Performance of Medicare ACOs through 2013

Background: In the Medicare Shared Savings Program (MSSP), participating ACOs share in savings with Medicare if they keep spending for an attributed population of fee-for-service beneficiaries sufficiently below a financial benchmark.  Unlike Pioneer ACOs, MSSP ACOs are not required to assume downside risk for spending in excess of benchmarks in initial contracts.  To set financial benchmarks, CMS uses national rates of spending growth to update spending for patients served by an ACO during a baseline period to a given contract year.  Because spending growth varies substantially across markets, savings or losses determined from comparisons with CMS benchmarks may be misleading, particularly when compared across geographically different subgroups of ACOs with different traits, times of program entry, or incentives. 

Methods: We used a difference-in-differences approach to compare MSSP ACO performance in 2013 (the first full year of the MSSP for providers entering in mid-2012 or 2013) with a counterfactual established by changes in performance among local non-ACO providers.  Specifically, we used linear regression to compare changes in Medicare spending and claims-based quality measures from the pre-contract to post-contract period between beneficiaries attributed to ACOs vs. non-ACO providers, adjusting for geography and beneficiaries’ sociodemographic and clinical characteristics.  We compared ACO savings by: organizational structure; baseline efficiency (assessed in 2008 to mitigate regression-to-the-mean); and concurrent entry into commercial ACO contracts.

Results: Spending levels and trends were similar for ACO and non-ACO providers in the pre-contract period.  The first full year of MSSP participation was associated with modest savings among ACOs entering in 2012 (1.4%) but not among those entering in 2013.  Savings in the 2012 cohort were only slightly lower than updated estimates for Pioneer ACOs (1.6-1.8% for years 1-2) and were driven by a 1.4% reduction in inpatient spending, a 2.1% reduction in hospital outpatient department spending (with an offsetting increase in office-based spending), and a 6.1% reduction in post-acute spending on skilled nursing facility care.  Aggregate savings in the 2012 cohort of MSSP ACOs ($238 million) were less than the shared-savings bonuses paid by Medicare ($248 million).  In contrast to savings determined from comparisons with benchmarks, our estimated savings were larger for independent primary care groups than for organizations vertically integrated with hospitals and larger for ACOs with higher baseline spending.  Savings were not systematically related to commercial ACO contracting.  Differential changes in claims-based quality measures suggested unchanged or slightly improved performance.  

Conclusions: One-sided ACO contracts may elicit successful efforts to lower spending, but modest savings do not accrue to Medicare in the absence of downside risk, and early gains observed among early MSSP participants may not generalize to later participants.  Under currently weak contract incentives, meaningful savings from the entire MSSP develop slowly as less advanced participants develop greater capacity to manage care, but rule changes could strengthen incentives.  Independent primary care groups, which have stronger incentives to limit inpatient and outpatient hospital care, achieved greater savings; early efficiencies from vertical integration were not evident.  Benchmarking policy must consider the importance of participation by ACOs with high baseline spending.