Uncompensated Care before and after Prospective Payment: The Role of Hospital Location and Ownership Cheryl I. Hultman Research was undertaken to determine the effects of hospital ownership, location, and Medicare's prospective payment system (PPS) on inpatient uncompensated care. A nonequivalent group design was used with repeated measures of uncompensated care (UNCC) on 137 system hospitals taken pre-and post-pps. Investorowned system hospitals demonstrated the largest increase in UNCC (37 percent) under the PPS. Results suggest that not-for-profit and investor-owned system hospitals are becoming more similar in levels of uncompensated care provided and that the PPS has had a negative effect on rural hospital profitability. BACKGROUND Concern has been expressed about the effect of cost-containment measures and the growth of system and investor-owned (IO) hospitals on access to health services for the poor. This concern is due to the fact that the cross-subsidies that have historically allowed hospitals to provide charity care without serious financial consequences have been The data reported here were obtained during a research project on the strategies, structures, and performance of multihospital systems conducted by Professor Stephen Shortell of the J. L. Kellogg School of Management and the Center for Health Services and Policy Research of Northwestern University. The project was funded under NCHSR Grant No. HS 5-159 and by the Robert Wood Johnson Foundation. Data analyses were partially supported by a Dissertation Year Grant from Northwestern University's Research Grants Committee. Address correspondence and requests for reprints to Cheryl I. Hultman, Ph.D., Research Health Scientist, Midwest Health Services Research and Development Field Program (151 H), Department of Veterans Affairs, Hines VA Hospital, Hines, IL 6141. This article, submitted to Health Services Research on July 13, 1989, was revised and accepted for publication on November 7, 199.
614 HSR: Health Services Research 26:5 (December 1991) gradually disappearing. It has been hypothesized that increasing hospital system involvement and the prospective payment system (PPS) under Medicare will have a negative effect on charity care; however, there has been little empirical evidence to support or reject such hypotheses. Past research that has dealt specifically with multihospital systems and access to care for the poor has not demonstrated much difference between system hospitals and independent voluntary hospitals. Ermann and Gabel's (1985) review of the multihospital system literature, for example, found no discernible differences between 1 system hospitals and matched independent hospitals in the percentages of revenue from Medicare and Medicaid (Sloan and Vraciu 1983), the availability of specific services, or diagnostic case mix (Pattison and Katz 1983). However, IO system hospitals were found to treat fewer charity patients than independent voluntary hospitals, although neither group serves substantial numbers of these patients (Ermann and Gabel 1985): the majority of such patients are treated by large urban public and teaching hospitals (Pattison and Katz 1983). More recent findings of Schlesinger, Bentkover, Blumenthal, et al. (1987) and Shortell et al. (1987) provide evidence that increasing competitive pressures and the growth of IO hospitals and multihospital systems may be accompanied by restrictions on access to hospital services for low-income patients. To date, research on the PPS does not suggest that it has caused access to deteriorate substantially. The one exception is patient dumping, which may have been exacerbated by PPS; however, national studies have not yet demonstrated a significant effect. The effect of PPS on the profitability of hospitals has been more obvious and has caused concern among providers and in Congress. While the first year of PPS was a profitable year for most hospitals, more recent studies indicate that Medicare operating margins have declined rapidly (Cassidy 1988). Rural hospital profit margins have been hurt especially by the wage index used in the PPS reimbursement formula. This differential has posed a severe problem for rural hospitals located near urban areas where they have to compete with urban hospitals for staff. Also, rural hospitals have a greater proportion of Medicare admissions and are therefore more dependent than urban hospitals on Medicare reimbursement (Tauke 1985). The research reported here examines the effects of the PPS and hospital ownership and location on the amount of inpatient uncompen-
Uncompensated Care 615 sated care that hospitals provide, and it also examines how hospital profitability has been affected by the PPS. METHODOLOGY SOURCES OF DATA The data were obtained from a national two-wave panel study of multihospital systems; the study was a joint effort conducted by the Center for Health Services and Policy Research (CHSPR) of Northwestern University and the J. L. Kellogg Graduate School of Management. The national study's first wave sample, conducted pre-pps in 1983-1984, consisted of 574 system hospitals representing eight systems, and ranging in size from very small (under ten hospitals) to very large (several hundred hospitals) (Shortell, Morrison, Hughes, et al. 1986). Of these eight systems, three were investor-owned and five were not-for-profit (NFP). The hospitals were located in 45 states and provided a good cross section for research purposes. There were 839 comparison hospitals in the first-wave sample. The second-wave sample, conducted post-pps in 1985-1986, was slightly larger, consisting of 57 system hospitals and 898 comparison hospitals. The response rates to the first wave were 94.6 percent (543) of the system hospitals and 65.7 percent (551) of the comparisons (Shortell, Morrison, Hughes, et al. 1986). The response rates for the second wave were similar but somewhat lower than the first-wave response rates. Ninety-two percent (524) of the system hospitals and 5.2 percent (451) of the comparison hospitals responded to the second-wave survey. The data used in this study contained information on hospital uncompensated care (bad debt and charity care) and the financial status of system-owned hospitals that responded to both waves of the CHSPR study. These data were collected directly from corporate offices and dealt with total revenue and total hospital uncompensated care (UNCC). This sample consisted of 77 (56 percent) IO and 6 (44 percent) NFP system hospitals. No market area competitors were included in this part of the national study. The variables uncompensated care and financial status were measured over time and across different hospital types, that is, IO system hospitals versus NFP system hospitals. Other variables examined included geographic location and hospital profitability. The geographic location variable, developed from Area Resource Files (DHHS 1985
616 HSR: Health Services Research 26:5 (December 1991) and 1987), included four categories: rural, collar county, urban, and central city. The profitability data included hospital operating margins and net income margins (income after federal income tax), as well as total revenue and uncompensated care values as a percentage of hospital total revenue. The research design used was a quasi experiment with a nonequivalent group design. In such experiments, there are treatments, outcome measurements, and experimental units, but use is not made of random assignment (Cook and Campbell 1979). In this design, the implementation of PPS was assumed to be the treatment, and outcomes included changes in hospital uncompensated care. The nonequivalent groups being considered were the IO and the NFP system hospitals. Both between-group and within-group changes over time were analyzed by analysis of variance (ANOVA) and analysis of covariance statistical techniques. Tukey B multiple comparison tests were also used in determining which groups were significantly different from each other at the.5 level. RESULTS As shown in Figure 1, the average percentage of uncompensated care increased by a statistically significant amount [F (1,96) = 31.1; p <.1 and the different slopes of the lines in Figure 1 illustrate what turned out to be a significant interaction between the hospital organization type and the UNCC variable. The interaction indicates that the different hospital organization types have had different trends in uncompensated care since the PPS began. The IO system hospitals experienced a greater increase than the NFP system hospitals; however, the 1 levels of uncompensated care stayed below the NFP levels of uncompensated care during both waves of the study. The results of the three-way ANOVA of UNCC by system type by location are presented in Figure 2. Although the numbers are small in some of the categories of Figure 2, the main effects of the location [F (3,11) = 4.97; p <.31 and system type [F (1,11) = 15.3; p <.] are statistically significant. It was found that rural system hospitals were significantly different from the other hospital groups at both waves of the study (N = 216 for 1983 and N = 16 for 1985), and they averaged the highest levels of UNCC during 1983 and 1985. Also apparent in Figure 2 are the statistically significant interac-
Uncompensated Care 617 Figure 1: Uncompensated Care by Organization Type a 1 SH 6 ~4 NFP SH 5 3 2 Px PPS POST PPS TIME The number of cases equaled 98 during each time frame. There were 43 not-for-profit system hospitals and 55 investor-owned system hospitals. tions of system type and location with UNCC. The difference in UNCC between wave 1 and wave 2 was also significant [F (1,11) = 42.2; p <.], reflecting the definite increase in its level over the course of the study. The sample hospital mean level of UNCC increased from 3.65 percent in 1983 to 5. percent in 1985. A comparison of the location and net income margin means of the sample hospitals from 1983 and 1985 suggests that central city system hospitals were the most profitable, while the rural system hospitals were the least profitable, on the average. Table 1 demonstrates that the sample rural system hospitals actually experienced a net decrease in their mean net income margins from 1983 to 1985. In summary, rural system hospital UNCC has increased since the PPS began. And, as alluded to earlier, the rural system hospitals of this study provided more uncompensated care with relatively less profit in comparison to hospitals in other locations; however, the location differences in net profits were not statistically significant. Additional analysis of American Hospital Association (AHA) annual hospital survey data collected in 1983 and 1985 support these results. As shown in Table 2, up to 1986, IO hospitals increased their
Q I IL ui U cn -o. a( Uż.>.._ U.. I I -. I I - I - a b ( ) ct ) DONn % NM U; o -Q oc. - r. > Cl U (L) C Cl)._ 1 U.. t3 k Q 2 - 'U- ^UU ' C)4 V 5-4 z..q HE:2 -o w~~ I I*I* ( I- * UI.- UI v) a a *^- U) lw ( cm onnn %
Mean Net Income Profit Margins of System Uncompensated Care 619 Table 1: Hospitals Percent Percent Percent Location * 1983 1985 Difference Central city 5.8 5.65 +.57 Urban 6.14 6.36 +.22 Collar county 6.1 6.27 +.26 Rural 4.44 3.75 -.69 Overall mean 5.53 5.66 +.13 *The number of the cases equaled 278, with the following location breakdown: 63 central, 87 urban, 73 collar, and 55 rural system hospitals. Table 2: AHA National Survey Data Uncompensated Care Hospital Type Percent Percent Percent 1983 1985 Change Not-for-profit Investor-owned 4.99 4.7 5.51 5.22 1.4 28.25 Rural Urban 5.86 6.53 6.48 6.93 1.6 6.1 Not-for-profit Bad debt Charity care 27.6* 17.9* Investor-owned Bad debt 58.7* Charity care 126.6* *Bad debt and charity care are measured separately in the AHA National Hospital Survey. levels of uncompensated care and total hospital charity care at a rate faster than the NFP hospitals. And the AHA sample of rural hospitals also increased their uncompensated care to a greater degree than did the sample urban hospitals. DISCUSSION LIMITATIONS Because the samples used in this analysis and in the larger CHSPR study were not randomly selected, the inferences that can be made to the population of U.S. hospitals are somewhat limited. Limitations of generalizability are due to two types of bias: selection and nonresponse
62 HSR: Health Services Research 26:5 (December 1991) bias. In this article, the IO system hospitals are overrepresented and NFP system hospitals are underrepresented. Also, comparisons of respondents to nonrespondents demonstrate that respondents to both waves of the study were more likely to be smaller hospitals, investorowned and from collar or urban locations. COMMENTS The results indicate that significant changes in uncompensated health care have occurred since implementation of the PPS. Sample system hospital uncompensated care increased by 37 percent between 1983 and 1985. If one assumes that the heterogeneity of the sample -that is, the large number of states and hospitals represented by the samplecancels out differences in uncompensated care due to state-level regulatory policies and hospital differences, then one can assume that the PPS, up until 1986, has been associated with an increase in UNCC. However, since UNCC has been found to represent about 75 percent bad debt and about 25 percent charity care (Wasserman 1986), the increase found in UNCC could have been due to an increase in bad debt with no accompanying change in charity care. Of interest is that the 1 system hospitals demonstrated the largest relative increase in UNCC between 1983 and 1985 compared to the NFP system hospitals. However, NFP system hospitals still provided more UNCC than IO system hospitals did during 1983 and 1985. Results indicate that between 1983 and 1985 the IO system hospitals were coming closer to the NFP system hospitals in terms of UNCC. Also, rural location and NFP system affiliation were related to the highest average levels of UNCC in both waves of the study. And the trends in UNCC from 1983-1985 varied more with hospital location than with hospital organizational type. The sample rural system hospitals also demonstrated the least profits after taxes of all of the hospital location categories; in fact, they sustained an average net loss in profits from 1983 to 1985. These results confirm previous reports that rural hospitals are being more negatively affected by the PPS than was anticipated (Rolph and Lindsey 1986). An additional contributing factor to the low profits of rural hospitals could be the nation's poor farm economy in recent years. Also, another drawback for rural hospitals is that they are more likely to be sole community providers unable to shift patients and thereby increase profits. Congress has recently become more aware of the financial difficulties facing rural hospitals. As a result, the fiscal year 1987 budget
Uncompensated Care 621 reconciliation law narrowed the difference in the PPS payments between urban and rural hospitals (Merrill 1987). IMPLICATIONS The present findings have implications for theories regarding the effects of the PPS and hospital ownership differences on uncompensated care. In contrast to the theory that the PPS would have little or no effect on uncompensated care, the data suggest that a statistically significant increase in uncompensated care has taken place since the PPS was implemented. Also, ANOVA analyses have demonstrated that hospital location is more important than system status as a determinant of uncompensated care. CONCLUSIONS Theories or assumptions about the differences in care for the poor between NFP and IO system hospitals should be reevaluated. Results suggest that these two hospital types are becoming more like each other in terms of uncompensated care. The results indicate that rural hospitals deserve the attention of policymakers. The rural/urban differences found substantiate claims that rural hospitals have made regarding the built-in inequities of the PPS, although it is also likely that the depressed farm economy has contributed to a relative increase in hospital bad debt and to the need for charity care in the rural setting. Additional areas for research should include (1) the specific benefits that NFP system hospitals provide over IO system hospitals, such as teaching and research; and (2) the examination of rural hospital profitability after recent PPS payment formula adjustments. Finally, this research highlights the need for more explicit measures of charity care by hospitals. If such measures were developed and implemented, it would be possible to evaluate more accurately the effects of various federal and state policies on the access by the poor to health care services. ACKNOWLEDGMENTS The author acknowledges the helpful comments of Professor Stephen Shortell and Kendon Conrad during the preparation of this manuscript.
622 HSR: Health Services Research 26:5 (December 1991) REFERENCES Cassidy, J. "Protect Medicare '9 Campaign Urges Adequate Hospital Payments." Health Progress 69, no. 1 (December 1988): 24-27. Cook, T. D., and D. T. Campbell. Quasi-Experimentation. Boston: Houghton Mifflin Co., 1979. Department of Health and Human Services. Area Resource Files, 1985, 1987. Washington, DC: DHHS, 1985, 1987. Ermann, D., and J. Gabel. "The Changing Face of American Healthcare: Multi-hospital Systems, Emergency Centers and Surgery Centers." Medical Care 23, no. 5 (1985): 41-2. Merrill, T. "Rural Issues Move Up on Congress's Agenda." Hospitals 61 (2 May 1987): 19-21. Pattison, R. V., and H. M. Katz. "Investor-owned and Not-for-profit Hospitals." New EnglandJournal of Medicine 39, no. 6 (1983): 347-53. Rolph, E., and P. Lindsey. "Medicare's Prospective Payment System: Healthcare Community Reaction and Perceptions." Prepared for Health Care Financing Administration and the U.S. Department of Health and Human Services. Washington, DC, 1986. Schlesinger, M., J. Bentkover, D. Blumenthal, R. Musacchio, and J. Willer. "The Privatization of Health Care and Physicians' Perceptions of Access to Hospital Services." Milbank Memorial Fund Quarterly 65, no. 1 (1987): 25-58. Shortell, S. M., E. M. Morrison, B. Friedman, and S. L. Hughes. A Study of the Strategies, Structures and Performance of Multihospital Systems, a NCHSR Assessment. Grant no. HS 5-159. Washington, DC: National Center for Health Services Research, December 31, 1987. Shortell, S. M., E. M. Morrison, S. L. Hughes, B. Friedman, J. Coverdill, and L. Berg. "The Effects of Hospital Ownership on Nontraditional Services." Health Affairs 5, no. 4 (1986): 97-111. Sloan, F. A., and R. A. Vraciu. "Investor-owned and Not-for-profit Hospitals: Addressing Some Issues." Health Affairs 2 (Spring 1983): 25-37. Tauke, T. J. "Rural Hospitals Suffer PP Inequities." Business and Health 2, no. 6 (1985): 37-41. Wasserman, J. "The New Jersey DRG System and Uncompensated Care: Some Empirical Results." Bulletin of the New York Academy of Medicine 2 (July-August 1986): 63-37.