Health Aff 2005 Horwitz 790 801

At the Intersection of Health, Health Care and Policy

doi: 10.1377/hlthaff.24.3.790

, 24, no.3 (2005):790-801Health AffairsHospitals

Making Profits And Providing Care: Comparing Nonprofit, For-Profit, And GovernmentJill R. Horwitz

Cite this article as:

http://content.healthaffairs.org/content/24/3/790.full.html

available at: The online version of this article, along with updated information and services, is

For Reprints, Links & Permissions: http://healthaffairs.org/1340_reprints.php

http://content.healthaffairs.org/subscriptions/etoc.dtlE-mail Alerts : http://content.healthaffairs.org/subscriptions/online.shtmlTo Subscribe:

from the Publisher. All rights reserved.including photocopying or by information storage or retrieval systems, without prior written permission may be reproduced, displayed, or transmitted in any form or by any means, electronic or mechanical,

Health Affairs Foundation. As provided by United States copyright law (Title 17, U.S. Code), no part of by Project HOPE - The People-to-People Health2005Bethesda, MD 20814-6133. Copyright ©

is published monthly by Project HOPE at 7500 Old Georgetown Road, Suite 600,Health Affairs

Not for commercial use or unauthorized distribution

by guest on August 26, 2014Health Affairs by content.healthaffairs.orgDownloaded from


http://www.healthaffairs.org

http://content.healthaffairs.org/content/24/3/790.full.html

http://healthaffairs.org/1340_reprints.php

http://content.healthaffairs.org/subscriptions/etoc.dtl

http://content.healthaffairs.org/subscriptions/online.shtml

http://content.healthaffairs.org/


M a r k e t Watc h

Making Profits And Providing Care: ComparingNonprofit, For-Profit, And Government HospitalsDiscussion of the value of nonprofit hospital ownership must accountfor the differences in service offerings among hospital types.

by Jill R. Horwitz

ABSTRACT: Three types of entities—nonprofit, for-profit, and government—own hospitals.Yet we know neither whether hospital types specialize in different medical services nor howservice profitability affects specialization. In this econometric analysis of American HospitalAssociation data for every U.S. urban, acute care hospital (1988–2000), more than thirtyservices were categorized as relatively profitable, unprofitable, or variable. For-profits aremost likely to offer relatively profitable medical services; government hospitals are mostlikely to offer relatively unprofitable services; nonprofits often fall in the middle. For-profitsare also more responsive to changes in service profitability than the other two types.

Ho w d o f o r- p r o f i t hospitals makeprofits? One theory is that they man-age their case-mix more carefully

than nonprofit and government hospitals do.For example, they differentially locate in ar-eas with relatively well-insured patients.1

Underlying this concept of profit making isthe assumption that all general hospitals, re-gardless of ownership, are alike in the types ofmedical services they provide. They merelydiffer in their patient mix.

This paper advances a different mechanismof profit making. It proposes that for-profithospitals are more likely than other types todecide which medical services to offer basedon service profitability. Under this theory, gen-eral hospitals do not provide a standard arrayof medical services. Instead, they specialize.Analyzing 1988–2000 data on medical serviceprovision for every U.S. urban, general hospi-tal, this paper tests whether hospital typesspecialize in services based on profitability.

Which profit-making method hospitals adopt,and to what degree hospital types employthem, matters a great deal for tax policy, struc-turing reimbursement, and quality of care.

Two-thirds of all U.S. urban hospitals arenonprofit, with the remainder split betweenfor-profit and government ownership. Thesehospital types operate under different legalrules. For-profits may distribute accountingprofits to shareholders, whereas governmentand nonprofit hospitals enjoy income andproperty tax exemptions. However, there isreason to expect all hospitals to provide a sim-ilar array of medical services: General hospi-tals all treat patients with a mix of needs, con-tract with the same insurers and governmentpayers, operate under the same health regula-tions, and employ staff with the same trainingand ethical obligations. Consequently, it is notsurprising that much of the empirical literatureon corporate ownership finds little differenceamong hospital types.2

7 9 0 M a y / J u n e 2 0 0 5

H e a l t h T r a c k i n g

Jill Horwitz ([email protected]) is an assistant professor of law at the University of Michigan Law School inAnn Arbor.

DOI 10.1377/hlthaff.24.3.790 ©2005 Project HOPE–The People-to-People Health Foundation, Inc.



Although the ownership literature is volu-minous and widely debated in the UnitedStates and abroad, it is incomplete.3 Studieshave mainly examined financial topics such ascosts, profits, billing, the value of uncompen-sated care, and responsiveness to financialpressure. Important subjects such as quality,physician control, and patient access havebeen studied less frequently, have focused on alimited number of services, and have generatedinconclusive results.4

Researchers have not previously lookedbroadly at whether any hospital types system-atically offer relatively profitable services andavoid relatively unprofitable ones. To that end,this study investigates how ownership affectsthe central activity of hospitals: medical careprovision. It evaluates more than thirty ser-vices, ranging from sports medicine to AIDStreatment, to ask whether ownership is corre-lated with offering services and how thosechoices relate to profit seeking.

Study Data And Methods� Hospital population. Data are from the

American Hospital Association’s (AHA’s) An-nual Surveys of Hospitals during 1988–2000and the 1990 U.S. census. The AHA survey hadalmost a 20 percent nonresponse rate, andnonrespondents were disproportionately for-profit hospitals (1988: 3 percent nonprofit, 4percent government, 18 percent for-profit;2000: 14 percent nonprofit, 20 percent govern-ment, 26 percent for-profit). Since the samehospitals were not missing in all years, how-ever, the results should be valid. The analysisincluded all nonrural, acute care hospitals thatoperate in metropolitan statistical areas(MSAs) with at least two general medical andsurgical hospitals, excluding military, uncate-gorized federal, and prison hospitals. I ex-cluded rural hospitals because there are rela-tively few for-profit hospitals in rural areas andthey provide a limited range of services. In 1995,for example, of the roughly 2,500 rural hospi-tals, approximately 8 percent were for-profit,and only a few provided open-heart surgery.

� Study variables. Hospitals were classi-fied by ownership: nonprofit, for-profit, and

government. They were asked whether eachservice was offered at the hospital or anotherhospital, or was not available (1988–1993), orwhether the service was offered at the hospitalor a subsidiary, another location in the system,the network, or a joint venture (1994–2000).Because the study assessed hospital-based ser-vices, the first categories in the two survey ver-sions were treated as equivalent. Roughly fortyvariables were included in the data on acutecare services, of which thirty-two were suffi-ciently well-defined and consistent across therelevant time period to be useable (Exhibit 1).

I imputed missing values using data fromthe years before and after the missing year. Forthe end years (1988, 2000) I based the imputa-tion on the next or previous two years. Whereseveral values were missing, I excluded the ob-servations. Generally, I imputed fewer than 4percent of observations for each service.

� Determining profitability. To test hos-pitals’ responsiveness to service profitability, Isorted services into three categories: relativelyhigh, low, and variable profitability (see Ex-hibit 1). The classifications were based primar-ily on the peer-reviewed medical and socialscience literature. Also, because Medicare pay-ments are the largest single source of hospitalrevenues, I consulted Medicare Payment Advi-sory Commission (MedPAC) and ProspectivePayment Assessment Commission (ProPAC)reports to Congress for the relevant years.5 In-terviews with hospital administrators anddoctors, as well as analyses of the socioeco-nomic or insurance status of patients likely todemand various services, supported the casethat a service was relatively profitable or un-profitable.

Because this project is concerned primarilywith hospitals’ behavior and motivation, Ichecked the scientific literature with a reviewof trade publications, business magazines, andnewspaper reports. The qualitative evaluationof relative service profitability was critical tothe project because perceptions of whether aservice would be profitable were likely as im-portant determinants of service investmentchoices as whether services turned out to beprofitable. The various sources yielded re-

M a r k e t W a t c h

H E A L T H A F F A I R S ~ V o l u m e 2 4 , N u m b e r 3 7 9 1



markably similar results.6 Abbreviated exam-ples of three services follow.

Open-heart surgery, for example, is rela-tively profitable. Like almost all surgical andrelated services, cardiac services—includingcardiac catheterization labs, angioplasty, andcoronary artery bypass graft (CABG)—arewidely known to be hospital profit centers.7

Insurers typically reimburse heart attacktreatments at high rates, patients receivingCABG are unusually well insured since mostare covered by Medicare, and there is high andincreasing spending on cardiac care.8 From1984 to 1994, the real price for bypass surgeryamong Medicare patients rose 2.3 percent an-nually, from $29,176 to $36,564 (1991 dollars),

7 9 2 M a y / J u n e 2 0 0 5


EXHIBIT 1Percentage Of Hospitals Offering Services And Relative Profitability Status, 1988–2000

Service PercentRelativelyprofitable

Relativelyunprofitable Variable

AIDS (outpatient)AIDS servicesAIDS unitAlcohol bedsAlcohol/drug (outpatient)

11544

3033

●

●

●

●

●

AngioplastyBirthing roomBurn treatmentCardiac catheterization labComputed tomography (CT) scanner

40695

5492

●

●

●

●

●

Child psychiatric servicesDiagnostic radioisotope facilityEmergency roomExtracorporeal shock-wave lithotripterFitness center

2581961724

●

●

●

●

●

HIV testHome healthMagnetic resonance imaging (MRI)Neonatal intensive careObstetrics (beds)

6044463573

●

●

●

●

●

Obstetrics (births)Open-heart surgeryOrthopedic surgeryPediatric intensive care unitPositron emission tomography

713492226

●

●

●

●

●

Psychiatric (inpatient)Psychiatric emergency servicesSkilled nursingSingle photon emission CT

49483545 ●

●

●

●

Sports medicineTrauma centerUltrasoundWomen’s center

32259647

●

●

●

●

SOURCE: Author’s analysis of American Hospital Association Annual Surveys, 1988–2000.

NOTE: Includes all nonrural, general medical and surgical hospitals, in metropolitan statistical areas with more than onehospital.



while the share of patients receiving the treat-ment increased by one percentage point annu-ally, from 5 percent to 15 percent.9 The costs ofsupplying CABG in real terms either were flator fell during the period.10 In 1991, becausespending on bypass surgery was so high, theCenters for Medicare and Medicaid Services(CMS, then HCFA) ran a pilot program inwhich hospitals and physicians negotiatedprices.11 During the study period, for-profitcorporations opened single-service cardiacsurgery centers, while neighboring hospitalscomplained of losing profitable business.12

Conversely, hospital-based, psychiatricemergency services are relatively unprofitablefor several reasons: (1) the emergency depart-ment (ED) is a comparatively unprofitable set-ting, which attracts patients whose admis-sions are more costly than those of patientswho are admitted to the hospital by othermeans; (2) psychiatric care reimbursement isuncertain and often low relative to cost; and(3) the care attracts a poor, poorly insured,sick, and difficult-to-manage population.13

Compared with that of acute care, the prof-itability of postacute services varied dramati-cally during the 1980s and 1990s. With legalchallenges to reimbursement resolved and thehospital prospective payment system (PPS)implemented, postacute services becamehighly profitable by the early 1990s.14 Unlikeacute services diagnosis-related group (DRG)payments in which hospitals receive a per epi-sode payment for each patient, Medicare paidhospitals a cost-related reimbursement forpostacute services. Hospitals could increasereimbursements by unbundling the servicesand transferring patients to postacute care atthe end of their hospital stay.15 There is consid-erable evidence of these transfers. From 1981 to1998 the number of inpatient days for Medi-care beneficiaries fell at an average annual rateof 4.1 percent; from 1986 to 1998 home healthvisits for Medicare beneficiaries grew at an av-erage annual rate of 15.6 percent.16 Paymentswere particularly generous to new entrants,with skilled nursing facilities (SNFs) andhome health services exempt from cost limitsfor the first years of operation. In fact, home

health payments grew from $3.9 billion tomore than $18.3 billion between 1990 and1996.17 Hospital administrators and regulatorsalike recognized the profitability potential ofpostacute care.18

With passage of the 1997 Balanced BudgetAct (BBA) of 1997, however, the profit-makingopportunities plummeted. Medicare pay-ments were reduced, the CMS developed aPPS for postacute services, and home healthcare spending fell by a factor of two.

� Statistical analysis. Using a probitmodel, I analyzed whether ownership was cor-related with the probability of offering each ofapproximately thirty medical services (Ex-hibit 1). Independent variables included own-ership, year, and the interaction of the two.

To correct for potential biases, I controlledfor hospital and market characteristics. Hospi-tal variables included size (measured by ad-missions), teaching status (measured by teach-ing association membership), and a dummyvariable for location by geographic region.Market characteristics included patients’ sex,race, household income, and age. These werecompiled using the ten-mile radii around thecenters of the hospitals’ ZIP codes, the meandistance that captures 75 percent of dischargesamong acute care urban hospitals.19

I adjusted for heteroskedasticity, and, be-cause the probability of offering a service is notindependent among years, I allowed for an ar-bitrary covariance matrix within each hospitalover time. By varying the hospital’s ownershiptype while holding the independent variablesconstant (at 1994 levels or the next-closestyear), I predicted the probabilities that eachhospital in each year would offer a service. Ithen averaged the individual predicted proba-bilities to obtain a single probability that ahospital type would offer a service each year.

I performed sensitivity tests on the threeservices described above. Because size is thebest predictor of service offering, I restrictedthe regressions to the observations in the top,bottom, and middle two quartiles of hospitalsmeasured by admissions.

To ensure that ownership, rather than ge-ography, explained these results, I ran several





sensitivity tests related to hospital regions.John Wennberg and others have observed thatmedical service provision varies considerablyby small region.20 Consequently, one mightthink that firm types chose where to operatebased on the local character of demand. Itested this alternative explanation for the re-sults by using a fixed-effects approach, includ-ing an indicator variable for the year 2000 Hos-pital Referral Regions (HRRs) in which eachhospital operates. I also altered the region vari-able to account for areas of high for-profit pen-etration (for example, the South andSouthwest) and included dummy variables forall nine AHA regions.

To test the sensitivity to other market char-acteristics, I added age-squared categories forthe percentage of the population over agesixty-five and over age eighty. Because statepayment policies for mental health servicesvary considerably, I included state dummiesand state-year interactions for the psychiatricemergency service estimations. In addition, totest variation within the government hospitalcategory, I excluded hospitals in the Depart-ment of Veterans Affairs (VA) system.

Sensitivity tests also included propensityscore analysis, a method used to make causalinferences when assignment to a group, suchas nonprofit ownership status, is not random.21

The results confirmed that I compared hospi-tals that differed primarily by ownership and

not other hospital characteristics such as hos-pital size. More specifically, I determined theconditional probability of corporate owner-ship (nonprofit versus for-profit; nonprofitversus government, government versus for-profit), given the observed characteristics usedin the Probit estimates (the propensity scores),created five subcategories defined by the esti-mated propensity score, and predicted theprobability of a hospital type’s offering a ser-vice in a given year, controlling for the propen-sity grouping.

Study ResultsThe results demonstrate that among com-

parable hospitals, for-profits are more likelythan nonprofits, which in turn are more likelythan government hospitals, to offer open-heartsurgery. The magnitude of these differencesseems large, given the importance of the deci-sion to offer the service and the constraints todoing so. For-profits are, on average, 13.0 per-centage points more likely than governmenthospitals (40.9 percent versus 27.9 percent, p <.001) and 7.3 percentage points more likelythan nonprofit hospitals (40.9 percent versus33.6 percent, p < .001) to offer open-heart sur-gery (Exhibit 2). Restricting the data to thesmallest hospitals—those in the bottom twoadmissions quartiles—the nonprofit-govern-ment difference for open-heart surgery was in-significant. This result is expected because so

7 9 4 M a y / J u n e 2 0 0 5


EXHIBIT 2Hospitals’ Probability Of Offering Open-Heart Surgery, By Ownership Type, 1988–2000

SOURCE: Author’s analysis of data from American Hospital Association Annual Surveys, 1988—2000.NOTES: Probit predicted probabilities include all nonrural, general medical and surgical hospitals in metropolitan statisticalareas with more than one hospital. values are based on the chi-square test of the differences between average predictedPprobabilities of offering services in 1988–2000 by hospital type. Not-for-profit versus for-profit, < .001; not-for-profit versuspgovernment, = .001; for-profit versus government, < .001.p p

50

40

30

20

Probability (percent)

1988 1990 1992 1994 1996 1998 2000

Not-for-profit

Government

For-profit



few small hospitals offer open-heart surgery.Unlike open-heart surgery, for-profits are

less likely than nonprofits, which in turn areless likely than government hospitals, to offerpsychiatric emergency care (Exhibit 3). On av-erage from 1988 to 2000, controlling for othercharacteristics, 41 percent of for-profit hospi-tals offered psychiatric emergency services,compared with 48 percent of nonprofit and 56percent of government hospitals. Again, themagnitude of these differences is large: For-profits are 15.0 percentage points less likelythan government hospitals and 8.4 percentagepoints less likely than nonprofit hospitals (p <.01) to offer psychiatric emergency care.Among the smallest hospitals, the nonprofitversus for-profit difference was insignificant,as was the difference between nonprofit andnon-VA government hospitals.

Not only did the probability of offeringhome health care vary by ownership, but therelative differences among types varied overtime. The probability of offering home healthservices when the service was profitable in-creased for all three hospital types. However,among for-profits, growth of home health carewhen profitable and decline when unprofit-able were particularly dramatic (Exhibit 4).From 1988 to 1996, the probability of a for-

profit hospital’s offering home health servicesmore than tripled (17.5 percent to 60.9 per-cent), controlling for hospital and marketcharacteristics. During the same period, theprobability of offering home health care onlygrew slightly more than ten percentage points(40.9 percent to 51.7 percent) for nonprofitand fourteen percentage points (38.1 percentto 51.9 percent) for government hospitals.From 1997 to 2000, as home health care be-came relatively unprofitable with the imple-mentation of the BBA of 1997, the probabilityof offering it fell a striking 37.5 percentagepoints among for-profit, 7.7 percentage pointsamong nonprofit, and 1.5 percentage pointsamong government hospitals.

� Other services. Tests of more thanthirty other services yielded similar results(Exhibit 5). While for-profit hospitals wereonly somewhat more likely than nonprofits tooffer relatively profitable services, both for-profit and nonprofit hospitals were consider-ably more likely than government hospitals tooffer relatively profitable services. For-profitswere less likely than nonprofits, which in turnwere less likely than government hospitals, tooffer relatively unprofitable services. For-profit hospitals were more responsive than theother types were to rapid changes in service



EXHIBIT 3Hospitals’ Probability Of Offering Psychiatric Emergency Services, By Ownership Type,1988–2000

SOURCE: Author’s analysis of data from American Hospital Association Annual Surveys, 1988—2000.NOTES: Probit predicted probabilities include all nonrural, general medical and surgical hospitals in metropolitan statisticalareas with more than one hospital. values are based on the chi-square test of the differences between average predictedPprobabilities of offering services in 1988–2000 by hospital type. Not-for-profit versus for-profit, = .001; not-for-profit versuspgovernment, < .001; for-profit versus government, < .001.p p

55

50

45


1988 1990 1992 1994 1996 1998 2000

Not-for-profit

Government

For-profit40

35



profitability (Exhibit 6). The average probabil-ity of a hospital type’s offering a medical ser-vice from 1988 through 2000, controlling forother characteristics, is reported in Exhibit 7.

Discussion And Policy ImplicationsMedical service offerings vary markedly by

ownership, likely because hospital types adoptor prioritize goals differently. Although allhospitals must earn sufficient profits to oper-ate, the evidence here suggests that for-profitsare more likely to respond to profitability thanthe other types are when making supply deci-sions. Since government hospitals are mostlikely to supply the unprofitable services thatare disproportionately needed by poor and un-derinsured patients, the evidence also suggeststhat such hospitals are caregivers of last resort.Nonprofit hospitals are often the intermediatetype in terms of balancing profit seeking andserving the poor through service choices.

Neither the profit-making versus charitabledivide (for-profit versus nonprofit and govern-ment) nor the public versus private divide(government versus nonprofit and for-profit)fully predicts behavior. These results suggestthat corporate form can be used as a regula-tory tool when regulations are designed to ac-

count for the motivations and relative respon-siveness to profitability of all three types. Forexample, ownership could be considered indesigning reimbursement policies. In addition,if they hope to secure a full range of services,states’ attorneys general should consider themix of types in a market when overseeing hos-pital conversions.

A further implication is that the measuretraditionally used to justify nonprofit tax ex-emptions—the provision of uncompensatedcare—is too limited. Recent calls to eliminatenonprofit subsidies, commonly founded on theclaim that there are no important differencesamong corporate types, should be rejected.22

Differences in service mix matter to all pa-tients, not only the uninsured and nonpayingpatients.

Although this paper has not addressedhealth outcomes, it does raise the question ofthe relationship between service profitabilityand medical appropriateness. As discussed inthe home health example above, for-profits’ re-sponsiveness to incentives is noteworthy forits magnitude and speed. However, we do notknow whether the most medically appropriatemix of hospital services is the most profitable.Public payment rates are set through a com-

7 9 6 M a y / J u n e 2 0 0 5


EXHIBIT 4Hospitals’ Probability Of Offering Home Health Services, By Ownership Type,1988–2000

SOURCE: Author’s analysis of data from American Hospital Association Annual Surveys, 1988—2000.NOTES: Probit predicted probabilities include all nonrural, general medical and surgical hospitals in metropolitan statisticalareas with more than one hospital. values are based on the chi-square test of the differences between average predictedPprobabilities of offering services in 1988–2000 by hospital type. Not-for-profit versus for-profit, < .001; not-for-profit versuspgovernment, = .071; for-profit versus government, < .001.p p

50

40

30


1988 1990 1992 1994 1996 1998 2000

Not-for-profit

GovernmentFor-profit

20

10

60



plex and changing process based on, amongother factors, the evolving judgment of ratesetters, imperfect adjustments for hospitalmarkets’ demographic and geographic charac-

teristics, and the political strength of inter-ested parties. Private payment rates resultfrom complex negotiations and relative bar-gaining power. We need more study on how



EXHIBIT 5Comparison Of Services Offered At Study Hospitals, By Ownership, 1988–2000

Relatively profitable services F > N F > G N > G

Angioplasty (1989–2000)Birthing rooma

Cardiac catheterization labComputed tomography (CT) scanner

Y***N*Y***N

Y***NY***Y

Y***YY***Y*

Diagnostic radioisotope facilityExtracorporeal shock-wave lithotripterFitness centerMagnetic resonance imaging (MRI)

N*Y***N**Y

Y***Y***NY***

Y***Y***Y**Y***

Neonatal intensive care (>0 beds)a

Open-heart surgeryOrthopedic surgery (1989–1993)Pediatric intensive care (>1 beds)a

Y***Y***NY***

Y***Y***Y***Y***

N***Y***Y***N***

Positron emission tomography (1990–2000)Single photon emission CT (1990–2000)Sports medicineUltrasoundWomen’s centera

YN**=N***Y***

Y*YY***NY***

YY***Y***YY*

Relatively unprofitable services

AIDS (outpatient) (1988–1993)AIDS services (1994–2000)AIDS unit (1988–1993)Alcohol/drug inpatient (>1 beds)

NN***Y**Y***

N***N***NY*

N***N***N***N***

Alcohol/drug outpatientBurn treatment (>0 beds)Child/adolescent psychiatric (>0 beds)a

Emergency roomEmergency rooma

N***YNN**N*

N***N*N*Y=

N***N***NY***Y

HIV test (1988–1991)Obstetrics (>2 beds)a

Obstetrics (≥100 births)a

Psychiatric inpatient (1989–2000, >1 beds)

NNN***Y**

N*NN**N***

N*NNN***

Psychiatric emergency servicesPsychiatric emergency servicesa

Trauma centerTrauma centera

N***N***N**N**

N***N***NN***

N***NYN***

SOURCE: Author’s analysis of data from American Hospital Association Annual Surveys, 1988–2000.

NOTES : F is for-profit. N is not-for-profit. G is government. Equal sign denotes that the difference of predicted probability ofoffering service between firms is less than .003. F > N means that, on average over the study period, for-profits were morelikely than nonprofits to offer the specified services, with a difference of at least .003.a Excluding Veterans Affairs (VA) hospitals.

*p < .10 **p < .05 ***p < .01



and, indeed, whether these processes produceincentives for hospitals to provide a medicallyappropriate service mix.

� Study limitations. Despite the method-ological strength of this work—particularlythe focus on medical services rather than fi-nancial behavior and the breadth of servicesstudied—it has limitations. First, as discussedabove, the nonrespondents were dispropor-tionately for-profit. The data are self-reportedand not independently verified, although thereis no reason to suspect that data reliability iscorrelated with ownership.

Second, despite the rigorous nature of theresearch used to determine profitability, a noteof caution is necessary. Profitability is not aninherent attribute of medical services; rather,it depends on institution-specific factors suchas management skills, case-mix, and local in-put costs. Further, even within a single hospi-tal, costs and charges differ, discounts vary byindividual payer, and allocation of joint costsblur the profitability picture. Despite thesecomplications, however, one can reasonablycompare the relative profitability of servicesdefined as bluntly as they are in the AHA data.

Wh e t h e r n o n p r o f i t organiza-tions behave differently fromother ownership types, particu-

larly in the hospital industry, has raised con-siderable debate. In 2004 more than fifty law-suits alleging that nonprofit hospitals haveviolated their charitable obligations werefiled in federal district courts alone.23 Con-

gress is considering an extensive overhaul ofthe nonprofit sector’s regulatory regime to in-crease accountability.24 Any discussion of thevalue of nonprofit hospital ownership mustaccount for the significant differences in ser-vice offerings among hospital types and howthose offerings vary according to profitability.

The author thanks Jia-Jia Ye for research assistanceand Paula Payton for administrative assistance. Shethanks Allan Brandt, Suzanne Cooper, David Cutler,Phoebe Ellsworth, Don Herzog, Nancy Keating, MaryBeth Landrum, Barbara McNeil, Martha Minow, CarlMorris, David Naylor, Joseph Newhouse, EdwardParson, Jonathan Skinner, Richard Zeckhauser, threeanonymous referees, and participants in the November2003 National Bureau of Economic Research (NBER)Health Care meetings for helpful comments. This studywas supported by a grant from the National Instituteon Aging to the NBER (Grant no. T32-AG00186); theHauser Center for Nonprofit Organizations; theHarvard University doctoral program in health policy;and the University of Michigan Law School, John M.Olin Center for Law and Economics.

7 9 8 M a y / J u n e 2 0 0 5


EXHIBIT 6Comparison Of Probability Of Offering Services With Variable Profits, By OwnershipType, 1992–1996 And 1997–2000

Profitable (1992–1996) Unprofitable (1997–2000)

Service F N G F N G

Home healthSkilled nursing

39.328.1

9.715.4

12.74.9

–37.62.8

–7.74.7

–1.59.7

SOURCE: Author’s analysis of data from American Hospital Association Annual Surveys, 1988–2000.

NOTES: Includes all nonrural, general medical and surgical hospitals, in metropolitan statistical areas with more than onehospital. F is for-profit. N is not-for-profit. G is government. Values are the percentage-point changes in probability of offeringservice during the years indicated.





EXHIBIT 7Hospitals’ Predicted Probability Of Offering Various Services, By Hospital OwnershipType, 1988–2000

Probability (%)

Service Not-for-profit Government For-profit

AIDS (outpatient) (1988–1993)AIDS services (1994–2000)AIDS test (1988–1992)AIDS unit (1988–1993)

8.153.163.0

3.3

26.265.366.3

7.5

6.945.362.3

5.5

Alcohol/drug inpatient (>1 beds)Alcohol/drug outpatientAngioplasty (1989–2000)Birthing rooma

27.732.038.873.4

34.041.834.171.5

37.426.648.070.3

Burn treatment (>0 beds)Computed tomography (CT) scanCardiac catheterization labChild psychiatry (>0 beds)a

3.792.653.225.6

7.590.348.828.4

4.692.260.224.0

Diagnostic radioisotope facilityExtracorporeal shock-wave lithotripterEmergency roomEmergency rooma

83.016.996.696.6

76.013.194.695.6

81.221.795.395.5

FitnessHome healthHome health (1988–1993)Home health (1995–1997)Home health (1998–2000)

25.246.842.152.549.5

21.445.939.151.853.9

20.833.020.559.526.2

Pediatric intensive care (>1 beds)a

Magnetic resonance imaging (MRI)Neonatal intensive care (>0 beds)Neonatal intensive care (>0 beds)a

18.648.433.133.1

25.340.933.338.8

31.150.745.445.4

Obstetrics bed (>2 beds)Obstetrics bed (>2 beds)a

Obstetrics births (≥100 births)Obstetrics births (≥100 births)a

75.676.274.374.8

64.976.563.075.6

73.474.569.470.5

Open-heart surgeryOrthopedic surgery (1989–1993)Positron emission tomography scan (1990–2000)Psychiatric inpatient (1989–2000, >1 beds)

33.693.9

5.646.6

27.988.4

5.258.5

40.992.6

7.251.1

Psychiatric emergency servicesPsychiatric emergency servicesa

Skilled nursing (>2 beds)Single photon emission CT (1990–2000)Sports medicine

47.546.935.046.433.6

55.949.537.740.925.7

40.840.435.041.833.4

TraumaTraumaa

UltrasoundWomen’s centerWomen’s centera

25.725.596.246.746.5

23.930.795.249.042.9

21.821.794.452.051.8

SOURCE: Author’s analysis of American Hospital Association Annual Survey data, 1988–2000.

NOTES: Probit predicted probabilities control for hospital, demographic, and geographic characteristics. Includes all generalmedical and surgical, nonrural hospitals in metropolitan statistical areas with more than one hospital.a Excluding Veterans Affairs (VA) hospitals.



NOTES1. E.C. Norton and D. Staiger, “How Hospital Own-

ership Affects Access to Care for the Uninsured,”RAND Journal of Economics 25, no. 1 (1994): 171–185.

2. F. Sloan, “Not-for-Profit Ownership and Hospi-tal Behavior,” in Handbook of Health Economics, vol. 1,ed. A.J. Culyer and J.P. Newhouse (Amsterdam:Elsevier Science B.V., 2000), 1141–1174. For evi-dence on differences in financial behavior, see E.M. Silverman, J.S. Skinner, and E.S. Fisher, “TheAssociation between For-Profit Hospital Own-ership and Increased Medicare Spending,” NewEngland Journal of Medicine 341, no. 6 (1999): 420–426; and R. Frank and D. Salkever, “MarketForces, Diversification of Activity, and the Mis-sion of Not-for-Profit Hospitals,” in The ChangingHospital Industry: Comparing Not-for-Profit and For-Profit Institutions, ed. D. Cutler (Chicago: Univer-sity of Chicago Press, 2000), 195–215. See alsoR.G. Frank and D.S. Salkever, “Nonprofit Organi-zation in the Health Sector,” Journal of EconomicPerspectives 8, no. 4 (1994): 129–144; M. Gaynorand D. Haas-Wilson, “Change, Consolidation,and Competition in Health Care Markets,” Jour-nal of Economic Perspectives 13, no. 1 (1999): 141–164;M.G. Duggan, “Hospital Ownership and PublicMedical Spending,” Quarterly Journal of Economics115, no. 4 (2000): 1343–1373; F.A. Sloan, “Com-mercialism in Nonprofit Hospitals,” Journal ofPolicy Analysis and Management 17, no. 2 (1998): 234–252; and G.J. Young, K. Desai, and C.V. Lucas,“Does the Sale of Nonprofit Hospitals ThreatenHealth Care for the Poor?” Health Affairs 16, no. 1(1997): 137–141.

3. D.C. Naylor, “Your Money and/or Your Life?” Ca-nadian Medical Association Journal 116, no. 11 (2002):1416–1417.

4. F.A. Sloan et al., “Hospital Ownership and Costand Quality of Care: Is There a Dime’s Worth ofDifference?” Journal of Health Economics 20, no. 1(2001): 1–21; P.J. Devereaux et al., “A SystematicReview and Meta-Analysis of Studies Compar-ing Mortality Rates of Private For-Profit and Pri-vate Not-for-Profit Hospitals,” Canadian MedicalAssociation Journal 166, no. 11 (2002): 1399–1406;and M. McClellan and D. Staiger, “ComparingHospital Quality at For-Profit and Not-for-ProfitHospitals,” in The Changing Hospital Industry, 93–112.

5. Medicare Payment Advisory Commission, Hospi-tal Inpatient Services Payment System (Washington:MedPAC, 13 July 2004).

6. A detailed research note on the relative profit-ability of all services in the study, including acomplete list of sources, is available online atcontent.healthaffairs.org/cgi/content/full/24/3/790/DC1.

7. R. Winslow, “Missing a Beat: How a Break-

through in Cardiac Broke Down for J&J,” WallStreet Journal, 18 September 1998; H. Stout, “AHealthy Bottom Line,” Business Journal, 11 May2001; and M. Wagner, “Cardiac AdministrationTaking Off at Hospitals,” Modern Healthcare (1 July1991): 28.

8. D.M. Cutler, M. McClellan, and J.P. Newhouse,“How Does Managed Care Do It?” RAND Journalof Economics 31, no. 3 (2000): 526–548.

9. D.M. Cutler et al., “Pricing Heart Attack Treat-ments,” in Medical Care Output and Productivity, ed.D.M. Cutler and E.R. Berndt (Chicago and Lon-don: University of Chicago Press, 2001), 305–362.

10. D.M. Cutler and R.S. Huckman, “TechnologicalDevelopment and Medical Productivity: The Dif-fusion of Angioplasty in New York State,” Journalof Health Economics 22, no. 2 (2003): 187–217.

11. J. Cromwell, D.A. Dayhoff, and A.H. Thoumaian,“Cost Savings and Physician Responses to GlobalBundled Payments for Medicare Heart BypassSurgery,” Health Care Financing Review 19, no. 1(1997): 41–57.

12. P.B. Ginsburg, “Are Focused Factories the Waveof the Future?” Frontiers of Health Services Manage-ment 16, no. 3 (2000): 41–46; K.J. Devers, L.R.Brewster, and P.B. Ginsburg, Specialty Hospitals: Fo-cused Factories or Cream Skimmers? (Washington:Center for Studying Health System Change,April 2003); H. Meyer, “Focused Factories: AreYou Ready for the Competition?” Hospitals andHealth Networks 72, no. 7 (1998): 24–26, 28–30; M.Gallagher, “Bitter Medicine: Presbyterian PlansPainful for Doctor,” Albuquerque Journal, 10 March1998; R. Winslow, “Coronary Bypass: Fed-UpCardiologists Invest in Own Hospital Just forHeart Disease—They’ll Regain Autonomy, butCritics See a Grab for Most-Profitable Care—AShowdown in Albuquerque,” Wall Street Journal,22 June 1999; S. Dang, “Ground Broken for HeartHospital,” Milwaukee Journal Sentinel, 16 August2002; and M. Romano, “The Week in Health-care,” Modern Healthcare (29 March 2004): 10–11.

13. Regarding ED costs, see, for example, Deloitteand Touche, U.S. Hospitals and the Future of HealthCare (Boston: Deloitte and Touche, June 1990);W.M. Gentry and J.R. Penrod, “The Tax Benefitsof Not-for-Profit Hospitals,” in The Changing Hospi-tal Industry, 285–324; and G.A. Melnick, C.A.Serrato, and J.M. Mann, “Prospective Paymentsto Hospitals: Should Emergency AdmissionsHave Higher Rates?” Health Care Financing Review10, no. 3 (1989): 29–39. Regarding psychiatriccare reimbursement, see A. Woodward et al.,“The Drug Abuse Treatment Gap: Recent Esti-mates,” Health Care Financing Review 18, no. 3(1997): 5–17. Regarding populations attracted,see, for example, D.M. Dhossche and S.O. Ghani,“A Study on Recidivism in the Psychiatric Emer-

8 0 0 M a y / J u n e 2 0 0 5




gency Room,” Annals of Clinical Psychiatry 10, no. 2(1998): 59–67; H. Shwed, “Teaching EmergencyRoom Psychiatry,” Hospital and Community Psychia-try 31, no. 8 (1980): 558–562; and L. Tye, “Beth Is-rael to Keep Psychiatric Unit,” Boston Globe, 2February 2001.

14. Duggan v. Bowen, 691 F. Supp. 1487 (1988).

15. J.P. Newhouse, Pricing the Priceless: A Health Care Co-nundrum (Cambridge, Mass.: MIT Press, 2002).

16. J.P. Newhouse, “Medicare,” in American EconomicPolicy in the 1990s, ed. J.A. Frankel and P.R. Orszag(Cambridge, Mass.: MIT Press, 2002), 899–955.

17. K. Liu et al., Medicare’s Post-Acute Benefit: Background,Trends, and Issues to Be Faced (Washington: UrbanInstitute, 1999).

18. C. Helbing and E.S. Cornelius, “SNFs,” Health CareFinancing Review 14 Supp. (1992): 97–123; C.Helbing, J.A. Sangl, and H.A. Silverman, “HomeHealth Agency Benefits,” Health Care Financing Re-view 14 Supp. (1992): 125–148; N.J. Scharmach,“Diversifying into Skilled Nursing Care: It CanFill Beds, Manage Medicare Costs, and Meet aNeed,” Modern Healthcare (30 April 1990): 30; andL. Wagner, “Hospitals Seeing Benefit in OfferingLong-Term Care,” Modern Healthcare (24 March1989): 40–42.

19. C.R. Gresenz, J. Rogowski, and J.J. Escarce, “Up-dated Variable-Radius Measures of HospitalCompetition,” Health Services Research 39, no. 2(2004): 417–430.

20. J.E. Wennberg, “Understanding GeographicVariations in Health Care Delivery,” New EnglandJournal of Medicine 340, no. 1 (1999): 52–53.

21. D.B. Rubin, “Estimating Causal Effects fromLarge Data Sets using Propensity Scores,” Annalsof Internal Medicine 127, no. 8, Part 2 (1997): 757–763; P.R. Rosenbaum and D.B. Rubin, “The Cen-tral Role of the Propensity Score in Observa-tional Studies for Causal Effects,” Biometrika 70,no. 1 (1983): 41–55; and P.R. Rosenbaum and D.B.Rubin, “Reducing Bias in Observational Studiesusing Subclassification on the Propensity Score,”Journal of the American Statistical Association 79, no.387 (1984): 516–524.

22. For review of these arguments, see J.R. Horwitz,“Why We Need the Independent Sector: The Be-havior, Law, and Ethics of Not-for-Profit Hospi-tals,” UCLA Law Review 50, no. 6 (2003): 1345–1411.

23. BNA’s Health Law Reporter, NonProfit HospitalCharity Care Litigation 13, no. 43 (2004): 1555–1575.

24. Senate Finance Committee, “Staff DiscussionDraft,” Senate Finance Committee Hearing—Charity Oversight and Reform: Keeping BadThings from Happening to Good Charities (22June 2004).





Documents

Health Aff 2005 Horwitz 790 801