Accountable Care Organizations have been promoted by CMS as a way to improve care and lower costs in the fee-for-service arm of the Medicare program. These ACOs have the potential to get bonuses if they hit certain savings targets and in some cases may be at downside risk as well. Previous reports have suggested that overall the ACOs were saving money for Medicare, but a new analysis in the Annals of Internal Medicine calls those findings into question. (Annals Article) Obviously, the physicians who choose to participate in ACOs may be more confident of their ability to deliver low-cost care and thus attain financial rewards, and those physicians who are higher cost when participating in an ACO may voluntarily drop out of the program or be pushed out by the ACO. These effects would obviously make the savings look better than they actually were. In this new research, the authors adjusted for the effect of such selection factors in determining whether savings actually occurred. Data from 2008 to 2014 was used. The primary outcome was price-standardized Medicare spending per beneficiary per calendar quarter, with a secondary analysis for sub-categories of spending. Quality measures were also examined. A particular physician’s proximity to other physicians in an ACO was used as an instrument variable to test the analysis outcomes.
The researchers were most interested in whether average spending per beneficiary was associated with a particular physician’s entry into or exit from an ACO and whether the physician’s patients then also left the ACO as a source of care. They also looked at whether the percent of beneficiaries in a Medicare Advantage plan in a county appeared related to either ACO penetration or savings. In general, beneficiaries who were attributed to an ACO appeared to have somewhat lower spending prior to such attribution than did those beneficiaries who did not end up with their care provided by an ACO. Prior to comprehensive adjustment, spending in an ACO was about $120 per beneficiary per quarter lower. When physician supply and participation trends were considered, however, the difference disappeared completely. This result appeared to be correlated with exit of higher-spending physicians from an ACO. There did not appear to be an association between MA penetration and ACO participation or spending. Similarly, while the unadjusted analysis showed an improvement in quality performance, the fully adjusted model did not. The highest cost clinicians were much more likely to exit an ACO and their patients generally also ceased being attributed to the ACO. The findings suggest that, as usual, provider organizations respond significantly to financial incentives, and will take steps to maximize their financial outcomes, but those steps may not result in real savings to the system.