This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

H E A L T H  T R A C K I N G :
M A R K E T W A T C H 
 
25 October 2005  
  

Effects Of Physician-Owned 
Limited-Service Hospitals: 
Evidence From Arizona

An examination of cardiac care provided in single-specialty
hospitals and community hospitals in two Arizona cities.


by Jean M. Mitchell


ABSTRACT:

In recent years physician ownership of so-called limited-service hospitals has become commonplace in many states lacking certificate-of-need regulations. Empirical evidence documenting the effects of these facilities is sparse. This study compares practice patterns of physician-owners of limited-service cardiac hospitals and physician-nonowners who treat cardiac patients at competing full-service community hospitals. Analyses of six years of Arizona inpatient discharge data show that physician-owners treat higher volumes of profitable cardiac surgical diagnosis-related groups (DRGs), higher percentages of low-severity cases, and higher percentages of cases with generous insurance compared with physician-nonowners who treat cardiac patients in community hospitals.

Prohibitions on physician self-referral were enacted during the early 1990s in response to several empirical studies’ findings that the financial incentives inherent in physician self-referral arrangements resulted in increased use of services and higher third-party reimbursements.¹ Federal and most state laws, however, exempt “whole hospitals” and ambulatory surgery centers (ASCs).² The basis for the former was that any referral made by a physician-investor would yield only small financial gains for each physician because hospitals typically provide a wide range of services. Although classified as whole hospitals, physician-owned “limited-service” hospitals are more akin to a specialized hospital department.³

Much of the concern about the recent growth of limited-service hospitals focuses on physician ownership issues.⁴ Hospitals are typically paid a facility fee for each patient treated; physicians bill patients separately for professional services. A referring physician with ownership interest in a limited-service hospital is compensated for professional services but also shares in any profit generated from facility fees. Thus, physician ownership of limited-service hospitals creates financial incentives that could influence physicians’ referral behavior. Second, under the diagnosis-related group (DRG) case-based payment approach, physician-owners of limited-service hospitals could have incentives to treat primarily low-acuity patients within DRGs that are more profitable but send clinically complex cases to full-service community hospitals. A third concern is whether physician-investors refer patients with generous insurance coverage to their own facilities and send those with limited or no coverage to community hospitals. If this is the case, then full-service community hospitals will have limited revenues to subsidize the costs of money-losing services (such as trauma and indigent care).

Proponents counter that physician ownership enables limited-service hospitals to secure high volume through referrals made by physician-investors.⁵ Advocates further contend that such arrangements result in better patient care and outcomes because physician-owners have direct control over the management decisions. Because care in such facilities is organized along product lines or by type of illness, economies of scale could occur and result in lower production costs.

Despite concerns regarding the increasing number of physician-owned limited-service hospitals, empirical evidence to date consists primarily of case studies.⁶ Comparisons of hospital types, however, do not focus on the individual physicians who make the decisions to admit patients. This study addresses this gap in knowledge by comparing practice patterns of physician-owners of limited-service cardiac hospitals and physician-nonowners who treat cardiac patients at competing full-service community hospitals. Much of the literature documenting the effects of physician self-referral arrangements compares physician-owners with nonowners.⁷

Study Data And Methods

The data for this study come from two sources: (1) inpatient discharge data from Arizona hospitals, spanning the years 1998–2003, obtained from the Arizona Department of Health Services, and (2) physician directory information obtained from the Arizona Medical Board. The discharge database contains detailed information on each patient discharge, including the name and state license number of the attending or operating physician responsible. For surgical DRGs, the identified physician is the person who performed the procedure, whereas for medical DRGs, the identified physician is the person who monitored the patient in the hospital. The Arizona Medical Board maintains a database that contains detailed information (such as specialty and medical school graduation date) on all physicians (active and retired) and residents practicing in the state.

With assistance from the Arizona Department of Health Services, the study team identified two physician-owned limited-service hospitals that specialize in the delivery of cardiac care services. The Tucson Heart Hospital began treating patients in October 1997. The Arizona Heart Hospital in Phoenix became operational in June 1998. We also identified four full-service community hospitals with substantial cardiac care programs that were operational before the Tucson Heart Hospital entered the market and that submitted valid inpatient discharge data to the state during 1997–2003. In the Phoenix market area, we identified seven hospitals that had substantial cardiac care programs before the Arizona Heart Hospital entered the market and two competing facilities that entered later.

Construction of physician-level analytical file. For each year (1998–2003) we selected all inpatient discharges with either a cardiac surgical or medical DRG code that were treated at either of the two heart hospitals or one of the competing full-service community hospitals in either city.⁸ We identified 236,590 inpatient discharges that met these criteria. Our first series of exclusions resulted in a sample of 215,435 cases (91 percent of the original sample). This sample included 33,060 cardiac cases treated at one of the heart hospitals and 182,375 cardiac DRG cases treated at one of the competing hospitals.⁹

The next step was to aggregate across cardiac DRG cases by physician license number and hospital provider number to construct an analytical file in which the physician-year is the unit of observation and the volume counts for each physician reflect the number of DRG cardiac cases (total, surgical, and medical) treated at each hospital where the physician practices. For physicians who treated cases at both a heart hospital and the community hospitals, we calculated two indicators of supply: the annual volume of cardiac DRG cases treated at the heart hospital, and the number of cardiac DRG cases treated in competing hospitals. For physicians who only treated patients in the community hospitals, we summed across facilities to calculate the volume of inpatient cardiac DRG cases treated at community hospitals in the market area.

Defining physician-owners and nonowners. Because physician ownership of health care facilities is not reported to the state, it was necessary to establish a set of criteria to define “owners” and “nonowners.” Considerable evidence indicates that ownership of limited-service hospitals is offered only to physicians who can both refer and treat patients at the facility.¹⁰ We used this information to define “owners” and “nonowners.” A physician-year observation is included in the sample of owners if (1) the physician treated at least six cardiac DRG cases in a given year across all hospitals in the market area, and (2) the physician treated at least 10 percent of his or her cardiac DRG cases at the heart hospital. Recognizing that this definition of owners could include some physicians who are nonowners, we conducted sensitivity analyses to evaluate the robustness of the results to more stringent definitions of ownership. Specifically, we increased the share of cardiac DRG cases that a physician-owner treated in a given year at the heart hospital to 20 percent, 30 percent, 40 percent, and 50 percent and then used higher minimum caseload volumes (11 and 21). We then replicated the statistical analyses using these more stringent definitions.

A physician was classified as a nonowner if he or she treated at least six cardiac DRG cases in one or more of the community hospitals and no such cases at the heart hospital in a given year. The volume counts of inpatient cardiac DRG cases that each physician-owner treated in competing full-service community hospitals were excluded from the primary comparisons but were analyzed later; 17,424 discharges (7.36 percent of the original sample of 236,590) met this criterion. The rationale for this exclusion was to create a pure control group comprising physicians who only treated patients at competing full-service community hospitals. Physicians who treated low volumes resulted in the exclusion of another 10 percent of the original sample.¹¹ The final sample used in constructing the indicators of physician supply contained 174,133 discharges (73.6 percent of the original sample); 32,032 cases treated at the heart hospitals; and 142,101 cases treated at the community hospitals.

Indicators of case-mix. We used the 3M APR-DRG software to assign a severity of illness class (minor, moderate, major) to each cardiac DRG case. Based on these results, we constructed a series of mutually exhaustive indicator (0–1) variables to identify the severity of each case. Next we summed the severity indicator variables to create, for each physician-year observation, a count of the number of cases treated in each severity class. Finally, we calculated the percentage of cardiac DRG cases in each severity-of-illness class (minor, moderate, major) treated by each physician in a given year.

More complicated case-mix can also be measured by identifying the presence of multiple comorbid conditions. We used the comorbidity software developed by Anne Elixhauser and colleagues to construct a series of comorbidity variables from secondary diagnosis codes reported on each discharge.¹² Based on the results, we assigned a count of the number of comorbid conditions to each case. Next we constructed a series of mutually exhaustive indicator variables to identify whether each case has zero, one, two, three, four, five, or six-plus comorbid conditions. Following the approach outlined for severity-of-illness class, we summed the comorbidity indicator variables across each physician in a given year to calculate the number of cases treated in each comorbidity category. The last step was to calculate the percentage of cases in each comorbidity category treated by each physician in a given year.

Indicators of payer mix. For each physician-year observation, we constructed several indicators of payer mix. These measure the percentage of cardiac DRG cases treated by each physician in a given year, by type of insurance coverage. For each physician, we calculated the percentages of cardiac DRG cases treated each year with the following types of insurance coverage: Medicare fee-for-service (FFS); commercial indemnity or preferred provider organization (PPO); Medicare health maintenance organization (HMO); Medicaid HMO; commercial HMO; and AHCCCS Health Care Group (a state-sponsored HMO available to self-employed people and small businesses). Arizona providers regard Medicare FFS and commercial indemnity/PPO plans as “generous” insurance coverage because each reimburses on either a FFS or discounted FFS basis. Conversely, providers consider Medicare HMO, Medicaid HMO, commercial HMO, and the state-sponsored AHCCCS plans as less generous because each pays lower rates than either Medicare FFS or commercial PPO/indemnity plans.¹³

Analyses comparing physician-owners and nonowners. Assuming that patients are randomly distributed, there should be no difference in the volume of cases treated, case-mix, and payer mix between owners and nonowners. The null hypothesis assumes that the financial incentives linked to physician ownership do not affect referral behavior. We first compared the annual volume of inpatient cardiac DRG cases (total, surgical, and medical) treated by owners versus nonowners using a two-tailed test for differences between the means, with a null hypothesis of no difference. We next evaluated the effects of physician ownership on case-mix by comparing the mean percentage of cardiac DRG cases within each severity-of-illness class treated by owners relative to nonowners. We also conducted t-tests to compare the mean percentages of cases treated within each comorbidity category by physician ownership status. Finally, we performed t-tests to compare the mean percentage of cardiac DRG cases with each type of insurance coverage treated by owners relative to nonowners, again to test the null hypothesis of no difference. We replicated these four sets of analyses controlling for market area and year.

Study Results

Sample sizes. The sample of physician-owners used in the volume and payer-mix analyses includes 426 physician-year observations (210 from Phoenix and 216 from Tucson). The sample of physician-nonowners used in the volume and payer-mix comparisons comprises 3,197 physician-year observations (2,164 from Phoenix and 1,033 from Tucson). The analyses comparing severity-of-illness class and comorbidity counts controlling for DRG type (surgical versus medical) are based on the subsamples of owners and nonowners who treated one or more surgical (medical) cases. The analyses of case-mix for surgical cases are based on a sample of 388 owner and 2,762 nonowner physician-year observations. The corresponding set of analyses for cardiac medical cases is based on a sample of 395 owner and 3,101 nonowner physician-year observations.

Physician supply comparisons. Physician-owners in Phoenix treated nearly twice as many cardiac DRG cases as nonowners in the study period; the annual volume was eighty-nine for owners in Phoenix compared with forty-five for nonowners (p < .01) (Exhibit 1). This difference in total cardiac DRG arises because owners treated close to 3.8 times as many surgical DRG cases as nonowners. The mean was seventy-two for owners compared with nineteen for nonowners (p < .01). The reverse pattern emerges for cardiac medical DRGs. Physician-owners in Tucson treated 47 percent more cardiac DRG cases per year than nonowners; the mean was 62 for owners compared with 42 for nonowners (p < .01). Comparisons of cardiac surgical DRG caseloads reveal that physician-owners treated more than double the number of surgical cases that nonowners treated (Exhibit 1). The average was 26.5 for owners and 12.2 for nonowners (p < .01). Results controlling for year and market area are similar to the pooled findings.¹⁴

Severity-of-illness comparisons. Exhibit 2 compares the severity-of-illness classes assigned to cardiac surgical DRG cases treated by owners and nonowners, stratified by market area. Physician-owners treated proportionately more surgical cases classified as “minor” in comparison with nonowners. The reverse holds true for surgical cases assigned either a “moderate” or “major” severity class. Although not depicted graphically, similar patterns are evident for cardiac medical DRG cases classified as either “minor” or “moderate” (p < .01). Also, the results comparing severity by ownership status controlling for both market area and year are similar.

Comorbidity comparisons. Exhibit 3 shows a comparison of comorbidity counts for cardiac surgical DRGs, controlling for physician ownership status. Physician-owners treated significantly higher percentages of surgical DRG cases with one or two comorbid conditions. For example, surgical DRG cases with one comorbidity account for nearly 21 percent of the cases treated by owners compared with about 10 percent of similar cases treated by nonowners (p < .01). The reverse pattern characterizes surgical DRG cases with four, five, or six-plus comorbid conditions. It is interesting to note that the distribution of surgical cases classified by comorbidity counts is normally distributed for nonowners. In contrast, the distribution of cases treated by owners is highly skewed toward low-severity cases. The results controlling for both market area and year are nearly identical.

Although not displayed graphically, examination of the comorbidity status of cardiac medical DRGs reveals a similar pattern. Compared to physician-nonowners, physician-owners treated significantly higher percentages of cardiac medical DRG cases with zero, one, or two comorbidities (p < .01). Physician-owners treated significantly lower percentages of medical DRG cases with three or more comorbidities than did nonowners (p < .01). Analogous comorbidity comparisons stratified by market area and year mirror the results for both market areas combined.

Payer-mix comparisons. Exhibit 4 shows the mean percentage of cardiac DRG cases treated under specific types of insurance coverage by physician ownership status. Physician-owners treated higher percentages of patients with generous insurance coverage (Medicare FFS and commercial PPO) but lower percentages of cases enrolled in HMO-type plans (p < .01). A similar comparison of payer mix controlling for market area reveals that two-thirds of the cardiac DRG cases treated by owners in Phoenix had Medicare FFS coverage, compared with 48 percent of such cases in Tucson. In contrast, Medicare FFS patients accounted for similar percentages of the cardiac DRG caseloads of nonowners: 36 percent in Phoenix and 39 percent in Tucson. On the other hand, commercially insured patients in Tucson accounted for 24 percent of cardiac cases treated by owners, compared with 4.5 percent of such cases treated by nonowners (p < .01). In Phoenix, nearly 20 percent of cardiac cases treated by physician-owners and 15 percent of such cases treated by nonowners were commercially insured (p < .01).

Discussion And Policy Implications

This study reports empirical evidence comparing the practice patterns of physician-owners of limited-service cardiac hospitals and nonowners who treat cardiac patients at competing full-service community hospitals. The analyses suggest that physician self-referral arrangements have major effects on physician practice patterns.

First, physician-owners of limited-service cardiac hospitals treated significantly higher volumes of the profitable cardiac surgical DRG cases than did nonowners treating patients at competing full-service community hospitals. Second, owners treated a less severe case-mix of both cardiac surgical and medical DRGs relative to nonowners. Irrespective of whether one measures case-mix by severity-of-illness class or a count of comorbid conditions, the results consistently show that the caseloads of physician-owners are low-acuity patients. In contrast, the case-mix of patients treated by nonowners is normally distributed. Finally, comparisons of payer mix reveal that physician-owners treated significantly higher percentages of patients with generous insurance (Medicare FFS and commercial indemnity/ PPO) but significantly lower percentages of patients enrolled in HMO-type plans.¹⁵ Importantly, significant effects of physician ownership are evident even after other confounding factors (specialty, experience, market area, and year) are controlled for. Moreover, these findings corroborate previous research examining the effects of physician self-referral.¹⁶

Study limitations. Although the findings reported here provide new evidence on the effects of physician ownership of limited-service cardiac hospitals, the study has some limitations. First, because Arizona does not require physicians to disclose ownership interests in limited-service hospitals, it was necessary to establish criteria to identify physician-owners and physician-nonowners. We recognized that the sample of owners might include some nonowner observations and thus tested the sensitivity of the results to alternative definitions of ownership. Although the analyses based on more stringent definitions reduced the number of physicians who qualified as owners, the differences in volume, case-mix, and payer mix by ownership were larger than those based on the 10 percent share criterion. Only 10 percent of the 426 physician-year observations were eliminated from the analyses when we increased the share treated by a physician-owner at the heart hospital from 10 percent to 20 percent. Furthermore, almost three-quarters of the 426 physician-year observations in the owner sample treated 50 percent or more of their cases at the heart hospital. Based on the sensitivity analyses, we contend the 10 percent threshold is conservative and thus biases downward the differences that exist between owners and nonowners.

A second limitation is that analyses reflect inpatient cardiac DRG cases treated. No data are available on outpatient surgeries and ancillary services rendered to cardiac patients at each hospital. However, the GAO report found that limited-service cardiac hospitals derived about 85 percent of their revenues from inpatient cases.¹⁷

Third, the analyses were based on limited-service cardiac hospitals located in two market areas in Arizona. The findings, therefore, might not be applicable to other types of physician-owned limited-service hospitals—in particular, those that specialize in the provision of orthopedic, spinal, or general surgical procedures. On the other hand, the findings are probably applicable to cardiac limited-service hospitals in other states. Both cardiac hospitals in Arizona are joint ventures between referring physician investors and a for-profit corporation that has established similar facilities in other states, including California, Arkansas, South Dakota, Ohio, and Texas.

A fourth limitation is that the findings provide no insights as to the benefits that could arise from high-volume specialization: namely, lower production costs and improved outcomes. A report by the Medicare Payment Advisory Commission (MedPAC), however, found that specialty hospitals do not have lower costs for Medicare patients than community hospitals, even though lengths-of-stay are shorter at specialty hospitals.¹⁸ Moreover, there is no evidence indicating that limited-service hospitals have better outcomes. Peter Cram and colleagues found that differences between specialty and general hospitals in mortality rates after cardiac revascularization were not significant after differences in patient characteristics and procedural volume were adjusted for.¹⁹ The Centers for Medicare and Medicaid Services (CMS) also analyzed specific mortality rates for four procedures and two conditions performed in heart hospitals and competing facilities. It found that both types of facilities performed better than expected, given the hospitals’ case-mix.²⁰

A final limitation is that because source of payment is not accurately reported for people without insurance, it was impossible to ascertain if physician-owned limited-service hospitals treat uninsured patients.

Other comparisons. We also compared cases that physician-owners treated in the heart hospitals versus the community hospitals, but for at least two reasons, we contend that such comparisons are less informative than comparisons of physician-owners and nonowners. First, a large percentage of owners did not treat any cases at the community hospitals. In Phoenix, almost 62 percent of physician-owner observations treated no patients at the community hospitals; 83 percent of such observations treated 80 percent or more of their cases at the heart hospital. In Tucson, more than 20 percent of the physician-owner observations treated no cases at the community hospitals. Since physician-owner observations who did not treat any patients at the community hospitals would be excluded from the analysis, such exclusions raise concerns about selection bias. Second, for those owners who treated patients at both types of facilities, a large percentage treated so few cases that these comparisons are hampered by selection bias. Physician-owners in Phoenix treated a mean of 3.5 surgical and 3.2 medical cases per year in community hospitals. Further analyses of the cases they treated in community hospitals show that the payer mix is similar to that of cases treated by physician-nonowners. Irrespective of whether physicians are owners or nonowners, the severity of the caseloads they treated at community hospitals is normally distributed. In contrast, the cases physician-owners treated in the heart hospital are highly skewed toward low-severity cases. These comparisons suggest that patient characteristics, not physician reputation, determine where physician-owners treat each patient.

Policy implications. The Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003 imposed an eighteen-month moratorium on physician self-referral of Medicare and Medicaid patients to new limited-service hospitals. The moratorium expired 8 June 2005.²¹ By administrative rule, the CMS extended the moratorium until January 2006 while it reviews whether physician-owned specialty facilities meet the definition of a whole hospital.²² In May 2005 Senators Charles Grassley (R-IA) and Max Baucus (D-MT) introduced legislation known as the Hospital Fair Competition Act of 2005, which would close the loophole and prohibit physicians from making referrals to limited-service hospitals in which they have an investment interest. The findings reported here provide evidence in support of these legislative actions. Nevertheless, if such a prohibition is to be effective, it should require that physician-owners divest their investments in existing limited-service hospitals within eighteen to twenty-four months.

The Grassley bill also requires that DRG weights be calculated at the hospital level rather than nationally, and that these weights be recalibrated on the basis of costs at least once every five years. Tying payment to patient severity within a given DRG class (that is, risk-adjusting) should help mitigate the financial incentives to treat low-acuity patients with specific DRG categories. Nonetheless, even if the DRG payment were risk-adjusted, physicians could compensate for the lower payment by recommending that the patient undergo more outpatient procedures and ancillary tests that are not covered by the DRG payment. Evidence documenting physicians’ responses to fee changes implemented under the Medicare fee schedule shows that cutting fees for profitable surgical procedures has spillover effects and causes physicians to increase the supply of services whose payments have not been reduced.²³ A prohibition on self-referral to the limited-service hospitals in conjunction with the payment adjustment should help mitigate undesirable spillover effects. It is important to recognize that the Grassley bill does not apply to nonelderly patients with private insurance coverage. Thus, the findings reported here should be of interest to state policymakers as well as insurance companies who are concerned about the conflict of interest associated with physicians’ self-referral arrangements.

The author thanks Greg Lostoski for his excellent programming expertise and Cynthia Schuster for invaluable research assistance. Research support was provided through an unrestricted educational research contract between the Sioux Valley Hospitals and Health System and Georgetown University. Additional support was provided through institutional research grants from Georgetown University.

NOTES

1. For a comprehensive review of the physician self-referral literature, see J.M. Mitchell, “Physician Joint Ventures and Self-Referral: An Empirical Perspective,” in Conflicts of Interest in Clinical Practice and Research, ed. R.G. Spece, D.S. Shimm, and A. Buchanan (New York: Oxford University Press, 1996), 299–317.
2. U.S. Government Accountability Office, Specialty Hospitals: Geographic Location, Services Provided, and Financial Performance, Pub. no. GAO-04-167 (Washington: GAO, October 2003).
3. Medicare Payment Advisory Commission, Report to the Congress: Physician-Owned Specialty Hospitals (Washington: MedPAC, March 2005); and J.K. Iglehart, “The Emergence of Physician-Owned Specialty Hospitals,” New England Journal of Medicine 352, no. 1 (2005): 78–83.
4. R. Abelson, “Hospitals Battle For-Profit Groups for Patients,” New York Times, 30 October 2002; R. Abelson, “Generous Medicare Payments Spur Specialty Hospital Boom,” New York Times, 26 October 2003; R. Abelson, “The Shifting Burden of Emergency Care,” New York Times, 3 February 2004; C.N. Kahn and J. Cohen, “Adverse Effects of Physician-Owned Limited Service Facilities: Healthy Competition Depends on Level Playing Field” (Paper presented at Conference on Specialty Hospitals, Ambulatory Surgery Centers, and General Hospitals: Charting a Wise Public Policy, Washington, D.C., September 2004); and American Hospital Association, “Impact of Limited-Service Providers on Communities and Full-Service Hospitals,” Trendwatch 6, no. 2 (2004): 1–8.
5. MedPAC, Report to the Congress; Iglehart, “The Emergence of Physician-Owned Specialty Hospitals”; and GAO, Specialty Hospitals.
6. L.P. Casalino, K.J. Devers, and L.R. Brewster, “Focused Factories? Physician-Owned Specialty Facilities,” Health Affairs 22, no. 6 (2003): 56–67.
7. Mitchell, “Physician Joint Ventures.”
8. Cardiac surgical and medical DRGs are listed in Supplemental Exhibit 1; see
content.healthaffairs.org/cgi/content/full/hlthaff.w5.481/DC2.
9. Supplemental Exhibit 2 provides details; ibid.
10. R.E. Tibbs et al., “Physician Ownership of Specialty Spine Hospitals,” Neurosurgery Focus 12, no. 4 (2002): 1–3; J. Goolsby, “The Evolution of the Heart Hospital in Austin: Why We Did It and What Happened,” American Heart Hospital Journal 1, no. 1 (2003): 97–103; MedPAC, Report to the Congress; and Iglehart, “The Emergence of Physician-Owned Specialty Hospitals.”
11. See Note 9.
12. A. Elixhauser et al., “Comorbidity Measures for Use with Administrative Data,” Medical Care 36, no. 1 (1998): 8–27.
13. We based this classification on extensive discussions with staff at the Arizona Department of Health Services and people employed by community hospitals with cardiac care programs, who are responsible for negotiating contracts with the insurers in either the Phoenix or Tucson markets.
14. Physician-owners in Phoenix seem to differ from owners in Tucson. One explanation is that owners in Tucson could have academic affiliations with University Medical Center, which has training programs in both cardiology and vascular surgery. There are no such training programs in the Phoenix market. Another contributing factor could be differences in ownership structure. The Tucson Heart Hospital is a joint venture among referring physicians, a community hospital, and a for-profit corporation, whereas the Arizona Heart Hospital does not involve a community hospital partner.
15. Several sources in Arizona confirmed that the heart hospitals opted not to contract with some of the HMOs because they pay lower rates than Medicare FFS and the commercial PPO plans.
16. Mitchell, “Physician Joint Ventures”; and MedPAC, Report to the Congress.
17. GAO, Specialty Hospitals.
18. MedPAC, Report to the Congress.
19. P. Cram, G.E. Rosenthal, and M.S. Vaughan-Sarrazin, “Cardiac Revascularization in Specialty and General Hospitals,” New England Journal of Medicine 352, no. 14 (2005): 1454–1462.
20. M.O. Leavitt, Study of Physician-Owned Specialty Hospitals Required in Section 507(c)(2) of the Medicare Prescription Drug, Improvement, and Modernization Act of 2003, May 2005,
www.cms.hhs.gov/media/press/files/052005/RTC-StudyofPhysOwnedSpecHosp.pdf (4 October 2005).
21. GAO, Specialty Hospital Moratorium, Pub. no. GAO-05-647R (Washington: GAO, May 2005).
22. M.B. McClellan, “Specialty Hospitals: Assessing Their Role in the Delivery of Quality Health Care,” Testimony before the House Energy and Commerce Committee, 12 May 2005.
23. J.M. Mitchell, J. Hadley, and D.J. Gaskin, “Spillover Effects of Medicare Fee Reductions: Evidence from Ophthalmology,” International Journal of Health Care Finance and Economics 2, no. 3 (2002): 171–188; and M.E. Miller, “MedPAC Recommendations on Imaging Services,” Testimony before the House Ways and Means Subcommittee on Health, 17 March 2005.

Read related papers by Jack Hadley and Stephen Zuckerman and Allen Dobson and Randall Haught.

Jean Mitchell (mitchejm{at}georgetown.edu) is a professor of public policy in the Georgetown Public Policy Institute, Georgetown University, in Washington, D.C.

DOI: 10.1377/hlthaff.w5.481
©2005 Project HOPE–The People-to-People Health Foundation, Inc.