharges increased slightly in both 1989 and 1991. The temporal order of capacity and volume assumed in Roemer's Law, that is, changes in volume are preceded by changes in capacity, was not observed in all hospital types. The number of discharges from private non profit psychiatric hospitals increased significantly in 1989 but were not imniediately preceded by capacity increases. However, capacity as well as discharges in this hospital type declined in 1991. Although public and non130

PAGE 139

profit general hospitals increased psychiatric bed capacity slightly in 1989, the number of discharges declined in 1989, then remained relatively stable in 1991. Length of Stay. Study findings do not support the hypothesis that length of stay would increase as a function of increases in capacity. LOS declined in all but one hospital type in 1989. LOS declined in all hospital types except public general hospitals in 1989, controlling for patient characteristics and source of payment. During the study period, a possible Roemer effect on LOS may have been masked by other factors such as more restrictive payer policies and changes in methods of treatment (Cooper, 1993). According to Cooper (1993), payer strategies, such as utilization review or per-case (rather than per-diem) reimbursement have been successfully used to reduce LOS for psychiatric treatment. Additionally, new treatment methods, which use partial hospitalization or outpatient treatment in conjunction with inpatient care, permit earlier discharge and thereby decrease LOS (Lutz, 1992). Patient Days. The hypothesis that the number of patient days would increase as a function of capacity increases was not supported. The total number of days patients were hospitalized for the treatment of mental illness (total patient days) in Colorado declined between 1987 and 1989. Most of the decline was due to decreasing public psychiatric hospital utilization: 131

PAGE 140

Significant declines in both the number of discharges and average LOS led to a 44 percent decrease in patient days from this hospital type in 1989. Total patient days also decreased in all general hospital types. In private psychiatric hospitals, however, total patient days increased significantly in 1989 as a result of the increased number of discharges from both nonand for-profit private psychiatric hospitals. Changes in patient days by hospital type illustrated a shift in locus of care to private psychiatric hospitals in 1989. In 1987, private psychiatric hospitals accounted for less than 18 percent of total psychiatric and substance abuse patient days. By 1989, however, private psychiatric hospitals accounted for a third of all patient days; general hospitals and public psychiatric hospitals each accounted for another third. Limitations of Roemer's Law This study found only limited evidence supporting the Roemer hypothesis in the Colorado experience. Underlying the Roemer hypothesis is the assumption that increases in bed supply alter the way physicians practice, which in tum increases the rate at which hospital services are utilized (Feldstein, 1981). Other factors, however, also influence physicians' decision making processes. Additionally, physicians have a more limited role in decisions to admit patients to private psychiatric hospitals. Typically, private 132

PAGE 141

psychiatric hospitals are more likely to market their services directly to the consumer than are general hospitals. Finally, the Roemer effect was not apparent among public psychiatric hospitals: Changes in utilization in this hospital type were driven by changes in public policy rather than by capacity changes. Physician's Role. Changes in methods of reimbursement and innovations in the treatment of mental illness at least partially explain why physicians in Colorado did not always alter their practice patterns in the manner predicted by Roemer's Law. Third-party payer strategies-such as utilization review, preadmission screening and case-management-that were designed to control costs have limited the physician's ability to create additional demand in response to bed supply increases. Managed care systems, such as health maintenance organizations and preferred provider organizations, financially reward providers for reducing the volume of services provided. Innovations in mental health treatment also affected utilization. New drugs developed for the treatment of mental illness enabled more patients to be served on an outpatient basis. Moreover, recent research suggests that, for many patients needing psychiatric or substance abuse treatment, outpatient care may be as just effective as hospitalization, at significantly lower cost (Lave, 1989). 133

PAGE 142

According to the Roemer hypothesis, changes in physician behavior are the key determinants influencing changes in utilization following capacity increases. In general hospitals, the physician usually make the decision to admit patients for treatment. However, among the general hospital types, only the for-profit hospitals demonstrated any Roemer effect. Evidence supporting a Roemer effect was most pronounced among private investor-owned psychiatric hospitals where physicians have a more limited role in admissions decisions than they do in general hospitals. Dorwart et al. (1989), for example, found that only 40 percent of the patients admitted to private psychiatric hospitals are referred by a physician. The authors note that: Psychiatric hospitals, moreover, are likely to market themselves more directly to patients, who are far more likely to admit themselves without advice of a physician than they are to seek admission for physical illness (p. 147). Marketing Psychiatric Services. By marketing their services directly to the public, private psychiatric hospitals play a major role in creating a demand for their services. Intensive marketing of psychiatric services may have an effect as important as increases in capacity on utilization. In Colorado, for 134

PAGE 143

example, private psychiatric hospitals were major advertisers in 1989, the year that admissions for psychiatric treatment increased most dramatically. There is some evidence that advertising can induce demand for inpatient psychiatric treatment. In a small study, Greer and Greenbaum (1992) found that advertising targeted at the parents of adolescents raised parental fears about adolescent behavior which could, in turn, lead to increased hospital admissions of adolescents. The authors also found that parents of relatively normal adolescents were more likely to respond to advertising than parents of troubled teenagers, suggesting that advertising may have induced demand for unnecessary or marginally useful psychiatric services. Additional research is needed to determine to what extent the marketing of inpatient psychiatric services leads to greater use of those services. It is also important to discern whether advertising contributes to the appropriate use of services or encourages greater use of services among patient groups not in need of inpatient care. Utilization in Public Psychiatric Hospitals. Public policy decisions, rather than capacity increases, explain the dramatic changes in utilization in public psychiatric hospitals during the study period. In public institutions, discharges and length of stay are largely driven by the ability to place patients in community settings after inpatient treatment (Ellis, personal communication, 1993). 135

PAGE 144

Despite a modest increase in bed capacity, public psychiatric hospitals experienced a significant decline in discharges in 1989. The volume of discharges decreased by one-third, suggesting that the state's ability to continue the deinstitutionalization of the chronically mentally ill was constrained. The sharp decline in the 1989 discharge rate triggered a reexamination of the criteria used to evaluate the appropriateness of long-term hospitalization (Ellis, personal communication, 1993). Based on the revised criteria, the institutionalized patient population was divided into two groups according to diagnosis and level of functioning. Efforts were then made to place the less impaired group of patients into community settings. The review process led to more aggressive efforts to locate resources for those patients deemed able to live in community settings. Hospital staff became more adept at qualifying public psychiatric hospital patients for assistance from federal disability programs for patients discharged to community settings, thereby shifting some of the costs of care for this population to federal programs such as Medicare and Medicaid (Ellis, personal communication, 1993). As a result of these "administrative discharges", total discharges from public psychiatric hospitals increased by more than one third in 1991. Public psychiatric hospital LOS declined 17 days in 1989, controlling for age group, diagnosis and gender. As a result of decreasing discharges and 136

PAGE 145

declining LOS, total patient days also declined significantly during the same period. Hospital Type and LOS Although previous research studies found that average length of hospital stay for the treatment of mental illness varied significantly by hospital type, little research has been done to determine why these variations occur. This research gap has continued because the growth of private investor-owned psychiatric hospitals is a recent phenomenon that raises complex questions about the interactions of specialization and investor-ownership on the delivery of treatment for mental illness. This study begins to bridge the research gap by addressing the effects of ownership form and degree of specialization on psychiatric LOS. Implications of LOS Variation by Hospital Type Until the mid-1980s, private psychiatric hospitals played a relatively minor role in the mental health care delivery system. General hospitals assumed the major role in the delivery of most inpatient psychiatric treatment, while public psychiatric hospitals continued to provide care for the most seriously disabled mentally ill. Accordingly, inpatient treatment in general hospital psychiatric units was of brief duration, while treatment in public psychiatric hospitals tended to be significantly longer-term. Thus, the 137

PAGE 146

differences in LOS between the general and public psychiatric hospitals were explained by the specialized role each played in the mental health care delivery system. Private psychiatric hospitals also reported longer average LOS than general hospitals (NIMH, l990a); however, this was not a serious concern until private psychiatric hospitals became major providers of services. Previously, it had been assumed that differences in LOS between general and private psychiatric hospitals were largely explained by differences in patient mix. Private psychiatric hospitals, for example, served a higher proportion of adolescent and seriously mentally ill patients, two patient populations with characteristically higher average LOS (CMHS & NIMH, 1992). LOS is a major determinant of the cost of inpatient care. If the longer LOS reported by private psychiatric hospitals is not explained by differences in patient characteristics or does not result in better outcomes, then the greater cost of treatment in these hospitals may not be justified. Assuming no difference in outcomes, a shift in locus of care from general to private psychiatric hospitals would increase the costs of treatment for mental illness without providing comparable benefit. Such a shift in locus of care from general to private psychiatric hospitals occurred in Colorado during the study period. The effects of this shift on length of hospital stay are discussed in the following section. 138

PAGE 147

Cost Impacts Study findings confirm the hypothesis that LOS for the treatment of mental illness varies by hospital type, controlling for differences in patient characteristics, payer type, and year of service. The reference group, non profit general hospitals, had shorter average LOS than any other hospital type. Private for-profit psychiatric hospitals had the longest LOS, which was on average, 11 days longer than LOS in non-profit general hospitals. Private psychiatric hospitals also expanded dramatically during the study period. One important effect of the emergence of investor-owned psychiatric hospitals and the corresponding shift from non-profit general hospitals to for profit psychiatric hospitals was that the total costs of providing treatment for mental illness in Colorado increased significantly during the study period. It is possible to make a rough estimate of the increased cost-as measured in patient days-of providing inpatient psychiatric treatment in for-profit psychiatric hospitals rather than non-profit general hospitals. The number of discharges from private for-profit psychiatric hospitals increased from 1,497 in 1987 to 4,388 in 1989, a difference of 2,891 discharges. When controlled for differences in patient characteristics, private for-profit psychiatric hospital LOS was an average 11.34 days longer than the average LOS in non-profit general 139

PAGE 148

hospitals. If the additional patients treated in the for-profit psychiatric hospitals had received treatment in non-profit general hospitals, an estimated 32,784 days of inpatient psychiatric care would have been saved. The average daily charge for treatment in private for-profit psychiatric hospitals was $699 in 1989. At a cost of $699 per day, the additional32,784 patient days spent in private for-profit psychiatric hospitals cost nearly $22,900,000 in Colorado in 1989 alone. (This estimate assumes that the private for-profit psychiatric hospital data used in the analysis are representative of discharges from the non-reporting private psychiatric hospitals and that the same volume of discharges would have occurred in 1989 if private psychiatric hospital capacity had not increased). The additional costs paid by private insurers were likely passed on to consumers in the form of higher insurance premiums. Differences in LOS between general and private for-profit psychiatric hospitals may be explained by differences in methods of treatment or treatment philosophy, however, further research is needed to determine whether the more-costly treatment in private investor-owned psychiatric facilities can be justified by better patient outcomes. Implications for Colorado The finding that private for-profit psychiatric hospitals have significantly longer LOS than non-profit general hospitals is particularly important in states, 140

PAGE 149

such as Colorado, that experienced a rapid growth in investor-owned psychiatric hospitals and a corresponding shift from general hospitals to these facilities. An increased number of patients admitted to private for-profit psychiatric hospitals for a longer average LOS resulted in higher costs for inpatient mental health care in Colorado during the study period. Differences in LOS among hospital types also raises questions about the effectiveness of third-party payer cost-containment policies aimed at reducing LOS. LOS did decrease in all hospital types but public general hospitals during the study period, suggesting that such policies had an effect on the LOS of individual hospital types. Potential cost-savings were offset, however, by the increased number of patients hospitalized for longer stays in private investor owned psychiatric hospitals in Colorado in 1989. To control costs more effectively, payers must encourage patients to use hospitals with lower average lengths of stay and must carefully monitor potential psychiatric treatment admissions to prevent unnecessary or marginally useful hospitalizations. Cream Skimming The emergence of private for-profit psychiatric hospitals has raised concerns about the effects of investor-ownership on the evolving mental health care delivery system. Previous studies have found that this hospital type typically caters to privately insured patients and provides little or no care to 141

PAGE 150

Medicaid or uninsured patients (NIMH, 1990a; Dorwart et al., 1991). Those patterns were confirmed by findings from this study indicating that over 80 percent of patients discharged from for-profit psychiatric hospitals were privately insured. In contrast, privately insured patients accounted for only 46 percent of totaldischarges from all hospital types in 1989. This study hypothesized that the increase in the number of private for profit psychiatric hospitals between 1987 and 1989 would result in cream skimming, that is, the diversion of privately insured patients from other types of hospitals providing inpatient psychiatric and substance abuse treatment. Study findings indicate that cream skimming may indeed have occurred; the payer mix of non-profit general hospitals was weakened by a major reduction in the proportion of commercially insured and Blue Cross/Blue Shield patients-from 41 percent in 1987 to 33 percent in 1989. The change in payer mix is particularly important because non-profit general hospitals served the largest number of patients during the study period-accounting for approximately 40 percent of total discharges annually. The decline in privately insured patients represents an erosion of the non profit general hospitals' abilities to shift the costs of caring for poorly reimbursed Medicaid patients and largely uncollectible self-pay accounts to privately insured patients. Dorwart et al. (1989) noted that a disproportionately large volume of low-income patients whose treatment costs 142

PAGE 151

are not fully reimbursed threatens access to care for this population because competitive pressures will eventually force non-profit general hospitals to eliminate unprofitable psychiatric services. The Medicaid program is now the single largest payer of treatment costs for persons with severe and chronic mental illness (Schlesinger & Mechanic, 1993). As previously discussed, Medicaid costs for the treatment of mental illness are likely to increase in Colorado because public psychiatric hospital staff have become more adept at shifting patients from institutional settings by using Medicaid financing to pay for community care (Ellis, personal communication, 1993). Federal and state restrictions on the use of Medicaid financing for treatment in private psychiatric hospitals make it likely that Medicaid patients will increasingly be treated in the non-profit general hospital sector. Federal Medicaid regulations, for example, prohibit the use of federal Medicaid funds for psychiatric treatment of patients 18 to 64 in private psychiatric hospitals (GAO, 1990). Colorado requires that private psychiatric hospitals be Medicaid-certified; however, only two private psychiatric hospitals have obtained Medicaid certification (Susan Rehak, Colorado Health Facilities Division, Colorado Department of Health, personal communication, 1992). Thus, it is likely that non-profit general hospitals will be required to serve an increasing number of Medicaid patients. 143

PAGE 152

These trends have important implications for state policymakers. The Medicaid program is the fastest-growing expense in most state budgets and the increasing number of Medicaid-eligible chronically mentally ill is likely to continue to raise program costs. If cost pressures force lawmakers to maintain Medicaid reimbursement at low levels, then non-profit general hospitals may be forced to reduce access to care for Medicaid patients or eliminate psychiatric services altogether. The Changing Role of Private Psychiatric Hospitals During the study period, the roles of the general and public psychiatric hospitals in Colorado's mental health care delivery system remained essentially unchanged; that is, general hospitals treated the majority of patients in brief hospital stays, while public psychiatric hospitals provided long-term treatment to the most seriously impaired patients. During the same period, however, private psychiatric hospitals evolved from institutions serving a relatively small number of primarily well-insured patients to major providers of inpatient care, competing with general hospitals for privately insured patients. The number of private psychiatric hospitals declined in Colorado in 1991. Despite increasing admissions, a strong payer mix, and high average daily costs among private psychiatric hospitals in 1989, only four out of 11 private psychiatric hospitals reported positive net income, according to 144

PAGE 153

financial statements from Medicare Cost Reports. Estimated net losses for all private psychiatric hospitals in the state for that year alone were over $2 million. Significant excess capacity led to declining profit margins, which were followed by private psychiatric hospital closings subsequent to 1989. Although the market did eventually clear some of the excess private psychiatric hospital capacity, the unrestricted growth in private psychiatric hospitals between 1987 and 1989 was an expensive experiment. During this period, private insurers paid most of the costs of increased utilization of services provided by private psychiatric hospitals. These costs were likely passed on to consumers in the form of higher insurance premiums. Losses were also incurred by investors in private psychiatric hospitals. Policy Recommendations In 1987, Colorado deregulated hospital construction. Subsequently, the number of private psychiatric hospitals surged, increasing the utilization and costs of inpatient mental health treatment. This study of capacity changes in the state's mental health care delivery system led to three public policy recommendations which are summarized as follows: 1. Changes in for-profit psychiatric hospital capacity led to changes in the volume of admissions to this hospital type during the study period. An analysis of patients admitted to for-profit psychiatric 145

PAGE 154

hospitals in 1987 and 1989 found the mixture to be predominantly privately insured, with a disproportionate percentage of adolescents. Accordingly, further research is needed to ascertain the extent to which increased admissions among well-insured.and adolescent patients to for-profit psychiatric hospitals are marginally useful or unnecessary. 2. This study found that LOS in for-profit psychiatric hospitals was, on average, significantly longer than LOS in non-profit general hospitals, controlling for differences in patient and payer mix. Further research is needed to ascertain whether the longer length of hospital stay typical of private psychiatric hospitals is cost-effective; that is, leads to better outcomes or reduced readmission rates. 3. Between 1987 and 1989, the proportion of privately insured patients treated in Colorado non-profit general hospitals declined substantially, thereby weakening the payer mix. This may force non-profit general hospitals serving a large proportion of poorly insured or uninsured patients to close their psychiatric units. Further research is needed to determine to what extent 146

PAGE 155

changes in payer mix threaten access to services for this vulnerable population. The mental health care delivery system changed dramatically during the last few decades. The growth in private psychiatric hospitals is but the most recent manifestation of the continuing privatization of the system. This study has explored issues surrounding the emergence of private psychiatric hospitals as major providers, and from these analyses, policy recommendations and suggestions for future research have been formulated. 147

PAGE 156

APPENDIX A Recoded Variables Age: Reference Group = 25-44 0-18 Years 19-24 Years 45-64 Years 65 and over Year: Reference Group = 1987 1989 Gender: Reference Group = Female Male Expected Source of Payment: Reference Group = Commercial Insurance Self-Pay Worker's Compensation Medicare Medicaid Other Government Blue Shield Diagnosis-Related Groups: Reference Group = 430 = Psychosis 424: Mental Disorder, Operating Room Procedure 425: Adjustment Reaction 426: Depression 427: Neuroses Except Depression 428: Personality Disorders, Impulse Control 429: Organic Disturbance or Mental Retardation 431: Childhood Mental Disorders 432: Other Mental Disorders 433: Alcohol/Drug Abuse/Left Against Med. Advice 434: th Detoxification 148 IAGE 1 IAGE2 IAGE3 IAGE4 IYEAR 1 ISEX 1 IPAY 1 IPAY2 IPAY3 IPAY4 IPAY6 IPAY7 IDRG 1 IDRG2 IDRG3 IDRG4 IDRG5 IDRG6 IDRG8 IDRG9 IDRG 10 IDRG 11

PAGE 157

* Hospital Type: Reference Group = Private Non-Profit General Public General Hospitals Private For-Profit General Hospitals Public Psychiatric Hospitals Private Non-Profit Psychiatric Hospitals Private For-Profit Psychiatric Hospitals 149 ITYPE 1 ITYPE 3 ITYPE 4 ITYPE 5 ITYPE 6

PAGE 158

APPENDIXB Table B.1 Regression on LOS, Psychiatric Discharges: All But Public Psychiatric Hospitals N = 27 744 = 12.16 R2 = 12.5% F = 158A 6.23* 3.98* 018 Years 11.06* 19 Years 1.05** 45-64 Years .34 -6.35* -4.40* -7.15* -1.12 Retardation -3.63* Childhood Mental Disorders .31 Other Mental Disorders 10.83* Med. Advice -8.21 -2.07* 150

PAGE 159

Table B.1 (cont.) Regression on LOS, Psychiatric Discharges: All But Public Psychiatric 'tals Independent Variables Self-Pay Worker's Medicare Medicaid Other Government Blue Cross/Blue Shield Ps .0001 ** Ps .005 *** Ps .05 151 Regression Coefficient -2.66* 2.45*** 1.40** .23 .91 2.64*

PAGE 160

Table B.2 018 Years 19-24 Years 45-64 Years 65 and Over Childhood Mental Disorders Other Mental Disorders Medicare Medicaid Other Government Discharges: Public General lnf"i:l> .. ,..,..nf"= 10.41 R2 = 8.3% F = 17.0 1.18 .88 .46 -2.62** -6.30* -5.12* -3.51 *** -5.24* -3.58** -8.61 *** -9.42* 3.21* -2.10 4.12* 4.68* 1.81 Ps .0001 *"' Ps .005 *** Ps .05 152

PAGE 161

Table B.3 018 Years 19-24 Years 45-64 Years 65 and Over Childhood Mental Disorders Other Mental Disorders Medicare Medicaid Other Government Ps .0001 ** Ps .005 Discharges: Priv. Non-Profit Gen. tPTt"Pnt= 14.07 R2 = 13.4% F= 107.5 8.64* .64 .03 -1.71* -7.65* -5.29* -7.47* -2.49** -3.97* -5.05* 11.56* -7.37* -1.82 .97*** -1.12* .75 *** Ps .05 153

PAGE 162

Table B.4 Regression on LOS, Psych. Discharges: Priv. For-Profit General N= =15.62 R2=18% F=34.6 018 Years 14.76* 45-64 Years 1.01 19-24 Years .38 65 and Over -9.31* -2.92** -3.14 Control -2.24 -7.04* Childhood Mental Disorders 7.17** Other Mental Disorders 2.65 Medicare .58 Medicaid 5.76* Blue Shield 2.84*** p::; .0001 ** p::; .005 *** p::; .05 154

PAGE 163

Table B.5 Regression on LOS, Psych. Discharges: Priv. Non-Profit Psych. N=4,295 Intercept=14.89 R2=13.6% F=21.1 018 Years 19-24 Years 45-64 Years 65 and Over Adjustment Reaction Control Organic Disturbance/Mental Retardation Childhood Mental Disorders Other Mental Disorders Alcohol/Drug Abuse/with Detoxification Self-Pay Worker's Compensation Medicaid Other Government Blue Cross/Blue Shield Ps .0001 ** Ps .005 155 Regression Coefficient 21.55* 2.93 -.51 4.38** .92 -4.43** -10.37* 4.01 -1.09 -2.38 -13.35 1.62 -2.10 22.57* -3.37 .75 8.07*

PAGE 164

Table B.6 Regression on LOS, Psych. Discharges: Private For-Profit Psych. N=1,428 =23.07 R2=14.1% F=154.3 Independent Variables 018 Years 19-24 Years 45-64 Years 65 and Over Disturbance/Mental Retardation Childhood Mental Disorders Other Mental Disorders Alcohol/Drug Abuse/Left Against Med. Advice Alcohol/Drug Abuse/with Detoxification Self-Pay Medicare Medicaid Ps .0001 ** Ps .005 *** Ps .05 156 Regression Coefficient 15.60* 1.59 -.72 10.16*** 7.56 4.57 13.59* 10.39 -9.99 -4.68 -11.61 ** 12.08 -1.75 2.99 5.81 ***

PAGE 165

Table B.7 Regression on LOS, Psychiatric Discharges: Public Psychiatric Hospitals N = 4,156 Intercept = 74.49 R2 = 1.6% F = 4.5 Independent Variables 018 Years 19-24 Years 45-64 Years 65 and Over Mental Disorder, Operating Room Procedure Neuroses Except Depression Personality Disorders Organic Disturbance/Mental Retardation Childhood Mental Disorders Other Mental Disorders Alcohol/Drug Abuse/Left Against Med. Advice Ps .0001 ** Ps .005 *** P:s; .05 157 Regression Coefficient 36.48** 5.36 7.06 -1.88 106.79** 45.46 -63.17*** -56.85** -32.48*** -23.46 2.09 83.88 -55.96 -26.27 10.65

PAGE 166

BIBLIOGRAPHY 1\ckerknecht, E. (1986). Private institutions in the genesis of psychiatry. Bulletin of the History of Medicine, 60(3), 387-395. !\merican Hospital Association. (1989). Hospital statistics. 1988. Chicago: American Hospital Publishing Inc. !\merican Hospital Association. (1993). Hospital statistics. 1992-1993. Chicago: American Hospital Publishing Inc. 1\rrow, K. (1963). Uncertainty and the welfare economics of medical care. American Economic Review, 53(3), 941-973. 1\shley, J. (1984). The impact of hospital regulatory programs on per capita costs, utilization and capital investment. Inquiry, 21(1), 45-59. 1\strachan, B., & Astrachan, J. (1989). Economics of practice and inpatient care. General Hospital Psychiatry, 11, 313-319. Bachrach, L. (1982). Young adult chronic patients: An analytical review of the literature. Hospital and Community Psychiatry, 33(3), 189-196. Bachrach, L. (1986). The homeless mentally ill in the general hospital: A question of fit. General Hospital Psychiatry, R, 340-347. Beigel, A, & Scharfstein, S. (1984). Mental health care providers: Not the only cause or only cure for rising health care costs. American Journal of Psychiatry, 141, 668-672. Bickman, L., & Dokecki, P. (1989). Public and private responsibility for mental health services. American Psychologist, 44(8), 1133-1137. Bittker, K. (1985). The industrialization of American psychiatry. American Journal of Psychiatry, 142, 149-154. Bluestone, H., & Melinek, M. (1982). Effects of the urban crisis on the community general hospital. Hospital and Community Psychiatzy, 33(6), 477-480. Brewer, W., & Friedman, M. (1982). Causes and implications of variations in hospital utilization. Journal of Public Health 445-454.

PAGE 167

:rodie, H., & Banner, L. (1987). Daniel in the lion's den: American medicine and the corporate solution. Psychiatric Hospital, 18(2), 61-69. :uck, R. (1987). For-profit psychiatric hospitals. American Journal of Psychiatry. 144(3), 395-396. :ada, C. (1990, May 29). State experiencing glut of facilities for treatment. The Denver Post, p. 6B. :astro, J. (1993, May 31). What price mental health? Time, pp. 59-60. :aton, C. (1981). The new chronic patient and the system of community care. Hospital and Community Psychiatry, 32(7), 475-478. :aton, C., & Gralnick, A (1987). A review of the issues surrounding length of psychiatric stay. Hospital and Community Psychiatry, 38(8), 858-863. for Mental HealthServices & National Institute of Mental Health. (1992). Mental health. United States. 1992. Washington, DC: U.S. Government Printing Office. G., Brenner, L., & Bryant, K (1991). Factors predicting inpatient length of stay in a CMHC. Hospital and Community Psychiatry, 42(8), 853-855. A (1986). Capital projects funding for psychiatric hospitals, 1972-1981. Hospital and Community Psychiatry, 'J1(4), 380-385. M. (1987). Substance abuse in the workplace. Hospitals, .2(12), 68-73. leary, P. (1989). The need and demand for mental health services. In C. Taube, D. Mechanic, & A. Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. blorado Department of Health, Division of Health Facilities. (1991). Directory of Colorado health facilities. 1990. Division of Local Government, Demographic Section. (1991). Unpublished data. Hospital Association. (1990). Unpublished data. 159

PAGE 168

:onnell, F., Day, R., & LoGerfo, J. (1981). Hospitalization of Medicaid children; Analysis of small area variations in admission rates. American Journal of Public Health, 71(6), 606-613. :ook, T., & Campbell, D. (1979). Quasi-Experimentation: Design and analysis issues for field settings. Boston: Houghton Mifflin. :ooper, H. (1993, March 16). Cost controls impel psychiatric hospitals to establish more outpatient programs. Wall Street Journal, p. Bl. :osteno, A., Dulcan, M., & Kalas, R. (1991). A checklist of hospitalization criteria for use with children. Hospital and Community Psychiatzy, 42(8), 823-828. P., Bene-Kociemba, A, & Cole; R. (1979). The effect of deinstitutional ization on a general hospital's inpatient psychiatric service. Hospital and Community Psychiatzy, 30(9), 609-612. :romwell, J., Harrow, B., McGuire, T., & Ellis, R. (1991). Medicare payment to psychiatric facilities: Unfair and inefficient? Health Mfairs, 10(2), 124-134. rosby, J. (1990, March). How and where the mentally ill are cared for. National Journal, p.4. Uster, W. (1990, February). Issues in mental health care. Employee Benefit Research Institute (Issue Brief). yr, J., & Haley, G. (1983). Use of demographic and clinical characteristics in predicting length of psychiatric hospital stay: A final evaluation. Journal of Consulting and Clinical Psychology, 51(4), 637-640. 1orwart, R. (1987). Shrink costs versus costing shrinks. Advances in Health Economics and Health Services 127-131. 1 0rwart, R. (1990). Managed mental health care: Myths and realities in the 1990's. Hospital and Community Psychiatzy, 41, 1087-1091. 1 orwart, R., Epstein, S., & Davidson, H. (1988). The shifting balance of public and private inpatient psychiatric services: Implications for administrators. Administration and Policy in Mental Health, 16, 4-13. 160

PAGE 169

torwart, R., Horgan, C., Schlesinger, M., & Davidson, H. (1989). In C. Taube, D. Mechanic, & A. Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. torwart, R., & Schlesinger, M. (1988). Privatization of psychiatric services. American Journal of Psychiatzy. 145(5), 543-553. torwart, R., Schlesinger, M., Davidson, H., Epstein, S., & Hoover, C. (1991). A national study of psychiatric hospital care. American Journal of Psychiatry, 148(2), 204-210. troste, T. (1988). Teens: Trading boarding schools for psych wards? Hospitals, 62(18), 74-75. isenberg, L. (1984). The case against for-profit hospitals. Hospital and Community Psychiatry. 35(10), 1009-1012. isenberg, L. (1986). Health care: For patients or profits. American Journal of Psychiatry. 143(8), 1015-1019. nglish, J., Kritzler, Z., & Scherl, D. (1984). Historical trends in the financing of psychiatric services. Psychiatric Annals, 14(5), 321-330. nglish, J., & McCarrick, R. (1986). DRGs: An overview of the issues. General Hospital Psychiatry. a, 359-364. rmann, D., & Gabel, J. (1984). Multihospital systems: Issues and empirical findings. Health Affairs, J.(1), 50-64. ssock, S. (1987). The influence of medical comorbidities and other patient characteristics on resource consumption on psychiatric wards. Advances in Health Economics and Health Services Research, a, 133-142. eldstein, M. (1981). Hospital costs and health insurance. Cambridge: Harvard University Press. erguson, T. (1990, May 22). Any wonder medical premiums are anything but shrinking? Wall Street Journal, p. Bl. tne, J., & Super, K. (1986, October). Repeal of some states' CON laws spurs psych hospital construction. Modem Healthcare, pp. 74-75. 161

PAGE 170

inkier, S. (1985). DRG problems and potential. Journal of Health and Human Services Administration, 1(1), 35-51. isher, W., Dorwart, R., Schlesinger, M., Epstein, S., & Davidson, H. (1992). The role of general hospitals in the privatization of inpatient treatment for serious mental illness. Hospital and Community Psychiatzy. 43(11), 1114-1119. isher, W., Geller, J., Costello, D., & Phillips, B. (1989). Projecting inpatient admissions to state facilities in the 1990's. Hospital and Community Psychiatzy, 40(7), 747-749. olland, S., & Stano, M. (1989). Sources of small area variations in use of medical care. Journal of Health Economics, .8., 85-107. rank, R. (1985). Pricing and location of physician services in mental health. Economic Inquizy, 23, 115-133. rank, R. (1989a). Regulatory responses to information deficiencies in the market for mental health services. In C. Taube, D. Mechanic, & A Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. rank, R. (1989b ). The medically indigent mentally ill: Approaches to financing. Hospital and Community Psychiatzy, 40(1), 9-12. rank, R., & Gertler, P. (1989). The effect of Medicaid policy on mental health and poverty. Inquizy. 26, 283-290. rank, R., & Kamlet, M. (1990). Economic aspects of patterns of mental health care: Cost variation by setting. General Hospital Psychiatzy, 12, 11-18. rank, R., & Lave, J. (1985). The impact of Medicaid benefit design on length of hospital stay and patient transfers. Hospital and Community Psychiatzy, 36(7), 749-753. rank, R., Lave, J., Taube, C., Rupp, A, & Goldman, H. (1987). The impact of Medicare's prospective payment system on psychiatric patients treated in scatterbeds. Advances in Health Economics and Health Services Research, 8a 23-47. 162

PAGE 171

'rank, R., & McGuire, T. (1986). A review of studies of the impact of insurance on the demand and utilization of specialty mental health services. Health Services Research, 21(2), 241-265. rank, R., & McGuire, T. (1990). Mandating employer coverage of mental health care. Health Affairs, .2(3), 31-41. rank, R., & Salkever, D. (1991). Non-profit providers of mental health care: Privileges and responsibilities. Administration and Policy in Mental Health, 18(3), 195-198. rank, R., Salkever, D., & Scharfstein, S. (1991). A new look at rising mental health insurance. Health Affairs, 10(2), 116-123. reiman, M. (1989). Reimbursement decision for hospital services: The case for psychiatric units under Medicare. Inquiry, 26, 399-405. reiman, M. (1990). Hospital financial performance under the prospective payment system by type of admission: Psychiatric or medical/surgical. Health Services Research, 25(5), 785-808. reiman, M., Ellis, R., & McGuire, T. (1989). Provider response to Medicare's PPS: Reductions in length of stay for psychiatric patients treated in scatter beds. Inquiry, 26, 192-201. reiman, M., Goldman, H., & Taube, C. (1990). Hospitalization for psychiatric illness under Medicare, 1985. Hospital and Communicy Psychiatry. 41(1), 51-58. reiman, M., Mitchell, J., & Rosenbach, M. (1987a). An analysis of DRG-based reimbursement for psychiatric admissions to general hospitals. American Journal of Psychiatry. 144(5), 603-609. reiman, M., Mitchell, J., & Rosenbach, M. (1987b). Modifications of the prospective payment system and payments for Medicare psychiatric admissions. Advances in Health Economics and Health Services Research, 8, 23-47. reund, D., Schactman, R., Ruffin, M., & Quade, D. (1985). Analysis of length of stay differences between investor-owned and voluntary hospitals. Inquiry, 22, 33-44. 163

PAGE 172

'uchs, V. (1979). Economic, health and post-industrial society. Milbank Memorial Fund Quarterly, 57, 153-167. iaylin, S. (1985). The coming of the corporation and the marketing of psychiatry. Hospital and Community Psychiatry, 36(2), 154-159. ieraty, R. (1989). Administrative issues in inpatient child and adolescent psychiatry. Journal of the American Academy of Child & Adolescent Psychiatry. 28(1), 21-25. ihiselli, W., & Frances, P. (1985). The impact of the baby boom on the mental health system. Hospital and Community Psychiatry, 36(5), 536-537. iianfrancesco, F. (1980). Hospital specialization and bed occupancy rates. Inquiry. 17, 260-267. Hnsburg, P., & Koretz, D. (1983). Bed availability and hospital utilization: Estimates of the "Roemer effect". Health Care Financing Review, 5.(1), 87-91. foldfarb, M., Hornbrook, M., & Higgins, C. (1983). Determinants of hospital use: A cross-diagnostic analysis. Medical Care, 21(1), 48-66. roldman, H., & Skinner, E. (1989). Specialty mental health services: Research on specialization and differentiation. In C. Taube, D. Mechanic, & A Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. foldsmith, J. (1984). Death of a paradigm: The challenge of competition. Health Affairs, .3_(3), 5-19. foodman, A. (1989). Estimation of the offset and income effects on the demand for mental health treatment. Inquiry, 26,, 235-248. foplerud, E. (1986). Effects of proprietary management in general hospital psychiatric units. Hospital and Community Psychiatry, 37, 832-836. fornick, M. (1982). Trends and regional variations in hospital use under Medicare. Health Care Financing Review, .3_(3), 41-73. fottleib, G. (1989). Diversity, uncertainty, and variations in practice: The behaviors and clinical decisionrnaking of mental health care providers. In C. Taube, D. 164

PAGE 173

Mechanic, & A. Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. rray, B., & McNerney, W. (1986). For-profit enterprise in health care. New England Journal of Medicine, 314, 1523-1528. rrazier, K., & McGuire, T. (1987). Payment systems and hospital resource use: A comparative analysis of psychiatric, medical and obstetrical services. Advances in Health Economics and Health Services Research, 8, 75-95. rreen, J. (1978). Physician-induced demand for medical care. Journal of Human Resources, .12, 21-34. rreer, S., & Greenbaum, P. (1992). Fear-based advertising and the increase in psychiatric hospitalization of adolescents. Hospital and Community Psychiatry, 43(10), 1038-1039. rriffith, J., Wilson, P., Wolfe, R., & Bischak, D. (1985). Clinical profiles of hospital discharge rates in local communities. Health Services Research, 2Q(2), 131-151. ruterman, S., & Dobson, A. (1986). Impact of the Medicare prospective payment system for hospitals. Health Care Financing Review, 1(3), 97-113. [argreaves, W., & Shumway, M. (1989). Effectiveness of mental health services for the severely mentally ill. In C. Taube, D. Mechanic, & A. Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. [ay, J., & Anderson, G. (1988). The hospital services market: A disequilibrium analysis. Southern Economic Journal, 54(3), 656-665. [ealth Facilities Division, Colorado Department of Health. (1991). Unpublished data. [err, B., Abrahain, H., & Anderson, W. (1991). Length of stay in a general hospital psychiatric unit. General Hospital Psychiatry, 13, 68-70. [ornbrook, M., & Goldfarb, M. (1983). A partial test of a hospital behavioral model. Social Science & Medicine, 17(10), 667-680 165

PAGE 174

[ughes, R. (1990). Public information on private practice: The availability of physician inpatient data. Hospitals and Health Services Administration, 35(1), 55-69. rerson, G., & Norpoth, H. (1987). Analysis of variance. California: Sage Publications. K., & Ellis, R. (1987). A comparison of psychiatric services utilization in single group practice under multiple insurance systems. Advances in Health Economics and Health Services Research, .8,, 175-194. Jskow, P. (1979). The effects of competition and regulation on hospital bed supply and the reservation quality of the hospital. Bell Journal of Economics, pp. 421-447. :anaan, S. (1991). Financing policy for mental health services. Health Affairs, 10(2), 207-211 . eefer, G. (1990). Frequency of diagnosis, psychiatric admissions, January to June, 1987. Unpublished raw data. University of Colorado Health Sciences Center . eill, S. (1986). The changing patient population in general hospital psychiatric services. General Hospital Psychiatry, .8,, 350-358. iesler, C., & Sibulkin, A. (1984). Episodic rate of mental hospitalization: Stable or increasing? American Journal of Psychiatry, 141, 44-48. iesler, C., & Sibulkin, A. (1983). Proportion of inpatient days for mental disorders. Hospital and Community Psychiatry, 34(7), 606-611. iesler, C., Sibulkin, A, Morton, T., & Simpkins, C. (1991). Characteristics of psychiatric discharges from non-federal, short term specialty hospitals and general hospitals with and without psychiatric and chemical dependency units: The hospital discharge survey data. Health Services Research, 25(6), 881-906. iesler, C., & Simpkins, C. (1991). Changes in psychiatric treatment of children and youth in general hospitals: 1980-1985. Hospital and Community Psychiatry, 46, 601-604. iesler, C., & Simpkins, C. (1992). Changes in diagnostic case mix in psychiatric care in general hospitals, 1980-1985. General Hospital Psychiatry, 14, 156-161. 166

PAGE 175

iesler, C., Simpkins, C., & Morton, T. ( 1990). Predicting length of hospital stay for psychiatric patients. Hospital and Community Psychiatry, 41(2), 149-154. im, H. (1989, February). Demand for psychiatric care is expected to maintain the current building boom. Modern Healthcare, pp. 38-39. im, H. (1990a, May). Psychiatric hospitals growth lags to second slowest pace since 1985. Modem Healthcare, pp. 57-61. im, H. (1990b, June). Psychiatric expansion slows as focus shifts. Modern Healthcare, pp. 56-62. irshner, L. (1982). Length of stay of psychiatric patients: A critical review and discussion. Journal of Nervous and Mental Disorders, 170(1), 27-33 .oco, L. (1988, June). Adolescent psych costs seen pressuring claims. National Underwriter, pp. 16-17 .oyanagi, C. (1990). The missed opportunities of Medicaid. Hospital and Community Psychiatzy, 41(2), 135-138. ralewski, J., Gifford, G., & Porter, J. (1988). Profit vs. public welfare goals in investor-owned and not-for-profit hospitals. Hospitals and Health Services Administration, 33(3), 311-329. agoe, R. (1986). A community-based analysis of regional differences in hospital stays by diagnosis-related groups. Inquizy, 23(3), 183-190. amb, H. (1981). What did we really expect from deinstitutionalization? Hospital and Community Psychiatry, 32(2), 105-109. amb, H. (1982). Young adult chronic patients: The new drifters. Hospital and Community Psychiatzy, 33( 6), 465-468. amb, H. (1992). Is it time for a moratorium on deinstitutionalization? Hospital and Community Psychiatry, 43(7), 669. ave, J. (1989). Commentary. In C. Taube, D. Mechanic, & A. Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. 167

PAGE 176

.ave, J., & Frank, R. (1990). Hospital supply response to prospective payment as measured by length of stay. Advances in Health Economics and Health Services Research, 11, 1-26 .ave, J., & Goldman, H. (1990). Medicare financing for mental health care. Health Affairs .2(2), 19-21. .ee, F. (1987, March). Purchasers address escalating psychiatric and substance abuse utilization. Employee Benefits Journal, pp. 9-13 .evenson, A. (1982). The growth of investor-owned psychiatric hospitals. American Journal of Psychiatcy, 139, 902-907 .evenson, A. (1983). Issues surrounding the ownership of private psychiatric hospitals by investor-owned hospital chains. Hospital and Community Psychiatcy, 34(12), 1127-1131. evenson, A. (1984). Challenges and opportunities for independent private psychiatric hospitals. Psychiatric Annals, 14,(5), 350-356. ewin, L., Derzon, R., & Marqulies, R. (1981). Investor-owned and for-profits differ in economic performance. Hospitals, 55, 52-53. ewin, L., Eckels, T., & Miller, L. (1988). The provision of uncompensated care in not-for-profit hospitals. New England Journal of Medicine, 318(18), 1212-1215. ight, D. (1986). Corporate medicine for profit. Scientific American, 255(6), 38-45. ivingstone, R., Taylor, J., & Crawford, S. (1990). Factors related to length of hospitalization of children with mental disorders. Hospital and Community Psychiatcy, 41(2), 193-195. utz, S. (1992). Psych providers buffeted by payer pressures, stricter scrutiny. Modern Healthcare, 22(20), 72-78. yles, C., & Young, J. (1987). Hospital ownership and psychiatric services: Implications of ownership and reimbursement on length of stay and availability of services. Journal of Medical Systems, 11(1), 45-57. 168

PAGE 177

l!armor, T., & Gill, K. (1990). The political and economic context of mental health care in the United States. In I. Marks & R. Scott (Eds.), Mental health care delivezy: Innovations. impediments and implementation. Cambridge: University Press. l!ason, J., Louks, J., & Backus, F. (1991). Patient-predicted length of stay and diagnostic-related-group reimbursement. American Journal of Psychiatzy. 142(3), 369-371. l!attes, J. (1982). The optimal length of hospitalization for psychiatric patients: A review of the literature. Hospital and Community Psychiatzy, 33(10), 824-828. !lcClure, W. (1982). Toward development and application of a qualitative theory of hospital utilization. Inquizy, 19, 117-135. ilcDowald, D., McGeary, R., Meidinger, E., & Hay, R. (1980). Interrupted time series analysis. California: Sage Publications. !lcGuire, T. (1981). Financing and demand for mental health services. Journal of Human Resources, 16(4), 501-522. !lcGuire, T. (1989a). Financing and reimbursement for mental health services. In C. Taube, D. Mechanic, & A Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. ilcGuire, T. (1989b ). Combining demand and supply-side cost sharing: The case of inpatient mental health care. Inquizy, 26, 292-303. !lcGuire, T., Dickey, B., Shively, G., & Strumwasser, I. (1987). Differences in resource use and cost among facilities treating alcohol, drug abuse and mental disorders: Implications for design of a prospective payment system. American Journal of Psychiatzy, 144(5), 616-620. 1cGuire, T., Mosakowski, W., & Radigan, L. (1990). Designing a state-level payment system for inpatient psychiatric services in Medicaid. Administration and Policy in Mental Health, 18(1), 43-54. 1cGuire, T., & Weisbrod, B. (1981). Perspectives on the economics of mental health. Journal of Human Resources, 16(4), 494-500. 169

PAGE 178

icLaughlin, C. (1988). Measuring small area variation in hospital use: Site of care versus patient origin data. Socio-Economic Planning Science, 22(4), 177-184. icLaughlin, C., Normolle, D., Wolfe, L., & Griffith, J. (1989). Small area variations in hospital discharge rates. Medical Care, 27(5), 507-521. iechanic, D. (1989a). Assumptions in economic modelling of the mental health services system. Inquiry, 26, 304-306. iechanic, D. (1989b). The evolution of mental health services delivery. In C. Taube, D. Mechanic, & H. Hohman (Eds.}, The future of mental health services research. Washington, DC: NIMH. fechanic, D. (1991). Strategies for integrating public mental health services. Hospital and Community Psychiatry, 42(8}, 797-801. fechanic, D., & Aiken, L. (1987). Improving the care of patients with chronic mental illness. New England Journal of Medicine, 317(26). lezochow, J., Miller, S., Seixas, F., & Frances, R. (1987). The impact of cost containment on alcohol and drug treatment. Inquiry, 238(5}, 506-510. lichalon, M., & Richman, A. (1990). Factors affecting length of stay in a psychiatric intensive care unit. General Hospital Psychiatry, 12, 303-308. Iiller, D. (1983). Handbook of research design and social measurement (4th ed.}. New York: Longman. Iiller, M. (1988). The role of substitutes in policy analysis: Acute care services in Medicaid programs. Journal of Health Politics. Policy and Law, 13(3}, 499525. loore, K., & Coddington, D. (1991). How to break the cost-shifting spiral. Journal of Healthcare Benefits, pp. 31-35. lullner, R., & Hadley, J. (1984). Interstate variations in the growth of chain operated proprietary hospitals. Inquiry, 21, 144-151. rational Association of Private Psychiatric Hospitals. (1990). 1988 NAPPH annual Survey (Final Report). Washington, DC: NAPPH. 170

PAGE 179

rational Institute of Mental Health. (1984). By Schulberg, H. The treatment of psychiatric patients in general hospitals: A research agenda and annotated bibliography (DHHS Publication No. ADM84-1294). Washington, DC: U.S. Government Printing Office. ational Institute of Mental Health. (1988). One month prevalence of mental disorders in the United States. Archives of General Psychiatzy, 45, 961. ational Institute of Mental Health. (1990a). Mental health. United States. 1990. Washington, DC: Government Printing Office. ational Institute of Mental Health. (1990b ). National plan for research on child and adolescent mental health disorders. (DHHS Publication No. ADM 901683). Washington, DC: U.S. Government Printing Office. azario, S. (1992, September 15). National Medical is cited in lawsuit by two main insurers. Wall Street Journal, p. A9. ilfson, M. (1990). Treatment of depressed patients in general hospitals with scatter beds, cluster beds, and psychiatric units. Hospital and Community Psychiatzy. 1106-1111. ardes, H. (1990). The demise of a major innovation: Carter's 1980 Community Mental Health Systems Act in Reagan's hands. In I. Marks & R. Scott (Eds.), Mental health care delivezy: Innovations. impediments. and implementation. Cambridge: University Press. attison, R., & Katz, H. (1983). Investor-owned and not-for-profit hospitals. New England Journal of Medicine, 309(6), 347-353. auly, M., & Redisch, M. (1973). The not for profit hospital as a physician's cooperative. American Economic Review, 63(1), 87-99. epper, R., Kirshner, M., & Ryglewicz, H. (1981). The young adult chronic patient: Overview of a population. Hospital and Community Psychiatzy, 32(7), 463-469. oole, C. (1990). Unpublished data, Colorado Office of State Planning and Budgeting. revost, J. (1982). Youthful chronicity: Paradox of the 80s. Hospital and Community Psychiatzy, 33(3), 173. 171

PAGE 180

'ruitt, D., & Weiner, J. (1989). The AACAP and hospitalization for children and adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 28, 21-25. tedick, R., Stroup, A, Witkin, M., Atay, J., & Manderscheid, R. (1989). Private psychiatric hospitals, United States: 1983-84 and 1986. Mental health statistical note no. 191. NIMH. tedick, R., Witkin, M., Atay, J., Fell, A., & Manderscheid, R. (1990). Patient care episodes in mental health organizations, United States:. Selected years between 1955 and 1986. Mental health statistical note no. 191. NIMH. teeves, P., Bergwall, D., & Woodside, N. (1984). Introduction to Health Planning. Virginia: Information Resources Press. tegister, C., Sharp, A., & Biven, D. (1985). Profit incentives and the hospital industry: Are we expecting too much? Health Services Research, 20(2), 225240. tehak, S. (1990). Consequences of the elimination of CON in Colorado. Unpublished data. Colorado Department of Health. telman, A. (1983). Investor-owned hospitals and health care costs. New England Journal of Medicine, 309, 370. tice, D., & Kelman, S. (1989). Measuring comorbidity and overlap in the hospitalization cost for alcohol and drug abuse and mental illness. Inquiry, 26, 249-260. tice, Kelman, S., Miller, L., & Dunmeyer, S. (1990). The economic costs of alcohol and drug abuse and mental illness: 1985. Report submitted to the Office of Financing and Coverage Policy of the Alcohol, Drug Abuse, and Mental Health Administration, U.S. Department of Health and Human Services. University of California: San Francisco. Uce, T. (1983). The impact of changing Medicare reimbursement rates on physician-induced demand. Medical Care, 21(8), 803-815. Uce, T. (1984). Physician-induced demand for medical care. Advances in Health Economics and Health Services Research, 55, 129-160. 172

PAGE 181

J., & Luft, H. (1985). The impact of hospital market structure on patient volume, average length of stay, and cost of care. Journal of Health Economics, 333-356. M. (1989). Psych: PPS may be out, but the pressure is on. Hospitals, 63(11), 28-29. M. (1961). Bed supply and hospital utilization. Hospitals, 35, 36-42. M. (1976). Hospital utilization and the health care system. American Journal of Public Health, 66(10), 953-954. J. (1988). Efficiency and the supply of hospital beds in metropolitan areas. Journal of Public Health Policy, 393-402. D. (1982). Regional variations in hospital use: Geographic and temporal patterns of care in the U.S. Massachusetts: Lexington Books. B. (1983). State mandates for mental health insurance: What is their cost? Hospital and Community Psychiatzy, 34(3), 207-208. A, Steinwachs, D., & Salkever, D. (1984). The effect of hospital payment methods on the pattern and cost of mental health care. Hospital and Community Psychiatzy, 35(5), 456-459. A., & Taube, C. (1989). Analysis of the effect of disproportionately large share of low-income patients on psychiatric costs in general hospitals. Inquizy, 26. 216-221. argent, M. (1989). Update on programs for the homeless mentally ill. Hospital and Community Psychiatzy, 40(10) 1015-1016. callet, L. (1990). Paying for public mental health care: Crucial questions. Health Affairs, 2(2), 117-124. charfstein, S. (1988). Privatization: Economic opportunity and public health. American Journal of Psychiatry, 145, 611-612. cheffler, R., & Miller, A. (1989). Demand analysis of mental health services use among ethnic subpopulations. Inquizy, 26, 202-215. 173

PAGE 182

:cheffler, R., & Watts, C. (1986). Determinants of inpatient mental health use in a heavily insured population. Journal of Human Resources, 21(3), 338-358. :cherl, D., & English, J. (1984). Current trends in the financing of psychiatric services: The initial response of psychiatry to prospective payment. Psychiatric Annals, 14(5), 332-339. :chlesinger, M., & Dorwart, R. (1984). Ownership and mental health services: A reappraisal of the shift toward privately owned facilities. New England Journal of Medicine, 311(5), 959-965. :chlesinger, M., & Mechanic, D. (1993). Challenges for managed competition from chronic illness. Health Affairs, 12, 123-137. ,chulberg, H., & Manderscheid, R. (1989). The changing network of mental health services delivery. In C. Taube, D. Mechanic, & A. Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. chulz, R., Greenley, J., & Peterson, R. (1984). Differences in the direct costs of public and private acute inpatient psychiatric services. Inquiry, 214(4), 380393. ederer, L., Katz, B., & Manschreck, T. (1984). Inpatient psychiatry: Perspectives from the general the private and the state hospital. General Hospital Psychiatry, 180-190. haughnessy, P. (1982). Methodological issues in per capita measurements in health care. Health Services Research, 17(1), -81. beets, J., Prevost, J., & Reihman, J. (1982). Young adult chronic patients: Three hypothesized subgroups. Hospital and Community Psychiatry, 33(3), 197-203. heingold, S. (1986). Unintended consequences of Medicare's national prospective payment rates. Health Affairs, 4 ), 5-22. hortell, S., & Hughes, E. (1988). The effects of regulation, competition and ownership on mortality rates among hospital inpatients. New England Journal of Medicine. 318(7), 1100-1107. loan, F., & Valvona, J. (1986). Why has hospital length of stay declined? An evaluation of alternative theories. Social Science and Medicine, 22(1), 63-73. 174

PAGE 183

loan, F., & Vraciu, R. (1983). Investor-owned and not-for-profit hospitals: Addressing some issues. Health Affairs, 25-37. tano, M. (1989). An analysis of the evidence of competition in the physicians services market. Journal of Health 197-211. :ano, M., Cromwell, J., Velkey, J., & Suad, A (1985). The effects of physician availability on fees and demands for doctor's services. Atlanta Economic Journal, 13(2), 51-60. :arr, P. (1982). The social transformation of american medicine. New York: Basic Books. :ate Issues Forum. (1989, July). State health planning report. Washington, DC: American Hospital Association. :einwachs, D. (1989). Patterns of care methodologies: Utility for assessing costs and outcomes of mental health care. In C. Taube, D. Mechanic, & A Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. :roup, A, & Manderschied, R. (1988). The development of the state mental hospital system in the U.S.: 1840-1980. Journal of the Washington Academy of Sciences, 78, 59-68. :rumwasser, I., Paranjpe, N., Udow, M., & Share, D. (1991). Appropriateness of psychiatric and substance abuse hospitalization: Implications for payment and utilization management. Medical Care, 22, 77-90. albott, J. (1981). The emerging crisis in chronic care. Hospital and Community Psychiatzy, 32(7), 447. aube, C. (1989). The economics of mental health services through 1986: Empirical studies. Administration and Policy in Mental Health, 16(3), 115-130. aube, C. (1990). Funding and expenditures for mental illness. In National Institute of Mental Health. Mental health. United States. 1990. Washington, DC: U.S. Government Printing Office. aube, C., & Goldman, A (1989). State strategies to restructure psychiatric hospitals: A review. Inquizy, 2Q, 146-156. 175

PAGE 184

aube, C., Goldman, H., & Salkever, D. (1990). Medicaid coverage for mental illness: Balancing access and costs. Health Affairs, 2(1), 5-18. aube, C., Lee, E., & Forthofer, R. (1984). Diagnosis-related groups for mental disorders, alcoholism and drug abuse: Evaluation and alternatives. Hospital and Community Psychiatry. 35, 452-455. hompson, J., Burns, B., & Taube, C. (1988). The severely mentally ill in general hospital psychiatric units. General Hospital Psychiatry, 10, 1-9. orrey, E. (1990). Economic barriers to widespread implementation of model programs for the seriously mentally ill. Hospital and Community Psychiatry, 41(5), 526-530. nited States General Accounting Office. (1988). Medicaid: Views on changes needed in mental health benefits (Report NO. GAO/HRD-88-96 FS). Washington, DC: U.S. Government Printing Office. nited States General Accmmting Office. (1990). Medicaid: Sources of information on mental health services (Report No. GAO/HRD-90-100). Washington, DC: U.S. Government Printing Office. nited States General Accounting Office. (1991). Available research shows that capacity is only weakly related to volume (Report No. GAO/PEMD-91-7). Washington, Dc: U.S. Government Printing Office. 'allen, J. (1987). Resource use by psychiatric patients in community hospitals: The influence of illness severity, physician specialty and presence of a psychiatric unit. Advances in Health Economics and Health Services Research, .8, 103126. 7eisbrod, B. (1987). Research issues in economics and mental health. Advances in Health Economics and Health Services Research, .8, 303-305. 'ells, K (1992). Medicare's PPS implementation and increase in the recoding of psychotic depressive disorder: Some thoughts on the reasons. General Hospital Psychiatry, 14, 153-155. 'ells, K, & Brook, R. (1989). The quality of mental health services: Past, present and future. In C. Taube, D. Mechanic, & A. Hohman (Eds.), The future of mental health services research. Washington, DC: NIMH. 176

PAGE 185

rennberg, J., Barnes, B., & Zubkoff, M. (1982). Professional uncertainty and the problem of supplier-induced demand. Social Science and Medicine, 16, 811824. ridem, P., Pincus, H., Goldman, H., & Jencks, S. (1984). Prospective payment for psychiatric hospitalization: Context and background. Hospital and Community Psychiatry, 35(5), 447-451. rilson, P. (1981). Hospital use by the aging population. Inquiry, 18, 332-344. rilson, P., & Tedeschi, P. (1984). Community correlates of hospital use. Health Services Research, 19(3), 333-355. rorthington, N., & Piro, P. (1982). The effects of hospital rate-setting programs on volumes of hospital services: A preliminary analysis. Health Care Financing Review, 47-65. 177