Medicare ACO Performance

The Centers for Medicare and Medicaid Services is locked in on proving the value of its Accountable Care Organization initiative.  The 20 “Pioneer” ACOs (who have to feel like it’s been a long rutted trail beset with hostile savages (i.e. government bureaucrats)) have shown great results, according to CMS, and the 333 regular Medicare Shared Savings Program ACOs aren’t far behind.   (CMS Statement)   In 2014 total savings from the program were $411 million and 97 ACOs got shared savings payments of $422 million in aggregate.   The Pioneer ACOs covered 622,265 beneficiaries, with 15 of these generating $120 million in savings, up 24% from the prior year.  3 Pioneers owe CMS a total of $9 million due to excess losses.  So just as happened this year, next year there will likely be 5 to 9 fewer participants in the Pioneer program.  Scores on quality measures were basically the same as in 2013, but are high, so a lot of improvement shouldn’t be expected.

The shared savings program ACOs had mixed results, as only 92 saved enough to get a shared savings payment.  Another 89 reduced health costs below the benchmark, but not enough to qualify for a payment.  If my math is right, that means that the remaining 152 didn’t save Medicare any money.  Expect many of those to drop out as well. Quality also showed improvement among the shared savings ACOs.  CMS claims that ACOs with more years in the program perform better, but this is clearly just a result of survivor bias–the ACOs that weren’t generating savings have dropped out.  And most of the ACOs are dissatisfied with the program, partly because of its failure to generate money for them and partly because of design factors like failure to require patients to be assigned to the ACO for care.  The reality is that this program is producing modest results at best, and CMS’ analyses aren’t adjusted for health status or other factors relating to the beneficiaries, and don’t take into account the likelihood that the most successful providers probably would show these results even if they weren’t an ACO.  I suppose it’s better than nothing, but hardly game-changing.