Population-based rates of surgical procedures by calendar year in hospital service areas (HSAs) where an ambulatory surgery center (ASC) was never present and HSAs that experienced an ASC opening after 1998. Procedures included cancer-directed breast surgery (A), cataract surgery (B), colonoscopies (C), and upper gastrointestinal tract endoscopies (D). Rates were derived from our primary analysis of Medicare-eligible persons 65 years or older. All rates were adjusted for age and sex using direct standardization. Trend lines were generated using fractional polynomial regression. Data were obtained from Florida files from the State Ambulatory Surgery Databases.
The effect of an ambulatory surgery center (ASC) opening on a competing full-service hospital's surgical volume and patient mix. Stacked bar graphs refer to the proportion of hospital patients with a given Charlson score the year before and after an ASC opened in the same HSA. Line plots refer to hospital outpatient department (OPD) mean relative value unit (RVU) totals the year before and after an ASC opened in the same hospital service area. Data were obtained from Florida files from the State Ambulatory Surgery Databases. GI indicates gastrointestinal.
Hollingsworth JM, Krein SL, Ye Z, Kim HM, Hollenbeck BK. Opening of Ambulatory Surgery Centers and Procedure Use in Elderly PatientsData From Florida. Arch Surg. 2011;146(2):187–193. doi:10.1001/archsurg.2010.335
Ambulatory surgery centers (ASCs) potentially deliver care more efficiently than hospitals. However, ASC proliferation may increase discretionary surgery use because of financial incentives for the physicians who staff them. To explore this possibility, we measured the impact of the opening of an ASC in a health care market, as defined by the hospital service area (HSA), on rates of procedure use.
With a 100% sample of outpatient surgery encounters, we measured annual rates of use for discretionary (cataract surgery, colonoscopy, and upper gastrointestinal [GI] tract endoscopy) and imperative (cancer-directed breast surgery) procedures among Medicare-eligible persons. Using a multiple time series research design, we compared the change in procedure use for HSAs where ASCs opened with that of HSAs where an ASC was never present.
All patients 65 years or older undergoing outpatient surgery from January 1, 1998, through December 31, 2006.
Main Outcome Measure
Adjusted HSA-level rates of procedure use.
In HSAs where an ASC opened, colonoscopy use increased by 1610 procedures per 100 000 per year (95% confidence interval, 736-2485; P < .001) and upper GI tract endoscopy use increased by 775 procedures per 100 000 per year (159-1391; P = .01). However, rates of cancer-directed breast surgery remained flat. Among HSAs where an ASC opened, the relative increases in colonoscopy and upper GI tract endoscopy use were approximately 117% and 93% higher, respectively, 4 years after the opening compared with HSAs without ASCs.
The opening of an ASC within an HSA is associated with significant increases in discretionary surgery use.
Nearly 57 million outpatient procedures are performed annually in the United States, 14 million of which occur in elderly patients.1 Increasing use of these minor, yet common, procedures contributes to rising health care expenditures.2 Once exclusively within the hospital's domain, more and more outpatient procedures are being performed in freestanding ambulatory surgery centers (ASCs), visits to which have increased 300% during the past decade.1 Concurrent with this growing demand, the number of ASCs has more than doubled since the 1990s, with more than 5000 facilities currently in operation nationwide.3
Ambulatory surgery centers offer improved efficiency in health care delivery,4,5 allowing patients to spend less time in the health care setting.6 Their quicker patient turnover rates may also increase provider productivity.7 Despite these benefits, almost all ASCs are owned, in part, by the physicians who staff them,8 and the financial incentives related to ownership may alter provider behavior. Indeed, owners are more likely to refer their well-insured patients to their facilities and route the underinsured ones to hospitals.9 Moreover, a physician's procedure use significantly increases after acquiring an ownership stake in an ASC.10 Insofar as financial incentives motivate owners to lower their treatment thresholds,11,12 ASC expansion may increase overall procedure use and, perhaps, drive health care costs higher.
To evaluate this possibility, we used a 100% sample of ambulatory surgery encounters from Florida. We concentrated on the following 2 types of health care markets: those in which an ASC was observed to open and those where an ASC was never present. Using a multiple time series research design,13 we measured whether the entry of an ASC into a market was associated with greater use of discretionary (cataract surgery, colonoscopy, and upper gastrointestinal tract [GI] endoscopy) and imperative (cancer-directed breast surgery) procedures. To assess the relative value of ASC availability, we determined the extent to which the opening of an ASC in an HSA affected the surgical indications for patients who underwent these procedures.
We obtained Florida data from the State Ambulatory Surgery Databases, which contain the universe of ambulatory surgery encounters, including records from every hospital and freestanding ASC.14 For our primary analysis, we restricted our study population to Medicare-eligible persons 65 years or older. Because Medicare is the largest purchaser of ambulatory surgical care,1 examining outpatient surgery trends among its beneficiaries helps to inform the policy debate on ASCs.
We used Current Procedural Terminology (CPT) codes to identify encounters for cataract surgery (66820-66900), colonoscopy (45355-45392), and upper GI tract endoscopy (43200-43325) from January 1, 1998, through December 31, 2006. We chose these procedures because they are commonly performed on elderly patients,15 span multiple medical disciplines, and, owing to their discretionary nature, are susceptible to the misapplication of financial incentives. We also identified encounters for cancer-directed breast surgery (CPT codes 19120, 19125, 19126, 19160, and 19162). Because the incidence of breast cancer essentially determines the need for surgery, these procedures served as a “control” to help us disentangle the effect of ASC opening from the addition of capacity to the market. Through the data source's hospital number, we ascertained all facilities at which 1 or more procedures were performed during the study period. We used explicit codes provided by the State Ambulatory Surgery Databases to distinguish between hospital and ASC discharges.
Next, we divided Florida into 114 health care markets using the hospital service area (HSA) boundaries specified by Wennberg.16 An HSA represents a collection of zip codes in which residents receive most of their hospitalizations from the hospitals in that area. We chose the HSA as our marker of the health care market because most ambulatory procedures are considered minor, and patients likely undergo them in the areas where they most commonly receive their health care. Based on its zip code, we assigned each hospital and ASC to one of the HSAs in which our procedures of interest were performed.
We sorted all HSAs into the following 3 mutually exclusive categories: (1) those where at least 1 ASC was present at the start of the study period (ie, performing a given procedure in 1998); (2) those that were initially without an ASC but had one open in 1999 or later; and (3) those where an ASC never opened within their boundaries. Because we were interested in measuring the association between the opening of an ASC and market-level procedure use, we focused our analyses on the second and third HSA categories. In terms of their population characteristics, we found no statistically significant differences between HSAs where an ASC was present at the start of the study period and those where an ASC opened after 1998.
We then calculated annual HSA-level rates of procedure use. The numerator for our rate calculation was the number of times that a given surgery was performed in an HSA during a specific calendar year. The denominator represented the number of people 65 years or older within the HSA that year. Using population estimates for the zip code tabulation areas,17 we summed across an HSA's constituent zip codes to determine the population at risk. We used direct standardization to adjust all rates by age and sex to the 2000 US population.
With the HSA serving as our unit of analysis and the opening of an ASC as our exposure, we made comparisons between HSA types based on economic data, the level of education and urbanization within them, and the availability of their health care resources. We also examined differences between HSAs with respect to the patients who underwent outpatient surgery within them. Specifically, we compared a variety of demographic characteristics (patient age, sex, race, primary payer, socioeconomic status,18 and comorbidity status19) for persons treated in the HSAs, using parametric and nonparametric statistics where appropriate.
We plotted the population-based rates of cancer-directed breast surgery, cataract surgery, colonoscopy, and upper GI tract endoscopy by calendar year for HSAs that experienced an ASC opening in 1999 or later and those where an ASC was never present. We assessed trends visually using fractional polynomial regression.20 Using a multiple time series research design, we measured the rate of change of surgery use across an HSA after the opening of an ASC in it. This approach reduces bias from the following 2 potential sources: (1) a difference between HSA types that is stable over time cannot be mistaken for an effect of an ASC opening because each HSA is compared with itself, and (2) changes over time that affect all HSAs similarly (eg, seasonal migration of elderly adults in Florida) cannot be mistaken for an effect of an ASC opening.21 Furthermore, the longitudinal nature of this approach allows us to infer directionality.
Given the potential correlation of observations, we used generalized estimating equations,22 assuming a first-order autoregressive correlation matrix structure with robust variance estimators.23 Our models included a time-varying indicator for the HSA type (set to 1 when an ASC opened in an HSA and 0 otherwise) and time, in years, since a new ASC entered an HSA (counting began when the first ASC opened in an HSA but remained 0 throughout for those HSAs in which no ASCs were present). We accounted for temporal trends by introducing calendar year as a continuous fixed effect. We adjusted for multiplicity of ASCs with a time-varying indicator for the presence of more than 1 ASC in an HSA during a calendar year. We also controlled for differences in market-level factors. To assess whether our results held for a younger population, we repeated our models to include adults older than 45 years.
To better understand the mechanisms underlying the observed trends, we evaluated the extent to which indications for these procedures changed in HSAs that experienced an opening of an ASC. For colonoscopy (CPT codes 45300, 45305, 45308, 45309, 45315, 45320, 45330, 45331, 45333, 45338, 45339, 45378, 45380, 45383, 45384, and 45385), we determined the proportion of cases in which the procedure was completed for screening purposes (ie, the patient undergoing it had no personal history of colon polyps [International Classification of Diseases, Ninth Revision (ICD-9) code V12.72; colorectal cancer [ICD-9 code V10.05]; inflammatory bowel disease [ICD-9 codes 555.x or 556.x]; GI symptoms [ie, abdominal pain (ICD-9 codes 787.3, 789.0x, or 789.6x), altered bowel habits (ICD-9 codes 564.0 or 787.x), or GI tract bleeding (ICD-9 codes 578.x or 792.1)]; weight loss [ICD-9 code 783.2]; or anemia [ICD-9 codes 280.x or 285.9]). For cataract surgery, we measured the frequency with which laser capsulotomy (CPT code 66821) for postcataract treatment (ICD-9 codes 336.50, 366.51, or 366.53) was performed. For upper GI tract endoscopy, we evaluated how often the procedure was performed for dyspepsia (ICD-9 codes 536.8 or 787.1) with no alarm features (ie, adults younger than 56 years without a concurrent diagnosis of persistent vomiting [ICD-9 codes 536.2, 787.0, 787.01, or 787.03], dysphagia [ICD-9 code 787.2x], weight loss, or anemia). A physician's decision to proceed with the latter 2 procedures is largely based on clinical judgment.24,25
To assess the effect of ASCs on competing hospitals, we identified hospitals located in HSAs that were initially without an ASC but then experienced one opening in 1999 or later. We then calculated annual surgical volumes at the competing hospitals within these HSAs. We assigned resource-based relative value units (RVUs) for each procedure (according to its level of complexity), and we tabulated RVU totals for these hospitals by procedure type and calendar year. Next, we measured the procedure-specific annual RVU totals at hospitals in the year before a competing ASC opened vs those in the year after it opened. Finally, we evaluated changes in the payer mix and level of disease severity of patients treated at hospitals from 1 year before to 1 year after a competing ASC opened within the same HSA.
All tests were 2 tailed and performed at a significance level of .05. In accordance with the Code of Federal Regulations Title 45, subpart A, section 46.101, paragraph b, subparagraph 4, institutional review board approval was waived for this study.
Ambulatory surgery centers generally entered markets with higher levels of education and urbanization and greater availability of health care resources (Table 1). As shown in Table 2, patients treated in an HSA that experienced an ASC opening were more likely to be white and to come from a higher socioeconomic class (P < .001 for each comparison). In addition, patients treated in these HSAs tended to be healthier (P < .001 for each comparison). With the exception of cancer-directed breast surgery, there were no differences in surgery rates between the HSAs at the start of the study period; however, Figure 1 illustrates that rates of cataract surgery, colonoscopy, and upper GI endoscopy rose for HSAs where an ASC opened and remained flat in those without an ASC.
On multivariable analysis, rates of colonoscopy for an HSA increased, on average, by 1610 procedures per 100 000 per year (95% confidence interval [CI], 736-2485; P < .001), and those for upper GI tract endoscopy increased by 775 procedures per 100 000 per year (159-1391; P = .01) after an ASC opened. A similar trend in rates was noted for cataract surgery, but this result was not statistically significant (1260 procedures per 100 000 per year; 95% CI, −208 to 2727; P = .09). Controlling for differences in community-level factors did not affect these estimates. No association between rates of cancer-directed breast surgery and opening of an ASC was found. The magnitudes of the rate increases for colonoscopy (1262 procedures per 100 000 per year; 95% CI, 43-2482; P = .04) and upper GI tract endoscopy (403 procedures per 100 000 per year; 20-786; P = .04) were attenuated when all adults older than 45 years were included.
Four years after their opening, the relative increases in the rates of colonoscopy and upper GI tract endoscopy were approximately 117% and 93% higher, respectively, for HSAs where ASCs opened compared with those without them (Table 3). As colonoscopy rates increased after ASC entry, the proportion of patients undergoing a screening procedure decreased from 80.2% the year before to 77.8% the year after opening (P < .001). Conversely, the proportion of patients who underwent upper GI tract endoscopy for dyspepsia without alarm features increased (79.6% to 89.1%; P = .008), as did the proportion of patients who received laser capsulotomy for postcataract treatment (14.5% to 22.4%; P < .001).
Although colonoscopy and upper GI tract endoscopy use increased in markets after an ASC opened, the hospitals within them witnessed declining surgical volumes (Figure 2). Relative to the year before a competing ASC opened, mean RVU totals at hospitals fell 40% for colonoscopy (3476 vs 2101 RVUs; P = .01 for the mean difference) and 35% for upper GI tract endoscopy (1523 vs 997; P = .04 for the mean difference) the year after an ASC entered the hospital's market. The decreases in mean RVU totals observed for cancer-directed breast surgery and cataract surgery were not statistically significant. As their annual caseloads declined, the level of disease severity for which hospitals provided care increased significantly across all 4 procedures (Figure 2). The largest absolute difference was evident for cancer-directed breast surgery in which the proportion of patients with a Charlson score of 0 fell from 58% the year before to 45% the year after a competing ASC opened (P < .001). Finally, there was a significant decrease, albeit small, in the proportion of patients treated at the hospital before and after ASC opening for whom Medicare was the primary payer (for colonoscopy, 88% vs 86%; P < .001; for upper GI tract endoscopy, 89% vs 86%; P < .001).
The opening of an ASC within a health care market is associated with significant increases in population-based rates of colonoscopy and upper GI tract endoscopy among patients 65 years or older. This growth is due, in part, to increased use of procedures for more discretionary indications (eg, upper GI tract endoscopy for dyspepsia with no alarm features). We observed a similar association for cataract surgery, but this was not statistically significant. In contrast, rates of cancer-directed breast surgery, which parallel the incidence of breast cancer, were independent of ASC openings and remained flat over time. This finding suggests that ASC expansion leads to increased procedure use beyond the simple addition of capacity to the market. Furthermore, our data show that the opening of an ASC leads to the off-loading of ambulatory procedures from nearby hospitals. Although statistically significant changes in the patient and payer mixes at hospitals resulted from ASC competition, the absolute differences were small and are unlikely of economic significance.
Although similar analyses exist in the specialty hospital literature,26,27 our study is the first, to our knowledge, to demonstrate the effect of ASC expansion on market-area procedure use. To provide context, the increases that we observed translated into an estimated additional 25 000 colonoscopies and 9800 upper GI tract endoscopies from 2002 to 2006 in HSAs in which an ASC opened. One potential explanation is that ASCs may enter markets in which there is a pent-up demand for surgery. As such, greater procedure use might reflect the elimination of surgical backlogs. Indeed, there is a robust literature on the role that the bed supply of local hospitals plays in differences in health care service utilization.28- 30 However, our results relative to cancer-directed breast surgery, for which categorical rules exist to determine its use, show no increase in market-area procedure use after the entry of an ASC. These data suggest that the latent need in these markets is small and that an ASC opening leads to greater procedure use beyond the simple addition of capacity to the market.
Alternatively, greater use of discretionary procedures may relate to the influence of physician ownership. Nearly 83% of the physicians who staff ASCs have an ownership stake in them8 and are financially committed to ensuring their success. This may motivate some owners to lower treatment thresholds. Indeed, there are numerous studies documenting increased use of health care resources related to ownership.31- 33 However, without knowing what the right market rate of surgery is, we cannot directly assess the extent to which our findings are explained by induced demand. The trends in upper GI tract endoscopy for dyspepsia without alarm features and laser capsulotomy provide some insight. A physician's decision to proceed with either procedure is largely based on clinical judgment.24,25 Thus, the greater proportion of patients undergoing these procedures after an ASC opening suggests that the overall increases in procedure use are at least partially influenced by lowered treatment thresholds. In the same context, declining use of screening colonoscopy after an ASC opening further supports the possibility that volume growth occurs in the marginal patient.
Evidence on case redistribution from the hospital to the ASC is limited.34,35 Competition between ASCs and hospitals is generally considered favorable from the perspective of patient empowerment. Moreover, ASCs were developed with the expressed intent of off-loading outpatient procedures from the hospital.36 Critics warn, however, that this redistribution may result in the skimming of the well-insured and the healthy patients from the hospital. If full-service hospitals are burdened with more uninsured and medically complex patients, then their ability to cross-subsidize safety-net care is potentially threatened.37 However, our data reveal that only small changes to hospitals' patient and payer mixes result from ASC competition.
We must recognize several limitations to our study. First, it is possible to detect the effects of other unmeasured factors that are coincident with an ASC presence and inappropriately attribute market and competing hospital changes to the opening of the ASC. To guard against this, we used an unexposed group in our analysis (ie, HSAs where an ASC was never present). Second, given that the State Ambulatory Surgery Database files are at the discharge level, we were unable to measure movement of patients from one HSA to another after ASC opening. Because procedure rates for the unexposed HSAs remained flat, it is unlikely that the increases in utilization associated with an ASC opening came at the expense of neighboring HSAs. Finally, the strictness of Florida's Certificate of Need program is low,38 and our results may not be generalizable to states with more stringent Certificate of Need legislation.
Despite these limitations, these data help inform the policy debate on ASCs. With total payments to ASCs having increased 190% during the past decade,39 these utilization trends may have considerable economic implications. The use of national data to assess more directly the implications of ASC opening and regulatory restrictions on procedure rates and costs will lend further insight into the underpinnings of the growth of outpatient surgery. If the savings accrued by lower costs per case40 are mitigated by increases in the total number of procedures performed, then revisions to current Stark law may be required. Alternatively, new delivery system models (eg, accountable care organizations) may reduce the incentive to do more altogether.
Correspondence: Brent K. Hollenbeck, MD, MS, Division of Oncology, Department of Urology, University of Michigan, Room 1032B, Michigan House, 2301 Commonwealth Blvd, Ann Arbor, MI 48105-2967 (firstname.lastname@example.org).
Accepted for Publication: January 13, 2010.
Author Contributions:Study concept and design: Hollingsworth. Acquisition of data: Hollingsworth and Ye. Analysis and interpretation of data: Hollingsworth, Krein, Ye, Kim, and Hollenbeck. Drafting of the manuscript: Hollingsworth and Hollenbeck. Critical revision of the manuscript for important intellectual content: Hollingsworth, Krein, Ye, Kim, and Hollenbeck. Statistical analysis: Hollingsworth, Krein, Ye, and Kim. Obtained funding: Hollingsworth and Hollenbeck. Administrative, technical, and material support: Hollenbeck. Study supervision: Kim and Hollenbeck.
Financial Disclosure: None reported.
Funding/Support: This study was supported by the Robert Wood Johnson Foundation Clinical Scholars Program (Dr Hollingsworth), the American Cancer Society Pennsylvania Division Dr William and Rita Conrady–mentored Research Scholar grant MSRG-07-006-01-CPHPS (Dr Hollenbeck), the American Urological Association Foundation (Dr Hollenbeck), and Astellas Pharma US, Inc (Dr Hollenbeck).
Disclaimer: The funding sources had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.
Additional Contributions: Mary A. M. Rogers, PhD, MS, reviewed our analytic methods. Sameer D. Saini, MD, MS, and Jeremy Sussman, MD, helped in the interpretation of our results.