First Glimpse at ACO Results

A cornerstone of health policy arising from health care reform is the formation of Accountable Care Organizations.  The premise is that an organizational structure focused on improving patient care and satisfaction, in a coordinated fashion, would improve quality and thus, reduce cost. The policy implication is that on a shared incentive basis, organizations at-risk, would create savings sufficient to share between the government and the ACO. In June of 2011, I wrote a post regarding my analysis of the ACO rules and their impact on the post-acute environment (http://wp.me/ptUlY-8H ).  Today, our first glimpse of how effective ACOs are/may be in producing outcomes and netting savings is available via a study released and published today in Health Affairs.

The study involved a five-year forecast or simulation of an average population of 65-75 year old, type II diabetics.  Using a tool that simulates “what” would occur with this group participating in an average ACO with representative populations, costs, and processes, synthesized outcomes were produced and evaluated.  Important to note, the tool simulated the outcomes for the set of individuals correlated to the Medicare required performance measures (blood pressure, cholesterol, glucose levels, etc.). The choice of diabetes as a disease focus is related to an easy to quantify population where known application of certain best practices and care management techniques can have a readily qualified improvement. In short, the group is a large, at-risk population with enormous potential for adverse events when care is not properly managed.  Diabetes represents a focal area of care improvement per the ACO intent under the reform law.

A key element for the ACO formation under the reform law is known as the Shared Saving Program.  Under this program, CMS specifies four domains for performance measures: patient/caregiver experience, care coordination and safety, preventative health, and at-risk populations.  For diabetes, seven of the twelve measures in the “at-risk” domain are tracked although only the composite hemoglobin A1c and poor control measures are scored by CMS for calculating the shared savings. Each of these measures is given a score by CMS and then compared against a “threshold” and a “benchmark”.  While CMS has yet to specify the limits for thresholds and benchmarks, the law provides that attainment of one or the other triggers the “shared savings” payments. In the first three years of ACO roll-out, organizations can choose between two forms of shared savings.  The first option is the organization can elect to receive 50% of the savings if costs decrease but not be “at-risk” if costs increase.  The second option provides for a 60% share of savings with the possibility of a penalty if costs increase.  By year three, the first option expires and in the interim, savings must exceed 2% before any “sharing” kicks-in.

The results of the study are intriguing and supportive of my general conclusions that the ACO concept while a good-start, can’t be terribly viable without overall reforms to Medicare.  Simply, layering a different model over-the-top of the current system is like reupholstering a broken down couch – it only looks better.  The core results showed that even a 10% improvement in the core measures for this population, while preventing up to 4% of adverse outcomes, nets savings of less than 2% (1.2% to Parts A and B).  If the costs associated with drug therapy (Part D is excluded from the ACO model) are included, the net to Medicare is zero or perhaps, a modest cost increase. In short, the likelihood of “shared savings” returns to the ACO are minimal and compared against heavy start-up costs, on average $1.7 million, a disheartening risk-return proposition.

As I have been watching this developing element of the health care landscape carefully and working on a variety of network initiatives, my conclusions are pretty consistent with the study authors, namely;

• CMS needs to look carefully at the weights assigned across the performance measures – now, equally calculated.  We know certain elements are more directly linked to savings.
• Drugs need to be factored into the shared savings program – big gains or losses can be found here.
• Thresholds and benchmarks are yet defined by CMS and they will need to look very carefully at these levels.
• Payments via shared savings need to be re-evaluated by level and delivery.  ACO development is expensive and as demonstrated, the current shared savings program provides little return on investment.
• ACOs will need to look at alternatives to produce higher volume savings and core operating efficiencies in order to make “real” money under the ACO model.
• Lastly, CMS needs to figure a methodology for a two-part payment approach – an incentive per capita for high performing organizations and then a return payment for results equal to a savings share (a bonus if you will).  The success should drive the next period payment levels rather than a “one size” fits all approach.