The Physician Group Practice Demonstration conducted by the Centers for Medicare and Medicaid Services selected ten large physician multispecialty groups and allowed them to retain up to 80% of any savings they created in caring for Medicare patients assigned to the groups, if they also made improvements on 32 quality measures. An article in the Journal of the American Medical Association reports on the results of the demonstration, which have taken on added significance since much of the rationale for the accountable care organization program is based on the PGP demo. (JAMA Article) In results released by CMS, most of the practices improved quality scores and some showed savings resulting in receipt of bonuses. According to CMS, spending was reduced by a total of $137 million over 5 years or less than $30 million a year. The authors of this article undertook their own savings analysis with various risk adjustment assumptions.
The researchers compared spending for participants in the demonstration and those in the same area but not in it. Various demographic adjustments were made and health status was adjusted by hierarchical condition categories and another risk adjuster. One comforting fact is that it did not appear that the PGP participants shifted the nature of beneficiaries they treated to game the system. Annual Medicare spending per beneficiary increase by $1206 during the intervention period in the PGP and by $1230 for the control beneficiaries. After adjustment, annual savings appeared to be about $496 per beneficiary under the HCC adjuster or around $114 per beneficiary under the other, more conservative risk adjustment method. Most of the saving appears to be in hospital care. The dual eligible population accounted for a significant part of the savings as well. While the demonstration appears to have achieved some savings, it is relatively small and did not seem to grow over time. Unless ACOs are a whole lot more effective, their ability to slow Medicare spending as much as it needs to slow should be in doubt.