INTRODUCTION

The aging baby boomer generation presents a unique financial challenge for the US healthcare system, as the Centers for Medicare and Medicaid Services (CMS) project healthcare spending to grow 5.5% per year from 2018-2027, reaching $60 trillion dollars and 19.4% of the United States’ GDP by 2027.1 Growing expenditures for increasingly utilized, function-restoring orthopedic surgeries like upper extremity arthroplasties and other major joint/limb reattachment procedures of upper extremities (Diagnosis-related group 483, DRG-483; See Supplement Table 1, Supplemental Digital Content 1, https://links.lww.com/COP/A75) demonstrate increasing general US healthcare costs.1–3 The average costs for total shoulder arthroplasties and total elbow arthroplasties increased annually by $867 and $1534 respectively between 1993 and 2007, with hospital charges demonstrated to increase even when adjusted for inflation.2 With increasing healthcare costs and the maturing baby boomer population projected to increase the Medicare enrollment to 20% of the United States population by 2029, initiatives to reduce individual and systemic costs become the focus of federal, state, and local policy-makers alike.

In an effort to reduce healthcare costs without compromising quality, it is critical to determine what hospital factors are associated with higher quality of care, and at what cost. Previous research demonstrates that lower extremity total joint arthroplasties performed at orthopedic specialty hospitals (OSH) are executed at lower surgical costs, readmission rates, and complication rates than non-specialty hospitals (NSH).4 Additionally, other studies suggest that higher surgical volume is associated with better patient outcomes and a lower cost.5,6 Where there appears to be a lack of research is the influence of hospital neighborhood population density and the volume of thoroughfare on surgical charges and Medicare payments for orthopedic procedures. This study seeks to determine the influence of volume, hospital demographics, and orthopedic specialization on the cost of DRG-483 procedures.

MATERIALS AND METHODS

Data were extracted from the Centers for Medicare and Medicaid Services (CMS) Medicare Provider Utilization and Payment Dataset for the years 2014 through 2018.7–11 These datasets contain finalized hospital-specific charges after claim adjustments for more than 3,000 U.S. hospitals that receive Medicare Inpatient Prospective Payment System (IPPS) payments based on a rate per discharge using the Medicare Severity Diagnosis Related Group. The CMS groups related diagnoses into Diagnosis Related Groups (DRGs) as part of a prospective payment framework designed to promote quality of care and responsible utilization of resources.12 In order to preserve CMS beneficiary anonymity, the datasets excluded low volume centers enrolled in the Medicare Inpatient Prospective Payment System (IPPS) with less than 10 discharges in a given Diagnosis-Related Group code (DRG) in which patient’s identity or protected health information could possibly be deduced secondary to diminutive representation. This also serves a secondary purpose in this study by eliminating institutions whose low volume could represent gross outliers.13 The relevant data extracted from each year’s dataset included basic hospital identification information (e.g. facility name and address) and all DRG-codes with their respective total discharges, average Medicare payments (AMP), average total payments (ATP), and average covered charges (ACC). Using a similar methodology to Bockhorn et al., DRG-specific discharge volume was utilized to assign each hospital an Orthopedic Specialty Ratio (OSR), defined as the ratio of annual musculoskeletal discharges (DRG 453-565) to total annual discharges (DRG 001-989).4 Hospitals were then divided into either Orthopedic Specialty Hospital (OSH = OSR > 0.99) or Non-Specialty Hospital (NSH = OSR ≤ 0.99) groups to allow for binary comparison. The cutoff for OSH was determined after individual hospital evaluation to include exclusively orthopedic specialty hospitals.

After categorizing the hospitals into OSH or NSH groups, all hospitals performing DRG-483 procedures were isolated within each year’s dataset. In 2014, DRG-483 (major joint/limb reattachment procedure of upper extremities) and DRG-484 (major joint & limb Reattachment procedure of upper extremity w/o CC/MCC) existed as distinct DRG codes. In order to include data from 2014, these two DRG codes were pooled for comparison with subsequent years. This was accomplished by summing the total number of discharges and calculating the weighted averages of the CMS cost metrics for each hospital that performed procedures coded as DRG-483, or DRG-484, or both. After 2014, CMS stopped reporting both codes separately, and began reporting both procedures under DRG-483. Hospitals that performed 10 or more of these procedures were queried for comparison. Annual summary statistics stratified by hospital specialty classification can be visualized in Table 1.

TABLE 1 - Hospital characteristics Variable 2014 2015 2016 2017 2018 Orthopedic Specialty Hospitals (OSR > 0.99)  Number of hospitals 64 69 78 75 90  All discharges 613.1 ± 213.3 598.6 ± 202.3 605.4 ± 187.0 649.0 ± 201.0 593.7 ± 171.6  MSK discharges 612.4 ± 212.1 598.3 ± 201.8 605.2 ± 186.6 648.8 ± 200.7 593.4 ± 171.2  DRG-483 discharges 46.0 ± 11.1 47.0 ± 10.5 52.4 ± 11.5 57.4 ± 12.0 60.9 ± 13.0  Average Medicare Payments $11696 ± 697 $12211 ± 402 $12328 ± 558 $12585 ± 448 $12494 ± 479  Average Total Payments $13802 ± 692 $14376 ± 443 $14654 ± 555 $15038 ± 488 $15158 ± 571  Average Covered Charges $50069 ± 4253 $52910 ± 4213 $54957 ± 4625 $57815 ± 5011 $59739 ± 5011 Nonspecialty Hospitals (Mean OSR = 0.18)  Number of hospitals 826 1150 1239 1279 1304  All discharges 4705.6 ± 245.4 4280.6 ± 198.3 4068.5 ± 186.6 4106.1 ± 185.1 4016.6 ± 183.0  MSK discharges 713.3 ± 34.8 624.2 ± 28.3 619.4 ± 28.0 616.1 ± 27.1 578.2 ± 26.2  DRG-483 discharges 36.0 ± 2.1 34.2 ± 1.7 38.0 ± 1.9 39.8 ± 2.0 42.5 ± 2.1  Average Medicare Payments $13904 ± 217 $14400 ± 194 $14372 ± 183 $14555 ± 183 $14821 ±184  Average Total Payments $16289 ± 240 $16847 ± 209 $16831 ± 192 $17079 ± 196 $17512 ± 200  Average Covered Charges $64312 ± 1894 $68631 ± 1890 $71582 ± 1946 $73570 ± 2010 $76831 ± 2135

DRG-483, Major joint/limb reattachment procedures of upper extremities; MSK, musculoskeletal; OSR, Orthopedic Specialty Ratio.

The cost metrics recorded in the Medicare Inpatient Prospective Payment System (IPPS) include Average Covered Charges (ACC: the hospital’s average charge for services covered by Medicare), Average Total Payments (ATP: payment to hospitals including teaching, disproportionate share, capital, outlier payments, co-payment, deductible amounts, and additional third-party payments), and Average Medicare Payments (AMP: average amount that Medicare pays to the hospital for Medicare’s share of the MS-DRG).13 AMP is the most direct measure of DRG-483 cost to society and the focus of the financial analysis in this study.

To characterize the degree of Urban/Rural environments for each facility, hospital zip codes were converted to USDA Rural-Urban Commuting Area codes (RUCA) which are based on population data, commuting flows, and bureau shapefiles from the 2010 US census as well as Esri’s ArcMap software.14 The primary RUCA codes 1 through 10 are categorically subdivided into four groups: metropolitan (1-3; urbanized area core with surrounding areas of low and high primary flow), micropolitan (4-6; large urban clusters of population size 10,000-49,999 with surrounding areas of low and high primary flow), small town (7-9; small urban clusters of population size 2,500-9,999 with surrounding areas of high and low primary flow), and rural (10; rural areas with primary flow to a tract outside an urban area or urban cluster).

Statistical analyses were performed using open-source R statistical software.15 Three linear regression models were fit for AMP, ATP, and ACC treated as continuous dependent variables. The independent variables included in the models were OSH, sum of DRG-483 discharges, sum of all surgical discharges, facility location (RUCA), and year. Deviations from the assumptions of linear regression were examined. A positively skewed distribution with non-normality and heteroscedastic residual errors were observed and tolerated because they were felt to represent true outliers rather than systematic error in the data.

RESULTS

The n=64-90 hospitals classified as OSH (Table 1) performed 20,051 DRG-483 procedures between 2014-2018, with an annual average of 52.7 procedures per hospital in the five years studied. The n=826-1304 NSH (Table 1) performed the remaining 83.5% of cases at an annual average of 38.1 procedures per hospital. The percentage difference in AMP per surgery between NSH and OSH in 2014, 2015, 2016, 2017, and 2018 was 15.9%, 15.2%, 14.2%, 13.5%, 15.7% respectively with a cumulative average difference of 15% over the years evaluated (Fig. 1). The cumulative average percentage differences per surgery for ATP and ACC were 13.6% and 22.5% respectively.

FIGURE 1:

Combined NSH vs. OSH DRG-483 Average Medicare Payment.

The results of the linear regression models are summarized in Table 2.

TABLE 2 - Regression results for cost and payment analyses Outcome Models Beta SE T P Model R2 Average Medicare Payment Constant $13,104 $127 102.91 <0.001 0.0629 OSH -$1,413 $183 −7.712 <0.001 DRG-483 Discharges -$3 $1 −2.29 <0.05 Total Surgical Discharges $0 $0 12.155 <0.001 Micropolitan $440 $149 2.959 <0.01 Small Town $2,961 $328 9.023 <0.001 2015 $557 $142 3.918 <0.001 2016 $576 $140 4.1 <0.001 2017 $759 $140 5.434 <0.001 2018 $1,024 $139 7.348 <0.001 Average Total Payment Constant $15,317 $137 112.009 <0.001 0.0712 OSH -$1,525 $197 −7.748 <0.001 DRG-483 Discharges $0 $1 −0.319 >0.1 Total Surgical Discharges $0 $0 13.144 <0.001 Micropolitan -$39 $160 −0.245 >0.1 Small Town $2,317 $352 6.575 <0.001 2015 $634 $153 4.152 <0.001 2016 $678 $151 4.496 <0.001 2017 $923 $150 6.148 <0.001 2018 $1,362 $150 9.096 <0.001 Average Covered Charges Constant $68,427 $1,344 50.932 <0.001 0.0522 OSH -$16,224 $1,934 −8.39 <0.001 DRG-483 Discharges -$56 $14 −4.112 <0.001 Total Surgical Discharges $0 $0 −0.231 >0.1 Micropolitan -$16,037 $1,568 −10.227 <0.001 Small Town -$28,524 $3,462 −8.239 <0.001 2015 $4,308 $1,501 2.87 <0.01 2016 $7,434 $1,481 5.018 <0.001 2017 $9,757 $1,474 6.617 <0.001 2018 $13,151 $1,471 8.941 <0.001

OSH, Orthopedic specialty hospital; SE, standard error.

The AMP model revealed that OSH classification resulted in an average of 10.8% reduction from the constant AMP reference of $13,103.92. The addition of DRG-483 discharge volume, total surgical discharges, facility location, and year as independent variables strengthened the total percentage of the variance accounted for in all three models. For each additional DRG-483 discharge, AMP was decreased by $2.95 (P<0.05), indicating that higher DRG-483 volume centers performed the procedures at a functionally insignificant lower average cost to CMS. By means of comparison, the average AMP difference between facilities that annually perform 38 cases and 52 cases (the average discharge volumes of NSH and OSH respectively) is predicted to be $41.30. The relationship between discharge volume and AMP can be visualized in Fig. 2. Conversely, the grand total number of surgical discharges (includes all DRG codes) of the studied hospitals accounted for a $0.17 (~$0 as represented in Table 2) increase in modeled DRG-483 AMP per discharge (P<0.001), with considerably increased costs to CMS in hospitals with high annual general surgical volume. When included in the model, the total number of orthopedic discharges displayed significant multicollinearity to other variables with a variance inflation factor of 4.77. For this reason, the orthopedic discharges variable was excluded from the optimized final linear model.

FIGURE 2:

Regression for Discharge Volume on DRG-483.

With regards to the influence of facility environment/population demographics, linear regression indicated a 3.36% difference in AMP between hospitals located in metropolitan and micropolitan centers (P<0.01) as well as a 22.6% AMP increase in small towns with populations less than 10,000 (P<0.001). Lastly, the models demonstrate a gradual increase in all three cost metrics in the five years studied with total increases from 2014-2018 in AMP, ATP, and ACC of 7.82% ($1,024), 8.9% ($1,362), 19.22% ($13,151) respectively (all P<0.001) (Table 2).

DISCUSSION

The purpose of this study was to investigate the effects of OSH, surgical volume, and hospital location on DRG-483 procedure costs. Our analysis demonstrated that OSH is associated with a higher surgical volume and lower costs, while higher DRG-483 operative volume itself was associated with only modestly lower costs. Finally, population density of hospital location was inversely proportional to DRG-483 AMP, with metropolitan centers performing the procedure at the lowest cost to the CMS.

These findings suggest that CMS payments are optimized at high arthroplasty volume, urban, orthopedic specialty hospitals. The inverse relationship between orthopedic specialization and cost to CMS was evident in the multivariable model and in isolation without any evidence of confounding or effect modification, suggesting this independent variable’s importance. In other words, it is not just because OSH operate at greater annual volume of DRG-483 procedures or are more likely to be located in urban centers that they are more cost effective; other factors related to the episode of care, such as length of stay e.g. may be playing a role. A better understanding of the OSH practices and processes that confer savings should be sought and targeted for further research.

The exact OSH practices that produce these striking decreases in cost remain elusive, but studies have suggested that several advantages of OSH exist in their system engineering and treatment culture. Higher average nurse-to-patient ratios, shorter operating room turnovers, newer equipment, and talent-attracting work environments have been found to promote both a culture of efficiency and cooperation between providers and allied-health professionals in OSH.16,17 In comparing its OSH to its tertiary referral center, the Rothman Orthopaedic Institute at Thomas Jefferson University Hospital identified that surgical time (skin incision to closure) was comparable for matched cohorts at the care centers, while anesthesia preparation time (patient in room to incision) and conclusion time (skin closed to patient out of room) were observed to be significantly longer at the tertiary referral center.18 Elimination of superfluous operating room time is a key cost-reduction strategy implemented in OSH that could account for decreased average surgical expenditure.19–23 Additionally, smaller inpatient census and comparatively less hierarchy fosters consistent management protocols and continuity of care that mitigates inpatient complication incidence in the critical early postoperative period. The value of specialization is economically clear in the institutional cost reduction at OSH but extends to NSH via healthy market competition, demonstrating a 4.5% reduction in NSH patient care costs with a competing specialty hospital in their market.24 Distilling the practice differences between OSH and NSH that yield the demonstrated cost reductions is an important endeavor for actionable system engineering.

Prevalence and costs for common orthopedic procedures have been increasing in the past several decades.1 The cost and utilization of upper extremity arthroplasty procedures have followed this trend, leading to an increased burden on the CMS. Hospital charges for common procedures like total hip arthroplasty and total knee arthroplasty have also been increasing.13 In the context of this rising economic burden, emulation of the aforementioned OSH practices represents a possible cost-saving measure.

In past literature, centers with higher arthroplasty surgical volume have been associated with lower costs and better outcomes.25 However, other studies have indicated that in some procedures, like revision arthroplasties, high volume centers have higher costs.26 The aforementioned study by Frisch et al. also indicates that high volume centers are less likely to be high-cost outliers and that increased Medicare reimbursements in these centers may be linked to greater operative complexity.4 These studies are weakly corroborated by the findings of our study, with increased DRG-483 surgical volume associated with modestly decreased AMP and total hospital surgical volume conversely associated with increased AMP. We speculate that higher volume centers may have increased costs to either offset financial losses associated with other procedures, urban indigent care, surgical patients with more comorbidities, and higher volume emergency centers or as an incentive for providing care to a high proportion of CMS patients. Supporting the latter speculation, studies demonstrate that government-owned hospitals, whose patient population are predominantly government-insured, are associated with more robust reimbursements.27 Our results, in the context of conflicting literature, indicate that higher upper extremity arthroplasty volume translates to clinically insignificant decreased costs to the payer, indicating that NSH should not focus on increasing their upper extremity arthroplasty volume to yield cost reductions.

Our study evidences that upper extremity procedures are performed at higher costs in small towns compared to micropolitan and metropolitan centers. Since hospitals have reimbursements adjusted based on the proportion of low-income patients and CMS-insured patients, the higher costs experienced in small towns may simply reflect lower-income areas and Medicare reimbursement policy. This is substantiated by the well-documented phenomenon of medical tourism, in which high-income residents of rural areas travel to urban hospitals for access to specialists and medical resources not available in their home community, decreasing the average socioeconomic status of the population served in the small town hospitals.28 While the relatively new operative technique of the reverse total shoulder arthroplasty has allowed orthopedic generalists to offer their patients pain-relieving and function-restoring surgeries regardless of rotator cuff integrity, this is likely only minimally offsetting medical tourism to urban medical centers where board-certified shoulder and elbow surgeons perform more technically challenging operations including anatomic shoulder arthroplasties.

Like many studies using data from large national databases, the large sample size in our analyses possesses great statistical power and generalizability. However, it is also limited by the multilevel analytical approach that utilizes aggregate data (hospital-level) and the inferential fallacies that can result. Without the ability to quantify and control for individual patient-level characteristics that may influence hospital costs, including illness severity, comorbidities, and mortality probability, residual confounding can occur. It is possible that there is a negligible difference in patient illness severity and comorbidities, similar to a finding by Blumenthal et. al.29 Moreover, while serving a useful function in the CMS prospective payment framework, DRG-483 is not exclusive to just upper extremity arthroplasties and hemiarthroplasties (commonly shoulder and elbow) and includes a wide variety of upper extremity replantations, interpositional arthroplasties, and sternoclavicular/acromioclavicular procedures (Supplement Table 1, Supplemental Digital Content 1, https://links.lww.com/COP/A75). While shoulder and elbow arthroplasties may be disproportionately represented within this DRG secondary to increasing demand and utilization, this aggregation of procedures inherently limits its generalizability to exclusively shoulder and elbow arthroplasties and direct conclusions cannot be made in isolation of the other procedures represented. Furthermore, while the CMS data used presents a useful picture of surgical volume and associated costs across the country at the hospital-level, it is bereft of complication data for the procedure in question. Data on complication and readmission rates has been published by the CMS for hip and knee arthroplasties but has yet to be made public for the increasingly utilized upper extremity arthroplasties. Volume and payment data are still critical components to understanding cost effectiveness of OSH, but complication data as a measure of quality would provide a key missing element in assessing and comparing value of care provided by hospitals. The addition of quality metrics data to this national database would contribute significantly to studies in this area and represents a potential avenue in furthering our understanding of the relationship between orthopedic specialization and expenditures.

CONCLUSION

It is critical in the current financial state of the US healthcare system to eliminate unnecessary expenditures and invest in cost-reduction strategies. With OSH performing DRG-483 procedures at a lower cost compared to NSH in all years studied, this research indicates that one way of potentially reducing cost is the emulation of OSH practices. Eliminating the AMP difference for the 222,519 DRG-483 procedures performed at NSH in the years studied would amount to over $480 million in CMS savings. Our findings indicate that this cost saving is not confounded by increased DRG-483 operative volume at OSH, but explained by other factors intrinsic to OSH operations. These findings add to mounting evidence that OSH practices represent a potential avenue to maximize CMS resources.

ACKNOWLEDGMENTS

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

REFERENCES 1. Sisko AM, Keehan SP, Poisal JA, et al. National Health Expenditure Projections, 2018–27: economic and demographic trends drive spending and enrollment growth. Health Aff. 2019; 38:491–501. 2. Day JS, Lau E, Ong KL, et al. Prevalence and projections of total shoulder and elbow arthroplasty in the United States to 2015. J Shoulder Elbow Surg. 2010; 19:1115–1120. 3. Kim SH, Wise BL, Zhang Y, et al. Increasing incidence of shoulder arthroplasty in the United States. Bone Joint Surg. 2011; 93:2249–2254. 4. Bockhorn LN, Goytia RN, Laughlin MS, et al. Increased orthopedic specialization lowers costs and improves outcomes in total joint arthroplasty. Arthroplasty. 2019;34:S76–S79. 5. Chowdhury MM, Dagash H, Pierro A. A systematic review of the impact of volume of surgery and specialization on patient outcome. British Surg. 2007;94:145–161. 6. Wilson S, Marx RG, Pan TJ, et al. Meaningful thresholds for the volume-outcome relationship in total knee arthroplasty. Bone Joint S - Am. 2016; 98:1683–1690. 7. Inpatient Charge Data FY 2014. Guidance Portal. Accessed 2020. https://www.hhs.gov/guidance 8. Inpatient Charge Data FY 2015. Guidance Portal. Accessed 2020. https://www.hhs.gov/guidance 9. Inpatient Charge Data FY 2016. Guidance Portal. Accessed 2020. https://www.hhs.gov/guidance 10. Inpatient Charge Data FY 2017. Guidance Portal. Accessed 2020. https://www.hhs.gov/guidance 11. Inpatient Charge Data FY 2018. Guidance Portal. Accessed 2021. https://www.hhs.gov/guidance 12. Design and development of the Diagnosis Related Group (DRG). Center for Medicare and Medicaid Services; October 2019. Published at CMS.gov. 13. Medicare Fee-For Service Provider Utilization and Payment Data Inpatient Public Use File: A Methodological Overview. Center for Medicare and Medicaid Services; 2019. Published at CMS.gov. 14. Rural-Urban Commuting Area Codes. Economic Research Service, U.S. Department of Agriculture; 2010. Published at ers.usda.gov. 15. Team RC. R: A Language and Environment for Statistical Computing. 2019. 16. Badlani N, Boden S, Phillips F. Orthopedic specialty hospitals: centers of excellence or greed machines? Orthopedics. 2012;35:e420–e425. 17. Kovner C, Jones C, Zhan C, et al. Nurse Staffing and postsurgical adverse events: an analysis of administrative data from a sample of U.S. hospitals, 1990–1996. Health Ser Res. 2002;37:611–629. 18. Padegimas EM, Hendy BA, Chan WW, et al. The effect of an orthopedic specialty hospital on operating room efficiency in shoulder arthroplasty. J Shoulder Elbow Surg. 2019;28:15–21. 19. Diab MA, Fernandez GN, Elsorafy K. Time and cost savings in arthroscopic subacromial decompression: the use of bipolar versus monopolar radiofrequency. Int Orthop. 2009; 33:175–179. 20. Li LT, Chuck C, Bokshan SL, et al. Cost comparison of open and arthroscopic treatment options for SLAP tears. Arthrosc Sports Med Rehabil. 2021; 3:e315–e322. 21. Beckerman D, Esparza M, Lee SI, et al. Cost Analysis of Single-Level Lumbar Fusions. Global Spine J. 2020;10:39–46. 22. Lu Y, Lavoie-Gagne O, Forlenza EM, et al. Duration of sare and operative time are the primary drivers of total charges after ambulatory hip arthroscopy: a machine learning analysis. Arthroscopy: The Journal of Arthroscopic & Related Surgery. 2022; 38:2204–2216.e3. 23. Gross CE, Scott D, Mather RC, et al. Effect of facility on operative costs of total ankle arthroplasties. J Surg Orthop Adv. 2018; 27:321–324. 24. Schneider JE, Ohsfeldt RL, Morrisey MA, et al. Effects of Specialty Hospitals on the Financial Performance of General Hospitals, 1997–2004. Inquiry. 2007;44:321–334. 25. Courtney PM, Frisch NB, Bohl DD, et al. Improving Value in Total Hip and Knee Arthroplasty: The Role of High Volume Hospitals. J Arthroplasty. 2018;33:1–5. 26. Frisch NB, Courtney PM, Darrith B, et al. Do higher-volume hospitals provide better value in revision hip and knee arthroplasty? Bone Joint J. 2017;99-B:1611–1617. 27. Padegimas EM, Verma K, Zmistowski B, et al. Medicare reimbursement for total joint arthroplasty: The driving forces. Bone Joint Surg - Am. 2016; 98:1007–1013. 28. Basu J, Mobley LR. Impact of Local Resources on hospitalization patterns of medicare beneficiaries and propensity to travel outside local markets. Rural Health. 2010; 26:20–29. 29. Blumenthal DM, Orav EJ, Jena AB, et al. Access, quality, and costs of care at physician owned hospitals in the United States: observational study. BMJ. 2015;351:h4466.