The University of Texas MD Anderson Cancer Center, in conjunction with UnitedHealthcare, implemented prospective bundled payments for patients with head and neck cancer. We successfully developed a method to identify and enroll patients and track performance. The biggest challenges were in claims management and in addressing new technologies.
Cancer conditions are good candidates for bundled reimbursement.
Within the provider organization, close collaboration among clinical, financial, and operational teams is essential for success.
The provider/payer relationship must have a foundation of trust and transparency.
Strong project management is necessary to ensure that deadlines are met and barriers addressed in a timely manner.
As health care moves from a fee-for-service system to alternative payment plans, there are few well-tested models. Cancer care, optimally delivered in a multidisciplinary setting, lends itself to a bundled reimbursement approach. However, bundled payments for cancer treatment are in the early stages of development with efforts to-date focused on targeted aspects of care. There is no evidence that bundles control costs or improve outcomes of cancer care — questions we will try to address.
MD Anderson partnered with UnitedHealthcare to test the feasibility of bundled reimbursement for multidisciplinary cancer care. We designed a single payment for one year of care for patients with newly diagnosed head and neck cancer. This group was chosen for the pilot due to efficient processes, strong care coordination, participation in prior cost studies, and the insurer’s preference.
First, we modeled the costs of care for a cohort of patients. We found that costs—the biggest being surgery, radiation, hospitalization, intensive care, and imaging — varied based on treatment plan and patient comorbidities, not cancer diagnosis and staging. We proposed four treatment-related payment bundles, with a payment modifier for patients with two or more comorbidities. During weekly meetings with the payer, we negotiated included services for each of the four bundles, patient eligibility criteria, and bundle prices.
We also agreed upon a stop-loss provision, lessening MD Anderson’s financial risk for the few patients (about 1%) incurring significantly higher costs (which could be 4 or 5 times the bundled price) for unanticipated complications.
We worked closely with the clinical team to develop a protocol to verify patient eligibility and define pilot operations. Clinical leaders emphasized the importance of implementing the pilot without impacting patient care.
The provider and payer teams included clinical, financial, and operational leaders, along with project managers. Lawyers and contracting experts were engaged as needed.
We developed measures to allow us to assess the value of care for enrolled patients — the outcomes and costs. We developed dashboards to track financial and clinical outcomes, including survival and recurrence; complications, such as reoperation and readmission; timeliness of care; time to return to normal activities; and functional status, such as the ability to speak and swallow. We developed systems to collect many of these outcomes via patient-reported outcome tools. Enrollment is under way.
Initial deliberations began in late 2013, with pilot development in 2014. The three-year pilot started in November 2014, with patient enrollment during the first two years.
The most unexpected hurdle was in claims management. Fee-for-service claims systems are not well configured to test alternative payment models. Bundled payments required manual workarounds by MD Anderson and UnitedHealthcare. Claims and payment processing improved with experience but still lagged behind the automated fee-for-service process. Additionally, proton therapy, a newer technology, was excluded from the bundle (and paid for separately) because it is not a covered service for the pilot population. Barriers to including emerging technologies and innovative therapies are one problem we are working to address in this bundle, and will similarly be a challenge in other bundled reimbursement programs as they evolve.
We will track our performance carefully, comparing the outcomes and costs (including time-driven activity-based costing) of enrolled patients with previously treated patients, to determine whether we are controlling costs and optimizing outcomes. We are also evaluating other cancer conditions that can be paid in this manner.
Where to Start
Within your own organization, identify health conditions that have well-coordinated, integrated care delivery. Engage clinicians in defining which services should be included in a bundled payment. Work with finance leaders to understand your costs of delivering care before establishing a bundle price. Key questions to ask are: Do the clinical, finance, and administrative teams have a good relationship? Can you build the systems to assess your performance in a bundle? How will you measure outcomes and costs, and set performance targets?
The authors thank Ron Walters, MD; Randy Weber, MD; and James Incalcaterra, PhD of MD Anderson Cancer Center and Lee Newcomer, MD, of UnitedHealthcare for their contributions to the development of the pilot.
This case study originally appeared in NEJM Catalyst on November 25, 2015.