Dissertation Title: "Strategic Behavior Among Provider Organizations in Risk-Sharing Contracts"
Medicare implemented population-based risk-sharing payment models that encouraged otherwise separate providers to collaborate and form Accountable Care Organizations (ACOs). In contrast to fee-for-service payment, ACOs shared financial risk in the total annual spending of their attributed beneficiaries, aiming to incentivize the delivery of higher value health care. However, various policies intended to support these models also created new opportunities for strategic behavior that could run counter to program goals. My dissertation studied whether provider organizations responded to these related policies and the potential implications for the future of Medicare’s largest and longest-standing ACO initiative, the Medicare Shared Savings Program (MSSP).
Paper 1 evaluated the effects of Medicare ACO participation on commercial primary care prices to shed light on organizations potentially engaged in joint negotiation. Antitrust guidance specifies that participation in Medicare ACOs meets clinical integration standards for separately owned providers to jointly negotiate with insurers. Accordingly, ACO participation may facilitate price increases through a less conventional, “softer” consolidation that would not be categorically challenged as price fixing. Using commercial claims and data on health system membership and ACO participation, we found some abrupt, large price increases for independent primary care practices that joined health system-led ACOs but were not acquired by systems. These price jumps were rare, however, increasing prices by just 4 percent on average among independent practices in system-led ACOs. Additional analyses suggested minimal increases in health systems’ primary care prices or market shares from ACO contracting. Thus, the price jumps were more consistent with an extension of existing pricing power from systems to some independent practices than with a major expansion of system market power. Nevertheless, the potential for growth of these arrangements through ACOs argues for closer monitoring and evaluation.
Paper 2 studied potential selection in the MSSP related to benchmarking changes introduced in 2017. The switch to “regionalized” benchmarks effectively subsidized the participation of ACOs and practices with negative baseline regional spending deviation (i.e., below-average spending in their region). To assess whether providers responded to these new selection incentives, I conducted a series of descriptive analyses and characterized changes in the composition of MSSP participants before versus after the benchmark regionalization rule. I found that ACOs entering the MSSP following benchmarking changes had on average 3.5 times lower baseline regional spending deviation compared to ACOs that entered prior. After earlier ACO cohorts entered the MSSP, they may have responded to selection incentives through subsequent ACO-level exit or within-ACO participant list reconfiguration. Prior to benchmarking changes, exits and reconfigurations tended to increase these cohorts’ average regional spending deviation, but these patterns reversed in 2017, coinciding with new selection incentives. Participation pattern changes also appeared to vary meaningfully by ACO type. While other factors besides benchmark regionalization may have explained observed changes in participation patterns, recent shifts nonetheless implied that a substantial portion of total net bonuses were awarded due to selective participation.
Paper 3 investigated whether integrated provider organizations used Medicare ACO contracts to redirect referrals from independent primary care providers (PCPs) to their own hospitals and specialists. Since separately owned providers participating in ACOs may coordinate care while sharing bonuses and resources, Medicare ACOs received special waivers from self-referral and anti-kickback laws to help ACOs address care leakage. However, some have raised concerns that this regulatory relief could also enable provider organizations to redirect or reinforce referrals to their own specialists and hospitals, even when other more efficient providers or care pathways exist. I investigated these concerns by employing a stacked difference-in-differences design. Measuring specialist and hospital referral patterns, I compared changes for independent PCPs joining ACOs led by specialist- or hospital-integrated provider organizations versus changes for non-ACO PCPs in the same ZIP codes. Since I could not directly observe referral decisions, I constructed proxy measures based on utilization flows. On average, I found no evidence of differential changes in specialty or hospital referral patterns associated with participation in ACOs led by an integrated provider organization. These findings are reassuring given concerns about possible exploitation of self-referral and anti-kickback exemptions, but changes could also require longer timeframes to manifest than observed in this study.