While much has been written, both skeptical and positive, in the days since the Center for Medicare and Medicaid Services (CMS) released performance data for its Medicare Accountable Care Organization (ACO) participants in 2016, one thing mysteriously remains unclear: How should we define success for ACO models? If ACOs were a clinical trial, we would have pre-specified the endpoints to measure, the duration of the trial, and the inclusion criteria for participants—but we haven’t done any of these things for ACOs. How can we even interpret their performance without a shared definition of success?
Ideally, success would come in three parts:
- Beneficiaries would benefit from higher quality care and less unnecessary or avoidable utilization;
- Medicare spending would be lower, providing net savings to the Trust Funds; and
- Successful ACOs would be fairly compensated for their efforts, and earn enough to cover their costs and justify staying in the program over time.
CMS and the health policy community have not done a good job articulating what combination(s) of these three pillars could qualify as success, and how to weigh tradeoffs between them. We should also be clear about the limitations of our data to address these areas; ACOs’ performance against benchmarks is not as robust as regional differences-in-differences analyses, as others have pointed out.
Moreover, discussion of ACO results has tended to lump all ACO models together and to focus on short-term results. But today’s ACO models have fundamental differences in their design—most notably in the amount of financial opportunity and risk being borne by participants—and should be viewed separately. Equally importantly, the timelines required for organizational transformation necessitates that we take a longer-term view instead of jumping to premature conclusions.
In any trial of a new drug, different doses would be evaluated separately. There are different “doses” of ACO models being tested: some with low-dose “upside-only” contracts, others with partial shared risk, and others with high-dose capitation or complete shared risk. However, upside-only ACOs continue to outnumber Shared Risk ACOs and, as a result, have disproportionately driven the narrative surrounding ACO results. This is obscuring the more encouraging results of two-sided models.
Despite the limitations of the performance data released by CMS, clear trends are beginning to emerge.