Percentage of patients who underwent carpal tunnel repair, by physician ownership status. Highlighted numbers indicate that the year-specific differences in frequency of use between owners and nonowners were statistically significant at P < .001.
Percentage of patients who underwent rotator cuff repair, by physician ownership status. Highlighted numbers indicate that the year-specific differences in frequency of use between owners and nonowners were statistically significant at P < .001.
Percentage of patients who underwent arthroscopic surgery of the knee, by physician ownership status. Highlighted numbers indicate that the year-specific differences in frequency of use between owners and nonowners were statistically significant at P < .001.
Mitchell JM. Effect of Physician Ownership of Specialty Hospitals and Ambulatory Surgery Centers on Frequency of Use of Outpatient Orthopedic Surgery. Arch Surg. 2010;145(8):732-738. doi:10.1001/archsurg.2010.149
Copyright 2010 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2010
Physician-owned specialty hospitals and ambulatory surgery centers have become commonplace in many markets throughout the United States. Little is known about whether the financial incentives linked to ownership affect frequency of outpatient surgery.
To evaluate if financial incentives linked to physician ownership influence frequency of outpatient orthopedic surgical procedures.
Design and Setting
We analyzed 5 years of claims data from a large private insurer in Idaho to compare frequency by orthopedic surgeon owners and nonowners of surgical procedures that could be performed in either ambulatory surgery centers or hospital outpatient surgery departments.
Main Outcome Measure
Frequency of use, calculated as number of patients treated with the specific diagnoses who received the surgical procedure of interest divided by the number of patients with such diagnoses treated by each physician.
Age- and sex-adjusted odds ratios indicate that the likelihood of having carpal tunnel repair was 54% to 129% higher for patients of surgeon owners compared with surgeon nonowners. For rotator cuff repair, the adjusted odds ratios of having surgery were 33% to 100% higher for patients treated by physician owners. The age- and sex-adjusted probability of arthroscopic surgery was 27% to 78% higher for patients of surgeon owners compared with surgeon nonowners.
The consistent finding of higher use rates by physician owners across time clearly suggests that financial incentives linked to ownership of either specialty hospitals or ambulatory surgery centers influence physicians' practice patterns.
Under existing federal law, it is illegal for physicians to refer Medicare and Medicaid patients to designated health care entities in which the physician has an ownership interest. About half the states have enacted similar prohibitions that apply to privately insured persons.1 The federal law and most state self-referral bans do not apply to “whole” hospitals and ambulatory surgery centers (ASCs).1,2 The whole-hospital exception was based on the presumption that because general hospitals offer a wide array of services, any referrals made by an individual physician investor would produce little personal economic gain.1 The ASCs (freestanding facilities that provide surgical procedures that do not require an overnight stay) were exempt from the self-referral prohibitions because they were viewed as an extension of the physician's practice. In fact, the federal safe-harbor guidelines require that physician owners of multispecialty ASCs perform at least one-third of their outpatient surgical procedures at the facility in which they have a financial stake. This safe-harbor guideline does not apply to single-specialty ASCs.3
The exceptions to federal and state self-referral prohibitions are one factor that has fostered the growth of specialty hospitals and ASCs that involve referring physician investors.1- 4 These so-called limited-service facilities have grown most rapidly in states with less stringent regulatory environments. Specialty hospitals have become increasingly common in about a dozen states that do not regulate the development of new hospitals or expansions of existing facilities.2,4,5 In contrast, the growth of ASCs has been more widespread because many state certificate-of-need laws have provisions that exempt ASCs.2,3 Other factors that have contributed to the growth of both types of facilities include physicians' desires to have greater involvement in management decisions, uninterrupted operating room schedules, and the potential to improve efficiency and quality that may result from specialization. Moreover, physician owners share in any profits that result from facility fees and ancillary procedures performed at the ASC or specialty hospital. Thus, ownership provides an avenue for physicians to supplement the income they generate from the provision of direct patient care.2,4,5 Although proponents tout the advantages physician-owned facilities provide (ie, more efficient management and scheduling of patients), critics maintain that ownership of specialty hospitals or ASCs represents an inherent conflict of interest for referring physician investors.2,4- 7 As owners, physicians have financial incentives to refer less-sick patients and those with good insurance to their own hospital or ASC. These financial incentives may result in increased utilization of profitable procedures and services, some of which may not be medically necessary. Because increased utilization is the driving force behind escalating health insurance premiums, such trends are likely to result in greater numbers of uninsured and underinsured persons. Such practices also may weaken the ability of competing general hospitals to provide safety net services.2,6,7
Nearly all prior research on physician-owned specialty hospitals is based on site visits and/or market-level analyses of Medicare inpatient claims.4- 11 Existing evidence based on inpatient procedures provides an incomplete picture because inpatient procedures account for only 20% of the services rendered at physician-owned spine, orthopedic, or surgical facilities.4 Although ASCs are geographically distributed throughout the United States, research documenting their effects on use of outpatient surgery is limited.12- 19 Previous research provides limited insight about whether patients who are treated by physician owners of ASCs or specialty hospitals are more likely to undergo outpatient surgical procedures compared with patients with the same clinical conditions who seek care from physician nonowners. To my knowledge, only 1 study to date has examined whether physician ownership of specialty hospitals influences the frequency of use of inpatient and outpatient surgery, but the analysis was restricted to workers' compensation cases that involved back or spine injuries.20 The present study addresses this significant gap in knowledge. I analyzed claims data from a large private insurer in Idaho to compare frequency (referral rates) by physician owners and physician nonowners of specific outpatient orthopedic surgical procedures that could be performed in either ASCs or hospital outpatient surgery departments.
Using data from state records and the insurer, I identified the freestanding ASCs, specialty hospitals, and full-service community hospitals located in Idaho. I compiled a database containing the incorporation date, location, ownership, and contact information for each facility. Nearly all of the growth in ASC availability had occurred since the mid-1990s. I also contacted each facility to obtain additional information on the types of surgical procedures performed, the number of operating rooms, and more details on physician investors for those facilities with incomplete ownership records. With further assistance from state documents on corporate filings and detailed physician records from the insurer, I was able to identify the physician owners of each ASC and specialty hospital in Idaho. I was also able to identify the date each physician-owned facility became operational, thus enabling us to accurately classify each physician by ownership status for each year.
The frequency-of-use comparisons are based on analyses of medical claims data for the years 2003 through 2007 from the large private insurer. The claims contain detailed information on each patient encounter, including the physician who performed the service, location, type of claim (professional or facility), and tax identification number; patient characteristics such as age, sex, and diagnoses; and details regarding the types of services received and type of coverage: deductible, copayments, amount billed, and amount reimbursed. In 2003, the insurer provided health insurance coverage (indemnity and preferred provider organization plans) for approximately 228 000 individuals. By 2007 it had covered an estimated 326 000 individuals, so the insurer's market share was about 40%.
I restricted the analysis to procedures performed at facilities located in Idaho. Services performed by out-of-state physicians were excluded because they account for a small share of all claims and do not directly compete with in-state physicians. The insurer uses a fee schedule to determine the amount it will pay for each procedure. Although the deductibles vary across plan types, the insurer pays 80% of the allowed fee for outpatient surgery and the patient is responsible for the remaining 20%. This cost-sharing arrangement applies to both the professional fee paid to the surgeon and the facility fee paid to the hospital or ASC. No enrollee was required to obtain plan approval before undergoing outpatient surgery. Individuals with Medicare supplemental coverage and those enrolled in Medicare health maintenance organizations were excluded from the analysis, but about 2% of patients were enrolled in a preferred provider organization Medicare Advantage plan.
I constructed frequency-of-use indicators for common outpatient surgical procedures performed by orthopedic surgeons: carpal tunnel repair, rotator cuff repair, and arthroscopic surgery of the knee. I developed the following algorithm to construct these indicators.
I identified for both orthopedic surgeon owners and nonowners the most commonly reported diagnoses associated with each surgical procedure of interest (Table 1).
For each orthopedic surgeon owner and nonowner, I identified all office visits of patients with these diagnoses that occurred within the 90-day period that preceded the surgical procedure of interest.
For each orthopedic surgeon owner and nonowner, I identified all patients with these diagnoses who had office visits but did not undergo the surgical procedure of interest.
I constructed frequency-of-use indicators; that is, the percentage of patients who received the surgical procedure of interest. The numerator is the number of patients who received the surgical procedure of interest performed by orthopedic surgeon owners (or nonowners) in each given year. The denominator is the sum of (1) patients who had an office visit with the designated diagnoses followed by the surgical procedure, (2) patients who had an office visit with the designated diagnoses but no surgery, and (3) patients who had no office visit with an orthopedic surgeon owner or nonowner but who underwent the surgical procedure for the designated diagnoses (these cases were infrequent but were included to obtain accurate counts of the number of procedures performed).
In calculating the frequency-of-use indicators, I addressed potential concerns to obtain accurate rates of use. First, to ensure that the orthopedic surgeons performed the procedures of interest on a regular basis, I eliminated cases associated with orthopedic surgeons who did not perform a minimum number of the surgical procedures of interest in a given year. The minimum was 5 for carpal tunnel repair, 1 for rotator cuff repair, and 5 for arthroscopic surgery of the knee. Although I dropped only a handful of physicians from each group, I also conducted sensitivity analyses by including these low-volume physicians. Second, I identified overlap cases, in which the office visit with the orthopedic surgeon owner or nonowner occurred in November or December but the surgery was performed in the following year. Each overlap case was assigned to the year the surgery was performed, but it was excluded from the denominator for the preceding year when the office visit occurred. Third, I excluded cases with patients who had an initial office visit with an orthopedic surgeon owner or nonowner during the fourth quarter of 2007 (October-December) but did not subsequently undergo surgery in 2007. These exclusions were made to allow for the possibility that these patients may have had surgery in 2008. Their inclusion in the denominator would be a source of potential bias in calculating frequency of use. Finally, I conducted sensitivity analyses using 60- and 120-day windows.
I stratified the frequency-of-use rates for each surgical procedure by ownership status and year. The null hypothesis assumes that there were no differences in the frequency of surgery performed by physician owners vs nonowners during a given year. I evaluated the null hypothesis for each surgical procedure by year (2003-2007) using a 2-tailed test of differences between proportions (P < .001). Statistically significant year-specific differences are highlighted in yellow on the figures (Figures 1, 2, and 3), depicting the comparisons. I also estimated a series of logistic regression models to predict the probability that a patient underwent the surgical procedure of interest in a given year, while controlling for physician ownership, age, and sex.
The health care marketplace in Idaho has undergone a dramatic transformation since 1995. Although the number of community hospitals remained stable at 37, 4 specialty hospitals owned solely by referring physicians entered the Idaho market. In 1995, 8 ASCs were located in Idaho and all but 1 were owned by referring physicians. By 2005, there was a 5-fold increase in the number of ASCs. Idaho currently has 42 ASCs, 39 of which are owned entirely by referring physicians. Physician-owned limited-service facilities account for 25% to 47% of the operating room capacity in the 6 regions of the state. About 25% of both owners and nonowners were practicing in the Boise area. The remainder in both groups, however, were geographically distributed throughout Idaho. Moreover, Idaho has no medical school and only 1 family practice residency program, so surgical procedures performed by nonowners are not concentrated in teaching institutions.
Table 2 reports the number of patients treated by orthopedic surgeon owners and nonowners through office visits and/or surgery by year and procedure type. The mean ages of men and women who underwent surgery performed by owners and nonowners were similar for carpal tunnel repair (48-56 years). For owner and nonowner physicians, women represented 65% to 73% of patients who had carpal tunnel repair. The age range of patients who underwent rotator cuff repair was similar for owners and nonowners; men were aged 53 to 58 years and women were aged 54 to 58 years. Between 36% and 50% of patients who had rotator cuff repair performed by nonowners were women, whereas women accounted for 40% to 46% of such procedures performed by physician owners. For patients who had arthroscopic surgery of the knee, those treated by owners were somewhat younger (aged 38-45 years for men and aged 41-45 years for women) than those treated by nonowners (aged 42-47 years for men and aged 45-50 years for women). Approximately 48% to 52% of patients who had arthroscopic surgery performed by nonowner physicians were women, compared with 44% to 46% for knee surgery performed by owners.
As shown in Figure 1, the frequency of carpal tunnel repair for patients treated by owners was significantly higher than the rate for patients treated by nonowners (P < .001). In 2003, the frequency of use by owners was 33.1% compared with 24.2% for nonowners, a difference of nearly 9 percentage points (P < .001). By 2007, this difference had widened to 15.6 percentage points: 32.8% for owners vs 17.2% for nonowners. Age- and sex-adjusted odds ratios (Table 3) show that the odds of having carpal tunnel repair were 54% to 129% higher for patients of owners relative to nonowners (P < .001).
Figure 2 depicts frequency-of-use comparisons for rotator cuff repair. In 2003-2004, there were no differences in frequency of use by ownership status. During the period 2005-2007, frequency of rotator cuff repair for patients treated by owners was 4.6 to 8.8 percentage points higher compared with nonowners (P < .001). Adjusted odds ratios (Table 3) indicate that the likelihood of undergoing rotator cuff repair during the period 2005-2007 was 33% to 100% higher for patients treated by owners compared with those treated by nonowners (P < .001).
Figure 3 presents comparisons for arthroscopic surgery of the knee. In 2003, differences between owners and nonowners were not significant. This pattern changed in 2004, when the frequency of use was 33.2% for owners compared with 24.5% for nonowners, a difference of close to 9 percentage points (P < .001). The difference widened further to nearly 13 percentage points in 2005 (P < .001). In 2006-2007, the differences in frequency of use between owners and nonowners had narrowed somewhat but remained statistically significant, at 7.8 and 5.5 percentage points, respectively (P < .001). Adjusted odds ratios (Table 3) show that during the period 2004-2007, patients treated by owners were 27% to 78% more likely to undergo knee surgery compared with those treated by nonowners (P < .001).
Since 2000, specialty hospitals and ASCs that involve referring physician investors have grown rapidly in several markets throughout the country. Whereas the pros and cons of physician ownership of limited-service facilities have been the subject of extensive debate among advocates and critics, the most controversial issue relates to the practice of physician self-referral.2,6,7,21 Three recent studies based on Medicare inpatient claims found that when a physician-owned specialty hospital enters the market, there is subsequently an increase in the use of procedures of interest.5,8,9 Other evidence indicates that specialty hospitals treat primarily low-acuity cases within specific diagnosis related groups as well as those with few comorbid conditions.4,5,10,11 Furthermore, relative to general hospitals, specialty facilities treat small percentages of Medicaid patients and negligible numbers of uninsured patients and are less likely to provide emergency care.4,10 Prior research on ASCs is composed of case studies,12,13 trend analyses documenting their growth,14 comparisons of the patient mix treated in freestanding ASCs and hospital outpatient surgery departments,15,16 and comparisons of annual surgical caseloads treated by owners vs nonowners.17- 19 Existing evidence, therefore, provides limited insight into whether the financial incentives associated with ownership of specialty hospitals or ASCs influence the frequency of use for outpatient surgical procedures.
This study analyzed 5 years of claims data from a large private insurer in Idaho to provide new evidence on this issue. The analyses evaluated the relationship between physician ownership and frequency of use for 3 common outpatient surgical procedures: carpal tunnel repair, rotator cuff repair, and arthroscopic surgery of the knee. The health care marketplace in Idaho is an appropriate laboratory to investigate these issues for several reasons. First, the state has almost equal numbers of physician-owned ASCs and competing general hospitals, along with 4 physician-owned specialty hospitals. Nearly all of these facilities were established before 2003, so the findings reflect a somewhat steady-state picture of the effects of physician ownership on frequency of use of outpatient surgical procedures. Second, I was able to accurately classify orthopedic surgeons as either owners or nonowners because the insurer requires physicians and facilities to report ownership and financial relationships. Finally, Idaho is an attractive marketplace for specialty hospitals and ASCs to locate because it has no state certificate-of-need laws and favorable reimbursement rates due to the absence of capitated managed care plans.
With few exceptions during the 5-year period, frequency of use for each of the orthopedic procedures examined was significantly higher for physician owners compared with physician nonowners. The age- and sex-adjusted probability of undergoing carpal tunnel repair was 54% to 129% higher for owners compared with nonowners. The age- and sex-adjusted likelihood of rotator cuff repair was 33% to 100% higher if the patient was treated by a physician owner. Age- and sex-adjusted odds ratios indicate that the likelihood of having arthroscopic surgery was 27% to 78% higher for patients of owners compared with nonowners. Moreover, the results were similar when I altered the episode window to 60 or 120 days. The consistent finding of higher adjusted use rates by physician owners across time clearly suggests that financial incentives linked to ownership of specialty hospitals or ASCs influence physicians' practice patterns.
I also investigated factors other than financial incentives that might account for the higher referral rates that characterize orthopedic surgeon owners. One consideration relates to the patient mix. Patients treated by physician owners might be healthier and therefore more likely to tolerate surgery. Alternatively, they may have more severe orthopedic problems, implying that surgery is warranted. Although such issues about patient mix cannot be addressed with claims data, it is important to recognize that information on physician ownership is generally not disclosed to patients. Considering the absence of full disclosure, there is no reason why healthier patients or those with more severe orthopedic injuries should be predisposed to seek care from physician owners rather than physician nonowners. A second possible explanation is that patients of physician owners may be enrolled in health insurance plans that require less cost sharing, and as a consequence these patients may be more prone to consent to elective surgery. This explanation, however, cannot account for the differences in use rates that exist between owners and nonowners because a 20% cost-sharing rate is applied to both the professional and facility payments for outpatient surgical procedures. A third possible explanation is patient preferences. Although I have no data on patient choice, I contend that this consideration would be dominated by financial incentives linked to physician ownership. I conducted analyses to evaluate the hypothesis that physician owners work primarily in their own facility, whereas nonowners perform surgery in nonowner facilities. For arthroscopic surgery of the knee, I found that physician owners perform 75% to 84% of these procedures in the ASC or specialty hospital in which they have a financial stake. The reverse characterizes the practice patterns of physician nonowners. Consistent with expectations, nonowners perform between 77% and 81% of these procedures in nonowner facilities. The other procedures showed similar patterns.
Although the results reported in this article provide new insights on how financial incentives affect physicians' practice patterns, the study has some limitations. First, the analyses are based on claims records from a large private insurer in Idaho. Thus, one consideration is whether similar patterns would be evident in other markets with a high concentration of physician-owned specialty hospitals and ASCs. Several studies have found higher use rates for inpatient procedures in markets with established physician-owned specialty hospitals,5,8,9 so it seems likely that analyses of outpatient surgical procedures would exhibit higher use rates as well. A related question is whether the findings are applicable to patients with other types of generous insurance. A recently published study comparing frequency of use of inpatient and outpatient procedures used to treat injured workers by physician owners and nonowners suggests that similar results are evident among patients with other generous insurance coverage.20 A third limitation is my inability to ascertain whether the higher frequency-of-use rates that characterize physician owners represent surgery that has little marginal benefit to patients who receive it. A recent study by Kirkley et al22 published in the New England Journal of Medicine suggests that this may be the case for at least 1 of the procedures examined in our article: arthroscopic surgery of the knee. The authors found that arthroscopic surgery of the knee for the treatment of osteoarthritis yielded no improvement in outcomes compared with patients who received only physical and medical therapy. These procedures are estimated to cost about $5000 per case, which is 2 to 3 times higher than a course of physical and medical therapy. If a substantial proportion of the increased frequency of surgery by physician owners is performed in marginal cases, then a share of the greater expenditures associated with self-referral represent inefficient spending.
In conclusion, these findings documenting the effects of physician self-referral on frequency of use of elective surgical procedures performed on well-insured individuals should be of concern to federal and state legislators, third-party payers, employers, and patients. The increased utilization that characterizes physician self-referral arrangements will continue to lead to higher health insurance premiums, which in turn is likely to result in more uninsured and underinsured persons.
Correspondence: Jean M. Mitchell, PhD, Georgetown Public Policy Institute, Georgetown University, 3520 Prospect St NW, Room 423, Washington, DC 20007 (email@example.com).
Accepted for Publication: April 28, 2009.
Financial Disclosure: None reported.
Funding/Support: Research support was provided through an unrestricted educational research contract between the Idaho Hospital Association and Georgetown University.