The Medicare Shared Savings Program (MSSP) does not improve spending or quality when adjusted for the nonrandom exit of high-cost clinicians, according to a study published in the Annals of Internal Medicine.

The investigators compared accountable care organizations (ACOs) participating in the MSSP vs control beneficiaries between 2008 to 2014. Total price-standardized Medicare spending, 4 quality indicators, and hospitalization for hip fracture (serving as a falsification outcome) were evaluated for both the ACO and control groups, and were adjusted for secular trends, market factors, and beneficiary characteristics. The MSSP supply, defined as the ratio of MSSP clinicians to all nearby clinicians, was used as an instrumental variable in the adjusted longitudinal model. The investigators further tested whether compositional changes among MSSP participants drive performance estimates in spending.

In adjusted longitudinal models, the MSSP was associated with reductions in spending of $118 per beneficiary per quarter (95% CI, -$151 to -$85) and demonstrated improvements in all quality indicators. In this model, savings were attributed to reductions in inpatient and outpatient services and skilled-nursing facility services. Using the MSSP supply as an instrumental variable, researchers found that the MSSP was not associated with significant changes in spending ($5 change; 95% CI, -$51 to $62) or quality. 

Continue Reading

In the adjusted model, the MSSP was associated with a substantial decrease in hip fracture outcomes (-0.24 hospitalizations for hip fracture per 1000 beneficiary-quarters; 95% CI, -0.32 to -0.16.), but not in the instrumental variable model (0.05 hospitalizations for hip fracture per 1000 beneficiary-quarters; 95% CI, -0.10 to 0.20). 

Related Articles

The nonrandom exit of high-cost clinicians drove ACO compositional changes. High-cost clinicians (99th percentile of average spending) had a 30.4% chance of leaving the MSSP in the next year compared with median-cost clinicians (50th percentile), who had a 13.8% chance.

Limitations to the study included its observational design and use of administrative data, in which unobserved factors may have confounded the relationship between MSSP supply and performance. Instrument variable models typically generate estimates that are less precise than adjusted modeling; changes to hip fracture hospitalization may represent improvements in beneficiary health. Growth in Medicare Advantage participation may affect MSSP evaluations over time.

After adjusting for the nonrandom exit of clinicians, the MSSP was not associated with improvements in spending, quality, or falsification outcomes. The researchers of the study concluded that the exiting of high-cost clinicians and their patient panels may be a major factor driving savings in the MSSP.


Markovitz AA, Hollingsworth JM, Ayanian JZ, Norton EC, Yan PL, Ryan AM. Performance in the Medicare Shared Savings Program after accounting for nonrandom exit: an instrumental variable analysis [published online June 18, 2019]. Ann Intern Med. doi:10.7326/M18-2539