Changes in Spending, Risk Selection, and the Response of Frontline Clinicians: Understanding Performance Mechanisms in the Medicare Shared Savings Program

Abstract

Confronted by an increasingly expensive and fragmented health care system, public and private payers have established a series of reforms designed to lower costs and improve quality. A leading example is the Medicare Shared Savings Program (MSSP) establishment of Accountable Care Organizations (ACOs), where groups of providers voluntarily assume responsibility for the spending and quality outcomes of a population of Medicare beneficiaries. Studies suggest MSSP ACOs have achieved modest savings. However, prior research may overstate savings if ACOs strategically select for low-cost clinicians or drop high-cost clinicians.

In Chapter I, we evaluated the effect of the MSSP on spending and quality while accounting for clinicians’ non-random entry and exit in the program, using the share of nearby clinicians in the MSSP as an instrumental variable. Hip fracture served as a falsification test. In instrumental variable models, the MSSP was not associated with spending ($6 per beneficiary-quarter; 95% confidence interval [CI], -$50, $63), any clinical quality indicator, or hip fracture (0.09 per 1000 beneficiary-quarters; CI: -0.07, 0.26). In contrast, adjusted longitudinal models — similar to what other ACO studies have estimated — failed the falsification test (hip fracture, -0.24; CI: -0.33, -0.17), an 11% decrease. Highest-cost clinicians (99th percentile spending) had a 30.0% chance of exiting the MSSP compared to a 14.7% chance among median (50th percentile) clinicians. Together, these results suggest that exit of high-cost clinicians may drive estimates of savings.

ACOs may also avoid sick or high-cost beneficiaries. To encourage ACOs to care for high-risk beneficiaries, MSSP savings benchmarks are adjusted by beneficiary risk score. To discourage coding intensification (“upcoding”), benchmarks are not adjusted upward if risk scores rise. In Chapter II, we evaluated the impact of this risk-adjustment policy. We examined whether MSSP exposure was associated with within-beneficiary risk score change and whether risk score was associated with beneficiary entry or exit. MSSP exposure was not associated with consistent changes in within-beneficiary risk score. Conversely, highest-risk beneficiaries (99th percentile of risk) had a 25.1% chance of exiting the MSSP compared to a 16.0% chance among median-risk beneficiaries. We conclude the decision to not upwardly adjust risk score has successfully deterred coding increases but may have led ACOs to avoid high-risk beneficiaries.

Why have ACOs struggled to improve spending and quality? One possibility is ACOs have not engaged frontline clinicians. In Chapter III, we surveyed 1,620 clinicians in the Physician Organization of Michigan ACO (response rate: 34%). We found limited ACO engagement: few clinicians participated in the decision to join the ACO (3%); few were aware of ACO incentives, including knowing the ACO was accountable for spending and quality (23%) or faced upside risk only (3%); and few reported the ACO changed compensation (20%) or practice (19%) or reduced inappropriate care (13%). However, a one standard deviation increase in ACO awareness was associated with decreased difficulty implementing recommendations against low-value care (-2.3 percentage point; CI: -3.8, -0.8). We conclude limited clinician engagement may hamper efforts to achieve MSSP objectives.

Collectively, these results suggest that the MSSP has had little impact on spending and quality. Instead, selection effects – including strategic dropping of high-cost clinicians and high-risk beneficiaries – may drive estimates of improved performance. Future research should investigate how Medicare can structure incentives that motivate ACOs to lower spending, care for high-risk beneficiaries, and engage frontline clinicians.