Journal of Health Care Finance

Journal of Health Care Finance (USPS 048-090) Indexing: JHCF is indexed in Academic (ISSN 1078-6767) is published quarterly by Aspen Search/CD-ROM, the Business Periodicals Index, the Publishers, 76 Ninth Avenue, New York, NY 10011. Cumulative Index to Nursing & Allied Health © 2010 Aspen Publishers. All rights reserved. Literature (CINAHL), EMBASE, Health Source, Index www.aspenpublishers.com. Reproduction in whole Medicus, MEDLINE, MEDLARS, the UP-TO-DATE or in part without permission is strictly prohibited. Library/Health Services Management and Wilson Postmaster: Send address changes to Subscription Dept. Business Abstracts®, Wilson Business Abstracts Full IP. P.O. Box 3000, Denville, NJ 07834. Text® by The H.W. Wilson Company. Currently Subscription rate is $319 (plus postage and handling) available on CD-ROM and online via the WilsonWeb. per year in the United States and Canada (four issues), For more information, please visit http://www. payable in advance. The two-year subscription rate is hwwilson.com. $542, the three-year subscription rate is $766, and the “This publication is designed to provide accurate and cost of one issue is $96. Subscribers may specify any authoritative information in regard to the Subject Matter issue to begin the subscription. Subscribers in the covered. It is sold with the understanding that the United States and Canada: Address inquiries to publisher is not engaged in rendering legal, accounting, Fulﬁ llment, Aspen Publishers, 7201 McKinney Circle, or other professional service. If legal advice or other Frederick, MD 21704, or call 1-800-234-1660. To place expert assistance is required, the services of a an order, call 1-800-638-8437. Subscribers in all competent professional person should be sought.” countries other than the United States, Canada, and (From a Declaration of Principles jointly adopted by a Japan: Address inquiries to Aspen Publishers, c/o Swets Committee of the American Bar Association and a & Zeitlinger, P.O. Box 825, 2160 SZ Lisse, The Committee of Publishers and Associations.) Netherlands, telephone 31 252 435111, fax 31 252 Issue: Vol. 36, No. 4, 9900610013 415888. ISSN: 1078-6767 Permission requests: For information on how to Printed in the United States of America obtain permission to reproduce content, please go to the Aspen Publishers Web site at www.aspenpublishers. The paper used in this publication meets the com/permissions. requirements of the American National Standard for Purchasing reprints: For customized article reprints, Information Sciences—Permanence of Paper for Printed please contact Wright’s Reprints at 1-877-652-5295 or Library Materials, ANSI Z39.48-1992, effective with go to the Wright’s Reprints Web site at www.wrights Volume 13, Issue 3. reprints.com.

For more information resources, directories, articles, and a searchable version of Aspen Publishers’ full catalog, visit us at: www.aspenpublishers.com

1 2 3 4 5 Editorial Board

Editors Editors Emeritus

James J. Unland, MBA Judith J. Baker, PhD, CPA President Partner The Health Capital Group Resource Group, Ltd. Chicago, IL Dallas, TX

Paul Gibson, Publisher William O. Cleverley, PhD Joanne Mitchell-George, Senior Managing Editor Professor Elizabeth Venturo, Managing Editor Ohio State University Dom Cervi, Marketing Director Columbus, OH

Editorial Board Scott Becker, JD, CPA, Partner, Ross & Hardies, Kevin T. Ponton, President, SprainBrook Group, Chicago, IL Hawthorne, NY Dana A. Forgione, PhD, CPA, CMA, CFE, Janey Elizabeth Simpkin, President, The Lowell S. Briscoe Endowed Chair in the Business of Group,Inc., Chicago, IL Health, and Professor of Accounting, College of Elaine Scheye, President, The Scheye Group,Ltd., Business, University of Texas at San Antonio,TX. Chicago, IL Ellen F. Hoye, MS, Principal, Hoye Consulting Pamela C. Smith, PhD, Associate Professor, Services, Elmhurst, IL Department of Accounting, The University of David Koontz, Senior Director of Medical Texas at San Antonio, San Antonio, TX Informatics, Evanston Northwestern Jonathan P. Tomes, JD, Partner, Tomes & Dvorak, Healthcare, Evanston, IL Leawood, KS Daniel R. Longo, ScD, Professor and Director Mustafa Z. Younis, Professor of Health Economics & of Research, ACORN Network Co-Director, Finance, Jackson State University, School of Department of Family Medicine, Virginia Health Sciences, Department of Health Policy & Commonwealth University, Richmond, VA Management, Jackson, MS Roger C. Nauert, BS, MBA, JD, President, RCN Associates, Houston, TX JHCF 36:4, Summer 2010 Contents

iv From the Editor James J. Unland 1 The Effects of Certiﬁ cate of Need Regulation on Hospital Costs Patrick A. Rivers, Myron D. Fottler, and Jemima A. Frimpong 17 The Utilization of Hospitalists Associated with Compensation: Insourcing Instead of Outsourcing Health Care Doohee Lee and Andrew Sikula, Sr. 27 Hospital Productivity and Information Technology Steven R. Eastaugh 38 Project Measurement and Success: A Balanced Scorecard Approach Kevin Devine, Timothy J. Kloppenborg, and Priscilla O’Clock 51 Contracting and Reimbursement in Transplantation Nicolas Jabbour, Ashish Singhal, and Marwan S. Abouljoud 65 The Financial Impact of Post Traumatic Stress Disorder on Returning US Military Personnel Jeffrey P. Harrison, Lynn F. Satterwhite, and Walter Ruday, Jr. 75 Health Care Policy and the HIV/AIDS Epidemic in the Developing World: More Questions Than Answers Paul J. Flaer, Paul L. Benjamin, Francisco I. Bastos, and Mustafa Z. Younis 80 Private Health Insurance in Australia: Community Rating, but at What Price(s)? Luke B. Connelly and Henry Shelton Brown III From the Editor—About This Issue

Once again, this issue of the Journal of —James J. Unland Health Care Finance is illustrative of the The Health Capital Group breadth of topics we cover. We are always 244 South Randall Road, Ste 124 interested in new article ideas that directly Elgin, IL 60123 or indirectly relate to health care ﬁ nance. To (800) 423-5157 submit ideas or articles, please send an email healthﬁ [email protected] to: [email protected].

Patrick A. Rivers, Myron D. Fottler, and Jemima A. Frimpong

This study examines the impact of Certifi cate of Need Regulation (CNR) on hospital costs (HC). Sec- ondary data from multiple sources were used for the analysis. A panel representing 2,168 short-term general, nonfederal US hospitals operating during the period 1999–2003 was analyzed. Results of our analysis indicate that the existence of a CNR program was not related to HC; however, the stringency of the CNR program was positively and signifi cantly related to HC. Implications from these results include the inability of CNR to contain HC as assumed or expected, and the possibility that CNR may actually increase HC, while reducing competition. Keywords: Certifi cate of Need Regulation (CNR), hospital costs (HC), HC per adjusted admissions, hospital competition.

he Certifi cate of Need Regulation as an alternative instrument to controlling (CNR) emerged in the early 1960s the increase in hospital capital expenditures T as a practice to contain health costs and the state Medicaid budgets.1 (HC) in American hospitals. The overarch- From a historic perspective, the fi rst CNR ing rationale was to regulate capital expen- law was enacted by New York State in 1964. ditures of health care providers by requiring New York was then followed by Rhode providers to obtain specifi c certifi cation Island and Maryland in 1968 and California showing the need for services and expendi- and Connecticut in 1969. In 1972, the US tures. As a result of the CNR, prior approval Congress modifi ed the Social Security Act of health care investments over certain dol- (SSA) by enacting a Public Law (Public Law lar limits became mandatory, though the No. 92–603) to resonate with the CNR. The threshold varies from state to state. In an SSA reinforced the orientation of various increasingly global competitive world economy, the necessity of containing HC cannot be overemphasized. However, assumptions and practices on how HC are contained Patrick A. Rivers, PhD, MBA, is a Professor and the merit a critical examination. Through such Director of Health Care Management at the College an examination, health policy makers and of Applied Sciences & Arts, Southern Illinois Univer- administrators in the health care industry sity, Carbondale, Illinois. are likely to become more informed and Myron D. Fottler, PhD, MBA, is a Professor and adaptive to the ever-changing economic Executive Director of Health Administration Pro- grams at The University of Central Florida, Depart- environment of health care. ment of Health Professions, College of Health and The CNR refl ects one response to the ris- Public Affairs, in Orlando, FL. He can be reached at ing cost of medical care and the existence of [email protected]. excess capacity within the US health care Jemima A. Frimpong, PhD, MPH, is an Assistant system, which are some of the major con- Professor in the Heilbrunn Department of Popula- cerns of health care policy makers. As a tion and Family Health, Mailman School of Public Health at Columbia University. She can be reached result of these concerns, state governments at [email protected]. have been compelled to become actively engaged in regulating health care expendi- J Health Care Finance 2010;36(4):1–16 tures. Indeed, the CNR has been embraced © 2010 Aspen Publishers

1 2 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

state CNR proposals by prohibiting the use the anticompetitive effect of such denial” of monies allocated for Medicare, Medic- is absent from most state CNR policies. In aid, and maternal and child health programs addition, the CNR programs necessitate that to make “unnecessary capital expenditures” a legally authorized government agency by the health care facilities or health main- offer written substantiation that a change for tenance organizations (HMOs). 2 The CNR service or project is needed. laws require that state regulatory agencies The “need,” often based on the require- approve both the entry of new hospitals ments of the public for an institution or and “large” capital expenditures by exist- for a service over a preset period of time, ing hospitals. By 1979, almost all states had may be diffi cult to quantify. Furthermore, enacted these laws. There is some empirical the review process that certifi es “need” also evidence that hospitals began some capi- varies from state to state. For example, some tal projects in anticipation of CNR.3 Once states require two while other states require enacted, CNR laws plausibly would have three reviews each by different bodies of greater effects after they had been in place the review board. There is also an appeal for a number of years. By 1999, most CNR process for institutions that want to appeal state laws had been in effect for at least the decisions of the review board. The struc- 13 years. ture of CNR legislation adopted by a state As of 2002, 36 states were active partici- also depends on the economic situation of pants of the CNR program or had passed the state and the relations between politi- some form of CNR legislation. Although the cal bodies such as legislators, government laws governing the administration of CNR regulators, planners, providers, and con- differ from state to state, they generally cover sumers.7 Each of these entities undoubtedly hospitals, nursing homes, ambulatory facili- holds a distinct purpose and objective in the ties, and laboratories.4 As a norm, the state CNR process. CNR laws require agencies that regulate the This article presents the results of an health care providers within states to approve empirical study on the effectiveness of CNR the investments over a certain dollar amount as a hospital cost containment practice in the made toward the construction of new facili- US hospital industry. The study examines ties and additional beds, investments in new prior research on CNR and HC, investigates services and equipment, and expenditures CNR and HC in light of more recent data, towards restoration and equipment to sustain and addresses the implications of the current existing services.5 study fi ndings on public policy and future However, the current normative imple- research. mentation of the CNR in various states has been criticized by some researchers. For Literature Review example, Campbell and Fournier 6 maintain Research Streams on CNR that “a clear, economic, and legal standard to distinguish between an action to deny Since the introduction of CNR as a mech- an applicant in order to prevent invest- anism for cost containment in health care, ments that would raise costs by unneces- there have been numerous studies in the sary duplication, and actions motivated by health care domain concerning the impact The Effects of Certifi cate of Need Regulation on Hospital Costs 3

of CNR efforts. Most studies published in care and have negatively affected the health the 1980s and 1990s have analyzed data care industry by reducing competition. Ex- from the 1970s and 1980s. This literature amination of CNR’s failure to control cost has examined the relationship within three has been based largely on the performance streams: of programs during the early years of their enactments.15 1. Between CNR and quality of health Some authors claim that the performance care;8 of many CNR programs has improved over 2. Between CNR and access;9 and time. 16 Donahue et al .17 acknowledged the 3. Between CNR and health care system importance of early evaluation of the per- costs.10 formance of CNR programs but concluded that the CNR programs generally have little Although we have seen some progress impact on overall cost infl ation of hospitals. in understanding the nature of CNR in the These authors pointed out that some suc- fi eld of health care and its impact on health cesses have been experienced in states that care—related outcomes from the above have cost control as the primary function of studies, the results have been quite mixed. 11 CNR programs. Sloan18 came to a similar For example, results from the fi rst research conclusion when he found that CNR laws stream (CNR and quality) suggest that no reduced cost per patient. However, his fi nd- clear conclusion concerning the impact of ing did not conclude that CNR laws have CNR on hospital quality is possible since considerable impact. data are old and results mixed. Results from Lanning, Morrisey, and Ohsfeldt 19 found the second stream (CNR and access) suffer contrary results associated with the pres- from the same limitations. However, while ence of CNR. According to these authors, the impact of CNR on quality and access the presence of a CNR increased hospital are important topics, the present research spending by 20.6 percent, personal serv- focuses on the third research stream (i.e ., the ices by 13.6 percent, and other health care impact of CNR on HC) where current stud- expenditures by 9 percent. In other cases, ies suggest inconsistent results. the absence of a CNR program is reported to have a negative effect on HC. For example, Research on CNR and HC using time series data to assess the effects Empirical studies have shown different of eliminating CNR, Conover and Sloan 20 and mixed impacts of CNR on HC. Data found that there is a 5 percent long-term gathered from the early 1980s suggest that decrease in acute care spending per capita CNR programs did little to contain cost.12 as a result of eliminating mature CNR pro- Although most of the past studies on CNR grams. In addition, these authors found no focused on hospital expenditures, CNR has signifi cant change in total per capita spend- been used by many states to plan and regu- ing. However, they also found that after the late facilities despite the apparent inability elimination of CNR, there was no increase of CNR programs to lower costs.13 Burda14 in the acquisition of facilities or costs, and states that CNR programs have not been there was a 2 percent reduction in bed sup- instrumental in controlling the cost of health ply. Finally, Younis, Rivers, and Fottler21 4 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

also found a positive relationship between CNR as a hospital cost containment practice the existence of CNR and HC. in the US hospital industry. While most studies have failed to clearly In addressing the limitations of previous delineate the usefulness of CNR regu- research on CNR and HC, this study takes lations in containing hospital and other a different, more sophisticated approach to health care costs, the case for deregulation looking at the relationship between CNR seems strong to some researchers.22 Some and HC. National data (1999–2003) encom- researchers believe that deregulation is passing all states in the United States were necessitated by the anti-competitive CNR used to assess the impact of CNR on HC. impact of protecting existing providers The impact of both existence and strin- from competition.23 Although assessment gency of CNR in the states where it exists of CNR programs does not show a signifi - was included in the analysis. The study cant impact on hospital expenditures, poli- also advances our knowledge base of CNR cymakers in many states are not inclined and extends the literature by controlling for to abolish CNR laws. Their prime con- a number of environmental, market, and cern is that eliminating the CNR program institutional variables, which have not been would result in increased health care capi- controlled in previous research. The study tal expenditures and operating expenses hypothesis examines the relationship of both despite data to the contrary. The motivat- the existence of CNR and the stringency of ing factor is that for a CNR program to be the regulation on HC: effective, it has to put restrictions on both Hypothesis: The existence of a Certifi - existing hospitals and those looking to cate-of-Need Regulation and the stringency enter the industry. of CNR will both negatively impact HC, The review of the literature refl ects an after controlling for environmental, market, ambiguity regarding the impact of CNR on and institutional characteristics. HC. Previous research suffers from a lack of recent data, failure to differentiate the Methodology various impacts of CNR (i.e ., on HC versus other impacts), inadequacies of the meas- Sources, Defi nitions, and Measures of Variables urement of CNR, insuffi cient research on CNR impact on HC, failure to control for the This study integrates data from different effects of managed care and other environ- but related sources and datasets to test the mental or market variables, and the lack of study hypothesis. The datasets used were national data in most of the earlier studies drawn from the databases of the Ameri- conducted. can Hospital Association Annual Survey While the question of CNR effectiveness (AHA),24 American Health Planning Associ- remains an area of public policy debate and ation (AHP),25 Area Resource File (ARF),26 an area that warrants the attention of health Centers for Medicare & Medicaid Services service researchers, it has been at least a dec- (CMS),27 CMS Case-Mix Index (CMI), and ade since research in this area has been done. InterStudy Data (ISD).28 The AHA dataset The purpose of the present study is to present contains data on an annual survey of non- a focused examination of the effectiveness of federal short-term general hospitals in the The Effects of Certifi cate of Need Regulation on Hospital Costs 5

United States. The analysis included data on using 1980s data. Finally, by 1999, the surveys conducted in 1999–2003. The AHA, effects of the Medicare’s Prospective Pay- AHP, ARF, CMS, and CMI datasets provided ment System and the Balanced Budget Act measures for capital investment, fi nancial of 1997 should have also stabilized, thereby factors, and operational characteristics while minimizing extraneous sources of variation the ISD dataset provided HMO penetration in the data. rate. The measures were used to obtain oper- CNR is defi ned as the primary independ- ational and market characteristics, and only ent construct with two variables: hospitals located in metropolitan statistical areas (MSAs) in 1999–2003 were included 1. The existence of CNR law in the state in the analysis. where the hospital is located; and While defi ning a hospital’s market can 2. The stringency score for the CNR pro- be problematic,29 for this study, a hospital’s gram of each state used. market is defi ned by the MSA for urban hospitals, and by county for non-MSA hospitals The stringency score is measured by the since rural hospitals may be in communities number of CNR-regulated services multi- too small to be included in an MSA. MSA plied by a weight based on reviewability is defi ned by the US Bureau of Census30 to thresholds. For the two CNR variables, (1) include central cities and their associated CNR laws are defi ned as 1 if hospital is suburbs. The use of only those hospitals located in a state that has a CNR law, and 0 operating in MSAs is valuable in that the otherwise; and (2) for CNR stringency (1 if a defi nitions of hospital markets and HMO state has the most stringent CNR thresholds, markets are reasonably clear, and enhance and 0 otherwise). the validity of hospital and HMO penetra- The states having the most stringent CNR tion measures. are Maine, Connecticut, West Virginia, The impact of CNR on HC was investi- Georgia, Alaska, Vermont, South Carolina, gated with the hospital as the unit of analy- and Missouri. If CNR programs are effective sis. Data 1999–2003 determined if current in containing cost, then it is expected that the fi ndings will refute or substantiate fi ndings regression coeffi cients for each of the two from earlier studies that used data from the CNR independent predictor variables will be 1980s. In addition to using more recent negative and signifi cant (see the analytical hospital data, this study takes into con- approach in next section). sideration the stabilization of the hospital The study defi nes the dependent construct industry in the implementation of CNR in with one variable, HC per adjusted admis- the United States. Those states that enacted sion. Previously, measures of HC have been CNR have not seen signifi cant changes in cost per day or cost per case. In some cases, these laws between the early 1990s and both of these indicators have been used. 31 In 2000. The period selected for this study is the present study, costs-per-adjusted admis- also particularly advantageous since there sion was used to measure HC. Since the were signifi cant changes in both the number expense data on the AHA Annual Survey of of HMOs and enrollment in HMOs than Hospitals included both the inpatient and out- what would have been captured in studies patient expenses, the admission was adjusted 6 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

to summarize the inpatient and outpatient for rural facilities.35 This study also meas- use into a single utilization measure. The ured the level of managed care penetration AHA calculated adjusted admissions attrib- in each market defi ned as the percent of the uted to outpatient services by multiplying population enrolled in HMOs. Market vari- admissions by the ratio of outpatient revenue ables also include per capita income and per- to inpatient revenue. centage of non-Whites in the market area. The HC measure was calculated in this The institutional control and operating study as operating expense or costs divided variables include percentages of Medicare by adjusted admissions. This choice of and Medicaid discharges from the hospital variable was conceptually consistent with as well as patient acuity [derived from CMS the goals of hospitals in the environment data on Medicaid and Medicare discharges], of increasing dominance of fi xed payment bed size, system affi liation, staffi ng intensity, reimbursement. Fixed payment reimburse- ownership status, occupancy rate, staffi ng ment caused hospitals to have as their objec- index, teaching status, and Medicare wage tive the minimization of the cost per episode index ( i.e., cost of hospital labor). of care. Operating expense or cost was calculated as the total facility expense minus non- Empirical Specifi cation operating expenses including depreciation, and Analytic Approach interest, and other non-operating losses.32 All variables used in the study are defi ned The analytic approach addresses several and listed in Figure 1. For all constructs Fig- important issues absent from any earlier ure 1 lists the variables, measures, means, single study. First, from the theoretical fram- and standard deviations of the variables and ing of the CNR program, HC are assumed data sources. to differ only in the values of the measured The specifi c market environmental, mar- attributes included as explanatory variables ket, and control variables were identifi ed and control variables. However, there exists through a review of previously cited litera- the possibility that hospitals have unmeas- ture regarding CNR regulation and HC,33 ured attributes that may affect HC. It is as well as the impact of these variables on often believed that these hospital-specifi c HC.34 The control variables included the variables are correlated with the variables models’ per capita income and percentage of of interest, and thus their exclusion leads non-White in the market as proxies for socio- to omitted variables bias problems.36 Sec- economic status. To examine the effect of ond, there might be year-specifi c effects.37 market competition on HC, the Herfi ndahl- Third, while market variables are assumed Hirschman index (HHI), defi ned as the sum as strictly exogenous, that is, uncorre- of squares of the market shares of all facili- lated with the error term in all time peri- ties in the market, is used. Hospital market ods, hospital-level variables are not strictly share is measured by the hospital’s acute- exogenous.38 care patient days divided by total acute-care Fourth, there is the possibility of “feedback patient days for the MSA in which the hos- effects” which are most easily thought of as pital was located for urban hospitals, and a type of endogeneity across time periods. total acute-care patient days in the county For example, a change in HC in period [t] The Effects of Certifi cate of Need Regulation on Hospital Costs 7

Figure 1. Variables, Measurement, Descriptive Statistics, and Data Source: 1999–2003

Std. Variable Measure Mean Deviation Source

Dependent Variable Hospital Costs Operating expense or costs divided 6,187.515 2554.44 AHA by adjusted admissions Independent Variables Certifi cate of Need 1, existence of CNR law; 0 otherwise 0.660 0.474 AHP Regulation (CNR) CNR Stringency 1, if a state has most stringent CNR 0.085 0.278 AHP thresholds; 0 otherwise Market Variables HMO Penetration % HMO enrollment as % of total 0.309 0.157 ARF MSA population HMO Competition Market shares based on distribution 0.681 0.206 Interstudy MSA of enrollees’ market (i.e., 1- value of HMO Herfi ndahl Index) Squared sum of (acute-care patient 0.819 0.185 ARF/CM days/total acute-care patient days in the market) Per Capita Income Log of per capita income in the market 27,775.020 7352.318 ARF % Non-White % Nonwhite population in the market 0.314 0.178 ARF Operating Variables For Profi t 1, for profi t; 0, otherwise 0.192 0.394 AHA Bed Size Number of staffed beds 229.886 189.659 AHA Teaching Status 1, for teaching; 0 otherwise 0.105 0.306 AHA Occupancy Rate Inpatient days/(staffed beds* 365) 0.571 0.171 AHA Staffi ng Intensity Health care workers full-time 13.691 5.567 AHA equivalents (FTEs) per 1,000 adjusted patient days Wage Index Cost of health care labor (i.e., ratio of 1.013 0.154 CM adjusted average hourly wage to mean of adjusted average hourly wage) System Affi liation 1, system affi liated; 0 freestanding 0.723 0.448 AHA % Medicare Medicare discharges/total discharges 0.412 0.129 Discharges % Medicaid Medicaid discharges/total discharges 0.139 0.101 CMS Discharges Case-Mix Index Medicare case-mix index 1.394 0.253 CM

Notes: AHP = American Health Planning Association; ARF = Area Resource File; AHA = American Hospital Association; CMS = Centers for Medicare & Medicaid Services; CM = CMS Case-Mix Index; ISD = InterStudy Data 8 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

may feed back to changes in bed size in where i is used to index the hospital period [t+1]. Such feedback effects violate and t is used to index the year (N = 2,168 the typical assumption of strict exogeneity. and T = 5 in our case). yit equals log of In this study, feedback effects are allowed hospital i’s costs per adjusted admis- by making the weaker assumption that hos- sion at year t, α is constant, X1 it equals pital-level regressors are predetermined: the CNR, X2 it equals CNR stringency, X3 it error term is uncorrelated with current and equals environmental/market variables, X4 past values of the predetermined regressors it equals operating variables, λi is unob- but potentially correlated with future values servable hospital-specifi c effect which is of regressors. constant across time, ηt is an time-specifi c To address the foregoing problems, a effect which varies across time, and uit fi xed effects model is employed to remove equals unexplained residual variation. α, the infl uence of such hospital heterogeneity β1, β2, β3, and β4 are coeffi cients needed and year-fi xed effects. Although one of the to be estimated, and they are estimated by commonly applied methods for fi xed-effects applying the IV estimation to the following models is the within-group transformation fi rst-differenced equation: in which the ordinary least squares (OLS) Δyit = β1ΔX1 it + β2ΔX2 it + β3 ΔX3 it + estimator is applied to data transformed by β4 ΔX4 it + Δηt +Δuit; i=1,2,…,N; taking deviations from time-series means for t=1,2,…,T, each cross-sectional unit, the within-group transformation yields inconsistent parameter where Δ denotes the difference operator. estimates if the model does not include strictly exogenous variables.39 Thus, the current study applies fi rst- Results difference transformation with the instru- Preliminary Tests ment variable (IV) estimation. After applying the fi rst-difference transformation to First, the study checked correlations eliminate the fi xed effects, the dependent among the study variables. While most had variable is regressed on the fi rst differences low correlations, some correlations coef- of the regressors. As consistent estimates fi cients were higher than others. However, may be obtained by using past values of the dropping one or more of the independent strictly exogenous regressors as instruments, variables in an effort to reduce multicolline- a two-year lagged value of the endogenous arity could lead to omitted variable bias.40 variable and one-year lagged values of the Since the study variables are properly cho- predetermined regressors are used as the sen based on theory and previous literature, instruments. all the variables were included in the subse- More specifi cally, the regression model is quent analyses. given below: Also important is the question of serial correlation. Serial correlation was tested yit = α + ß1X1 it + ß2X2 it + ß3X3 it + without strictly exogenous regressors. First, ß4X4 it + λi +ηt +uit; i=1,2,…,N; the simple OLS regression of the depend- t=1,2,…,T, ent variable on the independent variables The Effects of Certifi cate of Need Regulation on Hospital Costs 9

was run; and the OLS residual value was normality assumption of regression.41 We obtained. Second, the residual was regressed analyzed the data to test the hypothesis of on the lagged residual and all of the independ- the relationship between CNR construct ent variables. Finally, a heteroskedasticity- variables and HC performance variables (as robust version of the test was used to check indicated above). The results of the analysis the signifi cance of the coeffi cient for the of CNR on health system performance lagged residual. Since no signifi cant results (i.e ., HC per adjusted admission) are dis- were obtained, there is no evidence that the cussed below. The estimates of the coeffi - data have serial correlation problems. cients and standard errors from OLS results of the model regressions are presented in Descriptive Findings Figure 2. From the analysis, the adjusted R2 for Figure 1 displays the mean values and the model is 0.48. CNR stringency is standard deviations for all variables included signifi cantly and positively associated in the analysis of the 2,168 (36 percent of with costs per adjusted admission at the total number of hospitals) nonfederal short- .05 level. There was no signifi cant relation- term care general hospitals in the sample. ship between CNR laws and HC. The esti- Nineteen percent were for-profi t organiza- mated coeffi cient for the CNR law variable tions, the average number of staffed beds is 0.009. The positive signs indicate that all were 229, the occupancy rate was 57 per- else being equal, HC per adjusted admis- cent, and 10 percent were teaching hospitals. sion increase if the hospital is located in a HMO penetration in the market averaged state that has CNR law. Our fi ndings con- 30.9 percent in 1999–2003 and on average; cur with a number of studies conducted the hospitals were located in more com- with data from 1970s and 1980s, which petitive markets. In 1999–2003, 41 percent concluded that the CNR did not decrease of hospital discharges showed Medicare as HC in the 1970s.42 Our fi ndings are also payer and 13.9 percent showed Medicaid as in agreement with two other studies which payer. The mean of costs per adjusted admis- showed that CNR is associated with only sion was $6,187.52. a modest increases in HC in the 1980s. 43 Even though previous results separately Regression Results examined the 1970s and 1980s, these A regression model was used to determine results for 1999–2003 data are consistent the impact of CNR on costs per adjusted with those earlier studies. admission in hospitals. The model contained The results also showed that there are all the hospitals in MSAs in the sample. The several other variables that have a signifi - existence of CNR laws and CNR stringency cant impact on HC. Higher costs were found were used as independent variables. to be associated with hospitals with major The dependent variable used in the teaching functions, larger size, higher occu- regression is the natural logarithm (LOG) pancy rates, higher staff intensity, higher of HC per adjusted admission. The LOG percentage wage intensity, higher percent- is used to provide normal distributions of age of Medicare and Medicaid discharges, the dependent variable in order to meet the higher case-mix, and location in high income 10 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Figure 2. OLS Regressions with Robust Standard Errors—Dependent Variable: Log of Hospital Costs

Variables Coef. Std. Err. t Intercept 5.808 0.284 20.450 ** Certifi cate of Need Regulation (CNR) 0.009 0.013 0.750 CNR stringency 0.049 0.021 2.370 * HMO penetration 0.050 0.042 1.190 HMO competition MSA -0.038 0.032 -1.160 Hirschman-Herfi ndahl Index -0.082 0.038 -2.140 * Log [per capita income] 0.141 0.030 4.750** % Non-Whites 0.123 0.039 3.130 ** For profi t -0.050 0.015 -3.390 ** Bed size 0.000 0.000 7.220 ** Teaching status 0.221 0.021 10.410 ** Occupancy rate 0.101 0.040 2.530 * Staffi ng intensity 0.005 0.001 4.510 ** Wage index 0.457 0.049 9.330** System affi liation -0.021 0.012 -1.720 % Medicare discharges 0.276 0.053 5.170 ** % Medicaid discharges 0.343 0.060 5.730 ** Case-Mix Index 0.040 0.029 15.390 **

Adjusted R-square = 0.48 F-value = 40.70 * Signifi cant at 0.05 level **Signifi cant at 0.01 level areas and/or areas with a higher percentage 1. Contrary to expectation, the existence of non-Whites. HC were lower for hospitals of CNR law has no statistically signifi - located in more competitive hospital markets cant impact on HC per adjusted admis- as defi ned by the market share variable and sions for all hospitals; and HMO penetration. 2. Contrary to expectation, CNR stringency has a positive statistically signif- Discussion icant relationship with HC per adjusted admissions for all hospitals. The purpose of this study was to inves- tigate the impact of CNR on health care Previous health services research on organizational performance, as measured by the impact of CNR on HC has tended to HC. The main fi ndings of this study can be either use data that pre-dates the imple- summed up as follows: Based on the hypoth- mentation of the prospective payment sys- esis investigated: tem (PPS) in 1984 or predates the rise of The Effects of Certifi cate of Need Regulation on Hospital Costs 11

managed care during the 1990s. This has it is important to note that each state has made the generalizability of these previous different regulations and operates in dif- results to the current health care environ- ferent markets that are unique to the par- ment questionable. The present study went ticular state. A similar statistical analysis beyond previous research in a number of of all 50 states by Conover and Sloan47 ways. The CNR effects on HC were exam- reported that removing CNR did not have ined after establishing more sophisticated any overall effect on per capita health care controls for possible intervening environ- spending. mental, market, and institutional variables. Examining the impact of CNR, we con- In the current study, cost per adjusted trolled for all things being equal and the admission was used as a measure for HC. estimated coeffi cient showed a positive sign, HC were calculated in this study as oper- which illustrates that HC per adjusted admis- ating expense or costs divided by adjusted sion increase if the hospital is located in a admission. state that has a CNR law. Our fi ndings are Our results, as well as those of several pre- substantiated by previous studies. Lanning, vious studies, indicate that CNR programs do Morrisey, and Ohsfeldt48 also measured the not only fail to contain HC, but may actually effects of CNR on hospital expenditures and increase costs as well. Our results, together also found it to be positive and signifi cant. with those of previous research, heighten the The most signifi cant increase was for hos- debate whether CNR will ever be an effec- pital expenditure where CNR appeared to tive HC containment approach, and coun- increase per capita hospital expenditure by ter arguments that CNR programs could be 20.6 percent. They also found that CNR more effective after they have been in place raised hospital prices and they attributed for a period of time. this fi nding to the restraining of competi- Numerous studies, as referenced in this tion by CNR laws. Similar to our fi ndings, research, have made evident the ineffec- Sloan and Steinwald 49 found no evidence of tiveness of the CNR program in containing CNR impact on for-profi t hospitals. After HC. Studies conducted in the 1980s showed CNR repeal, for-profi t hospitals did not that CNR programs were not successful signifi cantly increase their costs or market in controlling hospital expenditures.44 The presence. fi ndings of our study are consistent with several studies conducted during the 1980s Limitations as well as some studies published in the 1990s.45 There are a number of limitations inherent Our fi ndings, together with results from in this study. Similar to studies that defi ned previous studies, raise the question of the hospital and HMO markets in the research impact of the abolishment of CNR on process,50 this study by defi nition excluded HC. To determine the impact of the abol- some hospitals. Hospitals that operate out- ishment of CNR programs, Mendelson side of an MSA were not included in this and Arnold46 reported that there was no study. Organizational strategy is another increase in cost in 12 states that abolished limiting factor; by using a geographic defi - CNR programs. Considering this fi nding, nition of the market, this study tends to 12 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

overestimate the competitiveness of markets HC. These fi ndings when combined sug- if segmentation is part of the market strat- gest CNR laws constrain competition more egy. That is, hospitals and HMOs may be than the lowering of hospital expenditures. located in the same MSA, yet due to market Similar to Conover and Sloan,52 these study segmentation, they may appear not to com- results refute the argument that the ending pete with others in that MSA since they cater of CNR laws will increase HC or costs of to different populations (e.g ., young families other health care services. The goals for versus older adults, white collar versus blue cost containment, in addition to increasing collar). access and quality sought by most CNR laws There are some issues that may be of con- do not achieve that end result, and may be cern but were not addressed in the design counterproductive. A recent study by Short, of this study. HMO enrollment data do not Aloia, and Ho53 examined how Certifi cate delineate which portions of the enrollees of Need (CON) infl uences cardiac mortal- are located within the MSA. Also, the study ity rates and reported that states that dropped data do not capture how the HMOs reim- CON had relatively lower rates for Coronary burse. The data do include the total number artery bypass graft (CABG) surgery, with no of enrollees and the service area (usually association between CNR and higher quality by county) of the HMO, requiring that the of care. enrollment for HMOs with service areas State goals for enhancing consumer access overlapping MSA and non-MSA counties and quality could be better achieved through be estimated. other programs such as provider or insurer Second, like all cross sectional studies, report cards.54 this study demonstrates only association The results indicate that CNR strin- and leaves open the question of causality. gency has a positive statistical relationship Third, by defi ning a market at the MSA to urban HC within the period 1999–2003. level, only a fraction of hospitals were Since the purpose of CNR legislation is included in the analysis. Hospitals located to contain or reduce such HC, we con- outside of defi ned MSAs would not be clude that CNR policies did not achieve captured by the measure. This biases the their stated objectives during the study sample toward urban areas and larger size period. As a consequence of the inability hospitals.51 Fourth, of the hospitals stud- of CNR laws to contain HC, many states in ied, the mean case-mix index is 1.34. This the United States are attempting to refi ne fi gure contrasts poorly with the nation as their CNR to better address the nature and a whole with a mean of 1.00. This differ- causes of HC infl ation. Future research ence could also bias the results of this study. should evaluate these initiatives in order Not withstanding the foregoing limitations, to determine which approaches are most this study provides further insight into CNR effective in achieving state objectives, and spurs further research that will seek to with particular attention to rural hospi- address these shortcomings. tals that experience a higher percentage From the current study and the fi ndings of Medicare and Medicaid discharges, of several earlier studies, it appears that higher case-mix, and higher percentage of CNR may stifl e competition and increase non-Whites. The Effects of Certifi cate of Need Regulation on Hospital Costs 13

REFERENCES

1. Hackey, RB, “New Wine in Old Bottles: Cer- Dimension,” Health Policy Quarterly , 1(4): tifi cate of Need Enters the 1990s,” Journal of 5–16 (1981); and Federal Trade Commission, Health Politics, Policy, and Law , 18(4): 927– Certifi cate of Need Regulation of Entry Into 935 (1993). HomeHealth Care Markets , Washington, DC: 2. US Department of Health, Education, and Federal Trade Commission (1986). Welfare, Certifi cate of Need Programs: A 9. E.g ., Weaver, JA, Certifi cate of Need: What Review, Analysis, and Annotated Bibliogra- Role Does it Have in Managed CareEnviron- phy of the Research Literature , Washington, ment”? in EG Gosfi eld (ed.), Health Law DC: US Department of Health, Education, Handbook , pp. 409–429 (1995); Mendelson, and Welfare (1978). DM, Arnold, J, Certifi cate of Need Revisited , 3. Sloan, FA., Steinwald, B, “Effects of Regulation Spectrum (Winter) 1993; Campbell, ES, and on Hospital Costs and Input Use,” Journal of Ahern, MW, “Have Procompeitive Changes Law and Economics , 23(1): 81–107 (1980). Altered Hospital Provision of Indigent of Care?” 4. Supra , n.2. Health Economics , 2(1): 281–289 (1993). 5. Coyte, PC, “Alternative Methods of Reimburs- 10. E.g ., Salkever, DS, Bice, TW, “Certifi cate ing Hospitals and the Impact of Certifi cate of of Need Legislation and Hospital Costs,” in Need and Rate Regulation for the Hospital Hospital Cost Containment , . M Zubkoff, IE Sector,” Southern Economic Journal , 53(4): Raskin, RS Hanft (eds.), New York, NY: Pro- 858–873 (1987). dist (1978); Sloan, FA, “Containing Health 6. Campbell, ES, Fournier, GM, “Certifi cate of Expenditures: Lessons Learned from Certifi - Need Deregulation and Indigent Hospital cate of Need Programs,” in Sloan, Blumstein, Care,” Journal of Health Politics, Policy, and and Perrin (eds.), Cost, Quality, and Access in Law , 18(4): 905–925 (2003). Health Care; New Roles for Health Planning 7. Supra , n.2. in a Competitive Environment , San Francisco: 8. E.g ., Luft, HS, Garnick, DW, Mark, DH, and Jossey-Bass, 1988; Conover, C, Sloan, FA, McPhee, SJ, Hospital Volume, Physician Vol- “Does Removing CNR Regulations Lead to a ume, and Patient Outcomes: Assessing the Surge in Health Care Spending?” Journal of Evidence, Ann Arbor, MI: Health Adminis- Health Politics, Policy and Law , 23(3): 455– tration Press (1990); Conover, C, Sloan, FA, 481 (1998); Lanning, J, Morrisey, ME, Ohs- “Does Removing CNR Regulations Lead to feldt, RL, Endogenous Hospital Regulations a Surge in Health Care Spending?” Journal and Its Effects on Hospital and Non-Hospital of Health Politics, Policy and Law , 23(3): Expenditures,” Journal of Regulatory Econom- 455–481 (1998); Griffi ths, RI, Powe, NR, ics , 3(2): 137–154 (1991); and Mendelson, Gaskin, DJ, Anderson, GF, Lissovoy, GV, and DM, Arnold, J, Certifi cate of Need Revisited , Whelton, PK, “The Production of Dialysis by Spectrum (Winter) 1993. For-Profi t Versus Not-For-Profi t Freestanding 11. Vaughan-Sarrazin, MS, Hannan, EL, Gorm- Renal Dialysis Facilities,” Health Services ley, CJ, Rosenthal, GE, “Mortality in Medi- Research , 29(4): 215–230 (1994); Lanning, J, care Benefi ciaries Following Coronary Artery Morrisey, ME, Ohsfeldt, RL, Endogenous Hos- Bypass Graft Surgery in States with and With- pital Regulations and Its Effects on Hospital out Certifi cate of Need Regulation,” JAMA and Non-Hospital Expenditures,” Journal of 288: 1859–1866 (2002); Ho, V, “Does Cer- Regulatory Economics , 3(2): 137–154 (1991); tifi cate of Need Affect Cardiac Outcomes and supra , n.1; Brown, EJ, Smith, JD, Sindelar, J, Costs?” Int J Health Care Finance Econ , 6(4): Can We Regulate the Quality of Care?: The 300–324 (2007). Case of Dialysis in Connecticut,” American 12. Sloan, FA, “Containing Health Expenditures: Journal of Kidney Diseases , XIX(6): 33–45 Lessons Learned from Certifi cate of Need Pro- (1992); Hershey, N, Robinson, D, “Health grams,” in Sloan, Blumstein, and Perrin (eds.), Planning and Certifi cate of Need: The Quality Cost, Quality, and Access in Health Care; 14 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

New Roles for Health Planning in a Competi- 21. Younis M, Rivers, PA, Fottler, MD, “The Impact tive Environment , San Francisco: Jossey-Bass, of HMO and Hospital Competition on Hos- 1988; Conover, C, Sloan, FA, “Does Remov- pital Costs,” Journal of Health Care Finance , ing CNR Regulations Lead to a Surge in Health 31(4): 60–74 (2005). Care Spending?” Journal of Health Politics, Pol- 22. Supra , n.6. icy and Law , 23(3): 455–481 (1998); Lanning, 23. Bernstein, AB, Gauthier, AK (eds.), “Issue J, Morrisey, ME, Ohsfeldt, RL, Endogenous Focus: Data Needs for Studies of Competition Hospital Regulations and Its Effects on Hospi- in Market Areas, Part II [Special Issue], Health tal and Non-Hospital Expenditures,” Journal of Services Research , 36 (1): 169–308 (2001). Regulatory Economics , 3(2): 137–154 (1991); 24. American Hospital Association (AHA), Annual and Mendelson, DM, Arnold, J, Certifi cate of Survey of Hospitals, 1999–2003 , Chicago: Need Revisited , Spectrum (Winter) 1993. American Hospital Association (1999–2003). 13. Kiel, JM, “How State Policy Affects Rural Hos- 25. American Health Planning Association, State pital Consortia: The Rural Health Care Deliv- Certifi cate of Need Programs , Falls Church, ery System,” The Milbank Quarterly , 71 (4): VA: American Health Planning Association 625–643 (1993). (1999–2003) . 14. Burda, D, “Conspiracies to Crush Competi- 26. US Health Resources and Services Adminis- tion: Hospitals Using CNR Laws to Thwart tration (HRSA), Offi ce of Research and Plan- Rivals’ Projects,” Modern Healthcare , July 8, ning, Bureau of Health Professions, Area pp. 28–34 (1991). Resource File , Washington, DC: USORP 15. Anderson, GF, “All Payor Rate Setting,” (1999–2003). Health Care Financing Review , Supplement: 27. Centers for Medicare & Medicaid Services, 35–41(1991); Anderson, G, Heyssel R, Dick- Research, Statistics, Data & System , Balti- ler, R, “Competition versus Regulation: Its more, MD: Centers for Medicare & Medicaid Effect on Hospitals,” Health Affairs , 12(1) 70- Services (1999–2003). 80. (1993); Schwartz , WB, “The Regulation 28. InterStudy, Update Report , Excelsior, MN: Strategy for Controlling Hospital Cost,” The InterStudy (1999–2003). New England Journal of Medicine , 305(21): 29. Clement, JP, Grazier, KL, “HMO penetra- 1249–1255 (1981); Merrill, J, McLaughlin, tion: Has It Hurt Public Hospitals?” Journal C, “Competition versus Regulation: Some of Health Care Finance , Fall, 28(1): 25–38 Empirical Evidence,” Journal of Health Poli- (2001). tics, Policy and Law , 10(4): 613–623 (1986). 30. US Bureau of the Census, Estimates of the 16. Supra , n.1. Population of the U.S. Regions, Divisions, 17. Donahue, JX, Williams, DC, Waters, WJ, and States by 5-Year Age Groups and Sex: DeBuono, BA, “Affordability Considera- Annual Time Series, July 1, 1999 to July 1, tions in Certifi cate of Need Hospital Capital 2003, Washington, DC: USBC (1999–2003). Expenditure Review Determinations,” Rhode 31. Robinson, JC, Luft, HF, “The Impact of Hos- Island Medicine , 75: 347–350 (1992). pital Market Structure on Patient Volume, 18. Sloan, FA, “Regulation and the Rising Cost Length of Stay, and Cost of Care 1972–1982, of Hospital Care,” Review of Economics and Journal of Health Economics , 4(1): 333–356 Statistics, 63(4): 479–487 (1981). (1985); Anderson, GA, Lave, JR, “Financing 19. Lanning, J, Morrisey, ME, Ohsfeldt, RL, Endog- Graduate Medical Education,” Inquiry , 23(3): enous Hospital Regulations and Its Effects on 322–336 (1986); Nguyen, NX, Madamba, Hospital and Non-Hospital Expenditures,” HR, Estimating The Impacts of Various Reim- Journal of Regulatory Economics , 3(2): 137– bursement Factors on the Medicare Operat- 154 (1991). ing Cost per Case in the Third Year of PPS , 20. Conover, C, Sloan, FA, “Does Removing CNR Research Note, Washington, DC: Consoli- Regulations Lead to a Surge in Health Care dated Consulting Group (1988); Thorpe, KE, Spending?” Journal of Health Politics, Policy “The Use of Regression Analysis to Determine and Law , 23(3): 455–481 (1998). Hospital Payment: The Case of Medicare’s The Effects of Certifi cate of Need Regulation on Hospital Costs 15

Indirect Teaching Adjustment,” Inquiry , 25(2): 41. Neter, J, Wasserman, W, Kutner, MH, Applied 116–125 (1988). Linear Regression Models , Boston, MA: Irwin 32. Rivers, PA, The Effects of Market Structure on (1989). Organizational Performance: An Empirical 42. Salkever DS, Bice TW, “The Impact of Cer- Analysis of Hospital, Dissertation, University tifi cate of Need Controls on Hospital Invest- of Alabama, Birmingham (1997). ments,” Milbank Mem Fund Q , Spring 1979 33. Federal Trade Commission, The Effect of State 54(2): 185–214; supra , n.3. Certifi cate of Need Laws on Hospital Costs: 43. Ashby, JL, “The Impact of Hospital Regulatory An Economic Policy , Washington, DC: Fed- Programs on Per Capital Costs, Utilization, and eral Trade Commission (1988); Hartley, D, Capital Investment,” Inquiry , 21(4): 120–132 Moscovice, I, Christianson, J, “Mobile Tech- (1984); Federal Trade Commission, The Effect nology in Rural Hospitals: The Case of the CT of State Certifi cate of Need Laws on Hospital Scanner,” Health Services Research , 31(2): Costs: An Economic Policy , Washington, DC: 213–234 (1996). Federal Trade Commission (1988). 34. Farley, DE, Competition Among Hospi- 44. Id. tals: Market Structure and Its Relation to 45. Antel, JJ, Ohsfeldt, RL, Becker, ER, “State Utilization, Costs, and Financial Position, Regulation and Hospital Costs,” Review of Research Note 7, Hospital Studies Program, Economics and Statistics , 77(3): 416–422 DHHS Publication No. (PHS) 85–3353, US (1995); Lanning, J, Morrisey, ME, Ohsfeldt, Department of Health and Human Services, RL, Endogenous Hospital Regulations and Its National Center for Health Services Research Effects on Hospital and Non-Hospital Expen- and Health Care Technology Assessment ditures,” Journal of Regulatory Economics , (1985); Robinson, JC, Luft, HF, “The Impact of 3(2): 137–154 (1991); Conover, C, Sloan, FA, Hospital Market Structure on Patient Volume, “Does Removing CNR Regulations Lead to a Length of Stay, and Cost of Care 1972–1982, Surge in Health Care Spending?” Journal of Journal of Health Economics , 4(1): 333–356 Health Politics, Policy and Law , 23(3): 455– (1985). 481 (1998). 35. McCue, MJ, Thompson, JM, Dodd-McCue, 46. Mendelson, DM, Arnold, J, Certifi cate of Need D, “Association of Market, Operational, and Revisited , Spectrum (Winter 1993). Financial Factors with Hospitals’ Level of 47. Conover, C, Sloan, FA, “Does Removing CNR Cash and Security Investments,” Inquiry 37(4): Regulations Lead to a Surge in Health Care 411–422 (2001/2002). Spending?” Journal of Health Politics, Policy 36. Id. and Law , 23(3): 455–481 (1998). 37. Calem, PS, Rizzo, JA, “Financing Constraints 48. Lanning, JA, Morrisey, MA, Ohsfeldt, RL, and Investment: New Evidence from Hospital Endogenous Hospital Regulations and Its Industry Data,” Journal of Money, Credit and Effects on Hospital and Non-Hospital Expen- Banking , 27(4): 1002–1014 (1995). ditures,” Journal of Regulatory Economics , 38. Wedig, GJ, Hassan, M, Morrisey, MA, “Tax- 3(2): 137–154 (1991). Exempt Debt and the Capital Structure of 49. Supra , n.3. Nonprofi t Organization: An Application to 50. E.g ., Chirikos, TH, White, SL, “Competition Hospital,” Journal of Finance , 51: 1247–1283 in Health Care Markets and the Development (1996). of Alternatives Forms of Service Delivery,” 39. Mark, BA, Harless, DW, McCue, M, Xu, Y, “A Health Policy , 8(3): 325–338 (1987); Wholey, Longitudinal Examination of Hospital Reg- D, Feldman, R, Christianson, J, “The Effect of istered Nurse Staffi ng and Quality of Care,” Market Structure on HMO Premium,” Journal Health Services Research , 39(2): 279–300 of Health Economics , 14(2): 81–105 (1995); (2004). Wholey, D, Feldman, R, Christianson, J, Eng- 40. Wooldridge, J, Econometric Analysis of Cross berg, J, “Scale and Scope Economies Among Section and Panel Data, MIT Press, Cam- Health Maintenance Organizations,” Journal bridge (2001). of Health Economics , 15(6): 657–684 (1996). 16 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

51. Noether, M, “Competition Among Hospitals,” 54. Fowles, JB, Kind, EA, Braun, BL, Knut- Journal of Health Economics , 7(2): 259–262 son, DJ, “Consumer Responses to Health (1988). Plan Report Cards in Two Markets,” Med 52. Supra , n.47. Care 38(5): 469–481 (2000); Werner, RM, 53. Short, MN, Aloia, TA, Ho, V, “Certiﬁ cate of Konetzka, RT, Kruse, GB, “Impact of Pub- Need Regulations and the Availability and lic Reporting on Unreported Quality of Use of Cancer Resections,” Annals of Surgical Care,” Health Serv Res , 44(2): 379–398 Oncology , 15(7) 1837–1845 (2008). (2008). The Utilization of Hospitalists Associated with Compensation: Insourcing Instead of Outsourcing Health Care

Doohee Lee and Andrew Sikula, Sr.

Objectives: The utilization of hospitalist is reversing an industrial and health care business model where outsourcing work has been the trend for the past several decades. This empirical analysis seeks to understand a link between hospitalist utilization and physician compensation affected by quality of care. Methods: We analyzed the secondary data from the 2004–2005 CTS Physician Survey (n = 6,628). A multivariate regression analysis was performed to estimate a link between compensation and the hospitalist model. Results: Of respondents, 66 percent reported the use of hospitalists one year prior to the survey. After controlling for other covariates, hospitalist users were those physicians concerned with patient satisfaction and quality of care associated with compensation, but were less concerned about compensation affecting personal fi nancial performance. Consistent with prior research, we found that hospitalist users were affi liated with managed care and capitation. Discussion: Future research is needed to understand factors improving physician compensation affected by productivity and fi nancial performance of practice. Keywords: insourcing health care, hospitalist, hospital medicine, compensation, quality care, managed care, capitation.

e are now seeing a new trend and Goldman,1 is a physician who specializes within health care where out- in seeing and treating other physicians’ W sourcing is beginning to be re- hospitalized patients in order to minimize placed by insourcing. For decades, medical the number of hospital visits by the patients’ communities have attempted to lower costs regular physicians. Such insourcing appears by reducing the length of hospital stays. Out- effi cient and effective as evidenced by the patient clinics fi rst, emergency care facilities literature.2 Managed care is a system of pro- later, and then day surgery centers recently viding health care (usually by an HMO or a have been part of this outsourcing of health PPO) that is designed to control costs through care movement. Hospices now routinely managed programs in which the physician handle health care outsourcing for near- accepts constraints on the amount charged death patients. However, there are signs that outsourcing health care services has run its Doohee Lee, PhD, is an Associate Professor in the course and trends may be reversing. Many Graduate School of Management, the Elizabeth health care organizations now fi nd it desir- McDowell Lewis College of Business, Marshall Uni- able and economical to have a physician as versity, Charleston, West Virginia. part of staff to reduce work time away from Andrew Sikula, Sr., PhD, is Associate Dean of the the offi ce or place of employment. Elizabeth McDowell Lewis College of Business at The development of hospitalists, managed Marshall University. He is also Director of the Marshall University Graduate School of Management. care, and capitation have made insourcing health care more plausible and perhaps even J Health Care Finance 2010;36(4):17–27 mandatory. A hospitalist, coined by Wachter © 2010 Aspen Publishers

17 18 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

for medical care and the patient is limited in to provide inpatient care. This same report the choice of a physician. Reducing choice indicates over half of US hospitals now have and variation are health care insourcing cost hospital medicine programs. The hospital- reduction tactics and strategy. Capitation also ist model has been widely touted to reduce encourages health care insourcing since it is health care costs and to improve quality of a fi xed per capita payment made periodically care.4 Specifi cally, these benefi ts include to a medical service provider (such as a phy- resource saving, reduced length of stay sician) by a managed care group (such as an (LOS), patient satisfaction, and medical HMO) in return for medical care provided to training and education. 5 A recent study by enrolled individuals. Pham et al . 6 revealed two primary motiva- There are other indicators and measures tors for hospitalist growth: of increased health care insourcing and less physical treatment outsourcing going on in 1. Reduction of LOS and per admission the economy and within society. Increasingly, costs; and terminally ill patients are being cared for at 2. Reimbursement pressures on primary home by family members and friends for care physicians (PCPs) so that they their last surviving months and weeks. Hos- can avoid traveling time and focus on pice personnel may visit home care providers outpatient services. and recipients, but patients staying full-time at hospices today is normally now a practice Notwithstanding the benefi ts and interests reserved for just the last few days of life. of hospitalists, little attention has focused What affects there will be on hospitalist on whether physician compensation is utilization and the health care insourcing associated with the utilization of hospital- movement because of the impending US ists. Understanding the association between federal health care system are indetermi- physician compensation and the hospitalist nable at this time. A prediction is that the model is an important research question for more a government option and one payer at least one reason: The hospitalist model is system results, the greater the expansions of relatively new and hospitals and practicing hospitalists usage and health care insourc- hospitalists must fi nd ways to be competitive ing mandates. What is certain is that there in the market in order to continue expanding will be a dollar trail. “Follow the money” hospital medicine and further market their is always a predictor of past, present, and unique quality services of inpatient care. A future behavior. Health care must make both recent survey of physicians7 shows the larg- physical sense and fi scal cents. As indicated est increases in compensation (7.32 percent) later in this article, physician compensation among hospitalists in the year 2007, sug- is related to hospitalist utilization and the gesting that demand for hospitalists exceeds implementation of both insourcing and out- supply in the current marketplace. The sourcing medical diagnoses, disease treat- compensation increase does not seem to be ment, and health care delivery alternatives. related to service quantity or productivity According to a recent survey, 3 there are of hospitalists. Several recent studies report about 23,000 hospitalists currently practicing fi nancial diffi culties among hospitals utiliz- in US hospital settings as active physicians ing hospitalists.8 A study by Hoff et al .9 also The Utilization of Hospitalists: Insourcing Instead of Outsourcing Health Care 19

revealed most hospitalists (75 percent) in sponsored by the Robert Wood Johnson their study having received no compensation Foundation (RWJF) has been conducting a linked to fi nancial incentives. It is unclear physician survey since 1996. The fi rst CTS whether the increase in compensation is physician survey was conducted in 1996, related to the hospitalist model. followed by a second round in 1998–1999, There are other factors that may affect the and a third effort in 2000–2001. utilization of hospitalists. Quality of care A total of 6,628 physicians in the United may be associated with the use of hospital- States participated in this telephone survey, ists.10 Also, under the current managed care using computer-assisted telephone interview- market system, capitation or the prospective ing technology. The study reports a response payment system is a primary reimbursement rate of 52.4 percent, and all participating method for medical providers and has been physicians received $25 for their time. A list popular for the past two decades. Managed of physicians was provided by the American care may be linked to the hospitalist model Medical Association (AMA) and the Ameri- as Coffman and Rundall11 concluded indicat- can Osteopathic Association (AOA), and the ing that hospitalists under managed care per- survey was conducted between June 2004 form better and generate positive outcomes. and July 2005. Understanding the development of how The study participants included American physician compensation is associated with physicians providing direct patient care for the use of hospitalists has largely gone unex- at least 20 hours a week. Certain physicians plained under the current managed health care excluded from the survey include federal delivery system. Therefore, the present empir- employees, specialists who do not provide ical research seeks to explore, using a nation- direct patient care, foreign medical graduates ally representative data of 6,628 practicing with temporary licenses, residents, interns, physicians, determinants of the use of hospi- fellows, and physicians whose names could talists in relation to physician compensation. not be disclosed to outsiders. The strati- The goal of this analysis is twofold: fi ed random sampling technique was used to study survey participants. More detailed 1. To estimate the prevalence rate of hos- information on the data collection and meth- pitalist utilization among physicians at odology are described elsewhere.13 the national level; and 2. To understand a link between hospi- Measurement talist utilization and physician compensation. Dependent Variable Hospitalist utilization has been increasing Methods in the past decade and understanding what is determining the use of hospitalists remains Data important. This study uses the extent to We analyzed secondary data from a 2004– which hospitalists were utilized in practice 2005 Community Tracking Study (CTS) as the dependent variable. The survey specif- Physician Survey in the United States.12 The ically asks participants, “What percentage of Center for Studying Health System Change your patients who were hospitalized last year 20 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

had a hospitalist involved in their inpatient 3 = $100,000–$149,999; care?” The response was in the range of zero 4 = $150,000–199,999; to 100 percent. 5 = $200,000–249,999; 6 = $250,000–299,999; and Covariate Variables 7 = > $300,000. Untangling the association between compensation and hospitalist utilization is an Control variables in the analysis in- important research question unexplored in cluded: previous studies. Respondents were asked • Age: whether physician compensation is affected 1 = 1940 or earlier; by: 2 = 1941–1945; 1. Own productivity; 3 = 1946–1950; 2. Satisfaction surveys completed by the 4 = 1951–1955; physician’s own patients; 5 = 1956–1960; 3. Specifi c measures of quality care; 6 = 1961–1965; 4. Practice profi ling; and 7 = 1966–1970; and 5. Overall fi nancial performance of one’s 8 = 1971 or later; practice. • Gender (male = 1, female = 2); and • Race (other = 1, white = 2). The response was categorical (no = 0, yes = 1). Given most hospitalists come from inter- The survey measured practice revenue by nal medicine, it is important to control spe- asking each respondent about the percent of cialty to detect interdependent relationships patient care practice revenue coming from: among variables:

1. Medicare; 1 = Internal medicine; 2. Medicaid; 2 = Family/general practice; 3. Capitation (prepaid basis); and 3 = Pediatrics; 4. All managed care. 4 = Medical specialties; 5 = Surgical specialties; The response was in the range of zero to 6 = Psychiatry; and 100 percent. 7 = OB/GYN. Annual income was assessed by the question: “During 2003, what was your own net Specialties thereby were controlled in the income from the practice of medicine to the analysis. nearest $1,000, after expenses but before Analysis taxes?” The response was in the range of 1–7: The statistical software package STATA 10.114 was used for all data analyses. 1 = $0–$49,999; Descriptive analyses were performed to gen- 2 = $50,000–$99,999; erate mean values and standard deviations The Utilization of Hospitalists: Insourcing Instead of Outsourcing Health Care 21

for all variables included in the analysis. A were positively linked to hospitalist usage. multivariate linear regression analysis was Finally, no other variables controlled in the conducted to identify determining factors of study were found signifi cant, except for hospitalist utilization. Several possible con- medical specialty, which was negatively founding factors (age, gender, race, and spe- associated with the use of hospitalists (β = cialty) were included and controlled in the -3.83, p < .001), suggesting that internal regression analysis. Using variance infl ation medicine specialists are more likely to factors (VIFs) statement command available approve the hospitalist model. in STATA, we tested for multicollinearity. No variable had a tolerance value lower than Discussion 0.1., suggesting that all variables analyzed in the regression model are stable. All of the To our knowledge, this analysis is the fi rst data analyses were fully adjusted, using the empirical effort, using a nationally repre- weight variables given in the data, in order to sentative survey data of 6,628 physicians, to represent a national sample. identify determining factors of the utilization of hospitalists in relation to compensation. Results We found an association between compensation and the use of hospitalists. Hospitalist Most respondents were male (72.07 per- users were concerned about compensation cent) and white (77.25 percent). About 34 affected by patient satisfaction and quality percent of the sample reported that they of care. Surprisingly, no prior study has tried did not utilize hospitalists one year prior to particularly to understand the effect of qual- the survey (not shown in fi gures). Figure 1 ity care on hospitalist compensation. Our presents descriptive statistics of the sample. study fi nding may be comparable to prior Figure 2 highlights results of a multivariate research15 that organizations may benefi t linear regression analysis, indicating deter- from utilizing hospitalists to improve qual- minants of the utilization of hospitalists in ity of care, which can be directly associated relation to compensation. Compensation with how physicians get reimbursed. affected by productivity and practice profi l- Some studies validate the importance of ing was not signifi cantly linked to the use of quality care and satisfaction linked to the hospitalists after controlling for other vari- utilization of hospitalists. In a recent study, ables (patient care revenues, annual income, Lopez et al. 16 found hospitals with hospi- gender, race, and specialty). talists better performing on quality indica- Compensation affected by patient satisfac- tors for acute myocardial infarction (AMI), tion survey (β = 5.74, p < .001) and quality of pneumonia, disease treatments and diag- care (β = 6.90, p <. 001) were positively asso- noses, counseling, and prevention. Compar- ciated with hospitalist utilization, whereas ing process and outcomes in relation to the compensation affected by fi nancial perform- inpatient care of 182 pediatric patients, Wells ance was negatively associated with the usage et al. 17 found that patients were more satis- of hospitalists (β = -2.92, p <. 020). Patient fi ed with care rendered by hospitalists. Halp- care revenues from capitation (β = .118, ert and colleagues18 also reported the similar p < .001) and managed care (β = .062, p < .006) fi nding of higher satisfaction among PCPs 22 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Figure 1. Descriptive Statistics of the Sample (n = 6,107)

Variables Number Mean Scores (SD) Minimum Maximum

Use of hospitalists 6,107 29.95 (37.71) 0 100 Compensation affected by productivity 5,046 .70 (.45) 0 1 Compensation affected by satisfaction 5,030 .24 (.43) 0 1 Compensation affected by quality care 5,034 .20 (.40) 0 1 Compensation affected by practice profi ling 5,009 .13 (.34) 0 1 Compensation affected by fi nancial 5,030 .68 (.46) 0 1 performance Patient care revenue from Medicare 6,628 31.62 (22.55) 0 100 Patient care revenue from Medicaid 6,628 16.67 (18.26) 0 100 Patient care revenue from capitation 6,628 13.46 (23.78) 0 100 Annual income* 6,622 4.18 (1.79) 1 7 Patient care revenue from all managed care 6,628 40.56 (28.08) 0 100 Age** 6,628 4.45 (1.92) 1 8 Gender 6,628 1.25 (.43) 1 2 Race/ethnicity 6,535 1.78 (.41) 1 2 Specialty 6,628 3.75 (1.68) 1 7

* A 7-point scale: (1) < $49,999; (2) $50,000–99,999; (3) $100,000–149,999; (4) $150,000–199,999; (5) $200,000–249,999; (6) $250,000–299,999; and (7) > $300,000. ** An 8-point scale: (1) 1940 or earlier; (2) 1941–1945; (3) 1946–1950; (4) 1951–1955; (5) 1956–1960; (6) 1961–1965; (7) 1966–1970; and (8) 1971 or later.

and improved quality of care provided to while productivity remained the same. 19 patients. Researchers of SHM suggested the shortage In our study, hospitalist users were less of hospitalists as a possible explanation of likely to be concerned about compensation rising hospitalist compensation. However, a affected by overall fi nancial performance review of the literature refl ects discrepancy of their practice, which is in line with eco- in understanding whether the hospitalist nomic theory of supply and demand in the model is linked to a favorable personal fi nan- current marketplace where labor demand for cial performance. Several studies report that hospitalists exceeds supply. As an example, many hospitals utilizing hospitalists experi- hospitalist salary has increased by 13 percent ence fi nancial diffi culties. Landrigan et al . 20 during the past two years (2007–2008). found effi ciency of the hospitalist model The 2007 Society of Hospital Medicine ( e.g., cost reduction, reduced LOS), but effi - (SHM) survey reports that hospitalist sala- ciency gains failed to generate revenues for ries increased signifi cantly in recent years the hospitalist programs and most hospitalist The Utilization of Hospitalists: Insourcing Instead of Outsourcing Health Care 23

Figure 2. Results of a Multivariate Regression Analysis on the Utilization of Hospitalists

B SE t P 95% CI Intercept 38.35 4.51 8.50 <.001 (29.50, 47.20) Compensation affected by productivity -1.822 1.27 -1.43 .152 (-4.31, 0.67) Compensation affected by satisfaction 5.74 1.65 3.48 <.001 (2.50, 8.97) Compensation affected by quality care 6.90 1.90 3.62 <.001 (3.16, 10.63) Compensation affected by practice profi ling 2.84 1.96 1.44 .149 (-1.01, 6.69) Compensation affected by fi nancial performance -2.92 1.26 -2.32 .020 (-5.40, -0.45) Patient care revenue from Medicare -0.006 .026 -0.23 .817 (-0.05, 0.04) Patient care revenue from Medicaid .006 .031 .20 .844 (-0.05, 0.06) Patient care revenue from capitation .118 .025 4.68 <.001 (0.06, 0.16) Annual income .271 .34 .78 .434 (-0.40, 0.95) Patient care revenue from all managed care .062 .022 2.77 .006 (0.01, 0.10) Gender 2.452 1.35 1.80 .071 (-0.21, 5.11) Age .069 .31 .22 .823 (-0.53, 0.67) Race -0.682 1.41 -0.48 .630 (-3.4, 2.09) Specialty -3.837 .36 -10.64 <.001 (-4.54, -3.13) R2 .074 Adjusted R2 .071 N 4,457

programs in their review faced fi nancial extensive review of the literature, Wachter diffi culties. A recent study by Tieder et al .21 and Goldman23 reported the fi nding of cost also reports a similar fi nding that pediatric reductions by utilizing hospitalists. hospitalist programs in a community hospital One core fi nding is that the role of experienced a substantial fi nancial defi cit. managed care is signifi cant in predicting Based on a systematic review of fi nancial hospitalist system patient care revenue. performance among pediatric hospitalists, This may be due to at least two reasons. another national survey22 reported that only The very fi rst hospitalist program started 12 percent of hospitalists were compensated in the high HMO penetration market 24 and through a model of 100 percent productivity thereby naturally developed with the man- incentives. aged care system. In addition to hospitalist Contrary to the aforementioned studies historical evolvement with managed care, negating the hospitalist model in fi nancial several studies support the linkage of man- performance, a number of different studies aged care affi liation with the usage of hos- support signifi cant reductions in resource pitalists. Harrison and Ogniewski 25 reported use ( e.g., hospital costs and LOS). In an that organizations utilizing hospitalists are 24 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

located in communities with higher HMO excluded from the survey. These physicians penetration. Molinari and Short26 revealed include federal employees, specialists who hospitalists being effi cient in managing do not provide direct patient care, foreign HMO patients. In a retrospective cohort medical graduates with temporary licenses, study to understand effi ciency in a managed residents, interns and fellows, and physicians care setting, Srivastava and colleagues27 whose names could not be disclosed to out- report hospitalists being effi cient and reduc- siders. Our fi ndings also may not represent ing costs in an HMO (e.g ., reduced LOS all hospitalists in practice as we assessed from two days to one day, which resulted only physicians who utilize hospitalists in an average cost-per-case reduction of instead of practicing hospitalists. Thus, cau- $105.51 for asthma patients). tion is needed when projecting our fi ndings Capitation-based patient revenue was also to practicing hospitalists. Finally, this study signifi cant in predicting the utilization of is cross-sectional and hence it did not inves- hospitalists, which can be explained by the tigate the causal relationship between phy- fact that capitation is a primary reimburse- sician compensation and hospitalist use. A ment method for physician services under longitudinal study may more correctly esti- the managed care delivery system. In our mate the causation between the two. study, nearly half (48 percent) of respondents In conclusion, this analysis extends the reported that their patient care revenue came hospitalist literature by investigating the from capitation or a fi xed payment system, association between physician compensa- which is in line with Coffman and Rundall.28 tion and the hospitalist model. We found In their careful review of the literature, Coff- that hospitalist users were concerned with man and Rundall concluded that hospitals compensation in relation to quality care under capitation would benefi t from using and satisfaction, but they were less likely hospitalists because hospitals maximize rev- to be concerned with compensation linked enue by reducing LOS. However, this fi nding to fi nancial performance and productivity. is not in line with Lindenauer et al .29 whose This may pose a threat to many hospitals study revealed that salary was the most com- with hospitalists as the Center for Medicare mon method of reimbursement and only & Medicaid Services (CMS) looks for evi- 3.6 percent of the surveyed hospitalist were dence-based practices and emphasizes the reimbursed based on capitation. The differ- importance of pay-for-performance (P4P). 31 ence may be explained by several factors A future study may benefi t from exploring including the rationale of the study, study alternatives of improving fi nancial per- design, and sample size. Particularly, practic- formance and productivity in the context ing hospitalists were studied by Lindenauer of hospitalist compensation. Our study also and colleagues,30 whereas hospitalist-using documents that managed care and personal physicians were assessed in our study. physician capitation are associated with This study has several limitations. Our the utilization of hospitalists. These afore- study is subject to response bias particularly mentioned fi ndings are important keys to as we measured respondent perceptions understanding the determinants and utiliza- which by nature are biased. Our fi ndings tion of hospitalists and their relationships may not represent those physicians who are to individual physician compensation. The Utilization of Hospitalists: Insourcing Instead of Outsourcing Health Care 25

REFERENCES

1. Wachter, RM, Goldman, L, “The Emerging 5. Kripalani, S, Pope, AC, Rask, K, Hunt, K, Role of ‘Hospitalists’ in the American Health- Dressler, DD, Branch, WT, Zhang, R, Wil- care System,” New England Journal of Medi- liams, MV, “Hospitalists as Teachers,” Journal cine, 335, 514–517 (1996). of General Internal Medicine, 19(1), 8–15 2. Halpert, AP, Pearson, SD, LeWine, HE, Mck- (2004). ean, SC, “The Impact of an Inpatient Physician 6. Pham, HH, Grossman, JM, Cohen, G, Boden- Program on Quality, Utilization, and Satisfac- helmer, T, “Hospitalists and Care Transitions: tion,” American Journal of Managed Care , The Divorce of Inpatient and Outpatient Care,” 6(5), 549–555 (2000); Kaboli, PJ, Barnett, MJ, Health Affairs, 27(5), 1315–1327 (2008). Rosenthal, GE, “Associations with Reduced 7. American Medical Group Association, 2008, Length of Stay and Costs on an Academic “Medical Group Compensation and Finan- Hospitalist Service,” American Journal of cial Survey,” retrieved January 11, 2010, from Managed Care, 10, 561–568 (2004); Wachter, http://www.amga.org/AboutAMGA/News/ RM, Goldman, L, “The Hospitalist Move- article_news.asp?k=279. ment 5 Years Later,” Journal of the American 8. Landrigan, CP, Conway, PH, Edwards, S, Medical Association, 287(4), 487–494 Srivastava, R, “Pediatric Hospitalists: A Sys- (2002). tematic Review of the Literature,” Pediatrics, 3. American Hospital Association, 2007, 117(5), 1736–1744 (2006); Tieder, JS, Migita, retrieved Jan. 4, 2010, from http://www. DS, Cowan, CA, and Melzer, SM, “Newborn hospitalmedicine.org/AM/Template.cfm? Care by Pediatric Hospitalists in a Commu- Section=Press_Releases&Template=/CM/Content nity Hospital: Effect on Physician Productivity Display.cfm&ContentID=21258. and Financial Performance,”Archives of Pedi- 4. Wachter, RM, Goldman, L, “The Hospital- atrics & Adolescent Medicine , 162(1), 74–78 ist Movement 5 Years Later,” Journal of the (2008). American Medical Association, 287(4), 487– 9. Hoff, TH, Whitcomb, WF, Willams, K, Nelson, 494. (2002); Lindenauer, PK, Rothberg, MB, JR, Cheesman, RA, “Characteristics and Work Pekow, PS, Kenwood, C, Benjamin, EM, Auer- Experiences of Hospitalists in the United bach, AD, “Outcomes of Care by Hospitalists, States,” Archives of Internal Medicine, 161(6), General Internists, and Family Physicians,” 851–858 (2001). New England Journal of Medicine, 357(25), 10. Lopez, L, Hicks, LS, Cohen, AP, McKean, S, 2589–2600 (2007); Halasyamani, LK, Valen- Weissman, JS, “Hospitalists and the Quality of stein, PN, Friendlander, MP, Cowen, ME, “A Care in Hospitals,” Archives of Internal Medi- Comparison of Two Hospitalist Models with cine, 169(15), 1389–1394 (2009); Wells, RD, Traditional Care in a Community Teaching Dahl, B, Wilson, SD, “Pediatric Hospitalists: Hospital,” American Journal of Medicine, Quality Care for the Underserved?” American 118(5), 536–543 (2005); Kripalani, S, Pope, Journal of Medical Quality, 16(5), 174–180 AC, Rask, K, Hunt, K, Dressler, DD, Branch, (2001). WT, Zhang, R, Williams, MV, “Hospitalists as 11. Coffman, J, Rundall, TG, “The Impact of Hos- Teachers,” Journal of General Internal Medi- pitalists on the Cost and Quality of Inpatient cine, 19(1), 8–15 (2004); Lopez, L, Hicks, LS, Care in the U.S.: A Research Synthesis,” Med- Cohen, AP, McKean, S, Weissman, JS, “Hos- ical Care Research and Review, 62(4), 379– pitalists and the Quality of Care in Hospitals,” 406 (2005). Archives of Internal Medicine, 169(15), 1389– 12. Community Tracking Study Physician Survey 1394 (2009); Hackner, D, Tu, G, Braunstein, Public Use File: User’s Guide, 2004–05, Tech- GD, Ault, M, Weingarten, S, Mohsenifar, Z, nical Publication No. 64, Center for Studying 2001, “The Value of a Hospitalist Service: Efﬁ - Health System Change, Washington, D.C. cient Care for the Aging Population?” Chest, (August 2006). 119(2), 580–589 (2001). 13. Id. 26 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

14. Stata Corporation, College Station, TX (2007). American Medical Association, 287(4), 487– 15. Supra, n.10. 494 (2002). 16. Lopez, L, Hicks, LS, Cohen, AP, McKean, S, 24. Freese, RB, 1999, “The Park Nicollet Experi- Weissman, JS, “Hospitalists and the Quality ence in Establishing a Hospitalist System,” of Care in Hospitals,” Archives of Internal Annals of Internal Medicine, 130, 350–354. Medicine, 169(15), 1389–1394 (2009). 25. Harrison, JP, Ogniewski, RJ, “The Hospitalist 17. Wells, RD, Dahl, B, Wilson, SD, “Pediat- Model: A Strategy for Success in US Hospi- ric Hospitalists: Quality Care for the Under- tals?” The Health Care Manager, 23(4), 310– served?” American Journal of Medical Quality, 317 (2004). 16(5), 174–180 (2001). 26. Molinari, C, Short, R, “Effects of an HMO 18. Halpert, AP, Pearson, SD, LeWine, HE, Mc- Hospitalist Program on Inpatient Utilization,” kean, SC, “The Impact of an Inpatient Phy- American Journal of Managed Care, 7(11), sician Program on Quality, Utilization, and 1051–1057 (1999). Satisfaction,” American Journal of Managed 27. Srivastava, R, Landrigan, CP, Ross-Degnan, Care , 6(5), 549–555 (2000). D, Soumerai, SB, Homer, CJ, Goldmann, DA, 19. Society of Hospital Medicine (SHM), retrieved Muret-Wagstaff, S, “Impact of a Hospitalist Jan. 11, 2009, from http://www.the-hospitalist. System on Length of Stay and Cost for Chil- org/details/article/186953/Hospitalist_Pay_ dren with Common Conditions,” Pediatrics , Up_Productivity_Steady_in_SHMs_Latest_ 120(2), 267–274 (2007). Survey.html. 28. Coffman, J, Rundall, TG, “The Impact of Hos- 20. Landrigan, CP, Conway, PH, Edwards, S, pitalists on the Cost and Quality of Inpatient Srivastava, R, “Pediatric Hospitalists: A Sys- Care in the U.S.: A Research Synthesis,” Med- tematic Review of the Literature,” Pediatrics, ical Care Research and Review, 62(4), 379– 117(5), 1736–1744 (2006). 406 (2005). 21. Tieder, JS, Migita, DS, Cowan, CA, and Melzer, 29. Lindenauer, PK, Pantilat, SZ, Katz, PP, Wachter, SM, “Newborn Care by Pediatric Hospitalists RM, “Hospitalists and the Practice of Inpa- in a Community Hospital: Effect on Physician tient Medicine: Results of a Survey of the Productivity and Financial Performance,” National Association of Inpatient Physicians,” Archives of Pediatrics & Adolescent Medi- Annals of Internal Medicine, 130, 343–349 cine , 162(1), 74–78 (2008). (1999). 22. Society of Hospital Medicine (SHM), News for 30. Id. Immediate Release (2004 Survey), retrieved 31. Tanenbaum, SJ, “Pay for Performance in Jan. 16, 2009, from www.hospitalmedicine.org. Medicare: Evidentiary Irony and the Politics 23. Wachter, RM, Goldman, L, “The Hospital- of Value,” Journal of Health Politics, Policy ist Movement 5 Years Later,” Journal of the and Law, 34(5), 717–746 (2009). Hospital Productivity and Information Technology

Steven R. Eastaugh

Information technology and linear programming help to control hospital costs without harming service quality or staff morale. This study presents production function results from a study of hospital output during the period from 2005 to 2008. The results suggest that productivity varies widely among the 58 hospitals as a function of stafﬁ ng patterns, methods of organization, and the degree of reliance on information support systems. Information technology (IT) can enhance the marginal value product of nurses and staff, so that they concentrate their workday around patient care activities. Financial incentives also help to enhance productivity. Incentive pay for nurses based on productivity gains is associated with improved productivity. One should get the greatest output for the least input effort, better balancing all factors of service delivery to achieve the most with the smallest resource effort. Key words: productivity, information technology (IT), nurse scheduling, productivity measures, incentive pay.

ask delegation and the allocation According to Peter Drucker, productiv- of staff within the hospital have ity is the fi rst test of management’s compe- T become major fi nancial issues. The tence.4 A hospital manager should get the traditional productivity assessment empha- greatest output for the least input effort, bet- sizes tasks, activities, and technical effi - ter balancing all factors of care delivery to ciency. A more global vision of productivity achieve the most with an optimal level of asks: What support systems can improve quality. A number of exogenous factors can productivity? Support systems that enhance affect productivity. For example, nurse staff productivity include: automated scheduling fl exibility is enhanced during tight low-wage systems utilizing linear programming and infl ation periods. When nurse wage infl a- electronic health records (EHRs). President tion is under control, nurses are less likely to Obama has pledged $50 billion over fi ve protest productivity improvement programs. years to help fund the transition to “stand- Implementing effi cient nurse scheduling ards-based” electronic health information systems and work-unit reorganizations, systems.1 Scheduling systems and EHRs can transform our health care system by making it safer, more effi cient, and more Steven R. Eastaugh is Professor of Finance and cost-effective. Information technology (IT) Health Economics, Department of Health Services support has the potential to improve qual- Management and Policy, George Washington Univer- sity, Washington, DC. ity, enhance staff morale, and reduce costs. Acknowledgments: I thank DHHS 100-94-0155 for The Institute of Medicine reports that EHRs the initial fi nancing of this project. I also wish to thank can reduce more than half the 1.5 million Warren Greenberg, Craig Rosenfeld, Amit Nanda, Americans injured every year by prescribing Phillip Reeves, and Ruth Hanft for helpful comments errors.2 In July 2008, the Congress passed a and Juan Acevedo, Luline Almonacy, Paymon Bagh- law providing Medicare bonuses to physi- eri, Tom Davy, Michelle Donovan, Oshrat Goldberg, cians who use electronic prescribing, and for Bonnie Horvath, Aniysha Nelpurackal, Janet Nug- gent, and Rick Tischler for their assistance in data penalties beginning in 2012 to those who do collection and analysis. not. The Department of Health and Human

especially when reinforced by an incentive- remedies, and holding and extending per- pay plan, can reduce costs signifi cantly. Pro- formance gains. The job of quality control ductive managers must be change agents, never ends because improvement criteria not overcommitted to the existing ways of require effort to maintain the new high levels doing things. The hidebound, tradition- of performance and productivity. By avoiding based hospital that does not adopt IT support mistakes and useless units of activity, gains in systems will not survive. Scheduling is key productivity occur as quality improves. in productivity enhancement. Three critical Hospital staffi ng ratios should be set in actors—the patient, the employee, and the proportion to forecast workloads. In sched- physician—must be scheduled for improved uling nurses, for example, if the workload on productivity. Better scheduling of all three day shifts is three times as high as that on groups can reduce unnecessary activity fl ow, night shifts, it would be illogical to provide reduce unit costs, improve patient satisfac- the same number of nurses over all three tion, and reduce waiting time for both pro- shifts. Similarly, even when workload is viders and patients. unscheduled, as in the emergency room, pat- Scheduling systems for physicians can terns of utilization are predictable. A sample reduce costs through reducing downtime survey over a few months demonstrates the (wasted time). When an operating room days and shifts that have the highest work- (OR) mishandles case scheduling, cost over- load, and staffi ng should be proportionate to runs result from either underutilization or the predicted demand. Additional adjustment overtime wages. The University of Michigan for seasonal changes and case-mix severity Medical Center has been a pioneer in OR can be made.5 scheduling. Operating room scheduling effi - A basic requirement of a scheduling system ciency is improved by scheduling on the basis is that it preserve morale and meet the per- of patient DRG, severity level, and surgeon sonal needs of employees for days off, vaca- (speed based on past experience). Schedul- tions, birthdays, and holidays. In addition, ing policies were made uniform across all employees must believe that the scheduling departments, and a service-specifi c sched- process is fair and impartial. Unfortunately, uler was assigned in each area (one person many hospitals use manual scheduling sys- for cardiac surgery, one for ophthalmology, tems that are unresponsive to subtle shifts and one for general surgery). The specialist in workload and that are perceived as being in scheduling knows how to collect informa- unfair. It is amazing that personnel are still tion, stagger schedule time blocks, minimize manually scheduled in an industry that spends misscheduled cases, call surgeons if neces- more than fi ve billion dollars each week. For sary, and collectively make optimal use of example, if 12 nurses are scheduled over a the 33 OR suites. (They experienced only 10 month so that each nurse works 22 days, percent downtime.) disregarding all constraints, there would be Productivity improvement involves more 1.5 million schedules possible.6 It is hard to than utilizing linear programming to set imagine that human ingenuity, even under schedules. The department manager and the no other constraints, could develop the best scheduler emphasize identifying problem schedule. A computerized scheduling system areas, team building, implementing successful can select the best schedule without hours of Hospital Productivity and Information Technology 29

paperwork, hassles, and appeals. The com- • Support clinical decisions, including puter can provide convincing documenta- warnings about drug interactions or tion of fairness, demonstrating that weekend contradictions. assignments and shift changes (AM to PM to nights) have been equalized. Frequently, one A basic EHR system is one that allows just fi nds capable employees being promoted to some of the fi rst three of these functions. The “scheduler” without even having been taught fourth function of an ideal EHR is computer- the importance of, or techniques for, effi cient ized physician order entry, or CPOE. When scheduling.7 Computer-generated schedules a physician uses CPOE to enter a prescrip- are guided by effi ciency and quality, and not tion, the system alerts him or her to poten- by interpersonal relationships. tial interactions with other drugs the patient Nurses should be allocated to account for is taking. Common dosages, contradictions acuity of the level of care a patient needs. such as pregnancy, and patients’ allergies are The goal is acuity-driven workload staffi ng, also fl agged. Goals set by the federal gov- not merely census-driven staffi ng. However, ernment call for EHRs to be standardized even the most refi ned industrial engineering and interoperable, meaning that multiple methodology can be undercut if manage- clinics and hospitals should be able to access ment eschews fl exibility in favor of a fi xed and update them as patients seek treatment decision rule; for example, that each nurse at multiple locations. 9 The 2005 Rand Cor- is allowed every other weekend off. Such a poration Study suggests reducing 404,000 rule results in overstaffi ng on the weekend unnecessary deaths through EHR improve- or on one or two of the weekdays. Nurs- ments, disease management, and prevention ing costs and morale can be improved by would save hospitals $51.7 billion.10 forecasting nontraditional staffi ng arrange- Nurse scheduling systems set staffi ng pat- ments.8 Flexible use of part-time staff and terns for registered nurses (RNs) and nurse combinations of two 12-hour shifts and two extenders (NEs). NE technicians became eight-hour shifts for some full-time nurses popular because the hospital sector experi- can ensure a better match between workload enced diffi culty in fi nding a suffi cient sup- and FTEs. ply of RNs for primary nursing staffs. Some nursing groups were not receptive to the NE concept because of fears that it represented Background a return to team nursing and under trained licensed practical nurses (LPNs) with a new A fully functional EHR system is defi ned job title. 11 However, task delegation to NEs as having the capability to: by itself does not undermine the standardiza- • Record patients’ clinical and demo- tion of nurse education. In fact, the realiza- graphic data; tion that the nation needs more caregivers • View and manage results of laboratory and that NEs will still be under the control of tests and imaging; the nursing department prompted the nursing • Manage order entry, including elec- literature to become less militant. Now an NE tronic prescription and the ability to is referred to in the literature as a “technical order tests and imaging; and assistant to an experienced RN as a primary 30 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

partnership,” or an “executive administra- complement or substitute for one another in tive assistant assisting the executive nurse.” 12 the service production process. A number of Such glowing titles may seem unimportant to recent studies have analyzed production func- economists, but in the workplace it is impor- tions in business and in the hospital sector.15 tant for job retention that NEs not be labeled The fi rst major study of American hospital reborn LPNs who do “scutwork” or “menial production functions involved a sample of tasks.” One profession’s menial task is another 60 Ohio nonteaching hospitals. Hellinger16 profession’s vital activity, so NEs spend most utilized a translog (transcendental logarith- of their workday performing a “noninter- mic) production function, which attenuates pretive” collection of vital signs, EKGs, lab or eliminates restrictions on the functional slips, and paperwork. One last factor that can form, thereby leaving as much general- impact productivity is outsourcing. ity and fl exibility in the service production We have seen a climb up the value chain estimation process (in contrast to the tradi- of services in recent years—from back offi ce tional Cobb-Douglas model). The translog support functions to what the industry calls form used in this study involves two basic “knowledge outsourcing.” Knowledge out- assumptions. First, managers monitor nurs- sourcing includes the offshore physician in ing costs when deciding the appropriate staff India offering diagnostic services— particular mix and range or level of hospital output imaging, such as X-rays and mammograms— and nurse workload. This assumption does and consultation specialists.13 Teleradiology not mean that nurse managers are perfect in particular, in which X-rays are taken at cost-minimizers operating at the production one location and then transmitted to doc- possibility frontier of 100 percent technical tors at another site can enhance productivity effi ciency. The second assumption is that and cut costs. Radiology images taken in the nursing departments exhibit constant returns middle of the night are still read right away to scale in producing their output. Conse- by a wide-awake radiologist working at the quently, there is no reason to presuppose that height of his or her own powers. In most nurses are any more productive in a 900-bed cases, the offshore clinicians are trained in hospital than 90-bed hospital. Previous hos- the United States. A second area where off- pital cost studies, not focused on the nursing shore outsourcing is working is in American department, report very shallow economies hospitals in hardware network management of scale of only 8 percent.17 and engineering design.14 In comparing isoquants—curves producing the same output for different quantities Data and Methods of inputs—two extreme situations can exist. Under perfect complementary production Production-function studies of technical between inputs, no substitution at all is pos- effi ciency (productivity) have been done by sible between inputs A and B, and inputs A economists since the 1930s. Production func- and B must always be used in fi xed proportions are useful to understand how resources tions (isoquants are straight downward slop- are combined by the department or fi rm ing lines). Under the opposite extreme, perfect (hospital) to produce some particular level substitutability between inputs defi nes the iso- of output and ascertain how these resources quants as perfect right angles. In the fi rst step Hospital Productivity and Information Technology 31

in the data analysis a translog production func- One last caveat must be presented con- tion will be estimated from data at 58 hospitals. cerning measurement error in this study: The second step measures the curvature of the measurement of capital inputs must avoid nursing isoquants and thereby the substitution the pitfalls of using depreciation charges to among inputs (the elasticity of substitution). more accurately reﬂ ect differences in the age Since nursing is a complex production and productivity of the capital stock. I have process, we will be assessing a production used the same index employed in a previous process with six-dimensional isoquants. study to adjust the capital expenses for dif- Between each pair of inputs partial elastici- ferences across the 58 sample hospitals in ties of substitution will be measured (e.g ., the average age of their capital stock.19 F o r RN x NE substitution). The six basic inputs each hospital the ratio of accumulated depre- studied include: ciation to total assets is taken as a measure of age. Age-adjusted capital input was cal- NE = nurse extender; culated as follows: RN = registered nurses; H = housestaff/residents and interns per- E = UA x Exp (M - R) (1) forming some nursing activities while understaffed; Where UA = unadjusted capital expenses A = clerks, LPNs, and nurse aides; R = ratio of accumulated depreciation to E = capital; and total assets ElecF = inputs for Electronic Health M = mean value of R for the sample Records, outsourcing radiology, Exp = inverse natural logarithm. and outsourcing laboratory tests. The sample is a convenience sample of Collection of data on labor inputs is hospitals with nursing activity research straightforward and has been done in a programs. Obviously, the sample is not number of previous studies. Nursing output generalizable to all American hospitals. is speciﬁ ed by a point-scoring system sold The more progressive hospitals, with active by the largest proprietary vendor of nurse support for health services research, may workload and nurse scheduling systems.18 have production technologies (scheduling This same system tracks work hours to meas- and staff education) that are more advanced ure the contribution of nonphysician labor than the average American hospital. Each inputs (input factors 1, 2, and 4). Housestaff/ of the sample hospitals subscribed to the resident and intern input was not measured same nurse workload system, and the hos- on an annual basis, but in two years, 2005 pitals ranged in size from 106 to 904 beds. to 2008. Filled residency slots have been The hypothetical frontier production can be largely time-invariant for the 16 sample expressed as: teaching hospitals, and physician labor in yij = II(Xijk) βk euij (2) nursing activities only ranges from 0.1 to 1.1 percent of nursing activities. To not include k this measured work input in the analysis where yij = the nurse output of the jth hos- would slightly overstate the productivity of pital in the ith period for peri- nursing departments in certain hospitals. ods 1–4 (2002–2005) 32 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Xijk = kth input applied by the jth hospital Empirical Results in the ith period. Maximum likelihood methods of esti- If the hospital realized its full techni- mation were applied to equation 4 and the cal effi ciency at 100 percent, then uj takes parameter estimates of the translog model the value zero, and if not, uj takes a value are presented in Figure 1. The ratio of hos- less than zero depending on the extent of pital specifi c variability in productivity was its lost productivity. The euij term provides signifi cant at the .1 level, indicating that a measure of hospital specifi c productivity, productivity dominates in explaining the and improvement in euij shall be refl ected in total variability of nurse output produced. higher mean productivity over time. Ineffi - Judging by the signifi cance of the dummy ciency is: variables, we can reject the hypothesis that Uij = ln yij – (Σβk ln Xijk + vij) (3) productivity was time-invariant over the four years. Most of the parameters not involving Estimation of uj and euj is possible once the two weakest variables (H and E) are sig- density functions for u and v are assumed. nifi cant at the .05 level. Let u follow a half-normal distribution and A second alternative partial elasticity v follow the full normal distribution. (The can also be derived. The Allen elasticity of validity of the half-normal distribution was substitution holds constant the quantities verifi ed at the end of the analysis by plotting of all other inputs, in addition to the level the combined residual (u + v), the hospital’s of nurse output. The Allen elasticities are technical effi ciency and the output levels.) related econometrically to the cross-price Equation 2 can be rewritten as elasticity of demand for factors; for example, the demand for input 1 (NEs) to change Yij=II (Xijk) βk eEij (4) the price of input 2 (RNs). The sign of a cross-price elasticity of demand (column 3 k of Figure 2) by itself is an indicator of where Eij =uj + vij. gross substitution—a negative sign indicating complementary factors, a positive sign The estimation of the maximum possible indicating substitution. As line 15 of Figure stochastic output, had the hospital realized 2 reveals, a negative sign on the elastic- its full technical effi ciency, is carried out by ity of demand for NE labor with respect to applying maximum likelihood methods20 to the price of RN labor indicates that as RN equation 4. With this model one can estimate labor becomes more costly, the labor of NEs individual hospital technical effi ciencies is used less extensively in place of RNs. together with the mean technical effi ciency On the positive side this suggests that NEs using four years of panel data (dummy vari- and RNs are complementary team mem- able D (0,1) for the three years after the base bers, not in competition with each other. year 2005). One can hopefully also target On the other hand, this suggests that a rap- some factors causing variation in technical idly infl ating and costly all-RN staff trades effi ciencies in nursing among the 58 sample effi ciency by avoiding the opportunity for hospitals. NE-induced productivity gains. Moreover, Hospital Productivity and Information Technology 33

Figure 1. Translog Production using nonemployee RNs, the temporary Function for Inpatient Hospital agency nurses, can cost many urban hospi- Services Delivery, 2005–2008 tals up to $60 to $110 per hour. The NEs substitute fairly well and fl u- Parameter Estimate* idly for clerks and LPNs (line 16 of Fig- Variable (Maximum Likelihood) ure 2) while complementing RNs. A posi- D1, 2006 0.028 tive sign in line 12 (see Figure 2) on the (17.8) elasticity of demand for NE labor with D2, 2007 0.039 respect to the price of household (resident) (30.1) labor indicates that, as housestaff labor (H) D3, 2008 0.043 becomes more costly per hour, the labor (38.8) of NEs is used more extensively in place ßElecF/RN 0.104 of residents. As some state regulators and (26.9) hospital managers have moved to restrict ßElecF/H 0.133 (41.5) housestaff work week—fewer hours at the ßElecF/A 1.154 same fi xed annual wage—this raises the (47.8) hourly wage of the housestaff and raises ßNE,E 0.130 the employment level of NEs. However, the (16.2) negative sign in line 13 of Figure 2 reveals ßH,E 0.122 that no increase in RN employment can be (14.9) expected as same states implement a maxi- ßH,NE 0.143 mum hourly workweek for housestaff/resi- (34.8) dents and interns. ßRN,H -0.486 (13.9) Lines 10, 11, 14, and 19 in Figure 2 ßRN,E 0.180 have the expected positive signs, indicat- (19.1) ing that labor can substitute for capital (.01 ßRN,NE -0.159 level of signifi cance). Line 11 has the high- (35.8) est observed elasticity, suggesting that the ßA,NE 0.155 highly skilled MD component of housestaff/ (20.5) residents—their technical diagnostic skill ßA,H 0.019 as doctors—partially substitutes for more (2.7) equipment and physical capital. This general- ßA,RN -0.030 ization may be increasingly true in the future (10.6) ßA,E 0.068 as more residents benefi t from economic (11.7) grand rounds, “think before ordering tests” Constant α 0.043 educational programs, and the cost-effective (9.2) clinical decision-making ethic of younger doctors trained in health economics. *t-values in parentheses; log likelihood = - 38.618 From observing the three dummy varia- Note: NE = nurse extenders, RNs; H = house staff/residents and interns; A = clerks, LPNs, bles at the top of Figure 1, more productivity and nurse aides; E = capital; ElecF = electronic for this sample of 58 hospitals was not time- health records and outsourcing (lab and radiology). invariant over the four-year period. Mean 34 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Figure 2. Partial Elasticities of Substitution for the Input Factors of Inpatient Hospital Service Delivery, 2005–2008

Input Pairs Partial Elasticity Relationship 1. NE/NE -0.189* Complements 2. RN/RN -0.162* Complements 3. H/H -0.170* Complements 4. A/A -0.125* Complements 5. E/E -0.254* Complements 6. ElecF/ElecF -0.228* Complements 7. ElecF/A 0.512 Substitutes 8. ElecF/H 0.348 Substitutes 9. ElecF/RN 0.193 Substitutes 10. NE/E 0.156 Substitutes 11. H/E 0.332 Substitutes 12. H/NE 0.579 Substitutes 13. RN/H -0.352 Complements 14. RN/E 0.385 Substitutes 15. RN/NE -0.438 Complements 16. A/NE 0.518 Substitutes 17. A/H -0.039 Complements 18. A/RN -0.240 Complements 19. A/E 0.081 Substitutes 20. NE/ElecF 0.057 Substitutes

*The own-price elasticities have the expected negative sign. Note: NE = nurse extenders, RNs; H = housestaff/residents and interns; A = clerks, LPNs, and nurse aides; E = capital; ElecF = electronic health records and outsourcing (lab and radiology). nurse productivity for each cross-section and the factor input (NEs) with the two equation improved from .74 to .81 from highest t-values (from Figure 1). Individ- 2005 to 2008. In the most recent year, nurs- ual nurse productivity ratings range from ing departments were realizing only .81 of .61 to .94. Figure 3 suggests discrete dif- their technical effi ciency (productivity) and ferences in production technologies as while averages are interesting, distributions well as differences in input mix. This wide are more policy relevant. range could in theory refl ect differences Figure 3 lists the average productivity in organizational effi ciency or differences level across the 58 nursing departments in the availability and use of factor inputs Hospital Productivity and Information Technology 35

Figure 3. Frequency Distribution of 3. A shortage of nurses did not drag Inpatient Floor Nursing Productivity down productivity levels in Figure 3 as and Nurse Extender Staffi ng-Mix, 2008 the six cities with the tightest nursing markets contained the nine hospitals Productivity Number of Ratio of NEs with the highest level of productivity; Level (Range)* Hospitals to RNs and 0.61–0.65 5 0.0 4. Employment of NEs reduces wasted 0.66–0.70 6 0.0 labor and enhances productivity. 0.71–0.75 7 0.29 0.76–0.80 9 0.53 The last of these four conclusions indicates 0.81–0.85 10 0.65 a number of avenues for future research. For example, the results at the end of the last col- 0.86–0.90 11 0.84 umn in Figure 3 weakly indicate that NEs, as 0.91–0.94 10 0.85 with any labor input, may approach a level Total 58 mean = 0.57 of diminishing returns. For example: * 0.81 = average productivity. None of the 28 hospitals with below average nurse • Does having eight to ten NEs per ten productivity offer incentive pay to nurses. RNs constitute a zone of diminishing returns? • Does a primary care nursing staff with (e.g ., a shortage of nurses). However, the greater than 70 percent BScN-trained eight hospitals with the worst nursing pro- RNs constitute an ineffi cient staff-mix ductivity at the top of Figure 3 employed of diminishing returns? no NE technicians, operated a 100 percent • Does deploying fi ve to eight NEs per RN primary care nursing organization, ten RNs harm patient care quality? and exhibited productivity 10 percent to 22 percent below average. The 20 hos- Judging by the deployment of NEs at pres- pitals in Figure 3 with the highest levels tigious teaching hospitals, task delegation 22 of nurse productivity made heavy use of can enhance the quality of patient care. NEs: nine used the team nursing organiza- Finally, what additional tasks can be del- tional concept, but 11 employed primary egated to NEs beyond obtaining vital signs care nursing with a 60 percent to 65 per- and EKG results, patient transport, procur- cent BScN-trained RN staff.21 ing supplies and equipment, procedural assistance, and paperwork ( e.g., lab slips)? Discussion and Conclusions Most of the 47 hospitals utilizing NEs have begun to utilize specialist technicians to In summary, the results suggest that: dress wounds, monitor pumps and catheters, administer tube and IV feedings, and assess 1. Primary care nursing can be either physical conditions. Progressive nurse man- highly productive or ineffi cient; agers will participate in careful studies to set 2. The all-RN staff, used in only 11 of the standards, study task delegation feasibility, 58 hospitals, reports the worst produc- and circumscribe the job descriptions for tivity performance; NE technicians.23 36 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Some nurse administrators believe that the hospitals in this current study. Consider reducing staff results in employees working the medication cycle times. The time it takes harder and translates into lower employee for an order to be fi lled and administered morale. Indeed, it is true that if the staff cuts to the patient was reduced in 2009 to eight appear abrupt and arbitrary and offer no minutes in one department from 84 minutes. incentive “carrot” to maintain performance, Physicians enjoy the productivity gains. morale might decline and the most outstand- After physicians make hospital rounds in the ing workers might look elsewhere for job morning, they do not have to call the nurse security. Staff cuts can improve morale, how- in the afternoon or at night to see how a par- ever, if employees share in the benefi ts of cost ticular patient is doing. They can look it up reduction and understand the new incentives themselves on the computer and see the cur- and why things must change. Sixteen of the rent patient information. top 21 high productivity nursing departments Physicians desire an integrated delivery in Figure 3 utilized productivity-based incen- system so the manner in which they enter tive pay systems to reward nurses. Imple- an order is the same in both locations (pri- mentation of incentives, following a study of vate offi ce, hospital). Pharmacists also love task requirements, common skills, job train- EHRs because the typical review time is ing, fl exible staffi ng, and scheduling, causes trimmed from 60 minutes to 15 minutes. the total organization to focus on making a Moreover, inpatient lab tests can be reduced given line of work most productive. With the by 10 percent due to elimination of dupli- help of incentives, a hospital can establish cate orders. and maintain a dedication to high level of The fi ve hospitals in this study that out- productivity and quality service. source radiology in Asia do not report a sig- At a time when hospital care is becoming nifi cant gain in department productivity. If more complex and patients are becoming all hospitals had a standard set of productiv- sicker, productivity enhancement is critical. ity measures in radiology, future researchers With future funding limitations, “barebones” could better focus on the effi cacy of out- reimbursement dictates that the recent tra- sourcing. dition of a 100-percent RN primary care Health care is the largest sector of the nursing must be abandoned. Development American economy. Thanks to methods to of an effi cient staff-mix criterion in nurs- enhance productivity, hospital productiv- ing should enhance nursing’s rising sense of ity gains have been better than average. In professionalism.24 Maximizing RN hospital the years 2000 to 2008 average productivity employment levels is hardly a desirable or grew 21.8 percent; and the real income of economical goal unless America has a gross working-age households declined 3.9 per- oversupply of nurses. Since no such over- cent ($2,640). Productivity in the hospital supply exists, increased reliance on NEs sector grew 26.2 percent from 2000 to 2008, is good economics, and good medicine. and the decline in real infl ation adjusted The current nursing shortage has not been wages was a mere $380. 26 Now, we need the eliminated.25 offer of gain-sharing incentive pay to give The productivity gains of information hospital workers wage enhancement in step technology and EHRs are obvious to most of with productivity gains.27 Hospital Productivity and Information Technology 37

REFERENCES

1. American Hospital Association, U.S.B.L.S U.K ., 109, 5682: 20–29 (Nov. 1999); Holling- infoweb.newsbank.com/iw-search/we/ worth, B, Dawson, P, Maniadako, N, “Effi - infoweb?p_action , Jan. 30, 2009. ciency Measurement of Health Care: A 2. Eastaugh, S, Health Care Finance and Eco- Review of Non-Parametric Methods and nomics , Sudbury, MA: Jones and Bartlett Applications,” Health Care Management (2006). Science , 2, 3: 161–172 (1999); McCue, M, 3. Incentives for Electronic Prescribing and Elec- Mark, B, Harless, D, “Nurse Staffi ng, Quality tronic Health Records, H.R. 6331, Pub. L. and Financial Performance,” Journal of Health No. 110-275, § 132, July 15, 2008. Care Finance , 29, 4: 54–76 (2003). 4. Drucker, P, Managing the Nonprofi t Organi- 16. Hellinger, F, “Specifi cation of a Hospital Pro- zation New York: Harper Collins (1999). duction Function,” Applied Economics , 7, 2: 5. American Nurses Association, “Hospital 149–160 (1999). Nurse Staffi ng,” www.nursingworld.org/gova/ 17. Moody, R, “Nurse Productivity Measures,” federal/agencies/ama.htm , Feb. 12, 2009. Health Care Management Rev ., 29, 2: 98–106 6. Matthews, C, “Using Linear Programming (2008). to Minimize the Cost of Nurse Personnel,” 18. Cardinal Health Information, Dublin, Ohio; Journal of Health Care Finance , 32, 1: 37–49 and MediQual, Atlas (2009). (2005). 19. Feldstein, M, Hospital Costs and Health Insur- 7. Eastaugh, S, “Hospital Nurse Productivity,” ance , Cambridge, MA: Harvard (1989). Journal of Health Care Finance , 29, 1: 14–22 20. Eastaugh Econometrics, and Ernst and Young, (2002). Study of the Financing of Graduate Medical 8. Rivers, P, Tsai, K, Munchus, G, “Financial Education , Volume 3 Report, DHHS 100-89- Impacts of Nursing Shortage,” Journal of 0155, Washington, DC: U.S. Department of Health Care Finance , 31, 3: 52–64 (2005). Health and Human Services (1999). 9. Goldstein, D, Groen, P, Ponkshe, S, Wine, M, 21. BScN = Bachelors of Science in nursing. Medical Informatics 20/20: Quality and EHR 22. Johnston, J, Econometric Methods , 3d ed. Through Collaboration, Open Solutions and London: McGraw-Hill (1989). Innovation, Sudbury, MA: Jones and Bartlett 23. Eastaugh, S, Health Economics: Effi ciency, (2008). Quality, and Equity, Westport, CT: Auburn 10. Bigelow, J, et al. , “Analysis of Healthcare (1999). Interventions that Change Patient Trajecto- 24. Walker, D, “Compensation: Rewarding Pro- ries,” Pub. No. MG-408 HLTH. Santa Monica, ductivity of the Knowledge Worker,” Journal CA: RAND (2009). of Ambulatory Care Management 23, 4: 48– 11. Eastaugh, S, Regan, M, “Nurse Extenders Offer 59 (2008). a Way to Trim Staff Expenses,” Healthcare 25. Rogers, A, Hwang, W, Scott, L, “Working Financial Management , 44, 4: 58–62 (1999). Hours of Hospital Staff Nurses and Patient 12. Gordon, S, Perfect Storm: Nursing In Twenty- Safety,” Health Affairs , 23, 4: 202–206 (2008); First Century America Nursing against the Buerhaus, P, Staiger, D, Auerbach, D, “New Odds, Ithaca, New York: Cornell (2009). Signs of a Strengthening Nurse Labor Mar- 13. Bradley, W, “Off-Site Teleradiology,” Radiol- ket?” Health Affairs , 23, 4: 526–531 (2008). ogy , 248, 2: 337–341 (2008). 26. American Nurses Association, Analysis of the 14. Linder, J, “Transformational Outsourcing,” Sloan American Nurses Association Staffi ng Survey, Management Review , 52–58 (Winter 2009). www.nurseweek.com/survey , Feb. 20, 2009. 15. Hollingsworth, B, Maniadakis, N, Thanas- 27. Report of the Presidents Council of Economic soulis, E, “Effi ciency: Measured Response, Advisors GPO, 3:09, Washington DC, Mar. DEA at 75 hospitals, Health Services Journal 2009. Project Measurement and Success: A Balanced Scorecard Approach

Kevin Devine, Timothy J. Kloppenborg, and Priscilla O’Clock

he current debate over health care 1. Growth/innovation; reform highlights the pressure on 2. Internal processes; T health care providers to continue 3. Customer; and attempts to control costs. In response to 4. Financial perspectives. these pressures the oversight of projects geared toward increasing both the effi - These dimensions were chosen because the ciency and effectiveness of health care authors believed that they were likely the most services has become a large component of critical aspects that enable organizations to management focus in today’s health care successfully implement their business strate- environment. Projects confronted by health gies. The name or number of the dimensions care professionals today may range from is not critical. However, when designing a projects aimed at improving or outsourcing balanced scorecard it is critical that manage- services, implementation of new software ment identify the dimensions that contribute or information systems, or the acquisition to the ability to successfully implement and and integration of new provider networks. monitor their business strategies. Examples The expanding importance placed on man- of using the BSC to enable organizations to agement of projects in all types of organiza- successfully implement strategy are wide- tions has coincided with a signifi cant body of spread.4 Norrie and Walker 5 examined the literature that has focused on determinants of benefi ts of using the balanced scorecard in project success.1 In spite of the great interest the management of a project when work- in project success the evidence of project fail- ing with a large telecommunications client. ure or at least projects that do not live up to The authors found that a project following expectations is far too common. The purpose a balanced scorecard approach performed of this article is to integrate factors shown to better on the key dimensions of on-time, on- increase project success within a balanced budget, and on-quality than a project follow- scorecard (BSC) perspective. The health ing traditional project performance metrics. care industry continues to confront a state of major change. A BSC approach will pro- Kevin Devine, is a Professor, Department of Account- vide those charged with carrying out major ancy, Xavier University, Cincinnati, OH. He can be changes with practical guidance designed to reached at [email protected]. increase project and organization success. Timothy J. Kloppenborg, is a Professor, Depart- The balanced scorecard was popularized ment of Management and Entrepreneurship, Xavier by Kaplan and Norton 2 as a broad-based University, Cincinnati, OH. He can be reached at performance report utilizing both fi nancial [email protected]. and non-fi nancial performance measures Priscilla O’Clock, is a Professor, Department of Accountancy, Xavier University Cincinnati, OH. She designed to assist fi rms in enhancing per- can be reached at [email protected]. formance and implementing strategy. The balanced scorecard proposed by Kaplan and J Health Care Finance 2010;36(4):38–50 3 Norton consisted of four dimensions: © 2010 Aspen Publishers 38 Project Measurement and Success: A Balanced Scorecard Approach 39

Figure 1 presents a basic framework Appendix at the end of this article. The new for a balanced scorecard for projects. The record system will provide a more effi cient framework provides a description of project and secure method of collecting, utilizing, assessment from the four classic score- and sharing dental information throughout card perspectives: customer, internal project the medical center in order to improve over- process, fi nancial, and growth/innovation. all patient care. The assessment criteria presented are broad but represent major considerations related to Life Cycles and Project Success project success developed by Shenhar and Dvir 6 and the Project Management Body Projects by their very defi nition are tem- of Knowledge.7 These assessment criteria porary endeavors. A project is frequently are later expanded upon to refl ect a more conceptualized as a series of life cycle comprehensive, balanced scorecard evalu- stages. The simplest life cycle model has ation approach designed to ensure project four stages of initiating, planning, execut- success. ing, and closing. At the end of each stage, an The remainder of this article is organized approval must be secured before the project into three parts. The fi rst section discusses continues. The model in this article includes the project life cycle and the corresponding project selection as part of the initiating stage opportunities for measuring project success. as well as leveraging project benefi ts as part The second section presents four dimensions of the closing stage. Figure 2 shows this of the balanced scorecard and links them to project life cycle model and the typical level organizational and project success factors of organizational effort and resources dedi- throughout the project life cycle. The fi nal cated to the project over its life cycle. The section presents a summary and conclusions. fi gure also depicts outcomes or stage ending An abbreviated example of using a BSC gates and other approvals and measures that approach to implement a centralized, elec- take place over the project life cycle. Many tronic dental record system is provided in the organizations have more detailed project life

Figure 1. Balanced Scorecard Approach to Project Monitoring and Control

Customer Internal Project Finance Growth/Innovation Scope Integration Schedule Participant development Quality Risk Cost Knowledge management Stakeholder satisfaction Communications Profi t Procurement ROI Market Share Source: Adapted from Kloppenborg, TJ, Contemporary Project Management, South- western Cengage Learning, Mason, OH (2009); Stewart, WE, “Balanced Scorecard for Projects,” Project Management Journal, vol. 32 (1) p. 45 (2001); and Norrie, J, Walker, DHT, “A Balanced Scorecard Approach to Project Management Leadership,” Project Management Journal, vol. 35 (4) pp. 52–53 (2004). 40 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Figure 2. Project Life Cycle with Measurement Points

Level of Effort

Stage Selecting Planning Executing Closing and and Initiating Leveraging

Stage Ending Charter Kick-Off Project Administrative Gates Result Closure

Benefits Other Progress Selection Measures Approvals Reports

cycle models, but the concept remains the of monitoring include the opportunity for same. project members and organizations to learn The life cycle of a project begins with the and grow from lessons learned while execut- selection and initiating phase. This phase ing the project. The completion cycle should would include feasibility studies and clearly include a post-audit of the degree to which identifi ed objectives, including scope, ben- initial objectives of stakeholders were, in efi ts, risks, and preliminary projections of fact, achieved. The process of monitoring cost and time. The selection and initiating and control, as well as evaluation of success, phase typically concludes with the approval and learning and growth can be enhanced of a project charter. Next, in the planning by measuring the progress of the project stage, the objectives subject to specifi c qual- throughout its life cycle using a balanced ity cost and time constraints are identifi ed. In scorecard approach. this phase a planning document of all actions necessary to execute the project is prepared. Evaluation of Project Success Using At the conclusion of the planning stage the a Balanced Scorecard project is launched and monitoring project progress occurs throughout the execution A project can be envisioned as a temporary and completion cycles. The main purpose organization. From this perspective it makes of monitoring and controlling a project is to sense to evaluate the project considering fac- enable the project manager and sponsor to tors that are used in assessing organizational be informed of the project’s progress and be performance. A project balanced scorecard able to take corrective or preventive steps to should be developed that maps into the keep the project on track. Secondary benefi ts organization’s scorecard utilizing either the Project Measurement and Success: A Balanced Scorecard Approach 41

four classic scorecard perspectives or those will vary across organizations and projects. perspectives unique to the project being An organization may develop measures for evaluated. For each perspective the score- each stage at the project outset. Alternatively, card should contain objectives, measures, the organization may adopt a rolling wave and targets to be met during the project’s life approach by developing measures at the end cycle. of each stage of the project before the next The scorecard should contain quantifi able stage begins. Regardless of the approach measures associated with each stage of a taken, passage through each approval gate project’s life cycle and should allow for qual- should include a report that looks back over itative inputs related to lessons learned. The the time period since the last report, looks at balanced scorecard motivates project man- the current time period between this report agers and their teams to concentrate on those and the next one, and looks further forward criteria and objectives identifi ed as most to future events in the project. A discussion critical to the project’s strategic success. of considerations and corresponding meas- Measures for the suggested scorecard ures related to each of the scorecard dimen- dimensions can be based on factors for sions is presented below. In addition, a brief project success as identifi ed by Shenhar and business case related to the implementation Dvir.8 These suggested measures of suc- of a new dental records system and the cor- cess typically focus on the closing stage, responding scorecard dimensions and meas- but identifi able evaluation measures must be ures is provided in the Appendix at the end delineated at each gate. Although these may of this article. mirror the fi nal evaluation metrics, the infor- Customer Perspective mation gathered from measurement throughout the project life cycle can identify potential Perhaps no perspective is more important problem areas, additional risks, and whether in managing projects than the needs of the the project is on the path to a successful customer. This is true regardless of whether completion. the customer is internal (the user of a new Thus, additional scorecard measures are information system, for example) or external included in the discussion below as adapted (related physician provider groups). From from best practices presented in A Guide to the customer perspective the BSC must the Project Management Body of Knowl- monitor scope and quality of the project as edge. 9 At each gate, one measure is often the well as include a constant assessment of key objective, and the sponsor may select how a project’s deliverables are aligned with up to four or fi ve additional criteria that are stakeholder expectations. Potential measures important at a given project stage. Figure associated with these critical project success 3 expands the basic framework presented factors are presented in Figure 3. earlier in Figure 1 and provides suggested BSC measurements during the project life Scope cycle. To prevent signifi cant deviations from This expansion presents potential key the project schedule or cost, the scope of objectives at each stage of the project. Of the project must be defi ned, documented, course, the importance of certain measures communicated, and controlled. Changes to 42 JOURNAL OF HEALTH CARE FINANCE/Summer 2010 Team pre-assignment Previous lessons learned Team ground rules Improve management of project meetings Project kick-off Team performance assessments Process improvement Replanning, Lessons learned application Celebration Reward Capture lessons learned Reassign workers Reapplication of lessons Systems • • • • • • • • • • • • • ve additional criteria. Not all the choices are Milestone schedule Summary budget Schedule baseline with resources Cost performance baseline Performance measures through earned value analysis Project termination decision Project termination decision ROI Contract closure Final project accounting Auditable result • • • • • • • • • • High-level risks Commitment Human resources plan Change management plan Risk management plan Risk register Quality management plan with metrics Procurement management plan Project management plan Contract awards Performance information Change requests Risk register updates Procurement documentation Complete project deliverables • Final transition of project deliverables Closed procurement Reuse • • • • • • • • • • • • • • • • • • cation ts ts realized • Scope overview Business case Stakeholder acceptance criteria Requirements documentation Scope baseline Work breakdown structure Communications of management plan Quality control measurements Stakeholder notifi and feedback Project reports and records Validated deliverables Accepted deliverables Initial realization of promised benefi Ongoing support Customer feedback Full benefi • • • • • • • • • • • Statement of Work Business Case Organization’s People and • • • • • Figure 3. Balanced Scorecard Measurement Suggestions During Project Life Cycle Suggestions During Project Measurement 3. Balanced Scorecard Figure End of Initiating Stage End of Planning Stage During Executing Project Gate/BSC CategoryInitial Project Selection Customer Internal Project Finance Growth/Innovation End of Executing Stage End Closing During Leveraging Note: At each gate, one measure is often the key objective, and sponsor may select up to about fi shown. Therefore, counting the key objective, no more than about six items would be reported at any gate. Project Measurement and Success: A Balanced Scorecard Approach 43

project scope should be processed through major project features that are developed the change control system and variance in relation to the project schedule over the analysis used to determine the impact of project life cycle. proposed changes. In addition, major com- Internal Project Perspective ponents of the project should be identifi ed, monitored, and checked off when parame- Internal project processes may be unique ters are accepted by the customer. Through- to the project or familiar to the organization. out the execution stage and closing, the For example, a hospital will have many spec- project scope must be verifi ed through a for- ifi ed processes/sequences that must be fol- mal process documenting acceptance of the lowed for any proposed expansion of patient project deliverables. services or hospital facilities. No doubt these processes will have to be adapted to Quality the current project; however, some processes A process to assure quality must be devel- may have to be changed or signifi cantly oped in the initiating and planning stages modifi ed due to the unique nature of the of the project. This process will ensure that project. work is performed correctly and the key On the other hand, a physicians’ practice stakeholders agree that the work is per- that is considering merging with an inte- formed correctly. Control of quality requires grated health care network for the fi rst time monitoring the project to ensure everything represents a totally new endeavor. In this is going according to plan, identifying when situation managing risk and effective com- preventative or corrective actions are neces- munication of all aspects of the integration sary, determining root causes of problems, effort are critical internal processes that providing specifi c measurements for quality must be considered. Additionally, on many assurance, and implementing change through projects vendors and suppliers work very the integrated change control system. The closely with internal resources and they need timing for inspections and quality audits may to be controlled accordingly be included in project schedules and checklists that become part of a project BSC. Integration The work that is expected must be commu- Managing Stakeholders nicated to each member of the project team Assumptions of stakeholders must be including a description of their role and how understood and managed at the outset of the it fi ts into the overall project. Trade-offs with project. Stakeholders’ expectations regard- respect to cost, time, quality, and scope must ing project deliverables, features of the be identifi ed, balanced, and managed. Fre- project, timelines, costs, quality measures, quently enhancements in one area of a project and actions must be documented. These lead to sacrifi ce in others. These trade-offs expectations must be clarifi ed, achieved, and must be managed and directed while closely reconfi rmed throughout the project. Confi r- monitoring project parameters defi ned in the mation procedures can be included in the scope statement. During execution this proc- project BSC and measured through the use ess is monitored and controlled primarily of project team and customer checklists for through the use of schedules and check lists. 44 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Performance report data will be collected of all three time horizons: past, present, and and actual compared with planned. When future. variances occur alternatives must be evaluated and corrective action taken. A process Procurement of change control should be developed that To the extent that external vendors are identiﬁ es, approves, and documents changes used on a project, several additional issues to the approved plan. The impact of any including vendor selection, contract signing, changes should be communicated through and monitoring of vendor processes need to revised schedules and updated completion be considered. Critical milestones associated checklists. Integration of the internal project with the selection and use of vendors may be process requires communication of meas- included in the project BSC. urements and reports throughout the project life cycle as indicated in Figure 3. Financial Perspective

Risk The fi nancial perspective of project performance is closely aligned with schedule Risks associated with roles and respon- and cost control. Therefore, the framework sibilities must be identifi ed, as well as risk for fi nancial performance measures for the associated with the scope, project require- project should be developed in the initiat- ments, and operations. Project risk should ing and planning stages. This framework be analyzed and prioritized and potential will provide the foundation to monitor the causes and potential solutions should be project through the execution phase and ulti- developed. Contingency plans to resolve mately evaluate project success in the clos- major project risks should be developed, ing and leverage phases. An example of the and if possible, a contingency time, budget, link between the project life cycle and fi nan- or other resource reserves to resolve unan- cial performance beginning with the devel- ticipated risk should be allocated. Tools opment of the project schedule, cost control, such as expected monetary value, statisti- and fi nancial performance is provided in cal sums, and/or decision trees are useful Figure 3. for evaluating the impact of risk. Measures related to the monitoring of risks and how Schedule Control each was resolved at appropriate life cycle Control of the schedule requires docu- stages should be maintained. mentation of the following: Communications • The work that should be done to date An assessment of the project team’s infor- of report; mation needs should be developed and infor- • A comparison with work that has been mation should be communicated accurately, done to date of report; promptly, and effectively. An informa- • The schedule variance (if any); tion retrieval and distribution system must • Reasons for any schedule variance be established. A progress report must be experienced; developed and shared with all stakeholders • The effi ciency of the project to date and include an analysis from the perspective with respect to schedule; and Project Measurement and Success: A Balanced Scorecard Approach 45

• Revised estimates (if necessary) with than from formalized training and educa- respect to project completion date. tion. Therefore, the BSC focuses on individual and organizational lessons learned Cost Control throughout the project life cycle. In addi- Control of the cost requires documentation, project team skill sets are monitored tion of the following: to ensure the project team has the necessary skill sets to meet future project life cycle • The planned cost of the project to date milestones. Figure 3 presents examples of of report; measurement considerations for the growth/ • A comparison with cost that has been innovation dimension over a project’s life incurred to date of report; cycle. • The cost variance (if any); • The reasons for any cost variance expe- Participant Development rienced; Role and responsibility assignments • The effi ciency of the project to date should be clearly communicated to the with respect to cost; and team members when they are assigned to • Revised estimate (if necessary) with a project. A recognition and reward system respect to project total cost. should be developed and participant performance reviewed at milestones throughout Profi t and Market Share the project life cycle. The degree to which In addition to cost control and earned the team members were motivated and sat- value, measures related to a project’s impact isfi ed should be determined and the morale on organizational profi t can be benefi cial. and energy of team members throughout When appropriate, measures of the project’s project life cycles should be evaluated. An return on investment (ROI), its impact on the interview of team members can be utilized organization’s market share, and operating to determine their personal growth from effi ciency can also contribute to the assess- participating in the project and the impact, ment of project success. if any, on the members’ desire to continue with the organization. An attempt to gather Growth/Innovation Perspective lessons learned and growth at each project The growth/innovation dimension of a milestone should be captured when feasible. project is largely centered on qualitative Such an approach assures the identifi cation individual and organizational measures of of a project member’s growth as witnessed growth. Learning and growth for projects by the expansion of the member’s skill set. is somewhat different from that of ongoing Preparing a list of project management organizations. Since individuals are fre- skills that were developed by team mem- quently selected for projects due to the skill bers participating in the various phases of a set they already possess, the growth/inno- project’s life cycle can be a valuable source vation dimension is more closely aligned of information for documenting employee with the development of the individual, as growth and can be useful when recruiting well as the organizational knowledge that and securing participants for future project results from producing the project rather teams. 46 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Knowledge Management concept that considers both short-term and The extent to which the project will con- long-term goals as well as the perspectives tribute to future projects should be evalu- of all stakeholders. ated. Identifi cation of project successes or This article has provided a framework failures is an important step in capturing for enhancing project success by relying on lessons learned. Documentation of causes of the incorporation of a balanced scorecard success and/or failure and what can be done to monitor project results at critical mile- to avoid future failures or to recreate simi- stones throughout a project’s life. Evalua- lar successes on future projects enhances tion of projects from the perspectives of (1) organizational learning and growth. In addi- customer, (2) internal project, (3) fi nancial, tion, identifi cation of new markets that may and (4) growth/innovation using a balanced be opened as a result of the project, new scorecard approach facilitates the success of or expanded services to be offered, and projects and ensures that project outcomes new technologies for future use should be are aligned with organizational goals and reported. Finally, improvements in manager strategies. capabilities and the contributions to new It is also important to realize that the business processes should be documented. groundwork for project success is often laid Similar to participant learning and growth, an long before any attempt to develop a bal- attempt should be made to document organi- anced scorecard is ever undertaken. The use zational lessons learned at each major mile- of a balanced scorecard approach to moni- stone of a project’s life cycle. This approach tor project success requires that projects are ensures that the knowledge is not lost in the carefully selected, aligned with organization ever present pressure to wrap up and move goals, and can be related to the organiza- on to the next project. tional strategy in a meaningful and measur- able way. Summary and Conclusions Once a project aligned with goals and strategies of the organization is chosen, The on-going public debate associ- project members selected, and proper com- ated with health care reform will no doubt munication of goals has been provided to the lead managers of health care organizations project team, it is possible to develop a bal- to continue to try to identify projects that anced scorecard to monitor the project over can increase the effi ciency of management its life cycle. The development of BSC meas- and provision of health care services while ures tied to the key milestones in a project’s increasing or at a minimum not diminishing life cycle can provide the opportunity to the effectiveness of these practices. Even the enhance the chance of project success. Ben- best of well intended projects can fail if the efi ts derived from the utilization of a BSC various stakeholders are not on board, or if approach to project evaluation and control the project fails to identify inherent risk, or include improved communication between appropriately consider the needs and con- project sponsors and project managers as cerns of the project’s end-users or custom- well as enhanced attention to customer needs ers. Shenhar and Dvir10 indicate that project and individual and organizational learning success is a multi-dimensional, strategic and growth. Project Measurement and Success: A Balanced Scorecard Approach 47

Appendix

An Example Project Scope Overview, Business Case, and Milestone Schedule with Acceptance Criteria

In this appendix a partial example of applying a BSC approach to project management is presented. The example is adopted from a real project and illustrates measurement criteria as determined at the end of the initiating stage (charter approval). From the customer perspective, the important criteria were scope overview, the rationale behind the business case, and acceptance criteria. The milestone schedule and the fi nancial aspect of the business case were the most important criteria from the fi nancial perspective. These steps are numbered 1 through 4, respectively, and also in Figure A1. Furthermore, the acceptance criteria in this example suggest additional items to be identifi ed and reported. These additional items are numbered 5 through 14 in Figure A1. Note that at least one item is identifi ed for each project stage. The project sponsor and manager would then decide what additional measures the project manager needs to report at each project gate. (1) Scope Overview This project will implement a centralized, electronic dental record system that will provide a more effi cient and secure method of collecting, utilizing, and sharing dental information throughout the medical center in order to improve overall patient care. This project will provide an ADT interface and will centralize scheduling and fi nancial information from existing systems. This project will not incorporate any scanning functionality to incorporate paper records of any kind. (2) Business Case This project will increase capability by providing access to dental records for multiple caregivers, incorporating clinical guidelines and safety alerts, and by providing digital radiography at all clinic locations. It will save an estimated $100,000 per year by eliminating paper charts, granting ability to access current and historical dental data at multiple sites, and providing a database for better administrative reporting and research. 48 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

(3) Milestones and (4) Acceptance Criteria

Milestone Date Acceptance Who Paper, non-centralized record Needs assessment Oct. 1, 2009 (5) List of needed features Dental management Vendor selection Apr. 15, 2010 (6) Vendor choice Dental management XYZ contract negotiation Aug. 8, 2010 (7) Signed contract Legal, IS Digital radiography vendor Sept. 1, 2010 (8) Software and hardware Dental management, and hardware selection choice with contract IS Installation and confi guration Oct. 1, 2010 (9) Functional software in test IS, director, environment application specialist Conversion Nov. 15, 2010 (10) Data from user IS, director, departments added to application specialist database Training Jan. 15, 2011 (11) Training team CIS/Director Go-live Jan. 15, 2011 (12) Successful use, no IS, user show-stoppers, customer departments, accepts management Support/Maintenance Jan. 15, 2011 (13) Support plan, user Service desk departments Electronic, centralized record Jan. 15, 2011 (14) Ability to enter and retrieve Sponsor information at all sites Project Measurement and Success: A Balanced Scorecard Approach 49 Percent of staff trained in new system ed in software training

Percent of training team certifi practices (11) nancial), Cost and schedule estimates of potential project Business case (fi milestone schedule, spending approvals, constraints Cost to percent completion comparison, project cost variance to budget Project cost variance to budget Increase in percent of reimbursed billing codes Revenue enhancement arising from improved service coding, project ROI Vendor selection Signed contract with vendor

Quality and productivity improvement (6) (7) Establish training team, Percent project completion to plan Billing errors verse baseline Billing accuracy improvement Scope overview Business case Milestones Acceptance criteria Approved list of software features Software and hardware choice with vendor Functional software in test environment Data from user departments added to database Successful use, no show- toppers, customer accepts deliverables Support plan, user departments Percent of records successfully entered and retrieved system wide.

(1) (2) (3) (4) (5) (9) Customer wants, needs, and satisfaction (10) (12) (13) (14) Figure A1. Balanced Scorecard Measurement Suggestions During Project Life Cycle Suggestions During Project Measurement A1. Balanced Scorecard Figure Project Gate/BSC CategoryInitial Project Selection by Executive Team End of Initiating Stage Customer/SupplierEnd of Planning Stage Internal ProjectDuring Executing (8) Finance Growth/Innovation End of Executing Stage End Closing Leverage Phase 50 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

REFERENCES

1. See , for example, Bessner, C, Hobbs, B, “The China Strategy,” Cost Management , vol. 20 Perceived Value and Potential Contribution of (4), pp. 9–19 (2006); Gumbus, A, Lyons, B, Project Management Practices to Project Suc- “The Balanced Scorecard at Philips Electron- cess,” Project Management Journal , vol. 37 ics,” Strategic Finance , vol. 84 (5), pp. 45–49 (3), pp. 37–48 (2006); Biehl, M, “Success Fac- (2002); and Kaplan, RS, Norton, DP, The Strat- tors for Implementing Global Information Sys- egy-focused Organization: How Balanced tems,” Communications of the ACM , vol. 50 (1), Scorecard Companies Thrive in the New Busi- pp. 53–57 (2007); Hyvari, I, “Success of Projects ness Environment , Harvard Business School in Different Organizational Conditions,” Project Press, Boston, MA, 2001. Management Journal , vol. 37 (4), pp. 31–41 5. Norrie, J, Walker, DHT, “A Balanced Score- (2006); Muller, R, Turner, R, “The Inﬂ uence of card Approach to Project Management Lead- Project Managers on Project Success Criteria by ership,” Project Management Journal , vol. 35 Type of Project,” European Management Journal , (4) pp. 52–53 (2004). vol. 25 (4) pp. 298–309 (2007); and Shenhar, AJ, 6. Shenhar, AJ, Dvir, D, Reinventing Project Dvir, D, Reinventing Project Management , Har- Management , Harvard Business School Press, vard Business School Press, Boston, MA, 2007. Boston, MA, 2007. 2. Kaplan, RS, Norton, DP, “The Balanced 7. A Guide to the Project Management Body Scorecard-Measures that Drive Performance,” of Knowledge (PMBOK© Guide), Fourth Edi- Harvard Business Review , January–February, tion, Project Management Institute, Newtown pp. 71–79 (1992). Square, PA, 2008. 3. Id. 8. See supra , n.6. 4. See , for example, Beiman, I, “Using the Bal- 9. See supra , n.7. anced Scorecard Methodology to Execute 10. See supra , n.6. Contracting and Reimbursement in Transplantation

Nicolas Jabbour, Ashish Singhal, Remzi Bag, and Marwan S. Abouljoud

The cost and quality of health care delivery are coming under increased scrutiny by both public and private payer sectors with the clear intent of increasing the ability of the consumer to use value-driven decisions in purchasing health care services. However, they are not that simple for transplantation, as there are peculiarities associated with both the cost accounting and reimbursement that differentiate transplantation from other health disciplines including complex “carve-out” contractual agreements and reimbursement methods, high per-unit cost, and organ-acquisition cost centers resulting in case rates from private payers for various phases of transplantation care. In this setting of fi xed reimbursement, the fi nancial success relies on the ability to manage its expenses and the revenue complexities effectively and effi ciently regardless of payer source. This integrity will be best protected when transplant physicians and surgeons understand their fi nancial environment and fi scal relationships of their clinical decisions and outcomes. Keywords: contracting, reimbursement, transplant center economics, center of excellence, global case rate, organ acquisition costs, fi nancial outcomes, MELD score.

s for any business, both cost and Contracting Strategies revenue estimates are central to A ensure fi nancial success of any The growth of managed care has sparked health care facility. However, they are not a renewed interest in the economics of solid that simple for transplantation, as transplant organ transplantation. It involves mix of pay- centers are resource intensive and expensive ers with numerous contract types and pay- enterprises with high overhead. Moreover, ment methodologies during various phases the economics of organ transplantation for each type of organ transplanted (see involve complex contracting strategies and Figure 1). In addition, to contract with a well reimbursement methods with multiple payer mix. They may have an adverse fi scal impact Nicolas Jabbour, MD, FACS, is the Medical Director if not managed appropriately. Therefore, to of the Nazih Zuhdi Transplant Institute at INTEGRIS direct it better, hospital administrators along Baptist Medical Center, Oklahoma City, Oklahoma. with transplant physicians must be aware of He performs liver transplantation, hepatobiliary, and their fi nancial environment, including types pancreatic surgery. He can be reached at nicolas. of payers, contracting terms, methodologies [email protected] . of payment and reimbursement, fi nancial Ashish Singhal, MD, Nazih Zuhdi Transplant Insti- tute, INTEGRIS Baptist Medical Center, Oklahoma impact of their clinical decisions and out- City, Oklahoma. comes, and most importantly, the ways to Remzi Bag, MD, Nazih Zuhdi Transplant Institute, optimize overall reimbursement. INTEGRIS Baptist Medical Center, Oklahoma City, We herein present an overview about Oklahoma. the contracting strategies and reimburse- Marwan S. Abouljoud, MD, Henry Ford Transplant ment methods involved in transplantation Institute, Detroit, Michigan. and also discuss the provider’s role (insti- Acknowledgment: The authors would like to thank tution, transplant physicians, and surgeons) Ms. Mansi Goel for her valuable contributions in the in achieving adequate reimbursement while preparation and the revision of this manuscript. maintaining volume, outcomes, and growth J Health Care Finance 2010;36(4):51–64 of the business. © 2010 Aspen Publishers

51 52 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

diversifi ed and larger payer group, transplant and Medicaid (Center for Medicare & centers need to meet the criteria for being a Medicaid Services, CMS),1 or commer- Center of Excellence (COE) (see Figure 2). cial payers, including transplant networks, This is a balancing act between the transplant insurance companies, third-party payers program wanting to stay fi nancially viable and reinsurers. The defi nition of providers and putting the necessary effort to become is limited to hospital and physicians, since a preferred COE as this is a critical way to most contractual agreements include these ensure suffi cient patient volume and revenue two groups for the provision of transplant stream. services. It should be clear that the relationship between hospitals and physicians might Contracting Parties vary greatly between institutions. The hospital may employ the physicians, and in this The contractual agreement involves pay- case, there is only one contracted provider; ers, hospitals, and physicians. The payers whereas, when physicians are employed can be regulatory payers, namely Medicare by a practice plan, or a school of medicine, there are two providers under contract who Figure 1. Phases of Transplantation may need to develop their own internal contractual agreements: the physicians and the hospital. For instance, a practice plan may Phases of Transplant charge a hospital for services rendered by physicians on behalf of the hospital’s transplant program. In this case, the hospital will receive payments from the payer and in Phase I: Pretransplant Phase—All pretransplant investigations and medical, financial, and social turn will compensate the practice plan for clearances. services rendered. Alternatively, the physicians and the hospital could both bill the

Phase II: Maintenance Phase—UNOS listing and Figure 2. Criteria for Center of Excellence waiting list management, including waiting period investigations and procedures. Volume

Phase III: Transplant Phase—All services related to transplant, including hospital and professional fees, OAC, and other costs from day of transplant CMS, UNOS, JCAHO to day of discharge or specified time. Transplant Team Support System

Cost Effectiveness Outcomes

Notes: CMS: Centers for Medicare & Medicaid Phase IV: Post-transplant Phase—From the day Services; UNOS: United Network for Organ after discharge to specified period of time. Sharing; JCAHO: Joint Commission on Accreditation of Healthcare Organizations. Contracting and Reimbursement in Transplantation 53

Figure 3. Overall Cost of a Transplant Episode

Overall Cost

Hospital Services Physician Services Organ Acquisition Cost

Fixed Cost Variable Cost

Salaries of Nurses and Other Operating Room, Laboratory Administrators, Building, Cost of Charges, Room, Pharmacy Equipment, Maintenance, Capital Cost payer separately, in which case each party vary greatly between institutions while recognizes revenue individually against payers use the local and national charge their respective charges. These peculiari- norms to determine negotiating contract ties will affect fi nancial evaluations greatly, pricing targets. This can be done using especially in the calculation of costs and in federal databases, databases maintained the recognition of revenue. by state insurance and health departments, and data from the Health Care Financing Cost Analysis Administration (HCFA) or other private The overall cost or the production cost groups; this would show how their organi- of the transplant episode or admission con- zation pricing compares with that of their sists of physician services, hospital services, competitors. and organ acquisition costs (OACs).2 The Hospitals should also identify those pay- total cost for the respective service is further ers that reimburse an unreasonably low divided into a fi xed component and variable fraction of full costs so that contractual component (see Figure 3). The transplant terms can be appropriately bargained. All center should capture these costs in real-time this requires a team approach for input by as it is the total cost against which reimburse- various departments including hospital ment will be recognized. leadership, contracting specialists, fi nance A competitive pricing analysis must be and billing departments, transplant sur- conducted as the cost accounting principles geons and physicians, legal staff, utilization 54 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

review, medical records, and managed care the hospital following transplantation, costs organizations. associated with these periods of care are not included in the calculation of cost for the Physician Services transplant procedure unless the contractual Physician services do not lend themselves arrangement does not allow for the separa- easily to cost analyses, since it is diffi cult tion of the cost related to these events. to allocate cost to a patient encounter. This Organ Acquisition Costs (OACs) would require defi ning an hourly cost of physician time and allocation of cost based OACs are unique to transplantation and purely on a percentage of salary including defi ned by a set of CMS regulations which benefi ts. Each physician or physician-group are subject to audit by the Offi ce of the 3 will then assign a ‘price’ for their services, Inspector General. In addition to the cost of also known as a ‘charge’ based on a percen- procuring all solid organs, OACs include any tile of national or regional standards pub- costs incurred in the evaluation of all poten- lished by groups. Thus, the real cost of the tial recipients and living donors at a spe- physicians is based on the aggregate cost of cifi c institution for a defi ned period of time, salaries and benefi ts, malpractice insurance, regardless of whether the patients become and other indirect costs, such as billing costs actual transplant recipients. It accounts and other practice expenses. An hourly for- for as much as 20 percent to 60 percent mula is used to bill the hospital for physician of total estimated billed charges of trans- 4 services rendered on behalf of the hospital plant for a respective organ (see Figure 4). cost centers, such as the organ acquisition This includes the costs involved in evalua- cost center (OACC), but this requires care- tion, selection, maintenance, reevaluation ful documentation and time studies that can of recipient candidates on waiting lists until withstand federal audits. Physicians will need transplantation occurred, and total cost of to complete such time studies diligently and organ acquisition, including the cost of the accurately in order to be compensated fairly organ plus direct and indirect expenses for for services rendered and be able to defend both the donor and the recipient. their charges, as well, if and when audited. The direct costs include tissue typing, donor and recipient evaluation, operating room and other inpatient ancillary services Hospital Services applicable to the donor, transportation, sur- Hospital services for transplantation can geons’ fees for recovering organs for trans- be defi ned in terms of the type of patient plantation, preservation and perfusion costs, encounter. The most straightforward con- and United Network for Organ Sharing sists of the transplant admission or episode. (UNOS) registration charges. Costs assigned to this episode typically The indirect costs include the salaries and include costs incurred from 24 hours prior to benefi ts of all personnel involved in these the transplant procedure to the time of dis- activities, as well as costs directly attribut- charge of the patient from the hospital. Thus, able to these activities, including offi ce rent, if the patient is in the hospital for any period computers, offi ce supplies, building and of time prior to transplant or readmitted to space, and even administrative and clinical Contracting and Reimbursement in Transplantation 55

Figure 4. Organ Wise Estimated Billed and Organ Procurement Charges

Estimated Billed Organ Organ Charges ($) Procurement ($) Kidney 259,000 67,500 Liver 523,400 73,600 Pancreas 275,500 68,400 Heart 787,700 94,300 Lung-Single 450,400 53,600 Lung-Double 687,800 96,500 Intestine 1,121,800 77,200 Kidney-Pancreas 439,000 122,300 Kidney-Heart 1,005,700 145,600 Liver-Kidney 763,500 127,000 Heart-Lung 1,123,800 151,900

Source: http://www.milliman.com/expertise/healthcare/ publications/rr/pdfs/2008-us-organ-tisse-RR4-1-08.pdf. salary expenses. In addition, physician may play a role in causing a great variability services necessary for these evaluations can among transplant centers for this cost ( see be charged to the OACC. Figure 5). All services mentioned above consist of a fi xed component, which is primarily com- Contracting Terms prised of overhead costs: this includes size of Contract language must be clear as it can the institution, the number of beds, salaries have a serious detrimental effect on revenue of nurses and other nonmedical staff such and must include concise defi nitions. Items, as administrators, building costs for the hos- such as new technology reimbursement and pital, the cost of equipment, and the cost of payment terms, need to be defi ned to ensure maintaining the facilities and the equipment, accurate and prompt payments (see Figure 6). as well as other costs, such as capital costs. When in doubt, the services of an actuary This constitutes one of the main reasons for may be of help in negotiating with payers. the variation in cost between institutions. In addition, the contract should mention the The variable component includes every cost language for “carved out” services and medi- directly or indirectly associated with the pro- cal needs not directly related to the transplant cedure. These costs would not be incurred if event. These include but are not limited to the procedures were not performed. They radiofrequency ablation for liver tumours, may include operating room costs, labora- chemotherapy or radiotherapy for tumours, tory costs, room and board costs, pharmacy post-transplant procedures, pharmaceu- costs, as well as other costs. Several factors ticals, non-transplant—related services, 56 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

dental, home health care, rehab, home infu- Reimbursement Methods sion, durable medical equipment, ventricular assist devices, registry fees, travel, lodging, The multiple payer mix is challenging unusual blood products, and viral treatment. with payment arrangements ranging from a single global case rate that must cover Figure 5. Factors Affecting Variable Cost all transplant-related services to individual • Geographic area payment arrangements for each stage of the • Transplant center volume and incidence of transplantation process (see Figures 7 and 8). complications Billed charges serve as the basis for negoti- • Average number of organs procured per donor ated prices; however, a contracted price, not • Underlying diagnosis and/or disease state a billed charge, serves as the basis for an • Medical management (may reduce costs economic transaction. Therefore, transplant particularly with respect to hospital charges) centers should be prepared to offer a global or • Use of cost control mechanisms, such as: single price for both hospital services (part A) – Greater donor and recipient selectivity; and professional services (part B). – Critical pathways to reduce inpatient LOS; A global fee includes not only the pro- – Aggressive use of outpatient therapies; cedure itself but also all related services – Reduction of medical errors; and and visits that occur within a designated – Pharmacoeconomics. time period prior and/or after the trans- • Compliance with federal and state regulations plant event. In some cases, the price may Note: LOS: Length of stay. include all services the patient requires up

Figure 6. Glossary of Contracting Terms

Activity-base costing (ABC): Bottom up approach to cost estimation and ties the allocation of cost to utilization of resources. Allowable days: Specifi ed number of days assigned by type of transplant, i.e., 12 days for kidney transplant and 24 days for liver transplant. Case rate: Rate that covers everything hospital provides during entire stay. Ceiling: “Lesser of” charges when charges exceed a certain level. Cost-to-charge ratio (CCR): Estimate costs assuming they are a percentage of charges. Floor: Clause for catastrophic cases, hospital will not be paid less than X percentage of charges Fully allocated cost: Cost of service after both direct costs and allocated costs are added up. Global case rate: Payer may “bundle” professional hospital and organ acquisition and defi ne the time period for included services (admission, 30 days, etc.). Outlier: Transplant charge that far exceeds case rate and typically an agreed upon threshold. Percent of charges: Payer pays percentage of charges from hospital/physicians. Per diem: Amount payer pays per hospital day, includes all services. Step-down method: Allocating costs “not paid for” to services that are. Stop loss: Level of charges above which provider is no longer liable. Contracting and Reimbursement in Transplantation 57

Figure 7. Common Contracting Models

Contracting Models Based on Phases of Transplant

Outpatient: Percentage Percentage of Percentage of Phase I of billed charges charges All Phases: Global charges case rate for all Inpatient: Per diem services for a case rate specified time

Percentage of Outpatient: Percentage Phase II Percentage of charges charges of billed charges Inpatient: Global per diem case rate

Global Phase III Global case rate for Pretransplant Inpatient: case rate specified time Per diem rate Transplant: Global case rate

Outpatient: Percentage Percentage of Phase IV of billed charges Outpatient: Percentage of charges billed charges Inpatient: Global per diem case rate Inpatient: per diem rate

to one year after the transplant procedure. and only rarely separate payment may be However, the contract language for case permitted for initial evaluation, for services rate time period may differ among payers. for unrelated problems, and reoperations for For example, Payer 1: “…if member is dis- related or unrelated complications. charged before case rate day maximum is reached, and readmitted to Hospital, reim- Medicare bursement will be covered under case rate In spite of the health care market move- until the maximum period is reached…”. ment to managed care, Medicare continues to Payer 2: “…includes Hospital, Physician, be the predominant payer. According to the and Ancillary services required from the Transplant Management Group, Medicare day prior to transplant through the discharge accounted for 51 percent of kidney transplants of member from Hospital…” (no readmis- nationally.5 In addition, a multi-organ sion language). Frequently, this situation is transplant program could have as much as very disadvantageous to transplant centers 40 percent of their total annual reimburse- as insurers pass on more risk to providers ment coming from Medicare. 58 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Figure 8. Speciﬁ c Reimbursement Mechanism Used by Individual Payer

REIMBURSEMENT MECHANISMS OF PAYERS

SINGLE CASE MEDICARE MEDICAID MCOs or COMMERCIAL SELF-PAY PAYERS CONTRACT

OAC + DRG-PPS + APC GLOBAL CASE RATE

OAC DRG PHYSICIAN’S REIMBURSEMENT

PERCENTAGE DISCOUNT FROM CHARGES

Notes: MCOs: managed care organizations; OAC: organ acquisition cost; DRG-PPS: Diagnosis Related Group-Prospective Payment System; APC: ambulatory payment classifi cations.

When Medicare is the primary payer, the fi nancial risk associated with potential the hospital is compensated for the actual complications and longer lengths of stay transplant procedure through the Part ‘A’ reside with the transplant program. Sur- Diagnosis Related Group-Prospective Pay- geons and physicians bill for their services ment System (DRG-PPS). DRG is a sys- during the inpatient stay through part B tem used by Medicare since 1983 to pay the Medicare and are paid 80 percent of the fee hospitals a predetermined set rate based on allowed by Medicare. the patient’s diagnosis rather than simply Organ acquisition costs (OAC), includ- reimbursing whatever costs they charged to ing pretransplant costs, are reimbursed on treat Medicare patients. This is based on the a dollar-per-dollar basis. When the organ is fact that patients within each category are removed from a living donor, surgeons and similar clinically and are expected to use the physicians are paid 100 percent of allowable same level of hospital resources. Medicare fees. This must be fully documented Each DRG has its own dollar value and and submitted via the Medicare Cost Report. covers all costs directly related to the opera- This reporting is also auditable by the Offi ce tion and inpatient hospitalization following of the Inspector General, and overreporting a transplant procedure. This means that can result in heavy fi nes for the institution. 6 Contracting and Reimbursement in Transplantation 59

To prevent overreporting, transplant pro- drivers. Thus, centers must still track and grams must be able to support and verify submit all incurred costs. all cost submissions. Items such as billing Tracking costs begins the day prior to the statements, pretransplant billing and regis- transplant and ends at the end of the global tration procedures, staff time studies, and case rate period, or if a covered individual transplant department records can be used is still inpatient at the end of the global case to ensure proof of reimbursed costs. 7 For rate period, on the date of discharge from the services in post-transplant period, reim- inpatient stay. If days for inpatient admission bursement is usually according to the outpa- exceed the global case rate period for trans- tient prospective payment system known as plant, the reimbursement will revert to the out- ambulatory payment classifi cations (APCs) lier per diem rate for transplant until the date of and the hospital may be paid for more than discharge from inpatient stay (see Figure 9). one APC in a single patient case. In a case where a referred transplant patient is not part of a policy in which the trans- Medicaid plant program contracts and the transplant Medicaid accounts for less than 10 per- program is outside of a transplant network, cent of the primary payer market for organ agreements can be negotiated on an individ- transplantation. As Medicaid is a partner- ual basis. Such an arrangement may benefi t ship program between federal and state par- both parties if the transplant center is able to ties, these programs can vary widely from negotiate a reasonable case rate, because the state to state. Some transplant programs are local program is usually much more cost- fi nancially unable to accept the Medicaid effective for the insurance company in terms payment rates for transplantation as reim- of savings on travel and lodging. bursement is signifi cantly below transplant When a commercial payer is primary, costs and it may be in the program’s best both the inpatient charge and the stand- interest to decline such an arrangement. On ard acquisition charge are submitted to the the other hand, some Medicaid programs carrier and Medicare becomes the second- are much more reasonable when it comes to ary payer through a process referred to as reimbursement and can supply an acceptable “coordination of benefi ts” for any portion amount of transplant volume to certain cent- of the inpatient care or standard acquisi- ers like Kentucky Medicaid.8 tion charge denied by the commercial payer (assuming that the recipient is eligible for Commercial Payers Medicare benefi ts). All-inclusive global contracts are the most In the case of living donor transplantation, often used payment method by commercial the commercial payer should be contacted payers. Global case rate is issued directly to prior to transplantation to determine whether the transplant facility where payment covers it will accept separate charges from physi- everything involved in the transplant process cians and surgeons for the care of the living from organ acquisition costs to the transplant donor. If the recipients’ commercial payer episode and for a set amount of time post- refuses altogether, the transplant hospital transplanation. However, many insurers will becomes payer of last resort by charging still want independent recognition of cost those services to its OAC center. 60 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Figure 9. Sample Case Rates

1. “Lesser of” % of billed charges or case rate plus outlier per diem for days over allowable days, and fl oor (never paid less than X % of billed charges). For example: Lesser of 90% of billed charges or case rate plus outlier no less than 60% of billed charges.

Case rate $170,000 Total billed charges $500,000 Days in case rate 12 Total length of stay 50 Outlier per diem $3,000 Outlier days 38

90% of billed charges $450,000 Case rate $170,000 + (38d x $3,000) $284,000 Floor 60% of billed charge $300,000 Reimbursement $300,000

2. Case rate plus 1st dollar stop loss: % of charges of entire admission when exceed agreed amount. Example 1. Lesser of 90% of billed charges or case rate plus outlier, or if billed charges are over $340,000, then entire case is paid at 80% of billed charges.

Case rate $170,000 Total billed charges $500,000 Days in case rate 12 Total length of stay 50 Outlier per diem $3,000 Outlier days 38

90% of billed charges $450,000 Case rate $170,000 + (38d x $3,000) $284,000 80% of billed charges $400,000 Reimbursement $400,000

Example 2. Lesser of 90% of billed charges or case rate plus outlier, or if billed charges are over $340,000, then entire case is paid at 80% of billed charges (lower charge).

Case rate $170,000 Total billed charges $300,000 Days in case rate 12 Total length of stay 15 Outlier per diem $3,000 Outlier days 3

90% of billed charges $270,000 Case rate $170,000 + (38d x $3,000) $284,000 80% of billed charge Not applicable Reimbursement $270,000 Continued Contracting and Reimbursement in Transplantation 61

Figure 9. Continued...

Charge over $340,000 x 0.5 = $80,000 Outlier case rate = $284,000 $364,000 Reimbursement $364,000

3. Case rate plus 2nd dollar stop loss: % of charges for billed above an agreed amount. Example: Lesser of 90% of billed charges or case rate plus outlier, and billed charges are over $340,000 are paid at 50% of billed charges.

Case rate $170,000 Total billed charges $500,000 Days in case rate 12 Total length of stay 50 Outlier per diem $3,000 Outlier days 38

90% of billed charges $450,000 Case rate $170,000 + (38d x $3,000) $284,000

Self-Pay be essentially “capped” under a case rate or Certain individuals are willing to pay out under a global payment schedule. of pocket for the complete cost of a trans- For Medicare, once charges exceed a plant procedure. This is particularly true for specifi ed amount above the DRG payment, international patients without health insur- reimbursement converts from a fi xed payment ance as well as uninsured, underinsured, and to a percentage of charges. Commercial pay- self-insured individuals. In such situations, ers may also have a stop-loss provision that both the patient and the hospital need to clar- takes effect at some specifi ed ceiling of costs ify the terms of payment and how to address above the global case payment. Once this catastrophic outcomes and costs. threshold is met, the entire case reverts to an agreed upon percentage of billed charges. Unpredictable Cost Alternatively, a fl at “per diem” hospital rate Almost all payer groups have provisions may be agreed upon as well. Although the for many unforeseen costs relating to an extra payment does help with expenses once individual transplant case for which no other a certain level of expenditures is reached, the fi nancial remedy can be found and the trans- gap between the case rate and the stop-loss plant institution is responsible; this is not payment threshold can often be fi scally prob- infrequently the situation in “complicated lematic. Costs that fall into this gap remain cases.” Under these circumstances, “outlier” unpaid and become the full burden of the payments may be allowed and is known as transplant center. Even when the cost ceiling outlier protection or stop loss. The trans- is reached, the percentage of charges that is plant center should have justifi ably incurred paid is usually insuffi cient to allow a margin excess expenditure relative to the contracted and may not even cover the costs of the hos- price for services and reimbursement may pitalization and transplant procedure. 62 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

A certain percentage of a global payment these contracts, transplant institutions must can be allocated into a special general ledger be aware of opportunities to keep expenses account. Transplant programs can use this low while remaining clinically successful. mechanism known as “consultant/risk pool” This mandates the in-depth understanding to help cover costs such as those in the outlier of the fi nancial environment by the hospital gap. Other examples include the receipt of a and transplant physicians and surgeons and physician bill for consultations not reported requires a collaborative approach to establish during the global period before claim adju- revenue optimization strategies. This will dication, and work-up costs for recipients certainly have a signifi cant effect on contract who received these services at an off-site, negotiation and may enable the hospital to out-of-network facility where an individual contract with more payers. agreement does not exist. This pool can be The existing contracting and reimburse- applied to a particular global arrangement or ment models in transplantation show a com- to individual agreements in which payment plete separation between the medical and is due. When and if the pool reaches a cer- fi nancial sides, putting more fi nancial risk on tain predetermined upper limit, no additional providers. This is more than just a concern as withholding would be taken against future the majority of transplant centers in the United accounts until the pool reaches a predeter- States are low volume centers for respective mined fl oor or upon distribution of funds at organs.10 This may negatively affect the fi s- the end of each quarter.9 cal health of transplant programs even with a small number of high-risk patients. Conse- Discussion quently, it may lead transplant centers to exer- cise strict selection criteria for these patients In the simplest form, a profi t-maximizing resulting in systematic exclusion of many fi rm will seek to simultaneously reduce costs potential candidates from the list who may while increasing revenue. For transplanta- have been transplanted otherwise.11 Sound tion, however, operations are not that simple contracting and continuous improvements in and it is reasonable to believe that a multi- practices would play a crucial role in provid- tude of factors are at play in achieving the ing fair treatment to patients and in receiving enviable fi nancial outcomes. Both payers and appropriate reimbursement to maintain the transplant centers need to reach a fi nancially transplant center’s operations.12 viable agreement for both parties. Reim- A relatively large body of literature is bursement and contracting vary between available on volume-outcome relationship;13 payers and can have an adverse economic however, there is a paucity of literature on impact if not managed appropriately. At the outcomes and their fi scal associations. As same time, cost estimates can be complicated the clinical outcomes are evaluated based and are not easily determined. On the other on disease severity like MELD (model for hand, transplant programs cannot afford to end stage liver disease) score, Buchanan and sign losing deals and must, at a minimum, colleagues have showed that MELD score obtain agreements in which the organiza- was a signifi cant cost driver for pretrans- tion will break even. Because of these inher- plant, transplant, and total charges. 14 They ent fi nancial risks and complicated nature of found that high-MELD score patients incur Contracting and Reimbursement in Transplantation 63

signifi cantly longer hospital stays and higher as equally important with other standard costs prior to and at the time of liver trans- quality measures such as volume, clinical plantation but did not have higher rates of re- outcomes, and patient satisfaction. admissions. This was an important fi nding as the transplant admission charges represented Conclusions approximately 50 percent of the total cost of liver transplantation. In the setting of fi xed reimbursement, the In addition, both the clinical and fi nancial fi nancial success of a transplant program outcomes following organ transplantation relies on the ability to manage its expenses depend on the quality of the graft; how- and the revenue complexities effectively ever, no standardized measure of graft qual- and effi ciently regardless of payer source. ity exists.15 Axelrod et al. 16 have previously The business strategy would aim to increase observed the increase in length of hospital volume by improving outcomes, securing stay and cost as the donor risk index increases, well-negotiated managed-care contracts, regardless of the MELD score. Furthermore, practicing revenue optimizing strategies, and current federal regulations mandate the use improving coverage of the market segments. of marginal organs for transplantation that Reimbursement schemes and pricing strate- are known to be associated with higher cost gies should incorporate an evaluation of the and inferior clinical outcomes. 17 Therefore, severity of illness because reimbursement- it would seem important to recognize the to-cost ratios translate directly into fi nancial impact of both recipient and donor variables outcomes. Lastly, physicians and surgeons on cost. are stewards of the transplant enterprise at It is reasonable to propose that risk- large as they make all clinical decisions. adjusted reimbursement should be deter- Functional integrity will be best protected mined based on the severity of illness; this when they understand the regulatory and fi s- is critical as hospital stay, complication cal relationships peculiar to transplantation if rates, overall results are directly related to they are to be competitive, viable, and fully disease severity. Alternatively, it may be reimbursed for the costs of their services. possible that despite risk adjustment and Hence, their participation in the fi scal proc- expected clinical outcomes including uti- ess and acceptance of a fi duciary role and lization of resources, the relevant variables responsibility are critical to fi nancial viabil- that can exacerbate the discrepancy between ity of a transplant center. This would man- expected and actual outcomes, not only clin- date their trust in and the collaboration of ically but also fi nancially, might not all be their administrative partners, matched with taken into account. However, this advocates transparency and an inclusive management that fi nancial outcomes should be measured style from hospital leadership.

REFERENCES

1. http://www.cms.hhs.gov. 3. www.oig.hhs.gov. 2. Abecassis, MM, “Financial Outcomes in 4. Milliman, Inc., 2008, U.S. Organ and Tissues Transplantation—A Provider’s Perspective,” Transplant Cost Estimates and Discussion, Am J Transpl , 6: 1257–1263 (2006). Apr. 2008, http://www.milliman.com/expertise/ 64 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

healthcare/publications/rr/pdfs/2008-us-organ- Complications Following Liver Transplant? A tisse-RR4-1-08.pdf. Business Case for Quality Improvement,” Am 5. Transplant Management Group, Professional J Transplant , 6: 2978–2982 (2006). Services Brochure, 2008, available at http:// 13. Dimick, JB, Pronovost, PJ, Cowan, JA, Lipsett, www.transplantmanagement.com/images/ PA, Complications and Costs After High-Risk TMG%20Professional%20Services%20Broch Surgery: Where Should We Focus Quality ure%202008.pdf. Improvement Initiatives?” J Am Coll Surg , 6. Health Economics Resource Center, Depart- 196: 671–678 (2003); Birkmeyer, JD, Stukel, ment of Veterans Affairs, 2006 http://www. TA, Siewers, AE, Goodney, PP, Wennberg, herc.research.va.gov/resources/glossary.asp?s DE, Lucas, FL, “Surgeon Volume and Opera- ort=Acronym%20DESC#. tive Mortality in the United States,” N Engl J 7. Burke, R, Medicare Report , Washington, DC: Med , 349: 2117–2127 (2003); Salvalaggio, Bureau of National Affairs, 15:1–4 (2004). PR, Modanlou, KA, Edwards, EB, Harper, AM, 8. Marshall B, Swaeringen JP, “Complexities Abecassis, MM, “Hepatic Artery Thrombosis in Transplant Revenue Management,” Prog After Adult Living Donor Liver Transplanta- Transpl , 17(2): 94–98 (2007). tion: The Effect of Center Volume,” Transplan- 9. Sinaiko, J, Jordan, V, Wollschlager, E, “Emerging tation , 84: 926–928 (2007). Compliance Iissues in Organ Transplant Reim- 14. Buchanan, P, Dzebisashvili, MS, Lentine, KL, bursement: Seven Questions Every Transplant Axelrod, DA, Schnitzler, MA, Salvalaggio, PR, Center Should Ask,” Chimera , 15: 11–13 (2003). “Liver Transplantation Cost in the MELD Era: 10. Weiss, ES, Meguid, RA, Patel, ND, et al ., Looking Beyond the Transplant Admission,” “Increased Mortality at Low-Volume Ortho- Liver Transpl , 15: 1270–1277 (2009). topic Heart Transplantation Centers: Should 15. Feng, S, Goodrich, NP, Bragg-Gresham, JL, Current Standards Change?” Ann Thorac Dykstra, DM, Punch, JF, DebRoy, MA, et al ., Surg , 86: 1250–1260 (2008); Reese, PP, Yeh, “Characteristics Associated with Liver Graft H, Thomasson, AM, et al ., “Transplant Center Failure: The Concept of a Donor Risk Index,” Volume and Outcomes After Liver Retrans- Am J Transplant , 6: 783–790 (2006). plantation,” American J Transpl , 8: 1–9 16. Axelrod, DA, Schnitzler, M, Salvalaggio, PR, (2008). Swindle, J, Abecassis, MM, “The Economic 11. Ammori, J, Pelletier, S, Lynch, R, Cohn, J, Ads, Impact of the Utilization of Liver Allografts Y, Campbell, D, et al ., “Incremental Costs with High Donor Risk Index,” Am J Trans- of Post-Liver Transplantation Complications,” plant , 7: 990–997 (2007). J Am Coll Surg , 206: 89–95 (2008). 17. Abecassis, M, “Making Dollars and Sense out 12. Englesbe, M, Dimick, J, Mathur, A, Ads, Y, Well- of Liver Transplantation,” Liver Transpl , 15: ing, T, Pelletier, S, et al ., “Who Pays for Biliary 1159–1161 (2009). The Financial Impact of Post Traumatic Stress Disorder on Returning US Military Personnel

Jeffrey P. Harrison, Lynn F. Satterwhite, and Walter Ruday, Jr.

This article addresses the ﬁ nancial impact of post traumatic stress disorder (PTSD) on US military personnel returning from service in Iraq and proposes a strategy to ensure that adequate resources are available to provide evidence-based PTSD care. Prolonged exposure to combat stress has produced high rates of veterans with PTSD and other psychiatric disorders. The study found that from 2003 to 2008 approximately 720,666 US military members deployed to Iraq. Based on that population at risk, if 15 percent of returning US military members will require health care services for PTSD, it is estimated that approximately 108,099 returning US military members will require treatment for PTSD. Based upon current deployment rates, government health care planners can anticipate the annual expenditure of $200 million on PTSD care. The study has managerial implications associated with ensuring high quality PTSD health care services for returning US military personnel. It has policy implications on the allocation of scarce health care resources within the Department of Defense (DoD) and Veterans Health Administration (VHA) health care systems to enhance the provision of PTSD services to military personnel and veterans. Keywords: post traumatic stress disorder (PTSD), US military personnel, veterans, evidence-based treatment, evidence-based medicine (EBM), transaction cost economics (TCE).

recent study by Seal et al . 1 found considerable impairment in function and that among the 289,328 returning become a signifi cant fi nancial burden for the A US veterans entering the Veterans United States. As a result, the Department of Health Administration (VHA) health care system between 2002 and 2008, 21. 8 per- Jeffrey P. Harrison, PhD, MBA, MHA, FACHE, is cent were diagnosed with post traumatic an Associate Professor, Department of Public Health, at the University of North Florida, Jacksonville, Flor- stress disorder (PTSD) and 17.4 percent ida. He can be reached at [email protected] were diagnosed with a depressive disor- and (904) 620-1440. 2 der. Similarly, the RAND study showed Lynn F. Satterwhite, RN, MS, CNS, ANP, is an high rates of PTSD in returning US mili- Instructor in the School of Medicine, Department of tary personnel. Based on that population Psychiatry, Virginia Commonwealth University, and at risk, the federal health care system must a Clinical Nurse Specialist and Nurse Practitioner at provide appropriate levels of PTSD serv- the Hunter Holmes McGuire Veterans Administration Medical Center, Richmond, Virginia. ices for the returning military service per- Walter Ruday, Jr., LMSW, ACSW, is the Administra- sonnel. Although traumatic brain injury is a tive Coordinator of the Jacksonville Substance Abuse signifi cant problem for returning US serv- Treatment Team Department of Veterans Affairs, Jack- ice members, it is beyond the scope of this sonville, Florida. article. Acknowledgement: The views presented in this arti- If signifi cant numbers of returning US mil- cle are the opinions of the authors. The article does itary members require health care services not represent the opinions of any organization that for PTSD, it is critical that the federal health the authors may have been affi liated with previously or are currently affi liated. care system be prepared to provide treatment. Without effective treatment, returning J Health Care Finance 2010;36(4):65–74 military personnel with PTSD could suffer © 2010 Aspen Publishers

65 66 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Defense (DoD) and VHA health care sys- Defi nition of Post Traumatic tems are focusing on their ability to provide Stress Disorder evidence-based PTSD services. This requires the building of partnerships among research- The American Psychiatric Association4 ers, clinicians, and health care providers to defi nes PTSD as an anxiety disorder that provide essential PTSD services. This large develops as a result of exposure to an extreme patient population supports further initia- traumatic event. This event involves actual tives in improving the quality and effi ciency or threatened death or serious injury to self of PTSD services. It also has policy impli- or others. A common response from the indi- cations on the distribution of health care vidual could be extreme helplessness, horror, resources to health care organizations which or intense fear. These individuals experience provide PTSD services to military personnel three clusters of symptoms which include and veterans. re-experiencing, avoidance, and arousal. As of July 18, 2009, US casualties Re-experiencing involves intrusive from Iraq include 4,319 dead and 31,446 thoughts, vivid nightmares, or fl ashbacks of wounded. Research would suggest that US the events. The individual with PTSD may military personnel in the Iraq and Afghani- act and feel as if the event is happening again. stan wars are receiving the best medical There are triggers that illicit intense psycho- care available in the theatre of battle.3 Most logical or physical responses from the indi- recently, the practice of medicine is putting vidual. Avoidance involves making efforts an increasing emphasis on adherence to to avoid thinking, feeling, or talking about standard clinical approaches developed the experience. Additionally, individuals through evidence-based medicine (EBM). may avoid anything that reminds them of the EBM has been adopted to provide the best trauma or not want to do the things they used medical practices based on sound empirical to enjoy. The individuals may feel different research. There is a growing body of knowl- from others and have diffi culty expressing edge about PTSD in US military personnel or having emotional feelings towards oth- and veterans, including a number of control- ers. Some individuals may have amnesia for led studies on the best practices for treat- part of the traumatic event. The fi nal cluster ment of PTSD. is arousal which involves sleep disturbances, By applying the research, it may be pos- irritability, anger, exaggerated startle refl ex, sible to estimate the health care resources and feeling as if they were looking over necessary to meet the projected demand their shoulder or on guard. In order to be for PTSD treatment. Thus providers who diagnosed with PTSD, the symptoms must are operating in an environment concerned impact the individual’s life in a negative way with cost-control can project the additional such as diffi culties in social, occupational, or resources necessary to give the highest qual- other important aspects of life.5 ity medical care. Such a systematic analysis Examples of extreme traumatic events of PTSD refl ects a commitment to excellence include childhood sexual trauma, physical in health care delivery and allows organiza- assault, serious accidents, terrorist attack, com- tions to measure performance and improve bat, and natural disasters such as fi re, tornado, quality. hurricane, fl ood, or earthquake. According The Financial Impact of Post Traumatic Stress Disorder 67

to the American Psychiatric Association, 6 military personnel fail to seek treatment for individuals who have symptoms for less than mental problems are: a month are considered to have acute traumatic stress disorder. After a month of symp- 1. The medications have signifi cant side toms patients are diagnosed with PTSD. effects, Individuals with continued symptoms of 2. Treatment could negatively affect their PTSD after three months are considered to career, have chronic PTSD. 3. Treatment could cause denial of security clearance, PTSD Within the Military Environment 4. Family and friends are more helpful than mental health providers; and Combat situations can have a nega- 5. Coworkers may lose confi dence in tive impact on the mental status of return- their ability.9 ing soldiers. There is a growing body of knowledge on the effects of PTSD on US Treatment Modalities for PTSD military personnel during the Vietnam, Persian Gulf, Iraq, and Afghanistan wars. Efforts to prevent development of PTSD Additionally, the National Vietnam Vet- have been implemented since the Vietnam erans Readjustment Study of 1983 was a War. For example, mental health specialists congressional mandate for studying the psy- are currently providing services including: chological effects of war. Results demon- education, briefi ngs on suicide prevention, strated approximately 15 percent of men and triage, and short-term treatment. However, 9 percent of women among Vietnam Veter- the use of critical incident stress debriefi ng ans were diagnosed with PTSD. The fi rst for early intervention to minimize the devel- Gulf War showed PTSD rates of 9 percent to opment of symptoms has not been supported 24 percent. 7 The current Iraq and Afghani- by research.10 stan confl icts are ongoing and as a result Treating PTSD has been widely stud- the full impact on mental health is not yet ied and there is a large body of evidence known. Since October 1, 2001, 1.64 mil- for best practices. There are clinical prac- lion US military personnel have deployed to tice guidelines developed by the VHA and these wars. Studies have shown that up to DoD11 for treatment of PTSD which are 21.8 percent of returning service members based on research from controlled and peer have PTSD and depression. 8 reviewed studies. These guidelines help cli- Research shows that among US troops nicians choose treatments that have demon- returning from Iraq and Afghanistan up to strated positive results and are cost effective. 300,000 may suffer from PTSD or major According to these guidelines, the four top depression. This research also found that the recommended psychotherapies for treatment majority of soldiers deployed were exposed of PTSD are: to a combat-related event such as ambushes, seeing dead bodies, being shot at, or know- 1. Cognitive therapy; ing someone who was seriously injured or 2. Exposure therapy; killed. The top fi ve reasons returning US 3. Stress inoculation training; and 68 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

4. Eye movement desensitization and then identifying the positive thoughts the reprocessing. patient would like to have when thinking of the trauma. There are bilateral stimula- These therapies have strong evidence for tion or eye movements as the individual’s improvements in global functioning and own brain makes sense of the disturbing reduction in symptom severity. Other psy- material.14 chotherapeutic interventions that have dem- First line medications recommended by onstrated some benefi t include: the VHA and DoD practice guidelines for • Imagery rehearsal therapy; PTSD treatment are the selective serotonin • Psychodynamic therapy; and reuptake inhibitors (SSRIs). The US Food • PTSD patient education. and Drug Administration has approved two medications for the treatment of PTSD; These therapies, as with most psycho- these are Sertraline (Zoloft) and Paroxetine therapy, can occur individually or in a group (Paxil), which are useful in depression and therapy format. anxiety. Other SSRIs have effi cacy in treat- Cognitive therapies work on chang- ment of PTSD. Benzodiazepines, which ing irrational beliefs that affect individuals have long been used for treatment of anxiety adversely and challenge them to revise those disorders, are not recommended.15 beliefs into something more positive. Forms The National Defense Authorization Act of Cognitive Therapy include Cognitive of 2008 required the VA and DoD to develop Restructuring, which assists the patient in a comprehensive approach for the manage- making sense of their bad memories. Expo- ment of health care and transition of service sure therapy is a behavioral therapy designed personnel. The US Government Account- to help reduce the level of fear and anxiety ing Offi ce Report to Congressional Com- associated with reminders of the trauma, mittees 16 outlines the progress to date and thereby also reducing avoidance. Stress Inoc- addresses the future activities necessary to ulation Training, another behavioral therapy, improve health care management. Substan- assists the patient in gaining self reliance in tial progress has been made in policy devel- his or her coping ability in order to overcome opment and many projects are being piloted. the anxiety and fear resulting from trauma The VHA and DoD now have policies for reminders.12 Behavioral theories also involve mandatory training on suicide prevention some cognitive therapies. These are based and screening for PTSD and other war- on the principal that the arousal response is related problems. Mechanisms are in place a conditioned, emotional response and that to track notifi cations to health care providers exposure to the trigger without a negative so that these problems may be addressed and outcome could relieve the arousal.13 later evaluated. Recovery plans for return- Eye Movement Desensitization and ing service members have been developed, Reprocessing (EMDR) is a controversial which include training of individuals who therapy, but there is substantial evidence supervise and provide the care. Access to indicating its effi cacy. EMDR involves care continues to be evaluated with mini- identifying the trauma, emotions, and nega- mum standards for access to non-urgent care tive thoughts associated with the trauma and other medical services. The Financial Impact of Post Traumatic Stress Disorder 69

Studies show that most people who live All of these problems could add to the total through a traumatic event do not get PTSD. cost of medical care in these veterans. However, risk factors include: Socially, individuals with PTSD have higher rates of absenteeism at work and are • Living through traumatic events; less productive. PTSD can affect families • A history of mental illness; resulting in domestic problems, diffi culties • Getting hurt or seeing others hurt; in parenting, and possibly suicide. The liter- • Horror or extreme fear; and ature contains much on the plights of home- • Having minimal social support after the less veterans and substance abuse. event. The RAND19 study found that the cost of Among combat soldiers, risk factors in- PTSD and major depression for two years clude: after deployment ranged from $5,900 to • Younger age; $25,760. As a result, it estimated that the • Combat exposure; national cost of PTSD and depression for • Rank; 1.64 million service members could range • Branch; and from $4.0 billion to $6.2 billion. Studies • Multiple deployments. have shown that if 100 percent of PTSD patients received evidence-based treatment, Another factor that plays an important a savings of approximately $1.7 billion or role is resiliency, which can reduce the risk $1,063 per case would be realized. Based on of being diagnosed with PTSD. Use of resil- their analysis, evidence-based treatment for iency includes seeking support from family PTSD and major depression would pay for and friends, fi nding a support group, feeling itself within two years. These results show good about one’s own actions in a dangerous that the use of EBM protocols for PTSD event, having good coping skills, and being and major depression would increase recov- able to respond effectively despite fear.17 ery rates, improve productivity, enhance the readiness of US military personnel, and pro- The Long-Term Effects of PTSD vide benefi ts to society.20

The costs and long-term effects of PTSD Theoretical Foundation can be signiﬁ cant. Biologically, patients with PTSD face higher rates of unhealthy behav- Transaction cost economics (TCE) theory ior, which could result in increased rates of suggests that organizational structures exist many chronic illnesses such as cardiovascu- to economize on the costs of exchanging lar disease, osteoporosis, arthritis, diabetes, goods and services in the marketplace while certain cancers, and periodontal disease. maximizing the quality of services provided. Vieweg et al . 18 studied body mass index According to Stiels, Mick, and Wise,21 health (BMI) in a convenience sample of military care is a complex sequence of transactions veterans and found that those with PTSD had among patients and providers which occurs far greater rates of obesity than the current in markets as well as within health systems. US population. Higher rates of unsafe sex Since health care transactions involve the are also associated with the affects of PTSD. production of care and the coordination of 70 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

that care, it is imperative that they are exe- we project future costs for PTSD patient cuted smoothly and effi ciently. care. TCE is a conceptual framework for ana- lyzing health care transactions and quantify- Results ing their impact on organizations, processes, and outcomes. In health care, these transac- This study documents that federal health tion costs come from the following: care organizations and other health care providers are facing a growing need to pro- • Monitoring outcomes; vide PTSD programs. Additionally, the lit- • Information gathering; erature clearly indicates that a greater use of • Administrative support; evidence-based treatment for PTSD patients • Negotiations; provides an opportunity for signifi cant im- • Transaction frequency; provement in health care quality. • Idiosyncrasy of the exchange; and As noted in Figure 1, the level of US forces • The degree of uncertainty in the market. deployed in Iraq during the study period As a result, TCE would support the use of ranged from a low of 132,444 in 2003 to a integrated health care delivery systems as an high of 156,000 in 2007. These data show effective method of expanding the scope of a combined total of 720,666 US military PTSD services as well as reducing the trans- members served in Iraq from 2003 to 2008. action costs associated with providing high Research on the rates of PTSD in prior quality health care services. wars show a range from 9 percent to 24 percent in returning military personnel. Figure 2 Methods shows the projected rates of PTSD for US military members in need of PTSD treat- To implement appropriate evidence-based ment using the conservative historic PTSD treatment for PTSD, practitioners need to be well informed on the treatment protocols and Figure 1. US Troop Forces health care leaders need to provide adequate Deployed in Iraq fi nancial support. As a result, health care organizations must develop standard treat- Year Mean ment processes for PTSD and incorporate 2003 132,444 these standards to ensure patient safety and quality of care. 2004 133,916 This study evaluates data on the number 2005 145,291 of US military personnel deployed in Iraq 2006 138,500 from 2003 to 2008. Based on the deployed 2007 156,000 population, we use historical rates of PTSD 2008 145,285 found in previous wars to project the number Total 720,666 of PTSD cases in US military personnel Source: Global Security (2008), Iraq—US returning from Iraq. Finally, using realistic Forces Order of Battle, retrieved June 12, projections on the cost of care per patient 2009 from www.globalsecurity.org/military/ ops/iraq_orbat.htm. developed during the RAND 2008 study, The Financial Impact of Post Traumatic Stress Disorder 71

rate of 15 percent. This shows that the pro- felt by those people who suffer the disorder jected number of new PTSD patients return- but also by families, employers, and society ing from Iraq range from a low of 19,866 in as a whole. This documents that PTSD has 2003 to a high of 23,400 in 2007. a negative economic impact on society due Additionally, the data show an aggregate to reduced personal income, lower work per- patient population of 108,099 with PTSD formance, and an increase in the utilization of from 2003 to 2008. These data show that treatment and support services. They believe responding health care organizations need cost-benefi t analysis (CBA) is the most ap- additional resources to meet the growing propriate method to evaluate the economic demand for PTSD care. It is also interesting impact of PTSD. This approach is unique in to note that the DoD and the VA health care that it addresses the extent to which a treat- organizations are increasingly recognizing ment or policy is socially worthwhile in the the critical importance of a comprehensive broadest sense, with all costs and benefi ts approach to providing ongoing PTSD health being valued in monetary units. If benefi ts care operations. Also, as military mem- exceed costs, the CBA would recommend bers resign or retire, other US health care providing the treatment. With two or more organizations can anticipate an increasing alternatives, CBA evaluation would recom- need for PTSD treatment within the general mend the one with the greatest net benefi t. population. Economic evaluations such as CBA exam- McCrone, Knapp, and Cawkill 22 in their ine cost in relation to the level of outcome research completed an economic impact to show the best alternative. Looking specifi - study of PTSD within a military population. cally at PTSD, an improved quality of life They found the impact of PTSD is not only could be evaluated consistent with resource consumption to provide a better understanding of the economics of the disorder and the Figure 2. US Troop Forces Deployed alternative treatment options. in Iraq and Projected PTSD This CBA approach for the economic analysis of PTSD is supported by RAND, 23 US Forces Projected which incorporated treatment costs, de- Year Deployed PTSD @ 15%* creased wages, lost work productivity, and 2003 132,444 19,866 suicide costs in its economic model to meas- 2004 133,916 20,087 ure the impact of PTSD. RAND24 found a 2005 145,291 21,793 baseline cost of $10,151 per case cost of 2006 138,500 20,775 PTSD for returning US military personnel. 2007 156,000 23,400 As noted in Figure 3, the projected 2008 145,285 21,792 cost of PTSD for US military personnel Total 720,666 108,099 returning from Iraq ranges from a low of

Source: Global Security (2008), Iraq—US $201,659,766 in 2003 to $237,533,400 in Forces Order of Battle, retrieved June 12, 2007. Most importantly, the potential total 2009 from www.globalsecurity.org/military/ cost of PTSD in returning US military ops/iraq_orbat.htm. personnel during the study period could *Calculated using PTSD rate of 15%. approach $1,097,312,949. 72 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Figure 3. Projection of Resources US military personnel with PTSD exceeds Necessary to Treat US Troop Forces $200 million annually. Additionally, as these Deployed in Iraq for PTSD resources are made available, it is essential that evidence-based treatment be provided PTSD Cases Projected Cost to maximize patient recovery rates. Finally, Year @ 15% of PTSD* it is important that the future allocation of 2003 19,866 $201,659,766 resources be made to ensure that discharged 2004 20,087 $203,903,137 military personnel are able to access care 2005 21,793 $221,220,743 within their local community.25 2006 20,775 $210,887,025 From an operational perspective, as addi- 2007 23,400 $237,533,400 tional US military forces are ordered into 2008 21,792 $221,210,592 Iraq, the research suggests that additional Total 108,099 $1,097,312,949 mental health providers will be required to Source: RAND (2008). care for a growing PTSD patient population. *Calculated using PTSD per case cost of It is important to note that our research is part $10,151. of an incremental process to examine current research and benchmark best practices Discussion in PTSD care. As a result, clinicians and health care leaders need to use such research Our research shows that the DoD, the VA, to improve PTSD care. Recent innovations and other health care providers are expe- assist health care professionals with clini- riencing a signifi cant growth in patients cal decision-making for PTSD by integrat- requiring PTSD care. TCE shows that costs ing patient information with recommended can be reduced and quality improved by clinical protocols to provide the best EBM. coordination of care across the continuum Additionally, electronic medical record of health care services. Fortunately, recent (EMR) systems can be linked with decision investments in health information technol- support systems to give alerts when patient ogy provide an opportunity for health care status changes. According to Harrison and organizations to improve the quality of care Palacio, 26 the Veterans Health Administra- they provide PTSD patients. This can be tion (VHA) EMR is currently in use in 1,300 done by sharing of best practices and par- VHA medical centers, outpatient clinics, ticipating in research on PTSD to improve and nursing homes. The VHA EMR allows the quality and effi ciency of the care they approximately 100,000 VHA clinical pro- provide. Additionally, we believe health care viders to access patient medical records as leadership should continue to refi ne out- well as place orders for services or medi- come measures utilized for benchmarking cations through this single interface. Such best practices in PTSD care. health information systems can be linked As shown in our research, improved treat- with the latest wireless health information ment of PTSD in returning US military technology being deployed on the battlefi eld personnel requires additional resources be to maximize the quality of care. Most impor- allocated to the health care treatment system. tantly, integrated databases provide a way to Specifi cally, we found the cost of returning project workload and plan for future resource The Financial Impact of Post Traumatic Stress Disorder 73

allocation. By adequately forecasting poten- providers have the potential to meet the tial demand for PTSD services, organi- growing need for PTSD care if adequate zational factors such as excessive clinical resources are provided. workload, inadequate resources, and poor Harrison, Nolin, and Suero 27 found that clinical practice can be minimized thereby greater coordination of clinical services improving the quality of PTSD care. enhances operational effi ciency and improves quality of care. This suggests that provid- Conclusions ing coordinated evidence-based PTSD care will improve outcomes and enhance effi - PTSD is an emotional illness that has ciency. It will also improve communications developed as a result of exposure to a signif- among patients, providers, and health care icant traumatic event. There is research that organizations. shows it has been a part of the aftermath of From a policy perspective, directing addi- combat as long as there have been wars, and tional resources into PTSD care has the its effects on the victim, the victim’s fam- potential to reduce health care costs while ily, as well as society are great. Specifi cally, improving the health status of patients with PTSD has been linked to increased rates PTSD. With the number of individuals with of domestic problems, parental problems, PTSD growing, the challenges of providing substance use problems, fi nancial prob- cost-effective PTSD care will increase. As a lems, occupational problems, and suicide. result, policymakers should support further The cost associated with treating PTSD is research in EBM to evaluate whether PTSD high although EBM and today’s technology can be treated by coordinating preventive allow a greater capacity to provide treat- services, outpatient care, and inpatient health ment for those in need. Without effective services. Such an approach will also allow treatment, the long-term costs and negative for the effi cient use of medical resources and consequences would be increased. Fortu- facilitate collaboration among all stakehold- nately, DoD, the VHA and other health care ers leading to improvements in PTSD care.

REFERENCES

1. Seal, KH, Metzler, TJ, Gima, KS, Bertenthal, globalsecurity.org/military/ops/iraq_casualties- D, Maguen, S, Marmar, CR, “Trends and refs.htm. Risk Factors for Mental Health Diagnoses 4. American Psychiatric Association: Diagnostic Among Iraq and Afghanistan Veterans Using and Statistical Manual of Mental Disorders Department of Veterans Affairs Health Care,” (DSM-IV-TR), 4th ed., Arlington, VA, Ameri- American Journal of Public Health , 99(9): can Psychiatric Publishing, Inc., 2000. 2002–2008 (2009). 5. Id. 2. RAND, “Invisible Wounds: Mental Health and 6. Id. Cognitive Care Needs of America’s Returning 7. Tull, M, “Rates of PTSD in Veterans,” Medical Veterans,” 2008 (RB No. 9336-CCF), retrieved Review Board 2009, retrieved June 25, 2009, June 25, 2009, http://www.rand.org/pubs/ from http://ptsd.about.com/od/prevalence/a/ research_briefs/RB9336/index1.html. MilitaryPTSD.htm. 3. “Global Security: US Casualties in Iraq,” 2009, 8. Supra , n.2, n.1. retrieved on June 12, 2009, from http://www. 9. Supra , n.2. 74 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

10. Pols, H, Oak, S, “Public Health Then and June 25, 2009, from http://www.nimh.nih. Now: War & Military Mental Health: The US gov/health/publications/post-traumatic-stress- Psychiatric Response in the 20th Century,” disorder-easy-to-read/index.shtml ; supra , n.1. American Journal of Public Health , 97(12): 18. Vieweg, WV, Julius, DA, Benesek, J, Satter- 2132–2142 (2007). white, L, Fernandez, A, Feuer, S, Pandurangi, 11. Veterans Health Administration, Department AK, “Posttraumatic Stress Disorder and Body of Defense, “VA/DoD Clinical Practice Guide- Mass Index in Military Veterans: Prelimi- line for the Management of Post-Traumatic nary Findings,” Progress in Neuro-Psychop- Stress,” 2004, Version 1.0, Washington (DC): harmacology & Biological Psychiatry , 30: Veterans Health Administration, Department 1150–1154 (2006); Vieweg, WV, Fernandez, of Defense, complete summary retrieved July A, Julius, D, Satterwhite, L, Benesek, J, Feuer, 27, 2009 from http://www.guideline.gov/ S, Oldham, R, Pandurangi, AK, “Body Mass sumary/sumary.aspx?ss=15&doc_id-5187. Index Relates to Males with Posttraumatic 12. NIMH, “Post-Traumatic Stress Disorder, Stress Disorder,” Journal of the National Med- National Institute of Mental Health publica- ical Association , 98(4): 580–586 (2006). tion number TR-08-6388, 2008, retrieved 19. Supra , n.2. June 25, 2009, from http://www.nimh.nih. 20. Eibner, C, “Invisible Wounds of War: Quanti- gov/health/publications/post-traumatic-stress- fying the Societal Costs of Psychological and disorder-easy-to-read/index.shtml; Cloitre, M, Cognitive Injuries, Congressional testimony “Effective Psychotherapies for Posttraumatic presented before the Joint Economic Com- Stress Disorder: A Review and Critique,” CNS mittee, RAND: Santa Monica, CT-309 (2008); Spectrums [The International Journal of Neu- supra , n.2. ropsychiatric Medicine], 14(1), Supplement 21. Stiels, RA, Mick, SS, Wise, CG, “The Logic of 1: 32–43 (2009). Transaction Cost Economics in Healthcare 13. Supra , n.1. Organization Theory,” Health Care Manage- 14. Cloitre, M, “Effective Psychotherapies for ment Review , 26(2): 85–92 (2001). Posttraumatic Stress Disorder: A Review and 22. McCrone P, Knapp M, Cawkill P, “Posttraumatic Critique,” CNS Spectrums [ The International Stress Disorder (PTSD) in the Armed Forces: Journal of Neuropsychiatric Medicine ] 14(1), Health Economic Considerations,” Journal of Supplement 1: 32–43 (2009). Traumatic Stress , 16(5): 519–522 (2003). 15. Supra , n.11. 23. Supra , n.2. 16. United States Government Accounting Ofﬁ ce, 24. Id. “Recovering Service Members: DOD and 25. Id. VA Have Jointly Developed the Majority of 26. Harrison, JP, Palacio, C, “The Role of Clini- Required Policies but Challenges Remain,” cal Information Systems in Healthcare Qual- Report to Congressional Committees, GAO- ity Improvement,” The Health Care Manager, 09-728 (2009). 25(3): 206–212 (2006). 17. NIMH, “Post-Traumatic Stress Disorder, 27. Harrison, JP, Nolin, J, Suero, E, “The Effect of National Institute of Mental Health publica- Case Management on U.S. Hospitals,” Nurs- tion number TR-08-6388, 2008, retrieved ing Economics , 22(2): 64–70 (2004). Health Care Policy and the HIV/AIDS Epidemic in the Developing World: More Questions Than Answers

Paul J. Flaer, Paul L. Benjamin, Francisco I. Bastos, and Mustafa Z. Younis

When the United Nations declared “health care for all” (at the conferences at Alma-Ata in 1978 and the Ottawa Charter in 1986),1 the declarations were largely premature to impact the upcoming HIV/AIDS epidemic. These UN declarations still apply today, as multitudes of humanity continue to die from what amounts now to be a treatable chronic disease. Can the wealthier, industrialized countries stand by and watch the decimation of the populations of the developing world by HIV/AIDS? The global “health 9/10 gap,” relates that only 10 percent of global heath resources go to developing countries—i.e., those having 90 percent of the poorest world populations.2 The World Bank/World Health Organization has been at the forefront of providing resources for the global HIV/AIDS epidemic,3 but for many countries of the developing world (especially Sub-Saharan Africa) it may be too little, too late. This work explores the application of an ecological model to global policy against HIV/AIDS, highlighting access to antiretroviral drugs (ARV). ARV distribution is constrained by patents and laws protecting the intellectual property rights of the international pharmaceutical corporations. In response to this situation, more questions arise. Will governments in the developing world invoke compulsory licensing (patent-breaking) in their negotiations with the international pharmaceutical corporations to provide medications against HIV/ AIDS in their countries? Can international political and ﬁ nancial negotiations with these pharmaceutical corporations speed the growing push for a solution to this solvable crisis? The answers may lie in the “Brazilian model,” that is a developing world government using all means available to provide ARV drugs for all its citizens with HIV/AIDS. The basis of this model includes negotiating with the pharmaceutical corporations over patent rights and importation of copied drugs from the Far East. Keywords: HIV/AIDS Policy, Brazil, ARV, Compulsory Licensing, World Bank/World Health Organization.

he application of public policy in the the accurate assessment of needs, access, provision of HIV/AIDS medications costs, and avenues of distribution. Moreover, T to a target population (i.e ., in this case, developing world countries) requires Fernando I. Bastos is a senior researcher at FIOCRUZ - Oswaldo Cruz Foundation, Rio de Janeiro, Brazil. He is the chairman of graduate studies on epidemiology at Paul J. Flaer is a doctoral fellow in the Stempel the National School of Public Health-FIOCRUZ and College of Public Health, Florida International Uni- a physician who has extensive experience working on versity, Miami, Florida. At the Dade County Dental studies assessing populations at high-risk of HIV infec- Research Clinic, Jackson Memorial Medical Center, tion in Brazil. He holds the degrees of Medical Doctor he provides pro bono dentistry to disadvantaged from State University, Rio de Janeiro, and a PhD from populations in Miami, Florida. He holds the degrees the National School of Public Health, FIOCRUZ. of Doctor of Dental Surgery from Medical College of Mustafa Z. Younis is a Professor of Health Econom- Virginia, Richmond, Virginia, and the degrees of Doc- ics & Finance at Jackson State University, Jackson, tor of Education and Master of Public Health from Mississippi. He teaches courses in Health Financial Florida International University. Management, Health Economics, Health Policy and Paul L. Benjamin is an attending practitioner provid- Comparative/International Health Systems, and has ing pro bono dentistry for disadvantaged populations extensive experience in international health. He holds and teaching of dental residents at the Dade County the degree of Doctor of Public Health from Tulane Dental Research Clinic/Jackson Memorial Medical University, New Orleans, Louisiana. Center in Miami, Florida. He holds the degree of Doctor of Medicine Denarius from the University of J Health Care Finance 2010;36(4):75–79 Florida, Gainesville. © 2010 Aspen Publishers 75 76 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

extent of adherence to prescribed medica- health ministries want drugs that are avail- tions, knowledge of side/adverse effects, able and distributed to patients at low cost or and the use of advanced drug regimens free of charge. ( i.e., combination drug therapy or HAART— Families and communities throughout the highly active antiretroviral therapy) are key developing world are being destroyed solely factors in HIV/AIDS program implemen- by the lack of access to life-saving HIV/ tation. Public programs, in an approach to AIDS medications. Entire cultures and ways accessing ARV (antiretroviral drugs) for of life are being decimated by the ravages countries of the developing world, are the of a killer disease easily treated with effec- means of last resort to access and treat the tive ARV drugs. Gone are the days of diffi - high morbidity and mortality of HIV/AIDS cult dosing and onerous side effects by ARV in the developing world. drugs that existed in the early years of the HIV/AIDS epidemic. Today, newer drugs Compulsory Licensing used in combination therapy (HAART) are well tolerated and have fewer side effects In Brazil, copying ARV drugs under com- than medications in the initial years of HIV/ pulsory licensing ( i.e ., patent-breaking) AIDS treatment.4 and importing ARV drugs manufactured in A possible solution to containing the highly competitive third world developing HIV/AIDS epidemic in developing countries countries such as Thailand, India, and China, may be a modifi ed approach to compulsory has subsequently lead to lowered prices and licensing of HIV/AIDS drugs. A worldwide- increased availability of the drugs to affected unifi ed approach to compulsory licensing populations. However, questions arise about on the “Brazilian model” could be accom- diffusion of this Brazilian innovation to other plished by UN agencies—especially the developing world countries: World Health Organization (WHO) acting in concert with the World Bank.5 UN involve- • Can developing countries apply this ment could present a single and unifi ed voice Brazilian system of market compe- to negotiate with the pharmaceutical corpo- tition to obtain ARVs and drugs for rations over the cost of ARV drugs for the the treatment of deadly opportunistic developing world. The economic, geopo- infections? litical importance, and relative success in • Moreover, can each developing country dealing with the HIV/AIDS crisis by Brazil effectively distribute ARV drugs; i.e ., serves as a model for other nations.6 In addi- do they have an adequate public health tion to leading the developing world in the infrastructure? HIV/AIDS crisis, Brazil sponsored a resolu- • Can these developing countries negotiate tion that placed access to ARV medications drug acquisition from the international as a fundamental human right in the context pharmaceutical corporations according of previous UN declarations of “health care to policies workable and favorable to for all.” 7 In Brazil, although use of ARV/ the interests of both parties? HAART resulted in moderately decreased The bottom line is that the pharmaceutical rates of transmission, it signifi cantly lowered corporations want a profi t margin, while the both morbidity and mortality of HIV/AIDS.8 Health Care Policy and the HIV/AIDS Epidemic in the Developing World 77

The HIV/AIDS epidemic is the great- India, and China, could successfully est public health crisis in modern history compete in the international market but has also become the greatest battle over in providing ARV medications at low intellectual property rights.9 Applying the prices that the populations in the devel- “Brazilian model” to health care of afﬂ icted oping world could afford. HIV/AIDS patients evokes consideration of 6. The presence or development of a the following elements of emerging policy functional public health infrastructure for treatment and prevention:10 for treatment and prevention of both HIV/AIDS and subsequent opportun- 1. Invoking a national policy of com- istic infections. pulsory licensing; that is, a system of 7. ARV/HAART therapy and medica- market competition that has been pre- tions for the treatment of opportunis- viously proved successful (by Brazil)— tic infections must be available at no leading to deep discounts on patented or low cost to the patient. Advanced ARV medications from the interna- second line, third line, and specialty tional pharmaceutical corporations drugs should be employed in HIV/ in the face of patent-breaking in the AIDS therapy as the rule rather than market economies of the developing the exception. world. 8. Capacity to produce ARV drugs in 2. Guiding the implementation of policy developing world countries, usually in by monitoring of modes of treatment the form of an existing pharmaceutical of HIV/AIDS and subsequent oppor- industry, is important for local produc- tunistic infections by means of effec- tion of HIV/AIDS drugs. (Note that tive surveillance systems. Brazil’s pharmaceutical industry is the 3. A progressive approach to HIV/AIDS tenth largest in the world.) intervention, using the media, open discussion in the public schools, and Discussion the targeting of those highest at-risk or living in marginalized communi- Article 25 of the United Nations’ Univer- ties; for example, IDUs (injection drug sal Declaration of Human Rights (1948)11 users), MSM (men who have sex with afﬁ rms the right of all humanity to receive men), pregnant women, youth, and needed medical care and services. This dec- sex-workers. laration, although preceding the HIV/AIDS 4. Harm reduction strategies such as pro- crisis, directly applies to the present day pan- motion of abstinence and use of bar- demic. In the developing world, treatment is rier protection along with the presence most often initiated with CD4+ counts less of needle exchange programs in the than 200 or until the patient becomes sympto- community. matic.12 In industrialized countries this count 5. Pharmaceutical laboratories manufac- is usually at a lower end of 500. Such delays, turing ARV medication in the highly besides their costs in human suffering, also competitive market environments of promote economic loss and instability due to developing countries, such as Thailand, the increased costs of treating more severe 78 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

cases, overall increases in morbidity/mortal- immune systems continue to fail due to HIV/ ity, and subsequent increased levels of trans- AIDS. Those in developing countries who mission within the population. 13 To optimally could have gone into remission with treat- meet global responsibilities of industrialized ment by ARV medications continue to die. 16 nations, the response to the HIV/AIDS epi- Addressing the HIV/AIDS epidemic in the demic in the developing world might include developing world brings attention to other the following characteristics:14 important social and health issues such as food supply, poverty, ability to work, and domestic 1. Governments in the developing world political stability. In addition, the prevalence working in a crisis mode to provide of co-infections, i.e ., a high morbidity of HIV/HAART to its citizens; tuberculosis, hepatitis, syphilis, and malaria, 2. Wealthier governments supporting the presents almost insurmountable problems for fi ght against HIV/AIDS in the devel- health care in countries already ravaged by the oping world; HIV pandemic. Without treatment with life- 3. International pharmaceutical corpora- saving ARV, the early deaths of HIV-infected tions providing free or low cost ARV patients have devastating effects on families, drugs—irrespective of corporate profi t communities, and ultimately pose a dilemma and the specter of compulsory licens- for the stability of developing nations. In this ing; and time of global fi nancial crisis, it makes good 4. Industrialized nations making the sense for nations in the developing world to worldwide and collective investment manufacture, instead of importing, HIV/AIDS to fi ght what effectively amounts to a drugs for treating their populations.17 Self- global health crisis. production of ARV/HAART subsidized by developing world governments saves money Conclusion otherwise earmarked for their importation and even stimulates local industry. Early in the Health care education and the Hippocratic HIV/AIDS crisis, Somerville and Gilmore at Oath teach practitioners to ease suffering, pre- the International Conference on AIDS (1989) 18 vent death, and above all “do no harm.” Since concluded that the HIV/AIDS epidemic “must the inception of the AIDS epidemic, many be examined within a framework that enables health practitioners in developing countries them to be comprehensively and thoroughly feel that they cannot work with integrity and analyzed and connected with wider health follow the Hippocratic Oath when life-sav- and social issues.” The “Spaceship Earth” ing drugs for HIV/AIDS are not available.15 concept applies here: We all live together on Despite global support by teams of health care an increasingly interdependent world where workers, controlled nutrition, and aggressive disease, poverty, and socio-political unrest in treatment of opportunistic infections, patients’ one country eventually impact us all.

REFERENCES

1. Alma-Ata, Declaration of Alma-Ata, Interna- The Ottawa Charter for Health Promotion, tional Conference on Primary Health Care, First International Conference on Health Pro- Alma-Ata, USSR (1978); and Ottawa Charter, motion, Ottawa, Canada (1986). Health Care Policy and the HIV/AIDS Epidemic in the Developing World 79

2. Fowler, Adhikari, Bhagwanjee, “Clinical Re- 9. Bastos, FI, Buchalla, CM, Ayres, JR, da Silva, view: Critical Care in the Global Context— LJ, “Brazilian Response to the HIV/AIDS Epi- Disparities in Burden of Illness, Access, and demic, 2001–2005,” Rev Saúde Publica , 40 Economics,” Critical Care , 12: 225–231 (Suppl): 3–4 (2006). (2008). 10. Supra , n.7; Nunn, AS, Fonseca, EM, Bas- 3. Beyrer, C, Gauri, V, Vaillancourt, D, “Evalu- tos, FI, Gruskin, S, Salomon, JA, “Evolution ation of the World Bank’s Assistance in of Antiretroviral Drug Costs in Brazil in the Responding to the AIDS Epidemic, 2005, Context of Free and Universal Access to AIDS Washington, D.C. : The World Bank , retrieved Treatment,” PLos Medicine , 4(11): e305–e324 June 27, 2009, from http://www.worldbank. (2007); Nunn, AS, Fonseca, EM, Gruskin, S, org/oed/aids/main_report.html. “Changing Global Essential Medicines Norms 4. Cohen, K, Cornick, R, Harley, B, Louis, F, to Improve Access to AIDS Treatment: Les- Orrell, C, Subotzky, E, “‘We’re Unable to Ful- sons from Brazil,” Global Public Health , 4(2): ﬁ ll Our Hippocratic Oath’: An Open Letter by 131–149 (2009). Doctors in the Public Sector, The Cape Times 11. United Nations Universal Declaration of (Apr. 14, 2003). Human Rights , New York, NY: United Nations 5. WHO, “Evidence for Action: Effectiveness General Assembly (1948). of Community-Based Outreach in Prevent- 12. Ford, N, Mills, E, Calmy, “A Rationing Antiret- ing HIV/AIDS Among Injecting Drug Users,” roviral Therapy in Africa—Treating Too Few, Geneva, Switzerland: World Health Organi- Too Late,” New England Journal of Medicine , zation (2004). 360(18): 1808–1810 (2009). 6. Celentano, DD, Beyrer, C (Eds.), Public Health 13. Id. Aspects of HIV/AIDS in Low and Middle 14. O’Brien, K, “Drug Companies and AIDS in Income Countries: Epidemiology, Prevention Africa,” National Catholic Weekly , 2002, and Care , Chapters 8, 9, and 29, New York: retrieved May 22, 2009, from http://www. Springer (2008). americamagazine/org/content/article.cfm? 7. Id. article_id=2625. 8. Kaiser Family Foundation, “The Role of 15. Supra , n.4. Part D for People with HIV/AIDS: Coverage 16. Id. and Cost of Antiretrovirals Under Medicare 17. Malone, B, “Africa Must Manufacture Own Drug Plans,” Medicare Drug Beneﬁ t, Report AIDS Drugs: AU,” Thomson Reuters Press #7548, 2006), retrieved July 1, 2008, from Release, May 7, 2009. http://www.kff.org/hivaids/7548.cfm ; Calmy, 18. Somerville, MA, Gilmore, N, “Physicians, Eth- A, Hirschel, B, Cooper, DA, Carr, A, “A New ics and AIDS: A Framework for Ethical Analy- Era of Antiretroviral Drug Toxicity,” Antiviral sis of HIV/AIDS,” International Conference on Therapy , 14(2): 165–179 (2009). AIDS , 5: 947 (1989). Private Health Insurance in Australia: Community Rating, but at What Price(s)?

Luke B. Connelly and Henry Shelton Brown III

Australia has voluntary private health insurance (PHI) markets in which open enrollment and community-rated premiums are mandated by government. Historically, adverse selection in these markets led to a substantial decline in coverage, giving voice to fears about the viability of PHI markets in the longer-run. In order to preserve community rating but improve the PHI pool, the Australian government instituted a novel scheme of age-based penalties (ABPs) for individuals who join a PHI fund later in life. This article computes the price of PHI under the so-called Lifetime Cover (LC) scheme and shows that the LC scheme per se is not appropriately calibrated to prevent another adverse selection death spiral.1 Based on our results, we recomputed age-based penalties that would result in a fair price of PHI for all age groups. The premium multipliers we derive suggest a premium ratio of 10:1 for the oldest and youngest joiners. Our premium multiplier sequence is well-approximated by a linear ABP scheme that is approximately three times that of the present LC scheme for older joiners. Key words: private health insurance (PHI), premium, community rating, Lifetime Cover (LC), subsidies, taxes.

n most countries, private health insur- 19 to 34 (40 percent uninsured) than for peo- ance (PHI) membership is not com- ple ages 55 to 64 (8 percent uninsured).3 O n e I pulsory. In voluntary PHI markets, concern has been with the denial of PHI cover an attendant problem is adverse selection: to people with chronic diseases, or PHI offers many individuals who would benefi t from that entail prohibitively high (and perhaps buying PHI are unwilling to do so, either unfair) premiums. Some states in the United because the premiums available exceed the States have reacted by making the denial of expected PHI benefi ts or because they are coverage illegal through guaranteed issue denied coverage due to discoverable risk and community rating laws. 4 Of course, the factors, such as chronic diseases. recently passed health care reform bill man- In Australia, where PHI is voluntary and dates private insurance for most individuals open enrollment is mandated, an adverse who are not eligible for public insurance.5 selection death spiral2 had characterized PHI markets. In response to this problem, Aus- Luke B. Connelly, PhD, is Professor of Health Econom- tralia has made a novel attempt to counteract ics, (UQ Node) Director, Australian Centre for Economic adverse selection through, inter alia, an age- Research on Health (ACERH UQ); Associate Director, Centre of National Research on Disability and Reha- based penalty (ABP) scheme. Individuals bilitation Medicine (CONROD); and School of Eco- who buy and maintain PHI avoid the ABP nomics, The University of Queensland, Mayne Medical altogether. Furthermore, those who main- School. He can be reached at [email protected]. tain PHI are guaranteed to have the option Henry Shelton Brown III, PhD, is Associate Profes- of being privately insured at the prevailing sor of Health Economics, Division of Management, community-rated premium, irrespective of Policy and Community Health, Michael & Susan Dell their risk profi le at any point in the future. Center for Advancement of Healthy Living, Institute for Health Policy, University of Texas School of Pub- In the United States, it is well-known that lic Health—Austin campus. He can be reached at many millions are uninsured. The percentage [email protected]. of uninsured persons in the United States in J Health Care Finance 2010;36(4):80–92 2001 was fi ve times higher for people ages © 2010 Aspen Publishers 80 Private Health Insurance in Australia: Community Rating, but at What Price(s)? 81

although some have vowed to challenge the (perhaps prohibitively) high risk–rated pre- so-called “personal mandate” provision on miums. The Australian government also constitutional grounds6 and for other, politi- has pursued two more conventional policies cal reasons.7 to promote the purchase of PHI: a tax on Another concern in the United States has middle-to-high-income earners for not buying been with the consequences of being reclas- PHI and a direct ad valorem subsidy on the sifi ed as high risk. The Health Insurance insurance premium for purchasers of PHI. Portability and Accountability Act (HIPAA) This article examines the extent to which of 1996 ensures that high-risk individu- a range of PHI policy measures in Australia als cannot be dropped or have their premi- have affected the price of PHI, defi ned as the ums increased by more than others insured ratio of premiums to expected benefi ts, by by that fi rm. These policies may have the age. Using data on PHI expenditures, by age, unintended consequence of increasing the as well as data on hospitalization frequencies, average risk level of those who are insured and episodic costs, we compute the price of although recent work has found little evi- PHI for each age group in the Australian pop- dence of this in individual markets.8 Adverse ulation with PHI and examine the dynam- selection implies that an increased average ics of prices and membership over time. risk level, in this case by mandating that pri- We conclude that the propensity for an vate insurers cover high risks at a premium adverse selection death spiral still exists in below their expected benefi t, may drive the Australian PHI markets, but that this has better risks among the insured out of the been kept in check by tax penalties that market or into low-risk pools in other fi rms.9 the Australian government has imposed on This may result in a net reduction in PHI middle-to-high-income earners who do not rates at the population level and may lower buy PHI. We also offer an alternative scheme economic welfare. of age-based taxes that is nevertheless con- The Australian ABP scheme offers the sistent with the general framework of the following alternative: for those who buy existing LC scheme. and maintain PHI when they are young and relatively healthy, there is guaranteed issue at Background the community-rated premium, independent of any subsequent risk level. If one becomes Australia has a system of public insurance high risk while not privately insured, how- that is universal and compulsory. Despite this ever, although there is still guaranteed issue, fact, more than 40 percent of the population an increasing ABP is also imposed on the holds private health insurance (PHI) for hos- PHI premium, and it applies for all sub- pitalization. This is one of the highest rates sequent PHI periods. Thus, in a sense, the of PHI coverage in the world for countries Australian arrangements are similar to those with universal health care. This relatively in US jurisdictions that have mandated guar- high rate of PHI coverage has, however, not anteed issue for the uninsured, but have not been achieved without considerable govern- regulated premiums. Both policies “punish” ment intervention. Between 1984 and the late new joiners, either by adding an ABP to the 1990s, PHI coverage fell from 50 percent to community-rated premium or by allowing 30.1 percent of the population as a result of 82 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

adverse selection in community-rated PHI PHI; it is calculated as 1.0 percent of tax- markets. This has been characterized as an able income.12 The ABPs essentially involve adverse selection death spiral. 10 Over the loading the base premium by a fraction that past decade, the Australian government has is calculated as the number of years beyond made a number of attempts to raise the level 30 that a person ﬁ rst took out PHI, times 0.02 of PHI coverage in Australia by introducing ( e.g., a person who joined at 40 years pays a a series of tax-and-subsidy measures and premium that is 1.2 times the base premium age-related late-joiner penalties to boost PHI for the duration of membership). coverage. The range of measures has been discussed in some detail in the existing lit- The Model erature.11 PHI coverage in Australia has not only grown, but also appears to have stabi- Under standard PHI theory, when faced lized in recent quarters. with a choice between a risky income dis- The measures in place in Australian PHI tribution with mean k and a certain income markets, at the time of writing, include: k, a risk-averse individual prefers the latter. Thus, the welfare of a risk-averse person is 1. The subsidy (or rebate) on PHI raised when he or she purchases PHI at the policies; actuarially fair price:13 2. The Medicare Levy Surcharge (MLS)

tax on mid-to-high-income earners Pi = zi Hi (1) who do not have PHI; and 3. The LC scheme, which penalizes peo- where Pi is the actuarially fair premium for ple who join the PHI pool later in life. the ith individual, z i is the probability of the The ABP is calculated by taking the loss event (e.g ., of hospitalization) for the number of years beyond 30 and mul- ith individual and Hi is the value of the loss tiplying this number by 0.02, or 2 per- ( e.g., the cost of the hospital episode) to the cent. Note that the penalty is paid in ith individual if the event occurs. Note that every year after initially purchasing PHI. the premium ( P i) is in fact the price of a PHI policy, not the price of PHI per se . The price Specifi cally, the rebate applies to all PHI of PHI ( p i) is the price per dollar of expected policies based on the age of the oldest per- benefi t or, equivalently, the ratio of the son covered. The subsidy is 30 percent of expected loss to the premium: the premium paid for individuals 65 years, 35 percent of the premium paid for 65 to 69 pi = Pi / zi Hi (2) year-olds, and 40 percent of the premium paid for individuals 70 years and older. The Note that it follows from (1) that Pi /zi Hi subsidy may be taken either as a reduction = 1.00 = p i , i.e. , by defi nition, a premium in the price of the policy at the time of pur- is actuarially fair if the price per dollar of chase, or as a tax rebate when an income expected benefi t is one. tax return is fi led. The MLS is payable by Actuarially unfair prices for PHI are, how- individuals who earn $50,000 or couples/ ever, common place in practice. Typically families that earn $100,000 but do not have emphasized reasons for this include the Private Health Insurance in Australia: Community Rating, but at What Price(s)? 83

existence of asymmetries of knowledge results in unfair prices that are are favorable between the insurer and insured about risk to h s and unfavorable to l s. However, note and loss expectations, for administrative that some l s may still fnd the purchase of PHI loadings, and monopoly pricing.14 Institu- welfare-maximizing. Whether or not this is tional arrangements such as mandated com- the case depends jointly on the degrees of munity rating, where low-expected-loss risk aversion and premium unfairness. individuals (l s) pay the same premium for a An aforementioned source of ineffi ciency policy as high-expected loss individuals (h s), that is associated with community rating is also result in unfair prices. There is also evi- adverse selection, wherein h-types are over- dence, though, that cross-subsidization also represented in the PHI pool. 15 occurs in experience-rated PHI markets. Let z represent the mean risk in the pop- Under community rating, the premium for ulation and be the mean loss. Thus, the PHI is invariant with respect to an individu- community-rated premium is zH. For an al’s risk and loss expectations. For precisely individual, the premium payable for a given this reason the price of PHI, defi ned as the policy, under the Australian ABP scheme price per dollar of expected PHI benefi t, may be represented as: is not uniform across risk and loss types. p Assuming only two risk types, l s and h s, Ci = (1+ A× 0.02)× zH − Ri (7) community rating may be characterized as a system of insurance cross-subsidies16 from l s where A is the number of years beyond 30 to hs. In a competitive community-rated PHI that a person fi rst took out PHI and R i is the market with equal proportions of l s and h s, applicable (age-based) rebate/subsidy. The no excess profts, and no taxes or subsidies of price of PHI, as previously defi ned, (2), thus PHI we may write: becomes

p p pi = Ci / zi Hi Cl = P + l P (3) = [(1+ A× 0.02)× zH − R ] (8) i /zi Hi p Ch = P − h P (4) The imposition of the ABPs is intended l = h (5) to attract and maintain customers from an early age. In particular, the healthy young C p > C p l h (6) may insure against future penalties, which would otherwise arise due to the ABP, by p where C l is the price charged to low- maintaining PHI when p i is less than one. p expected-loss individuals, Ch is the price Thus, it is useful to sum (8) across all ages charged to high-expected-loss individuals, after 30 and compute the implicit (real)

ll is the premium loading for ls, and lh is the annual price of PHI when it is computed for premium discount for h s. Under the assump- a particular joining age. This necessitates p p tions above, C l > $1.00 and C h < $1.00, i.e., adding subscript A , which corresponds to for every dollar of expected beneﬁ t, ls pay the period of initial purchase past age 30, more than one dollar and hs pay less than and L , which corresponds to the point of one dollar. Thus, the uniform premium ( P ) death, as 84 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

L L p The utility functions (10) and (11) are not pi = ∑∑CiA / ziA HiA A==11A estimable. However, recall from (1) that a risk-averse individual will always choose = [(1+ A× 0.02)× zH − RiA ] L the certain equivalent (11) in preference / z H to a risky distribution (10) with the same ∑ iA iA (9) A=1 expected value. By extension, one can infer

= (L − A) [ziA H iA /(1+ A× 0.02) that if the value of (8) < $1.00, the individual L will prefer to purchase PHI rather than self- × zH − R ] ∑ iA insure. Of course, values of (8) > $1.00 are A=1 sometimes consistent with welfare-maxi- Now assume, for simplicity, that all three mizing purchase of PHI for the reasons out- PHI measures apply to all consumers (1. and lined above. We also estimate (9) based on 3., in fact, always do) and that this policy the youngest cohort in the sample. In this article, we use these facts and the available covers the entire loss Hi of insured individuals, otherwise known as full PHI. data to estimate the impact of an array of A utility-maximizing individual’s deci- Australian government policies on the desir- sion whether or not to buy PHI may be ability of PHI. represented as a choice between an uncertain (uninsured state) income distribution with Methods and Data expected income-utility: Using PHI industry data18 on member- E(U ) = zU (Y − H −T) ship and expenditures, along with industry + (1− z)U (Y −T); (10) premium data, we estimate the price of PHI T = f (Y;Y > 50,000) by age and gender, simulate the effects of the applicable taxes and subsidies on those where T is the MLS tax penalty, and the cer- prices, and compare them to actuarially fair tain (insured state) income-utility: (but community-rated) premiums for ﬁ ve- year age groups. Essentially, we estimate (8) and (2) and compare the results of these to U = U (Y − ()1+ A× 0.02)× zH − Ri (11) show whether the price of PHI is actuarially

Recall that Ri is positive for all purchas- fair in each age group. In addition, we com- ers of PHI and T ≥ 0, depending on income. pute the expected income and certain income Thus, neither of these measures is predicted components of (9) and (10) for income levels to lower the likelihood that good risks buy where the tax penalties apply and ask whether PHI. On the other hand, as has been empha- or not, solely on the basis of the income sized elsewhere,17 the incentives produced by tax implications, might individuals/fami- the age-based penalties of the ABP scheme lies purchase a PHI policy even though the do not uniformly increase the attractiveness premium is unfair (i.e ., when (8) > (2))? of PHI (i.e ., increase (11)). The ABPs could, Finally, we produce some illustrative for example, discourage older “low-risk” results of the ABP scheme and tax provi- individuals from initially taking out PHI sions for low-risk old individuals who have while they remain low-risk. not previously held PHI. The only other Private Health Insurance in Australia: Community Rating, but at What Price(s)? 85

work of this nature that has been conducted from the largest Australian private health for Australia was undertaken by Butler.19 H i s insurer, Medibank Private Ltd. The policy work was, however, primarily concerned chosen is called “First Choice Hospital” and with estimating elasticities for PHI and was it contains the most basic inclusions this conducted for a pre-ABP period. insurer offers at the premium of $586.79 in The PHIAC 20 data available to us are fi ve- Australia’s most populous state, New South year age-group aggregates, by gender, dis- Wales.21 There are several reasons that this is aggregated into hospital and ancillary PHI a conservative assumption. First, this policy cover. The data do not enable us to deter- is unlikely to be attractive to older consum- mine which members hold both hospital and ers, high-expected-loss types, and couples ancillary cover, so our exclusive focus is on planning to use private hospital services for the hospital PHI. Quarterly observations are childbirth. Second, and notwithstanding our available on: assumption that full PHI is available, this policy has some coinsurance provisions. (i) The number of members; Finally, note that we do not know how the (ii) The proportion of the Australian benefi ts data are distributed as between indi- population with PHI; viduals and households, let alone the family (iii) The total PHI benefi ts paid; and composition of households that have PHI. (iv) The number of hospital episodes. Although family premiums are computed simply as twice the singles premium, unfor- Since the insured event is hospitalization, the tunately the effect is not a simple linear trans- mean probability of the insured event in the formation with respect to the computation jth age group (zj) may be derived by divid- of our price (8) variable. Specifi cally, when ing (iv) by (i). Similarly, the mean cost per we compute the price of PHI for children of insured event for the (insured) members of dependent age, we essentially ignore the fact the j th age group (zj) may be derived by divid- that the majority of these children must be ing (iii) by (i). This is the mean fair premium covered by parents’ or guardians’ policies. for the jth age group (zj Hj), derived. All price This is quite an important limitation, which data were converted to constant 2006–2007 we address by constructing family unit sce- Australian dollars. Due to seasonality in the narios towards the end of this article and re- quarterly series, we report annual means for computing the price of PHI. our series. Detailed data on the characteristics (e.g ., Results the inclusions and coinsurance provisions) of hospital policies purchased were not availa- Figure 1 presents the central results on ble to us, nor was the value of premiums col- the price of PHI, by age group and gender, lected. Furthermore no matching time-series with and without the rebate. Recall that a of premiums is available. Thus, in order to premium (P i ) is actuarially fair if it results compute the price of PHI, we must make in a price ( p i ) = $1.00 per dollar of expected an assumption about the policy types pur- benefi t: a price of more than $1.00 suggests chased. To be conservative, we selected the a premium that is actuarially unfair and unfa- lowest-price hospital PHI policy available vorable to the insured, while a price of less 86 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

than $1.00 suggests a premium that is unfair groups are, on average, unfair and unfavo- but favorable to the insured. The effect of the rable, with or without the subsidy. The pre- rebate is, universally, to reduce the price of miums for women and persons ages 25 to 39 PHI. However, Table 1 shows that the pre- are unfair but favorable, presumably due to miums for the zero to four to 45 to 49 age the predominance of obstetrics and related

Figure 1. Estimated Mean Prices of PHI in Australia, with and Without the PHI Rebate, by Gender and Age

Mean annual prices (pA) and mean annual subsidized prices (pA-RA) of private health insurance (price per dollar of expected benefi t) for the Ath age group

Age Persons Males Females

pA pA-RA pA pA-RA pA pA-RA 0–4 $2.40 $1.68 $2.17 $1.52 $2.71 $1.90 5–9 $9.43 $6.60 $8.62 $6.04 $10.47 $7.33 10–14 $8.81 $6.17 $8.91 $6.23 $8.71 $6.10 15–19 $3.59 $2.51 $3.88 $2.72 $3.33 $2.33 20–24 $2.36 $1.65 $2.92 $2.04 $2.00 $1.40 25–29 $1.30 $0.91 $3.10 $2.17 $0.89 $0.62 30–34 $1.03 $0.72 $3.22 $2.25 $0.66 $0.46 35–39 $1.26 $0.88 $2.70 $1.89 $0.85 $0.60 40–44 $1.57 $1.10 $2.22 $1.55 $1.24 $0.87 45–49 $1.36 $0.96 $1.63 $1.14 $1.18 $0.83 50–54 $1.04 $0.73 $1.11 $0.78 $0.98 $0.68 55–59 $0.75 $0.53 $0.74 $0.52 $0.77 $0.54 60–64 $0.52 $0.36 $0.49 $0.34 $0.56 $0.39 65–69 $0.36 $0.24 $0.33 $0.22 $0.40 $0.26 70–74 $0.26 $0.16 $0.24 $0.14 $0.30 $0.18 75–79 $0.21 $0.13 $0.19 $0.11 $0.23 $0.14 80–84 $0.19 $0.11 $0.17 $0.10 $0.20 $0.12 85–89 $0.17 $0.10 $0.16 $0.10 $0.18 $0.11 90-94 $0.17 $0.10 $0.16 $0.09 $0.17 $0.10 95+ $0.19 $0.11 $0.17 $0.10 $0.19 $0.11

Note: Data are presented in 2006 Australian dollars (AUD1 = USD0.74; 31 June 2006) and were computed using Equation (8). Sources: Computed from PHIAC (2007) and Medibank Private Ltd (2007). Private Health Insurance Administration Council (PHIAC), “Coverage of Hospital Insurance Tables Offered by Registered Health Benefi ts Organisations by Age Cohort,” http://www.phiac.gov.au/statistics/membershipcoverage/agecohort.htm, accessed Apr. 20, 2007, Canberra: PHIAC (2007); Medibank Private Ltd, Health Cover Options, https://secure.medibank.com.au/join/join_result.asp, accessed Feb. 14, 2007, Medibank Private Ltd: Melbourne (2007). Private Health Insurance in Australia: Community Rating, but at What Price(s)? 87

services in this age group. Gross premiums by these groups still generates an actuarially are closest-to-fair in the 50 to 54 years age unfair and favorable price for the oldest of group, but in all, older age groups are unfair the old. Note that the lifetime average price but favorable to the insured. of PHI for those over 30 is $0.45 overall, The effect of the rebate on actuarial fair- $0.67 for men, and $0.39 for women. ness, around this age group, is the most note- Figure 2 presents the results in Figure 1 worthy: prices net of the subsidy become with the ABPs added by sex and by age that actuarially unfair and favorable to the insured around this age point. Notably, the Figure 2. Estimated Annual Mean Prices of price of PHI for the oldest old is extremely PHI in Australia by Gender and Joining Age, low, with or without the rebate. Without the with 2006 Age-Based Penalties (ABPs) PHI rebate, 70+ year olds were paying 23 cents or less per dollar of expected PHI ben- Persons efi t. With the 40 percent PHI rebate, this age (male/female group now pays less than 14 cents per dollar Joining means) Males Females age p p p of expected benefi t. A A A Finally, note that the fi rst two columns of 32 $1.00 $2.15 $0.68 price data on “persons” provides an effective 37 $1.36 $1.92 $1.08 way of considering whether or not PHI 42 $1.29 $1.53 $1.11 premiums are fair, on average, for an adult 47 $1.05 $1.12 $0.98 couple of the same age. The PHI premium 52 $0.82 $0.80 $0.83 for a couple is simply double that of the sin- 57 $0.59 $0.56 $0.64 gles premium. 62 $0.41 $0.37 $0.44 Of course, the data presented in Figure 1 67 $0.27 $0.24 $0.31 depend on several simplifying assumptions, 72 $0.22 $0.19 $0.24 the most important of which are that: 77 $0.19 $0.17 $0.20 82 $0.17 $0.17 $0.19 (i) The insured population buys a pre- 87 $0.17 $0.15 $0.17 scribed individual, rather than family 92 $0.19 $0.17 $0.19 policy; and (ii) That the policy chosen is the most Note: Data are presented in 2006 Australian dollars (AUD1 = USD0.74; 31 June 2006) and frugal available. were computed using Equation (8). Sources: Computed from PHIAC (2007) and Additionally, the unfair prices in Figure 1 Medibank Private Ltd (2007). Private Health Insurance Administration Council (PHIAC), ignore the ABPs, which make policies for all “Coverage of Hospital Insurance Tables age groups > 25 to 29 more expensive. Note, Offered by Registered Health Benefi ts Organi- though, that the magnitudes of premiums for sations by Age Cohort,” http://www.phiac.gov. au/statistics/membershipcoverage/agecohort. the oldest old suggest that even the applying htm, accessed Apr. 20, 2007, Canberra: the maximum ABP scheme penalty, a 70 per- PHIAC (2007); Medibank Private Ltd, Health cent premium loading for people who join at Cover Options, https://secure.medibank.com. 65 years or older may not be particularly dis- au/join/join_result.asp, accessed Feb. 14, 2007, Medibank Private Ltd: Melbourne (2007). suasive: 1.7 times the prices currently paid 88 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

a person initially purchased PHI. The sub- $50,000 and couples/families that earn in sidy and ABPs are included. The fi rst column excess of $100,000, is the obvious explana- lists the price of PHI p using equation (8). tion. Note, for example, that the price of the For simplicity, we report the ages initially PHI product we selected, net of the 30 per- joined at the mid-points of the age intervals cent subsidy is $410.75 (= 0.70 x $586.79), in Figure 1. For instance, for all persons, the while the tax penalty for not having PHI is price of PHI for a 42 year old buying pri- $500 for a single person who earns $50,001. vate PHI for the fi rst time is $1.29, which is Clearly, there is a strong fi nancial incentive actuarially unfair and unfavorable. Note that for individuals with mid-to-high income to the tax mainly affects the price of PHI at the buy PHI even when the price is actuarially highest ages, which is not surprising because unfair and unfavorable. Taxing individuals it increases with age. This is also the average into PHI is unlikely to be effi cient though, price paid for PHI by age. for reasons that are well-established.22 For instance, a person buying PHI for the fi rst time at age 57 would pay an average of Other Scenarios $0.59 for PHI between the ages of 57 and 92, which is higher than the average price of As was outlined above, a serious limita- $0.45 per year if he or she initially bought PHI tion of the estimates presented in Figures 1 at age 30. The price is higher than he or she and 2 is that they ignore the fact that, under would pay without ABP or if he or she was family policies, dependents essentially get a an insured individual at some point between “free ride.” This is an important issue because ages 30 and 57. However, PHI is still over- the effective price for families with depend- whelmingly actuarially unfair but favorable. ents will be lower than is suggested by the Clearly, the initial tax is too high, which data in Figures 1 and 2. In this section, we may keep low risk individuals from initially present some indicative simulations of fam- buying PHI after age 30. This may in fact ily purchasers of PHI. The simulations are drive out low-risk individuals under age 40 conducted only for those adult age groups who, for whatever reason (liquidity con- in Figure 1 for whom the price of PHI sug- straints, temporary unemployment), did not gested unfair individual premiums. buy PHI at age 30 and therefore must pay The simulations assume that no LC penal- the tax to initiate coverage. At older ages, ties are incurred. Dependents are assumed, the increased expected expenditures far conservatively, to be in the zero to four age outweigh the tax penalties for not maintain- range for household units with up to three ing HI. dependents. This is the dependent age range Thus, this combination of subsidies and with the highest expected benefi t per insured. ABPs may entice only the most risk averse Simulations with a fourth dependent assume to take out and maintain private PHI from that the fourth child is in the fi ve to nine age 30. Why, then, has PHI membership year age range. The results are presented in recently stabilized as a proportion of the Figure 3. Australian population? The income tax Under these scenarios, most household penalty, calculated at one percent of tax- units face actuarially unfair but favorable able income for individuals who earn over prices, i.e ., substantial cross-subsidization Private Health Insurance in Australia: Community Rating, but at What Price(s)? 89

Figure 3. Simulated Annual Mean Prices of PHI in Australia, with and Without the PHI Rebate, by Couple Age, for Couples with Dependent Children

Couples with …

Age of oldest 1 Child 2 Children 3 Children 4 Children

adult on policy pA pA-RA pA pA-RA pA pA-RA pA pA-RA 20–24 $1.08 $0.76 $0.81 $ 0.57 $ 0.65 $0.46 $0.62 $0.44 25–29 $0.70 $0.49 $0.58 $ 0.40 $ 0.49 $0.34 $0.47 $0.33 30–34 $0.58 $0.41 $0.49 $ 0.35 $ 0.43 $0.30 $0.42 $0.29 35–39 $0.69 $0.48 $0.57 $ 0.40 $ 0.48 $0.34 $0.47 $0.33

Note: (i) Computed using Equation (8) and assuming that the household consists of two adult parents/guardians in the same age range, plus dependent children; (ii) based on the following assumptions: in family units three dependent children or fewer, it is assumed that all dependents are in the zero to four years age group (this is the dependent age group with the highest expected benefi t per insured), while for family units with four dependents the eldest dependent is assumed to be fi ve to nine years of age, and the remaining three dependents are assumed to be zero to four years of age; and (iii) presented in 2006 Australian dollars (AUD1 = USD0.74; June 30, 2006). from individuals to families with dependents as the summation of individual, fair, lifetime is evident under the current arrangements. policies. An obvious way to introduce greater fairness Figure 4 presents (second column) the in the premium structure is to re-compute premium multipliers that, if applied for a family premiums as summations of fair age- lifetime of PHI cover, result in fair annual based premiums for individuals. In the next premiums for joiners in each age group, section, we produce the premium multipliers on average. It also shows (third and fourth that enable this to be done, given the present columns) the material effects of these mul- composition of the PHI pool. tipliers on premiums for two PHI products that were offered by Medibank Private Ltd Lifetime Fairness in 2006. These computations suggest, for example, that children who are enrolled from What alternative scheme might work birth (ages zero to four) and maintain cover that would improve effi ciency, but does not would pay just over half of the community- depend on harsh income tax penalties? Tak- rated premium, while individuals ages 70+ ing both community rating and the LC prin- pay more than fi ve times the community- ciples as institutional constraints, our answer rated (or “base” premium). Note that this is to impose a premium structure that creates penalties-and-discounts scheme means that fair lifetime premiums. This can be achieved the oldest fi rst-joiners pay approximately ten for any given joining age, by solving (9) for times the premium that is paid by individuals

pi = $1.00. Furthermore, assume that cross- who have held PHI since early childhood. subsidization from singles to families is also Interestingly, for the ﬁ ve-year age groups abolished: family premiums are determined represented in Figure 4, a linear regression 90 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

Figure 4. Fair Lifetime Premium Multipliers and Annual Premiums by Joining Age

Fair Lifetime Estimated Premiums: Estimated Premiums: Joining Premium Multiplier Most Basic Policy, Most Comprehensive Age for Joining Age No Rebate* Policy, No Rebate** 0–4 0.54 $317 $787 5–9 0.55 $323 $800 10–14 0.72 $420 $1,042 15–19 1.04 $611 $1,515 20–24 1.26 $737 $1,828 25–29 1.45 $848 $2,103 30–34 1.54 $905 $2,244 35–39 1.61 $948 $2,351 40–44 1.77 $1,037 $2,572 45–49 2.11 $1,237 $3,067 50–54 2.59 $1,520 $3,771 55–59 3.19 $1,873 $4,645 60–64 3.86 $2,268 $5,625 65–69 4.52 $2,650 $6,574 70–74 5.04 $2,959 $7,339 75–79 5.38 $3,155 $7,826 80–84 5.56 $3,260 $8,087 85–89 5.66 $3,321 $8,239 90–94 5.56 $3,260 $8,087 95+ 5.26 $3,088 $7,661

Note: (i) *Based on the “First Choice Hospital” policy, with a current base premium of $586.79; (ii) **based on the “Blue Ribbon Hospital” policy, with a current base pre-

mium of $1455.57; (iii) premium multipliers are computed by setting pi = 1 in Equation (9); and (iv) presented in 2006 Australian dollars (AUD1 = USD0.74; June 30, 2006). Sources: Computed from PHIAC (2007) and Medibank Private Ltd (2007). Private Health Insurance Administration Council (PHIAC), “Coverage of Hospital Insurance Tables Offered by Registered Health Benefi ts Organisations by Age Cohort,” http:// www.phiac.gov.au/statistics/membershipcoverage/agecohort.htm, accessed Apr. 20, 2007, Canberra: PHIAC (2007); Medibank Private Ltd, Health Cover Options, https://secure.medibank.com.au/join/join_result.asp, accessed Feb. 14, 2007, Med- ibank Private Ltd: Melbourne (2007). also fi ts the data very well: with only a this coeffi cient suggests that an ABP (or dis- constant and age-step “trend,” the coeffi - count, as the case may be) of the order of 6 cient of determination (R2) is approximately percent per annum on average is a reasona- 0.95. The age-based trend coeffi cient is 0.31. ble approximation to our premium multipli- Since our regression uses fi ve-year intervals, ers. This ABP is three times the magnitude Private Health Insurance in Australia: Community Rating, but at What Price(s)? 91

of that which currently applies under the people may be too harshly penalized by the Australian LC scheme, even when extended scheme (or have earned entitlements that we to individuals ages < 31. do not account for); and so forth. A system of transfer payments may be an effi cient way Discussion to address equity concerns of this kind. One important limitation of our study is In the United States., many states have that the data available to us pertain to pur- guaranteed issue laws for the uninsured. In chasers of PHI. One would typically expect the individual market, high-risk individu- the self-selected pool to constitute an adverse als cannot be denied the right to purchase selection of individuals, although some recent PHI. For those with PHI, HIPPA regulations evidence25 suggests that the tax penalties on ensure that persons currently insured cannot mid-to-high income individuals for not pur- be dropped nor have their premiums raised chasing PHI may confound this prediction. to prohibitively high levels. While these laws In any event, improvements in the risk- increase access for high-risk individuals, PHI composition of the pool would necessitate is too harsh on initial non-purchasers and too recalibrations of the age-based tax rates we lenient on those initially buying PHI at higher have proposed here. Moreover, one would ages. Applying, for example, Herring and wish to base the initial calibration and recali- Pauly’s23 fi nding that high-risk older males bration on more fi nely disaggregated indus- consume approximately eight times more try data than are publicly available. benefi ts than low-risk males, one can easily Finally, the Australian government has see that a cap on the maximum difference recently increased the MLS income thresh- between premiums paid of 0.7 times will not olds from $50,000 to $73,000 for singles and induce low-risk individuals to purchase and from $100,000 to $146,000 for couples/fam- maintain PHI. Indeed, it is interesting to note ilies. Our results suggest that many young that our proposed premium multipliers lead individuals (ages < 31) who earn less than to premium relativities that are of the order the threshold amounts will have a strong discovered by Herring and Pauly. 24 fi nancial incentive to drop their PHI cover- Our proposal is likely to strike protestation age as a result of the threshold changes. By on equity grounds. One may, for example, contrast, the modifi cations of the ABPs we object that young adults should not be cap- propose would make the price of PHI fairer tives of the historical decisions of their par- and encourage younger people to join and ents (i.e ., be ineligible for a discount because maintain PHI, without imposing punitive their parents did not buy PHI); that elderly taxes on non-joiners.

REFERENCES

1. Butler, James RG, “Policy Change and Private 3. Hoffman, Catherine, Wang, Marie, “Health Health Insurance: Did the Cheapest Policy Insurance Coverage in America: 2001 Data Do the Trick?” Australia Health Review , 25, Update,” Washington, DC, The Kaiser Com- 33–41 (2002). mission on Medicaid and the Uninsured, 2. Id. 2003, http://www.k_.org/uninsured/4154. 92 JOURNAL OF HEALTH CARE FINANCE/Summer 2010

cfm , accessed Sept. 17, 2007, Henry J. Kaiser 13. Arrow, Kenneth J, “Uncertainty and the Wel- Family Foundation: Washington DC. fare Economics of Medical Care,” The Ameri- 4. Herring, Bradley, Pauly, Mark V, “The Effect of can Economic Review , 53: 941–973 (1963). State Community Rating Regulations on Pre- 14. Supra , n.9. miums and Coverage in the Individual Health 15. Herring, Bradley and Pauly, Mark V, “Incen- Insurance Market,” National Bureau of Eco- tive-Compatible Guaranteed Renewable nomic Research (NBER) Working Paper, New Health Insurance Premiums,” Journal of York: NBER (2006). Health Economics , 25, 395–417 (2006). 5. United States Department of Health and 16. Pauly, Mark V, “The Welfare Economics of Human Services, “Health Insurance Reform Community Rating, The Journal of Risk and and You,” http://www.healthreform.gov/ , accessed Insurance , 37, 407–418 (1970); Pupp, Roger Apr.28, 2010, Washington: Department of L, “Community Rate and Cross Subsidies in Health and Human Services (2010). Health Insurance,” The Journal of Risk and 6. Balkin, Jack, M, “The Constitutionality of the Insurance , 48, 610–627. (1981). Individual Mandate for Health Insurance,” 17. Brown III, H Shelton, Connelly, Luke B, “Life- New England Journal of Medicine , 362: time Cover in Private Insurance Markets,” Inter- 482–483. national Journal of Health Care, Finance and 7. Baker, Peter, “Obama, in Maine, Needles Economics , 5, 75–88; Connelly, Luke B, Brown Republicans Over Calls to Repeal Health III, H, Shelton, “Lifetime Subsidies in Australian Care Overhaul,” The New York Times , http:// Private Health Insurance Markets with Commu- www.nytimes.com/2010/04/02/us/politics/ nity Rating,” Geneva Papers on Risk and Insur- 02obama.html , accessed Apr.28 2010, New ance-Issues and Practice, 31, 705–719 (2006). York: The New York Times Company. 18. Private Health Insurance Administration Coun- 8. Supra , n.4. cil (PHIAC), “Coverage of Hospital Insurance 9. Rothschild, Michael, Stiglitz, Joseph, E, “Equi- Tables Offered by Registered Health Beneﬁ ts librium in Competitive Insurance Markets: An Organisations by Age Cohort,” http://www. Essay on the Economics of Imperfect Informa- phiac.gov.au/statistics/membershipcoverage/ tion,” The Quarterly Journal of Economics , agecohort.htm , accessed Apr. 20, 2007, Can- 90, 630–649, Nov. 1976. berra: PHIAC (2007). 10. Supra , n.1. 19. Butler, James RG, “Estimating Elasticities of 11. See, e.g., Hall, J, De Abreu Lourenco, R, Demand for Private Health Insurance in Aus- Viney, R, “Carrots and Sticks the Fall and tralia,” National Centre for Epidemiology and Fall of Private Health Insurance in Australia,” Population Health (NCEPH) Working Paper Health Economics , 8, 653–660 (1999); supra , No. 43, Canberra: NCEPH (1999). n.1; Brown III, H Shelton, Connelly, Luke B, 20. Supra , n.18. “Lifetime Cover in Private Insurance Markets,” 21. See www.medibank.com.au for details. Med- International Journal of Health Care, Finance ibank Private Ltd, Health Cover Options, and Economics , 5, 75–88; Connelly, Luke B, https://secure.medibank.com.au/join/join_ Brown III, H, Shelton, “Lifetime Subsidies in result.asp , accessed Feb. 14, 2007, Medibank Australian Private Health Insurance Markets Private Ltd: Melbourne (2007). with Community Rating,” Geneva Papers on 22. Pauly, Mark V, “The Welfare Economics of Risk and Insurance-Issues and Practice , 31, Community Rating, The Journal of Risk and 705–719 (2006). Insurance , 37, 407–418 (1970). 12. These thresholds were, for the ﬁ rst time since 23. Supra , n.15. the introduction of the MLS, adjusted recently. 24. Id. The thresholds are now set at $73,000 25. Australian Bureau of Statistics, National for individuals and $146,000 for couples/ Health Survey-Summary of Results 2004–05 , families. Canberra: ABS (2006).