Performance in the Medicare Shared Savings Program After Accounting for Nonrandom Exit: An Instrumental Variable...

Area of Science:

• Health economics and policy
• Healthcare quality improvement
• Observational study design

Background:

• Accountable Care Organizations (ACOs) in the Medicare Shared Savings Program (MSSP) are linked to modest cost savings.
• Previous studies may overestimate MSSP effectiveness due to unaddressed clinician selection bias, specifically the exit of high-cost providers.

Purpose of the Study:

• To assess the true impact of the MSSP on healthcare spending and quality.
• To account for nonrandom clinician exit from ACOs to correct for selection bias in evaluating MSSP outcomes.

Main Methods:

• Utilized longitudinal models comparing MSSP ACO participants with control beneficiaries (2008-2014, 20% Medicare sample).
• Employed an instrumental variable approach using the share of nearby clinicians in MSSP to address selection effects.
• Analyzed total spending, four quality indicators, and hip fracture hospitalizations as a falsification outcome.

Main Results:

• Adjusted models showed MSSP associated with spending reductions (-$118/beneficiary-quarter) and quality improvements.
• Instrumental variable models found no significant association between MSSP participation and spending or quality.
• High-cost clinicians were more likely to exit ACOs (30.4% vs. 13.8%), driving compositional changes.

Conclusions:

• After adjusting for nonrandom clinician exit, the MSSP demonstrated no significant improvements in spending or quality.
• Selection effects, including the departure of high-cost clinicians, likely inflate the estimated savings attributed to the MSSP.