HMO Penetration, Hospital Competition, and Growth of               Ambulatory Surgery Centers                             ...
nities. They concluded that hospital outpa­        accompanied by increasingly consolidatedtient surgery volume declined a...
for surgeons to steer patients away from                          bypass grafts and $4,865 less for cesarean-community hos...
MeTHODS                                           Dependent variableData Sources                                         T...
by Baker (1997). The number of HMO                     Multivariate Statistical analysisenrollees (all ages combined) were...
Figure 1      Growth of Ambulatory Service Centers (ASCs) in Metropolitan Statistical Areas: 1992-2001                    ...
Table 1                                                                       Trends of Growth of Ambulatory Surgery Cente...
Figure 2Differential Growth of Ambulatory Surgery Centers (ASCs) Corresponding to Differential Changes                    ...
Table 2      Estimated Effects of Market Characteristics on Growth of Ambulatory Surgery Centers (ASCs)        From Least ...
1,150 to over 2,900 in the MSAs studied.          have been sufficient competition amongTwo hypotheses were advanced to ex...
to be directed at understanding the market             Iglehart, J.K.: The Emergence of Physician-Ownedforces at play, and...
U.S. Department of Justice: The Herfindahl-Hirschman Index. Internet address:
Upcoming SlideShare
Loading in …5

Hospital growth and competition


Published on

Published in: Health & Medicine
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Hospital growth and competition

  1. 1. HMO Penetration, Hospital Competition, and Growth of Ambulatory Surgery Centers John Bian, Ph.D. and Michael A. Morrisey, Ph.D. Using metropolitan statistical area (MSA) argued that such facilities pose unfair costpanel data from 1992-2001 constructed advantages to hospitals by drawing profit­from the 2002 Medicare Online Survey able surgeries and procedures away fromCertification and Reporting (OSCAR) hospitals (Winter, 2003). There are alsoSystem, we estimate the market effects of concerns that ASCs may lead to unneces­health maintenance organization (HMO) sary surgeries and procedures because ofpenetration and hospital competition on the the financial incentives inherent in physi­growth of freestanding ambulatory surgery cian-owned ASCs and that their narrowcenters (ASCs). Our regression models with service availability may compromise qual­MSA and year fixed effects suggest that a 10­ ity of care (Casalino, Devers, and Brewster,percentage-point increase in HMO penetra­ 2003; Mitchell and Sass, 2006).tion is associated with a decrease of 3 ASCs While the growth of specialty hospitalsper 1 million population. A decrease from has been contentious with concerns cen­5 to 4 equal-market-shared hospitals in a tering on physician ownership and favor­market is associated with an increase of 2.5 able selection of healthy patients (Mitchell,ASCs per 1 million population. 2006; Guterman, 2006; Stensland and Winter, 2006; Greenwald et al., 2006), theINTRODUCTION growth of ASCs has been much more rapid. Surprisingly, there has been little Freestanding ASCs have been a grow­ empirical research on ASCs and virtuallying phenomenon in the U.S. health care none that has examined the effects of mar­market for the past 20 years. Winter (2003) ket conditions on the growth of ASCs. Theindicated that the number of Medicare- concerns about the growth of ASCs, how­certified freestanding ASCs had increased ever, are much the same. Winter (2003)from about 400 in 1983 to over 3,300 in examined differences in the case mix of2001. These facilities typically consist of Medicare patients receiving ambulatorya small number of operating rooms and surgeries and procedures between hospitalprovide a specific set of surgical proce­ outpatient departments and ASCs, suggest­dures (Casalino, Devers, and Brewster, ing that ASCs might have treated patients2003). Their expanding role in the U.S. with less intense case mix. Two studieshealth care delivery system has been con­ examined the association of ASCs withtroversial. Some have argued that these hospital surgery markets. Casalino et al.“focused factories” lower the unit cost of (2003) reported the perceptions of medicalsurgical care by performing a narrow set group, hospital and health plan executivesof procedures in a remarkably efficient on the effects of ASCs on hospital markets,fashion (Herzlinger, 2004). Others have but were not able to quantitatively sup­The authors are with the University of Alabama at Birmingham. port these perceptions. Lynk and LongleyThe statements expressed are those of the authors and donot necessarily reflect the views or policies of the University (2002) examined the impact of ASC entryof Alabama at Birmingham or the Centers for Medicare &Medicaid Services (CMS). on hospital surgery volume in two commu-HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4 111
  2. 2. nities. They concluded that hospital outpa­ accompanied by increasingly consolidatedtient surgery volume declined as a result hospital markets and emerging hospitalof the new ASC entry and that doctors with systems (Cueller and Gertler, 2003), mayan ownership position in the new ASCs have also encouraged providers to shiftreduced outpatient surgery volume of hos­ deliveries of surgeries and procedurespitals where the doctors had admitting from the inpatient to ambulatory settingsprivileges. Two other studies explained as a method of cost control.the factors influencing the growth of ASCs. ASCs have developed over time, as enti­Casalino and colleagues (2003) suggested ties that are organizationally distinct formthat the absence of State certificate-of-need hospital outpatient surgery departments.(CON) laws and the presence of large Most ASCs are freestanding facilitiessingle-specialty groups in a health care that are not owned outright by were likely factors associated with However, they are required by Medicarethe development of ASCs. A preliminary to be licensed by the States in order toreport by the Medicare Payment Advisory be Medicare-certified providers (CasalinoCommission (2004a), using cross-sectional et al., 2003). Almost all ASCs are for-data, suggested that markets with high­ profit, located in large metropolitan areas,er managed care penetration had slower and equipped with two or more operat­growth of ASCs. ing rooms (Medicare Payment Advisory This study contributes to the existing lit­ Commission, 2004b). Many ASCs special­erature on ASCs. In this analysis, we start ize in one or two types of surgical serviceswith an overview of rapid growth of ASCs such as ophthalmology, gastroenterology,over the past two decades. We then present and orthopedic surgeries or proceduresa model of how changes in health care mar­ (Winter, 2003; Medicare Payment Advisoryket characteristics may affect the growth Commission, 2004b). ASCs are assumedof ASCs. In particular, we focus on two to be a lower cost alternative to hospitalkey market characteristics—managed care outpatient surgery departments possiblypenetration and hospital market concentra­ because of their specialization and lowertion. Finally, we empirically analyze the overhead costs. However, they are paideffects of the two key market characteris­ more generously by Medicare in 8 oftics on the growth of Medicare-certified the 10 surgical procedure categories thatfreestanding ASCs, using a balanced 1992­ account for the highest share of Medicare2001 MSA level panel dataset constructed payments to ASCs (Winter, 2003). Thus,from the 2002 Medicare OSCAR system. differences in Medicare facility payment rates between ASCs and hospital outpatientBaCKgROUND surgery departments may create potential incentives for ASCs to selectively perform Advances in anesthetics and the devel­ certain types of more profitable surgeriesopment of minimally invasive surgical or procedures (Winter, 2003).techniques since the 1980s have made Medicare pays surgeons the same pro­it possible to move many surgeries and fessional surgical services regardless ofprocedures from an inpatient to an ambu­ delivery settings. Surgeons who have anlatory setting (Detmer and Gelijns, 1994). ownership interest in an ASC, however,Contemporaneous changes in health care can earn a share of profit from their invest­market characteristics, particularly the ment in the ACS in addition to their profes­proliferation of managed care activities sional fee. Thus, there is some incentive112 HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4
  3. 3. for surgeons to steer patients away from bypass grafts and $4,865 less for cesarean-community hospital outpatient facilities to sections than did the indemnity plans alsoASCs where they have an ownership inter­ offered to employees.est. Federal laws (Stark I and II) generally If ASCs have a cost advantage over hospi­prohibit physicians from referring their tal outpatient departments because of theirpatients to facilities in which the physi­ quality and efficiency (Casalino, Devers,cians have an ownership (U.S. Department and Brewster, 2003; Herzlinger, 2004), oneof Health and Human Services, 2004). would expect managed care plans to seekHowever, ASCs are exempted from this contracts with them. This demand wouldprohibition (Iglehart, 2005). lead more ASCs to enter the market. Thus, Some State regulations such as CON laws we hypothesize that greater managed careare likely to influence the growth in ASCs penetration will lead to more ASCs in a(Casalino et al., 2003). CON laws require market.that a covered entity obtain approval from A tenant of standard economics is thatthe State before undertaking major capital prices are more likely to be driven downinvestments such as new construction, to marginal costs when there are morerenovation, or expansion into new service competitors in the market. Melnick andlines. Some States do not have CON laws colleagues (1992) were the first to showwhile in others the dollar threshold for that a managed care plan (in California)investments to come under CON reviews was able to negotiate lower hospital pricesis set higher than the amounts needed when there were more hospitals in theto open some types of ASCs. However, local market. Bamezai et al. (1999) showeda preliminary report suggests that CON that managed care penetration had a largerlaws, as currently in place, may only be restraining effect on hospital cost growthweakly associated with the growth of ASCs when there was greater hospital competi­(Medicare Payment Advisory Commission, tion in the market. Their results implied2004a). that a market with four equally sized hos­ pitals would have hospital cost growthCONCePTUal OveRvIew that was 6 percentage points lower than a similar market with only two equal sized While advances in surgical technolo­ hospitals (Morrisey, 2001). More recently,gies undoubtedly have driven the overall the U.S. Government Accountability Officegrowth of ASCs, changing market charac­ (2005) examined the transaction pricesteristics may have much to do with the dif­ paid to hospitals by insurers participatingferential growth of ASCs across markets. in the Federal Employees Health BenefitsTwo factors of particular interest are the Plan. They found that hospitals in the leastgrowth in managed care influence and the competitive quartile of MSAs had pricesconsolidation of hospital markets. that were 18 percent higher than those in The comparative advantage of managed the most competitive quartile of plans is their ability to negotiate lower Other things equal, higher prices shouldprices in exchange for greater service lead to the entry of new competitors. ASCsvolume.1 Altman, Cutler, and Zeckhauser serve as substitutes to hospitals for outpa­(2003) found that HMOs offered to tient surgery. Thus, we hypothesize thatMassachusetts State employees in 1994­ greater hospital concentration will lead to1995 paid $20,808 less for coronary artery more ASCs in a market.1 For a review of hospital markets, refer to Morrisey (2001).HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4 113
  4. 4. MeTHODS Dependent variableData Sources The dependent variable is ASCs per 10,000 population, calculated as the num­ We used four secondary data sourc­ ber of ASCs in an MSA divided by thees for our empirical analysis. The main MSA population and multiplied by 10, source is an extract from the 2002 We were unable to capture two impor­Medicare OSCAR system, which reported tant pieces of information on ASCs thatdata on all Medicare-certified freestand­ were potentially relevant to this study. Theing ASCs in operation at the end of 2001. first is specialties of ASCs. Market effects(Hereafter, ASCs refers to Medicare-certi­ might vary by type of procedure becausefied freestanding ASCs.) Relevant informa­ of their differential profitability implica­tion in the OSCAR includes facility opening tions to ASCs. Thus, analyzing the growthdates as well as State and county location. of ASCs only in aggregate may bias ourHowever, the OSCAR data provide no infor­ results to the null. The second is mergersmation on the volume of services provided and closures information on ASCs. Thereby ASCs. have been ASC mergers and closures. For The three complementary data sources example, during the period from 1997-2002,included an HMO enrollment file that there was an average of 58 ASC mergersreported the number of HMO enrollees and/or closures per year, while 279 newat the county level from 1992-2001, the ASCs were opened per year (MedicareAmerican Hospital Association (AHA) Payment Advisory Commission, 2004a).annual survey of hospitals including infor­ While information on new and terminatedmation such as the number of hospital providers is available in the OSCAR data,admissions from 1992-2001, and the Area complete information on mergers is moreResource Files (ARF), a public use file, that difficult to obtain.compiles county-level information such asthe supply of physicians, population esti­ explanator y variablesmates, and demographic and economiccharacteristics from 1993 and 1995-2003. Two market characteristics are key(We used multiple-year ARFs to construct to our hypotheses about ASC growth.a longitudinal database.) The first was managed care penetration. We defined an MSA as the health care mar­ Managed care includes HMOs, preferredket and identified them based on the 2001 provider organizations, and their deriv­designations by the Office of Management atives. However, the literature suggestsand Budget. We aggregated all county-level that preferred provider organizations havedata to the MSA level and constructed a been much less effective than HMOs in1992-2001 MSA-level balanced panel dataset controlling health care costs (Bamezaiby merging ASC data with HMO penetra­ et al., 1999; Morrisey, Jenen, and Gabel,tion, AHA survey files, and ARF data. (Data 2003). Thus, we focused only on HMOfrom non-MSA counties were excluded from penetration in this study. Publicly availablethe analysis.) In 2001, there were a total of HMO enrollment data are reported by the322 MSAs in the U.S., but the final panel location of the headquarters of the HMOdataset included only 317 MSAs each year and, therefore, are misleading. We usebecause HMO penetration data were not a penetration measure constructed fromavailable in 5 MSAs. HMO enrollment data previously reported114 HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4
  5. 5. by Baker (1997). The number of HMO Multivariate Statistical analysisenrollees (all ages combined) were report­ed once in 1992 and 1993 using data from The main concern about estimating thethe Group Health Association of American market effects on ASCs is that unobserv­(GHAA), twice (July by the Interstudy and able market heterogeneity and secularDecember by GHAA) in 1994, and twice time trends (i.e., unobservable factors cor­(January and July both by the Interstudy) related with both the dependent variablefrom 1995-2001. We calculated HMO pen­ and the two key explanatory variables)etration as a ratio of the total number of could yield biased estimates. With thisHMO enrollees to the total population in concern in mind, we estimate ordinaryeach MSA. For the years during which the least squares regression models with MSAnumber of HMO enrollees were reported and year fixed effects. This two-way fixedtwice, we used the average number of effects specification may mitigate potentialHMO enrollees. biases arising from any time-invariant MSA- The second key market characteristic level covariates (e.g., geographic locationwas community hospital concentration, of ASCs, differential preference or tastes ofmeasured by the Herfindahl-Hirschman surgery delivery setting) as well as impor­Index (HHI) from AHA admissions data tant time trends (e.g., advances in medical(U.S. Department of Justice Web site). The technology, nationwide policy changes). InHHI is defined as the sum of the squared addition, we control for a set of observedadmission market shares of all community time-varying MSA-level demand and sup­hospitals in an MSA. (The value of HHI ply covariates in the regression models.ranges from 0 to 1. The higher value of Our regression model took the followingHHI indicates a more concentrated mar­ form:ket.) A potential limitation of using indi­ ASCit = α + β HMOit + γ HHIit + δ Zit +vidual hospital market share to measure σi + τt + μitcompetition is that it may overstate mar­ where the number of ASCs per 10,000 pop­ket competition by failing to account for ulation in MSA i in year t is a function ofthe rapid development of hospital systems HMO penetration, hospital concentration(Cuellar and Gertler, 2003). (HHI), a vector of other market conditions Other MSA-level covariates, all construct­ (Z) as well as year (τ) and MSA (σ) fixeded from ARF data, included per capita spe­ effects. The Z vector includes measurescialty surgeons (specializing in colon/rectal of the number of specialty surgeons persurgery, general surgery, neurological sur­ 10,000 population, the number of non-gery, obstetrics-gynecology subspecialties, Federal physicians per 10,000 population,ophthalmology, orthopedic surgery, otolar­ the proportion of the MSA population ageyngology, plastic surgery, thoracic surgery, 65 or over, per capita income, and theand urology), per capita total non-Federal unemployment rate. Standard errors werephysicians (i.e., excluding physicians full­ adjusted via Huber robust standard errorstime employed by the Federal Government), correction (White, 1980).the proportion of population age 65 or over, In sensitivity analyses, we ran the modelper capita income, and the unemployment using lagged right hand side variable torate among those age 16 or over. account for ASC delays in adjusting to market conditions. In addition, we ranHealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4 115
  6. 6. Figure 1 Growth of Ambulatory Service Centers (ASCs) in Metropolitan Statistical Areas: 1992-2001 3,000 2,916 2,967 Number of ASCs Number of ASCs in the U.S.1 2,694 2,743 2,455 2,502 2,500 2,239 2,279 2,026 2,064 2,000 1,814 1,846 Number of ASCs 1,652 1,680 1,499 1,521 1,500 1,320 1,339 1,156 1,173 1,000 500 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Year 1 Reported by the Online Survey Certification and Reporting System (OSCAR). SOURCE: 2002 Medicare OSCAR System.the model interacting HMO penetration HMO penetration almost doubled fromand hospital concentration to see if ASCs 11.8 percent in 1992 to 20.1 percent in 1999,depended on the interaction of HMO pen­ but declined slightly after 1999. Hospitaletration and hospital competition. market concentration remained relatively stable until 1996, but began to increaseReSUlTS afterward, reflecting an increase in hos­ pital mergers in the late 1990s (e.g., theTrends and Market Characteristics number of community hospitals used to calculate HHI in this study decreased from Table 1 shows the total number of ASCs 3,037 in 1992 to 2,791 in 2001).in MSAs growth rate from 1,156 in 1992 Figure 2 shows the results of two bivari­to 2,916 in 2001, whereas the correspond­ ate comparisons relating changes in HMOing number of ASCs in MSAs reported by penetration and hospital market concen­the 2002 OSCAR grew from 1,173 to 2,967 tration to the growth in ASCs per 10,000(Figure 1). Thus, our data captured almost population because the MSA fixed effectsall operating ASCs in MSAs in 2001. The approach only uses within-MSA varia­number of ASCs per 10,000 population tion for estimation. For each of the threeincreased at a similar rate from 0.07 to 0.17 variables, we first calculated the absoluteduring the same period. (Twenty of the change within MSAs between 1992 and317 MSAs had no ASC during the study 2001. We then plotted the change in ASCsperiod.) per 10,000 population by the lowest and116 HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4
  7. 7. Table 1 Trends of Growth of Ambulatory Surgery Centers (ASCs) and Market Characteristics at the Metropolitan Statistical Area (MSA)1 Level: 1992-2001 Characteristic 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Total Number of ASCs 1,156 1,320 1,499 1,652 1,814 2,026 2,239 2,455 2,694 2,916 ASCs Per 10,000 Population 0.066 0.074 0.084 0.092 0.108 0.122 0.134 0.147 0.158 0.169 HMO Penetration 0.118 0.127 0.138 0.147 0.165 0.182 0.194 0.201 0.196 0.190 Hospital Concentration (HHI) 0.361 0.363 0.366 0.368 0.378 0.386 0.391 0.394 0.398 0.399 Per Capita Specialty Surgeons2 (Per 10,000 Population) 4.431 4.492 4.446 4.570 4.709 4.811 4.872 4.723 4.781 4.829 Per Capita Non-Federal Physicians (Per 10,000 Population) 23.971 24.612 24.627 26.205 26.867 27.561 28.088 28.723 28.682 28.801 Proportion of Population > 64 Years 0.125 0.126 0.127 0.127 0.127 0.127 0.128 0.128 0.126 0.126 Per Capita Income (in $10,000) 1.781 1.880 1.946 2.039 2.177 2.285 2.362 2.552 2.666 2.734 Unemployment Rate 0.072 0.067 0.061 0.056 0.054 0.050 0.046 0.043 0.042 0.048 1 N=317. 2Surgery specialties include colon/rectal surgery, general surgery, neurological surgery, obstetrics-gynecology subspecialties, ophthalmology, orthopedic surgery, otolaryngology, plastic surgery, thoracic sur­ gery, and urology. NOTES: HMO is health maintenance organization. HHI is Herfindalh-Hirschman Index (higher HHI means less competitive a market). SOURCES: 2002 Medicare OSCAR System, HMO Enrollment File, 1992-2001 American Hospital Association Annual Survey Files, and Area Resource Files.HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4117
  8. 8. Figure 2Differential Growth of Ambulatory Surgery Centers (ASCs) Corresponding to Differential Changes in Market Characteristics: 1992 and 2001 0.16 Lowest Quartile Highest Quartile 0.14 0.131 0.1313209 0.131 0.1307792 People Changes in ASCs per 10,000 people 0.12 0.10 0.080 0.0795391 0.08 0.071 0.0713082 0.06 Changes 0.04 0.02 0.0 Δ HMO Penetration Δ HHI Changes in Market Characteristics NOTES: HMO is health maintenance organization. HHI is Hertindahl-Hirschman Index. SOURCES: 2002 Medicare OSCAR System, HMO Enrollment File, and 1992-2001 American Hospital Annual Survey File.highest quartiles of the change in each mates of both models were similar in mag­of the two market variables. Although the nitude, direction, and significance, we onlytwo bivariate comparisons of changes from focused on the results of the concurrentthe beginning (1992) to the end (2001) model. After controlling for the numbers ofof the study period did not use any of the surgeons and physicians and demograph­information from the years in between, ics and economic characteristics, greaternor did they control for other explanatory HMO penetration was associated with avariables that may be correlated with both reduction in ASCs per 10,000 population (pASCs and the two market characteris­ < 0.01). To put this in context, a 10-percent­tics. However, the figure suggests that the age-point increase in HMO penetrationgrowth of ASCs had a negative association was estimated to result in 3.0 fewer ASCswith increased HMO penetration and a per million population.positive association with increased hospital Greater hospital concentration was asso­market concentration. ciated with greater ASC presence (p < 0.01). An increase in the value of the HHI of 0.05Results of Multivariable Regression is the equivalent of a reduction from 5 to 4analysis equal sized hospitals in a market. The coef­ ficient estimate implies that this increase The estimates from the concurrent and in hospital concentration would result inlag models with MSA and year fixed effects 2.5 more ASCs per million population in aare presented in Table 2. Because the esti­ market. In a model that interacted the HHI118 HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4
  9. 9. Table 2 Estimated Effects of Market Characteristics on Growth of Ambulatory Surgery Centers (ASCs) From Least Squares Regression with Metropolitan Statistical Areas (MSAs) Level and Year Fixed Effects: 1993-2001Dependent Variable1 Concurrent Model Laggged ModelHMO Penetration ***0.3.00 ***-0.278 (0.065) (0.069)Hospital Concentration (HHI) ***0.494 **0.435 (0.156) (0.170)Per Capita Specialty Surgeons (Per 10,000 Population) -0.020 -0.013 (0.012) (0.012)Per Capita Non-Federal Physicians (Per 10,000 Population) 0.003 0.002 (0.004) (0.003)Proportion of Population > 64 Years **-3.098 ***-3.740 (1.249) (1.412)Per Capita Income (in $10,000) ***-0.060 ***-0.079 (0.019) (0.025)Unemployment Rate **-1.065 -0.693 (0.446) (0.520)Year1993 0.013 __ (0.009) __1994 *0.021 *0.018 (0.011) (0.011)1995 **0.031 *0.030 (0.014) (0.013)1996 ***0.055 ***0.052 (0.015) (0.015)1997 ***0.071 ***0.076 (0.019) (0.018)1998 ***0.088 ***0.095 (0.021) (0.020)1999 ***0.104 ***0.114 (0.023) (0.023)2000 ***0.112 ***0.134 (0.021) (0.025)2001 ***0.132 ***0.145 (0.022) (0.024)Number of MSAs 317 317Number of Years 10 9Huber Standard Errors Correction Yes Yes1 ASCs per 10,000 population.* Statistical significance at 10 percent.** Statistical significance at 5 percent.*** Statistical significance at 1 percent.NOTES: Standard errors are in parentheses. HMO is health maintenance organization. HHI is Herfindahl-Hirschman Index. MSA is metropolitanstatistical area.SOURCES: 2002 Medicare OSCAR System, HMO Enrollment File, 1992-2001 American Hospital Association Annual Survey Files, and Area Resource Files.with HMO penetration, the coefficient of (p < 0.05). All the estimates of year indica-the interaction term was statistically insig- tors were positive, statistically significant,nificant and the other coefficients in the and monotonically increasing from 1994­model were essentially unchanged. 2001, indicating a strong secular trend of The estimated effects of other covariates increased ASCs over that the growth of ASCs is also sig­nificantly associated with demographic and DISCUSSIONeconomic characteristics. ASCs were lesslikely to enter a market with a higher pro- This study has sought to provide insightportion of the elderly (p < 0.05), a market into the growth of ASCs in the U.S. fromwith higher per capita income (p < 0.01), or 1992-2001, during which period the num­a market with a higher unemployment rate ber of ASCs increased from approximatelyHealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4 119
  10. 10. 1,150 to over 2,900 in the MSAs studied. have been sufficient competition amongTwo hypotheses were advanced to explain local community hospitals to keep ambu­the diffusion of ASCs across metropoli­ latory surgical prices low. Again, withouttan areas. First, we argued that greater information on prices, we are unable tomanaged care penetration would result in directly confirm this speculation.greater entry of ASCs. This hypothesis Our study has some limitations. First,was rejected by the analysis. We found we only have information on the numberthat ASCs were less likely to enter markets of ASCs in each year. We do not know thewith greater HMO penetration. There are specialty or specialties of each ASC, norat least three explanations for this result. do we know the volume of services pro­One explanation is that ASCs are not the vided. As a result, we have only the crudestefficient, low-cost providers of care that sense of the market presence of each ASC.their advocates claim. If HMOs aggressive­ Clearly, further work would benefit fromly seek value in contracting, the negative the use of claims data from CMS or privaterelationship between HMO penetration sectors that would allow a more detailedand ASC entry would suggest that HMOs examination of the role that ASCs play indo not find ASCs attracting contracting providing outpatient surgical care. Second,partners. A second explanation is that because of the crudeness of the ASC data,HMOs have been able to negotiate lower we have not invested in developing anprices with existing hospital providers of MSA specific mapping of the hospital net­outpatient surgeries. As a result, the nego­ work formation that has occurred over thistiated lower prices have deterred ASCs period. As a result, our measures of hos­to enter and service the market. Without pital contraction are understated. Third,information on prices we are unable to although a preliminary report (Medicaretest this speculation. A third explanation Payment Advisory Commission, 2004a)is that greater HMO penetration has led to using cross-sectional data suggested thatgreater reliance on ASCs as outpatient sur­ CON regulations did not appear to be agery providers, but our data are too crude major factor in ASC growth, we have notto show this reliance. We only know the tested the impact of CON regulations, butnumber of ASCs in each MSA in each year. instead have relied on MSA fixed effects toIf managed care penetration is associated control for these generally stable legislativewith exclusive contracting with one large programs. However, if CON regulationsASC, to the exclusion of others, we could have an impact on the growth of ASCs, andhave fewer (but larger) ASCs in the local there were changes in CON regulationsmarket and all our data would indicate specifically aimed at curbing the growth ofwould be fewer ASCs. ASCs, our estimated effects of HMO pen­ The second hypothesis was that there etration and hospital concentration mightwould be more ASCs in more concentrated be markets. This hypothesis was con­ The traditional hospital market is infirmed by the analysis. A metropolitan mar­ the midst of significant changes with theket with 4 equal sized hospitals rather than ongoing diffusion of ASCs, the entry of5, other things equal, was likely to have 2.5 specialty hospitals, and the development ofadditional ASCs per million people. This other traditionally hospital based servicessuggests that ASCs have been more likely that are being offered independently ofto enter markets in which there may not the hospital. Much more research needs120 HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4
  11. 11. to be directed at understanding the market Iglehart, J.K.: The Emergence of Physician-Ownedforces at play, and the effects of these new Specialty Hospitals. New England Journal of Medicine 352(1):78-84, January 2005.types of providers on the volume and qual­ Lynk, W.J. and Longley, C.S.: The Effect of Physician-ity of care provided. Owned Surgicalcenters on Hospital Outpatient Surgery. Health Affairs 21(4):215-221, July/AugustaCKNOwleDgMeNT 2002. Melnick, G.A., Zwanziger, J., Bambezai, A. et al.: The authors would like to thank Laurence The Effects of Market Structure and Bargaining Position on Hospital Prices. Journal of HealthBaker and Kathleen Dalton for generously Economics 11(3):217-233, October 1992.providing part of the data for our analysis. Mitchell, J.M.: Effects of Physician-Owned Limited-In addition, we appreciate many construc­ Service Hospitals: Evidence From Arizona. Healthtive comments from Kathleen Dalton on Affairs W5:481-488, October 2005. Internet address:this manuscript. http://content.healthaf hlthaff.w5.481v1. (Accessed 2006.) Mitchell, J.M. and Sass, T.R.: Physician OwnershipReFeReNCeS of Ancillary Services: Indirect Demand Inducement or Quality Assurance? Journal of Health EconomicsAltman, D., Cutler, D., and Zeckhauser, R.: Enrollee 14(3):263-289, August 1995.Mix, Treatment Intensity, and Cost in CompetingIndemnity and HMO Plans. Journal of Health Medicare Payment Advisor y Commission:Economics 22(1):23-45, January 2003. Characteristics of Independent Diagnostic Testing Facilities and Ambulatory Surgical Centers. AprilBaker, L.C.: The Effect of HMOs on Fee-for-Service 2004a. Internet address: Expenditures: Evidence From Medicare. public_meetings/transcripts/0404_allcombined_Journal of Health Economics 16(4):453-481, August transcripts.pdf. (Accessed 2006.)1997. Medicare Payment Advisor y Commission:Bamezai, A., Zwanziger, J., Melnick, G.A., et al.: Ambulatory Surgical Center Services. Report toPrice Competition and Hospital Cost Growth in Congress: Medicare Payment Policy. Washington,the United States, 1989-1994. Journal of Health DC. 2004b.Economics 8(3):233-243, May 1999. Morrisey, M.A.: Competition in Hospital and HealthCasalino, L.P., Devers, K.J., and Brewster, L.R.: Insurance Markets: A Review and Research Agenda.Focused Factories? Physician-Owned Specialty Health Services Research 36(1 Pt 2):191-221, AprilFacilities. Health Affairs 22(6):56-67, November/ 2001.December 2003. Morrisey, M.A., Jensen, G.A., and Gabel, J.: ManagedCuellar, A.E. and Gertler, P.J.: Trends of Hospital Care and Employer Premiums. International JournalConsolidation: The Formation of Local Systems. of Health Care Finance and Economics 3(2):95-116,Health Affairs 22(6):77-87, November/December June 2003.2003. Stensland, J. and Winter, A.: Do Physician-OwnedDetmer, D.E. and Gelijns, A.C.: Ambulatory Surgery: Cardiac Hospitals Increase Utilization? HealthA More Cost-Effective Treatment Strategy? Archives Affairs 25(1):119-129, January/February 2006.of Surgery 129(2):123-127, February 1994. U.S. Government Accountability Office: CompetitionGreenwald, L., Cromwell, J., Adamache, W., et al.: and Other Factors Linked to Wide Variation inSpecialty Versus Community Hospitals: Referrals, Health Care Prices. GAO Report Number 05-856,Quality, and Community Benefits. Health Affairs August 2005.25(1):106-118, January/February 2006. U.S. Department of Health and Human Services:Guterman, S.: Specialty Hospitals: A Problem or a Medicare Program: Physicians’ Referrals to HealthSymptom? Health Affairs 25(1):95-105, January/ Care Entities with Which They Have FinancialFebruary 2006. Relationships (Phase II). March 2004. InternetHerzlinger, R.E.: Specialization and Its Discounts: address: terly­The Pernicious Impact of Regulations Against providerupdated/downloads/CMS18101FC.pdfSpecialization and Physician Ownership on the U.S. (Accessed 2006.)Healthcare System. Circulation 109(20):2376-2378,May 2004.HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4 121
  12. 12. U.S. Department of Justice: The Herfindahl-Hirschman Index. Internet address: 2006.)White, H.: A Heteroskedasticity-ConsistentCovariate Matrix Estimator and a Direct Test forHeteroskedasticity. Econometrica 48(4):817-838,May 1980.Winter, A.: Comparing the Mix of Patients inVarious Outpatient Surgery Settings. Health Affairs22(6):68-75, November/December 2003.Reprint Requests: John Bian, Ph.D., University of Alabama atBirmingham, MT 640, 1530 3rd Avenue S, Birmingham, AL35294. E-mail: jbian@uab.edu122 HealTH CaRe FINaNCINg RevIew/Summer 2006/Volume 27, Number 4