Evaluation of CMS’ ACO Predecessor Program

Before the Centers for Medicare and Medicaid Services began its accountable care organization program, it had several related demonstration projects, including the Physician Group Practice evaluation.     (CMS Evaluation)   An article in the Medicare and Medicaid Research Review analyses the results of the program, which ran from 2005 to 2010.  Like the ACO program, the PGP demonstration gave physician groups incentives for controlling spending and improving quality.   There were ten participating organizations, all having at least 200 doctors.  It was completely up to each practice how it would meet the objectives set by CMS and the participants took a variety of approaches.  Most of the practices had or through the course of the demonstration developed good electronic medical record systems and they implemented a variety of process changes to encourage greater use of evidence-based medicine and care coordination.  Most also had care management programs aimed at specific diseases or high-cost patients.  The researchers analyzed claims from 2001 to 2010 for Medicare beneficiaries who received most of their primary care at one of the PGPs.  The researchers constructed a comparison group and used a pre-post design to evaluate the effects of the demonstration.  The three primary outcomes were total Medicare spending and spending on six major categories of care, utilization of inpatient hospital and ER visits and quality measures for diabetes, heart failure, coronary artery disease and preventive care.

The intervention group of beneficiaries (the patients assigned to a PGP) had slightly higher risk scores than the comparison group, but this may be due to better coding by these large, sophisticated practices.  The average savings across all sites was $171 per beneficiary per year, or 2% of spending.  CMS paid the groups mean bonuses of $102 per person, so the average net savings were $69 per beneficiary per year.   But only six of the groups showed savings before bonuses, one with very large savings and one with very small savings.  The savings were largely achieved by reducing hospital and nursing facility use.  The most savings were seen for stroke and heart failure patients, and there were significant savings for most chronic diseases.   The most savings, about $1900 were seen for the beneficiaries in the top 10% of risk scores.  The intervention groups all did better on quality measures, but the gains were not huge, which may be due to the general attention to quality measures during this period and to these practices probably doing a better job that most physicians at the start of the demonstration.  Four key qualitative lessons the authors identified were the importance of patient engagement, the need to expand support for care management, improving care transitions and using non-physician clinicians more effectively.    Note that all of these raise costs, except the last one.   Based on the PGP experience, ACOs are unlikely to generate large cost savings for Medicare.  And remember that the PGP practices were all large sophisticated groups.  ACOs are much more heterogenous and many are newly-formed organizations with little infrastructure or management experience.