New pricing approaches for bundled payments: Leveraging clinical standards and regional variations to target avoidable utilization
Introduction
While health care payment systems based on case mix-adjusted activity—such as Diagnosis-Related Groups—continue to serve as policy mainstays in most industrialized health systems [1], payers have increasingly found these systems wanting on two important fronts. First, these systems are often cited as perpetuating fragmented health care delivery systems by paying à la carte for discrete services furnished by individual provider entities, offering few incentives for improving coordination of care [2], [3]. Second, the prices set for services under these systems are typically based on the average historical costs of providing the service, regardless of whether the care was provided efficiently or effectively. This retrospective pricing approach can potentially perpetuate wasteful patterns of historical utilization by “baking in” the costs of suboptimal care into the pricing model [4], [5], [6].
In the hopes of overcoming the aforementioned first issue of fragmented care, governments and insurers in several countries have begun to experiment with varieties of bundled payment, where a single payment is issued for a patient's complete episode of care over a pre-defined period of time. The reimbursed episode can span multiple entities such as hospitals, physicians and post-acute care providers, offering shared incentives for delivering coordinated and efficient care across the entire course of treatment [7], [8], [9]. While the empirical evidence supporting bundled payments is still emerging, demonstration projects have shown promising results in reducing costs and improving coordination of care [10], [11], [12], [13]. In the US, the Medicare system has put bundled payments on centre stage by launching a new series of pilots through the Bundled Payment for Care Improvement (BPCI) initiative legislated under the Affordable Care and Patient Protection Act [14]. The BPCI gives providers the flexibility to select from a set of eligible health conditions and to choose a 30, 60 or 90 day post-discharge time window for the reimbursed episode of care.
While bundled payment models such as Medicare's BPCI initiative show promise for bridging fragmented payment systems, pilot programs to date have done little to address the second limitation of traditional payment models: pricing models driven by historical average costs. The BPCI program sets prices for episodes based on the discounted sum of historical payments for services previously provided within the episode, without consideration for whether such services were necessary or effective [15]. With a growing body of research demonstrating wide regional variations in the costs of episodes of care driven by unwarranted variations in utilization of discretionary or potentially avoidable services [16], [17], payers that continue to rely on historical costs for pricing bundled payments run the risk of perpetuating patterns of suboptimal utilization.
To date, payers seeking to experiment with more innovative pricing approaches for bundled payments have had little in the way of guidance from the health services literature. One notable exception is the PROMETHEUS Payment pilots, which reimburse episodes of care with pricing adjustments for potentially avoidable complications such as unplanned hospital admissions [18], [19]. While these elements are promising, the PROMETHEUS methodology does not address unwarranted historical variations between providers in the way they utilize different health care settings—a key driver of avoidable variations in costs [20].
This study addresses this gap in the literature by developing a set of alternative models for pricing episodes of stroke care. Stroke is one of the most devastating and costly conditions included under Medicare's BCPI program, and has been suggested as a good candidate condition for bundled payment reforms. We capitalize on payment reforms underway in the Canadian province of Ontario, where an expert panel of clinicians has provided clinical parameters for defining best practice episodes of stroke care, including identifying areas of potentially avoidable utilization [21]. Using linked administrative data spanning hospitals, physicians and post-acute care providers, episodes of care are constructed for incident ischemic and hemorrhagic strokes. Based on the panel's recommended best practice standards, costs are estimated for five clinically defined areas of potentially avoidable utilization. Finally, the study examines how regional variations in these areas of utilization can be exploited to establish best practice benchmarks for pricing bundled payments.
The pricing strategies explored in this study are pertinent to payers and policymakers seeking to move beyond traditional activity-based payment models toward paying for efficient, high quality care across providers and settings. Moving forward, bundled payments may provide a powerful platform to tackle unwarranted practice variations through use of pricing models that expressly target potentially avoidable utilization defined through evidence-based clinical standards. The findings in this study will also be of interest to providers of stroke care and to students of regional variations in health service utilization.
In 2012, the Canadian province of Ontario, home to 13.2 million residents, began implementing a new activity-based payment model for hospital-based services. While resembling a conventional DRG-like hospital payment system in its current form, the Ontario government has announced two ambitious policy directions for the payment model's evolution. First, payments will eventually reimburse the entire episode of care for a health condition, potentially spanning multiple providers and settings [22]. Second, prices under the new model will be based on the expected costs of best practice care, excluding the costs of services that are ineffective or the result of quality problems [23].
Well aware of the perils of administrators attempting to define clinical best practice, the Ontario government has commissioned expert panels of leading clinicians to develop best practice standards for a set of high volume conditions funded under the new model. Lessons learned from the PROMETHEUS pilots have reinforced the importance of engaging clinicians early on in new payment reforms in order to provide input into the methodology and build buy-in for the reforms among their communities of practice [20].
As with Medicare's BCPI program, stroke has been selected by the Ontario government as an early adopter for payment reform. Prior studies have noted that stroke is well-suited to bundled payment [24], with an episode of care that typically spans multiple settings, putting patients at risk for experiencing costly adverse events as they transition between providers [25], [26], [27], [28]. Ontario practice leaders have also identified opportunities for reducing avoidable utilization in stroke care, including excess hospital days [29] and excess use of inpatient rehabilitation [28].
The stroke expert panel drew from these prior analyses and published guidelines [28], [29], [30] to formulate a set of best practice standards spanning the acute, rehabilitation and community phases of the stroke episode [21]. Many of these best practices cannot be readily translated into routinely reported administrative data elements, the building blocks of provider payment systems. However, out of the total list of recommendations, five key best practice standards (described in Table 1) were identified for their amenability to reliable measurement in Ontario administrative datasets. Acknowledging that this limited set of standards can capture only a fraction of all the elements of good quality stroke care, we model four illustrative options for applying these standards to adjust the prices paid for episodes of care in order to explore possible approaches toward normative pricing of bundled payments.
Section snippets
Cohort and episode development
This analysis is based on retrospective observational health care utilization and claims data for a cohort of incident Ontario stroke episodes spanning four fiscal years, from April 1st 2008 to March 31st 2012. The cohort is constructed from the entire population of strokes in Ontario that result in an acute hospitalization. Episodes of care are created by linking index stroke hospitalizations with subsequent health service utilization over different periods of time and across multiple
Stroke episode costs
Across the four study years, there were an average annual 7431 ischemic and 1826 hemorrhagic stroke episodes. The total annual public expenditures on 30 and 90 day episodes of stroke care, respectively, were $177 and $231 million for ischemic stroke and $45 and $56 million for hemorrhagic stroke.
Fig. 1 illustrates average cumulative spending on stroke episodes over time, combining ischemic and hemorrhagic strokes from index events through 30 and 90 day episode costs. The intensity of
Discussion
Several influential publications have concluded that a substantial share of health care costs are the result of wasteful utilization [6], [46], [47], with estimates running as high as 29% of total US health care spending [48]. These analyses frequently arrive at their estimates through comparing observed differences in the per capita costs of care between regions [46], [48]. Criticizing this top-down approach as overly simplistic, Reschovsky et al. [49] performed a more nuanced analysis that
Acknowledgements
Preliminary results have been shared with members of the Stroke Quality-Based Procedure Expert Panel, and the Ontario Ministry of Health and Long-Term Care (MOHLTC). Financial support for this project was partially provided by the Canadian Institutes of Health Research (CIHR). The last author is a Scholar of the Michael Smith Foundation for Health Research (MSFHR). The views expressed in the paper do not represent the views of CIHR, MSFHR or the MOHLTC.
References (59)
- et al.
The average hospital
Accounting, Organizations and Society
(2005) - et al.
Readmission after stroke: recurrence or infection?
Journal of the Chinese Medical Association
(2013) - et al.
Bundles: an opportunity to align incentives for continuing care in
Health Policy
(2012) - et al.
Challenges of rehabilitation case mix measurement in Ontario hospitals
Health Policy
(2008) - et al.
Australian diagnosis related groups: drivers of complexity adjustment
Health Policy
(2015) - et al.
- et al.
Paying for hospital services: a hard look at the options. Toronto
(2013) From volume to value: better ways to pay for health care
Health Affairs (Millwood)
(2009)- et al.
Activity based financing in England: the need for continual refinement of payment by results
Health Economics, Policy and Law
(2007) - et al.
Controlling costly care: a billion-dollar hospital opportunity. Melbourne
(2014)