This is the accessible text file for GAO report number GAO-05-743T entitled 'Nonprofit, For-Profit, and Government Hospitals: Uncompensated Care and Other Community Benefits' which was released on May 26, 2005. This text file was formatted by the U.S. Government Accountability Office (GAO) to be accessible to users with visual impairments, as part of a longer term project to improve GAO products' accessibility. Every attempt has been made to maintain the structural and data integrity of the original printed product. Accessibility features, such as text descriptions of tables, consecutively numbered footnotes placed at the end of the file, and the text of agency comment letters, are provided but may not exactly duplicate the presentation or format of the printed version. The portable document format (PDF) file is an exact electronic replica of the printed version. We welcome your feedback. Please E-mail your comments regarding the contents or accessibility features of this document to Webmaster@gao.gov. This is a work of the U.S. government and is not subject to copyright protection in the United States. It may be reproduced and distributed in its entirety without further permission from GAO. Because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately. Testimony: Before the Committee on Ways and Means, House of Representatives: United States Government Accountability Office: GAO: For Release on Delivery Expected at 10:00 a.m. EDT: Thursday, May 26, 2005: Nonprofit, For-Profit, and Government Hospitals: Uncompensated Care and Other Community Benefits: Statement of David M. Walker: Comptroller General of the United States: GAO-05-743T: GAO Highlights: Highlights of GAO-05-743T, a testimony before the Committee on Ways and Means, House of Representatives Why GAO Did This Study: Before 1969, IRS required hospitals to provide charity care to qualify for tax-exempt status. Since then, however, IRS has not specifically required such care, as long as the hospital provides benefits to the community in other ways. Seeking a better understanding of the benefits provided by nonprofit hospitals, this Committee requested that GAO examine whether nonprofit hospitals provide levels of uncompensated care and other community benefits that are different from other hospitals. This statement focuses on, by ownership group, hospitals’ (1) provision of uncompensated care, which consists of charity care and bad debt, and (2) reporting of other community benefits. The hospital ownership groups were (nonfederal) government, nonprofit, and for- profit. To compare the three hospital ownership groups, GAO obtained 2003 data from five geographically diverse states with substantial representation of the three ownership groups in each state. GAO analyzed cost data from two perspectives—each hospital group’s percentage of (1) total uncompensated care costs in a state and (2) patient operating expenses devoted to uncompensated care. What GAO Found: Government hospitals generally devoted substantially larger shares of their patient operating expenses to uncompensated care than did nonprofit and for-profit hospitals. The nonprofit groups’ share was higher than that of the for-profit groups in four of the five states, but the difference was small relative to the difference found when making comparisons with the government hospital group. Further, within each group, the burden of uncompensated care costs was not evenly distributed among hospitals but instead was concentrated in a small number of hospitals. This meant that a small number of nonprofit hospitals accounted for substantially more of the uncompensated care burden than did others receiving the same tax preference. Figure: Average Percent of Patient Operating Expenses Devoted to Uncompensated Care, by Hospital Ownership Type, 2003: [See PDF for image] [End of figure] Hospitals in the five states—nonprofit, for-profit, and government hospitals—reported providing a variety of services and activities, which the hospitals themselves defined as community benefits. Community benefits include such services as the provision of health education and screening services to specific vulnerable populations within a community, as well as activities that benefit the greater public good, such as education for medical professionals and medical research. GAO was unable to assess the value of these benefits or make systematic comparisons between hospitals or across states. These observations illustrate a larger point—namely, that current tax policy lacks specific criteria with respect to tax exemptions for charitable entities and detail on how that tax exemption is conferred. If these criteria are articulated in accordance with desired goals, standards could be established that would allow nonprofit hospitals to be held accountable for providing services and benefits to the public commensurate with their favored tax status. www.gao.gov/cgi-bin/getrpt?GAO-05-743T. To view the full product, including the scope and methodology, click on the link above. For more information, contact A. Bruce Steinwald at (202) 512-7101. [End of section] Mr. Chairman and Members of the Committee: I am pleased to be here today as you discuss issues regarding tax exemptions for nonprofit hospitals. At this Committee's recent hearing on the tax-exempt sector as a whole, I emphasized the importance of reviewing this sector, drawing parallels to our agency's call to reexamine all major federal policies and programs in light of 21st century challenges.[Footnote 1,2]. Provisions granting federally recognized tax-exempt status and associated policies have been layered on one another to respond to challenges at the time, but they need to be reviewed and revised to reflect 21st century changes and challenges. On a broad scale, a comprehensive reexamination could help address whether exempt entities are providing services and benefits to the public commensurate with their favored tax status, whether the current number and nature of exemptions continue to make sense, whether the conditions and restrictions on the activities of tax-exempt entities remain relevant, and whether the framework for ensuring that exempt entities adhere to the requirements attendant to their status is satisfactory. There are a number of issues that merit reexamination, including whether nonprofit hospitals perform sufficiently different services of benefit to the public to justify their tax exemption. To examine these hospitals' tax-exempt status, we must look back several decades. Before 1969, the Internal Revenue Service (IRS) required hospitals to provide charity care to qualify for tax-exempt status. Since then, however, IRS has not specifically required such care for a hospital to be exempt from federal taxation and have access to tax-exempt bond financing and charitable donations, as long as the hospital provides benefits to the community in other ways. Community benefits include such services as the provision of health education and screening services to specific vulnerable populations within a community, as well as activities that benefit the greater public good, such as education for medical professionals and medical research. Nonprofit hospitals may also be exempt under state law from state and local taxes. Seeking a better understanding of the benefits provided by nonprofit hospitals, this Committee requested that we examine whether nonprofit hospitals provide levels of uncompensated care--care provided to a patient that a hospital is not reimbursed for--and other community benefits that are different from other hospitals. My remarks today will focus on our examination, for selected states, of (1) the provision of uncompensated care by state and local government-owned, nonprofit, and for-profit hospitals and (2) hospitals' reporting of other community benefits. To examine the provision of uncompensated care by the three hospital ownership groups,[Footnote 3] we analyzed cost data from two perspectives, namely each hospital group's percentage of (1) total uncompensated care costs in a state and (2) patient operating expenses devoted to uncompensated care. We obtained 2003 data from five states- -California, Florida, Georgia, Indiana, and Texas. Hospitals in these states include 46 percent of the nation's for-profit hospitals and more than a quarter of all hospitals in the three ownership groups. We selected these states because they represented geographically diverse areas; had a number of hospitals in each ownership group sufficient to make comparisons; and collected hospital-specific uncompensated care data, which not all states maintain.[Footnote 4] We compared each hospital ownership group's provision of uncompensated care by examining each group's uncompensated care costs[Footnote 5] as a percentage of its total patient operating expenses.[Footnote 6] Our measure of uncompensated care includes the cost of charity care as well as bad debt and deducts any payments made by or on behalf of individual patients. We limited our analysis of uncompensated care to nonfederal, short-term, acute care general hospitals.[Footnote 7] In doing our work, we tested the reliability of the state data and determined they were adequate for our purposes.[Footnote 8] To examine hospitals' provision of community benefits other than uncompensated care, we reviewed 21 hospital or hospital systems' reports and Web sites for information about such benefits. These reports and Web sites covered nonprofit, for-profit, and government hospitals in the five states. We also examined laws in the five states regarding community benefit requirements for nonprofit hospitals, reviewed the literature, and interviewed state officials and state hospital association representatives. In addition, we interviewed officials from the Centers for Medicare & Medicaid Services (CMS), the American Hospital Association, and the Federation of American Hospitals. We conducted our work from February 2005 through May 2005 in accordance with generally accepted government auditing standards. (See app. I for more detail on our scope and methodology.) In summary, the cost burden of providing uncompensated care varied among the three hospital groups, but the burden was generally concentrated in a small number of hospitals. In four of the five states, government hospitals, as a group, devoted substantially larger shares of their patient operating expenses to uncompensated care than did nonprofit and for-profit hospitals. The nonprofit hospitals' uncompensated care costs, as a percentage of patient operating expenses, were higher on average than those of the for-profit hospitals in four of the five states, but the differences were generally not as great as the differences between the government hospitals and both these groups. Further, the burden of uncompensated care costs was not evenly distributed within each hospital group but instead was concentrated in a small number of hospitals. For example, in California's nonprofit hospital group, the top quarter of hospitals, ranked by uncompensated care as a percentage of patient operating expenses, averaged 7.2 percent devoted to uncompensated care compared with an average of 1.4 percent for hospitals in the bottom quarter. Regardless of ownership status, the hospitals we reviewed reported providing a wide range of other community benefits, from health education to clinic services specifically for the community's indigent population. Variations in the types of community benefits hospitals in the five states reported providing could be explained by differences in the services hospitals chose to provide as well as by variation in the applicability, specificity, and breadth of state requirements. Background: In 2003, of the roughly 3,900 nonfederal, short-term, acute care general hospitals in the United States,[Footnote 9] the majority--about 62 percent--were nonprofit. The rest included government hospitals (20 percent) and for-profit hospitals (18 percent). States varied-- generally by region of the country--in their percentages of nonprofit hospitals (see fig. 1). For example, states in the Northeast and Midwest had relatively high concentrations of nonprofit hospitals, whereas in the South the concentration was relatively low. Figure 1: Geographic Distribution of Nonprofit Hospitals in 2003: [See PDF for image] Note: Hospitals include nonfederal, short-term, acute care general hospitals, but not critical access hospitals that provide general acute care. [End of figure] The five states we reviewed varied in number and ownership composition of hospitals (see table 1). For example, in California and Indiana, nonprofit hospitals accounted for over half of each state's hospitals. In Texas, government hospitals made up the state's largest percentage, although the distribution between nonprofit, for-profit, and government hospitals was similar; in Florida, most hospitals were either nonprofit or for-profit, while 11 percent were government. Table 1: Distribution of Hospitals Reviewed, by Ownership Type, 2003: California; Total number of hospitals: 331; Percent nonprofit: 51%; Percent for-profit: 27%; Percent state and local government: 22%. Florida; Total number of hospitals: 169; Percent nonprofit: 43%; Percent for-profit: 46%; Percent state and local government: 11%. Georgia; Total number of hospitals: 133; Percent nonprofit: 43%; Percent for-profit: 21%; Percent state and local government: 36%. Indiana; Total number of hospitals: 97; Percent nonprofit: 56%; Percent for-profit: 9%; Percent state and local government: 35%. Texas; Total number of hospitals: 332; Percent nonprofit: 33%; Percent for-profit: 32%; Percent state and local government: 35%. Source: GAO analysis of state and CMS data. Note: Hospitals include nonfederal, short-term, acute care general hospitals. [End of table] The average size of hospitals in our study, as measured by patient operating expenses, varied across the three ownership groups. (See table 2.) On average, nonprofit hospitals were larger than for-profit hospitals. The pattern held in all five states but the magnitude of the difference varied. For example, in California, nonprofit hospitals were twice as large as for-profit hospitals, whereas in Texas, this difference was smaller. Table 2: Average Hospital Size as Measured by Patient Operating Expenses, 2003: California; Average patient operating expenses (in millions): For-profit: $71.7; Average patient operating expenses (in millions): Nonprofit: $143.4; Average patient operating expenses (in millions): State and local government: $141.2. Florida; Average patient operating expenses (in millions): For-profit: $90.8; Average patient operating expenses (in millions): Nonprofit: $181.8; Average patient operating expenses (in millions): State and local government: $229.3. Georgia; Average patient operating expenses (in millions): For-profit: $52.7; Average patient operating expenses (in millions): Nonprofit: $91.8; Average patient operating expenses (in millions): State and local government: $72.4. Indiana; Average patient operating expenses (in millions): For-profit: $62.1; Average patient operating expenses (in millions): Nonprofit: $116.1; Average patient operating expenses (in millions): State and local government: $47.6. Texas; Average patient operating expenses (in millions): For-profit: $73.9; Average patient operating expenses (in millions): Nonprofit: $112.9; Average patient operating expenses (in millions): State and local government: $43.0. Source: GAO analysis of state and CMS data. Note: Hospitals include nonfederal, short-term, acute care general hospitals. [End of table] Hospitals' Qualifications for Federal and State Tax-exempt Status: Hospitals may be extended a federal tax exemption by IRS if they meet the Internal Revenue Code's qualifications for charitable organizations under section 501(c)(3).[Footnote 10] Hospitals that qualify for nonprofit status are exempt from federal income taxes and typically receive other advantages, including access to charitable donations-- which are tax deductible for the individual or corporate donor--and tax- exempt bond financing. To qualify for federal tax-exempt status, a hospital must demonstrate that it is organized and operated for a "charitable purpose," that no part of its net earnings inure to the benefit of any private shareholder or individual, and that it does not participate in political campaigns on behalf of any candidate or conduct substantial lobbying activities.[Footnote 11] Before 1969, IRS required hospitals to provide charity care to qualify for tax-exempt status.[Footnote 12] Since then, however, IRS has not specifically required such care, as long as the hospital provides benefits to the community in other ways. This "community benefit" standard came into existence with an IRS ruling, which concluded that a hospital's operation of an emergency room open to all members of the community without regard to ability to pay promoted health in a way consistent with other activities--such as advancement of education and religion--that qualify other organizations as charitable.[Footnote 13] In addition, the 1969 ruling identified other factors that might support a hospital's tax-exempt status, such as having a governance board composed of community members and using surplus revenue to improve facilities, patient care, medical training, education, and research. Nonprofit hospitals may also receive exemptions from state and local income, property, and sales taxes, which, in some cases, are of greater value than the federal income tax exemption. Some states have defined community benefits for nonprofit hospitals, but their statutes vary considerably in their specificity and scope. Appendix II provides more information on statutory definitions of community benefits in the states we reviewed. Government Payments for Uncompensated Care and Other Costs: Hospitals may receive direct payments from different government sources to help cover their unreimbursed costs, including those for charity care, bad debt, and low-income patients. For example, Medicare and Medicaid make payments to hospitals that serve a disproportionate share of low-income patients under their respective disproportionate share hospital (DSH) programs. Medicare bad debt reimbursement partially reimburses hospitals for bad debt incurred for Medicare patients. Other state payments may also be available to hospitals, although their specific types vary widely. For example, hospitals may receive payments from special revenues such as tobacco settlement funds, uncompensated care pools that are funded by provider contributions, and payment programs targeted at certain services such as emergency services. (See app. III for more information on payments for uncompensated care and other costs.) Burden of Providing Uncompensated Care Varied among Hospital Groups, but Burden Was Generally Concentrated in a Small Number of Hospitals: In our review of hospitals' provision of uncompensated care in five states, we analyzed cost data from two perspectives--namely, each hospital group's percentage of (1) total uncompensated care costs in a state and (2) patient operating expenses devoted to uncompensated care. The former relationship showed hospitals' uncompensated care costs in dollars, aggregated by groups; whereas the latter relationship showed hospitals' uncompensated care costs as a proportion of their operating expenses, thereby accounting for differences in hospital number and size among the hospital groups. In general, government hospitals, as a group, accounted for the largest percentage of total uncompensated care costs and devoted the largest share of patient operating expenses to uncompensated care costs. The uncompensated care cost burden was not evenly distributed within each hospital group but instead was concentrated in a small number of hospitals. Government Hospitals Generally Accounted for the Largest Percentage of the Uncompensated Care Costs in States Reviewed: Government hospitals, as a group, accounted for the largest percentage of the total uncompensated care costs in three of the five states-- California, Georgia, and Texas. Nonprofit hospitals, as a group, accounted for the largest percentage of the uncompensated care costs in Florida and Indiana. For-profit hospitals, as a group, provided 20 percent or less of total uncompensated care costs in each state we reviewed. (See table 3). Table 3: Total Uncompensated Care Costs Incurred by Hospitals Reviewed, by State, 2003: California; Total uncompensated care costs (in millions): $2,307; Nonprofit (percent of total): 34%; For-profit (percent of total): 9%; State and local government (percent of total): 57%. Florida; Total uncompensated care costs (in millions): $1,561; Nonprofit (percent of total): 46%; For-profit (percent of total): 20%; State and local government (percent of total): 34%. Georgia; Total uncompensated care costs (in millions): $830; Nonprofit (percent of total): 43%; For-profit (percent of total): 10%; State and local government (percent of total): 47%. Indiana; Total uncompensated care costs (in millions): $342; Nonprofit (percent of total): 79%; For-profit (percent of total): 3%; State and local government (percent of total): 17%. Texas; Total uncompensated care costs (in millions): $2,101; Nonprofit (percent of total): 39%; For-profit (percent of total): 18%; State and local government (percent of total): 43%. Source: GAO analysis of state and CMS data. Note: Hospitals include nonfederal, short-term, acute care general hospitals. [End of table] In each of the five states, the nonprofit hospital groups accounted for a larger percentage of total uncompensated costs compared with the For- profit hospital groups. This difference was due, in part, to the larger number of nonprofit hospitals and their larger size relative to the for- profit hospitals. For example, in California, the nonprofit group's percentage of total uncompensated care costs was almost four times higher than that of the for-profit group, but this is not surprising, as nonprofit hospitals outnumbered for-profit hospitals almost 2 to 1 and were twice the size in patient operating expenses. Government Hospital Groups Generally Devoted Largest Share of Patient Operating Expenses to Uncompensated Care, but Shares Varied across States: In four of the five states reviewed, government hospitals devoted substantially larger shares, on average, of their patient operating expenses to uncompensated care than did nonprofit and for-profit hospitals.[Footnote 14] (See fig. 2.) In those four states, the differences in average percentages between the government hospital groups and the nonprofit hospital groups ranged from about 4.3 percentage points in Georgia to 11.3 percentage points in Texas. In contrast, in the fifth state, Indiana, the nonprofit hospital group devoted the largest share, on average, of patient operating expenses to uncompensated care. Between the nonprofit and for-profit hospital groups, the nonprofit hospitals' average percentages were greater in four of the five states--ranging from 1.2 percentage points greater in Florida to 2.3 percentage points greater in Indiana. In contrast, in the fifth state, California, the nonprofit group's average percentage was similar to that of the for-profit group. Figure 2: Average Percent of Patient Operating Expenses Devoted to Uncompensated Care, by Hospital Ownership Type, 2003: [See PDF for image] Notes: The average percent of patient operating expenses devoted to uncompensated care for a hospital ownership group is calculated by dividing the sum of uncompensated care costs for hospitals in that group by the sum of the group's total patient operating expenses. Hospitals include nonfederal, short-term, acute care general hospitals. [End of figure] The five states varied in their hospitals' shares of patient operating expenses devoted to uncompensated care, ranging from an average 4.1 percent for all Indiana hospitals to an average 8.3 percent for Texas hospitals. (See table 4.) Similar state-to-state variation found in other studies was due, in part, to differences in states' proportions of uninsured populations, variation in Medicaid eligibility or payment levels, and the presence of state programs that provide health insurance to low-income uninsured individuals.[Footnote 15] Specifically, prior research showed that hospitals located in states with more uninsured individuals and hospitals in states with relatively more eligibility-restricted Medicaid programs may have higher levels of uncompensated care. Our data are consistent with these studies' findings on the uninsured. For example, in our five-state review, Texas had the highest percentage of uninsured--25 percent--and the highest share, on average, of patient operating expenses devoted to uncompensated care, whereas Indiana had the lowest percentage of uninsured--13 percent--and the lowest average share. Table 4: Average Percentage of Patient Operating Expenses Devoted to Uncompensated Care, by State, 2003: California; Average percentage of patient operating expenses devoted to uncompensated care: 5.6%. Florida; Average percentage of patient operating expenses devoted to uncompensated care: 6.4%. Georgia; Average percentage of patient operating expenses devoted to uncompensated care: 8.2%. Indiana; Average percentage of patient operating expenses devoted to uncompensated care: 4.1%. Texas; Average percentage of patient operating expenses devoted to uncompensated care: 8.3%. Source: GAO analysis of state and CMS data. Notes: We calculated the average percent of patient operating expenses devoted to uncompensated care for each state by dividing the sum of uncompensated care costs for hospitals in the state by the sum of the hospitals' total patient operating expenses in the state. Hospitals include nonfederal, short-term, acute care general hospitals. [End of table] For Each Hospital Group, Uncompensated Care Costs Were Concentrated in a Small Number of Hospitals: For each group, uncompensated care costs were concentrated in a small number of hospitals. We observed this pattern when examining the percentages of patient operating expenses devoted to uncompensated care costs as well as hospitals' shares of total uncompensated care costs in a state. For the three hospital ownership groups, we ranked hospitals according to their share of patient operating expenses devoted to uncompensated care. We found that, for all three hospital groups, the top quarter of hospitals devoted substantially greater percentages of their patient operating expenses to uncompensated care, on average, compared with the bottom quarter of hospitals. (See fig. 3.) For example, in California's nonprofit hospital group, the top quarter of hospitals devoted an average of 7.2 percent compared with 1.4 percent for the bottom quarter of hospitals. Similarly, in Florida's government hospital group, the top quarter of hospitals devoted an average 19.6 percent compared with an average 5.2 percent for the bottom quarter of hospitals. From state to state, the difference in ranges between top and bottom quarters was also substantial. For example, in Indiana's government group, the average share of operating expenses devoted to uncompensated care for hospitals in the top quarter was about 3 times larger than for those in the bottom quarter; whereas in California, the average share for the top quarter of hospitals was almost 13 times higher than that of the bottom quarter. Figure 3: Average Share of Patient Operating Expenses Devoted to Uncompensated Care for Hospitals Ranked in Top and Bottom Quarters, by Ownership Type, 2003: [See PDF for image] Notes: Hospitals were ranked by percentage of patient operating expenses devoted to uncompensated care. The average percent of patient operating expenses devoted to uncompensated care for a hospital ownership group is calculated by dividing the sum of uncompensated care costs for hospitals in that group by the sum of the group's total patient operating expenses. Hospitals include nonfederal, short-term, acute care general hospitals. [End of figure] When examining hospitals' shares of total uncompensated care costs in a state, we found that uncompensated care costs remained concentrated in a disproportionately small number of hospitals. Specifically, each state's top quarter of hospitals accounted for a disproportionately large share of the state's uncompensated care costs. For example, in Texas, the top quarter of hospitals accounted for about 50 percent of total uncompensated care costs, yet accounted for only 18 percent of the total beds. (See table 5). Moreover, in Texas, six major government teaching institutions accounted for 34 percent of total uncompensated care costs, which amounted to over half of the contribution of the hospitals in the top quarter. This pattern was also true for California, Florida, and Georgia. For example, in California, 13 major teaching hospitals accounted for 42 percent of total uncompensated care costs.[Footnote 16] In contrast, in Indiana, total uncompensated care costs were distributed more evenly across a greater number of hospitals. Table 5: Percentage of Total Uncompensated Care Costs in a State for Hospitals Ranked in Top Quarter, 2003: State: California; Percentage of state's total uncompensated care: 68%; Percentage of states' hospital beds: 25%. State: Florida; Percentage of state's total uncompensated care: 47%; Percentage of states' hospital beds: 22%. State: Georgia; Percentage of state's total uncompensated care: 39%; Percentage of states' hospital beds: 19%. State: Indiana; Percentage of state's total uncompensated care: 21%; Percentage of states' hospital beds: 14%. State: Texas; Percentage of state's total uncompensated care: 50%; Percentage of states' hospital beds: 18%. Source: GAO analysis of state and CMS data. Notes: Hospitals were ranked by percentage of patient operating expenses devoted to uncompensated care. Hospitals include nonfederal, short-term, acute care general hospitals. [End of table] Several factors explain which hospitals were likely to be in their group's top and bottom quarters. For example, in our five-state analysis, we found that whether a hospital was a teaching institution was an important predictor of whether it would be in the top quarter of a state's government hospital group. Hospitals that had teaching programs were more likely to be in the top quarter of a government hospital group. In contrast, teaching status was not an important predictor for either the nonprofit or for-profit hospital groups' top quarter. For nonprofits, hospitals in rural areas were more likely to be in the top quarter than hospitals located in urban areas. Other factors that were outside the scope of this study, such as differences in the proportion of uninsured populations in the hospital market, may have also influenced the likelihood of a hospital's inclusion in the top or bottom quarter. Hospitals Reported Providing a Wide Range of Other Community Benefits: In addition to providing uncompensated care, hospitals may provide other services to their communities for which they are not reimbursed. In our review of hospitals' Web sites and reports about community benefits--published documents specifying the types and value of services hospitals provide to communities--we found that, regardless of ownership status, hospitals reported providing a wide range of community benefits.[Footnote 17] Variations in the types of community benefits hospitals reported providing could be explained by differences in the community benefits hospitals chose to provide as well as by variations in the applicability, specificity, and breadth of state requirements. Certain hospital industry guidance defines community benefits as the unreimbursed goods and services hospitals provide that address their communities' health needs, including health education, screening, and clinic services, among others. Consistent with this industry definition, we found through our review of reports and Web sites that hospitals reported providing similar types of services, including: * community health education such as parenting education, smoking cessation, fitness and nutrition, health fairs, and diabetes management; * health screening services such as screening for high cholesterol, cancer, and diabetes; * clinic services, including clinics targeted to specific groups in the community, such as indigent patients; * medical education for physicians, nurses, and other health professionals; * financial contributions, including cash donations and grants, to community organizations; * coordination of community events and in-kind donations--such as food, clothing, and meeting room space--to community organizations; and: * hospital facility and other infrastructure improvements. Community health education and health screenings were listed by most of the reports and Web sites we reviewed. Clinic services, support groups, community event coordination, cash contributions to charities, and medical education for health professionals were listed by over half of the reports we reviewed.[Footnote 18] Because of the wide variation in hospitals' reporting of community benefits, we were not able to discern clear patterns in the provision of these benefits across hospital ownership groups. The variation could be explained by differences in the community benefits hospitals chose to provide as well as by variations in the applicability, specificity, and breadth of state requirements. Specifically, the five states reviewed require all hospitals to report financial data, including data on the cost of charity care they provide. However, as shown in table 6, California, Indiana, and Texas also have statutory requirements for nonprofit hospitals to develop plans for meeting their communities' health needs and to report annually on the types and value of the community benefits they provide.[Footnote 19] Of these three states, only Texas and Indiana require nonprofit hospitals to report using standardized forms and have the explicit statutory authority to impose fines for noncompliance as part of the requirements.[Footnote 20] The Texas form is more specific, as it includes line-items that capture the hospitals' unreimbursed costs associated with providing traditionally "unprofitable" health services such as trauma care and community clinics, education of medical professionals, medical research, and cash and in-kind donations made by the hospital to local charities. Indiana's form provides nonprofit hospitals more flexibility in delineating the types and value of their community benefits but includes supplementary guidance to nonprofit hospitals about what should be considered community benefits, including financial or in-kind support of public health programs, community-orientated wellness and health promotion programs, and outreach clinics in economically depressed communities. California has no form for annual community benefit reports but requires that hospitals classify the services provided into broad, statutorily defined categories, including cash and in-kind donations to public health programs, efforts to contain health care costs and enhance access, and services that help maintain a person's health. Table 6: Community Benefit Requirements for Nonprofit Hospitals: State: California[A]; Description of requirements: Maintain community benefit plans that include measurable objectives for meeting the community's needs within specified time frames and mechanisms to evaluate effectiveness. In addition, report annually on the plans, as well as the types and value of community benefits provided; Penalties for noncompliance: None explicitly authorized as part of requirements. State: Florida; Description of requirements: None; Penalties for noncompliance: Not applicable. State: Georgia[B]; Description of requirements: None; Penalties for noncompliance: Not applicable. State: Indiana[C]; Description of requirements: Maintain and report annually on community benefit plans that include measurable objectives for meeting the community's health care needs within a specified time frames, evaluation strategies, and a budget. In addition, must describe the types and value of any additional community benefits; Penalties for noncompliance: Fines explicitly authorized as part of requirements for failure to make annual report. State: Texas[D]; Description of requirements: Maintain and report annually on community benefit plans that include measurable objectives for meeting the community's health care needs within specified timeframes, mechanisms for evaluating effectiveness, and a budget. In addition, must describe the types and value of community benefits provided; At a minimum, hospitals are required to provide: (1) charity and government-sponsored indigent care at a level that is reasonable in relation to community needs, the available resources of the hospital, and the tax-exempt benefits received; (2) charity and government-sponsored indigent health care equal to 100 percent of state tax-exempt benefits; or; (3) charity care and other community benefits equal to at least 5 percent of net patient revenue, provided that charity care and government-sponsored indigent health care are provided in an amount equal to at least 4 percent; Penalties for noncompliance: Fines explicitly authorized as part of requirements for failure to make annual report; Hospitals that fail to provide the required community benefits must be reported annually to attorney general and comptroller. Source: GAO analysis. [A] CAL. HEALTH & SAFETY CODE §§ 127345, 127350, and 127355 (2004). [B] Georgia requires all "hospital authorities," which create or operate nonprofit hospitals, to submit "community benefit reports" that disclose the cost of charity and indigent care provided. GA. CODE ANN. § 31-7-90.1 (2004). However, this information is otherwise required of hospitals in all groups in Georgia as part of financial reporting requirements. GA. CODE ANN. § 31-6-70 (2004). [C] IND. CODE § 16-21-9-4 - 16-21-9-8 (2004). [D] TEX. HEALTH & SAFETY CODE ANN. §§ 311.043 - 311.047 (2004). [End of table] According to state officials or state hospital association representatives in the five states we reviewed, for-profit and government hospitals are not required to report on the community benefits they provide outside of the requirements to report financial data, including data on the cost of charity care they provide. However, as we found through our review, some of these hospitals report publicly--for promotional purposes--on the community benefits they provide, either through published reports or by posting general information on their Web sites. Moreover, the three states with community benefit reporting requirements--California, Indiana, and Texas--conduct limited monitoring of nonprofit hospitals' community benefit reports. For example, according to officials from state agencies, none of the three states conducts audits of nonprofit hospitals' self-reported community benefits information, although Texas reviews the reports to ensure that "reasonable" types of services are listed as community benefits. In addition, these states do not routinely use the data collected through community benefit reports to review hospitals' tax-exempt status. Concluding Observations: Our comparison of the hospital ownership groups' uncompensated care costs, as a percentage of patient operating expenses, was instructive. Differences between the nonprofit and for-profit groups were often small when compared with the substantial differences between the government group and the other two groups. Moreover, the burden of uncompensated care costs was not evenly distributed among hospitals, which meant that a small number of nonprofit hospitals accounted for substantially more of the uncompensated care burden than did others receiving the same tax preference. As for the other community benefits hospitals reported providing, we were not able to discern a clear distinction among the government, nonprofit, and for-profit hospital groups. Hospitals in the five states reported conducting a variety of activities, which the hospitals themselves considered community benefits. We were unable to assess the value of these benefits or make systematic comparisons between hospitals or across states. These observations illustrate a larger point that I and others raised at the hearing last month--namely, that current tax policy lacks specific criteria with respect to tax exemptions for charitable entities and detail on how that tax exemption is conferred. If these criteria are articulated in accordance with desired goals, standards could be established that would allow nonprofit hospitals to be held accountable for providing services of benefit to the public commensurate with their favored tax status. Mr. Chairman, this concludes my prepared statement. I will be happy to answer questions you or the other Committee Members may have. Contact and Acknowledgments: For further information regarding this testimony, please contact A. Bruce Steinwald at (202) 512-7101. Kristi Peterson, Thomas Walke, Joanna Hiatt, Kelly DeMots, Mary Giffin, Emily Rowe, Craig Winslow, and Hannah Fein contributed to this statement. [End of section] Appendix I: Scope and Methodology: To examine the provision of uncompensated care by the three hospital ownership groups, we obtained 2003 uncompensated care data from five states--California, Florida, Georgia, Indiana, and Texas. We obtained all other data, such as cost-to-charge ratios,[Footnote 21] patient operating expenses,[Footnote 22] and all descriptive statistics, from 2002 and 2003 Medicare hospital cost reports.[Footnote 23] We selected the five states because they represented geographically diverse areas; had a number of hospitals in each ownership group sufficient to make comparisons; and collected hospital-specific uncompensated care data, which not all states maintain.[Footnote 24] The 2003 state uncompensated care data and 2002 and 2003 Medicare hospital cost reports were the most recent available at the time of our analysis. We also interviewed health officials from all five states as well as officials from the Centers for Medicare & Medicaid Services (CMS), the American Hospital Association, and the Federation of American Hospitals. We limited our analysis to nonfederal, short-term, acute care general hospitals for which a cost report was available.[Footnote 25] This analysis included critical access hospitals that provide general acute care. Our study included about 92 percent of nonfederal, short-term, acute care hospitals in the five states. We defined uncompensated care as the sum of charity care and bad debt costs as reported in the state data. To determine uncompensated care costs, we multiplied uncompensated care charges by a hospital-specific cost-to-charge ratio. Although specific definitions of charity care varied, states generally defined it as charges for patients deemed unable to pay all or part of their bill, less any payments made by, or on behalf of, that specific patient. States generally defined bad debt as the uncollectible payment that a patient is expected to, but does not pay. Our definition of uncompensated care does not include any contractual allowances or cost shortfalls.[Footnote 26] In addition, we did not subtract any charity care-specific block grants or donations a hospital may receive, as this information was not available for all states. We analyzed uncompensated care cost data from two perspectives---- namely, each hospital ownership[Footnote 27] group's percentage of (1) total uncompensated care costs in a state, and (2) average patient operating expenses devoted to uncompensated care. To examine factors that could explain differences in the provision of uncompensated care by hospital ownership groups, we examined certain hospital characteristics including a hospital's size, teaching status, and location. We used patient operating expenses to measure hospital size. For teaching status, we defined major teaching hospitals as those hospitals having an intern/resident-to-bed ratio of 0.25 or more and minor teaching hospitals as those having an intern/resident-to-bed ratio greater than 0 and less than 0.25. We defined a hospital as urban if it was located in a metropolitan statistical area and as rural if it was not located in a metropolitan statistical area. We supplemented our analysis with a review of the literature to determine other factors that could explain differences in the provision of uncompensated care by hospital ownership groups. We assessed the reliability of the hospital Medicare cost reports and the reliability of state uncompensated care cost data from California, Florida, Georgia, Indiana, and Texas in several ways. First, we performed tests of data elements. For example, we examined the values for uncompensated care costs and patient operating expenses to determine whether these data were complete and reasonable. We also verified that the dollar amount of uncompensated care in the 2003 data was consistent with the amount in 2002. Second, we reviewed existing information about the data elements. For example, we compared descriptive statistics we calculated from the Medicare hospitals cost reports with statistics published by CMS. Third, we interviewed state and agency officials knowledgeable about the data in our analyses and knowledgeable about hospital uncompensated care costs. We determined that CMS and all five states performed quality assurance tests on the data before releasing them. Overall, we determined that the data we used in our analyses were sufficiently reliable for our purposes. To examine hospitals' provision of community benefits other than uncompensated care, we reviewed 21 hospital reports and Web sites for information about such benefits in five states. Specifically, we reviewed 12 publicly available reports about the community benefits provided by nonprofit and for-profit hospitals and 3 reports for For- profit hospital systems representing multiple hospitals. We also reviewed 6 government hospitals' Web sites to determine the extent to which they publicized the provision of services that are generally considered community benefits. We also examined laws in five states regarding community benefit requirements for nonprofit hospitals, reviewed the literature, and interviewed state officials and hospital association representatives. We conducted our work from February 2005 through May 2005 in accordance with generally accepted government auditing standards. [End of section] Appendix II: Statutory Definitions of Community Benefits in the Five States Reviewed: Table 7 summarizes the statutory definitions of community benefits for nonprofit hospitals in the states we reviewed. We found that the statutes vary considerably in their specificity and scope. In addition, of the five states we reviewed, only the Texas statute contains an explicit link between the statutory definition of community benefits and hospitals' qualifications for state tax exemptions. Table 7: Statutory Definitions of Community Benefit for Purposes of Requirements Specific to Nonprofit Hospitals: State: California[A]; Statutory definition of community benefit: Hospital activities to address community needs and priorities through disease prevention and improvement of health status, including, but not limited to: (1) health care services, rendered to vulnerable populations (e.g., charity care and unreimbursed costs of providing services to uninsured and underinsured); (2) health promotion, prevention services, adult day care, child care, medical research and education, nursing and other professional training, home delivered meals, aid to the homeless, and outreach clinics; (3) financial or in- kind support of public health programs; (4) donation of funds, property, or other resources for a community priority; (5) health care cost containment; (6) enhancement of access to health care; (7) services offered without regard to profitability to meet a community need; and; (8) goods and services to help maintain a person's health; Cross-reference to tax exemption in community benefit provisions: No provisions explicitly cross-referencing definitions and related requirements to tax exemption. State: Florida; Statutory definition of community benefit: Not defined; Cross-reference to tax exemption in community benefit provisions: Not applicable. State: Georgia[B]; Statutory definition of community benefit: Not defined, but community benefit reporting requirement refers to charity and indigent care; Cross-reference to tax exemption in community benefit provisions: No provisions explicitly cross-referencing definitions and related requirements to tax exemption. State: Indiana[C]; Statutory definition of community benefit: Unreimbursed cost to hospitals of providing charity care, government- sponsored indigent care, donations, education, government-sponsored program services, research, and subsidized health services. Does not include hospital taxes or other government assessments; Cross-reference to tax exemption in community benefit provisions: No provisions explicitly cross-referencing definitions and related requirements to tax exemption.[D]. State: Texas[E]; Statutory definition of community benefit: Unreimbursed cost to hospitals of providing charity care, government- sponsored indigent health care, donations, education, government- sponsored program services, research, and subsidized health services, but not hospital taxes or other government assessments; Cross-reference to tax exemption in community benefit provisions: Numerous provisions cross-referencing definition of community benefit and related requirements to tax exemption. Source: GAO analysis. [A] CAL. HEALTH & SAFETY CODE §§ 127340 and 127345(c) (2004). [B] GA. CODE ANN. § 31-7-90.1 (2004). [C] IND. CODE ANN. §§ 16-18-2-64.5 and 16-21-9-1 (2004). [D] There are no provisions explicitly cross-referencing community benefits to nonprofit hospitals' tax exemption, but hospital-owned physician offices or practices, or other property not substantially related to inpatient facilities, must provide or support charity care or community benefits, as it is defined above, to qualify for property tax exemption. IND. CODE ANN. § 6-1.1-10-18.5 (2004). [E] TEX. HEALTH & SAFETY CODE ANN. §§ 311.042 and 311.045 (2004). [End of table] [End of section] Appendix III: Government Payments for Uncompensated Care and Other Unreimbursed Costs: Hospitals may receive direct payments from different government sources to help cover their unreimbursed costs. Such payments may include special Medicare and Medicaid payments, known as disproportionate share hospital (DSH) payments, Medicare bad debt reimbursement, and other state payments. Medicare DSH: The Medicare DSH adjustment provides payments to hospitals that serve a disproportionate share of low-income patients. The Congress mandated this adjustment in 1986 to address the concern that hospitals that serve such patients have higher Medicare costs per case because they have higher overhead and labor costs and their patients are in poorer health with more complications and secondary diagnoses. Hospitals qualify for the Medicare DSH adjustment based on their low-income patient share.[Footnote 28] The low-income patient share is computed as the percentage of a hospital's Medicare inpatient days attributable to patients that are eligible for both Medicare part A and Supplemental Security Income[Footnote 29] plus the percentage of total inpatient days attributable to patients eligible for Medicaid, but not Medicare part A. For hospitals that qualify for a DSH adjustment, their actual adjustment is based on several factors, including the number of acute care beds, number of patient days for low- income patients, and location (rural or urban). See table 8 for Medicare DSH payments in 2003 to the hospitals in the selected states we analyzed. Table 8: Medicare DSH Payments to Hospitals Reviewed, 2003 (in millions): State: California; Medicare DSH payment to hospitals (in millions): $1,122. State: Florida; Medicare DSH payment to hospitals (in millions): $486. State: Georgia; Medicare DSH payment to hospitals (in millions): $209. State: Indiana; Medicare DSH payment to hospitals (in millions): $94. State: Texas; Medicare DSH payment to hospitals (in millions): $637. Source: GAO analysis of state and CMS data. Note: Hospitals include nonfederal, short-term, acute care general hospitals. [End of table] Medicaid DSH: The Medicaid statute requires that states make DSH adjustments to the payment rates of certain hospitals treating large numbers of low-income and Medicaid patients. The Medicaid DSH adjustment was established by the Congress in 1981 and establishes broad guidelines for hospital eligibility to receive Medicaid DSH and for the methods used to compute the amount of payment. States have discretion in designating DSH hospitals and calculating adjustments for them.[Footnote 30] States also vary in terms of program rules and resource levels as well as the degree to which they target payments to different types of hospitals.[Footnote 31] Medicaid DSH is the largest source of financial support for hospital uncompensated care and is funded jointly by the states and the federal government. State approaches to financing the state portion of Medicaid DSH include obtaining funds from hospitals through provider taxes or intergovernmental transfers in order to establish the state's contribution required to obtain the federal match for Medicaid DSH funding. Therefore, it is not always possible to determine what portion of Medicaid DSH payments to individual hospitals is the net additional payment to the hospital. Medicare bad debt reimbursement: Medicare partially reimburses acute care hospitals for bad debts resulting from Medicare beneficiaries' nonpayment of deductibles and copayments after providers have made reasonable efforts to collect unpaid amounts. If a hospital can document that a Medicare patient is indigent, the hospital can then forgo collection efforts from the patient. Medicare pays hospitals 70 percent of their reimbursable bad debts, except critical access hospitals,[Footnote 32] for which it pays 100 percent of their reimbursable bad debts. See table 9 for total Medicare bad debt reimbursements in 2003 to the hospitals in the selected states we analyzed. Table 9: Medicare Bad Debt Reimbursements to Hospitals Reviewed, 2003: State: California; Medicare bad debt reimbursement to hospitals (dollars in millions): $160. State: Florida; Medicare bad debt reimbursement to hospitals (dollars in millions): $55. State: Georgia; Medicare bad debt reimbursement to hospitals (dollars in millions): $45. State: Indiana; Medicare bad debt reimbursement to hospitals (dollars in millions): $20. State: Texas; Medicare bad debt reimbursement to hospitals (dollars in millions): $78. Source: GAO analysis of state and CMS data. Note: Hospitals include nonfederal, short-term, acute care general hospitals. [End of table] Other state sources: Other state sources of payment to hospitals for uncompensated or unreimbursed care vary widely, and may include special revenues such as tobacco settlement funds, uncompensated care pools that are funded by provider contributions, and payment programs targeted at certain services such as emergency services. For example, Massachusetts has used a portion of the state's tobacco settlement fund to help cover uncompensated care costs. FOOTNOTES [1] GAO, Tax-Exempt Sector: Governance, Transparency, and Oversight Are Critical for Maintaining Public Trust, GAO-05-561T (Washington, D.C.: Apr. 20, 2005). [2] GAO, 21st Century Challenges: Reexamining the Base of the Federal Government, GAO-05-325SP (Washington, D.C.: February 2005). [3] The state and local government-owned hospitals in this statement refer to state-owned hospitals, such as those at state universities, and locally owned hospitals, such as county and city hospitals. In this statement we will refer to these as government hospitals. Federal hospitals, such as those operated by the Department of Veterans Affairs, are not included in this definition. [4] Reliable, hospital-specific data were not available nationwide. In addition, some states do not have sufficient diversity in hospital ownership to make comparisons for the purpose of this analysis; in particular, some states have very few for-profit hospitals. [5] To obtain uncompensated care costs, we multiplied hospitals' uncompensated care charges reported in the state data by hospital- specific, cost-to-charge ratios from Medicare hospital cost reports. These cost-to-charge ratios are specific to hospital costs and charges as a whole, not to Medicare costs and charges. [6] Patient operating expenses include those expenses incurred for patient care. They exclude such expenses as those incurred for operating a parking garage, gift shop, and certain other nonmedical expenses. [7] Cost, charge, and other data obtained from the states and other sources are for individual hospitals, even if a hospital is part of a larger hospital system. [8] We excluded 8 percent of the hospitals in the five states because certain key information, such as total patient operating expenses, was not available. [9] This total does not include critical access hospitals that provide general acute care. Critical access hospitals are small, rural hospitals that receive payment for their reasonable costs of providing inpatient and outpatient services to Medicare beneficiaries, rather than being paid fixed amounts under Medicare's prospective payment systems. By excluding critical access hospitals, which are numerous and small, we removed the effect that they would have on the distribution of hospitals by ownership group. [10] Section 501(c) specifies 28 types of entities that are eligible for tax-exempt status. Over 1.5 million entities have been recognized as exempt by IRS. [11] Charitable activities may include those that relieve the poor, distressed, or underprivileged; those that lessen the burdens of government; and those that promote social welfare. [12] See, for example, IRS Rev. Rul. 56-185, 1956-1 C.B. 202. [13] See IRS Rev. Rul. 69-545, 1969-2 C.B. 117. A revenue ruling is a formally published interpretation of tax law by the IRS upon which taxpayers are entitled to rely. [14] These results are consistent with studies showing a similar relationship. See L. Fishman, "What Types of Hospitals Form the Safety Net?" Health Affairs, vol. 16, no. 4 (July/August 1997); J. Mann, et al., "A Profile of Uncompensated Hospital Care, 1983-1995," Health Affairs, vol. 16, no. 4 (July/August 1997); and S. Zuckerman, et al., "How Did Safety-Net Hospitals Cope in the 1990s?" Health Affairs, vol. 20, no. 4 (July/August 2001). [15] See G. Atkinson, W. Helms, and J. Needleman, "State Trends in Hospital Uncompensated Care," Health Affairs, vol. 16, no. 4 (July/ August 1997); L. Fishman, "What Types of Hospitals Form the Safety Net?" Health Affairs, vol. 16, no. 4 (July/August 1997); A. Davidoff, A. LoSasso, G. Bazzoli, and S. Zuckerman, "The Effect of Changing State Health Policy on Hospital Uncompensated Care," Inquiry, vol. 37 (Fall 2000); K. Thorpe, E. Seiber, and C. Florence, "The Impact of HMOs on Hospital-Based Uncompensated Care," Journal of Health Politics, Policy and Law, vol. 26, no. 3 (June, 2001); and GAO, Nonprofit Hospitals: Better Standards Needed for Tax Exemption, GAO/HRD-90-84 (Washington, D.C.: May 30, 1990). [16] We defined major teaching hospitals as those hospitals having an intern or resident-to-bed ratio of 0.25 or more and minor teaching hospitals as those having an intern or resident-to-bed ratio greater than 0 and less than 0.25. [17] To determine the types of community benefits hospitals reported providing, we reviewed 15 publicly available reports about community benefits for nonprofit and for-profit hospitals and six government hospitals' Web sites. [18] Our findings on the types of community benefits hospitals reported providing are consistent with our findings in GAO/HRD-90-84 and industry publications. [19] Georgia requires all "hospital authorities," which create or operate nonprofit hospitals, to submit "community benefit reports" that disclose the cost of charity and indigent care provided. GA. CODE ANN. § 31-7-90.1 (2004). However, this information is otherwise required of hospitals in all groups in Georgia as part of financial reporting requirements. GA. CODE ANN. § 31-6-70 (2004). [20] In Texas, for-profit and government hospitals receiving Medicaid DSH payments are generally required to meet the same community benefit reporting requirements as nonprofit hospitals. TEX. HEALTH & SAFETY CODE ANN. § 311.046(e) (2004). [21] These cost-to-charge ratios are specific to hospital costs and charges as a whole, not to Medicare costs and charges. [22] Patient operating expenses include those expenses incurred for patient care. They exclude such expenses as those incurred for operating a parking garage, gift shop, and certain other nonmedical expenses. [23] The reporting period of certain hospitals differed between the state data and the cost reports. Therefore, we combined the 2003 state data with the cost report, either 2002 or 2003, that best overlapped the state data's reporting period. [24] Reliable, hospital-specific data were not available nationwide. In addition, some states do not have sufficient diversity in hospital ownership to make comparisons for the purpose of this analysis; in particular, some states have very few for-profit hospitals. [25] Cost, charge, and other data obtained from the states and other sources are for individual hospitals, even if a hospital is part of a larger hospital system. [26] Contractual allowances are the difference between a hospital's full charges for a service and the payment it has agreed to accept for that service from a particular insurer. Cost shortfalls are the difference between the accepted payment for a service and the actual cost of that service, in the case that the payment is less than the cost. [27] In order to determine a hospital's ownership status, we compared its ownership from the state data (if available) to that from the Medicare cost report data. Where the two sources did not match, we used the 2002-2003 AHA Guide to confirm one of the sources as correct. If possible, we also confirmed ownership status using the hospital's Web site. [28] To qualify for Medicare DSH, a hospital must have a share of low- income patients that exceeds 15 percent. Alternately, large hospitals located in urban areas can qualify if more than 30 percent of their total net inpatient care revenue is for indigent care and comes from state and local governments (excluding Medicare and Medicaid funds). [29] Medicare Part A pays for inpatient hospital stays, care in a skilled nursing facility, hospice care, and some home health care. The Supplemental Security Income program makes payments to people with low income who are at least 65 or are blind or have a disability. [30] Congressional Research Service, Medicaid Disproportionate Share Payments (Washington, D.C.: 2005), 15. [31] Congressional Research Service, Medicaid Reimbursement Policy (Washington, D.C.: 2004), 36. [32] See GAO, Medicare: Modest Eligibility Expansion for Critical Access Hospital Program Should Be Considered, GAO-03-948 (Washington, D.C.: September 2003).