- Email: [email protected]
Digital Breast Tomosynthesis Implementation: Considerations for Emerging Breast Cancer Screening Bundled Payment Models
ORIGINAL ARTICLE
Digital Breast Tomosynthesis Implementation: Considerations for Emerging Breast Cancer Screening Bundled Payment Models Margaret M. Fleming, MD, MSc a , Danny R. Hughes, PhD b, Lauren P. Golding, MD c, Geraldine McGinty, MD, MBA d, Dan MacFarlane, BS, CPA a, Richard Duszak Jr, MD a Abstract Purpose: Bundled payments have been touted as mechanisms to optimize quality and costs. A recent feasibility study evaluating bundled payments for screening mammography episodes predated widespread adoption of digital breast tomosynthesis (DBT). We explore a similar model reflecting emerging acceptance of DBT in breast cancer screening. Methods: Using 4-year data for 59,094 screening episodes from two large facilities within a large academic health system, we utilized published methodology to calibrate Medicare national allowable reference prices for women undergoing screening mammography before and after practice-wide implementation of DBT. Results: Excluding DBT, Medicare-normalized bundled prices for traditional breast imaging 364 days downstream to screening mammography are extremely similar pre- and post-DBT implementation ($182.86 in 2013; $182.68 in 2015). The addition of DBT increased a DBT-inclusive bundled price by $53.16 (an amount lower than the $56.13 Medicare allowable fee for screening DBT) but was associated with significantly reduced recall rates (13.0% versus 9.4%; P < .0001). Without or with DBT, screening episode bundled prices remained sensitive to bundle-included services and varied little by patient age, race, or insurance status. Conclusions: Prior non-DBT approaches to bundled payment models for breast cancer screening remain viable as DBT becomes the standard of care, with bundle prices varying little by patient age, race, or insurance status. Higher DBT-inclusive bundled prices, however, highlight the need to explore societal costs more broadly (eg, reduced time away from work from fewer recalls) as bundled payment models evolve. Key Words: Bundled payments, alternative payment models, mammography, digital breast tomosynthesis, cancer screening, breast cancer J Am Coll Radiol 2019;-:---. Copyright 2018 American College of Radiology
INTRODUCTION In response to escalating health care costs in the United States, payers and policy makers have increasingly emphasized accelerating the transition from traditional fee-for-service to alternative payment models (APMs) [1], a
Emory University School of Medicine, Atlanta, Georgia. Harvey L. Neiman Health Policy Institute, Reston, Virginia. c Triad Radiology Associates, Winston-Salem, North Carolina. d Weill Cornell Medicine, New York, New York. Corresponding author and reprints: Margaret M. Fleming, Emory University School of Medicine, 1701 Uppergate Drive NE, Ste C1104, Atlanta, GA 30322; e-mail: margaret.fl[email protected]. Dr Duszak’s work was supported by a research grant from the Harvey L. Neiman Health Policy Institute. The other authors state that they have no conflict of interest related to the material discussed in this article. b
ª 2018 American College of Radiology 1546-1440/19/$36.00 n https://doi.org/10.1016/j.jacr.2018.11.025
which aim to emphasize quality and value over the volume of services performed. Bundled episodic payments are an important type of APM and incentivize cost savings within discrete episodes of care [2,3]. Such bundled payments require defined start and end points for an episode and price all interval services, inclusive of those two points, within the bundle. To date, most bundled payment initiatives, such as those for cardiac disease and elective joint replacements, have focused on medical and surgical conditions [4-7]. More recent interest, however, has been expressed in developing outpatient imaging-based cancer screening episodes [8]. To that end, Hughes et al recently explored the feasibility of screening mammography as a potential
1
bundled payment model [9]. Breast cancer screening is an attractive candidate for an imaging-focused bundled payment model because the start point (ie, the screening event) and end point (ie, 364 days afterward or until a positive diagnosis is established) are both clearly identifiable, as are included breast imaging services. Using Medicare claims data and a study population largely consisting of Medicare fee-for-service beneficiaries, that team calibrated bundled prices using national average Medicare Physician Fee Schedule (MPFS) benchmark prices for screening mammography and various downstream diagnostic breast imaging services and highlighted the feasibility of such an APM. The claims data used by Hughes et al, however, predated the implementation of Current Procedural Terminology (CPT) and Healthcare Common Procedure Coding System (HCPCS) codes to report digital breast tomosynthesis (DBT), and thus did not in any way consider the potential impact of this emerging service on their proposed bundled prices. Since receiving approval from the US FDA in 2011 [10], DBT has gained widespread acceptance [11]. Studies have now shown that DBT with mammography has both higher sensitivity and specificity compared with full-field digital mammography alone [12] and results in decreased recall rates and increased cancer detection rates [12-15]. Since its first commercial implementation, DBT has been used increasingly across our academic health system to the extent that now all of our screening mammograms are performed with DBT. Our ability to craft centermatched pre- and post-DBT bundles and uniquely identify DBT services before its availability in large, commonly used payer data sets represents an opportunity to build upon Hughes’ work and explore bundled payment scenarios inclusive of DBT. We thus aim to explore an episodic bundled payment model for breast cancer screening that reflects the emerging widespread adoption of DBT.
METHODS This analysis of patient-deidentified health system billing records was deemed to not constitute human subject research by our institutional review board. As such, institutional review board oversight was not required. In our large academic health system, screening mammography is performed at outpatient sites affiliated with four separate hospitals, which have all now since adopted DBT as part of routine screening. For this analysis, we focused on the two large hospitals that had 2
no DBT capabilities in 2013, but which subsequently performed DBT routinely as part of their screening examinations in 2015 (2014 was a transition year for both). Because our goal was to assess how bundled payment episodes change in conjunction with DBT, we excluded data from another hospital that performed DBT all along as well as another one that did not implement universal screening DBT until after 2015. Non-DBT bundles were constructed using billing claims data from our health system data warehouse using methods recently described by Hughes et al with Medicare claims and another health system’s data [9]. Specifically, claims data for all women undergoing a screening mammogram at any time in calendar year 2013 were analyzed to identify any breast imaging study performed up through the earlier of (1) 364 days after the initial screening examination or (2) the date before the next screening mammogram if it was scheduled between 11 and 12 months after the initial event. Episodes not triggered by a screening event (eg, first presentation with a diagnostic mammography for a lump) were not considered parts of screening episodes. Breast imaging services within screening episodes were identified using CPT and HCPCS codes and grouped into specific breast imaging service categories: (1) screening mammography, (2) diagnostic mammography, (3) mammography computer-assisted detection, (4) ultrasound, and (5) MRI. DBT bundles were constructed in a nearly identical manner for all women undergoing screening mammography any time in calendar year 2015 at the two sites that had since implemented universal DBT as a routine part of screening. The bundle breast imaging service groups, however, were now expanded to include DBT. For screening and postscreening services in both the pre- and post-DBT years, we identified possible subsequent care pathways and tabulated the frequencies with which they occurred. Like Hughes et al [9], we ignored the timing and order of the services within episodes because payers are not typically concerned with timing but rather just costs. For methodological consistency, and in the same manner previously described, bundle prices were calculated by applying the 2015 MPFS reference prices associated with each service during either window and whether patients were Medicare beneficiaries or not. For each identified distinct screening mammography care pathway, we multiplied the MPFS service price by the number of services for each distinct CPT and HCPCS code (ie, original screening mammogram, original screening DBT if Journal of the American College of Radiology Volume - n Number - n Month 2019
Table 1. Demographics of study populations Demographic
2013
2015
Age Mean 59.8 60.1 Median 59.7 60.3 Race African American or black 12,004 (54%) 11,967 (55%) Asian 410 (2%) 404 (2%) Caucasian or white 8,870 (40%) 8,301 (38%) Unknown or other 959 (4%) 1,221 (6%) Payer Medicare or Medicaid 7,873 (35%) 8,080 (37%) Private insurance 14,163 (64%) 13,636 (62%) Self-Pay 207 (1%) 176 (1%)
performed, and any subsequent breast imaging services) in that pathway and summed the total across all services in the pathway to produce the total cost of screening for that care pathway. The frequencyweighted average of each distinct mammography care pathway was then calculated to produce a single bundle price for the entire patient population. We estimated global bundle prices by adding the Medicare professional and technical component bundle prices. We calculated a bundle price for all services and pathways for the 2013 patient cohort and two bundle prices (one with and one without DBT) for the 2015 cohort. Additionally, bundle prices were calculated with (1) screening and diagnostic Table 2. Bundled payment pricing Bundled Breast Imaging Services All screening and diagnostic mammography ($) Professional component Technical component Global All screening and diagnostic mammography and ultrasound ($) Professional component Technical component Global All screening and diagnostic mammography, ultrasound, and MRI ($) Professional component Technical component Global
2013 2015 (Without (Without DBT) DBT)
2015 (With DBT)
44.44 120.46 164.90
44.75 120.83 165.58
73.54 145.20 218.74
46.56 126.34 172.90
47.76 125.55 173.31
76.54 149.93 226.47
48.10 134.76 182.86
49.21 133.47 182.68
78.00 157.85 235.84
DBT ¼ digital breast tomosynthesis.
Journal of the American College of Radiology Fleming et al n Digital Breast Tomosynthesis Implementation
mammography, (2) mammography and ultrasound, and (3) all three modalities (mammography, ultrasound, and MRI). The robustness of the model was evaluated by crafting prices for various patient subsets for which sample sizes were greater than 400. Recall rates were calculated by identifying screening patients who returned for subsequent breast imaging. We also report the demographic data associated with these two cohorts. All analyses were performed using Excel (Microsoft Corporation, Redmond, Washington, USA).
RESULTS Our 2013 cohort consisted of 25,278 women and our 2015 cohort consisted of 33,776 women. Both were similar in sociodemographic and payer characteristics (Table 1). Bundled prices without DBT were similar in 2013 and 2015 (Table 2) with global MPFS benchmarked pricing varying less than $1.00 between the 2 years for bundles including mammography alone ($164.90 versus $165.58), mammography with ultrasound ($172.90 versus $173.31), and mammography with both ultrasound and MRI ($182.86 versus $182.68). Bundle prices were sensitive to included services, rising slightly as each additional modality was included. When DBT was added in 2015, prices increased for all three bundle constructs ($165.58 to $218.74 for mammography alone, $173.31 to $226.47 when ultrasound was included, and $182.68 to $235.84 when both ultrasound and MRI were included; Table 2). However, the increase in each bundle price when DBT was added was consistently less than the MPFS price of DBT. Overall, the addition of DBT to each bundle increased the total bundle price by $53.16, or 29.1% (Fig. 1).
Fig 1. Incremental increase in screening mammography bundle with the addition of digital breast tomosynthesis (DBT).
3
78.53 159.59 238.13 77.03 154.76 231.79 77.54 158.01 235.55 DBT ¼ digital breast tomosynthesis; US ¼ ultrasound.
73.14 142.70 215.84 76.41 153.69 230.10
78.19 157.35 235.54
77.31 159.31 236.62
79.35 161.13 240.48
76.85 150.44 227.29 75.97 148.98 224.95 75.76 148.27 224.03 73.14 142.70 215.84 75.24 147.32 222.56
77.01 150.90 227.92
75.04 146.94 221.98
77.78 152.59 230.36
73.62 145.37 218.99 73.36 144.88 218.24 74.22 146.99 221.21 72.93 143.83 216.75 72.61 143.13 215.74 71.59 140.27 211.86
All screening and diagnostic mammography ($) Professional component 75.22 73.27 72.98 Technical component 149.75 144.40 143.73 Global 224.96 217.67 216.71 All screening and diagnostic mammography and US ($) Professional component 79.94 76.09 75.45 Technical component 157.17 148.83 147.60 Global 237.10 224.92 223.05 All screening and diagnostic mammography, US, and MRI ($) Professional component 81.47 77.67 76.75 Technical component 165.51 157.43 154.73 Global 246.98 235.10 231.48
73.00 143.79 216.79
73.92 146.05 219.96
Caucasian Bundled Breast Imaging Services
40s
50s
60s
70s
>80
African American
Asian
Unknown or Other
Payer Medicare Private or Medicaid Insurance Race Age
Table 3. Bundled payment pricing by demographic (2015 with DBT)
4
Of note, the recall rate in 2013 before DBT implementation was 13.0%; this declined to 9.4% after universal implementation of DBT at the time of screening mammography. Across patient race, there was little difference in bundled prices (Table 3), ranging from $235.54 (African American cohort) and $235.55 (Caucasian cohort) to $240.48 (unknown or other cohort) for the most inclusive bundled. In patients age 40 and older, bundled prices by age decile were similar but did have a tendency to decline as age increased (Table 3). Of note, no one in the over-80 group underwent an MRI, and thus bundled prices with or without MRI were the same in that cohort. Similarly, bundle prices were similar by payer, ranging from $231.79 (Medicare and Medicaid cohort) to $238.13 (commercial insurance cohort).
DISCUSSION Using billing data from two busy metropolitan hospital outpatient imaging centers both before and after universal facility-level implementation of DBT for screening, we used a previously described model to assess (1) the feasibility of that prior bundled pricing model in such a large health care system, (2) the feasibility of applying that bundled pricing model across different patient sociodemographic groups, and (3) the feasibility of that model to include DBT as it becomes standard of care during breast cancer screening episodes. We found that (1) the published, largely Medicare-based model was relevant across an urban health system with a broad payer mix; (2) it is relevant across different patient age, race, and payer populations; and (3) it remains relevant after broad implementation of DBT. As payers and policymakers alike seek to identify episode- and specialtyspecific APMs for demonstration projects, those observations together suggest that such efforts focusing on breast cancer screening could be meaningful, generalizable, and executable. With the emergence of alternative payment models, the radiology community has struggled to determine how best to participate in these new payment systems. Radiologists often believe they have little control over costs associated with many diseases and clinical conditions. Lack of data linking imaging with downstream costs and outcomes is an additional challenge for our specialty to meaningful and fairly participate in APMs. Breast cancer screening is an episode of care that can be attributed to a radiologist and is an opportunity for breast imagers to Journal of the American College of Radiology Volume - n Number - n Month 2019
accept accountability for the quality and costs of patient care. Breast imagers have fewer steps, relative to other diagnostic radiology specialties, in transitioning to accountability given their central and well-defined role in diagnosing diseases of the breast, as well as their typical role in communicating findings and recommendations directly to patients. Both CMS and private payers have clearly embraced APMs as the goal of value-based care; the Merit Based Incentive Payment System is intended to be an intermediate step in the path to more comprehensive riskbased and population-based payments [16]. Nonetheless, the vast majority of radiologists participate in the Merit Based Incentive Payment System arm of the Quality Payment Program [17]. The APMs currently available are almost exclusively designed for primary care, with limited opportunity for specialist participation. The Medicare Access and CHIP Reauthorization Act of 2015 legislation created a pathway for specialists to develop their own payment models as a way to address this problem. Many of the physician focused payment models (PFPMs) that have been proposed to date have been built on episodic bundled payments. Unfortunately, the Health and Human Services Secretary has rejected all of the PFPMs that were recommended by the PFPM Technical Advisory Committee, a Congressionally created advisory committee tasked with evaluating models submitted by health systems, specialty societies, and other stakeholders. Regardless of the fate of specific PFPMs, the use of bundled payment methodology as a cornerstone of APMs has been ubiquitous in both CMS-developed and nonCMS-developed models. Our work establishing the feasibility of a screening mammography bundle is an important step in transparency and defining radiologists’ contribution to patient care in a value-based system. This type of bundle could serve as the basis for a PFPM, as risk-based contracting with commercial payers, or as a way to quantify shared savings allocation within a larger clinically integrated network or accountable care organization. We focused on screening mammography because it is a common service, with potential to impact approximately half of the screening-age population. The concept of an imaging-based bundle could be expanded to other cancer screening services such as colorectal cancer or lung cancer screening. The calculated bundle price previously set forth by Hughes et al using 2012 MPFS pricing ranged between $176.78 for the Medicare population and $192.05 for the private health system [9]. Our subsequent bundle
priced model uses the more recent 2015 MPFS rather than the 2012, and our price of $182.86 for 2013 before DBT implementation lies between the costs of two populations noted in that prior study. Furthermore, our bundle price of $182.68 for 2015 removing DBT from the calculation also lies within the range noted on the prior study. Although that prior work included health system data for validation, it focused largely on Medicare claims data. Our work adds credence to that validation, supporting the contention that the model works as well for other payers and for patients younger than the Medicare population and across a variety of patient demographics. Of note, bundled prices did have a tendency to decline slightly as age increased. This could in part be due to higher recall rates with baseline examinations (those without prior mammograms for comparison) [15,16]. Presumably, as patients age, the likelihood of having prior examinations for comparison increases and thus decreases the potential for further workups. Universal implementation of DBT during screening does increase a bundled price, but less than the price of DBT itself. It seems that this is likely related to reduced recall rates. As seen in other studies in the literature, our recall rates decreased after DBT implementation [6-9]. We focused on costs of imaging services paid by insurers or patients, but, like most pursuing bundled payment work, did not look at broader societal costs, which may be particularly relevant in light of those reduced recall rates. Mammography recalls incur a variety of other costs (eg, travel, missed work, daycare), and those should be considered in future works exploring more fully the value of DBT in a more patient-centric manner. Additionally, like Hughes et al, we focused only on imaging, but believe that the downstream imaging savings (vis-à-vis reduced recalls) may carry over further into fewer biopsies and other invasive procedures. Further bundled works should consider other services, including procedures, downstream to initial and follow-up imaging. The implementation of DBT seems to have a savings evidenced by reduced downstream imaging that may balance its incremental price increase to the bundle. Our study has several limitations. First, although our results were based on two calendar years at two separate hospitals, they may not be generalizable in other communities based on differences in patient population, age, and recall rates. Second, because bundle prices were calculated using the national average 2015 MPFS pricing, appropriate prices for other institutions and practices will
Journal of the American College of Radiology Fleming et al n Digital Breast Tomosynthesis Implementation
5
need to take into account locality adjustments and thencurrent reimbursement rates. Third, because our study focused on two large and busy hospitals, we recognize that episode prices for smaller practices may be more sensitive to ultrasound and MRI utilization in their patient populations. Fourth, our model included both average and high-risk populations. Previous work has noted that excluding high-risk population from the bundle price did not materially change the pricing [9]. Therefore, we did not include or exclude patients based on their risk (eg, prior breast cancer, family history, known gene mutation). And finally, it should be noted that we specifically studied only imaging in our health system and have no way of identifying patients who may have undergone follow-up imaging elsewhere. Accordingly, patients who change health systems within an episode will need to be accounted for and such instances may require reverting to fee-for-service pricing.
TAKE-HOME POINTS -
-
-
6
Prior, largely population-based work demonstrating the feasibility of bundled payment frameworks for breast cancer screening seems feasible as well at the individual institutional level, across patient age, race, and payer groups. Prior non-DBT approaches to bundled payment models for breast cancer imaging screening episodes remain viable as DBT becomes the standard of care. The addition of DBT increases DBT-inclusive bundled prices, but by an amount lower than Medicare allowable fees, and seems to result in significantly reduced recall rates. Higher DBT-inclusive bundled prices highlight the need to explore societal costs more broadly (eg, reduced time away from work from fewer recalls) as bundled payment models evolve.
REFERENCES 1. Burwell SM. Setting value-based payment goals—HHS efforts to improve U.S. health care. N Engl J Med 2015;372:897-9. 2. Cutler DM, Ghosh K. The potential for cost savings through bundled episode payments. N Engl J Med 2012;366:1075-7. 3. Edwards PK, Mears SC, Barnes CL. BPCI: everyone wins, including the patient. J Arthroplasty 2017;32:1728-31. 4. CMS. Bundled Payments for Care Improvement (BPCI) Initiative 2018. Available at: https://innovation.cms.gov/initiatives/bundledpayments/. Accessed December 15, 2018. 5. Medicare Program; Comprehensive Care for Joint Replacement Payment Model for Acute Care Hospitals Furnishing Lower Extremity Joint Replacement Services. Final rule. Fed Regist 2015;80:73273-554. 6. Falit BP, Chernew ME, Mantz CA. Design and implementation of bundled payment systems for cancer care and radiation therapy. Int J Radiat Oncol Biol Phys 2014;89:950-3. 7. Brill JV, Jain R, Margolis PS, et al. A bundled payment framework for colonoscopy performed for colorectal cancer screening or surveillance. Gastroenterology 2014;146:849-53.e9. 8. Liao JM, Basu A, Lee CI. The value of outpatient imaging-based cancer screening episodes. J Gen Intern Med 2018;33:1571-3. 9. Hughes DR, Jiang M, McGinty G, Shetty SK, Duszak R. An empirical framework for breast screening bundled payments. J Am Coll Radiol 2017;14:17-23 e1. 10. US FDA. Radiation-Emitting Products Digital Accreditation 2017. Available at: https://www.fda.gov/radiation-emittingproducts/mammography qualitystandardsactandprogram/facilitycertificationandinspection/ ucm114148.htm. Updated June 14, 2017. Accessed August 31, 2017. 11. Durand MA, Haas BM, Yao X, et al. Early clinical experience with digital breast tomosynthesis for screening mammography. Radiology 2015;274:85-92. 12. Skaane P. Breast cancer screening with digital breast tomosynthesis. Breast Cancer 2017;24:32-41. 13. Skaane P, Bandos AI, Gullien R, et al. Comparison of digital mammography alone and digital mammography plus tomosynthesis in a population-based screening program. Radiology 2013;267:47-56. 14. Haas BM, Kalra V, Geisel J, et al. Comparison of tomosynthesis plus digital mammography and digital mammography alone for breast cancer screening. Radiology 2013;269:694-700. 15. McDonald ES, McCarthy AM, Akhtar AL, et al. Baseline screening mammography: performance of full-field digital mammography versus digital breast tomosynthesis. AJR Am J Roentgenol 2015;205:1143-8. 16. Health Care Payment Learning and Action Network Alternative Payment Model Framework and Progress Tracking Work Group. Alternative Payment Model (APM) Framework. January 12, 2016. Available at: https://hcp-lan.org/workproducts/apm-whitepaper.pdf. Accessed Dec 15, 2018. 17. Rosenkrantz AB, Goldberg JE, Duszak R Jr, Nicola GN. Merit-Based Incentive Payment System participation: radiologists can run but cannot hide. J Am Coll Radiol 2018;15:674-80.
Journal of the American College of Radiology Volume - n Number - n Month 2019
Digital Breast Tomosynthesis Implementation: Considerations for Emerging Breast Cancer Screening Bundled Payment Models Margaret M. Fleming, MD, MSc a , Danny R. Hughes, PhD b, Lauren P. Golding, MD c, Geraldine McGinty, MD, MBA d, Dan MacFarlane, BS, CPA a, Richard Duszak Jr, MD a Abstract Purpose: Bundled payments have been touted as mechanisms to optimize quality and costs. A recent feasibility study evaluating bundled payments for screening mammography episodes predated widespread adoption of digital breast tomosynthesis (DBT). We explore a similar model reflecting emerging acceptance of DBT in breast cancer screening. Methods: Using 4-year data for 59,094 screening episodes from two large facilities within a large academic health system, we utilized published methodology to calibrate Medicare national allowable reference prices for women undergoing screening mammography before and after practice-wide implementation of DBT. Results: Excluding DBT, Medicare-normalized bundled prices for traditional breast imaging 364 days downstream to screening mammography are extremely similar pre- and post-DBT implementation ($182.86 in 2013; $182.68 in 2015). The addition of DBT increased a DBT-inclusive bundled price by $53.16 (an amount lower than the $56.13 Medicare allowable fee for screening DBT) but was associated with significantly reduced recall rates (13.0% versus 9.4%; P < .0001). Without or with DBT, screening episode bundled prices remained sensitive to bundle-included services and varied little by patient age, race, or insurance status. Conclusions: Prior non-DBT approaches to bundled payment models for breast cancer screening remain viable as DBT becomes the standard of care, with bundle prices varying little by patient age, race, or insurance status. Higher DBT-inclusive bundled prices, however, highlight the need to explore societal costs more broadly (eg, reduced time away from work from fewer recalls) as bundled payment models evolve. Key Words: Bundled payments, alternative payment models, mammography, digital breast tomosynthesis, cancer screening, breast cancer J Am Coll Radiol 2019;-:---. Copyright 2018 American College of Radiology
INTRODUCTION In response to escalating health care costs in the United States, payers and policy makers have increasingly emphasized accelerating the transition from traditional fee-for-service to alternative payment models (APMs) [1], a
Emory University School of Medicine, Atlanta, Georgia. Harvey L. Neiman Health Policy Institute, Reston, Virginia. c Triad Radiology Associates, Winston-Salem, North Carolina. d Weill Cornell Medicine, New York, New York. Corresponding author and reprints: Margaret M. Fleming, Emory University School of Medicine, 1701 Uppergate Drive NE, Ste C1104, Atlanta, GA 30322; e-mail: margaret.fl[email protected]. Dr Duszak’s work was supported by a research grant from the Harvey L. Neiman Health Policy Institute. The other authors state that they have no conflict of interest related to the material discussed in this article. b
ª 2018 American College of Radiology 1546-1440/19/$36.00 n https://doi.org/10.1016/j.jacr.2018.11.025
which aim to emphasize quality and value over the volume of services performed. Bundled episodic payments are an important type of APM and incentivize cost savings within discrete episodes of care [2,3]. Such bundled payments require defined start and end points for an episode and price all interval services, inclusive of those two points, within the bundle. To date, most bundled payment initiatives, such as those for cardiac disease and elective joint replacements, have focused on medical and surgical conditions [4-7]. More recent interest, however, has been expressed in developing outpatient imaging-based cancer screening episodes [8]. To that end, Hughes et al recently explored the feasibility of screening mammography as a potential
1
bundled payment model [9]. Breast cancer screening is an attractive candidate for an imaging-focused bundled payment model because the start point (ie, the screening event) and end point (ie, 364 days afterward or until a positive diagnosis is established) are both clearly identifiable, as are included breast imaging services. Using Medicare claims data and a study population largely consisting of Medicare fee-for-service beneficiaries, that team calibrated bundled prices using national average Medicare Physician Fee Schedule (MPFS) benchmark prices for screening mammography and various downstream diagnostic breast imaging services and highlighted the feasibility of such an APM. The claims data used by Hughes et al, however, predated the implementation of Current Procedural Terminology (CPT) and Healthcare Common Procedure Coding System (HCPCS) codes to report digital breast tomosynthesis (DBT), and thus did not in any way consider the potential impact of this emerging service on their proposed bundled prices. Since receiving approval from the US FDA in 2011 [10], DBT has gained widespread acceptance [11]. Studies have now shown that DBT with mammography has both higher sensitivity and specificity compared with full-field digital mammography alone [12] and results in decreased recall rates and increased cancer detection rates [12-15]. Since its first commercial implementation, DBT has been used increasingly across our academic health system to the extent that now all of our screening mammograms are performed with DBT. Our ability to craft centermatched pre- and post-DBT bundles and uniquely identify DBT services before its availability in large, commonly used payer data sets represents an opportunity to build upon Hughes’ work and explore bundled payment scenarios inclusive of DBT. We thus aim to explore an episodic bundled payment model for breast cancer screening that reflects the emerging widespread adoption of DBT.
METHODS This analysis of patient-deidentified health system billing records was deemed to not constitute human subject research by our institutional review board. As such, institutional review board oversight was not required. In our large academic health system, screening mammography is performed at outpatient sites affiliated with four separate hospitals, which have all now since adopted DBT as part of routine screening. For this analysis, we focused on the two large hospitals that had 2
no DBT capabilities in 2013, but which subsequently performed DBT routinely as part of their screening examinations in 2015 (2014 was a transition year for both). Because our goal was to assess how bundled payment episodes change in conjunction with DBT, we excluded data from another hospital that performed DBT all along as well as another one that did not implement universal screening DBT until after 2015. Non-DBT bundles were constructed using billing claims data from our health system data warehouse using methods recently described by Hughes et al with Medicare claims and another health system’s data [9]. Specifically, claims data for all women undergoing a screening mammogram at any time in calendar year 2013 were analyzed to identify any breast imaging study performed up through the earlier of (1) 364 days after the initial screening examination or (2) the date before the next screening mammogram if it was scheduled between 11 and 12 months after the initial event. Episodes not triggered by a screening event (eg, first presentation with a diagnostic mammography for a lump) were not considered parts of screening episodes. Breast imaging services within screening episodes were identified using CPT and HCPCS codes and grouped into specific breast imaging service categories: (1) screening mammography, (2) diagnostic mammography, (3) mammography computer-assisted detection, (4) ultrasound, and (5) MRI. DBT bundles were constructed in a nearly identical manner for all women undergoing screening mammography any time in calendar year 2015 at the two sites that had since implemented universal DBT as a routine part of screening. The bundle breast imaging service groups, however, were now expanded to include DBT. For screening and postscreening services in both the pre- and post-DBT years, we identified possible subsequent care pathways and tabulated the frequencies with which they occurred. Like Hughes et al [9], we ignored the timing and order of the services within episodes because payers are not typically concerned with timing but rather just costs. For methodological consistency, and in the same manner previously described, bundle prices were calculated by applying the 2015 MPFS reference prices associated with each service during either window and whether patients were Medicare beneficiaries or not. For each identified distinct screening mammography care pathway, we multiplied the MPFS service price by the number of services for each distinct CPT and HCPCS code (ie, original screening mammogram, original screening DBT if Journal of the American College of Radiology Volume - n Number - n Month 2019
Table 1. Demographics of study populations Demographic
2013
2015
Age Mean 59.8 60.1 Median 59.7 60.3 Race African American or black 12,004 (54%) 11,967 (55%) Asian 410 (2%) 404 (2%) Caucasian or white 8,870 (40%) 8,301 (38%) Unknown or other 959 (4%) 1,221 (6%) Payer Medicare or Medicaid 7,873 (35%) 8,080 (37%) Private insurance 14,163 (64%) 13,636 (62%) Self-Pay 207 (1%) 176 (1%)
performed, and any subsequent breast imaging services) in that pathway and summed the total across all services in the pathway to produce the total cost of screening for that care pathway. The frequencyweighted average of each distinct mammography care pathway was then calculated to produce a single bundle price for the entire patient population. We estimated global bundle prices by adding the Medicare professional and technical component bundle prices. We calculated a bundle price for all services and pathways for the 2013 patient cohort and two bundle prices (one with and one without DBT) for the 2015 cohort. Additionally, bundle prices were calculated with (1) screening and diagnostic Table 2. Bundled payment pricing Bundled Breast Imaging Services All screening and diagnostic mammography ($) Professional component Technical component Global All screening and diagnostic mammography and ultrasound ($) Professional component Technical component Global All screening and diagnostic mammography, ultrasound, and MRI ($) Professional component Technical component Global
2013 2015 (Without (Without DBT) DBT)
2015 (With DBT)
44.44 120.46 164.90
44.75 120.83 165.58
73.54 145.20 218.74
46.56 126.34 172.90
47.76 125.55 173.31
76.54 149.93 226.47
48.10 134.76 182.86
49.21 133.47 182.68
78.00 157.85 235.84
DBT ¼ digital breast tomosynthesis.
Journal of the American College of Radiology Fleming et al n Digital Breast Tomosynthesis Implementation
mammography, (2) mammography and ultrasound, and (3) all three modalities (mammography, ultrasound, and MRI). The robustness of the model was evaluated by crafting prices for various patient subsets for which sample sizes were greater than 400. Recall rates were calculated by identifying screening patients who returned for subsequent breast imaging. We also report the demographic data associated with these two cohorts. All analyses were performed using Excel (Microsoft Corporation, Redmond, Washington, USA).
RESULTS Our 2013 cohort consisted of 25,278 women and our 2015 cohort consisted of 33,776 women. Both were similar in sociodemographic and payer characteristics (Table 1). Bundled prices without DBT were similar in 2013 and 2015 (Table 2) with global MPFS benchmarked pricing varying less than $1.00 between the 2 years for bundles including mammography alone ($164.90 versus $165.58), mammography with ultrasound ($172.90 versus $173.31), and mammography with both ultrasound and MRI ($182.86 versus $182.68). Bundle prices were sensitive to included services, rising slightly as each additional modality was included. When DBT was added in 2015, prices increased for all three bundle constructs ($165.58 to $218.74 for mammography alone, $173.31 to $226.47 when ultrasound was included, and $182.68 to $235.84 when both ultrasound and MRI were included; Table 2). However, the increase in each bundle price when DBT was added was consistently less than the MPFS price of DBT. Overall, the addition of DBT to each bundle increased the total bundle price by $53.16, or 29.1% (Fig. 1).
Fig 1. Incremental increase in screening mammography bundle with the addition of digital breast tomosynthesis (DBT).
3
78.53 159.59 238.13 77.03 154.76 231.79 77.54 158.01 235.55 DBT ¼ digital breast tomosynthesis; US ¼ ultrasound.
73.14 142.70 215.84 76.41 153.69 230.10
78.19 157.35 235.54
77.31 159.31 236.62
79.35 161.13 240.48
76.85 150.44 227.29 75.97 148.98 224.95 75.76 148.27 224.03 73.14 142.70 215.84 75.24 147.32 222.56
77.01 150.90 227.92
75.04 146.94 221.98
77.78 152.59 230.36
73.62 145.37 218.99 73.36 144.88 218.24 74.22 146.99 221.21 72.93 143.83 216.75 72.61 143.13 215.74 71.59 140.27 211.86
All screening and diagnostic mammography ($) Professional component 75.22 73.27 72.98 Technical component 149.75 144.40 143.73 Global 224.96 217.67 216.71 All screening and diagnostic mammography and US ($) Professional component 79.94 76.09 75.45 Technical component 157.17 148.83 147.60 Global 237.10 224.92 223.05 All screening and diagnostic mammography, US, and MRI ($) Professional component 81.47 77.67 76.75 Technical component 165.51 157.43 154.73 Global 246.98 235.10 231.48
73.00 143.79 216.79
73.92 146.05 219.96
Caucasian Bundled Breast Imaging Services
40s
50s
60s
70s
>80
African American
Asian
Unknown or Other
Payer Medicare Private or Medicaid Insurance Race Age
Table 3. Bundled payment pricing by demographic (2015 with DBT)
4
Of note, the recall rate in 2013 before DBT implementation was 13.0%; this declined to 9.4% after universal implementation of DBT at the time of screening mammography. Across patient race, there was little difference in bundled prices (Table 3), ranging from $235.54 (African American cohort) and $235.55 (Caucasian cohort) to $240.48 (unknown or other cohort) for the most inclusive bundled. In patients age 40 and older, bundled prices by age decile were similar but did have a tendency to decline as age increased (Table 3). Of note, no one in the over-80 group underwent an MRI, and thus bundled prices with or without MRI were the same in that cohort. Similarly, bundle prices were similar by payer, ranging from $231.79 (Medicare and Medicaid cohort) to $238.13 (commercial insurance cohort).
DISCUSSION Using billing data from two busy metropolitan hospital outpatient imaging centers both before and after universal facility-level implementation of DBT for screening, we used a previously described model to assess (1) the feasibility of that prior bundled pricing model in such a large health care system, (2) the feasibility of applying that bundled pricing model across different patient sociodemographic groups, and (3) the feasibility of that model to include DBT as it becomes standard of care during breast cancer screening episodes. We found that (1) the published, largely Medicare-based model was relevant across an urban health system with a broad payer mix; (2) it is relevant across different patient age, race, and payer populations; and (3) it remains relevant after broad implementation of DBT. As payers and policymakers alike seek to identify episode- and specialtyspecific APMs for demonstration projects, those observations together suggest that such efforts focusing on breast cancer screening could be meaningful, generalizable, and executable. With the emergence of alternative payment models, the radiology community has struggled to determine how best to participate in these new payment systems. Radiologists often believe they have little control over costs associated with many diseases and clinical conditions. Lack of data linking imaging with downstream costs and outcomes is an additional challenge for our specialty to meaningful and fairly participate in APMs. Breast cancer screening is an episode of care that can be attributed to a radiologist and is an opportunity for breast imagers to Journal of the American College of Radiology Volume - n Number - n Month 2019
accept accountability for the quality and costs of patient care. Breast imagers have fewer steps, relative to other diagnostic radiology specialties, in transitioning to accountability given their central and well-defined role in diagnosing diseases of the breast, as well as their typical role in communicating findings and recommendations directly to patients. Both CMS and private payers have clearly embraced APMs as the goal of value-based care; the Merit Based Incentive Payment System is intended to be an intermediate step in the path to more comprehensive riskbased and population-based payments [16]. Nonetheless, the vast majority of radiologists participate in the Merit Based Incentive Payment System arm of the Quality Payment Program [17]. The APMs currently available are almost exclusively designed for primary care, with limited opportunity for specialist participation. The Medicare Access and CHIP Reauthorization Act of 2015 legislation created a pathway for specialists to develop their own payment models as a way to address this problem. Many of the physician focused payment models (PFPMs) that have been proposed to date have been built on episodic bundled payments. Unfortunately, the Health and Human Services Secretary has rejected all of the PFPMs that were recommended by the PFPM Technical Advisory Committee, a Congressionally created advisory committee tasked with evaluating models submitted by health systems, specialty societies, and other stakeholders. Regardless of the fate of specific PFPMs, the use of bundled payment methodology as a cornerstone of APMs has been ubiquitous in both CMS-developed and nonCMS-developed models. Our work establishing the feasibility of a screening mammography bundle is an important step in transparency and defining radiologists’ contribution to patient care in a value-based system. This type of bundle could serve as the basis for a PFPM, as risk-based contracting with commercial payers, or as a way to quantify shared savings allocation within a larger clinically integrated network or accountable care organization. We focused on screening mammography because it is a common service, with potential to impact approximately half of the screening-age population. The concept of an imaging-based bundle could be expanded to other cancer screening services such as colorectal cancer or lung cancer screening. The calculated bundle price previously set forth by Hughes et al using 2012 MPFS pricing ranged between $176.78 for the Medicare population and $192.05 for the private health system [9]. Our subsequent bundle
priced model uses the more recent 2015 MPFS rather than the 2012, and our price of $182.86 for 2013 before DBT implementation lies between the costs of two populations noted in that prior study. Furthermore, our bundle price of $182.68 for 2015 removing DBT from the calculation also lies within the range noted on the prior study. Although that prior work included health system data for validation, it focused largely on Medicare claims data. Our work adds credence to that validation, supporting the contention that the model works as well for other payers and for patients younger than the Medicare population and across a variety of patient demographics. Of note, bundled prices did have a tendency to decline slightly as age increased. This could in part be due to higher recall rates with baseline examinations (those without prior mammograms for comparison) [15,16]. Presumably, as patients age, the likelihood of having prior examinations for comparison increases and thus decreases the potential for further workups. Universal implementation of DBT during screening does increase a bundled price, but less than the price of DBT itself. It seems that this is likely related to reduced recall rates. As seen in other studies in the literature, our recall rates decreased after DBT implementation [6-9]. We focused on costs of imaging services paid by insurers or patients, but, like most pursuing bundled payment work, did not look at broader societal costs, which may be particularly relevant in light of those reduced recall rates. Mammography recalls incur a variety of other costs (eg, travel, missed work, daycare), and those should be considered in future works exploring more fully the value of DBT in a more patient-centric manner. Additionally, like Hughes et al, we focused only on imaging, but believe that the downstream imaging savings (vis-à-vis reduced recalls) may carry over further into fewer biopsies and other invasive procedures. Further bundled works should consider other services, including procedures, downstream to initial and follow-up imaging. The implementation of DBT seems to have a savings evidenced by reduced downstream imaging that may balance its incremental price increase to the bundle. Our study has several limitations. First, although our results were based on two calendar years at two separate hospitals, they may not be generalizable in other communities based on differences in patient population, age, and recall rates. Second, because bundle prices were calculated using the national average 2015 MPFS pricing, appropriate prices for other institutions and practices will
Journal of the American College of Radiology Fleming et al n Digital Breast Tomosynthesis Implementation
5
need to take into account locality adjustments and thencurrent reimbursement rates. Third, because our study focused on two large and busy hospitals, we recognize that episode prices for smaller practices may be more sensitive to ultrasound and MRI utilization in their patient populations. Fourth, our model included both average and high-risk populations. Previous work has noted that excluding high-risk population from the bundle price did not materially change the pricing [9]. Therefore, we did not include or exclude patients based on their risk (eg, prior breast cancer, family history, known gene mutation). And finally, it should be noted that we specifically studied only imaging in our health system and have no way of identifying patients who may have undergone follow-up imaging elsewhere. Accordingly, patients who change health systems within an episode will need to be accounted for and such instances may require reverting to fee-for-service pricing.
TAKE-HOME POINTS -
-
-
6
Prior, largely population-based work demonstrating the feasibility of bundled payment frameworks for breast cancer screening seems feasible as well at the individual institutional level, across patient age, race, and payer groups. Prior non-DBT approaches to bundled payment models for breast cancer imaging screening episodes remain viable as DBT becomes the standard of care. The addition of DBT increases DBT-inclusive bundled prices, but by an amount lower than Medicare allowable fees, and seems to result in significantly reduced recall rates. Higher DBT-inclusive bundled prices highlight the need to explore societal costs more broadly (eg, reduced time away from work from fewer recalls) as bundled payment models evolve.
REFERENCES 1. Burwell SM. Setting value-based payment goals—HHS efforts to improve U.S. health care. N Engl J Med 2015;372:897-9. 2. Cutler DM, Ghosh K. The potential for cost savings through bundled episode payments. N Engl J Med 2012;366:1075-7. 3. Edwards PK, Mears SC, Barnes CL. BPCI: everyone wins, including the patient. J Arthroplasty 2017;32:1728-31. 4. CMS. Bundled Payments for Care Improvement (BPCI) Initiative 2018. Available at: https://innovation.cms.gov/initiatives/bundledpayments/. Accessed December 15, 2018. 5. Medicare Program; Comprehensive Care for Joint Replacement Payment Model for Acute Care Hospitals Furnishing Lower Extremity Joint Replacement Services. Final rule. Fed Regist 2015;80:73273-554. 6. Falit BP, Chernew ME, Mantz CA. Design and implementation of bundled payment systems for cancer care and radiation therapy. Int J Radiat Oncol Biol Phys 2014;89:950-3. 7. Brill JV, Jain R, Margolis PS, et al. A bundled payment framework for colonoscopy performed for colorectal cancer screening or surveillance. Gastroenterology 2014;146:849-53.e9. 8. Liao JM, Basu A, Lee CI. The value of outpatient imaging-based cancer screening episodes. J Gen Intern Med 2018;33:1571-3. 9. Hughes DR, Jiang M, McGinty G, Shetty SK, Duszak R. An empirical framework for breast screening bundled payments. J Am Coll Radiol 2017;14:17-23 e1. 10. US FDA. Radiation-Emitting Products Digital Accreditation 2017. Available at: https://www.fda.gov/radiation-emittingproducts/mammography qualitystandardsactandprogram/facilitycertificationandinspection/ ucm114148.htm. Updated June 14, 2017. Accessed August 31, 2017. 11. Durand MA, Haas BM, Yao X, et al. Early clinical experience with digital breast tomosynthesis for screening mammography. Radiology 2015;274:85-92. 12. Skaane P. Breast cancer screening with digital breast tomosynthesis. Breast Cancer 2017;24:32-41. 13. Skaane P, Bandos AI, Gullien R, et al. Comparison of digital mammography alone and digital mammography plus tomosynthesis in a population-based screening program. Radiology 2013;267:47-56. 14. Haas BM, Kalra V, Geisel J, et al. Comparison of tomosynthesis plus digital mammography and digital mammography alone for breast cancer screening. Radiology 2013;269:694-700. 15. McDonald ES, McCarthy AM, Akhtar AL, et al. Baseline screening mammography: performance of full-field digital mammography versus digital breast tomosynthesis. AJR Am J Roentgenol 2015;205:1143-8. 16. Health Care Payment Learning and Action Network Alternative Payment Model Framework and Progress Tracking Work Group. Alternative Payment Model (APM) Framework. January 12, 2016. Available at: https://hcp-lan.org/workproducts/apm-whitepaper.pdf. Accessed Dec 15, 2018. 17. Rosenkrantz AB, Goldberg JE, Duszak R Jr, Nicola GN. Merit-Based Incentive Payment System participation: radiologists can run but cannot hide. J Am Coll Radiol 2018;15:674-80.
Journal of the American College of Radiology Volume - n Number - n Month 2019