Secondary Logo

Journal Logo

Research Reports

A Comparison of Costs: How California Teaching Hospitals Achieved Slower Growth Than Nonteaching Hospitals in Operating Room Costs From 2005 to 2014

Childers, Christopher P. MD, PhD; Maggard-Gibbons, Melinda MD, MSHS; Nuckols, Teryl MD, MSHS

Author Information
doi: 10.1097/ACM.0000000000002844

Abstract

In 2016, the United States spent $3.3 trillion on health care, accounting for 17.9% of the nation’s gross domestic product.1 Surgical care accounts for one-third of all health care spending.2 Despite operations only taking a few hours, costs accrued in the operating room (OR) are the second most expensive part of a surgical patient’s inpatient stay, only eclipsed by room and board.3 As a result of this cost density, administrators, surgeons, and investigators may find the OR a target for cost reduction efforts. However, most efforts to date have focused on standardizing surgical supplies, which represent a small fraction of the total costs.4,5

Analyses in the 1980s and 1990s found that teaching hospitals spent up to 44% more per inpatient episode of care than their nonteaching counterparts, a finding believed to be related to greater intensity of care provided by teaching hospitals over and beyond the amount expected for a more complex patient population.6–9 However, these analyses are nearly 20 years old and included both medical and surgical patients. To our knowledge, little is currently known about the differences in the cost of surgical care between teaching and nonteaching hospitals. The few studies that have evaluated surgical care have either focused on Medicare payments to hospitals (i.e., reimbursement)10,11 or have been limited to indirect measures of cost, such as time per operation.12,13 The internal costs of delivering a service are distinct from the reimbursement a hospital receives from an insurer and also differ from the bill (i.e., charges) a hospital generates. Understanding the drivers of, and potential ways to curb, internal costs are critical to a hospital’s financial solvency.

From a policy standpoint, public and private payers are increasingly applying value-based payment policies that incentivize improvements in quality and reductions in cost.14 If teaching hospitals have higher costs than their counterparts, this would create a substantial impediment to their ability to compete under these emerging policies. However, in a recent analysis of data from California, we estimated the mean cost of OR time in hospital settings at $37 per minute, and surprisingly, unadjusted analyses suggested that costs may be lower and may have grown more slowly at teaching hospitals between fiscal years (FYs) 2005 and 2014.5 If confirmed, these findings would suggest that some teaching hospitals may fare better than expected under value-based payment models.

For this study, we first compared OR costs per minute between teaching and nonteaching hospitals, including examining major cost components and adjusting for factors that may contribute to differences by teaching status. Next, we performed a detailed longitudinal assessment of cost components from 2005 to 2014 to determine whether growth in OR costs differed between teaching and nonteaching hospitals, and to identify what drove any temporal changes. Finally, we explored longitudinal changes in surgical volume at teaching and nonteaching hospitals to assess how volume shifts may have contributed to changes in certain types of costs.

Method

Data sources and ethics review

We obtained annual financial statements from all licensed California hospitals between 2005 and 2014, compiled by the Office of Statewide Health Planning and Development (OSHPD). These statements provide a comprehensive summary of hospitals’ financial activities during the previous year. Each statement undergoes at least 1 audit by the OSHPD before public release. These statements provided most covariates of interest, including hospital ownership (government owned, for-profit, and not-for-profit), teaching status, and location. Teaching status was a dichotomous variable assigned by the OSHPD, based on data from the American Medical Association’s Graduate Medical Education Directory, hospital size, and the number of residents/fellows15 (see Supplemental Digital Appendix 1, available at http://links.lww.com/ACADMED/A702). Geographic location was operationalized using the federal Health Service Area (HSA) designation, which divides the state of California into 14 regions. We merged these financial statements with an additional hospital-level discharge file from the OSHPD that contained each hospital’s annual case mix index (CMI). A hospital’s annual CMI is the all-payer weighted average of the Medicare Severity Diagnosis Related Group codes attributed to a hospital in a given year. As a result, hospitals with higher CMIs generally care for a more clinically complex population.16 The CMI is not specific to surgery but instead reflects the entire inpatient population.

After merging the financial statements with the CMI file, we limited the analysis to short-term hospitals that reported complete surgical expenditure data to the OSHPD in a uniform accounting fashion. Short-term hospitals included general, specialty, and pediatric hospitals but excluded psychiatric hospitals and long-term care facilities. Some hospitals, such as Kaiser Permanente and Shriners hospitals, have been granted OSHPD waivers from submitting complete financial statements. Additional details regarding selection criteria are reported in our prior publication.5 The UCLA Institutional Review Board determined that the study was not human subjects research.

Deriving OR cost per minute and components of the OR cost per minute

This study focused on the “surgery and recovery” revenue center, which captures the costs of OR and recovery room care for inpatient ORs (located in the hospital but serving both inpatients and outpatients).17 We did not study ambulatory ORs or other revenue centers relevant to surgical care, such as room and board, radiology, laboratory, pathology, or blood bank.

The primary outcome was the average OR cost per minute for each hospital in a given year. Each hospital reports total annual costs for the surgery and recovery revenue center including both direct and indirect costs. Each hospital also reports the total minutes of surgery performed during the year, defined as the time from anesthesia start to anesthesia end (or, when no anesthesia is provided, surgery start to surgery end). This allowed us to divide the total costs by the total minutes of surgery to generate a hospital-level average cost per minute for the year. Using the cost per minute has 2 analytic merits. First, it controls for the fact that total costs will vary within a hospital from year to year as volume fluctuates, and second, it creates a normally distributed variable that improves model efficiency.

The OSHPD further defines 5 major cost components that together comprise total costs: 4 types of direct costs (wages, benefits, supplies, and other direct costs) and indirect costs.17 Direct costs are attributable to the patient care services provided by the surgery and recovery revenue center. Wages include both staff physically present in the OR (e.g., scrub and circulating nurses) and staff who keep the OR running (e.g., managers, technicians, aides, clerical staff). The OSHPD does not include surgeon or resident wages among direct costs because physicians are paid directly by insurers. The OSHPD also provides data on the components of wage-related costs, including the number of full-time equivalents (FTEs) overall and within each employment category (e.g., technicians) and the average hourly pay rate of workers overall and within each employment category. Costs related to worker benefits include vacation and sick leave, health insurance, and Federal Insurance Contribution Act taxes. Cost related to supplies includes the purchase price of nonbillable items such as sutures and drapes. Billable items, such as implants and mesh, are not included because they are accounted for in a separate revenue center. Finally, the other direct cost category includes costs associated with equipment depreciation, leases, utilities, insurance, and travel.17

In contrast to direct costs, by definition, indirect costs do not vary day to day with the volume of clinical services provided. Indirect costs are generated by non-revenue-producing centers, such as data processing, administration, billing, and security, that support revenue-producing centers, such as surgery and recovery. Under standard accounting procedures, indirect costs are allocated among each hospital’s revenue-producing centers.18

All 5 major cost components were standardized to the cost per minute. For wage-related components, we examined hourly pay rates, and we standardized FTEs to the number per 10,000 minutes of surgery.

Comparison of adjusted OR costs and OR cost components between teaching and nonteaching hospitals

To isolate the difference in OR costs between hospitals due to teaching status, we used 2014 data and performed multivariate linear regression to predict total OR cost per minute (dependent variable), adjusted for ownership, location, and CMI. After examining the total OR cost per minute between teaching and nonteaching hospitals, we then evaluated the 5 major OR cost components.

Comparison of temporal trends in OR costs and OR cost components between teaching and nonteaching hospitals

Next, we assessed changes in OR costs over time from 2005 to 2014. Mixed-effects linear regression was used to predict total OR cost per minute (dependent variable) using teaching status, year, a year-by-teaching-status interaction term (the primary independent variable), ownership, location, and CMI as fixed effects, and individual hospital random effects. The year-by-teaching-status interaction term enabled us to perform a difference-in-difference analysis that estimated changes over time at teaching and nonteaching hospitals (Δ) and then compared these changes between teaching and nonteaching hospitals (ΔΔ).

We replicated this analysis for each of the 5 major cost components (wages, benefits, supplies, other direct, and indirect) and the subcomponents of wages (hourly pay rate, FTEs per 10,000 minutes of surgery, and FTEs per 10,000 minutes in individual employment categories).

Comparison of temporal trends in surgical volumes between teaching and nonteaching hospitals

As described above, standardizing costs per minute controls for small changes in volume within and between hospitals. However, it may not allow examination of macroscopic changes. For example, a shift in volume away from nonteaching hospitals and into teaching hospitals may lower the cost per minute at teaching hospitals, not because they actively sought to reduce costs but, rather, because costs were diluted by increasing volume. This would be especially relevant for fixed costs such as indirect costs.

We explored this by comparing temporal trends in surgical volume (number of procedures per year and total minutes of OR time) between teaching and nonteaching hospitals from 2005 to 2014. We analyzed the data at the state level by aggregating volumes at all teaching hospitals and all nonteaching hospitals and then comparing changes in statewide volumes.

Sensitivity analyses

We performed 2 sensitivity analyses. First, because 16 hospitals changed teaching status over the study time period, we repeated our longitudinal analysis excluding those hospitals. Second, we repeated our analyses excluding 15 pediatric and specialty hospitals because these hospitals may have had systematically different OR costs than general acute care hospitals.

Statistical analysis

All statistical analyses were completed using STATA statistical software, version 15.1 (StataCorp, College Station, Texas). We did not adjust costs for inflation because OR costs grew faster than standard inflation indices and because the difference-in-difference analysis rendered this adjustment unnecessary.5 To remove outliers, all outcomes (e.g., costs, volumes) were winsorized to the 5th and 95th percentiles, by year. All regression models used complete case analysis with sample sizes indicated throughout.

Using results from regression analyses, we generated plots to graphically illustrate findings. We used 2-sided tests and α = 0.05. For the evaluation of temporal trends in volume, after winsorization at the hospital level, we summed volumes across all teaching hospitals and all nonteaching hospitals.

Results

Descriptive profile of teaching and nonteaching hospitals in California

Of 445 hospitals with financial statements in 2014, 302 were eligible for this analysis, including 30 (10.0%) teaching hospitals (Table 1). Compared with nonteaching hospitals, teaching hospitals had significantly more beds, discharges, and inpatient ORs (all P < .001) and cared for more complex patients (median CMI [interquartile range], 1.50 [1.23–1.71] vs 1.27 [1.11–1.42], P < .001). Teaching hospitals were more likely to be government owned (37% [11/30] vs 14% [38/272]) and less likely to be for-profit (0% [0/30] vs 29% [78/272]) than nonteaching hospitals (P < .001).

Table 1
Table 1:
Characteristics of Study Hospitals in Fiscal Year 2014, Stratified by Teaching Status, From a Comparative Study of Risk-Adjusted Operating Room Costs in California, 2005–2014

For the longitudinal analyses, 2,992 individual financial statements were available for eligible hospitals across the 10 years (range per year, 291–309), with 8.4% (251/2,992) coming from teaching hospitals (range per year, 7.3%–10.3%).

Adjusted OR costs and OR cost components between teaching and nonteaching hospitals in FY 2014

At teaching hospitals, the risk-adjusted total cost of OR time per minute was $28.93 (standard error $3.07), including $8.41 ($0.69) for wages, $3.46 ($0.35) for benefits, $1.81 ($0.53) for supplies, $2.73 ($0.41) for other direct costs, and $11.33 ($1.50) for indirect costs. At nonteaching hospitals, the total adjusted cost of OR time per minute was $38.37 ($0.93), including $10.20 ($0.21) for wages, $4.06 ($0.11) for benefits, $2.62 ($0.16) for supplies, $3.45 ($0.13) for other direct costs, and $16.99 ($0.45) for indirect costs (Figure 1).

Figure 1
Figure 1:
The components of operating room cost in fiscal year 2014, stratified by teaching status, from a comparative study of risk-adjusted operating room costs in California, 2005–2014. Error bars represent upper boundary of 95% confidence intervals. Estimates are based on complete case analysis: Of the 302 facilities reporting surgical expense data to the Office of Statewide Health Planning and Development,15 300 reported wages and benefits, 295 reported supplies, 294 reported other direct costs, and 302 reported indirect (allocated) costs. Abbreviation: USD indicates U.S. dollars.

The adjusted total cost of OR time was $9.44 per minute less in teaching compared with nonteaching hospitals (95% CI, $3.03–$15.85, P = .004), including $1.79 per minute lower wages (95% CI, $0.34–$3.23, P = .01) and $5.66 per minute lower indirect costs (95% CI, $2.52–$8.81, P < .001). The cost per minute of benefits, supplies, or other direct expenses was similar between teaching and nonteaching hospitals (Figure 1).

Total OR costs were not associated with hospital ownership or CMI quartile. There were small differences in mean costs by geographic region (i.e., HSA; see Supplemental Digital Appendix 2, available at http://links.lww.com/ACADMED/A702).

Temporal trends in OR costs and OR cost components at teaching and nonteaching hospitals

From 2005 to 2014, adjusted costs per minute rose from $23.86 to $29.99 at teaching hospitals (Δ, $6.13 [95% CI, $2.84–$9.42], P < .001) and from $23.56 to $39.75 at nonteaching hospitals (Δ, $16.19 [95% CI, $14.96–$17.43], P < .001).

At teaching compared with nonteaching hospitals, growth in adjusted OR costs per minute was significantly slower from 2005 to 2014 (ΔΔ, −$1.11 per year [95% CI, −$1.50 to −$0.74], P < .001, Figure 2A). Factors contributing to the slower growth in adjusted OR costs per minute at teaching hospitals included slower growth in indirect costs per minute (−$0.77 per year [95% CI, −$0.96 to −$0.58], P < .001, Figure 2B) and wages per minute (−$0.13 per year [95% CI, −$0.22 to −$0.04], P = .005, Figure 2C). Increases in benefits (Figure 2D), supplies (Figure 2E), or other direct costs (Figure 2F) from 2005 to 2014 did not differ between teaching and nonteaching hospitals.

Figure 2
Figure 2:
Trends (FY 2005–FY 2014) in the components of operating room cost between teaching and nonteaching hospitals, from a comparative study of risk-adjusted operating room costs in California, 2005–2014. Figure 2A indicates the total operating room costs; 2B the total indirect costs; and 2C–2F the total direct costs: 2C, wages costs; 2D, benefits costs; 2E, supplies costs; and 2F other direct costs. Graphs derived as postestimation linear margins following mixed-effects regression, including 95% confidence intervals around the margin. Fixed effects included year, teaching status, an interaction between continuous time and teaching status, and case mix index, as well as indicators for hospital ownership and Health Service Area; a random intercept was included for each facility to adjust for the longitudinal analysis. Models used complete case analysis with no attempt to impute missing values; the sample size for all 6 graphs was 2,992 observations from 337 unique facilities. Abbreviations: FY indicates fiscal year; USD, U.S. dollars.

Temporal trends in wage subcomponents

The slower growth in OR wages per minute at teaching hospitals was achieved by less growth in the number of FTEs per 10,000 minutes of surgery (−0.002 FTE/10,000 minutes/year, P = .001, Figure 3A) despite larger growth in average hourly pay ($0.26/hour/year, P = .008, Figure 3B). Over the study time period, teaching hospitals increased average hourly pay by $2.38 (95% CI, $0.63–$4.14), or 18% more than at nonteaching hospitals.

Figure 3
Figure 3:
Trends (FY 2005–FY 2014) in wage growth between teaching and nonteaching hospitals, from a comparative study of risk-adjusted operating room costs in California, 2005–2014. Graphs derived as postestimation linear margins following mixed-effects regression, including 95% confidence intervals around the margin. Fixed effects included year, teaching status, an interaction between continuous time and teaching status, and case mix index, as well as indicators for hospital ownership and Health Service Area; a random intercept was included for each facility to adjust for the longitudinal analysis. Models used complete case analysis with no attempt to impute missing values; the sample size for all graphs was 2,992 observations from 337 unique facilities. Figure 3A: Hourly rate is a weighted average of hourly remuneration for all staff in the surgery and recovery cost center; for example, if 50% of staff were paid $30/hour and the other 50% were paid $60/hour, the weighted average would be $45/hour. Figure 3B: The number of FTEs is generated by dividing the number of all productive and nonproductive hours by 2,080. Nonproductive hours included paid time off. The number of FTEs was then divided by the number of surgical minutes and multiplied by 10,000 to generate the number of FTEs per 10,000 minutes of surgery. Abbreviations: FY indicates fiscal year; USD, U.S. dollars; FTE, full-time equivalent.

The growth in the number of FTEs per 10,000 minutes of surgery was slower across multiple employment categories (management, technicians, registered nurses, and aides) with no identifiable difference in growth for clerical staff between teaching and nonteaching hospitals (see Supplemental Digital Appendix 3, available at http://links.lww.com/ACADMED/A702).

Temporal trends in surgical volumes between teaching and nonteaching hospitals

Across California, surgical volume increased at teaching hospitals between 2005 and 2014, both in terms of number of procedures per year (roughly 276,000 to 304,000) and number of minutes of OR time (34.7 million in FY 2005 to 41.4 million in FY 2014). At nonteaching hospitals, surgical volume decreased (procedures: 1.45 million to 1.19 million; minutes: 147 million to 130 million) (Figure 4).

Figure 4
Figure 4:
Trends (FY 2005–FY 2014) in state-level surgical volume between teaching and nonteaching hospitals, from a comparative study of risk-adjusted operating room costs in California, 2005–2014. Values represent the sum of all teaching and nonteaching hospitals in a given year after winsorization at the hospital level. Lines represent the best linear trend. Sample sizes vary by year. Abbreviations: FY indicates fiscal year; OR, operating room.

Sensitivity analyses

Neither of the sensitivity analyses substantively changed the results (see Supplemental Digital Appendixes 4 and 5, available at http://links.lww.com/ACADMED/A702).

Discussion

Our findings show that of the institutions we analyzed, in 2014, California’s teaching hospitals spent about 20% less per minute to operate their inpatient ORs than their nonteaching counterparts. A decade earlier, OR costs per minute were similar at teaching and nonteaching institutions.

This study demonstrates that teaching hospitals in California have gained a cost advantage in the OR through slower growth in per-minute wages and indirect costs. The teaching hospitals in this study were able to complete more surgery with the same staff than their nonteaching counterparts. A potential explanation for this increase in relative productivity is increasing growth in wages at teaching hospitals that may reflect a more highly trained staff. The gains identified in indirect costs likely resulted from economies of scale, with teaching hospitals diluting fixed costs over more procedures and more operative minutes. Whether these changes were intentional is unclear and may reflect either gains in productivity at teaching hospitals or losses at nonteaching hospitals. Nonetheless, it appears that across the study period, teaching hospitals in California were outcompeting nonteaching hospitals with respect to surgical volume. Hospitals that lose volume face an unenviable choice of sacrificing their bottom line or slashing costs—with the largest driver of costs being labor. Loss of volume may explain, in part, why unaffiliated and rural hospitals are struggling to maintain financial solvency.18

Reducing cost and improving quality are the ultimate goals of value-based payment programs. Two examples relevant to surgery are accountable care organizations (ACOs) and bundled payments. Under the ACO model, providers and/or hospitals voluntarily agree to coordinate care for a population of assigned patients. If, over the course of the year, payments are lower for the population than expected, the ACO receives a portion of the savings. The cost of surgical care is included in this equation.19 Bundled payments provide a fixed reimbursement for the perioperative and post–acute care period. Ultimately, both payment policies capitate spending, either for a population (ACO) or an individual surgical episode (bundled payment). To compete financially under these models, hospitals must, therefore, lower costs. Contrary to the existing belief that teaching hospitals are more costly—and therefore may be disadvantaged in these systems—our findings suggest the opposite. Combined with evidence that the quality of health care is better at teaching hospitals, these sites may fare better than has been expected under value-based systems.20,21

One argument for why teaching hospitals have better surgical outcomes is the positive association between higher surgical volume and improved quality.22,23 Our findings suggest that a similar albeit inverse relationship may exist between higher volume and lower cost. While the economic benefits of greater productivity are clear, the clinical implications have not been evaluated. In fact, data on staffing and quality of care in the OR are scant. One cross-sectional retrospective study found no relationship between total personnel or nursing personnel and surgical complications.24 Future research should examine the clinical effects as well as the costs of different OR staffing models.

This study has several limitations. First, our risk adjustment methods may not have fully accounted for differences between teaching and nonteaching hospitals, such as unmeasured patient complexity. As described, CMI is an aggregate measure of complexity for all inpatients, not just surgical patients. However, the risk adjustment we did include increased the adjusted difference in cost between teaching and nonteaching hospitals. Therefore, additional granularity with respect to patient complexity may make the difference more pronounced. Second, while this does not undermine the validity of our analysis, our estimates do not capture the full cost of surgical episodes, omitting anesthesia, room and board, billable supplies, radiology, pathology, and postacute care. With a push toward shorter lengths of stay, the fraction of spending dedicated to the OR will likely increase, making this an even more important target for cost reduction. Finally, data were only available for 1 state, and the reduced costs identified in teaching hospitals in this state may not generalize to other states or the nation at large. Further, because of exemptions, some hospital types (e.g., Kaiser Permanente and Shriners hospitals) were not included in this analysis, and therefore our findings may not apply to these unique institutions. Nonetheless, California is a large and diverse state, and the OSHPD resource is unique such that few, if any, other datasets would permit us to examine this question.

Conclusions

Teaching hospitals in California spent over 20% less per minute to run their ORs than nonteaching hospitals in FY 2014, a difference that was achieved over the past decade through slower growth in wages and indirect costs. This cost difference appears to reflect both relative gains in productivity and a statewide shift of volume favoring teaching hospitals over nonteaching hospitals. The clinical and fiscal implications of shifting surgery from nonteaching to teaching hospitals warrant further evaluation, particularly for nonteaching hospitals and the patients who may have difficulty accessing care at teaching hospitals.

References

1. Centers for Medicare & Medicaid Services. National health expenditure fact sheet. https://www.cms.gov/research-statistics-data-and-systems/statistics-trends-and-reports/nationalhealthexpenddata/nhe-fact-sheet.html. Accessed June 3, 2019.
2. Muñoz E, Muñoz W 3rd, Wise L. National and surgical health care expenditures, 2005–2025. Ann Surg. 2010;251:195–200.
3. Stey AM, Brook RH, Needleman J, et al. Hospital costs by cost center of inpatient hospitalization for Medicare patients undergoing major abdominal surgery. J Am Coll Surg. 2015;220:207–217.e211.
4. Childers CP, Showen A, Nuckols T, Maggard-Gibbons M. Interventions to reduce intraoperative costs: A systematic review. Ann Surg. 2018;268:48–57.
5. Childers CP, Maggard-Gibbons M. Understanding costs of care in the operating room. JAMA Surg. 2018;153:e176233.
6. Sloan FA, Valvona J. Uncovering the high costs of teaching hospitals. Health Aff (Millwood). 1986;5:68–85.
7. Mechanic R, Coleman K, Dobson A. Teaching hospital costs: Implications for academic missions in a competitive market. JAMA. 1998;280:1015–1019.
8. Koenig L, Dobson A, Ho S, Siegel JM, Blumenthal D, Weissman JS. Estimating the mission-related costs of teaching hospitals. Health Aff (Millwood). 2003;22:112–122.
9. Newhouse JP. Accounting for teaching hospitals’ higher costs and what to do about them. Health Aff (Millwood). 2003;22:126–129.
10. Chen Q, Bagante F, Merath K, et al. Hospital teaching status and Medicare expenditures for hepato-pancreato-biliary surgery. World J Surg. 2018;42:2969–2979.
11. Pradarelli JC, Scally CP, Nathan H, Thumma JR, Dimick JB. Hospital teaching status and Medicare expenditures for complex surgery. Ann Surg. 2017;265:502–513.
12. Bridges M, Diamond DL. The financial impact of teaching surgical residents in the operating room. Am J Surg. 1999;177:28–32.
13. Williams JR, Matthews MC, Hassan M. Cost differences between academic and nonacademic hospitals: A case study of surgical procedures. Hosp Top. 2007;85:3–10.
14. Nathan H, Dimick JB. Medicare’s shift to mandatory alternative payment models: Why surgeons should care. JAMA Surg. 2017;152:125–126.
15. CHHS Open Data. State of California. Office of Statewide Health Planning and Development. Hospital annual financial data. Selected data file documentation. https://data.chhs.ca.gov/dataset/hospital-annual-financial-data-selected-data-pivot-tables/resource/842c8bcb-bfac-4aa9-9de8-b11eb049f6de. Published 2018. Accessed June 3, 2019.
16. Steinwald B, Dummit LA. Hospital case-mix change: Sicker patients or DRG creep? Health Aff (Millwood). 1989;8:35–47.
17. Office of Statewide Health Planning and Development. Submit financial data (SIERA). https://oshpd.ca.gov/data-and-reports/submit-data/financial-reporting/#hospital-manual. Accessed June 12, 2019.
18. Bai G, Anderson GF. A more detailed understanding of factors associated with hospital profitability. Health Aff (Millwood). 2016;35:889–897.
19. Hawken SR, Ryan AM, Miller DC. Surgery and Medicare shared savings program accountable care organizations. JAMA Surg. 2016;151:5–6.
20. Kupersmith J. Quality of care in teaching hospitals: A literature review. Acad Med. 2005;80:458–466.
21. Burke LG, Frakt AB, Khullar D, Orav EJ, Jha AK. Association between teaching status and mortality in US hospitals. JAMA. 2017;317:2105–2113.
22. Urbach DR. Pledging to eliminate low-volume surgery. N Engl J Med. 2015;373:1388–1390.
23. Finks JF, Osborne NH, Birkmeyer JD. Trends in hospital volume and operative mortality for high-risk surgery. N Engl J Med. 2011;364:2128–2137.
24. Talsma A, Anderson C, Geun H, Guo Y, Campbell DA. Evaluation of OR staffing and postoperative patient outcomes. AORN J. 2013;97:230–242.

Supplemental Digital Content

Copyright © 2019 by the Association of American Medical Colleges