Bundled payments are a means of reimbursement in which a pre-specified lump sum payment is provided for an episode of care. Hospitals, physicians, and other providers then divide this payment on their own terms. This model is intended to reduce the high levels of spending attributed to fee-for-service reimbursement by incentivizing providers to cooperate to minimize the cost of care. The Centers for Medicare and Medicaid Services is experimenting with bundling arrangements.1,2 With the Bundled Payments for Care Improvement initiative, the Centers for Medicare and Medicaid Services launched 4 distinct bundling models encompassing a variety of time frames during hospitalization and after discharge.1 Total joint arthroplasty is the largest expense for a single condition among Medicare beneficiaries, constituting approximately 5.7% of annual Medicare expenditures.3 As a result, it has been identified as an attractive target for bundled payments, both within the Centers for Medicare and Medicaid Services demonstration projects and beyond.3–5 The Centers for Medicare and Medicaid Services launched mandatory bundled payments for total joint arthroplasty in 67 “metropolitan” regions under the Comprehensive Care for Joint Replacement program on April 1, 2016.
Various novel payment programs have been attempted, with a similar goal: incentivizing providers to reduce costs. However, many of these reimbursement plans, such as health management organizations, have been criticized for generating adverse incentives for providers to withhold care that may benefit patients. This risk is also possible with bundled payments, but theoretically, the temptation to discourage appropriate care can be avoided through risk adjustment.7–9 A 2012 review showed that slightly more than half of the early bundling initiatives across all fields of medicine and all payers featured some method of risk adjustment, but most of these were outside the specialty of orthopedics.4 The rest relied on “flat-rate” (unadjusted) bundled payments for a number of reasons, such as logistical ease for both payers and providers. The Medicare Bundled Payments for Care Improvement programs, which have been more relevant for orthopedics, include risk adjustment only if the participating health care organization proposes a risk stratification technique based on Medicare claims data, and this technique must be approved by the Centers for Medicare and Medicaid Services.10 The Comprehensive Care for Joint Replacement program involves a rudimentary risk adjustment plan that controls for patients with Major Complications/Comorbid Conditions (MCC) criteria (typically representing major complications) and patients treated for fractures, with the latter being much more relevant for total hip arthroplasty than for total knee arthroplasty (TKA). The MCC criteria include major medical complications, such as sepsis, myocardial infarction, and pulmonary embolus, but they also apply to some chronic conditions, such as end-stage renal disease.11
Traditionally, the Centers for Medicare and Medicaid Services has adjusted hospital payments based on patient and facility factors by calculating a Medicare Severity-Diagnosis Related Group weight for each hospitalization. This system provides increased reimbursement levels for complex patients who are typically more expensive to treat. For example, patients undergoing total joint arthroplasty who meet MCC criteria are coded under Medicare Severity-Diagnosis Related Group 469 rather than 470; the former carries a higher Medicare Severity-Diagnosis Related Group weight and therefore provides increased reimbursement.11 However, traditionally, surgeons have received relatively constant reimbursement levels for a given service, such as TKA (although modifiers occasionally elicit higher payments for particularly challenging cases). This fee-for-service system may result in some degree of “cherry-picking” because surgeons are not regularly paid more for cases that require more work. However, the system is largely resistant to financial incentives that affect patient selection because hospitals bear the brunt of excess costs associated with challenging patients, whereas surgeons are generally responsible for choosing surgical candidates and can expect to earn a set fee without the ramifications of expensive postoperative care. However, this level of protection may no longer be present with bundled payments because surgeons will become directly affected by the costs of intra- and postoperative care through gain sharing. This study was conducted to determine how bundled payments for TKA should be adjusted for risk.
Materials and Methods
All patients of Medicare-eligible age (≥65 years) undergoing elective unilateral primary TKA at an urban academic medical center over a 2-year period were identified retrospectively by International Classification of Diseases, Ninth Revision, procedure code 81.54 (n=941). Institutional review board approval was obtained. Variable costs (the sum of all marginal costs accrued in a patient's care) for each patient's operative hospitalization as well as any unplanned readmissions within 30 days of discharge were determined from the hospital's cost accounting database, which is amassed and maintained with Horizon Performance Manager software (McKesson Corp, San Francisco, California). Rather than relying on hospital charges and cost-to-charge ratios, this cost accounting system correlates all hospital expenditures with specific services provided, ultimately calculating the cost of each patient encounter as the sum of all associated outlays. Clinical and demographic factors, including age, American Society of Anesthesiologists (ASA) physical status classification, body mass index, sex, race, and MCC criteria were obtained from the university's clinical data warehouse. By considering surgical admissions and 30-day unplanned readmissions, the study replicated Medicare Bundled Payments for Care Improvement Model 4.1
Contribution margin, which is generally accepted as a gauge of an entity's financial incentives, is calculated as the difference between reimbursement and variable costs.12 Therefore, to keep the contribution margin equal between patients (ie, to avoid incentives to cherry-pick), reimbursement should be adjusted for each patient to match expected differences in variable cost. Thus, this study used variable cost as a proxy for the appropriate level of risk adjustment for each patient factor.
Statistical analysis was performed with SPSS version 23 software (IBM Corp, Armonk, New York). Nonparametric variables are reported as median and interquartile range. Parametric variables are reported as mean and standard deviation. Multivariate linear regression was performed in stepwise fashion to identify correlations with variable costs, with nonsignificant variables incrementally excluded. Statistical significance was set at alpha≤0.05.
Patients included in the study had an average age of 73 years and body mass index of 30.9 kg/m2. Most of the patients were assigned ASA class II (56%) or III (43%) and did not meet MCC criteria (93%). The population was predominantly female (69%) and white (65%) (Table 1). Median variable cost was $7345, with interquartile range of $6724 to $8655. Age, ASA class, and MCC criteria were correlated with variable cost and therefore were included in the final multivariate regression: (P=.003, P=.001, and P<.001, respectively). Regression coefficients of the multivariate analysis suggest that variable costs increase by $57 per year of life beyond 65 years, $729 per ASA class beyond I, and $3122 for patients with MCC criteria. Body mass index, sex, and race were not significantly associated with differences in costs (P=.107, P=.393, and P=.056–.891, respectively) (Table 2).
Patient Characteristics (n=941)
Correlations With Variable Cost
The field of orthopedics has been an early target for bundled payments, including the Medicare Bundled Payments for Care Improvement initiative,4,7,9,13 and the Centers for Medicare and Medicaid Services mandated compulsory total joint arthroplasty bundling under the Comprehensive Care for Joint Replacement program in certain markets as of April 2016.14 Bundling has great potential to align hospitals, surgeons, and other providers on the goal of cost containment, but gain sharing also carries the risk of cherry-picking because surgeons, who are typically responsible for patient selection, will be more directly affected by the costs of care. Careful risk adjustment should prevent this unintended consequence, but most orthopedic bundling programs have been unadjusted. Similarly, the Medicare Bundled Payments for Care Improvement initiative uses risk adjustment only if it is designed by the participating institution, and the Comprehensive Care for Joint Replacement program stipulates adjustments only for MCC criteria and for patients with traumatic injury, with the latter relevant primarily for total hip arthroplasty, not TKA.
The study findings suggest that the cost of acute inpatient care for Medicare patients undergoing TKA is associated with age, ASA class, and MCC criteria (typically representing major medical complications). If the current findings are generalizable, these patients should provide increased reimbursement by $57 per year for age greater than 65 years, $729 per ASA class above I, and $3122 for MCC criteria to prevent financial incentives to cherry-pick younger, healthier patients. The finding on MCC criteria supports the decision by the Centers for Medicare and Medicaid Services to use this data point for risk adjustment of Comprehensive Care for Joint Replacement bundled payments. The findings on age and ASA class are consistent with recent findings identifying these variables as important predictors of cost in bundled payments for arthroplasty.15,16
The current findings reflect only 1 bundling model allowed under the Medicare Bundled Payments for Care Improvement initiative. This model includes acute in-patient care (surgical hospitalization and any unplanned readmissions within 30 days). These findings are likely conservative compared with some other Medicare Bundled Payments for Care Improvement models and the Comprehensive Care for Joint Replacement program, which encompass more of the postdischarge care in the bundle. The cost of postdischarge care (eg, inpatient rehabilitation) has been estimated as 36% of total costs,7 and older or sicker patients likely require higher financial outlays during this period. These factors could make the financial incentive to avoid these patients in more comprehensive bundling models stronger than that found in this study. Future research on the costs associated with other bundling models will be valuable in planning adequate risk adjustment.
Another important direction for ongoing research is advancing the understanding of the factors that predict costs of care, preferably across multiple geographic regions. The coefficient of determination reported in the current study (R2) suggests that this model explains only 8% of the variation in costs between patients. Because much of the unexplained variation is likely the result of inefficiencies in delivery of care, it cannot be explained by patient factors. This inefficiency is the target of initiatives such as bundled payments that encourage providers to cut costs by streamlining processes and reducing waste. Still, by evaluating additional variables, such as specific comorbidities, smoking status, and socioeconomic proxies, future researchers (or potentially the Centers for Medicare and Medicaid Services) may be able to construct multivariate models to explain more of this variation. Ideally, these models would be updated in real time, or at least annually, to account for changes in costs, patient populations, and evidence-based care pathways. This type of rigorous risk adjustment would minimize the risk of adverse incentives against any specific patient subpopulations. However, the limiting factor will likely be accurate cost figures, which are not available to the Centers for Medicare and Medicaid Services through standard claims data and therefore are more easily studied by institutions and providers.
Even though MCC modifiers predict costs and will be used by the Centers for Medicare and Medicaid Services to adjust arthroplasty bundled payments under the Comprehensive Care for Joint Replacement program, this metric retrospectively identifies complications and therefore does not maximize the potential of bundled payments to improve quality of care. Bundling is intended to drive providers to avoid complications (and thereby costs) to capture as much profit as possible. This goal would be accomplished more effectively with true risk factors (ie, patient characteristics that are identified preoperatively) to adjust payments rather than adverse events identified postoperatively (eg, MCC criteria). When adjustments are made preoperatively, providers have a much stronger incentive to reduce complications and cut costs. For this reason, research is needed to determine additional criteria (beyond age and ASA class) that predict the cost of care. However, until such data are available to facilitate rigorous risk adjustment, the use of a well-tested marker of cost, such as MCC criteria, is wise to reduce the risk of cherry-picking.
Even when risk adjustment is based on true preoperative risk factors, it can be performed in several ways, at either the patient or the population level, and either prospectively or retrospectively.1,13 Patient-level adjustment involves varying the reimbursement amount on a case-by-case basis, depending on the individual patient's risk profile. If instituted correctly, this approach should protect against cherry-picking when performed either prospectively or retrospectively. Prospective plans generally require providers to assume more risk and therefore should allocate more profit to providers when costs are curbed successfully.
Risk adjustment for bundled payments in orthopedics generally has functioned at the population level.7,9,13 With this system, a provider's lump sum payment is adjusted according to the case mix of the overall patient population. Although this approach is logistically simpler than patient-level risk adjustment, it carries a greater risk of adverse incentives to withhold care from unprofitable patients. When performed retrospectively, overall reimbursement is adjusted at the end of a prespecified period (eg, 1 year in the Comprehensive Care for Joint Replacement program). If designed well, this type of system should assure providers that they will be compensated adequately for the care of expensive patients while simultaneously encouraging them to minimize costs. Prospective risk adjustment at the population level is more prone to adverse incentives, but it also may provide the greatest potential economic reward for providers who successfully reduce spending because providers assume more of the risk (and potential gain). The only safeguard against cherry-picking in this scenario is the knowledge that the case mix at the end of the adjustment cycle will likely determine the bundled payment rates for the next cycle. Therefore, cherry-picking will increase margins in the short term but could lead to lower reimbursement in the future.
This study had a number of limitations. It relied on administrative data, which are imperfect because coding errors and incomplete follow-up can affect results.17 Additionally, this study was performed at a single institution, limiting its generalizability. For example, there are well-documented geographic differences in health care costs.18,19 Also, although this study assessed several important patient factors, including some shown to predict complications in previous studies,20–25 these variables explained only a small percentage of variations in costs. As discussed earlier, future research assessing more patient factors will be essential in facilitating adequate risk adjustment. In addition, as discussed earlier, this study reflects only 1 model for bundling that does not incorporate the costs of postdischarge care other than unplanned readmissions, and these costs have been estimated to exceed one-third of total expenses.7 As a result, the current estimates for risk adjustment (eg, $57 per year for age ≥65 years) are likely conservative relative to other bundling models. Finally, cost data require assumptions and therefore may be imprecise. However, this problem was minimized by the use of a relatively rigorous cost accounting system based on actual expenses rather than charges and by the exclusion of fixed costs, which are more susceptible to assumptions.
The primary strength of this work is the use of actual cost figures. Earlier studies attempted to explain variations in Medicare reimbursement in patients undergoing TKA in the context of bundled payments.26,27 However, without assessing costs, which are poorly understood in health care in general, it is not possible to depict provider incentives accurately. Other methods besides risk adjustment have been considered and tested in bundled payment programs to decrease adverse incentives. These include required reinsurance, outlier payments, and gain and loss caps.5 However, although these approaches may make providers comfortable participating in bundling programs, only rigorous risk adjustment with preoperative patient factors can show the full potential of bundled payments to reduce costs and improve quality while avoiding adverse incentives.
The current findings suggest that providing TKA for older patients, those with higher ASA classifications, and those who meet MCC criteria results in increased hospital costs. If these results are generalizable, bundles encompassing all acute inpatient care without proper risk adjustment are likely to create adverse incentives against these populations. Moreover, the current findings likely underestimated this incentive for other bundling models that incorporate more postdischarge care, such as the Comprehensive Care for Joint Replacement program. A crucial direction for future research is to characterize the risk factors for high TKA costs by considering more patient factors and more postdischarge care than the current study evaluated. The Centers for Medicare and Medicaid Services can use this information to implement rigorous risk adjustment methods to maximize the potential benefits of bundled payments and minimize adverse incentives against any specific patient populations.
- Centers for Medicare & Medicaid Services. Bundled Payments for Care Improvement (BPCI) initiative: general information. http://innovation.cms.gov/initiatives/bundled-payments. Accessed November 28, 2014.
- Centers for Medicare & Medicaid Services. Medicare Acute Care Episode (ACE) demonstration. http://innovation.cms.gov/initiatives/ACE. Accessed November 28, 2014.
- Cutler DM, Ghosh K. The potential for cost savings through bundled episode payments. N Engl J Med. 2012; 366(12):1075–1077. doi:10.1056/NEJMp1113361 [CrossRef]
- Painter MW, Burns ME, Bailit MH. Bundled payments across the U.S. today: status of implementations and operational findings. Health Care Incentives Improvement Institute. http://www.hci3.org/wp-content/uploads/files/files/HCI-IssueBrief-4-2012.pdf. Accessed November 24, 2014.
- Sood N, Huckfeldt PJ, Escarce JJ, Grabowski DC, Newhouse JP. Medicare's bundled payment pilot for acute and postacute care: analysis and recommendations on where to begin. Health Aff (Millwood). 2011; 30(9):1708–1717. doi:10.1377/hlthaff.2010.0394 [CrossRef]
- Centers for Medicare & Medicaid Services. Comprehensive care for joint replacement model. https://innovation.cms.gov/initiatives/cjr. Accessed January 2, 2016.
- Bozic KJ, Ward L, Vail TP, Maze M. Bundled payments in total joint arthroplasty: targeting opportunities for quality improvement and cost reduction. Clin Orthop Relat Res. 2014; 472(1):188–193. doi:10.1007/s11999-013-3034-3 [CrossRef]
- Luft HS. Economic incentives to promote innovation in healthcare delivery. Clin Orthop Relat Res. 2009; 467(10):2497–2505. doi:10.1007/s11999-009-0930-7 [CrossRef]
- Rana AJ, Bozic KJ. Bundled payments in orthopaedics. Clin Orthop Relat Res. 2015; 473(2):422–425. doi:10.1007/s11999-014-3520-2 [CrossRef]
- Center for Medicare & Medicaid Innovation. Bundled payments for care improvement initiative request for application. http://innovation.cms.gov/Files/x/Bundled-Payments-for-Care-Improvement-Request-for-Applications.pdf. Accessed November 28, 2014.
- Centers for Medicare & Medicaid Services. FY 2015 final rule tables. https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/FY2015-IPPS-Final-Rule-Home-Page-Items/FY2015-Final-Rule-Tables.html. Accessed August 29, 2015.
- Eappen S, Lane BH, Rosenberg B, et al. Relationship between occurrence of surgical complications and hospital finances. JAMA. 2013; 309(15):1599–1606. doi:10.1001/jama.2013.2773 [CrossRef]
- Froimson MI, Rana A, White RE Jr, et al. Bundled payments for care improvement initiative: the next evolution of payment formulations. AAHKS Bundled Payment Task Force. J Arthroplasty. 2013; 28(suppl 8):157–165. doi:10.1016/j.arth.2013.07.012 [CrossRef]
- U.S. Department of Health & Human Services. CMS finalizes bundled payment initiative for hip and knee replacements. http://www.hhs.gov/about/news/2015/11/16/cms-finalizes-bundled-payment-initiative-hip-and-knee-replacements.html. Accessed December 6, 2015.
- SooHoo NF, Li Z, Chan V, Chenok K, Bozic KJ. The importance of risk adjustment in reporting total joint arthroplasty outcomes. J Arthroplasty. 2016; 31(3):590–595. doi:10.1016/j.arth.2015.09.041 [CrossRef]
- Clement RC, Kheir MM, Soo AE, Derman PB, Levin LS, Fleisher L. What financial incentives will be created by Medicare bundled payments for total hip arthroplasty?J Arthroplasty. Epub ahead of print.
- Bozic KJ, Chiu VW, Takemoto SK, et al. The validity of using administrative claims data in total joint arthroplasty outcomes research. J Arthroplasty. 2010; 25(suppl 6):58–61. doi:10.1016/j.arth.2010.04.006 [CrossRef]
- Goz V, Rane A, Abtahi AM, Lawrence BD, Brodke DS, Spiker WR. Geographic variations in the cost of spine surgery. Spine (Phila Pa 1976). 2015; 40(17):1380–1389. doi:10.1097/BRS.0000000000001022 [CrossRef]
- Hadley J, Reschovsky JD, O'Malley JA, Landon BE. Factors associated with geographic variation in cost per episode of care for three medical conditions. Health Econ Rev. 2014; 4:8. doi:10.1186/s13561-014-0008-4 [CrossRef]
- Belmont PJ Jr, Goodman GP, Waterman BR, Bader JO, Schoenfeld AJ. Thirty-day postoperative complications and mortality following total knee arthroplasty: incidence and risk factors among a national sample of 15,321 patients. J Bone Joint Surg Am. 2014; 96(1):20–26. doi:10.2106/JBJS.M.00018 [CrossRef]
- Soohoo NF, Zingmond DS, Lieberman JR, Ko CY. Primary total knee arthroplasty in California 1991 to 2001: does hospital volume affect outcomes?J Arthroplasty. 2006; 21(2):199–205. doi:10.1016/j.arth.2005.03.027 [CrossRef]
- White RH, Henderson MC. Risk factors for venous thromboembolism after total hip and knee replacement surgery. Curr Opin Pulm Med. 2002; 8(5):365–371. doi:10.1097/00063198-200209000-00004 [CrossRef]
- White RH, Romano PS, Zhou H, Rodrigo J, Bargar W. Incidence and time course of thromboembolic outcomes following total hip or knee arthroplasty. Arch Intern Med. 1998; 158(14):1525–1531. doi:10.1001/archinte.158.14.1525 [CrossRef]
- Mahomed NN, Barrett J, Katz JN, Baron JA, Wright J, Losina E. Epidemiology of total knee replacement in the United States Medicare population. J Bone Joint Surg Am. 2005; 87(6):1222–1228. doi:10.2106/JBJS.D.02546 [CrossRef]
- Vincent KR, Lee LW, Weng J, Alfano AP, Vincent HK. A preliminary examination of the CMS eligibility criteria in total-joint arthroplasty. Am J Phys Med Rehabil. 2006; 85(11):872–881. doi:10.1097/01.phm.0000242647.81882.5c [CrossRef]
- Cram P, Lu X, Li Y. Bundled payments for elective primary total knee arthroplasty: an analysis of Medicare administrative data. Geriatr Orthop Surg Rehabil. 2015; 6(1):3–10. doi:10.1177/2151458514559832 [CrossRef]
- Cram P, Ravi B, Vaughan-Sarrazin MS, Lu X, Li Y, Hawker G. What drives variation in episode-of-care payments for primary TKA? An analysis of Medicare administrative data. Clin Orthop Relat Res. 2015; 473(11):3337–3347. doi:10.1007/s11999-015-4445-0 [CrossRef]
Patient Characteristics (n=941)
|Age, mean (SD), y||73.3 (6.3)|
|American Society of Anesthesiologists class, No.|
| I||8 (0.9%)|
| II||525 (55.9%)|
| III||400 (42.6%)|
| IV||6 (0.6%)|
|Body mass index, mean (SD), kg/m2||30.9 (6.4)|
| Male||294 (31.2%)|
| Female||647 (68.8%)|
|MCC modifier, No.|
| Absent||877 (93.2%)|
| Present||64 (6.8%)|
| Asian||23 (2.4%)|
| Black||263 (27.9%)|
| Native American||1 (0.1%)|
| White||612 (65.0%)|
| Other||25 (2.8%)|
| Unknown||17 (1.8%)|
Correlations With Variable Cost
|Patient Factor||Median Variable Cost||Interquartile Range||Regression Coefficient||P|
|American Society of Anesthesiologists classa||$729||.001|
|Body mass index||.107|
| Native American||$7457||$6816–$8859||.891|