1 STATE COMMUNITY BENEFIT REQUIREMENTS AND PRIVATE HOSPITALS PROVISION OF UNCOMPENSATED CARE By YUE ZHONG A DISSERTATION PRESENTED TO THE S UPERVISORY COMMITTEE OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE RE QUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012
2 2012 Yue Zhong
3 To my husband, Xingdi Hu, and to my parents, Daojin Zhong and Lixin Chen
4 ACKNOWLEDGMENTS I will always be thankful for having had the opportunity to study at the University of Florida and be come a health services researcher. First of all I give my special thanks to my advisor and supervisory committee chair Dr. Niccie Lee McKay, for her guidance, support and patience during my graduat e studies at UF and through the whole process of dissertation writing Most importantly, her mentorship provided a well structured research experience in fulfilling my long term research goals. For me, Dr. McKay is more than an advisor. She is a friend who sets my heart aglow and lights up my student life at UF. I also thank my other committee members. My sincere thanks go to Dr. Jeffery Harman for his assistance and guidance in my methodology and results interpretation, and patience in providing me with th e foundation for becoming a health services researcher. I thank Dr. Paul Duncan for his encouragement through my entire time in the program and suggestions policy background and on the structure of my study Dr. Jessica Schumacher and Dr. Babette Brumback helped me refine my study design and models. Their advice greatly improved my study. Finally and most importantly, I thank my husband, Xingdi Hu, who has been an enormous help during the past four years. His love and support give me the strength and confi dence to make me strong and courageous so to face the challenge head on. My special thanks go to my parents for their faith in me and sacrifice to let me study in the U.S. and establish a future career in health services research.
5 TABLE OF CONTENTS pag e ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 Background ................................ ................................ ................................ ............. 14 Rationale and Research Questions ................................ ................................ ........ 15 Contribution to Literature and Policy Implication ................................ ..................... 16 2 BACKGROUND AND LITERATURE REVIEW ................................ ....................... 18 Overview of the Uninsured and the Provision of Uncompensated Care ................. 18 Uncompensated Care ................................ ................................ ...................... 19 Determinants of Uncompensated Care : Empirical Findings ............................. 21 Overview of Governmen t Policy Related to Uncompensated Care ......................... 23 Rationale for Government Intervention ................................ ............................. 23 Brief History of Policy Related to Universal He alth Care ................................ .. 25 Government Policy on Uncompensated Care ................................ .................. 27 Ownership Differences, Federal Tax Exemptions and Community Benefit ............. 33 Differences in Hospital Ownership Structure ................................ .................... 33 A History of Federal Regulation Regarding Tax Exemption s ............................ 34 Provision of Community Benefit by Nonprofit Hospitals ................................ ... 38 State Community Benefit Requirements ................................ ................................ 42 Overview ................................ ................................ ................................ .......... 42 States with the Strictest Level of Community Benefit Requirements ................ 45 States with a Moderately Strict Level of Comm unity Benefit Requirements ..... 47 States with a Less Strict Level of Community Benefit Requirements ............... 48 States with a Minimal Level of Community Benefit Requirements .................... 49 Empirical Evidence on State Community Benefit Laws ................................ .... 50 3 THEORETICAL FRAMEWORK AND HYPOTHESES ................................ ............ 55 Overview ................................ ................................ ................................ ................. 55 The Perfectly Competitive Market and Market Failure in the Hospital Market ......... 55 Provision of Uncompensated Care Hospital Level Determinants .......................... 59
6 Hospital Mission ................................ ................................ ............................... 59 Hospital Characteristic s ................................ ................................ .................... 63 Provision of Uncompensated Care Market Level Determinants ............................ 66 Economic Models ................................ ................................ ............................. 66 Empirical Findings ................................ ................................ ............................ 67 Provision of Uncompensated Care Impact of Government Policy ......................... 68 Resource Dependence Theor y ................................ ................................ ......... 69 Institutional Theory ................................ ................................ ........................... 71 Theory Integration and Research Hypotheses ................................ ................. 74 4 METHODOLOGY ................................ ................................ ................................ ... 80 Data Sources ................................ ................................ ................................ .......... 80 Empirical Models ................................ ................................ ................................ .... 81 Resear ch Hypothesis 1 and 2 ................................ ................................ .......... 81 Research Hypothesis 3 and 4 ................................ ................................ .......... 82 Variable s and Study Design ................................ ................................ .................... 82 Dependent Variable ................................ ................................ .......................... 82 Independent Variables ................................ ................................ ..................... 84 Covariates ................................ ................................ ................................ ........ 84 Study Design ................................ ................................ ................................ .... 86 Analytical Data Set ................................ ................................ ................................ 87 Methods and Sensitivity Analysis ................................ ................................ ............ 88 Econometric Models ................................ ................................ ......................... 88 Sensitivity Analysis ................................ ................................ ........................... 93 5 RESULTS ................................ ................................ ................................ ............. 100 Descriptive Statistics ................................ ................................ ............................. 100 Dependent and Independent Variables ................................ .......................... 100 Hospital Mission and Characteristics ................................ .............................. 102 Market Characteristics ................................ ................................ .................... 102 Multivariate Analysis ................................ ................................ ............................. 103 Model 1 ................................ ................................ ................................ ........... 104 Model 2 ................................ ................................ ................................ ........... 104 Model 3 ................................ ................................ ................................ ........... 105 Model 4 ................................ ................................ ................................ ........... 106 Other Findings ................................ ................................ ................................ 106 Sensitivity Analysis ................................ ................................ ............................... 106 6 CONCLUSION AND DISCUSSION ................................ ................................ ...... 127 Summary of Major Study Findings ................................ ................................ ........ 127 Conclusions ................................ ................................ ................................ .......... 129 Policy Implications ................................ ................................ ................................ 131 Limitations ................................ ................................ ................................ ............. 133 Future Research ................................ ................................ ................................ ... 134
7 APPENDIX : EXAMPLE OF THE MCR WORKSHEET S 10: HOSPI TAL UNCOMPENSATED CARE ................................ ................................ .................. 135 LIST OF REFERENCES ................................ ................................ ............................. 136 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 147
8 LIST OF TABLES Table page 2 1 Categories of states on level of community benefit requirements ....................... 54 3 1 Characteristics of four market structur es ................................ ............................ 77 4 1 List of variables ................................ ................................ ................................ ... 95 4 2 Descriptive statistics comparing the whole sample and the analytical data set .. 98 4 3 Descriptive statistics comparing the whole sample and the analytical data set .. 99 5 1 Descriptive statistics for study sample, categoric al variables (percentage, %) 110 5 2 Descriptive statistics for the study sample, continuous variables (mean(sd)) ... 111 5 3 The HL M results for Model 1. Dependent variable: log (UC_BED). n=6,544 .... 112 5 4 The HLM results for Model 2. Dependent variable: log (UC_BED). n=4,871. ... 113 5 5 The HLM results for Model 3. Dependent variable: log (UC_BED). n=3,574. ... 114 5 6 The HLM results for Model 4. Dependent variable: log (UC_BED). n=2,680. ... 115 5 7 Other findings comparing results for states with and without community benefit requirements. Dependent variable: UC_BED ................................ ....... 116 5 8 Sensiti vity analysis: Model 1 results with different dependent variables ........... 117 5 9 Sensitivity analysis: Model 2 results with different dependent variables ........... 118 5 10 Sensitivity analysis: Model 3 results with different dependent variables ........... 119 5 11 Sensitivity analysis: Model 4 results with different dependent variables ........... 120 5 12 Sensitivity analysis: comparing results for states with and without community benefit requirements. Dependent variable: TOTALUC_BED .......................... 121 5 13 Sensitivity analysis: Model 3 results with different independent variables ........ 122 5 14 Sensitivity analysis: Model 4 res ults with different independent variables ........ 123 5 15 Sensitivity analysis: comparing the study sample with the imputed sample ..... 124
9 LIST OF FIGURES Figure page 2 1 State community benefit requirements ................................ ............................... 54 3 1 The profit maximization model ................................ ................................ ............ 78 3 2 The utility maximization model quantity maximization ................................ ....... 78 3 3 The trade off between quality and quantity ................................ ......................... 79 5 1 Uncompensated care cost per bed, CB=1/0, 2005 2009 ................................ .. 125 5 2 Uncompensated care cost per bed, STRICT=1 4, 2005 2009 .......................... 125 5 3 U ncompensated care cost per bed trend, CB=1/0, 2005 2009 ......................... 126 5 4 Uncompensated care cost per bed trend, STRICT=1 4, 2005 2009 ................ 126
10 LIST OF ABBREVIATION S AALL Am erican Association of Labor Legislation AHA American Hospital Association AMA American Medica l Association ARF Area Resource File BEA Bureau of Economic Analysis BLS Bureau of Labor Statistics CBO Congressional Budget Office CCMC Committee on Costs of Me dical Care CHA Catholic Health Association CMS Centers for Medicare & Medicaid Services CON Certificate of Need CPI Consumer Price Index DSH Disproportionate Share Hospital EMTALA Emergency Medical Treatment and Active Labor Act ESRD End stage renal diseas e facilities GAO United States Government Accountability Office G ME G raduate M edical E ducation HCRIS Healthcare Cost Reporting Information System HFMA Healthcare Financial Management Association HHA Home health agencies HLM Hierarchical Linear Model HRSA Health Resources and Administration Services IME Indirect Medical Education IPPS Inpatient prospective payment system
11 IRS Internal Revenue Service JCT Joint Committee on Taxation LAUS Local Area Unemployment Statistics LOS L ength of stay MAR Missing at Ra ndom MCAR Missing Completely at Random MCR Med icare Cost Report MMACS Medicare/Medicaid Automated Certification Survey MS DRGs Medicare Severity Diagnosis R elated G roups PMATF Public Medical Assistance Trust Fund PPACA Patient Protection and Affordable Car e Act PPS P rospective p ayment s ystem SAHIE Small Area Health Insurance Estimates SID State inpatient data SNF Skilled nursing facilities TQM Total Quality Management UPL Upper Payment Limit
12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy STATE COMMUNITY BENEFIT REQUIREMENTS AND PRIVATE HOSPITALS PROVISION OF UNCOMPENSATED CARE By Yue Zhong August 2012 Chair : Niccie L ee McKay Major : Health Services Research Uncompensated care is an overall measure of hospital care provided for which no payment was received from the patient or an insurer. As the uninsured population increases each year so does the amount of uncompensa ted care provided In the U.S., a nonprofit health provider with tax exempt status is required by the IRS to provide a certain amount of benefit to the community, including uncompensated care. Several states have stricter standards and evaluation criteria to ensure that nonprofit hospitals justify their tax benefits The objective of this study was to examine the association between state community benefit requirements and uncompensated care. The study was a longitudinal stu dy using panel data for general acute care nonprofit and for profit hospitals in the Healthcare Cost Report Information System (HCRIS) from 2005 to 2009. Uncompensated care was measured as uncompensated care costs per bed adjusted for inflation using the 2005 Consumer Price Index (CPI) I ndependent variables of primary interest were whether or not hospitals were located in states with community benefit requirements and the level of strictness of community benefit requirements.
13 Results from Hierarchical Li near Models showed no evidence of a systematic benefit provision of uncompensated care. However, when combing costs of both public programs and uncompensated care, state community b enefits requirements were significantly positively related with private ho spitals provision of uncompensated care Results also showed that differences in the provision of uncompensated care between nonprofit and for profit hospitals were considerably gre ater i n states with community benefit requirements than states without such policies. The study findings have several implications for policymakers. First, states may need to reconsider the benefit derived from the requirements and whether the benefit exce ed the spending on administration of the program. Second, states community benefit requirements may be necessary to maintain organizational legitimacy for nonprofit hospitals. Finally, this study provides empirical support for the requirements in the rece nt health care reform regarding community health needs assessment and reporting guidelines
14 CHAPTER 1 INTRODUCTION Background The number of uninsured in the United States is rising, which poses serious health policy issues. The uninsured lack access to pr imary care doctors and preventive care, and are more likely to seek care in emergency departments and therefore incur higher costs when being treated. In the US, most of the financial burden of providing medical services to the uninsured is borne by hospit als for which no payment is received These costs for uncompensated care, plus uninsured patients lack of access to health care, are two key issues that have been addressed by federal and state governments through various health policies and in the recen t ly passed health care reform. The amounts of uncompensated care provided are not evenly distributed across different types of hospitals. Public hospitals generally account for a larger share of the uncompensated care burden than nonprofit and for profit hospitals, although nonprofit hospitals are the most common type of hospital in the US. For nonprofit hospitals, their mission requires them to distribute any surplus to the community rather than to shareholders. Nonprofit hospitals which evolved from alm shouses poorhouses and voluntary hospitals, and therefore have historically, cared for the poor and the sick. In addition, tax benefits s are provided to nonprofit hospitals to encourage their provision of charity care. However, whether or not nonprofit h ospitals justify their tax exemptions is a controversial issue. Carreyrou and Martinez ( 2008) argued that the gap between nonprofit hospitals wealth and what they g iv e back the community is growing and question ed the b illions of tax exempt dollars they receive each year Th is paper used one Chicago nonprofit
15 hospital as an example to illustrate this problem. The hospital spent over $ 1 billion to rebuild the entire campus and open a new hospital. In contrast the hospital only spent $20.8 million on chari ty care, which accounted for less than 2% of its revenues and only a small proportion of what it received in tax breaks, compared to public hospitals in the same area, which spent 14% of their revenues on charity care. The federal government has issued sev eral standards to regulate the behavior of nonprofit hospitals with respect to community benefit However, there is no official definition of community benefit and the Internal Revenue Service (IRS) does not require a minimum standard for the amount of com munity benefit a nonprofit hospital must provide in order to be tax exempt. In addition, with the growth of third party pay ment, the introduction of the Medicare prospective payment system (PPS), and managed care penetration some have argued that the beha vior al differences between nonprofit and for profit hospitals are lessening According to Scott and Davis (2007) for example, the control of health care organizations has been shifted from trustees and physicians to managerial and financial team s which h a ve changed their focus from community service and clinical and medical science to cost containment and efficiency. Rationale and Research Questions In considering the issues above, several states have established guidelines that require annual reports of community benefit provided by nonprofit hospitals. Some states provide a detailed definition and examples of community benefit Among these states, some have specific instructions on the minimum standards for community benefit reporting instruments, and penalties for noncompliance. Provisions regarding uncompensated care are always the major category in community benefit requirements.
16 Because some states have implemented community benefit laws and some have not t h e major purpose of this study is to exami ne whether private hospitals in states with community benefit requirements provide more uncompensated care than comparable hospitals in states without such requirements In addition, states are classified into five categories based on the level of strictne ss of community benefit requirements. The second purpose of this study is to examine whether the level of uncompensated care provided by private hospitals will be higher in states with stricter level of community benefit requirements than in those with les s strict requirements Thus, t he two research questions to be addressed are: Question 1. What is the association between state community benefit requirements and private hospitals provision of uncompensated care? Question 2. What is the association betwe en levels of strictness of state community benefit requirements and private hospitals provision of uncompensated care? Contribution s to Literature and Policy Implication Two studies have examined the effect of community benefit requirements on the communi ty health orientation of a hospital and health promotion services (Ginn and Moseley 2006; Gin n Shen, and Moseley 2009). However, the stud ies did not address the laws effect on uncompensated care nor did they provide a standard definition of community ben efit requirements. Several studies have examined the level of community benefit in certain states (Bazzoli, Clement, and Hsieh 2010 ; Congressional Budget Office 2006 ; Cram et al. 2010; Nicholson et al. 2000; Norton and Staiger 1994; Schneider 2007; Sutton and Stensland 2003 ). Among these studies, only two studies mentioned state community benefit requirements In addition, they were based on limited data. Consequently, the report by the Congressional Budget Office (2006) stated
17 that no firm conclusions coul d be drawn as to the effect of state community benefit requirements based on data from only a limited number of states. This study contributes to the literature by examining, for the first time, the relationship between state community benefit requirements for nonprofit hospitals and the level of uncompensated care provided. Moreover, the study defines community benefit requirements based on two national reports (Hellinger 2009 ; United States Government Accountability Office 2008 ) and classif ies states into five categories based on the level of strictness of requirements. The study thus also contributes to the literature on the definition and strictness of community benefit requirements. Finally, no study to date has used the Medicare Cost Report (MCR) to an alyze uncompensated care at the national level which may be due to the fact that the Centers for Medicare & Medicaid Services (CMS) did not start to collect data on uncompensated care until 2003. This study used the MCR from 2004 to 20 09 to examine the i nfluence of state community benefit requirements. The policy implications of the study will address whether or not those states without community benefit requirements should implement similar laws. The study also will provide evidence as to whether or not state and local governments should construct stricter community benefit standards when they are facing the direct loss of revenue from nonprofit hospitals tax exemptions. The empirical results will also inform policy makers about the possibilities of est ablishing stricter evaluation criteria and a minimum level of uncompensated care to be provided by nonprofit hospitals to justify their tax benefits
18 CHAPTER 2 BACKGROUND AND LITER ATURE REVIEW This study examines the effect of state community benefit r equirements on private hospital provision of uncompensated care. This chapter first provides a general overview of the uninsured and uncompensated care in the US. After describing government policy related to uncompensated care the chapter then discusse s the ownership structure differences between private nonprofit and for profit hospitals, including federal regulations on tax exemptions an d community benefit. The final section of the chapter summarizes state community benefit requirements. Overview of the Uninsured and the Provision of Uncompensated Care Americans obtain health insurance in various ways. More than half the adult s under 65 years old receive health insurance through their employers (Kaiser Family Foundation 2011), w hile Medicare covers t hose above 65 years old. Medicaid covers low income famil ies with children, as well as the aged, blind and disabled. Other s may purchase health insurance in the individual market but at a higher price. Still many in the U .S. are uninsured During the rec ent economic downturn, many lost their jobs and thus their private insurance coverage, but were not eligible for public programs and could not afford the cost in the individual market. In 2010, a total of 49.1 million nonelderly people in the United States were uninsured compared to 45.5 million in 2004 ( Kaiser Family Foundation 2011) Studies have showed that most of the uninsured are poor and come from working families with y oung adults, minorities and persons with less education being more likely to be uninsured. The high rate of uninsurance is a serious health threat because the uninsured receive only half the medical services received by those who have health insurance
19 coverage ; t he uninsured also receive less preventive care and therapeutic treatme nt and are more likely to be sicker when diagnosed ( Davis and Rowland 1983) The poorer health outcome s of the uninsured may subsequently result in higher e xpenses for hospital s and the government. Moreover, due to the limited ability to pay for care, the uninsured often face higher charge s for hospital services than those covered by private or public health insurance (Weissman 2005) When an uninsured individual is hospitalized, part of the cost is covered by out of pocket payment, but the remainder typica lly ends up as uncompensated care. Uncompensated Care Uncompensated care occurs when services are provided for which no payment is received from the patient or an insurer and is composed of two parts c harity care and b ad debt. C harity care is care prov ided to patients who cannot afford to pay, whereas bad debt occurs when institutions cannot obtain expected reimbursement for care provided for patients who are able but unwilling to pay their bill (American Hospital Association 2009) Only the level of ch arity care measures the provision of community benefit by health care providers because virtually all businesses incur bad debt However, in practice it is not possible to reliably distinguish between the two categories and thus health services researchers typically use uncompensated care as an approximation for charity care. Although uncompensated care is defined as services for which no payment is received, hospitals and other health institutions usually receive additional revenues from the government or other sources to offset revenues lost from those who do not pay for care. Thus it is important to note that uncompensated care only refers to payment that is not received from the patients receiving the care (Duncan 1992).
20 In 2008, the total cost for unco mpensated care wa s about $57 billion, compared to $40.7 billion in 2004 (Kaiser Family Foundation 2011) In general, those who were uninsured for a full year received less than half the amount of care received by those covered by private health insurance but paid more out of pocket. If those who were uninsured in 2008 were fully covered, the incremental cost of the extra care they would receive would be approximately $122.6 billion which is about 5% of the total health care spending and 0 .8% of GDP altho ugh possible saving s could be generated through improving efficiency in the health system ( Hadley 2008). In the U S hospitals account for over 60% of care for the uninsured with o ffice based physicians and clinics accounting for less than 20% each (Had ley and Holahan 2003). A study by Zwanziger and Khan (2008) using national data found that, on average uncompensated care accounted for 4% of total hospital expenses. It is important to note that those estimates refer to direct cost s whereas uncompensate d care also takes up resources that could create more profits for hospitals ; this is the opportunity cost of uncompensated care ( Ferrier, Rosko, and Valdmanis 2006). The major financial s upport f or uncompensated care comes from public funds, which covered approximately 75% of total uncompensated care in 2008 (Hadley et al. 2008). Medicaid Disproportionate Share Hospital ( DSH ) payments and other supplemental payment program s support hospitals with large number s of low income patients while Medicare provide s funds through Medicare DSH payment s and Indirect Medical Education (IME) hospital payments. Stat e and local government s also subsidize uncompensated care through tax exempt ions and public assistance programs. Private funds also may cover part of the cost of uncompensated care through
21 cost shifting and higher charge s for patients with private insurance (Wessman 2005) Studies have found that the amount of available funds directly influences the provision of uncompensated care. For example, Dunn and Chen (1 994) found that increased government payment for uncompensated care improved patients access to medical care and reduced the uneven distribution across hospitals. Determinants of Uncompensated Care : Empirical Findings This section describes studies that have investigated factors influencing the amount of uncompensated care provided by hopsitals. The focus here is on the chronological development of this literature ; a later chapter again draws on these articles in providing the rationale for the control va riables included in the study. Thorpe and Brecher (1987) investigated the relationship between public hospitals and access to care for the uninsured poor. The data was from a joint survey in 1982 by the American Hospital Association (AHA) and the Urban Ins titute The study found that a presence of public hospitals increased the amount of uncompensated care offered. Thorpe, Seiber and Florence (2001) used national data from 1991 to 1997 to investigate the impact of h ealth maintenance organization s (HMO s ) on the provision of uncompensated care. The study predicted that, because uncompensated care was financed by profits derived from insured patients, the growth in managed care and HMOs would reduce the profits and care for the uninsured. They found that if ma naged care penetration increased by a ten percentage point, hospital total profit margin reduced by two percentage point and the level of uncompensated care decreased by 0 .6 percentage. Weissman, Gaskin, and Reuter (2003) examined the relationship between managed care penetration and the care of uninsured patients in teaching hospitals
22 during the 1990s. Results indicated that the market share of uninsured patients grew at teaching hospitals compared to other hospitals and these increases were primarily inc urred primarily in areas with higher managed care penetration. A study by Rosko (2004) investigate d factors influencing the level of uncompensated care. Using panel data from 1995 1998 on Pennsylvania hospitals the study found that the provision of uncom pensated care was positively related to hospital profitability, the care of the uninsured by neighboring hospitals hospital size, number of emergency room visits, and the unemployment rate. Ferrier, Rosko, and Valdmanis (2006) examined how uncompensated care affect ed profitability. Using a sample of Pennsylvania hospitals in 2002, results indicated that hospitals had they not had the burden of uncompensated care could have produced 7% more output at the best practice frontier. Thus, even if h ospitals could operate efficiently, they would still face financial stress from providing uncompensated care. Based on data for Florida hospitals from 1998 to 2002, McKay and Meng (2007) examined the effect of managed care on hospitals provision of uncom pensated care. The results indicated that higher levels of managed care penetration were associated with decreased uncompensated care. Zwanziger and Khan (200 8 ) studied safety net hospitals and provided a guideline to identify these hospitals Using the A HA annual survey from 1990 to 2000, the study developed three measures: the socioeconomic status of the local area, Medicaid intensity, and uncompensated care burden and market share. Using these measures, t he study distinguished the group of safety net ho spitals from the non safety net
23 hospitals. However, the data driven result s were not stable. Thus, there was no stable group of hospitals that could always be designated as safety net hospitals. A study by Hsieh, Clement, and Bazzoli (2010) examined the e ffects of market and organizational characteristics on provision of uncompensated care. Analyzing longitudinal data from Virginia from 1998 to 2004, results indicated that hospital efficiency and market demand were positively associated with the provision of uncompensated care. Hospitals with more Medicare or Medicaid patients were more likely to provide less uncompensated care. Overview of Government Policy Related to Uncompensated Care This section provides an overview of government policy rel ated to uncompensated care in the United States. However, it i s necessary to first understand the role of the federal and state government s in the history of the healthcare system and their impact on the financing and deliver y of medical care. Thus, a fund amental issue is the rationale for government intervention in the health care system. The next section discusses government policy on uncompensated care, such as the Hill Burton Act, DSH and UPL programs uncompensated care pools, and other related health policies. Rationale for Government Intervention Public i nterest t heory. Under certain conditions, p erfec tly competitive private markets lead to the most efficient allocation of goods and services ( Santerre and Neun 2000 ). One rationale for government inte rvention stems from market failure in health care, such as the existence of asymmetric information between patients and providers. The public interest theory posits that legislators and policy makers will act on behalf of the interest of the public and as sumes that the two basic objectives of government intervention are to correct inefficient market practices and to redistribute medical
24 resources based on need instead of affordability (Feldstein 2006 ; Neun and Santerre 2000 ). Health care suppliers in impe rfect markets such as monopoly and oligopoly operate i nefficient ly because the market price exceeds marginal cost (Feldstein 2006). Inefficiencies also may occur due to externalities In the redistribution part of the theory government intervention resu lts in the redistribution of wealth through cross subsidization among different groups of people (Feldstein 2007). For instance, Medicare and Medicaid provide medical care for their respective populations by taxing those with higher income s Economic t heo ry. The economic theory of regulation stemmed from the failure of the public interest theory to explain certain phenomen a such as the establishment of entry barriers (Feldstein 2006). The economic theory assumes that the policy market is similar to the ec onomic market in that l egislators are suppliers who provide legislative benefit to special interest groups in return for better chances of reelection. The demanders in the market are the special interest groups who hope to use the regulatory power of gover nment in their self interest For example, Certificate of Need (CON) law s can be viewed as an entry barrier for hospitals prohibit ing further entry of competitors and thus benefit ing existing hospitals. Social j ustice t heory. Equity in the health care s ystem may be defined as the delivery of medical care to those who need it, based on demographic s (age and gender) personal view s and clinically evaluated need s while inequity occurs when the main determinants of access to health care depend on social str ucture (race or ethnicity), health beliefs and enabling factors (income or wealth) (Andersen 1995). The
25 social justice theory, based on the principles of equality and social solidarity argues that government intervention is necessary in order to secure th e well being of everyone (Powers and Faden 2006). Brief History of Policy Related to Universal Health Care Uncompensated care would not be a policy concern under a system of universal health care Thus, a brief review of policy related to universal health care is in order. In 1915, the American Association of Labor Legislation (AALL) for the first time proposed government sponsored national health insurance but the proposal failed due to opposition from the American Medical Association ( AMA ) (Starr 1982). Again in 1932 the AMA strongly protested recommendations from the Committee on Costs of Medical Care (CCMC) on delivery of comprehensive medical care and extension of all basic public health services for the entire population (Ross 2002). The Social Secu rit y Act of 1935, originally included national health insurance but was negotiated down to only provide funds for maternal and infant care The federal government became more involved after the Depression and World War II. In 1948, President Truman propos ed a national health insurance program Although was defeated again largely through opposition by the AMA the proposal led to the implementation of the Hospital Survey and Construction Act, commonly known as the Hill Burton Act A fter 1965, with the ena ctment of Medicare and Medicaid, the federal govern ment continued to expand its role in subsiding hospitals provision of medical care to the elderly and the poor. However, once the expenditures on those public programs skyrocketed and exceed ed the expecte d budget, several pieces of legislation were implemented to control costs For example, Medicare changed the way it reimburse d
26 hospital s from a retrospective, cost based payment system to Medicare Severity Diagnosis R elated G roups (MS DRG s ), which is a pro spective reimbursement system based on clinical problems that require similar amount s of hospital resources ( Medicare Payment Advisory Commission 2011). Additionally, m any states have limited Medicaid eligibility criteria and have reduced Medicaid payments to hospitals and physicians in order to stem rapidly increasing costs and decreasing tax revenues Another proposal for universal health care, by the Clinton administration, was also unsuccessful The Clinton plan had an employer mandate, requir ing all e mployers to contribute to insurance premiums for the ir employee s and managed competition, through which people would have the opportunity to choose from a variety of health plans (Oberlander 2007). The Clinton plan would have expanded government power in regulating insurance practices (such as the prohibition of advers e selection) and cost control The failure of th is legislation reflected incrementalism as the princip al mean s of the U.S. system of government, as well as the strong involvement of special i nterest groups in the policy making process (Patel and Rushefsky 2006). The Patient Protection and Affordable Care Act (PPACA) was signed into law by President Obama and enacted on March 23, 2010 (Kaiser Family Foundation 2011) It focuses on expanding a ccess to insurance coverage, control ling health care cost s and improving quality of care by health care system reform. It emphasize s the individual insurance market, the small business insurance market, and the uninsured. One of the key elements regarding universal health care is the expansion of public programs like Medicaid and SCHIP and subsidies for American s and legal residents who are uninsured to pay for health insurance. However, the reform is not fully effectiv e until
27 2014 and it is estimated that about 20 million will be still uninsured under the PPACA ( Buettgens and Hall 2011). Government Policy on Uncompensated Care Uncompensated care was formally acknowledged by the federal government as a policy concern in 1946 with the passage of the Hill Bu rton Act (Blewett et al. 2003; Weissman 1996 ). The Act provide d public funds and loans for public and nonprofit hospital construction in response to a hospital shortage after the Depression and World War II, particular ly in rural areas. To ensure patients their ability to pay the obligation of a hospital qualified under the Hill Burton Act is to provide a reasonable volume of free medical care to the indigent population (Weissman 1996). The law also require d that ho spital s receiving grants or loans under the Act provide medical services to eligible individuals for a twenty year period The amount of uncompensated care 10% of the federal assistance they received, adjusted for inflatio n, or 3% of their annual operating c osts, minus the amount of reimbursement they rece ( Coleman 2005 ). Facilities receiving Hill Burton funds must set up specific plans to fulfill the annual obligation of providing a certain level of uncompensated services They also need to publish an allocation plan, a written individual notice as well as a posted notice to the whole community, including a description of services at a free or reduced cost to indigent patients, the monetary v alue of services to be provided, and the eligibility criteria. These eligib ility criteri a are primarily based on the federal poverty level although facilities can set their own level up to twice the federal guideline Such strict requirement s were not fav ored by hospitals
28 and many dropped out of the program ; fewer than 900 hospitals remain in the Hill Burton program ( Weissman 1996 ). All h ospitals have an obligation to provide a certain level of charity care ; they cannot turn away patients from the emergen cy room due to the Emergency Medical Treatment and Active Labor Act ( EMTALA ) requirement. And n onprofit and public hospitals must provide a certain level of community benefit including charity care as part of the organization s mission. State and federal governments have implemented a variety of p rograms to subsidize hospitals that bear a higher uncompensated care burden (Jocobson et al. 2005) Davidoff et al. (2000) investigated the effects of changes in payment generosity, Medicaid eligibility and Medica re managed care on uncompensated care using the AHA annual survey of hospitals from 1990 through 1995. Results indicated that nonprofit and public hospitals responded positively to public payment generosity by increasing the level of uncompensated care wh ereas nonprofit hospitals responded negatively to Medicaid managed care penetration. And expansion of Medicaid eligibility reduced uncompensated care costs among for profit and public hospitals. Holahan and Zedlewski (1991) also concluded that Medicaid exp ansion could significantly reduce the number of uninsured The sections below discuss other government programs that affect provision of uncompensated care. States are required to make additional payment s to hospitals serving a disproportionate share of Med icaid and low income patients. Some s tates also established uncompensated care pool s to provide additional p aymen ts to hospitals with greater uncompensated care costs and provided direct tax subsidies to
29 public and teaching hospital to fund ca re for the uninsured (Blewett et al. 2003 ; Gaskin 1997 ). Some states established state subsidized programs to expand health insurance coverage, thus reducing the number of uninsured patients. Certification of Need (CON) laws and r egulations governing hospi tal conversions are two additional regulations at the state level that might directly influence the provision of uncompensated care Disproportionate Share Hospital and Upper Payment Limit Programs DSH payme nts, first introduced in the Omnibus Budget Rec onciliation Act of 1981 are the major source subsidiz ing (Coughlin and Liska 1997). The Medicare DSH program was originally intended to compensate hospitals treating a higher volume of low income Medicare patient s ( Fagnani and Tolbert 1999). With the passage of time, however, Medicare DSH payments have been used to ease hospitals accessibility to all low income and uninsured patients. Medicaid is funded by state governments with matching payments from the federal government. The Medicaid DSH program require s that states take into acc ount the situation of hospitals with higher share s of indigent care by raising payment rat es for those hospitals (Hearne 2005). In m any states, Medicaid DSH funds are crucial to the survival of safety net hospitals (Rousseau and Schneider 2004). DSH payments, which skyrocketed from less than $10 billion in 1991 to over $17 billion in 1992, were a major determinant of the rapid growth i n Medicaid spending during the early 1990s (Mechanic 2004). T h e Medicaid Upper Payment Limit (UPL) program is another way for states to draw down extra federal funding UPL programs allow states to make supplemental payments to providers serving a large n umber of uninsured and underinsured patients.
30 The total payment can exceed Medicaid rates as long as it is below the level of Medicare reimbursement. In 2002, 23 states had from one to nine UPL programs (Coughlin, Bruen and King 2004). The presumption is that DSH payments will ease hospitals financial burden from unprofitable medical services to the indigent and thus encourage them to provide more uncompensated care However, using a national data set based on the AHA annual survey of hospitals from 1990 to 2000, Lo Sasso and Seamster (2007) found that DSH spending did not have a significant effect on uncompensated care. Another study, by McKay, Zhong and Hu (2011), used data on Florida hospitals o ver the period of 2004 2008 to investigate the impact of Medicaid supplemental payments, including DSH payments and UPL programs, on the provision of uncompensated care. Results showed that hospitals receiving supplemental payments provid ed more uncompensated care. A study by Davidoff et al. (2000) showed that r esponses to public payments differed according to hospital owner type with nonprofit and public hospitals respond ing positively to payment generosity. Fonkych and Melnick (2010) used data for California hospitals from 2001 through 2006 to examine the impa ct of DSH subsidies on prices paid by uninsured patients in association with hospital ownership. The study found that the majority of uninsured care was provided by non DSH nonprofit hospitals. In addition, hospitals with DSH subsidies charged only a sligh tly lower price to uninsured patients than those without subsidies. Uncompensated c are po ol s. Anothe r way states have subsidized hospital provision of uncompensated car e is by using uncompensated care pools to provide additional payment s to hospitals for the provi sion of car e to the poor and/or uninsured
31 and by provid ing direct tax subsidies to public and teaching hospita ls to fund care for the uninsured (Blewett et al. 2003). In New Jersey the Uncompensated Care Trust Fund (which expired in the 1990s) wa s designed to spread the costs of uncompensated care across all hospitals through the establishment of a uniform uncompensated care surcharge on hospital bills (Bovbjerg, Cuellar and Holahan 2000; Derek 2007) The state government collected revenue from th e f und and redistributed it to hospitals based on the level of uncompensated care provided; the objective was to e liminate the financial disincentives incurred in the provision of care to indigent patients Thus, hospitals with a lower uncompensated care b urden were net contributors, while those with a higher burden receive d subsidies from the pool. Gaskin (1997) examined the effect of the Trust Fund on hospital provision of uncompensated care and found that the Trust Fund encouraged hospitals to increase the provision of uncompensated care Dunn and Chen (1994) found that the program resulted in a more even distribution of uncompensated care. In addition to New Jersey, Massachusetts and New York implemented uncompensated care pools in 1985 and 1983, respec tively. Evidence from New York State also indicated that the uncompensated care pool did affect the distribution of care for the uninsured, although the effect was moderate (Spen c er 1998). Coverage e xpansions. Minnesota implemented health care reform in 1 992 with legislation establishing the MinnesotaCare (still in operation), a health insuran ce program funded by the state government through taxes collected from all health providers (Blewett et al. 2003) By expand ing health insurance coverage and decreas i ng the population of the uninsured, one objective of the program is to reduce
32 the burden of hospital provision of uncompensated care in Minnesota. Blewett el al. (2003) examined the impact of MinnesotaCare and found that higher enrollment s in the Minnesota Care program were associated with reduced expenditures on uncompensated care; t he 5 year (1992 1996) savings in hospital specific uncompensated care expenses were estimated to be $58.6 million In 1984, Florida established a revenue pool, known as the Publ ic Medical Assistance Trust Fund (PMATF), for the medically indigent. The PMATF addressed both the uneven uncompensated care distribution across hospitals and health insurance expansion. Descriptive data showed that the number of people eligible for Medica id increased by almost 1 million in the eight years after the passage of the PMATF (Mitchell and Norton 1996), and it achieved the goal of leveling the playing field among different kinds of hospitals by distributing more money to public and nonprofit hosp itals. Certificate of Need (CON) l aws. CON regulation has been adopted by many states to restrict the entry of hospitals or expansion of hospital services, with the goal of controlling hospital costs and preventing duplicat ed facilities (Hellinger 2009). Based on interviews of current and former administrators of Florida s CON program, Campbell and Fournier (1993) concluded that CON laws improved access to hospital care for the indigent. First, hospitals with a higher share of indigent care were more likel y to be awarded CON approvals; thus CON laws may be viewed as an incentive to hospitals providing large amounts of indigent care Second, CON laws prevent ed the entry of potential competitors, hence ensuring the financial viability of e xisting providers to sustain unprofitable services, like care for the uninsured.
33 Regulations on h ospital c onversions In the 1990s, a number of hospitals converted to a different ownership type. Because p ublic and nonprofit hospitals have traditionally provided more communit y benefit including uncompensated care for the uninsured, changes in ownership might have an effect on the provision of community benefit especially for hospitals converting from public or nonprofit ownership to for profit ownership. A survey conducted b y the United States Government Accountability Office (GAO) (1997) found that more than 20 states had enacted legislation to address public concerns about hospital conversions, including questions about the continued provision of uncompensated care. Empiric al evidence from California, Florida and Texas ( Young and Desai 199 8 ) showed that, in the short term, nonprofit hospitals kept the same level of uncompensated care following conversion to for profit ownership, suggesting that a hospital s historical missio n and its role in the community played a key role in determining the provision of uncompensated care. Other studies came to similar conclusions based on analyses of a single state (Needleman, Lamphere and Chollet 1999; Young, Desai, and Lukas 1997). Owners hip Differences, Federal Tax Exemptions and Community Benefit Differences in Hospital Ownership Structure Hospitals in the U.S. have three types of ownership: public (government owned), private nonprofit and for profit Nonprofit hospitals are the most co mmon ownership form, accounting for over 60 % of the approximately 3,900 nonfederal, short term, acute care general hospitals in the U.S in 2003, with public and for profit hospitals each accounting for about 20% (GAO 2005). Because p ublic hospitals are gov ernment owned, they generally cannot refuse care to any patient. As mentioned before, they empirically provide more charity care than private hospitals (Cunningham and Tu 1997;
34 GAO 2005; Mann et al. 1997; McKay and Meng 2007; Thorpe and Brecher 1987; Weine r et al. 2008; Zwanziger and Khan 2008 ) Although private nonprofit hospitals may earn a surplus in revenues over costs the community is the owner and nonprofit hospitals must distribute the surplus to re investment in the hospital or to community servic e In general n on profit hospitals are assumed to maximize utility, subject to a net income constraint and the amount of need by the medically indigent in the community Thus nonprofit hospitals should u se some of their financial surpluses to fund the pr ovision of uncompensated care and provide more uncompensated care when the need is greater. For example, i n Pennsylvania where there are no public general hospitals, nonprofit hospitals accounted for most of th e burden of uncompensated care (Rosko 2004). F or profit hospitals first appeared in the late 1960s (Sloan 1998). As pure profit seekers, they seek to maximize shareholders and view uncompensated care as a cost rather a mission Thus, in theory for profit hospitals will provide less uncompensa ted care than nonprofit hospitals. The evidence indicates that for profit hospitals tend to be located in areas where the likelihood of having to provide charity care is lower and where there are more alternative suppliers for such care (Barro 1998 ; Clemen t and White 2002 ; Norton and Staiger 1994; Pauly 1987; Thorpe and Phelps 1991). A History of Federal Regulation Regarding Tax Exemption s A nonprofit health provider with tax exempt status is expected to provide a certain amount of benefit to the community For those area s like Pennsylvania where nonprofit hospitals are the major supplier s their tax benefits can be thought of as supplements
35 for public goods, an indirect form of government purchased care for the indigent population (CBO 2006 ; Pauly 1987 ) The tax exemptions provided to nonprofit hospitals contain six major categories. At the federal level, the exemption categories are federal corporate income tax, tax exempt bond financing and charitable contributions and at the state and local level, the categories are state corporate income tax, state and local sales tax, and local property tax es Nonprofit hospitals also benefit from tax exemptions received by donors. A ccording to the Joint Committee on Taxation (JCT), the estimated total value of tax e xemptions provided to nonprofit hospitals reached $12.6 billion nationwide in 2002 The largest category was forgone local property tax es which accounted for nearly 25% of the total tax exemptions (CBO, 2006). N ot every nonprofit entity is a charitable i nstitution and tax exemption status depends on the provision of care for public health purposes. From 1956 to 1998, t he IRS issued several standards to define which hospitals would qualify for tax exempt status under section 501(c ) ( 3) In the original law, a hospital that is operated to the extent of its financial ability for those not able to pay for the services rendered and not exclusively for those who are able to and expected to pay was entitled to tax exempt status. However, as described below, in 1969 the IRS removed the charity care provision requirement and added the emergency room operation requirement. Then in 1983, it went a step further to eas e the requirement by eliminating the emergency room requirement (Hellinger, 2009 ). Revenue Ruling 56 185 The IRS implemented Revenue Ruling 56 185 in 1956 describing four key featur es of hospitals exempt from tax First, the hospital must be a
36 nonprofit organization organized and operated exclusively for charitable purpose s and for the care of the sick. must be operated to the extent of its financial ability for those not able to pay for the services rendered and not exclusively for those who are able and expected to pay The hospital cannot refuse patients in need of hospital care who cannot pay for such services. facilities must not be restricted to a particular group of physicians and surgeons, such as a medical partnership or association, to the exclusion of all other quali fied doctors Finally, t he hospital must not distribute its net benefit to any private shareholder or individual (Rev. Rul. 56 185, 1956) Historically, hospitals were primarily institution s for care of the sick and poor, so the IRS emphasized the chari table purpose of nonprofit organizations as promotion of health like the relief of poverty and the advancement of education and religion Especially before the introduction of Medicare and Medicaid in 1965, for profit hospitals were rare and m ost hos pitals were charitable institutions s er ving indigent persons. The IRS also indicated th at hospitals could charge patients who were able to pay, in order to support medical care for those who were unable to pay. Revenue Ruling 69 545. With the development of medical science and technological progress, hospitals began to attract more middle and upper income patients and were transformed from purely charitable organizations to organizations that could generate profits. After Medicare and Medicaid were first i ntroduced in 1965, the government began funding hospital care for the poor and elderly which released the burden on hospitals and increased patients access for medical care. Revenue Ruling 69 545, which was issued by t he IRS in 1969 first transformed th e emphasis from
37 charity care for the poor to benefit for the community as a whole even though the class of beneficiaries eligible to receive a direct benefit from its activities does not include all members of the community, such as indigent members of t he community Free emergency care regardless of affordability was treated as a key determinant of hospitals contribution to the community benefit and qualification for tax exemption. The law also modified the Revenue Ruling in 1956 and removed the requir ements relating to caring for patients without charge or at rates below cost (Rev.Rul. 69 545 1969) Revenue Ruling s 83 157 and 98 15 I n 1983 Revenue Ruling 83 157 eliminated the requirement that a hospital must opera te an emergency room open to all t o be tax exempt. The ruling listed other significant factors, including a board of directors drawn from the community, an open medical staff policy, treatment of persons paying their bills with the aid of public programs like Medicare and Medicaid, and th e application of any surplus to improving facilities, equipment, patient care as indicators that the hospital was operating exclusively to benefit the community (Rev.Rul. 83 157 1983). Then in 1998, Revenue Ruling 98 15 analyzed the issue of joint ventu re s between tax exempt nonprofit and for profit corporations and stated that the nonprofit entity could keep its tax exempt status as long as participation in the partnership furthers a charitable purpose, and the partnership arrangement permits the exemp t organization to act exclusively in furtherance of its exempt purpose and only incidentally for the benefit of the for profit partners ( Rev.Rul. 98 15 19 98). Recent a ctions by the IRS. As tax exempt organizations nonprofit hospitals are required to file IRS Form 990, which is the primary tool to acquire financial and other information to oversee the behavior of Â§ 501 (c) (3) organizations Form 990 was
38 revised in 2007 as a result of growing concerns about providing a more accurate measure of hospitals p rovision of community benefit For the first time, the form identified specific activities that could be classified as community benefit including charity care costs, Medicaid shortfalls and other community benefit (Bazzoli, Clement, and Hsieh 2010) The most significant change in Form 990 is Schedule H, which requires nonprofit hospitals to report the number of medically indigent receiving charity care, the percentage of operating expenses attribut able to charity care, information about Medicaid shortfall s and the community health improvement programs (Hellinger 2009). However, as recommended by the AHA, the IRS postponed implementing Form 990 until fiscal year 2010. The PPACA (2010) modified the tax exemption standards applicable to tax exempt hospitals, requiring organizations to assess community benefit needs at least every three years and propose at strategy to meet the community needs; the legislation is effective for taxable years after March, 2012. Hospitals are subject to a $50,000 excise tax every year if they fail to submit their community benefit needs assessment. community benefit activities will be reviewed by t he Secretary of the Treasury at least once every three years. Tax exempt hospitals should establ ish and implement a financial assistance policy and limit the charges for emergency care f or other patients who are eligible for financial assistance. Provision of Community Benefit by Nonprofit Hospitals Definition of c ommunity benefit Although t he IRS rulings emphasize the community benefit requirement for nonprofit hospitals, they do not provide a clear definition of what constitutes community benefit Two widely adopted definitions come from the Catholic Health Association (CHA) and the AHA. The CHA d efined community
39 benefit as c harity care unreimbursed Medicaid services, as well as activities relating to such commun ity health improvement programs as health profession s education and training, community health education and wellness programs, disease s creeni ng free clinics and school physicals ( CBO 2006 ). Bad debt was not included in the definition nor were unreimbursed Medicare costs which might be due to inefficient hospital operations rather than underpayment, given that Medicare usually pays at a higher rate than Medicaid. The AHA considered both charity care and bad debt as community benefit because poor patients took up a large share of bad debt (AHA 2010). It also included both Medicare and Medicaid shortfalls in that the AHA felt that hospita ls underpayment by Medicare and Medicaid did relieve the governmental financial burden (Bazzoli, Clement and Hsieh 2010) Empirical f indings. W ith the emergence of third party payment, managed care and the Medicare PPS some have argued that fewer behavi or al differences remain between nonp rofit and for profit hospitals (Cram et al. 2010; CBO 2006; Weissman 1996) Accordingly, nonprofit hospitals are facing increased scrutiny to justify their tax exempt status (Bazzoli, Clement, and Hsieh 2010; Cram et al 2010; Nicholson et al. 2000; Norton and Staiger 1994; Schneider 2007 ). The evidence regarding the provision of community benefit by nonprofit hospitals varies among studies, often reflecting different definitions of community benefit Norton and Staiger (1994), for example analyzed the effect of hospital characteristics on the amount of charity care provided to the uninsured Based on national data from the 1981 AHA annual survey of hospital s t hey found that for profit hospitals served a similar
40 uninsu red patient volume as thei r counterparts in the same area. However, for profit hospitals were more likely to be locate d in better insured markets in order to avoid the uninsured patients. Using national data from the AHA, MCR and financial data from the t hree largest for profit hospital systems (1996 1998), Nicholson et al. (2000) applied a broader definition when examining nonprofit hospitals provision of community benefit The analysis was based on an economic concept of public goods. Measurement of p ub lic good services included provision of uncompensated care, Medicaid and Medicare shortfalls, expenses for medical research and education, price discounts to privately insured patients, and other unbilled public good services, like AIDS prevention clinics and health screening for the poor. While most studies measured whether nonprofit hospitals matched the tax benefits by providing the same value of community benefit this study used a different benchmark by comparing nonprofits actual level of community b enefit to the sum of for profit hospitals community benefit spending, tax payments and after tax profit in the same market ( the expected level of community benefit that nonprofit hospitals should provide) The final result indicated that nonprofit hospita ls spent less in community benefit than expected. T h e CBO ( 2006 ) measured the level of community benefit in five states (California, Florida, Georgia, Indiana, and Texas) using the MCR and states hospital financial data in 2003 Community benefit included the provision of both charity care and bad debt (as a share of total operating expenses) Medicaid covered services and four specialized services (identified as unprofitable procedures in other studies). Adjusted results showed that in terms of operating expenses, the uncompensated care cost at nonprofit
41 hospitals was only slightly ( 0 6 %) higher than for profit hospitals. Additionally, n onprofit hospitals even provided less Medicaid covered days than did for profit hospitals while they were more likely to provide emergency room care and labor and delivery services. Schneider (2007) investigated the effect of hospital ownership on the delivery of medical services to the indigent population across different definitions of community benefit. Uncompensated car e, as measured by pure charity, bad debt and federal and state income taxes, was the narrow definition of community benefit, while the broad definition added Medicare and Medicaid shortfalls, price discounts and policy discounts. If the narrow definition w as adopted, nonprofit hospitals supplied a significantly lower level of community benefit than for profit hospitals in the same market. However, the difference disappeared when adopting the broad definition of community benefit Cram et al. (2010) used 200 5 state inpatient data (SID) from 10 states to compare the number of uninsured patients in three conditions (AMI, CABG, or childbirth) treated by nonprofit, for profit, and public hospitals. For each condition, uncompensated care was measured as the propor tion of uninsured admissions to each hospital s total admissions. W hile public hospitals served significant more uninsured patients, the study failed to find a significant difference between the amount of uninsured care provided by nonprofit and for profit hospitals. Using financial data from 2005 for California and Florida hospitals, Bazzoli, Clement, and Hsieh (2010) first applied the community benefit definition from the IRS Form 990. When restricted to the IRS definition of community benefit, results s howed that nonprofit hospitals did not provide adequate community benefit relative to for profit hospitals after considering their tax benefits However, nonprofit hospitals did match for
42 profit hospitals after including bad debt and unreimbursed Medicare costs in the community benefit measurement. The study noted that, when examin ing the impact of hospital ownership on the provision of community benefit conclusions strongly depended on the definition adopted. Based on empirical findings, several studies ( Bazzoli, Clement, and Hsieh 2010; CBO, 2006; Schneider 2007 ) suggested that state and local government s should construct stricter charity care standards given that they are at the forefront in terms of the loss of revenue due to hospitals tax exempt stat us. They also recommended that state governments should establish evaluation criteria and a minimum level of charity care standards to ensure that nonprofit hospitals justify their tax benefits s Local governments, because of the fact that they are more fa miliar with the community and residents, should decide whether the benefit s from charity care equal or exceed the forgone tax revenue from nonprofit hospitals (CBO 2006; Folkerts 2009) State Community Benefit Requirements Overview Although federal regula tions require the provision of community benefit by nonprofit hospitals for tax exemptions, the definition of community benefit has not been standardized and adopted in the IRS rulings. Also, the IRS set s no specific guideline as to the level of community benefit that nonprofit hospitals should provide in order to meet the local community needs or match the tax exem ption benefit they receive, and before health care reform, set s no punishment if hospitals fail to comply with the regulations. Approximately half the states have passed legislation requiring nonprofit hospitals to report the amount of community benefit provided annually. Some states further establish a certain amount of community benefit that hospitals have to offer for each
43 year. According to the GAO (2008), a total of 15 states have set requirements in statutes or regulations on community benefit These states typically provide a detailed definition of community benefit which is usually consistent with the CHA definition. Among them, 10 state s have detailed instructions on the provision of and reporting instrument for community benefit ; five states have standards as to the minimum level of community benefit (or charity care) required for nonprofit hospitals to obtain tax exempt status, and fou r states impose penalties on nonprofit hospitals that fail to comply with the requirements. Variations exist in the definition of community benefit between the government standards and the hospital industry guideline (GAO 2008; Hellinger 2009). However, t he definition commonly includes such activities as charity care, the unreimbursed cost of public programs like Medicaid, and other research and education programs that would benefit the local community, while consensus does not exist in defining bad debt a nd unreimbursed Medicare cost as benefit to the community. Specifically, the CHA and Healthcare Financial Management Association ( HFMA ) define charity care as community benefit while the AHA and CMS collect information on both charity care and bad debt in the annual hospital reports. According to the GAO (2008), among the 15 states with community benefit requirements, 14 of them either use an explicit definition of community benefit to include charity care or list that charity care is one of the examples o f community benefit A study by Hellinger (2009) reported that 16 states had enacted legislation to require nonprofit hospitals to report their provision of community benefit since 1990, and 10 of them required hospitals to report the level of charity care provided.
44 The report by the GAO (2008) is the only study that has a definition of the state community benefit requirement, namely hospital to provide health care services or benefit to the community served by the hospital as a condition of maintaining tax exempt status or qualifying as a nonprofit benefit to obtain hospital licensure and legislation that indirectly affects community be nefit like EMTALA and vaccination requirements. States were excluded if they merely repeated the standards in section 501 (c) (3) of the IRS rule, or if they only required an annual report on community benefit activities instead of actually requiring hosp itals to provide a certain level of community benefit The study by Hellinger (2009) included all states with some form of community benefit requirements, but did not provide standardized inclusion criteria. Both studies provided tables that described stat e laws on the provision of community benefit This study will categorize states according to the strictness of their community benefit benefit requirements and the review by Hellinge r (2009). States are classified into five categories: (a) states with the strictest level of community benefit requirements; (b) states with moderately strict community benefit requirements; (c) states with less strict community benefit requirements; ( d ) s tates with a minimal level of requirements on community benefit provisions; and (e) states with no community benefit requirements. States in the first three categories must have implemented laws that require hospitals to provide services or benefit to the community to qualify for tax exemptions and maintain nonprofit ownership status. States in the first two categories must include one or more
45 of the following: a definition or examples of community benefit activities, guideline for bene fit plan s a community benefit annual reporting requirement, a minimum standard, and penalties for noncompliance. Table 2 1 and Figure 2 1 show the classification of states into these five categories. States with the Strictest Level of Community Benefit Re quirements The first category includes states with the strictest legal requirements on community benefit Specifically, these states must have detailed instructions on both the provision of and reporting instrument for community benefit In addition, they must require a minimum standard of community benefit that nonprofit hospitals must meet and/ or implement certain penalties for those that fail to comply with the requirements, or both. Six states fall into this category: Pennsylvania, Texas, West Virginia, Illinois, Indiana, and Maryland. Texas is the only state that has both minimum requirements and penalties. In 1993, Texas became the first state to pass legislation requiring nonprofit hospitals to provide a certain level of community benefit as a share o f total patient revenues (Wood 2000). In order to qualify for tax exempt status, a nonprofit hospital must provide community benefit sponsored indigent health care and other distributed community benefit such as donatio ns, education programs, government sponsored program services, research, and subsidized health servi an amount that must meet one or more of the following standards: First, community benefit must be provided based on local community nee ds, which are determined hospital system, and the tax exempt benefit received by the hospital or hospital system benefit must be equal to or more
46 than 100 % exempt benefit excluding federal income tax. Third, nonprofit hospitals must provide a combined amount of community benefit that sponsored indigent health care are provided in an amount equal to at least four percent of net patient revenue (Texas Health and Safety Code Â§ 311 itals in Texas must file an annual report to the state government and clarify which of the standard(s) have been satisfied and how much uncompensated care has been provided to the community ( Texas Health and Safety Code Â§ 311 ). Hospitals that fail to submit an annual report are subject to a penalty that does not exceed $1,000 per day. Pennsylvania and West Virginia are the other two states that have minimum standard requirements. Pennsylvania passed the Institute of Purely Public Charity Act in 1997, requiri ng that tax exempt institutions must advance a charitable purpose and fulfill any one or more of the following community benefit advancement and provision of education, advancement of religion, prevention and treatment of dise ase or injury, government or municipal purposes, and accomplishment of a purpose that is recognized as important and beneficial to the public ( Bowman 2003 minimum standard requirem ents. For example, provision of uncompensated services or goods must equal at least 75% of the hospital net operating income, but not less than 3% of its total operating expenses. At the aggregate level, provision of uncompensated goods or services should be equal to at least 5% of the hospital cost of providing goods or services.
47 Similarly, in West Virginia ( W.V. Code of State Rules Â§ 110 3 24), the l egislation explicitly lists the activities included in the definition of community benefit such as charity care, Medicaid shortfalls, health promotion activities, public education programs, donations, etc. Nonprofit hospitals must operate primarily for charitable purposes to qualify for tax exemptions. Unlike Pennsylvania, the West Virginia government requires the nonprofit board of trustees to review the charity care plan every two years to ascertain its effectiveness and determine a specific le vel of charity care based on community needs. Hospitals must achieve this level and provide annual reports. Besides Texas, three states (Illinois, Indiana, and Maryland) impose penalties for noncompliance with community benefit requirements. The state legislatures provide the definition of community benefit which usually includes charity care services, government spons ored indigent health care, health promotion services and health education and research, and requires an annual report of community benefit activities. Nonprofit hospitals in these states would face civil penalties if they fail ed to file the annual report ( GAO 2008). States with a Moderately Strict Level of Community Benefit Requirements The states in this category are those with detailed community benefit requirements but without a minimum standard or penalty for noncompliance (California, Idaho, New Hamps hire, and New York). They usually provide a definition or examples of community benefit requirements for a community benefit plan and annual reports on community benefit activities. California defines community benefit e intended to address community needs and priorities primarily through disease prevention and improvement of health status (California Health and Safety Code Â§ 127340
48 nonprofit hospitals must issue an organizational mission statement, and adop t and annually update a community benefit plan. They must investigate local community needs by consulting with community groups or government officials every three years, and annually report their activities in identifying and addressing the community need s within their mission and financial capacity. Similar ly to California, New Hampshire and New York also require a hospital mission statement, community needs assessment, and annual report of community benefit activities with h benefit p lans being available to the public (GAO 2008). Idaho identifies several activities in the definition of community benefit including charity care, bad debt, under reimbursed public program services, donations and other services provided below actual costs to the community. Tax exempt hospitals with at least 150 patients must file an annual report regarding provision of community benefit ( GAO 2008 ). States with a Less Strict Level of Community Benefit Requirements Five states have less detailed community b enefit requirements (Alabama, Colorado, Mississippi, North Dakota, and Wyoming). They all implemented legislation addressing the provision of community benefit and non profit hospital tax exempt status, but they do not provide a clear definition for activi ties that could be considered as community benefit In addition, they may not require hospitals to file an annual report or provide an annual community benefit plan. Alabama and Mississippi only address the provision of hospital services to charity patien ts in their community benefit requirement (GAO 2008). In order to be exempt from ad valorem taxation, hospitals in Alab ama need to provide at least 15 %
49 to the treatment of charity patients. Mississippi requires that tax exempt hospitals m ust maintain one or more charity wards for charity patients. Colorado, North Dakota, and Wyoming do not provide a detailed description of community benefit services. For example, Colorado defines community benefit for the benefit of an indefini te number of persons by relieving their bodies from disease, GAO 2008, p 58 requires that hospital community services should be for the benefit conferred upon the public and lead to a consequen t relief of the burden upon the state. Indigent care is also included, but will GAO 2008, p 64 ) States with a Minimal Level of Community Benefit Requirements Based on the definition of the GAO (2008), states in this category are identified as states that do not have legal requirements related to community benefit but address community benefit in some way. Some states include community benefit provisions in the context of hospital licensure for all hospitals; examples are Massachusetts, New Mexico, and Rhode Island. Massachusetts and New Mexico require all hospitals to provide free general health care and free emergent care to nonpaying and low income patients ( GAO 2008 ). In order to acquire initial and continued licensure, hospitals in Rhode Island need to meet the statewide standard for the provision of uncompensated care and other community bene fits Connecticut, Georgia, Minnesota, Nevada, and Oregon are states with only community benefit reporting provisions. The GAO (2008) did not classify these states as having community benefit requirements since they either have a voluntary community
50 benef it provisions or only include regulations on reporting without explicitly requiring the provision of community benefit Finally, some states address community benefit outside of codified statutes and regulations (GAO 2008). Utah, for example, discusses co mmunity benefit in it s property tax exemption standards of practice, requiring exempt hospitals to provide a total gift (indigent care, donation of time and money, and etc.) to the community exceeding its property tax liability. Empirical Evidence of State Community Benefit Laws Using two years of data (1994 and 1997) for the state of Texas Wood (2000) benefit In 1994, the total amount of community benefit was $468.6 million, 8% as a s hare of net patient revenues. The amount of charity care provided by nonprofit hospitals in 1994 accounted for almost 50% of total community benefit Three years later (1997), the total amount of community benefit increased to about $1.1 billion, which rep resented 16.8% of reported net patient revenues by nonprofit hospitals. Other community benefit health promotion programs, accounted for the largest share of total community benefit in 1997. Several problems were found when reviewing th e data reported in 1994 and 1997. First, only a small number of hospitals reported the tax benefits from tax exemptions. Thus, the government could not accurately determine whether nonprofit hospitals provided community benefit at least equal to the tax be nefits they received. This was partly due to the fact that no specific and uniform standards existed for determining the level of community benefit not welcomed by the hospital industry. It was hard for smaller hospitals to devote the
51 same percentage of charity care without taking into account location and patient demographic characteristics Kennedy et al. (2010) tested the impact of the Texas community benefit law on hospital charity care using data from 1992 through 1997. Results showed that the law did not lead to increased charity care spending by nonprofit hospitals. More specifically, descriptive statistics showed that nonprofit hospitals spending below the threshold increased the provision of charity to meet the standard, while those above the t hreshold actually decreased charity care spending. Two multivariate regression analyses examined the relationship between state regulations and the volume of uncompensated care across two or more states (CBO 2006; Sutton and Stensland 2003).The CBO (2006) examined the volume of nonprofit Texas. Results indicated that nonprofit hospitals in Indiana and Texas had a significant la rger share of uncompensated care than for profit hospitals. Among the five states, California, Indiana and Texas have requirements that nonprofit hospitals must maintain and report information about community benefit and Texas and Indiana impose penaltie s on nonprofit hospitals that do not report on community benefit In addition, Texas is the only state of the five that requires a minimum volume of community benefit to be provided by nonprofit hospitals. The study concluded that it was difficult to draw firm conclusions regarding possible links between state community benefit requirements for nonprofit hospitals and the volume of community benefit based on data from only five states. However, it is worth noting the positive effect in the Texas and Indiana results.
52 The study by Sutton and Stensland (2003) used data from1996 1998 to compare charity care provision across California, Washington State and Texas. Charity care was th the provision of charity care and bad debt, and uncompensated care deductions from characteristics, Texas hospitals provided a significant larger volume of charity care than did the other two states. And Texas is the only state that implements the stricter standards in reporting and providing community benefit which suggested that hospitals in states with stricter community benefit or charity care requirements may provi de more such services. However, the study was based on data from only three states, and did not included control factors for other market and hospital characteristics that may affect the le vel of charity care provisions Two other studies examined the infl uence of the state community benefit laws on community health orientation and other health promotion services. A study by Ginn and Moseley (2006) examined the effect of state community benefit laws on the provision of community health orientation activitie s. Based on the national AHA annual survey in 2000, ten states (California, Idaho, Indiana, Massachusetts, Missouri, Minnesota, New York, Pennsylvania, Texas and Utah) were identified as states with laws or guidelines that required or encouraged community health promotion services prior to the year 2000. The standards were based on the definition of the Coalition for Nonprofit Health Care and two other studies (Minnesota Statutes Annotated 2002 ; Noble, Hyams and Kane 1998 ). Two dependent variables were inco rporated in the analysis, the community health orientation of a hospital (mission, resources for community benefit activities, etc)
53 and health promotion services (breast cancer screening, child wellness, etc). Results indicated that community benefit laws decreased ownership related differences in found no significant evidence supporting the hypothes i s that community benefit laws had a positive effect on the provision of c ommunity health promotion services. The major limitation of the study is that it is inappropriate to draw strong inferences based on one year cross sectional data. In considering the limitations of the previous study, Ginn, Shen and Moseley (2009) conducte d a longitudinal study from 1996 to 2000 to measure the long term impact of state community benefit laws on community health orientation and hospital based promotion services. They adopted a quasi experimental panel design to compare the changes of the two factors between states with community benefit laws and those without. After adjusting for the control variables, they found that nonprofit hospitals in the states with community benefit laws provided a similar level of community health orientation activit ies and promotion services as that of the nonprofit hospitals in the states without such laws
54 Figure 2 1. State c ommunity benefit r equirements Table 2 1 Categories of s tates on l evel of c ommunity benefit r equirements Levels of Community benefit Requirem ents Total Number Examples of States States with the strictest level of community benefit requirements 6 Illinois, Indiana, Maryland, Pennsylvania, Texas, and West Virginia States with moderately strict community benefit requirements 4 California, Id aho, New Hampshire, and New York States with less strict community benefit requirements 5 Alabama, Colorado, Mississippi, North Dakota, and Wyoming States with a minimal level of requirements on community benefit provisions 9 Connecticut, Georgia, Mass achusetts, Minnesota, New Mexico, Nevada, Oregon, Rhode Island, and Utah States with no community benefit requirements 26 The remaining states in the U.S.
55 CHAPTER 3 THEORETICAL FRAMEWOR K AND HYPOTHESES Overview Th is study combines and adapts economi c theory and organizational theory to examine the factors that influence a hospital s decision making in providing community benefit The model assumes that a hospital s provision of community benefit is influenced by the regulatory environment (the state community benefit requirements) and other control factors. A general form of the model is: c ommunity benefit ( u ncompensated c are) = f ( state community benefit requirements; control factors) The first part of this chapter discusses the perfectly competitive market, market failure and the structure of hospital markets. The next part provides two economic models for hospitals with different ownership, and conceptualize s the relationship between uncompensated care provision and control factors from an economic perspective. Control factors are categorized into three major categories: hospital mission, hospital characteristics and market demand. The last part of the chapter examines the main effect of state community benefit requirements on the level of uncompensa ted care through t he integration of two organizational theories: Institutional Theory and Resource Dependence Theory. The Perfectly Competitive Market and Market Failure in the Hospital Market The main focus of this study is hospitals provision of uncomp ensated care. Hospital supply behavior in general depends on two factors: a hospital s cost structure and the market structure. A perfect ly competitive market possesses seven key features: a. price sensitive consumers; b. a profit maximization goal; c. a l arge number of
56 suppliers and demanders; d. a homogeneous product; e. no entry barriers; f. perfect information for all economic agents; g. nondecr easing costs of production (Santerre and Neun 2000). The first requirement implies an elastic demand such that consumers are sensitive to price changes in the market. The second characteristic implies that all firms want to maximize their profits and therefore they will strive to produce at a minimum cost. The remaining characteristics all focus on market power. F irms in a perfectly competitive market are price taker s ; no firm possesses enough market power to set price s A large number of suppliers with a homogeneous product implies that consumers can always find a perfect substitute for the product of any one firm No entry barriers ensure that nothing prevents new sellers from entering the market. Perfect information occurs when consumers are fully informed and can choose wisely among different sellers. Thus, sellers have to provide better products at a lower pric e in order to increase profits. Under these circumstances, markets deliver the socially optimal level of output. One exception that fails to meet the above criteria is the hospital market Most consumers of hospital services are insured and thus insulated from the full effects of price changes. The hospital market includes private nonprofit and government ownership, as well as hospitals operated for profit. Hospital services are highly heterogeneous, and few hospital markets have a large number of supplier s. Consumers are unlikely to be fully informed due to the technical nature of the services supplied. Hospital markets thus cannot be characterized as perfectly competitive. The other three types of market structure are monopolistic competition, oligopoly, and monopoly. Table 3 1 describes differences among market types. For purposes of this study, a choice of a
57 specific type of market is unnecessary. Rather, the important point is that hospitals will tend to have at least some market powers. The economic j ustification for government intervention in a market is the existence of some form of market failure. Market failure makes the task of making products and delivering services effectively, efficiently and equitably more difficult (Henderson 1999). Market fa ilure may stem from market power, externalities, barriers of entry, and/or imperfect information Market p ower. While pure monopoly is relatively rare, hospitals often have some level of market power that they use to avoid competition (Hendersen 1999). Th ey may differentiate their products or services, provide specialized hospital care, or focus on competition in quality and quantity instead of price. When hospitals possess substantial market power, they will set higher prices and raising prices will reduc e quantity sold. Consumers, however, may not reduce quantity demanded proportionately if they are relatively insensitive to price change ( i.e., demand is inelastic ), thus resulting in higher revenues for suppliers. Externalit ies Sometimes, actions by in dividuals or firms may have spillover effects on other parties (Hendersen 1999). Externalities can be either positive or negative, and on either supply side or on the demand side. An example of an externality in health care is a public health program. Acti vities like the provision of clean water and clean air reduce the incidence of disease and improve public health. In hospital markets, the u ninsured and underinsured incur medical bills that exceed their ab ility to pay, which then results in higher prices for other consumers of hospital services. This is a negative externality because the uninsured raise the price of hospital care for others.
58 Barriers to e ntry. In a perfectly competitive market, entry or exit is unrestricted (Hendersen 1999). Firms enter i nto the market when they see profit opportunities, which drives down prices and increases quantity produced. If profits are lower than expected, firms leave the market, thus driving up prices and profits for firms remaining in the market. Restricted entry in a market results in increased market power for established firms. In addition to licensing and accreditation requirements that pose entry barriers for hospitals, state CON law s are examples of entry barriers in hospital markets. In order to avoid du pli cation of services and waste of resources in many states, hospitals are required to receive government approval before entering into a new market or expanding capacity. However, restricted entry benefit firms in the market in the form of increased market power. Imperfect i nformation. In the medical sector, information is unequally distributed between the medical profession and patients; this is referred to as asymmetric information (Hendersen 1999). Patients are aware of the ir symptoms while they seldom un derstand the underlying medical causes. Thus patients are unable to choose wisely based on price and quality differences among providers, even when providers charge higher prices or provide lower quality services. In this situation, providers may not act o n behalf of the patient s best interests; rather they may put their own interests first. For example, physicians may induce patients to purchase more medica l care than they actually need Summary. Hospital markets thus exhibit market failure. Most hospita ls possess some level of market power. The uninsured pose a negative externality on other
59 consumers of hospital services. Barriers to entry protect hospitals from increased competition, and consumers of hospital services have imperfect information about qu ality and price. Overall, there is a strong case for government intervention in hospital markets. Provision of Uncompensated Care Hospital Level Determinants Hospital Mission A hospital s mission depends largely on its ownership and teaching status. Any o rganization would monitor its managers in light of organizational objectives (Feldstein 2007). Because public hospitals are required to provide uncompensated care and are not influenced by the s tate community benefit requirement s this study focuses on the behavior of private hospitals, including both private for profit hospitals and nonprofit hospitals. Teaching s tatus Teaching status is expected to affect the level of uncompensated care provided While all hospitals seek to provide cost efficient patient care, teaching hospitals also have the objectives of training physicians and care for the poor as core missions Thus, the hospital administrator must consider not only elements that affect operating costs to achieve cost efficiency but also must maintain academic and philanthropic missions. Patients who are unable to pay their bills, are more likely to be accepted by teaching hospitals for teaching and philanthropic purposes. In theory, teaching hospitals would be expected to provide more uncompensated ca re than general nonteaching hospitals. Hospital o wnership. Compared to nonprofit hospitals for profit hospitals have a straightforward objective, namely profit making and its managers are monitored by the shareholders for the maximization of profit (Fel dstein 2007). Because for profit hospitals
60 operate in an imperfectly competitive market, they are expected to set prices and select quantity to maximize their profits, at the level where the marginal revenue equals to the marginal cost (Figure 3 1 ). Unlik e for profit hospitals, nonprofit hospitals have multiple objectives, and they make decisions to maximize the utility of hospital decision makers ( Santerre and Neun 2000 ). Newhouse (1970) discussed the utility maximization model for nonprofit hospitals and postulated that the first element in the utility function was quantity of services provided. Based on the legal status of nonprofit hospitals, profits earned cannot be distributed as dividends to shareholders, whereas profits can be used to expand the org anization and improve the welfare of a community. If their salaries and bonuses are more correlated with firm size and community oriented goals than with profits, managers would seek to provide more services until consumers are satisfied. The public would be also better off if nonprofit hospitals provide more community benefit Figure 3 2 depicts a quantity maximizing nonprofit hospital producing quantity up to the level of Q where the average revenue equals the average cost. Hospitals producing quantity mo re than Q would incur an economic loss. Quantity maximization would allow nonprofit hospitals to increase their market share as well as maintain a good neighborhood image in the community, which would make it easier to gather community support such as dona tions grants, and volunteers, in the future ( Neun and Santerre 2000 ). Another element of the utility maximization model is quality (Newhouse 1970). Since nonprofit hospitals do not distribute profits, the performance of the administrator must be evaluate d by his contribution to the enhancement of the firm s prestige. Quantity of services provided can be one factor affecting prestige, but the quality of the
61 product is probably more important. Quality is difficult to measure, but can be reflected through th e structure, process and outcomes of an organization (Donabedian 1980). For example, the adoption of a new medical technology would enhance the quality of patient care. And any improvement in quality of care could influence such patient outcomes as the hos pital mortality and morbidity rate s Decision makers of nonprofit hospitals need to consider a tradeoff between quantity and quality (Newhouse 1970). H igher quality increases average cost, which would reduce the total quantity, and vice versa Facing a bu dget constraint and this trade off, the managers must jointly determine a mixture of quality and quantity (point A instead of point B) to maximize the total utility function, as shown in Figure 3 3. Based on the economic models for for profit and nonprofi t hosp itals, Banks, Paterson and Wendel (1997) proposed two alternative models for the impact of nonprofit and for profit ownership on uncompensated care. They assumed that in order to maximize utility, nonprofit hospitals would maximize provision of unco mpensated care when facing a zero profit constraint so where = the nonprofit hospital s profit function; P(Q;d)= the average revenue for compensated care; Q= patient days of compensated care; d= demand curve shift parameter; U= unc ompensated care C(Q,U)= variable cost of producing Q and U;
62 F= fixed cost. T he second model assumed that the major goal of for profit hospitals is to maximize profits although fail ing to meet the community needs could negatively impact profits (Banks, Pat erson and Wendel 1997). The hospital s board of trustees need s to make cost effective decision s on the provision of uncompensated care to maximize profits as well as to avoid the expected penalty from the lack of such services. Thus, the second model is: where = the profit hospital s profit function; L(e U)= the function of the gap between the community s expectation of the level of uncompensated care (e) and the hospital s actual supply of uncompensated care (U); P(Q;d), C(Q,U) Q and F are defined as above. Based on the two models above, we predict that for nonprofit hospitals, profits influence their ability to finance uncompensated care. A decrease d market demand would reduce profits available for the supply of uncompensated care. While for for profit hospital s, they would produce the optimal level of uncompensated care when they operate in a cost effective model such that marginal cost was equal to the hospital s marginal revenue for providing uncompensated care (MRu=MCu). E mpirical e vidence. Empirical evidence shows that the uneven distribution of uncompensated care burden among hospitals may cause problems o f patient access hospital competitive disadvantage, and financial difficulties ( Weissman 1996). P ublic hospitals gene rally provide proportionately more uncompensated care than private
63 hospitals (Cunningham and Tu 1997; GAO 2005; Mann et al. 1997; McKay and Meng 2007; Thorpe and Brecher 1987; Weiner et al. 2008; Zwanziger and Khan 2008 ) A study by the GAO (2005), for exa mple, found that uncompensated care as a share of patient operating expenses was 13% for government hospitals, considerably higher than nonprofit hospitals (average 5.3%) and for profit hospitals (average 4.0%). Some researchers have found that n onprofit h ospitals provide more uncompensated care than for profit hospitals ( CBO 2006; GAO 2005; Mann et al. 1997; McKay and Meng 2007; Weiner et al. 2008 ; Zwanziger and Khan 2008 ), although some studies have found no differences or a lower level of uncompensated c are provided by nonprofit hospitals (Bazzoli, Clement, and Hsieh 2010; Cram et al. 2010; Nicholson et al. 2000; Norton and Staiger 1994; Schneider 2007). Empirical studies on teaching status found that p ublic institutions with higher teaching intensity pro vided proportionately the highest levels of uncompensated care compared to other hospital types ( Weissman, Gaskin, and Reuter 2003 ; Zwanziger and Khan 2008). Hospital Characteristics Economic m odels. The economic models from Banks, Paterson and Wendel (199 7) indicated the influence of hospital characteristics and market characteristics on hospital profitability which in turn could affect the provision of uncompensated care. This section first examines hospital characteristics associated with hospital finan cial performance Two studies were used in this section because they included a variety of organizational and market factors from previous studies and provided empirical evidence using datasets in the post PPS era t herefore results were more relevant to h ospitals in the current payment system.
64 Using a 1989 data set of private acute care hospitals in Florida Gapenski, Langland Orban and Vogel (1993) examined the determinants of hospital profitability Hospital characteristics were categorized into organiz ational, managerial, and patient mix variables. Among organizational variables, hospital size significantly influenced hospital profitability, as did ownership and teaching status. Among the managerial variables, the labor intensity, measured by full time employees per adjusted inpatient day, was negatively related to hospital profit margin. In the patient mix variables, Medicaid mix and Medicare mix were two important factors in determining hospital financial performance both being negatively associated w ith hospital profit margin. Younis, Rice, and Barkoulas (2001) examined issues related to hospital profitability using a national data set from the post PPS period ( years 1991 and 1995). The labor intensity and average length of stay (LOS) were found to b e negatively related to hospital profitability. Hospital size, as measured by the num ber of beds, showed consistent agreement with the study by Gapenski, Langland Orban and Vogel (1993). Empirical f indings. In 1986, the US Congr ess passed the Emergency M edical Treatment and Active Labor Act ( EMTALA ) which require s h ospitals and ambulatory services to provide emergency health care regardless of a patient s ability to pay. A survey of Florida hospital emergency department s in 1998 showed that all emergency department s provided substantial amount s of uncompensated care ( Langland Orban, Pracht, and Salyani 2005). Thus, h ospitals with more emergency visits are expected to b ear more uncompensated care. Previous research also has found that u ncompensated
65 care ha s a positive rel ationship with hospital size ( Bazzoli, Clement, and Hsieh 2010; Gaskin 1997 ; Thorpe et al. 2001 ; Rosko 2004) Hospital p rofitability financially supports the provision of uncompensated care and thus higher profitability would be expected t o be associated with higher levels of uncompensated care. S ome studies found a positive relationship between profitability and uncompensated care (Banks 1997 ; Rosko 2004; McKay and Meng 2007 ) although Gaskin ( 1997) found no significant relationship betw een the two Another Medicaid patients also plays a role, with hospitals having a higher percentage of Medicaid patients also having m ore uncompensated care ( Mann et al. 1997; McKay and Meng 2007). M anaged car e also could affect the level of uncompensated hospital care provided (Cunningham and Tu 1997) Increased m anaged care in a market increase s price competition in the market and competition influences (Ban ks, Paterson, and Wendel 1997) and thus presumably a ability to subsidize uncompensated care. Using national data over an eight year period (AHA 1991 1997), Thorpe Seiber, and Florence ( 2001) found that a 10% point increase in managed care penetr ation was associated w ith a 0.6 % decrease in uncompensated care. Other studies showed similar results (Davidoff et al. 2000; Hsieh, Clement; and Bazzoli 2010 ; Lo Sasso and Seamster 2007; McKay and Meng 2007 ; Weissman et al. 2003 ). In contrast studies using data for hospitals i n California and Minnesota found no significant relationship between managed care volume and levels of uncompensated care (Blewett et al. 2003 ; Clement and White 2002 )
66 Provision of Uncompensated Care Market Level Determinants Economic Models Market level characteristics are also related to hospital profitability, which influences the level of uncompensated care provided Gapenski, Langland Orban and Vogel (1993) examined two factors relevant to this study. Area income level was significantly associated w ith hospi tal profitability. People in higher income area s are more likely to be insured and may demand less uncompensated care. Hospital concentration, on the other hand, had no effect on profitability Results from earlier studies, however, varied on the relationship between hospital concentration and profitability, and standard economic theory predicts that increases in competition will reduce excess profits (Scherer 1980). Younis, Rice, and Barkoulas (2001) found that geographic location significantly a ffected profitability. Hospitals located in the southern region of the U.S. were more profitable. They also found that being the only provider in the community gave hospitals monopoly power to earn higher profits. Two utility maximizing models by Frank an d Salkever (1991) and Gruber (1994), discuss a nonprofit hospital s provision of charity care using different assumptions about hospital motivations and the influence of other hospitals supply of charity care in the same market The first model is the p ure altruism model in which a nonprofit hospital s utility relies on the level of a community s unmet need s in the health care market. In this model, a given nonprofit hospital w ou l d reduce its effort to provide charity care if other hospitals ( especiall y public hospitals) in the same market increase supply.
67 The second model is termed the impure altruism model which assumes that the nonprofit hospital s utility relies on both the level of community unmet need and the credits received from supplying charity care (Frank and Salkever 1991; Gruber 1994). In this model, increases in the supply of charity care by other hospitals will not crowd out the level of indigent care in the given nonprofit hospital due to rivalry for public goodwill. T he model prop osed by Banks, Paterson and Wendel (1997) explains the behavior of for profit hospitals in selecting where to locate Since failure to meet community expectations regarding the supply of uncompensated care could lead to a profit reduction and penalties, f or profit hospitals would prefer locations with lowe r expectations or health care market s with the presence of other hospitals such as nonprofit or public hospitals or both Empirical Findings An increase in competition in a market would be expected to af fect the provision of uncompensated care via its effect on hospital profitability However, a profile of uncompensated care over ten years (1983 1995) by Mann et al. (1997) found that hospitals in the most competitive market s shared the greatest amount of uncompensated care with those in the least competitive market s A study by Gruber (1994) argued that revenues and incomes rose more quickly in the most concentrated (less competitive) markets and found that hospitals in more concentrated markets provided more uncompensated care. The presence of a public hospital in the market has been found to decrease the amount of uncompensated care provided by both nonprofit and for profit hospitals (Thorpe and Brecher 1987) In particular, for profit hospitals provis ion of
68 uncompensated care is most sensitive to the presence of a government hospital in the market (Banks Paterson, and Wendel 1997 ; McKay and Meng 2007 ). Other market level characteristics include i ncome level s and unemployment rates, which reflect pati ents ability to pay. Higher income levels and lower unemployment rates also may indicate fewer uninsured persons in the local market, thus leading to lower levels of uncompensated care. Two studies found a positive relationship between unemployment rate a nd the provision of un compensated care ( Banks Paterson, and Wendel 1997 ; Rosko 2004) while others found no significant effect (Blewett et al. 2002; Lo Sasso and Seamster 2007). Provision of Uncompensated Care Impact of Government Policy Th is study uses a macro perspective to examin e the influence of the external environment (state community benefit requirements) on organizational decisions (the amount of uncompensated care to be provided by hospitals) According to Schuler, Rehbein, and Cramer (2002), g overnment policy the structure of markets (through barriers to entry and changes in cost structures due to regulations, subsidies, and taxation); the offerings of goods and services that are permissible; and the sizes of markets Government policy can be viewed either as a mechanism to change external resources that might be crucial for organizations or an institutional pressure to force organizational conformity to the legal en vironment. Thus, th is study integrates two organizational theories as the main theoretical framework: Resource Dependence Theory and Institutional Theory Both theories are used to explain organizational reaction s to change s in the external environment bu t each has limitations. We combine them so that the limitations of one theory are addressed by the other
69 Resource Dependence Theory Resource dependence theory, developed by Pfeffer and Salancik (1978) emphasizes that the environment in which the organi zation is embedded determines organizational activities and outcomes Without control over all essential resources the organization must seek and establish relationship s with other organizations. Therefore, organizations that are more dependent on other o rganizations do not fully control all of the conditions necessary for their survival and need to make accommodations to maintain the exchange relationships. T o avoid this, organizations will make efforts to minimize their dependence on other s and maximize others dependence on them. To achieve that, they wi ll adjust organizational structure s and fit strategies to the external environment in order to acquire the necessary resources to survive and grow. The first basis for control over a resource is possessio n ( Pfeffer and Salancik 1978) For example, knowledge is one resource. Professionals such as doctors have medical knowledge and thus possess power with respect to their patients. Ownership rights are a lso means of possessing resources. Access is another ba sis for control over a resource. It is possible to regulate the access of resources without owning them. For example, the government regulates hospitals opening in a new market or expanding their capacities through CON laws. The third basis for control is the actual use of the resource and the actual users since resources can be used by other than owners. For example while physicians usually are not employees of hospitals, hospitals own medical resources. Therefore, physicians are user s of resources but d o not own them Resource dependence theory states that three core factors explain the way that organizations manage interdependent relations (Scott and Davis 2007). First, social context determines how organizations respond to the world of other organizat ions in
70 open systems. Second, varied strategies can be implemented to pursue the organization s best interest and enhance the autonomy. Finally, power is important since it is a determinant of both strategies within the organization and how they react to t he external environment. Power dependence, first introduced by Emerson (1962), suggests that asymmetric dependence leads to asymmetric power relations between organizations. If a given organization A is dependent on some resources provided by the other organization B B has power over A. As government regulations or interventions become more common, organizations are more likely to adapt the existing environment rather than to change environments ( Pfeffer and Salancik 1978). Rules and regulations ar e important tools of control that can affect access possession and actual use of resources. They can determine the existence and concentration of power over resources. Rules and regulations can also determine the extent to which interdependence relations can be used in the achievement of an action or outcome. The interdependence in the political arena becomes the relationship between those making decisions and those affected by those decisions. The policy maker most often does not directly experience the c onsequence of the decision. Thus, there i s a power imbalance between the two groups. In the health care sector, Alexander and Morrisey (1989) used Resource Dependence Theory to address the determinants of hospital entry into management contracts with outsi de organizations. They assumed that, based on the theory, hospitals would seek affiliation with other organizations only when the resource exchanges were vital to hospitals. Hospitals without the ability or internal expertise to operate efficiently and aut onomously would choose contract management and expect benefit from
71 resource exchange with the management organization, including highly trained managers and financial, accounting, and purchasing professionals. The study used a sample of 321 hospitals with contract management arrangements and a comparison group of 963 hospitals without such arrangements during the period 1980 1983. They found that management effectiveness, regulatory environment, and hospital ownership were all important determinants of cont ract management. Using the Medicare/Medicaid Automated Certification Survey (MMACS) from 1991 to 1992, Banaszak Holl, Zin n, and Mor (1996) applied the resource dependence framework to study the impact of market and organizational characteristics on nursin g home special care units. According to the theory, facilities depend on the resources for survival, and therefore they will make accommodation to guarantee exchange relationships. In this study, hospitals facing the growth in managed care sought to promot e the use of alternatives to inpatient care such as subacute care. Nursing care facilities we re pressured to accommodate market needs and develop subacute care unit in response to hospital demand. In addition, managed care plans preferred less costly alte rnatives to hospitals for patients with special needs, which encouraged nursing care facilities to develop their own subacute care units. Therefore, they found that nursing care facilities in areas with greater HMO penetration and greater hospital supply w ere more likely to operate a subacute care unit. Institutional Theory Institutional theory emphasizes the importance of legitimacy to organizations (Scott and Davis 2007). Legitimacy is defined as a generalized perception or assumption that the actions o f an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions (Suchman
72 1995 p 574). Institutional theorists believe that internal organizational structures are relatively inertial. Sim ilar to Resource Dependence Theory, they suggest that the sources of change are from the external environment. However, this theory addresses the external forces valued by society and by certain institutions in society and how these influence the organizat ion s behavior in searching for social legitimacy rather than a power balance between organizations. Institutional theorists classify social forces into three institutional elements ; regulative systems, normative systems, and cultural cognitive systems. R egulative systems are institutions as systems of rules or governance. Normative systems provide a moral framework for the organizational behavi or ; a nd cultural cognitive systems emphasize the role of common symbolic systems and shared values (Scott and Dav is 2007) DiMaggio and Powell (1983) introduced the concept of institutional isomorphism which is the homogeneity of organizational structure in the process of seeking legitimacy valued by the public or by certain institutions. They pointed out that bes ides resources and customers, organizations also compete for political power and institutional legitimacy, both in social and economic issues The mechanisms involved in institutional isomorphic change s are coercive isomorphism mimetic processes, and norm ative pressures. Coercive isomorphism the major mechanism of regulatory systems, assumes that organizational behavior can be viewed as legitimate if conforming to existing rules or laws (DiMaggio and Powell 1983) Public policy is a kind of coercive powe r that forces firms to increase the homogeneity in given domains to conform to government regulation s Usually, firms will be punished for not obeying the policy or rules.
73 Mimetic process is a process by which organizations imitate each other to acquire m ore stability when facing uncertainty (DiMaggio and Powell 1983) It is the mechanism of cultural cognitive systems. For example, a skilled labor force migration would transfer knowledge to another organization, which leads to a mimetic process As the ma jor mechanism of normative systems, n ormative pressures stem from professionalism (DiMaggio and Powell 1983) Behavior is guided by norms of what is appropriate and morally governed. Normative pressures can be developed during education For example, peopl e with the same education backgrounds usually have similar approach to problems, and that is normative pressure to homogeneity In the health care sector, Ruef and Scott (1998), using a sample of 143 hospitals in San Francisco over the period of 1945 and 1990, examined the extent to which factors of hospital legitimacy had affected hospital survival. Three levels of factors were included: technical legitimacy that focused on technology transforming inputs into outputs, managerial legitimacy such as person nel management and structure of the administrative staff, and an institutional level that related organizations to the environment with social legitimacy Based on Institutional Theory, the study predicted that both technical and managerial legitimacy woul d improve organizational survival chanc es, while the strength of the effects might be dependent on the institutional environment. The analyses also demonstrated that managerial legitimacy enhanced hospitals survival more in regimes that place more emphasi s on extensive formalized relations. And likewise, hospitals focusing more on technical legitimacy were more likely to survive in regimes with centralized regulation and controls on funding
74 Ginn, Shen, and Moseley (2009) used Institutional theory to iden tify the specific isomorphic changes associated with the enactment of state community benefit laws. They predicted that coercive power would force tax exempt nonprofit hospitals to increase provision of community orientation activities and health promotion services, while mimetic and normative pressures would force other hospitals to increase these services. However, using an AHA survey from 1996 to 2000, they found that nonprofit hospitals had increased community health orientation and promotion services i n a similar pattern to that of nonprofit hospitals in the states without such laws. Theory Integration and Research Hypotheses Institutional theory has been criticized for its focus only on wider institutional pressures without considering the role of or ganizational self interests and active agency in organizational responses to institutional pressures and expectations (Oliver, 1991). R esource Dependence Theory complements this by stating that organizations actively pursue and balance the relationship s w ith various stakeholders and other organizations in order to gain more power. And both theories acknowledge the multiple external pressures faced by organizations in decision making processes. O rganizational responses to the collective and interconnected e nvironment s determine their survival and stability. The integration of the two theories thus incorporates social legitimacy as well as interest driven organizational behavior (Oliver, 1991). Zinn, Weech, and Brannon (1998) examined the contextual factors that influenced nursing home adoption of Total Quality Management (TQM) from an integrativ e perspective. TQM referred to a program or information system affecting the quality and quantity of services for a continuous improvement in the decision making proc ess. It was an administrative innovation with both rational adaptive and institutional aspects
75 (p.264) The study applied the two complementary theories Resource Dependence Theory and Institutional Theory, and predicted that both organizational capabili ties and institutional environments would influence TQM adoption. The main source of data was a survey of nursing home administrators in Pennsylvania during 1994 1995. The study found that facilities with a greater proportion of Medicare recipients, locate d in more competitive markets and areas with a higher proportion of Medicare hospital discharges were more likely to adopt TQM. According to Resource Dependence Theory, n onprofit organizations are more dependent on government grants, and are more likely t o be influenced by government rules and policies on public funds and tax regulations (Heimovics, Herman, and Coughlin 1993). N onprofit hospitals increasing ly dependent upon their tax revenues, have to adapt to the government policy. Thus t ax exempt nonpr ofit hospitals would adopt an intentional strategy like providing more uncompensated care as required, as a means of enhanc ing their current competition level and stability As stated above c oercive i somorphism is as a result of regulation, rules, and mo nitoring processes. For the board s of nonprofit hospitals, failure to conform to regulatory institutional expectations may have serious effects on organizational legitimacy ( Miller Millesen 2003). That is, they could ultimately lose their tax exempt status as well as donations and grants. Pressure may come from legal and organizational sources According to normative pressures, organizations should act to enhance normative and moral legitimacy. Since serving the community is the mission of nonprofit hospita ls and community member s are involved in the activities of the board of
76 directors, nonprofit hospitals are expected to fill the community s unmet health care needs. Mimetic isomorphism is ano ther institutional isomorphism that relates to the study It sug gests that organizations imitat e each other due to uncertainty caused by the changing environment. Also based on Resource Dependence Theory firms that have same environment dependencies are more likely to adopt the same strategies ( Hillman, Withers, and C ollins 2009) When nonprofit hospitals change strategies such as to provide more uncompensated care in order to meet government requirement, th eir for profit counterparts locate d in the same area may also increase the provision of uncompensated care in ord er to attain a good neighbor image and to acquire more customer resources. Based on the two theories, the hypotheses for this study are: H ypothesis 1 Ceteris paribus, private hospitals in states with community benefit requirements will provide more uncom pensated care than comparable hospitals in states without such requirements H ypothesis 2. Ceteris paribus, nonprofit hospitals in states with community benefit requirements will provide more uncompensated care than comparable hospitals in states without s uch requirements Hypothesis 3. Ceteris paribus, the level of uncompensated care provided by private hospitals will be higher in states with stricter community benefit requirements than those with in states with less strict requirements Hypothesis 4. Cete ris paribus, the level of uncompensated care provided by nonprofit hospitals will be higher in states with stricter community benefit requirements than those with in states with less strict requirements
77 Table 3 1 Characteristics of f our m arket s tructur es Market Structure Type of Product Number of sellers Entry/Exist Restriction Perfect Competition Homogeneous product Lots of sellers Free/Unrestricted entry/exit Monopolistic Competition Differentiated product Lots of sellers Free/Unrestricted entry /exit Monopoly N/A One seller Completely restricted entry/exit Oligopoly Homogeneous or Differentiate product Few sellers Partially restricted entry/exit
78 Figure 3 1. The p rofit m aximization m odel Figure 3 2. The u tility m aximization m odel q uantity m aximization
79 Figure 3 3. The t rade off b etween q uality and q uantity
80 CHAPTER4 METHODOLOGY This chapter describes the methodology used to assess whether private hospitals in states with different levels of strictness of community benefit requirements provide different levels of uncompensated care. First, the chapter describes the data sources, then specifies the empirical models for the analyses. The next step i s to define variables and the study design. Finally, this chapter discusses econometric models, treatment of missing data and sensitivity analysis Data Sources The primary data source of the study is the Medicare Cost Report (MCR) from the Center s for Me dicare & Medicaid Services ( CMS ). CMS maintains the MCR in the Healthcare Cost Reporting Information System (HCRIS), which is on a federal fiscal year basis. For example, the 2006 hospital cost report contains data from October 1, 2005 to September 30, 200 6. HCRIS includes information on hospitals, skilled nursing facilities (SNF), home health agencies (HHA), end stage renal disease facilities (ESRD) and hospice. All Medicare certified hospitals in the U.S. are required to file an annual cost report with CM S. The MCR provide s information on hospital characteristics such as ownership, teaching status, and financial data The reports also provide information on hospital size and location. The data base includes annual cost reports from 1996 2010. Each year has information for over 6,000 hospitals across the U S Beginning in 2003, CMS added Worksheet S 10, which requires all inpatient prospective payment system (IPPS) hospitals to report their uncompensated care to comply with the Balanced Budget
81 Refinement Ac t (BBRA) of 1999 ( Schuhmann 2008 ). A sample of Worksheet S 10 is presented in the appendix (Appendix A). Data on market characteristics came from several national datasets. The county level unemployment rate came from the Local Area Unemployment Statistic s (LAUS) program which is provided by t he Bureau of Labor Statistics (BLS) of the U.S. Department of Labor The county level per capita income was accessed through the Bureau of Economic Analysis (BEA) of the U.S. Department of Commerce. The Census Bureau 's Small Area Health Insurance Estimates (SAHIE) program provided health insurance coverage information at the state and county level. SAHIE estimated, for the first time, the number and proportion of people without health insurance coverage for all ages a t the county level. The study data set covers a five year period from 2005 2009 to examine the influence of the policy. The four data sources were merged into the final dataset, using county and state where a hospital was located as the merging criteria. T he sample includes all general short term acute care hospitals in the U.S. with more than 50 beds filing a Medicare annual hospital report. The unit of analysis is the individual hospital Empirical Models Research Hypothes e s 1 and 2 Based on H ypothes e s 1 and 2 in Chapter 3, the first set of empirical model s are : Model 1: For all private hospitals, u ncompensated care (UC ) = f [ state with/without community benefit requirements ( CB ) hospital mission, hospital characteristics, and market and population charac teristics ]
82 Model 2: For nonprofit hospitals, u ncompensated care (UC ) = f [ state with/without community benefit requirements (CB) hospital mission (OWNERSHIP=NONPROFIT), hospital characteristics, and market and population characteristics] Research Hypoth es e s 3 and 4 Based on H ypothes e s 3 and 4 in Chapter 3, t he second set of empirical model s are : Model 3: For all private hospitals with CB=1, u ncompensated care (UC ) = f [ the strictness of state community benefit requirements (STRICT), hospital mission, hos pital characteristics, and market and population characteristics ] Model 4: For nonprofit hospitals with CB=1, Uncompensated care (UC ) = f [ the strictness of state community benefit requirements (STRICT), hospital mission (OWNERSHIP=NONPROFIT), hospital cha racteristics, and market and population characteristics ) Variable s and Study Design This section describes variables used in the analysis (see Table 4 1 for summary). Dependent Variable The dependent variable of the study is the level of uncompensated car e provided by a given hospital in a given year. The decision to use uncompensated care instead of charity care was based on two reasons First, although CMS defines charity care and bad debt separately it only collects data on uncompensated care. Second, one problem with the hospital file is that hospitals classify unpaid bills sometimes as charity care or sometimes as bad debt (B uczko 1994). T he criteria for classification may also va ry in specificity across states. For these reasons, the research literat ure typically measures uncompensated care as the sum of charity care and bad debt.
83 Previous studies have used different ways to measure uncompensated care. It has been measured as the proportion of net patient revenue devoted to uncompensated care costs ( Kennedy et al. 20 10 ; Sutton and Sten sland 2004; Weissman, Gaskin, and Reuter 2003; ) total uncompensated care, deflated by the hospital charge to cost ratio (Blewett et al. 2003; Davidoff et al. 2000; Lo Sasso and Seamster 2007 ; M orrisey, Wedig, and Hassan 1996 ), adjusted uncompensated care admissions (Gaskin 1997; Hsieh, Clement, and Bazzoli 2010; Rosko 2004; Thorpe and Brecher 1987 ), the percentage of a hospital s expenses (Clement, White, and Valdmanis 2002; McKay and Meng 2007 ; Th orpe, Seiber, and Flore nce 2001 ), or uncompensated care days per bed (Banks, Paterson, and Wendel 1997) CMS defines uncompensated care as the sum of charity care and bad debt, which excludes courtesy allowances and discounts to patients. In Worksheet S 10, line 31 measures tota l hospital uncompensated care costs. To avoid having to adjust for inflation, some studies calculated the level of uncompensated care as the ratio of uncompensated care costs to net patient revenues ( Sutton and Stensland 2004 ; We issman, Gaskin, and Reuter 2003 ). But for a longitudinal study, net patient revenue may be influenced by the overall economic environment and thus bias the proportional measure of uncompensated care. For example, the U.S. market experienced an economic downturn starting in 2007. Dur ing this period, hospitals in a market with a larger number of low income people may have had a lower level of net patient revenue, and had to provide more uncompensated care, which would exaggerate the overall effect if measuring the level of uncompensate d care as the proportion of net patient revenue. Consequently this study uses adjusted uncompensated care costs per bed
84 because hospital size is not a time varying variable. Uncompensated care costs were adjusted for inflation using the Consumer Price In dex (CPI) which can be accessed through the BLS Independent Variables For Models 1 and 2: The independent variable for Models 1 and 2 is CB, which is a dummy variable indicating whether or not the hospital is located in a state that has implemented comm unity benefit requirements. CB=1 indicates states with community benefit requirements and CB=0 indicates states without community benefit requirements. For Models 3 and 4 : The independent variable for Models 3 and 4 is STRICT, which is the level of strict ness of state community benefit requirements As explained in Chapter 2, s tates were classified into several categories according to the strictness of their community benefit requirements Thus, STRICT is a categorical variable: (a) STRICT=4 for states wit h the strictest level of community benefit requirements; (b) STRICT=3 for states with moderately strict community benefit requirements; (c) STRICT=2 for states with the least strict community benefit requirements; and ( d ) STRICT=1 for states with a minimal level of requirements for community benefit provisions Covariates Based on the conceptual framework variables were include d to control for hospital mission hospital characteristics, and market characteristics. Hospital m ission Variables for hospital mission include ownership and teaching status. Ownership is a dummy variable capturing private hospital ownership ( 0=Nonprofit hospitals, and 1=for profit hospitals ). Teaching is also a categorical variable measuring hospital teaching status, so Teach=1 de notes teaching hospital and
85 Teach=0 denotes non teaching hospitals. CMS defines t eaching hospital s as any institutions that receive Medicare direct or indirect graduate medical education (GME) payments Hospital c haracteristics. Variables included as contr ols for hospital characteristics are MSA, STATE, SIZE, MEDICAID, MEDICARE, HMO_MM, ED, PROFIT and DSH. MSA measures the location of hospitals (MSA=1 if hospitals are operated in Metropolitan Statistical Area s, =0 otherwise). STATE is included to account fo r the unmeasured factors at the state level. The total number of hospital beds is categorized i nto three levels to measure hospital size: fewer than 100 beds, 100 199 beds, and 200 or more beds ( to approximate for small, medium and large hospitals ) MEDICA ID/MEDICARE is calculated by the number of Medicare/Medicaid patient days as a proportion of total patient days. The two continuous variables measure the proportion of patient days that are from patients with Medicare or Medicaid as their primary insurance Since the HCRIS data set does not provide information on private managed care information, th is study uses the sum of Medicare and Medicaid HMO patient days divided by total hospital patient days as a proxy for overall managed care penetration. ED is em ergency room expenses as a proportion of total expenses for an individual hospital. Hospital profit margin (PROFIT) is measured by the difference between total hospital revenues and total hospital expenses as a proportion of total hospital revenues. The mo del also control s for DSH payment s since hospitals receiving DSH payments serve a disproportionate ly higher percentage of uninsured population (DSH=1 denotes hosp itals receiving DSH payments).
86 Market c h aracteristics Market and population characteristics include INCOME UE, HHI, UNINSURED, GOVFP, and GOVNFP INCOME is per capita income at the county level after adjustment for inflation using the Consumer Price Index (CPI); 2005 is the reference group. UE is the unemployment and UNINSURED is the uninsurance rate respectively, at the county level Finally, GOVFP and GOVNFP indicate the presence of a public hospital in the same market for for profit and nonprofit hospita ls, respectively. The study uses the Herfindahl Hirschman Index (HHI) to measure market co mpetition. HHI is commonly used to measure the degree of competition given a relevant market area and is calculated by summing the squared market shares of each hospital given the defined market area: HHI= where, in this study S i is defin ed as an individual hospital s nu mber of beds as a proportion of the total number of beds in the same MSA. A higher value of HHI indicates a higher market concentration and less market competition (HHI=0 would be a perfectly competitive market, while HH=1 indicates a monopoly). Study Design The study is a cross sectional study design using panel data. According to Baltagi (2001), using panel data controls for individual heterogeneity by providing more information and more variability. In addition, panel da ta is suited to study the dynamics of adjustment to economic changes (p.6). That is, panel data can identify and capture effects that cannot be explained by cross section al or simple time series data sets In this study, for example, cross sectional data can only provide a snapshot of hospitals provision of uncompensated care at a point in time without providing trend analysis Although repeated cross sectional data can show how this provision changes over time
87 o nly panel data can be used to examine wha t proportion of these hospitals that provide a certain level of uncompensated care in one period remain at that level in another period. Analytical Data Set In this study, the outcome variable was calculated by dividing inflation adjusted uncompensated ca re costs by the total number of beds for the individual hospital. Some missing values were due to lack of informatio n on uncompensated costs or total number of beds Due to lack of data, Medicaid and Medicare managed care penetration replaced p rivate manag ed care penetration. Not every hospital reported their Medicaid and Medicare managed care information, there were missing observations for the variable HMO_MM. These two variables were missed by around 20% of missing values, respectively. Data also were mi ssing for other continuous independent variables (around 3% of total observations). Listwise dele tion was used to eliminate all observations with any missing values (Howell 2008). In most studies, it is the standard and simplest approach to treat ing missi ng data although it is based on the assumption of Missing Completely at Random (MCAR), which assumes missing observations are unrelated to both observed and unobserved variables. Listwise deletion often results in a substantial reduction in the study powe r and sample size. If MCAR assumptions are not met, it can lead to biased parameter estimates. The sample as a whole has a total of 11,735 observations, and the analytical sample has 6,544 observations. Examination of Table 4 2 and 4 3 shows no systematic pattern of differences of variables between the whole sample and the analytical data set for the dependent variable, the independent variables of interest, and most control variables. H owever, hospitals in the analytical data set were somewhat
88 more likely to be larger, teaching hospitals, and in urban areas than hospitals in the whole sample. Methods and Sensitivity Analysis Econometric Models As mentioned previously, the basic unit of analysis is the individual hospital, with hospitals being nested withi n states with different levels of strictness of community benefit requirements. Thus, we need to build a multilevel model structure in which hospitals within states can be tracked over time. Nested data has two characteristics: h ospitals nested within a st ate tend to be more alike than hospitals selected at random and the dynamics of changes by states will tend to affect individual hospitals within the corresponding states as well. Thus, the study conduct s a multi level analysis to address the nested hospi tal effect s within states by using a Hierarchical Linear Model (HLM) to capture each level of analysis. HLM s (also called mixed effect models) have become widely used in analyzing longitudinal studies in which outcomes for the same subject are measured r epeatedly during the study period. An ordinary linear model assumes all observations are independent and identically distributed (iid). However, correlations may appear among some observations in two design features, repeated measures and cluster sampling (Cheng et al. 2008). In studies using longitudinal data, the assumption of independence between two observations from different subjects remain valid while two observations from the same subject may be correlated by sharing the same characteristics over ti me and thus become not independent from each other. A study using cluster sampling with subjects from many groups provides an example in which groups are define d as the independent sampling unit instead of the subject itself. This study faces more
89 challeng es in terms of multi level clustering in that observations from the same hospital are correlated while hospitals from the same state may also be correlated HLM or mixed effect models are good analytical tool s to address the above issues due to their advan tages in modeling missing data, repeated measures, and heterogeneity between clusters. Typically, HLM includes both fixed and random effects, which are important components for longitudinal data analysis estimat ing within subjects and be tween subjects var iance (Singer 1998). The fixed part of HLM is the same as in ordinary regression models. The coefficient provides estimates for the average effect for a certain variable, which represents the rate of change of the outcome variable as a function of changes in the independent variable. The random part of HLM explains the subject specif ic random effects (Cheng et al. 2008). In this study, it accounts for variance heterogeneity among responses from hospitals clustered in different states as well as the covarian ce pattern among responses within a hospital. An appropriate covariance model is important for valid fixed effects even though the covariance structure is not the major interest. HLM in this study can be viewed as a two level model. The level_1 model inclu des hospital level predictors, which is
90 w here i is at the hospital level, j is at the state level All continuous variables are rescaled by mean centering (i.e. use ) to make the intercept more meaningful and to avoid potential collinearity. The level 2 model is the state level model, whic h is n=1 2, 3, 20 We can combine the level 1 and level 2 models to yield:
92 The model has two random effects. One represents the variation among hospitals within states. Since hospitals are clustered by states, this study also specifies states as a nested random effect to account for the variation between states. All hospital level
93 predictors are considered fixed effects because they are not the major interest in this study We also treat time as a fixed effect by creating dummy variables from year 2006 to year 2009, with year 2005 as the reference group. In the selection of HL M, assumptions regulating normality and linearity should be checked with residuals. Results from visual diagnostics and the Hosmer Lemeshow Test showed that the model with the analytical data set met the linearity assumption. In order to meet the normality assumption the outcome variable was transformed by taking the natural log of the variable. Sensitivity Analysis Three types of sensitivity analyses are presented in this study. The first sensitivity analysis includes another three factors in the outcom e variable. Based on Medicare Hospital Cost Report Worksheet S 10, total state and local indigent care program cost (line 25), total SCHIP cost (line 27), and total gross Medicaid cost (line 29), together with uncompensated care cost s are combined to measu re the total uncompensated care cost to the hospital Costs related to Medicaid, SCHIP and the government indigent program are not included in the original model since they are not uncompensated care costs as defined by CMS. We include the three components in the outcome variable for the sensitivity analysis because states community benefit requirements also apply to these programs. The dependent variable for the first sensitivity analysis is TOTALUC_BED, adjusted for inflation using the 2005 Consumer Pric e Index (U.S. Department of Labor) The second sensitivity analysis is performed by reclassifying the variable STRICT. STRICT=3 and 4 are grouped together to represent states with stricter community benefit requirements while STRICT=1 and 2 are grouped to stand for states with less
94 strict community benefit requirements. The new categorical variable is used to assess the impact of different categorization alternatives of STRICT on the models. The final sensitivity analysis uses multiple imputation, a maxim um likelihood method, to estimate the missing values based on the existing data ( Rubin 1987; Schafer 1999). MI uses the Monte Carlo technique to replace the missing values by m>1 repeated imputation measures. We define that m equals five in this study, whi ch means five complete simulated datasets with each analyzed by standard methods. Then results are aggregated to produce parameter estimates and confidence intervals incorporat ing uncertainty from missing data. MI is based on the assumption of Missing at R andom (MAR), which is less restrictive than MCAR and assumes missing data only depends on observed data instead of unobserved data.
95 Table 4 1. List of v ariables Variable Description Type Category Data Source UC _BED Adjusted u ncompensated care costs per bed Continuous Calculated CB Hospital located in state with/without community benefit requirements Categorical 0= No 1=Yes Calculated( See Chap 2) STRICT Level of strictness of state community benefit requirements Categorical 1=Strictest 2=Moderately st rict 3=The least strict 4=A minimal level of requirements on community benefit provisions Calculated( See Chap 2) Ownership Hospital ownership Categorical 0=Nonprofit hospitals 1=For profit hospitals HCRIS TEACH Teaching intensity Categorical 0=Non teac hing hospitals 1= Teaching hospital s HCRIS MSA Location of h ospital Categorical 0=Non MSA 1=MSA HCRIS SIZE Number of licensed beds Categorical 1=beds<=100 2=100=200 HCRIS STATE States in which hospitals are located Categorical HCRIS
96 Table 4 1 Continued Variable Description Type Category Data Source DSH DSH payments from the government Categorical 0=Hospitals receiving DSH payments 1=Hospitals not receiving DSH payments HCRIS MEDICAID Medicaid patient days as percentage of the tota l patient days Continuous HCRIS MEDICARE Medicare patient days as percentage of the total patient days Continuous HCRIS HMO _MM Medicaid and Medicare Managed care patient days as percentage of the total patient days Continuous HCRIS PROFIT Total profi t margin=(Total Revenues Total Expenses)/Total Revenues Countinuous HCRIS ED Emergency room expenses as a percentage of total hospital expenses Continuous HCRIS I ncome Per capita income by county Continuous BEA*
97 Table 4 1 Continued Variable Descrip tion Type Category Data Source UE Unemployment rate by county Continuous LAUS U NINSURED Uninsurance rate by county Continuous SAHIE HHI Hospital concentration by MSA Continuous Calculated GOVFP P resence of government hospital in the same MSA with for profit hospital Continuous Calculated GOVNFP P resence of government hospital in the same MSA with nonprofit hospital Continuous Calculated *Adjusted for inflation using Consumer Price Index, 200 5 =100 (U.S. Department of Labor 201 2 )
98 Table 4 2 De scriptive s tatistics c omparing the w hole s ample and the analytical data set Categorical Variable Whole Sample n=11735 Analytical Data Set n=6544 CB With CB Without CB 5 6 0% 4 4 0% 54.6% 45.4% Strict Strictest Moderately strict Least Strict A minimal l evel 41.4 % 30. 5% 8.6% 19. 5% 43.0% 30.6% 7.3% 19. 1% NFP Nonprofit For profit 74.7% 25.3% 74.4% 25.6% TEACH Teaching Non teaching 37.6% 62.4% 45.7% 54.3% DSH With DSH No DSH 79.5% 20.5% 84.2% 15.8% Size Beds<=100 100=200 23.6% 36% 40.4% 16.7% 35.0% 48.3% MSA MSA hospitals Non MSA 76.0% 24.0% 83.9% 16.1% GOVFP GOVFP=1 GOVFP=0 13.7% 86.3% 14.8% 85.2% GOVNFP GOVNFP=1 GOVNFP=0 31.1% 68.9% 32.7% 67.3%
99 Table 4 3 Descriptive s tatistics c omparing the w ho le s ample and the analytical data set Continuous Variable Whole Sample n=11 735 Analytical Data Set n=6 544 UC_BED (n= 8 642 ) 30235.81 30 573 .49 MEDICAID (n= 11 565 ) (%) 13.4 12 6 MEDICARE (n= 11 733 ) (%) 43.9 42 .5 ED (n= 11 735 ) (%) 2.8 2 .7 PROFIT (n= 11 713 ) (%) .027 .03 4 HMO_MM (n= 9 034 ) (%) 11.1 1 1.0 INCOME (n= 11 340 ) $ 35 553.07 $ 35 483.93 UE (n= 11 331 ) (%) 6.0 6.1 UNINSURE (n= 11 352 ) (%) 16.7 17.0 HHI (n= 11 735 ) .363 .315 *Adjusted for inflation using Consumer Price Index, 200 5 =100 (U.S. Departm ent of Labor 201 2 )
100 CHAPTER 5 RESULTS This chapter presents the results of the study in three sections The first section reports and discusses the descriptive statistics for dependent and independent variables, hospital mission and characteristics, and ma rket characteristics. The second section presents the results of the four models and other interesting findings, while the last section presents the sensitivity analysis. Descriptive Statistics Dependent and Independent Variables The main objective of the study is to assess the association between community benefit requirements and private hospitals provision of uncompensated care. The variable CB determines whether hospitals are located in states with community benefit requirements, and the variable STRI CT represents the level of strictness of requirements on the level of uncompensated care. As shown in Table 5 1, over the five year period of analysis (2005 2009), there were a total of 6,544 general acute care hospitals in the study sample, which include d 1,259 hospitals i n 200 5 1,263 hospitals in 200 6, 1,268 hospitals in 2007, 1,298 hospitals in 2008, and 1,456 hospitals in 2009. Overall, 3,574 (54.6%) hospitals were located in states with community benefit requirements (CB=1). Of these, 683 hospitals we re in states that had a minimal level of requirements on community benefit provisions (19.1%); 261 hospitals were in states with the least strict community benefit requirements (7.3%); 1,094 hospitals were in states with moderately strict community benefit requirements (30.6%); and 1,537 hospitals were in states with the strictest level of community benefit requirements (43.0%).
101 The main dependent variable is UC_BED, which measures the inflation adjusted uncompensated care cost per bed. For the study sampl e (n=6 54 4 ), the average uncompensated care costs per bed was $ 30 ,577 across the five years ( Table 5 2), with t he mean uncompensated care costs per bed after adjusting for inflation increas ing over the study period, from $ 26 7 34 in 2005 to $ 33 7 62in 2009. Figure 5 1 shows that during the study period, among hospitals in states with community benefit requirements, the average uncompensated care cost per bed was $ 28 7 74, compared to $ 32 ,746 for those in states without any. For both groups, the level of unco mpensated care trended upward from 2005 to 2009, and the average amount of uncompensated care cost per bed was higher in each year in states without requirements. Figure 5 2 shows a nonlinear relationship between UC_BED and STRICT. Hospitals in states wit h a stricter level of community benefit requirements (STRICT=4 and 3) shared a lower level of uncompensated care than those with less strict level of community benefit requirements (STRICT 1 and 2). For states with the strictest and less strict level of co mmunity benefit requirements, the trend for the provision of uncompensated care was growing over 5 years. However, for the other two groups, the level of uncompensated care was relatively stable. The average amount of UC_BED for states with the strictest a nd moderately strict level of community benefit requirements w as $ 2 8,985 and $ 24 3 80, respectively for the five year s compared to $ 37 ,131 for states with the least strict community benefit requirements and $ 32 ,166 for states with a minimal level of requi rements on community benefit provisions.
102 Hospital Mission and Characteristics Of the hospitals in the study, 74.4% were nonprofit hospitals, and 45.7% were teaching hospitals (Table 5 1 and 5 2). T he proportion of teaching hospitals in the sample decrease d over the five years, from 48% in 2005 to 42% in 2009.The majority (84.2%) of the sample received DSH payments and most hospitals were located in a n MSA (83.9%). Of all hospitals, 16.7% had fewer than 100 beds; 35% had 100 to 199 beds; and 48.3% had over 200 beds. Medicaid and Medicare patient days accounted for 12.6% and 42.5% of the total patient days, respectively. The average hospital proportion of Medicaid patient days decreased from 13.4% to 12% in the first three years then increased to 12.7% in 20 09. The hospital proportion of Medicare patient days decreased from 45.1% to 40.3% during the five years The average proportion of patient days accounted for by Medicaid and Medicare managed care was 11%, growing from 8.3% to 13.8% over the period of anal ysis On average, the emergency room expense was 2.7% of total hospital expenses. Finally, the average hospital profit margin was 0 .034 over five years, and the trend for hospital profitability was stable except for 2006 when the hospital profit margin dec reased to the lowest point 0 .006 across five years. Market Characteristics Four market level variables were included in the study: INCOME, UE, UNINSURE D and HHI. After adjusting for inflation, the average per capita income at the county level was $ 3 6,237 which increased from $ 35 478 in 2005 to $ 37 ,004 in 2008, and then decreased to $ 35 59 0 in 2009. The mean unemployment rate by county was 6.1%, ranging from a low of 4. 8% in 2006 and 2007 to a high of 9. 4% in 2009. On average, the uninsured population acc ounted for approximately 17% of the total
103 population in the county. For for profit hospitals, 14.8% were in a market with a government hospital, while for nonprofit hospitals. 32.7% were in a market with a government hospital. Hospital concentration was st able, at about 0 .3 across the five years. According to t he Antitrust Division of the Department of Justice (2012), a market in which the HHI is above 0 .25 is considered to be concentrated. The hospital market is therefore relatively concentrated, and as ma rket concentration increases, the competition between hospitals decreases and the market power of hospitals increases. Multivariate Analysis This section describes the results of the four models as discussed in C hapter 4. SAS PROC MIXED was used to analyz e the HLM regressions for the four models. The use of HLM represents a substantial difference from the traditional one level model. Within SAS PROC MIXED, the MODEL statement indicates the fixed effects, similar to the ge neral linear regression (Singer 199 8). The RANDOM option i ndicates random effects, which are the lowest level (within states) residual and the second level (between states) residual The REPEATED command specifies different covariance structures of repeated measures. Thi s study assumes an unstructured matrix for the form of the covariance matrix after comparing with other covariance matrix structures ( command symmetry and autoregressive structure). A test for correlation measured the extent to which the variability of on e variable is shared by others Results showed that there were no perfect correlations in any of two variables in this study.
104 Model 1 Table 5 3 summarizes the HLM results for the first model. For all private hospitals (n=6 544), being in a state with com munity benefit requirements was not significantly associated with providing a higher level of uncompensated care ( p >.05) after controlling for hospital mission, hospital characteristics and market characteristics. Compared to for profit hospitals, nonprof it hospitals spent 39.1% more on uncompensated care per bed ( 0 .391= p <.0001). Teaching hospitals provided 12.8% more uncompensated care than non teaching hospitals ( p = 0 .01). In general, hospitals in an MSA ( p = 0 .03) with a higher proportion of Medicaid patients ( p = 0 .01) and higher emergency room expenditures ( p < 0 .0001) had a higher level of uncompensated care. Having more Medicaid and Medicare managed care patients was positive ly and statistically significant ly associated with a higher provision of uncompensated care ( p = 0 .05), while a higher proportion of Medicare patien ts was negatively associated with the level of uncompensated care provided ( p < 0 .01). For market characteristics, the coefficient for per capita income was significant ( p < 0 .01) although the magnitude was effectively zero. Hospitals in count ies with a highe r unemployment rate ( p = 0 .01) and higher market concentration level ( p < 0 .0001) provided more uncompensated care. Model 2 The second model only included nonprofit hospitals (n=4 871) since community benefit requirements directly affected on ly nonprofit owne rship. As reported in Table 5 4, nonprofit hospitals in states with community benefit requirements did not differ in the provision of uncompensated care compared to those in states without requirements ( p = 0 .69 )
105 Significant results were found for several covariates in this model Nonprofit teaching hospitals provided 11.6% more uncompensated care than nonprofit non teaching hospitals ( p = 0 .04). Similar ly to the first model, nonprofit hospitals with more Medicaid patients ( p < 0 .01), higher Medicaid and Medicar e managed care penetration ( p = 0 .01) and more emergency room expenditures ( p = 0 .05) were associated with a higher level of uncompensated care, while the proportion of Medicare patients was negatively associated with the provision of uncompensated care ( p < 0 .0 1). Nonprofit hospitals in counties with higher unemployment rate provided significantly more uncompensated care ( p = 0 .01). A higher level of market competition was associated with an increased provision of uncompensated care ( p < 0 .01). Model 3 Model 3 ( n=3 575) focused on the relationship between the strictness of state requirements on community benefit provisions and the level of uncompensated care, and thus excluded hospitals in states without any requirements on community benefit provisions. For Model 3 as shown in Table 5 5, the level of strictness was not significantly associated with providing more uncompensated care ( p = 0 .99). Nonprofit hospitals in states with requirements provided 55% more uncompensated care cost per bed than for profit hospital s ( p < 0 .0001). Teaching hospitals in these states spent 13.9 % more on uncompensated care per bed than nonteaching hospitals. Similar ly to previous models, the coefficient s for Medicaid patient days ( p = 0 .01) and Medicaid and Medicare managed care penetration ( p = 0 .04) were positive and statistically significant. A h igher unemployment rate ( p = 0 .04) and higher market concentration ( p = 0 .04) w ere positive ly associated with the provision of uncompensated care
106 Model 4 Model 4 was a subset of Model 3 s sample inclu ding only nonprofit hospitals (n=2 680). Results reported in Table 5 6 showed that, holding other factors constant, states strictness of requirements on community benefit provisions had no significant effect on nonprofit hospitals level of uncompensated care ( p = 0 .99). Nonprofit hospitals with a higher proportion of Medicaid patient days ( p = 0 .02) and more Medicaid and Medicare managed care patient days ( p = 0 .02) significantly pr ovided more uncompensated care; the coefficient of Medicare patient days w as neg ative and statistically significant ( p = 0 .04). Other Findings This study found that differences in uncompensated care shares between nonprofit and for profit hospitals in states with community benefit requirements were much larger than the differences betw een two ownership types in states without such requirements. Controlling other covariates, the coefficient for OWNERSHIP was 0 .44 in states with policies on community benefit provisions ( p < 0 .001), which indicated nonprofit hospitals in these states provid ed 55.5% more uncompensated care than for profit hospitals. However the coefficient for OWNERSHIP was 0 .28 for states without requirements ( p < 0 .001), indicating that nonprofit hospitals in these state provided 32.3% more uncompensated care than did for pr ofit hospitals (Table 5 7). Sensitivity Analysis Three different sensitivity analyses were performed as discussed in Chapter 4. The first sensitivity analysis added another three components to the outcome variable to reveal any change in the final results The second sensitivity analysis re classified the
107 outcome variable STRICT to analyze the influence of different classifications on the outcome variable. The last sensitivity analysis presents the results for imputing missing values Sensitivity a nalysis 1 The first sensitivity analysis changed the dependent variable from UC_BED to TOTALUC_BED (Tables5 8 through 5 11) For Model s 1, 3 and 4, adding total state and local indigent care program cost, total SCHIP cost, and total gross Medicaid cost did not ch ange the significance of the final results for the independent variables of primary interest (i.e., CB in Model 1 and STRICT in Models 3 and 4) However, the coefficient of CB became positive and significant for Model 2, suggesting that state community ben efit requirements w ere positively associated with nonprofit hospitals provision of uncompensated care and other indigent program services. In addition, positive and significant associations were found between hospital DSH payment s and profitability and to tal amount of uncompensated care, while a negative association was found as hospital size increased The relationship between t he strictness of state community benefit requirements and total cost of uncompensated care per bed changed from positive to negat ive in the third model, but was statistically significant in neither (Table 5 10). The direction and significance of the main effect did not change in Model 4 although it was larger in magnitude again neither coefficient was statistically significant (Ta ble 5 11). Changes also occurred in other covariates. For Model 3, hospital profitability had a positive and significant relationship with the total amount of uncompensated care provided For both Models 3 and 4 hospital DSH payment was positively and si gnificantly associated with the total level of uncompensated care provided while
108 hospital size and the proportion of Medicare patient days were negatively associated with the outcome variable. In Model 4, t he coefficient of GOVNFP became negative and sign ificant, suggesting that nonprofit hospitals decrease d their provision of total uncompensated care when there were government hospita ls in the same market. As shown in Table 5 12, after including public program patients in the model, the coefficient for OW NERSHIP was 0 .30 in states with policies on community benefit provisions (p<.001), which indicated that nonprofit hospitals in thes e states provided 35.0% more public program services and uncompensated care than for profit hospitals. Holding other factor s constant, t he coefficient for OWNERSHIP was 0 .15 for states without requirements ( p < 0 .001), on the other hand, indicating nonprofit hospitals in these state provided 16.2% more public program services and uncompensated care than did for profit hospitals Sensitivity a nalysis 2. In the second sensitivity analysis, the independent variable STRICT was compressed from four to two levels: a. STRICT=3 and 4 was combined to stand for states with a stricter level of community benefit requirements and b. STRICT=1 and 2 was combined to stand for states with less strict community benefit requirements. In general, the new categorization did not change the direction or significance of the relationship of interest for Model 3 (Table 5 13) or Model 4 ( Table 5 14). For th e covariates, b esides the previous ly significant coefficients, teaching status and hospital DSH payment were positively related with the level of uncompensated care after compression while such relationships changed to negative for SIZE and GOVNFP. Sensi tivity a nalysis 3. SAS PROC MI was used for multiple imputation, which study sample and the imputed sample. For Model s 1
109 and 2, similar results were found for the main independent variable CB, comparing the imputed sa mple to the study sample That is, r e sults were negative and insignificant, suggesting that being in a state with community benefit requirements w as not significantly associated with the provision of uncompensated care in both the study sample and the imputed sample Similarly, the coefficie nt of STRICT was not statistically significant for the third and fourth models in the imputed sample, although it changed from positive to negative in Model 3 and the magnitude increased in Model 4. Results from both models indicated that the level of stri ctness was not associated with the level of uncompensated care.
110 Table 5 1 Descriptive statistics for study sample, categorical variables (percentage, %) Variable All n=6 544 2005 n=1 259 2006 n=1 263 2007 n=1 268 2008 n=1 298 2009 n=1 456 CB With CB Wi thout CB 54.6 45.4 5 6.6 43.4 5 5.7 4 4.3 54. 4 45. 6 5 3.8 46.2 5 2.9 4 7.1 Strict Strictest Moderately strict Least s trict M inimal level 43.0 30.6 7.3 19. 1 42.1 32.9 7.2 17.8 41.8 3 2.1 6.7 19. 4 43.3 30. 5 7.1 19. 1 43.7 29.5 7.3 19. 5 44.0 28.4 7 .9 19. 7 NFP Nonprofit For profit 74.4 25.6 74. 4 25. 6 74.0 26.0 74. 1 25.9 7 5.5 2 4.5 74. 1 25. 9 TEACH Teaching Non teaching 45.7 54.3 48.0 52.0 47.6 52.4 46.4 53.6 44.8 55.2 42.2 57.8 DSH With DSH No DSH 84.2 15.8 83.8 16.2 84.2 15.8 84.2 15.8 83.7 16.3 84.8 15.2 Size (beds) <=100 100 199 200 + 16.7 35.0 48.3 15.6 36.5 47.9 14.7 36.4 48.9 16.1 3 4.5 49.4 17.7 34.1 4 8.2 19.0 34.0 47.0 MSA MSA hospitals Non MSA 83.9 16.1 84.9 15.1 85.7 14.3 85.0 15.0 82.7 17.3 81.8 18.2 GOVFP GOVFP=1 GOVFP=0 14.8 85.2 14.7 85.3 14.3 85.7 15.2 84.8 14.4 8 5.6 15.5 84.5 GOVNFP GOVNFP=1 GOVNFP=0 32.7 67.3 32.1 67.9 3 2.9 67.1 33.7 6 6.3 31.6 68.4 33.0 67.0
111 Table 5 2 Descriptive statistics for the study sample, continu ous variables (mean(sd)) Variable All n=6544 2005 n=1259 2006 n=1263 2007 n=1268 2008 n=1298 2009 n=1456 UC_BED 30 57 6 85 ( 34 00 4 37 ) 26 733.73 ( 29 155.56 ) 29 294.25 ( 37 132.32 ) 30 879.16 ( 36 509.62 ) 31 684.36 ( 35 355.05 ) 33 7 61.98 ( 31 119.01 ) MEDICAID (%) 12 6 ( 9 7 ) 13 4 ( 10 2 ) 12 .9 ( 9 6 ) 12 0 ( 9 .4) 12 .1 ( 9 4 ) 12 .7 (9 .9) MEDICARE (%) 42 .5 ( 12 .6) 45 1 ( 13 1 ) 43 8 ( 12 7 ) 42 .5 ( 12 3 ) 41 .4 ( 12 2 ) 40 3 (1 1 9 ) ED (%) 2 .7 ( 1 1 ) 2 .7 ( 1 1 ) 2 .7 ( 1 .1) 2 7 ( 1 1 ) 2 7 ( 1 .1) 2 7 ( 1 1 ) PROFIT 0.03 4 ( 0.12 4) 0.03 4 ( 0.14 4) 0.0 40 ( 0.12 8) 0.03 2 ( 0.11 8) 0.00 6 ( 0.11 8) 0.055 ( 0.10 4) HMO_MM (%) 1 1.0 ( 9 .7) 8 3 ( 8 .2) 9 .4 ( 8 .7) 1 1.0 ( 9 .5) 11 .9 ( 10 .3) 13 .8 ( 10 5 ) INCOME 36 237.09 ( 10 265.02 ) 35 478.00 ( 9 919.74 ) 36 589.57 ( 10 815.00 ) 37 003.92 ( 10 880.49 ) 36 601.15 ( 10 335.34 ) 35 590.21 ( 9 351.12 ) UE (%) 6.1 ( 2. 5) 5.2 ( 1.3 ) 4. 8 ( 1. 3) 4.8 ( 1. 3) 6.0 ( 1.5 ) 9.4 ( 2. 5) UNINS URE D (%) 17.0 ( 5. 9) 16. 8 ( 5.7 ) 17. 7 ( 6. 1) 17.0 ( 5. 9) 16. 4 ( 5. 9) 17.3 ( 5.8 ) HHI 0.315 ( 0.32 9) 0.301 ( 0.320 ) 0.297 ( 0.318 ) 0.310 ( 0.32 6) 0.32 7 ( 0.335 ) 0.337 ( 0.340 ) UC_BED and INCOME were adjusted for inflation using Consumer Price Index, 200 5 =100 (U.S. Department of Labor)
112 Table 5 3 The HLM results for Model 1. D ependent variable : l og ( UC_BED). n=6 544 Variable Coef. S.E. p value CB 0.08 0.15 0.60 OWNERSHIP 0.33 *** 0.07 < 0 .0001 TEACH 0.12 ** 0.05 0.01 SIZE 0.00 0.03 0.87 STATE 0.00 0.01 0.60 MCAI D 0.55 ** 0.23 0.01 MCARE 0.74 ** 0.20 0.00 ED 6.81 *** 1.65 < 0 .0001 DSH 0.08 0.05 0.07 MSA 0.18 0.08 0.03 PROFIT 0.03 0.09 0. 70 GOVNFP 0.07 0.05 0.17 GOVFP 0.12 0.07 0.10 HMO 0.36 0.19 0.05 INCOME 0.00 ** 0.00 0.00 UE 2.5 7** 0.92 0.01 UNINSURED 0.34 0.44 0.44 HHI 0.45 *** 0.10 < 0 .0001 Year2006 0.04 0.02 0.13 Year2007 0.14 *** 0.02 < 0 .0001 Year2008 0.18 *** 0.03 < 0 .0001 Year2009 0.28 *** 0.03 < 0 .0001 p<0.05 ** p 0.01 *** p 0.001
113 Table 5 4 The HLM r esults f or Model 2. D ependent v ariable: l og ( UC_BED). n=4 871. Variable Coef. S.E. p value CB 0.06 0.15 0.69 TEACH 0.11 0.05 0.04 SIZE 0.02 0.04 0.50 STATE 0.00 0.01 0.59 MCAID 0.95 ** 0.29 0.00 MCARE 0.72 ** 0.24 0.00 ED 4.00 2.02 0.05 DSH 0.05 0.05 0.36 MSA 0.16 0.09 0.08 PROFIT 0.03 0 .10 0.77 GOVNFP 0.03 0.05 0.62 HMO 0.59 ** 0.23 0.01 INCOME 0.00 ** 0.00 0.00 UE 2.84 ** 1.11 0.01 UNINSURED 0.88 0.53 0.09 HHI 0.39 ** 0.11 0.00 Year2006 0.07 0.03 0.02 Year2007 0.14 *** 0.03 <.0001 Year2008 0.18 *** 0.03 <.0001 Year2009 0.28 *** 0.03 <.0001 p< 0 .05 **p 0 .01 ***p 0 .001
114 Table 5 5 The HLM results for Model 3. D ependent variable : l og ( UC_BED). n=3 574. Variable Coef. S.E. p value STRICT 0.03 0.11 0.80 OWNERSHIP 0.44 *** 0.11 < 0 .0001 TEACH 0.1 3* 0.07 0.0 5 SIZE 0.04 0.04 0.3 6 STATE 0.00 0.01 0.84 MCAID 0.71 0.29 0.0 2 MCARE 0.47 0.27 0.09 ED 4. 12 2.29 0.0 7 DSH 0.02 0.07 0.77 MSA 0.1 1 0.13 0.3 7 PROFIT 0.00 0.13 0.9 9 GOVNFP 0.01 0.07 0.85 GOVFP 0.13 0.13 0.31 HMO 0.51 0.25 0.04 INCOME 0.00 ** 0.00 0.00 UE 2.9 3* 1.43 0.04 UNINSURED 0.04 0.59 0.95 HHI 0.32 0.16 0.04 Year2006 0.00 0.04 0.9 8 Year2007 0.06 0.03 0.0 6 Year2008 0.10 ** 0.03 0.00 Year2009 0.21 *** 0.03 < 0 .0001 p< 0 .05 **p 0 .01 ***p 0 .001
115 Table 5 6 The HLM results for Model 4. D ependent variable : l og ( UC_BED). n=2 680. Variable Coef. S.E. p value STRICT 0.00 ( 0.0009 ) 0.12 0.99 TEACH 0.13 0.08 0.08 SIZE 0.05 0.05 0.28 STATE 0.00 0.01 0.85 MCAID 0.86 0.37 0.02 MCAR E 0.69 0.34 0.04 ED 2.30 2.81 0.41 DSH 0.02 0.08 0.82 MSA 0.09 0.15 0.55 PROFIT 0.01 0.17 0.94 GOVNFP 0.03 0.08 0.66 HMO 0.68 0.30 0.02 INCOME 0.00 ** 0.00 0.01 UE 2.86 1.74 0.10 UNINSURED 0.26 0.71 0.72 HHI 0.33 0.18 0.06 Year2006 0.02 0.04 0.70 Year2007 0.05 0.04 0.18 Year2008 0.08 0.04 0.04 Year2009 0.19 *** 0.04 < 0 .0001 p<0.05 ** p 0.01 *** p 0.001
116 Table 5 7 Other findings comparing results for states with and without community benefit requirements. Dependent variable: UC_BED Variable Hospitals in states CB=1 N =3,574 Hospitals in states CB=0 N = 2 970 Coef. Coef. OWNERSHIP 0.44* ** 0.28*** TEACH 0.12 0.15** SIZE 0.04 0.01 STATE 0.00 0.01 MCAID 0.71** 0.39 MCARE 0.47 1.14*** ED 4.13 10.39*** DSH 0.02 0.14* MSA 0.11 0.21* PROFIT 0.00 0.07 GOVNFP 0.01 0.15* GOVFP 0.13 0.05 HMO 0.51* 0.17 INCOME 0.00** 0.00 UE 2.93 1.79 UNINSURED 0.02 0.35 HHI 0.31* 0.56*** Year2006 0.01 0.07** Year2007 0.04 0.21*** Year2008 0.03 0.22*** Year2009 0.04 0.22** p<0.05 ** p 0.01 *** p 0.001
117 Table 5 8 Sensitivity analysis: Model 1 results with different dependent variables Variable DV=log(UC_BED) DV=log(TOTALUC_BED) Coef. Coef. CB 0.08 0.08 OWNERSHIP 0.33 *** 0.23 *** TEACH 0.12 ** 0.12 *** SIZE 0.00 0.04 STATE 0. 00 0.00 MCAID 0.55 ** 1.97 *** MCARE 0.74 ** 0.71 *** ED 6.81 *** 1.67 DSH 0.08 0.18 *** MSA 0.18 0.12 ** PROFIT 0.03 0.12 ** GOVNFP 0.07 0.01 GOVFP 0.12 0.04 HMO 0.36 0.74 *** INCOME 0.00 ** 0.00 UE 2.5 7** 1.95 *** UNINSURED 0.34 0.11 HHI 0.45 ** 0.29 *** Year2006 0.04 0.02 Year2007 0.14 *** 0.10 *** Year2008 0.18 *** 0.11 *** Year2009 0.28 *** 0.13 *** p<0.05 ** p 0.01 *** p 0.001 TOTALUC_BED = ( u ncompensated cost+ t otal state and local indigent care programs cost + t otal SCHIP cost + t otal gross Medicaid cost)/BEDS, adjusted for inflation using Consumer Price Index, 200 5 =100 (U.S. Department of Labor)
118 Table 5 9 Sensitivity analysis: Model 2 results with different dependent variables Variable DV=log(UC_BED) Coef. DV=log(TOTALUC_BED) Coef. CB 0.06 0.13* TEACH 0.11 0.13*** SIZE 0.02 0.04 STATE 0.00 0.00 MCAID 0.95 ** 1.92*** MCARE 0.72 ** 0.81*** E D 4.00 0.70 DSH 0.05 0.15*** MSA 0.16 0.07 PROFIT 0.03 0.15** GOVNFP 0.03 0.01 HMO 0.59 ** 0.64*** INCOME 0.00 ** 0.00 UE 2.84 ** 1.95*** UNINSURED 0.88 0.00 HHI 0.39 ** 0.23*** Year2006 0.07 0.03 Year2007 0.14 *** 0.12*** Year2008 0.18 *** 0.14** Year2009 0.28 *** 0.14*** p<0.05 ** p 0.01 *** p 0.001 TOTALUC_BED = ( u ncompensated cost+ t otal state and local indigent care programs cost + t otal SCHIP cost + t otal gross Medicaid cost)/BEDS, adjusted for inflation using Consumer Price Index, 200 5 =100 (U.S. Department of Labor)
119 Table 5 10 Sensitivity analysis: Model 3 results with different dependent variables Variable DV=log(UC_BED) Coef. DV=log(TOTALUC_BED) Coef. STRICT 0.03 0.00( .003) OWNERSHIP 0.44 *** 0.29*** TEACH 0.1 3* 0.15*** SIZE 0.04 0.05* STATE 0.00 0.00 MCAID 0.7 1 1.98*** MCARE 0.47 0.38** ED 4. 12 0.18 DSH 0.02 0.15*** MSA 0.1 1 0.07 PROFIT 0.00 0.14* GOVNFP 0.01 0.05 GOVFP 0.13 0.07 HMO 0.51 0.82*** INCOME 0.00 ** 0.00 UE 2.9 3* 2.37*** UNINSURED 0.04 0.14 HHI 0.32 0.14 Year2006 0.00 0.01 Y ear2007 0.06 0.12*** Year2008 0.10 ** 0.12*** Year2009 0.21 *** 0.13*** p<0.05 ** p 0.01 *** p 0.001 TOTALUC_BED = ( u ncompensated cost+ t otal state and local indigent care programs cost + t otal SCHIP cost + t otal gross Medicaid cost)/BEDS, adjusted for inflation using Consumer Price Index, 200 5 =100 (U.S. Department of Labor)
120 Table 5 11 Sensitivity analysis: Model 4 results with different dependent variables Variable DV=log(UC_BED) Coef. DV=log(TOTALUC_BED) Coef. STRICT 0.00 ( 0.0009 ) 0.03 TEACH 0.13 0.19*** SIZE 0.05 0.05* STATE 0.00 0.00 MCAID 0.86 1.68*** MCARE 0.69 0.63*** ED 2.30 0.46 DSH 0.02 0.12** MSA 0.09 0.01 PROFIT 0.01 0.16* GOVNFP 0.03 0.08* HMO 0.68 0.64*** INCOME 0.00 ** 0.00 UE 2.86 1.80* UNINSURED 0.26 0.05 HHI 0.33 0.08 Year2006 0.02 0.00 Year2007 0.05 0.13*** Year2008 0.08 0.14*** Year2009 0.19 *** 0.15*** p<0.05 ** p 0.01 *** p 0.001 TOTALUC_BED = ( u ncompensated cost+ t otal state and local indigent care programs cost + t otal SCHIP cost + t otal gross Medicaid cost)/BEDS, adjusted for inflation using Consumer Price Index, 200 5 =100 (U.S. Department of Labor)
121 Table 5 1 2 Sensitivity analysis : comparing results for states with and without community benefit requirements. Dependent variable: TOTALUC_BED Variable Hospitals in states CB=1 N =3,574 Coef. Hospitals in states CB=0 N = 2 970 Coef. OWNERSHIP 0.30*** 0.15*** TEA CH 0.15*** 0.10** SIZE 0.05* 0.03 STATE 0.00 0.00* MCAID 1.98*** 1.94*** MCARE 0.38*** 1.11*** ED 0.17 4.36*** DSH 0.15*** 0.17*** MSA 0.07 0.15* PROFIT 0.14* 0.06 GOVNFP 0.05 0.06 GOVFP 0.07 0.03 HMO 0.82*** 0.66*** INCOME 0.00 0.00 U E 2.37*** 1.77** UNINSURED 0.13 0.06 HHI 0.14 0.40*** Year2006 0.01 0.04** Year2007 0.12*** 0.10*** Year2008 0.12*** 0.10*** Year2009 0.13*** 0.11** p<0.05 ** p 0.01 *** p 0.001 TOTALUC_BED = ( u ncompensated cost+ t otal state and local indigent care programs cost + t otal SCHIP cost + t otal gross Medicaid cost)/BEDS
122 Table 5 13 Sensitivity analysis: Model 3 results with different independent variables Variable I V=STRICT (Four Levels) Coef. IV=STRICT(Two Levels) Coef. STRICT 0.03 0.04 OWNERSHIP 0.44 *** 0.29 *** TEACH 0.1 3* 0.15 *** SIZE 0.04 0.05 STATE 0.00 0.00 MCAID 0.71 1.98 *** MCARE 0.47 0.34 ** ED 4. 12 0.52 DSH 0.02 0.13 *** MSA 0.1 1 0.08 PROFI T 0.00 0.13 GOVNFP 0.01 0.08 GOVFP 0.13 0.09 HMO 0.51 0.91 *** INCOME 0.00 ** 0.00 UE 2.9 3* 1.97 ** UNINSURED 0.04 0.06 HHI 0.32 0.12 Year2006 0.00 0.00 Year2007 0.06 0.11 *** Year2008 0.10 ** 0.15 *** Year2009 0.21 *** 0.25 *** p<0.05 ** p 0.01 *** p 0.001
123 Table 5 14. Sensitivity analysis: Model 4 results with different independent variables Variable IV=STRICT (Four Levels) Coef. IV=STRICT(Two Levels) Coef. STRICT 0.00 ( 0.0009 ) 0.01 TEACH 0.13 0.19 *** SIZE 0.05 0.05 STATE 0.00 0. 00 MCAID 0.86 1.72 *** MCARE 0.69 0.65 *** ED 2.30 0.16 DSH 0.02 0.08 MSA 0.09 0.00 PROFIT 0.01 0.12 GOVNFP 0.03 0.11 ** HMO 0.68 0.72 *** INCOME 0.00 ** 0.00 UE 2.86 1.28 UNINSURED 0.26 0.05 HHI 0.33 0.07 Year2006 0.02 0.00 Year2007 0.05 0.13 *** Year2008 0.08 0.17 *** Year2009 0.19 *** 0.26 *** p<0.05 ** p 0.01 *** p 0.001
124 Table 5 15 Sensitivity analysis: comparing the study sample with the imputed sample Variable The Study Sample The Imputed Sample Coef. p value Coef. p value CB 0.08 0.60 0.0 8 0. 50 CB 0.06 0.69 0.0 8 0. 49 STRICT 0.03 0.80 0.0 4 0. 59 STRICT 0.0009 0.99 0.0 5 0. 49
125 Figure 5 1. Uncompensated care cost per bed, CB=1/0, 2005 2009 Figure 5 2. Uncompensated care cost per bed, STRICT=1 4, 2005 200 9 26953 31092 34008 34166 36407 32746 26566 27866 28258 29551 31406 28774 0 5000 10000 15000 20000 25000 30000 35000 40000 2005 2006 2007 2008 2009 Avg. Uncompensated cost per bed Year No CBL With CBL 25952 26697 28528 30257 32912 28985 22862 25331 23596 24672 25501 24380 35009 35565 37374 38709 38597 37131 31447 31917 31680 31925 33629 32166 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 2005 2006 2007 2008 2009 Avg. Uncompensated cost per bed Year The Strictest Moderately Strict Less Strict A minimal level
126 Figure 5 3 Uncompensated care cost per bed trend CB=1/0, 2005 2009 Figure 5 4 Uncompensated care cost per bed trend STRICT=1 4, 2005 200 9 26953 31092 34008 34166 36407 26566 27866 28258 29551 31406 0 5000 10000 15000 20000 25000 30000 35000 40000 2005 2006 2007 2008 2009 Uncompensated cost per bed Year No CBL With CBL 25952 26697 28528 30257 32912 22862 25331 23596 24672 25501 35009 35565 37374 38709 38597 31447 31917 31680 31925 33629 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 2005 2006 2007 2008 2009 Uncompensated cost per bed Year The Strictest Moderate ly Strict Less Strict
127 CHAPTER 6 CONCLUSION S AND DISCUSSION This chapter begins with a summary of the study results sensitivity analys e s and other findings. Then it draws conclusion s regarding states requirements on community benefit provisions and discusses policy implication s The last part of th e chapter describes the study s limitations and possibilities for future research. Summary of Major Study Findings To regulate the behavior of nonprofit hospitals with re gard to the provision of community benefit states have implemented different standards of requirements. One of the major indicators of community benefit is the level of uncompensated care provided. The purpose of this study is to examine the association b etween the state health policy on community benefit and nonprofit hospitals provision of uncompensated care. D escriptive statistics indicate that without controlling other factors, hospitals in states with community benefit requirements provided less un compensated care than those without requirements across the five year study period States with the strictest and moderately strict requirements also provided less uncompensated care than states with less strict and a minimal level of requirements for comm unity benefit provisions. Hypothesis 1: Ceteris paribus, private hospitals in states with community benefit requirements will provide more uncompensated care than comparable hospitals in states without such requirements This hypothesis was not supported by the study results. Overall, holding other factors constant hospitals in states with community benefit requirements were not significant ly different in their level of uncompensated care
128 provided from comparable hospitals in states without such requireme nts during the study period. H ypothesis 2. Ceteris paribus, nonprofit hospitals in states with community benefit requirements will provide more uncompensated care than comparable hospitals in states without such requirements This hypothesis was not suppo rted by the study results. Nonprofit hospitals in states with community benefit requirements did not provide a significant ly higher level of uncompensated care than comparable hospitals in states without such requirements. Hypothesis 3. Ceteris paribus, th e level of uncompensated care provided by private hospitals will be higher in states with stricter community benefit requirements than those with in states with less strict requirements After restricting the study population to private hospitals in states with community benefit requirements, private hospitals in states with more strict community benefit requirements did not provide a significant ly higher level of uncompensated care than comparable hospitals in states with less strict requirements. Hypoth esis 4. Ceteris paribus, the level of uncompensated care provided by nonprofit hospitals will be higher in states with stricter community benefit requirements than those with less strict requirements T his hypothesis also was supported by the results. Nonp rofit hospitals in states with stricter community benefit requirements did not vary significantly in the provision of uncompensated care as compared to their counterparts in states with less strict community benefit requirements. Other f indings. For H ypot hesis 2, after considering costs of both public program s and uncompensated care patients, nonprofit hospitals in states with community benefit
129 requirements provided 35% more public program services and uncompensated care than for profit hospitals. Another finding was that t he additional amount of uncompensated care provided by nonprofit hospitals compared to for profit hospitals was 56% and 32% for states with and without community benefit requirements, respectively. Taking into account both public programs (state and local indigent programs, SCHIP and Medicaid) and uncompensated care patients, the difference between the additional amount of uncompensated care provided by nonprofit hospitals compared to for profit hospitals w as more than twice as great in st ates with community benefit requirements than in the states without any requirements (35% vs. 16%) respectively Conclusions In conclusion, the study found no evidence of a systematic relationship between states community benefit requirements and private hospitals provision of uncompensated care. One possible explanation is the legal environment may not be a major determinant of the level of uncompensated care. That is, such o ther factors as hospital ownership, the local unemployment rate, market competi tive forces and community expectations may be the dominant factors determining the provision of uncompensated care by private hospitals. Another possibility is that the penalties for noncompliance with the community benefit requirements are not large enoug h to affect hospital behavior For example, nonprofit hospitals in Texas are subject to a penalty that does not exceed $1,000 per day if they fail to submit an annual report which is a relative small amount of expenses for most hospitals. Second, the stu dy finding s could be partially attributed to the fact that states without community benefit requirements may have implemented laws that indirectly
130 influence d the level of uncompensated care. Certificate of need (CON) legislation is one possible example Al though there are no empirical studies that have shown an effect of CON laws on provision of uncompensated care, an interview of administrators of Florida s CON program (Campbell and Fournier 1993) indicated that hospitals were more likely to be approved th rough CON laws if they had a higher share of indigent care in the market ; Florida is one of the s t ates without any community benefit requirements. The major study finding i s consistent with the results of the study by Ginn, Shen and Mosely (2009), which u sed a quasi experimental design to assess the effect of community benefit laws on community orientation activities and health promotion services offered by nonprofit hospitals. Comparing states by the presence or absence of such laws they found no signifi cant effect of community benefit requirements on the amount of the two services provided by nonprofit hospitals. Even f or profit hospitals in the states with community benefit laws even experienced a significant decrease in providing community orientation and health promotion services compared to comparable hospitals in states without community benefit laws The authors suggested that the findings could be attributed to the fact that nonprofit hospital managers in states without community benefit laws could be influenced by other managers in states with such laws through professional socialization. Alternatively the finding s could be due to states without laws attempt ing to forestall such regulation after realizing that other states and the federal govern ment were considering passing or implementing community benefit laws to regulate nonprofit hospitals. Findings on the association between the level of strictness of community benefit requirements and the provision of uncompensated care were similar to the study by
131 Kennedy et al. (2010), which tested the Texas community benefit law s effect on hospital charity care spending. In 1993, Texas implemented the strictest law among the states, requiring a minimum standard and imposing punishment for noncompliance However, the study results showed that the 1993 Texas law resulted in a significant decrease (although the magnitude was close to 0) in the average level of charity care provided by nonprofit hospitals that did not receive any DSH payment. The authors sug gested that hospitals below the threshold may increase the provision of charity care, while hospitals above the threshold may decrease charity care provision to the required level. It is important to note that the lack of an association between states c ommunity benefit requirements and private hospitals provision of uncompensated care does not necessarily mean that the se requirements do not affect the level of uncompensated care. That is, other findings showed that differences in the provision of uncomp ensated care between nonprofit and for profit hospitals were much larger in states with community benefit requirements than states without such policies. This suggests that the requirements may be central to nonprofit hospitals efforts to distinguish them selves from for profit ownership in the provision of uncompensated care. Similarly, the CBO (2006) examined nonprofit hospitals and the community benefit requirements in five states. They found that the differences in the provision of uncompensated care be tween nonprofit and for profit hospitals were the largest in Texas and Indiana which imposed a strict level of community benefit requirements on nonprofit hospitals. Policy Implications The study findings have several implications for policymakers. First states may need to reconsider the benefit derived from the requirements and whether the benefit
132 exceed the spending on administration of the program. States should clearly define community benefit and/or provide examples of what could be counted for bene fit for the local community in terms of the reporting mechanism and performance evaluation. Second, states community benefit requirements may be necessary to maintain organizational legitimacy for nonprofit hospitals. Nonprofit hospitals in these states m ay make greater efforts to provid e uncompensated care and unprofitable public program services to differentiate themselves from for profit ownership. Related empirical studies (Bazzoli, Clement, and Hsieh 2010; CBO 2006 ; Schneider 2007) suggested that st ate and local government s should construct stricter charity care standards given that they are at the forefront in terms of the loss of revenue due to hospitals tax exempt statu s. Results from this study however, suggest that implementing a minimum stan dard might have a double edge d effect on nonprofit hospitals. That is if the minimum standard required is below the average level of uncompensated care in the state or nationwide, implementing the strictest level of community benefit laws with minimum sta ndard requirements may have no effect o r even decrease the provision of uncompensated care by nonprofit hospitals. In addition, the minimum standard should be based on the amount of charity care needed in the community instead of a static amount of communi ty benefit across all communities. The passage of the PPACA significantly modified existing policies regarding community benefit reporting and needs assessment. Although the expansion of Medicaid eligibilit y and the individual health insurance mandate may reduce the provision of uncompensated care it is estimated that 23 million Americans might still be uninsured under health reform ( Buettgens and Hall 2011 ) This study provides empirical
133 support for the requirements in the recent health care reform regar ding community health needs assessment and reporting guidelines. It also suggests the government should take an evidence based strategy to measur ing and improv ing community benefit Limitations One of the study limitation s is measurement error. The MCR dat a is based on self reported information from Medicare certified h ospitals. Hospitals may vary on how they assessment of costs and resource use. In addition the MCR only provides information on hospitals uncompensated care without differentiating between charity care and bad debt, and not every hospital reported their uncompensated care cost Charity care is more precise at measuring care provided to patients with no ability to pay. Missing data also might be a problem in this study although multiple imp utation was conducted to address this issue to some extent Some hospitals did not report their uncompensated care cost during the study period. And managed care penetration was replaced by Medicaid and Medicare managed care penetration due to the lack of data. Multiple imputation was used to adjust for missing data. The underlying assumption was Missing at Random esults may be biased if this assumption fails to hold The last limitation is the cross sectional s tudy design. Although a cross sectional comparison between states with community benefit requirements and states without such requirements illuminated the association between health policies and provision uncompensated care, a quasi experimental study woul d be required to make more causal inferences to examine the impact of policies, comparing the change of provisions of uncompensated care before and after the implementation of community benefit requirements.
134 Future Research Future studies are needed to use a quasi experimental study design to examine the effect of state community benefit requirements on the provision of charity care by nonprofit hospitals. In addition, with the passage of health care reform, its effect on the provision of community benef it including both uncompensated care and other community oriented services, should also be assessed. Moreover, since communities may differ in their perception of medical resource needs it is necessary to construct performance indicators that reflect a c ommunity priorities (Gale 2010).
135 APPENDIX EXAMPLE OF THE MCR WORKSHEET S 10: HOSPITAL UNCOMPENSATED CARE
136 LIST OF REFERENCES Alexander J A Morrisey M A A resource dependence model of hospital contract management Health services research 1989;24(2) :259 284. American Hospital Association. Uncompensated hospital care cost fact sheet. policy. 2010 ; Available at www.aha.org/content/00 10/10uncompensatedcar e.pdf. Accessed October 10, 201 1 Andersen RM. Revisiting the behavioral model and access to medic al care: does it matter? J Health Soc Behav 1995;36:1 10. Antel JJ, Ohsfeldt RL, Becker ER. State regulation and hospital costs The Review of Economics and Statistics 1995;77(3):416. Atkinson G, Helms WD, Needleman J. State trends in hospital uncompensa ted care. Health Affairs 1997;16(4):233 241. Baltagi B H. Econometric Analysis of Panel Data New York: Wiley ; 1995 Banaszak Holl J, Zinn JS, Mor V. The impact of market and organizational characteristics on nursing care facility service innovation. a re source dependency perspective. Health Services Research 1996; 31 :97 116. Banks D A Paterson M, Wendel J. Uncompensated hospital care: charitable mission or profitable business decision? Health economics 1997;6(2):133 43. Bazzoli GJ, Clement JP, Hsieh H C ommunity benefit activities of private, nonprofit hospitals. Journal of health politics, policy and law 2010;35(6):999 1026. Bazzoli GJ, Lindrooth RC, Kang R, Hasnain Wynia R. The influence of health policy and market factors on the hospital safety net. H ealth services research 2006;41(4 Pt 1):1159 80. Blewett LA, Davidson G, Brown ME, Maude Griffin R. Hospital provision of uncompensated care and public program enrollment. Medical care research and review: MCRR 2003;60(4):509 27. Bovbjerg R, Cuellar A, H olahan J. Market Competition and Uncompensated Care Pools Report. Washington, DC: Urban Institute ; 2000. Bowman HW. Reexamining the property tax exemption Land Lines 2003;15(3). Buczko W. Factors affecting charity care and bad debt charges in Washington hospitals. Hospital & Health Services Administration. 1994; 39 (2): 179+. Buettgens M Hall MA Who will be uninsured after health insurance reform? Robert Wood Johnson Foundation 2011.
137 Burns J. Are Nonprofit hospitals really charitable?: taking the quest ion to the state and local level Journal of Corporation Law 2004;29(3):665 683. California Health and Safety Code Â§ 127340 Available at http://www.oshpd.ca.gov/HID/SubmitData/CommunityBenefit/SB697CommBenefit s.pdf Accessed October, 2011. Callahan M. Hea lthcare providers constricted by financial, legislative, and regulatory issues. 24(3):143 6. Campbell ES, Fournier GM. Certificate of need deregulation and indigent hospital care. Journal of health politics, policy and law 1993;18(4):905 25. Carreyrou J, Martinez B. Nonprofit hospital s,once for the poor, strike it rich. Wall Street Journal 2008:10. Congressional Budget Office Nonprofit hospitals and the provision of community benefit 2006 Available at: http: //www.cbo.gov/ftpdocs/76xx/doc7695/12 06 Nonprofit.pdf Accessed October 10, 2011. Chen L W, Shambaugh Miller MD, Meza JL, et al. Inpatient hospital service utilization of the uninsured in Nebraska: policy implications for addressing uncompensated care. J ournal of health care for the poor and underserved 2009;20(2):444 57. Cheng J et al. Real longitudinal data analysis for real people: b uilding a good enough mixed model. Stat Med. 2010 ; 29(4):504 20. Clement JP, White KR, Valdmanis V. Charity care: do not for profits influence for profits ? Medical Care Research and Review 2002;59(1):59 78. Coleman SK. The Hill Burton uncompensated services program 2005. Available at: http://assets.opencrs.com/rpts/98 968_20050523.pdf. Accessed October 10, 2011. Colombo J D. The role of tax exemption in a competitive health care market. Journal of health politics, policy and law 2006;31(3):623 42. Cook K, Shortell SM, Conrad DA, Morrisey MA A theory of organizational response to regulation: the case of hospitals Academy of Management 2011;8(2):193 205. use of Medicaid UPL and DSH financing mechanisms Health Affairs 2004;23(2):245 257. Coughlin TA, Liska D The Medicaid Disproportionate Share Hospital payment program: background an d issues. 1997. Available at http://www.urban.org/url.cfm?ID=307025 Accessed October 2011. Coughlin TA, Zuckerman S, McFeeters J. Restoring fiscal integrity to Medicaid financing? Health affairs 26(5):1469 80.
138 Cram P, Bayman L, Popescu I, et al. Uncompen sated care provided by for profit, not for profit, and government owned hospitals. BMC health services research 2010;10:90. Cunningham PJ, Tu HT. A changing picture of uncompensated care. Health affairs 1997;16(4):167 75. Davidoff AJ, LoSasso AT, Bazzoli GJ, Zuckerman S. The effect of changing state health policy on hospital uncompensated care. Inquiry 2000;37(3):253 67. Davis K, Rowland D. Uninsured and underserved: inequities in health care in the United States. Health and society 1983;61(2):149 76. D erek D. Evaluation of the hospital charity care program in New Jersey 2007. Available at: http://www.cshp.rutgers.edu/Downloads/7060.pdf. Accessed October 10, 201 1 DiMaggio PJ, Powell W. 1983. The iron cage revisited: i nstitutional isomorphism and collec tive rationality in organizational fields. American Sociologi cal Review. 48: 147 160 Donabedian A. Explorations in Quality Assessment and Monitoring: The Definition Of Quality And Approaches To Its Assessment. Ann Arbor, MI: Health Administration Press; 19 80. Duncan RP. Uncompensated hospital care Medical Care Research and Review 1992;49(3):265 330. Dunn DL, Chen M. Uncompensated hospital care payment and access for the uninsured: evidence from New Jersey. Health services research 1994;29(1):113 30. Dusa nsky R, Kalman PJ. Toward an economic model of the teaching hospital Journal of Economic Theory 1974;7:210 223. Emerson RM. Power dependence relations American Sociological Review. 1962;27(1):31 40. Fagnani L, Tolbert J The Dependence of Safety Net Hospitals and Health Systems on the Medicare and Medicaid Disproportionate Share Hospital Payment Programs New York: The Commonwealth Fund; 1999. Feldstein M. Hospital Cost Inflation: A study of Nonprofit Price Dynamics. The American economic review 1971;6 1(5):853 872. Feldstein PJ, The politics of health legislation. An Economic Perspective Health Administration Press, Chicago Ill ; 2006.
139 Feldstein PJ. Health Policy Issues: An Economic Perspective on Health Reform. 4th ed Chicago, Ill: Health Administrat ion Press; 2007. Ferrier GD, Rosko MD, Valdmanis VG. Analysis of uncompensated hospital care using a DEA model of output congestion. Health Care Management Science 2006;9(2):181 188. Folkerts LL. Do standards for proving deservedness of federal tax exemptions The Journal of Corporation Law 2009;34(2):611 640 Fonkych K, Melnick G. Disproportionate share hospital subsidies for treating the uninsured. Medical care 2010;48(9):809 14. Frank RG, Salkever DS. The supply of charity services by nonprofit hospitals: motives and market structure. The Rand journal of economics 1991;22(3):430 45. Gale GA. Community benefit under health reform: focus on community needs and engagement 2010. Available at http://flexmonitoring.org/documents/CommunityBenefit Engagement_NCSFPMeeting071410_Gale.pdf Accessed May, 2012. Gapenski LC Langland Orban B, Vogel WB The determin ants of hospital profitabilit y. Hospital and Health Services Administration 1993;38(1): 63 80. Gaskin DJ. Altruism or moral hazard: t he impact of hospital uncompensated care pools. Journal of health economics 1997;16:397 416. Ginn GO, Moseley CB. The impact of state community benefi t laws on the community health orientation and health promotion services of hospitals. Journal of health politics, policy and law 2006;31(2):321 44. Ginn GO Shen JJ, Moseley CB. Community benefit laws, hospital ownership, community orientation activitie s, and health promotion services Health Care Manage Rev 2009; 34 : 109 118 Grogan CM, Gusmano MK. Political strategies of safety net providers in response to medicaid managed care reforms. Journal of health politics, policy and law 2009;34(1):5 35. Gruber J. The effect of competitive pressure on charity: hospital responses to price shopping in California. Journal of health economics 1994;13(2):183 212. Hadley J, Holahan J. How much medical care do the uninsured use and who pays for It? Health Affairs Web Exclusive. 2003 Available at http://www.healthaffairs.org/WebExclusives/Hadley_Web_Excl_021203.htm Accessed October 10, 2011.
140 Hadley J Holahan J. The cost Children 2004. Available at: http://www.kff.org/uninsured/upload/The Cost of Care for the Uninsured What Do We Spend Who Pays and What Would Full Coverage Add to Medical Spending.pdf. Accessed October 10, 2011. Hadley J, Holahan J, Coughlin T, Miller D. Covering the uninsured in 2008: current costs, sources of payment, and incremental costs. Health affairs 2008;27(5):w399 415. Hearne J. Medic aid disproportionate share payments. CRS report for Congress. 2005. Available at http://www.law.umaryland.edu/marshall/crsreports/crsdocuments/97 48301102005.pdf Accessed October, 2011. Heimovics RD, Herman RD, Coughlin CJ. Executive leadership and resour ce dependence in nonprofit organizations: a frame analysis Public Administration Review 1993;53(5):419 427. Hellinger FJ. Tax exempt hospitals and community benefit : a review of state reporting requirements. Journal of health politics, policy and law 20 09;34(1):37 61. Hellinger FJ. The effect of Certificate Of Need laws on hospital beds and healthcare expenditures: an empirical analysis. The American journal of managed care 2009;15(10):737 44. Henderson JW. Health E conomics and P olicy Florence KY: Sout h Western College Publishing; 1999. Herfindahl Hirschman Index t he Antitrust Division of the Department of Justice Available at http://www.justice.gov/atr/public/guidelines/hhi.html Accessed October, 2011. Hillman AJ, Withers MC, Collins BJ. Resource de pendence theory: a review Journal of Management 2009;35(6):1404 1427. Holahan J Zedlewski S. Expanding Medicaid to cover uninsured Americans. Health Affairs 1991;10(1):45 61 Holahan J, Garrett B. The Cost of Uncompensated Care W ith and Without Health Reform Washington, DC: The Urban Institute ; 2010. Horwitz JR. Making profits and providing care: comparing nonprofit, for profit, and government hospitals. Health Aff (Millwood). 2005 ;24(3):790 801. Howell DC. The A nalysis of M issing D ata. Handbook of So cial Science Methodology London: Sage 2008;
141 Hsieh H M, Clement DG, Bazzoli GJ. Impacts of market and organizational characteristics on hospital efficiency and uncompensated care. Health care management review 2010;35(1):77 87. Jacobson PD, Dalton VK, Berson Grand J, Weisman CS. Survival strategies for Health services research 2005;40(3):923 40. Johnson BL. Resource Dependence Theory: A Political Economy Model of Organizations University of Utah: Department of Educational Administration ; 1 995. Kaiser Family Foundation 2011 The uninsured: a primer 2011; Available at http://www.kff.org/uninsured/upload/7451 07.pdf Accessed October 10, 2011 Kennedy FA, Burney LL, Troyer JL, Stroup JC Do non profit hospitals provide more charity care when faced with a mandatory minimum standard? e vidence from Texas. Journal of Accounting and Public Policy 2010;29(3):242 258. Langland Orban B, Pracht E, Salyani S. Uncompensated care provided by emerge ncy physicians in Florida emergency departments. Health care management review 2005;30(4):315 21. Lo Sasso AT, Seamster DG. How federal and state policies affected hospital uncompensated care provision in the 1990s. MCRR 2007;64(6):731 44. Mann JM, Melnick G a, Bamezai a, Zwanziger J. A profile of uncompensated hospital care, 1983 1995. Health Affairs 1997;16(4):223 232. McClyde. Not for profit hospitals conversion issues prompt in creased state oversight 1997 ; Available at: http://www.gao.gov/archive/1998/he98024.pdf Accessed October 10, 2011. McKay uncompensated care. Inquiry 2007;44(1):114 24. McKay NL, Zhong Y, Hu X. The provision of uncompensated care (working paper) University of Florida. 2011 Mechanic RE. Medicaid's Disproportionate Share Hospital p rogram: c omplex s tructure, critical payments Washington, DC: National Health Policy Forum; 2004. Medicare Payment Advisory Commission (MedPAC). Report to the Congress: Medicare payment policy. Available at: http://www.me dpac.gov/documents/Mar10_EntireReport.pdf. Accessed October 31, 2011.
142 Miller Millesen JL. U nderstanding the behavior of nonprofit boards of directors: a theory based approach Nonprofit and Voluntary Sector Quarterly 2003;32(4):521 547. Minnesota Statutes Annotated. Vol. 11, p. 210, sec. 144.698, subdivision 1, para. 5 (2002). Mitchell JM, Nort on S a. Provider assessments, the uninsured, and uncompensated care The Milbank Quarterly 1996;74(4):545. Morrisey MA, Wedig GJ, Hassan M. Do nonprofit hospitals pay their way? Health Affairs 1996;15(4):13 2 144. Moseley CB. Community benefit laws, hospital ownership, community orientation activities and health promotion services. Health Care Management Review. 2009;34(2):109 118. Needleman J, Lamphere J, Chollet D. Uncompensated care and hospital conversi ons in Florida. Health affairs 1999;18(4):125 33. Needleman J. The role of nonprofits in health care. Journal of health politics, policy and law 2001;26(5):1113 30. Neun S, Santerre R. Health Economics Theories, Insights, and Industry Studies Orlando, F L: The Dryden Press; 2000. Kaiser Family Foundation. New Health Reform Law: Patient Protection and Affordable Care Act. Washington, DC: 2011 ; Available at : http://www.kff.org/healthreform/upload/8061.pdf Accessed December 2011. Newhouse NP. Toward a theo ry of nonprofit institutions: an economic model of a hospita l. 1970;60(1):64 74. Nicholson S, Pauly MV, Burns LR, Baumritter B, Asch D. Measuring community benefit provided by for profit and nonprofit hospitals. Health Affairs 2000;19(6):168 177. Noble A a, Hyams AL, Kane NM. Charitable Hospital Accountability: A Review and Analysis of Legal and Policy Initiatives. The Journal of Law, Medicine & Ethics 1998;26(2):116 137. Norton EC, Staiger DO. How hospital ownership affects access to care for the uninsur ed. The Rand journal of economics 1994;25(1):171 85. Oberlander J. Learning from failure in health care reform. N Engl J Med 2007;357(17):1677 9. Oliver C. Strategic r esponse to institutional Processes. Academy of Management Review 1991;16(1):145 179.
143 P atel K., Rushefsky ME Health Care Politics a nd Policy in America New York: M.E. Sharp; 2006. Pauly MV. Nonprofit firms in medical markets. The American economic review 1987;77(2):257 62. Pfeffer J, Salancik GR. The External Control of Organizations: A R esource Dependence Perspective New York: Harper & Row;1978. Phelps CE. Health Economics 3 rd Ed Adison Wesley; 2002. Powers M, Faden RR Social Justice: The Moral Foundations o f Public Health a nd Health Policy Oxford University Press; New York, NY ; 2006. Re v. Rul. 56 185, 1956 1 Cumulative Bulletin 202, 203 Available at http://taxlinks.com/rulings/1956/revrul56 185.htm A ccessed October 2011. Re v. Rul. 69 545 1969 2 C.B. 117 Available at http://www.irs.gov/pub/irs tege/rr69 545.pdf A ccessed October 2011. Rev. Rul. 83 157, 1983 2 C.B. 94 Available at http://www.irs.gov/pub/irs tege/rr83 157.pdf A ccessed October 2011. Rev. Rul. 98 15, 1998 12 Available at www.irs.gov/pub/irs drop/rr 98 15.pdf Accessed October 2011. Rosenau PV, Linder SH. Tw o Decades of Research Comparing For Profit and Nonprofit Health Provider Performance in the United States*. Social Science Quarterly 2003;84(2):219 241. Rosko MD. The supply of uncompensated care in Pennsylvania hospitals: motives and financial consequenc es. Health care management review 2004;29(3):229 39. Ross JS. The committee on the costs of medical care and the history of health insurance in the United States. The Einstein Quarterly Journal of Biology and Medicine 2002;19: 129 134. Rousseau D Schnei der A. Current issues in Medicaid financing anoverview of IGTs, UPLs, and DHS. Kaiser Commission on Medicaid and the Uninsured. 2004. Available at http://www.kff.org/medicaid/7071.cfm Accessed October, 2011. Rubin D B. Multiple Imputation for Nonresponse i n Surveys New York : J. Wiley & Sons ; 1987. Ruef M Scott W R. A multidimensional model of organizational legitimacy: hospital survival in changing institutional environments Administrative Science Quarterly. 1998;43:877 904.
144 Saidel J. Resource interdepen dence: the relationship between state agencies and nonprofit organizations Public Administration Review 1991;51(6):543 553. Santerre R, Neun S. Health Economics Theories, Insights, and Industry Studies Orlando, FL: The Dryden Press; 2000. Santerre RE, V ernon JA. The consumer welfare implications of the hospital ownership Health Economics 2006;1199:1187 1199. Schafer JL. Multiple Imputation: a primer. Statistical Methods in Medical Research 1999; 8: 3 15 Scherer FM. Industrial Market Structure and Economic Performance 2 nd ed., Rand McNally, Chicago, IL;1980. Schneider H. Paying their way? do nonprofit hospitals justify their favorable tax treatment? Inquiry 2007; 44:187 199. Schuhmann TM. National trends in uncompens ated care and profitability. Healthc Financ Manage. 2008;62(9):110 6, 118. S chuler DA, Rehbein K, Cramer RD Pursuing strategic advantage through political means : a multivariate approach The Academy of Management Journal 2002;45(4):659 672. Scott WR, Da vis GF Organizations and organizing: Rational, natural, and open systems perspectives Upper Saddle River, NJ: Prentice Hall ;2007 Simpson JB. State Certificate of Need American Journal of Public Health 1985;75(10):1225 1229 Singer JD. Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. Journal of Educational and Behavioral Statistics. 1998;24(4):323 355. Sloan, F. Not for profit ownership and hospital behaviour, in A. J. Culyer and J. Newhouse (eds), The Handbook of Health Economics, Amersterdam: North Holland, Vol 1B ;2000 Spencer CS. Do uncompensated care pools change the distribution of hospital care to the uninsured? Journal of health politics, policy and law 1998;23(1):53 7 3. Starr P. The Social Transformation of American Medicine New York: Basic; 1982. Suchman M. Managing legitimacy: strategic a nd institutional approaches. Academy of Management Review. 1995;20:571 610.
145 Sugin L, Hill F, Jones D. T ax exempt organizations and exempt New England Law Review 2010:7 10. Sutton JP, Stensland J. Promoting accountability: hospital charity care in California, Washington and Texas. Journal of Health Care for the Poor and Underserved 2004;15(2):237 250. Texas Health and Safety Code 311 Available at www.dshs.state.tx.us/chs/hosp/ch311.pdf Accessed October 2011. The Kaiser Family Foundation. The uninsured a primer 2011; Available at: http://www.kff.org/uninsured/upload/7451 07.pdf. Accessed October, 20 11. The United States Department of Justice. Herfindahl Hirschman Index 2012; Available at http://www.justice.gov/atr/public/guidelines/hhi.html Accessed May, 2012. Thorpe KE, Brecher C. Improved access to care for the uninsured poor in large cities: do public hospitals make a difference? Journal of health politics, policy and law 1987;12(2):313 24. Thorpe KE, Phelps CE. The social role of not for profit organizations: hospital provisi on of charty care. Economic inquiry 1991;29(3):472 484. Thorpe KE, Seiber EE, Florence CS. The impact of HMOs on hospital based uncompensated care. Journal of health politics, policy and law 2001;26(3):543 55. Thorpe KE, Spencer C. How do uncompensated c are pools affect the level and type of care? r esults from New York s tate. Journal of health politics, policy and law 1991;16(2):363 81. United States Government Accountability Office. Variation in standards and guidance limits comparison of how hospitals meet community benefit requirements. 2008; Available at: http://www.gao.gov/products/GAO 08 880 Accessed October, 2011. US General Accounting Office. Long standing concerns about inappropriate state arrangements support need for improved federal oversight 2007 ; Available at: http://www.gao.gov/new.items/d08650t.pdf Accessed October, 2011. Weiner SJ, VanGeest J, Abrams RI, Moswin A, Warnecke R. Managing the unmanaged: a case study of intra institutional determinants of uncompensated care at healthcare ins titutions with differing ownership models. Medical care 2008;46(8):821 8. Weissman JS, Gaskin DJ, Reuter J. Hospitals' care of uninsured patients during the 1990s: the relation of teaching status and managed care to changes in market share and market conc entration. Inquiry 2003; 40(1):84 93.
146 Weissman JS. The trouble with uncompensated hospital care. The New England journal of medicine 2005;352(12):1171 3. Weissman JS. Uncompensated hospital care. Will it be there if we need it? JAMA 1996; 276(10):823 8. West Virginia Code of State Rules Â§ 110 3 24 Available at http://www.communitycatalyst.org/projects/hap/free_care?id=0048 Accessed October, 2011. Wood KM. Legislatively Mandated Charity Care for Nonprofit Hospitals: Does Government Intervention Make any Difference? Review of Litigation 2000;20(3):709 742. Young GJ, Desai KR. Nonprofit hospital conversions and community benefit : new evidence from three states. Health affairs (Project Hope) 1999;18(5):146 55. Young GJ, Desai KR, Lukas CV. Does the sale o f nonprofit hospitals threaten health care for the poor? Health affairs 1997;16(1):137 141 Younis M, Rice J, Barkoulas J. An empirical investigationof hospital profitability in the post PPS era. J Health Care Finance 2001;28:65 73. Zinn JS, Weech RJ, Bran non D. Resource dependence and institutional elements in nursing home TQM adoption. Health services research 1998;33(2 Pt 1):261 73. Zwanziger J, Khan N. Safety net hospitals. 2008; 65(4):478 95.
147 BIOGRAPHICAL SKET CH Yue(Wendy) Zhong was born in Taiyuan, Shanxi province of China. Being influe nced by her parents, she became interested in medical science and later received the Bachelor of Medicine from the West China Medical Center of the Sichuan University During the residency at the West China Medical Center, she found that poor people were more likely to avoid seeking health care and less likely to p urchase needed medicine when they got sick This phenomenon not only led to serious health problem s but also resul ted in higher health expenditure s These experiences inspire d her future research in the area of health po licy and health system Following the undergraduate study she decided to pursue a PhD degree in H ealth S ervices R esearch at the University of Florida At UF, she received rigorous training s in the area of health policy, health economics health insurance as well as patient outcome r esearch In addition to coursework, s he served as a g raduate research assistant in projects pertaining to Florida Medi caid Reform. Finally she successfully defended her dissertation with the title s tate community benefit requirements and private hospitals provision of uncompensated care and graduated in summer 2012.