The Medicare program has made a big push on accountable care organization models, first with the pioneer program and more recently with the shared savings programs for ACOs. The pioneer program started in 2012, shared savings started in 2013. CMS released preliminary results for 2013 for both programs and an article in the Journal of the American Medical Association discussed the pioneer program results. (JAMA Article) (CMS Fact Sheet) Looking first at the pioneer program, originally there were 32 organizations from 18 states. After the first year, 12 quit and 9 of those went to the shared savings programs. This 36% dropout rate is likely due to either the operational hassle and cost of being in the program or poor financial results. The program generated about $147 million in savings in 2012 for CMS but apparently many organizations did not get the financial returns they hoped for. The costs of being in the program would likely offset any shared savings. Quality performance improved on all 15 measures in the first year and on 28 of 33 in the second year. In addition, the ACOs saved Medicare about $41 million. Not clear if there would have been savings if the same 32 organizations had been in for the second year as were in the first. Per capita spending growth was lower in the ACOs, 1.4%, than in general fee-for-service Medicare, 1.85%, but that is without any health status or other adjustment.
The 220 shared savings program ACOs had very mixed results. Only 53 met expectations, keeping spending $652 million below their targets in the first year, earning $300 million in shared savings payments and saving the Medicare trust funds $345 million. Another 52 ACOs kept spending below benchmarks but did not meet a minimum savings threshold. That means more than half of the ACOS did not meet the benchmark, which is why so many have dropped out of the shared savings program for the second or third year. Quality of care also improved in the shared savings program, with this group of organizations improving performance on 17 of 22 measures compared to fee-for-service Medicare providers. Both groups are likely performing below Medicare Advantage plans. And it is not clear that the program quality measures are really related to health outcomes or improvement. Most of these analyses are not conducted on a health-status or other adjusted basis. Pretty modest performance, but it is early in the program. The primary problem that needs to be worked on is the patient attribution formula, and the real issue there is the need to assign or have beneficiaries enroll in a specific ACO to get care. It seems unlikely that quality performance will improve much more and so far the savings are pretty negligible.