The need for matching pre-dates the world of OkCupid and Tinder. But according to Nobel prize winning economist Alan Roth there are lots of endeavors where what we seek isn't available as a commodity and there are certain criteria that determine an optimal match. He has been behind using matching across disparate set of problems from kidney exchange to school choice within most of the large metro areas in the United States. At a larger scale we can see a matching market at work in Federal Trade Commission (FTC) auctions for radio spectrum. Bids represent complex bundles across multiple players, and the matches are determined based on selection of bundles that provide the highest utility. Is there a way we can apply some of these principles to healthcare payment and delivery?

Under the Affordable Care Act (ACA) Center for Medicare and Medicaid Service (CMS) launched an array of experiments under Bundled Payments for Care Improvement (BPCI), which selected 48 different clinical episodes that provider organizations could choose from. The sub-parts of the BPCI experiment allowed the participants to engage in increasing levels of sophistication across two axes a) time horizon - retrospective vs. prospective and b) bundling across care settings (acute, post-acute). With the newly announced Comprehensive Care for Joint Replacement (CCJR) experiment, CMS has focused its attention on an area that has shown a lot of variations in cost and outcomes. Other experiments such as the Oncology Care Model (OCM) utilize the bundle payment model for non-surgical care around chemotherapy episodes.

With questions around the ROI for various ACO experiments, a lot of the organizations have pulled out of their contracts and are unsure about further participation. The skittishness around this is due to the fact that the savings have been negligible to non-existent. Secondly, most of the ACO contracts both for CMS and commercials had no real risk and hence no strong incentive to push participating provider organizations to assess whether they were really prepared to take on risk. Fact is, most provider organizations have proven to be ill-equipped to bear risk on behalf of the patients. Lack of top-down leadership, organizational nous, data mobility, capability to engage the patient, changing and challenging health IT regulations are just a few of the several reasons for these failures.

The bundle payment experiments have shown more promise with a narrower focus on care episodes. However, this is in no way a slam-dunk. While provider organizations have an easier task at bookending care delivery, organizations still have to keep track of overall medical resource utilization that might affect the episode outcomes. The questions on when to move the patient across care delivery boundaries and how to disburse payment will continue to plague the organizations that come together to provide these bundles. Overall, both the ACO and bundle payment experiments represent the need for skills that delivery organizations do not naturally possess as part of their DNA. The regulatory framework should not only govern reporting requirements but also provide guidelines used to vet partners who will result in achieving the desired result.

Let's take market design perspective and assume that these experiments by CMS represent an attempt at arriving upon a blueprint for matching disparate organizations that have not worked with each other, to provide care under the pay-for-performance paradigm. There are multiple types of matches possible based upon the intended outcome but the construct provided by CMS and other commercials is still evolving. As per Alan Roth this would be a dysfunctional matching market where the rules of this marketplace is still evolving and make it unsafe to participate, and by that consequence the market is not thick. From the payers perspective this requires putting a clearer guidelines that allow for providers to find each other to maximize their chance of extracting the best outcome for their patients as well as their organizations.