The American Journal of Managed Care June 2011
Hospital Market Concentration, Pricing, and Profitability in Orthopedic Surgery and Interventional Cardiology
Objective: To examine the association between hospital market concentration and pricing.
Background: Hospitals have been merging into systems that potentially wield bargaining power over private health insurers. Concern is growing among policy makers that these systems may respond to provisions of the 2010 health reform legislation by further increasing consolidation and prices.
Methods: Multivariate statistical methods were used to evaluate the association between hospital market concentration, prices, and profits (contribution margins) for commercially insured patients admitted for any of 6 major cardiac and orthopedic surgery procedures, adjusting for characteristics of the patient (diagnoses, comorbidities,
complications) and of the hospital (size, patient volume, teaching status). Data were obtained on 11,330 patients treated in 61 hospitals in 27 markets across 8 states in 2008.
Results: Hospital prices for patients in concentrated markets were higher than hospital prices for otherwise-comparable patients in competitive markets by 25.1% for coronary angioplasty, 13.0% for cardiac rhythm management (CRM) device insertion, 19.2% for total knee replacement, 24.1% for total hip replacement, 19.3% for lumbar spine fusion, and 22.7% for cervical spine fusion (P <.05). Contribution margins were higher in concentrated than in competitive hospital markets by $5259 for angioplasty, $3417 for CRM device insertion, $4123 for total knee replacement, $5889 for total hip replacement, $7931 for lumbar spine fusion, and $4663 for cervical spine fusion (P <.05).
Conclusion: Hospitals in concentrated markets charge significantly higher prices and earn significantly higher margins from private insurers than do hospitals in competitive markets.
(Am J Manag Care. 2011;17(6):e241-e248)
This study found that hospitals in more concentrated markets (with fewer competitors, after adjusting for urban/rural location and chain ownership) charge significantly higher prices to private insurers than do otherwise-similar hospitals in more competitive markets.
- Price differentials ranged from 15% to 26% for coronary angioplasty, cardiac rhythm management, knee replacement, hip replacement, lumbar spine fusion, and cervical spine fusion.
- The further consolidation of hospital markets, encouraged by the health reform legislation under the rubric of “accountable care organizations,” may have the unintended effect of increasing pricing leverage by hospitals.
The association between hospital market concentration and pricing has been a long-standing focus for federal antitrust agencies concerned with the rising costs of care. It is gaining new attention as a result of 2 provisions in the 2010 health reform legislation. First, the expansion of coverage is to be financed in part via a slowdown in Medicare payment rate updates relative to predicted trends, which may lead to further increases in the prices charged by hospitals to private insurers.5 Second, the consolidation of local markets may be accelerated by the provisions of the legislation that encourage hospitals and physicians in local markets to integrate and form accountable care organizations that provide the full spectrum of inpatient and outpatient services.6 It is unclear whether integration with physicians gives hospitals pricing power over and above what they achieve through integration with other nearby hospitals, but the DOJ and FTC have raised concerns about this ongoing process.7
DATA AND METHODS
Data were obtained on patients admitted to 61 hospitals in 2008 for coronary angioplasty with drug-eluting stent, insertion of cardiac rhythm management (CRM) device (pacemaker or implantable cardioverter defibrillator), total knee replacement, total hip replacement, lumbar spine fusion, or cervical spine fusion. These facilities were participants in the value-based purchasing initiative of the Integrated Healthcare Association, a coalition of large hospitals, medical groups, and health plans in California, or worked on value purchasing with Aspen Health Metrics, a hospital consulting firm.
Additional data on the hospitals where the procedures were performed were obtained from the American Hospital Association’s 2008 Annual Survey of Hospitals, including number of staffed beds, average annual earnings for hospital staff, and teaching status of the institution. The market for each hospital was identified as the Hospital Referral Region, developed by the Dartmouth Atlas based on patient flow data for Medicare patients.8 The Dartmouth Atlas assigns every hospital in the United States to one of 306 markets. The 61 hospitals used in this study are distributed across 27 of those markets, spanning 8 states. To control for the effect of market size, I also measured the population of the metropolitan regions served by each hospital.
The extent of consolidation of the local hospital market was measured in terms of the Herfindahl-Hirschman Index (HHI),9 the standard measure used in economic analyses of market competition. It is constructed by dividing the number of staffed beds for each facility by the total number of beds within the market to obtain each hospital’s share. The share of each facility then is squared and the shares of all hospitals are summed to create an index that potentially ranges from zero (many competitors, each with a negligible share) to 10,000 (1 facility, with 100% market share). For ease of interpretation, the index was scaled here so it ranged from a potential low of zero to a potential high of 100.
Some analyses of hospital market structure implicitly assume that every facility competes with every other facility in the same market, in that they construct the HHI using market shares for each facility separately. Given the importance of local chain ownership, this assumption likely is invalid. In this study, data from the American Hospital Association survey were used to identify the chain with which each hospital in the study was affiliated, if any. All hospitals owned by the same chain in the same market then were treated as part of the same organization in calculating market shares and HHI.
The hospitals included in this study were located in 27 local markets, as defined by the Hospital Referral Regions of the Dartmouth Atlas. These markets varied substantially in terms of the number of hospitals within them, from a low of 2 to a high of 92, with an average of 15.6. However, many of these individual hospitals within particular markets belonged to the same hospital chains and so were not competing with one another on the basis of price. When chain ownership was taken into account, the number of competitors ranged from a low of 2 to a high of 54, with a mean of 11.3. The traditional HHI, measured using each facility as a competitor regardless of chain ownership, had a mean of 21.1, while the chain-adjusted HHI had a mean of 25.1. By way of comparison, in 2008 the average HHI across all 306 markets in the nation was 19.5, and the chain-adjusted HHI was 25.2 (calculated using the American Hospital Association Annual Survey of Hospitals). The 6 procedures studied here are all performed on an inpatient rather than outpatient basis, so the relevant measure of market structure is that of the hospital rather than including, for instance, the presence of ambulatory surgery centers.
Prices charged by the hospitals were measured in terms of the amount collected from the private insurer for each patient, after all contractual discounts. Collected revenues per patient are a more valid indicator of true prices than are the hospital’s billed charges, which typically never are collected in full. In the multivariate statistical analyses, prices are measured on a logarithmic scale to facilitate interpretation as the association between market concentration and percentage (rather than dollar) differences in prices. I measured the profitability to the hospital of each patient in terms of the difference between the insurer’s payment and the hospital’s direct costs for treating that patient. This “contribution margin” measures the profitability of each particular patient exclusive of indirect hospital costs such as administrative overhead, depreciation of capital investments, and the charity care provided to uninsured patients.
The data included information on each patient’s principal diagnoses, comorbidities, age, discharge destination, and in-hospital complications. For hip and knee replacement procedures, coded diagnoses included osteoarthritis, rheumatoid arthritis, aseptic necrosis, and fractures. The knee and hip replacement analyses were limited to patients undergoing primary, rather than revision, surgery. For lumbar and cervical spine fusion, diagnoses included fracture, spondylolisthesis, and intervertebral disk disorder. For angioplasty, the analysis was limited to patients receiving a drug-eluting stent (as distinct from a bare metal stent or no stent) and adjusted for the number of stents used in the case. For CRM device insertion, the statistical analyses were adjusted for whether the implant was a single-chamber or dual-chamber pacemaker, a pacemaker with cardiac resynchronization therapy capability, an implantable cardioverter defibrillator, or an implantable cardioverter defibrillator with resynchronization capability. In these data, comorbidities were defined as preexisting conditions that resulted in an increase in the length of stay by at least 1 day. For orthopedic joint replacement and spine procedures, complications were defined as in-hospital events serious enough to result in at least 1 extra day of hospital stay. For angioplasty and CRM device insertion, complications were defined in terms of those serious enough to cause a shift in the patient’s diagnosis-related group assignment. Our measure of complications only captured events that occurred during the hospital stay; I had no data on events that occurred after discharge.
To examine the bivariate association between market structure and hospital performance, I divided hospitals according to whether their index of market concentration (HHI) was above or below the median for all study hospitals, and calculated average prices and contribution margins for patients undergoing each of the 6 procedures. I then conducted multivariate regression analyses of hospital prices and contribution margins as a function of market structure (HHI, population size), procedure volume (number of study procedures performed in the hospitalduring 2008), hospital characteristics (number of staffed beds, teaching status, average staff salary), and patient characteristics (principal diagnoses, age, comorbidities, complications, discharge destination). I modified the calculation of standard errors for the multivariate regression analyses to cluster for within-hospital correlation of prices and margins across patients. It is to be expected that unmeasured determinants of prices and margins will be correlated for patients treated at the same hospital.10
Table 1 presents means and standard deviations for the market, hospital, and patient characteristics used in the study. Average prices per procedure ranged from $21,125 for cervical fusion to $47,085 for lumbar fusion. These procedures were highly profitable, with contribution margins per patient ranging from $9483 for cervical spine fusion to $22,690 for lumbarspine fusion. In percentage terms, the contribution margins were 56% for angioplasty, 44% for CRM, 47% for knee replacement, 47% for hip replacement, 48% for lumbar fusion, and 45% for cervical fusion.
The bivariate association between hospital market concentration on the one hand and the prices charged and contribution margins earned on the other is presented in Table 2. Defining concentrated markets as those with an HHI above the median and competitive markets as those with an HHI below the median, the average price per procedure was significantly higher in concentrated markets than in competitive markets for all 6 procedures. The difference in price associated with market structure ranged from 29.3% for cervical fusion to 56.2% for CRM device insertion (P <.01 for all procedures).
The market-related differences in prices charged to commercial insurers were associated with substantial marketrelated differences in the contribution margins earned from commercial insurers. The average difference in contribution margins earned in concentrated markets compared with competitive markets was $9561 (90%) for angioplasty, $12,816 (116%) for CRM device insertion, $8147 (126%) for knee replacement, $9362 (133%) for hip replacement, $13,690 (95%) for lumbar fusion, and $4561 (64%) for cervical fusion (P < .01 for all procedures). It is interesting to note that these procedures generated positive contribution margins even in competitive markets where hospitals’ pricing leverage is weak. As indicated in Table 2, the average contribution margins for patients treated in competitive hospital markets was 49% for angioplasty, 36% for CRM insertion, 35% for knee replacement, 36% for hip replacement, 36% for lumbar fusion, and 39% for cervical fusion.