ACO Concept Wins Praise, Though Early Results Are Inconclusive

Evidence-Based Diabetes Management, October 2014, Volume 20, Issue SP13

How goes the most ambitious reform effort in the history of American diabetes management? The early evidence shows mixed results from health service providers that are teaming up to coordinate patient care and wide disparities between the best and worst teams.

The early evidence, however, is paper thin.

CMS last released outcomes data for some accountable care organizations (ACOs) in February 2014. It was 1 of only 2 announcements the government has ever made about the performance of such operations, which now treat several million Americans, and the numbers were already more than a year old. The dearth of information precludes detailed analysis of actual ACO results, leaving only a general consensus about the transformative potential of coordinated care and some anecdotal evidence about events on the ground.

“The ACO concept is extremely logical. Research clearly shows that people with diabetes and other chronic ailments enjoy dramatically better health, and consume dramatically fewer resources, when their various care providers intelligently coordinate treatment,” said Robert E. Ratner, MD, FACP, FACE, chief scientific and medical officer of the American Diabetes Association (ADA).

“We just don’t know many details about how this logical concept is working in practice. We can’t make any general statements about where ACOs have done well and where they have done poorly. We can’t say whether some types of ACO produce better results than others, or explain what separates the best and worst performers.”

“All we can really say is that the initial results were encouraging and that we’d really like to see much more data. We think greater transparency, not just in the data the government collects but also in the internal data from the ACOs themselves, would benefit everyone by helping best practices spread far more quickly,” added Ratner. The Affordable Care Act actually provides for several different types of ACOs that are variations of the same general idea: healthcare providers that work together to coordinate wellness programs for chronically ill patients get 50% to 60% of the difference between expenditure benchmarks and actual outlays.

Unlike the traditional fee-for-service model, which pays caregivers more when patients get sick, the ACO payment model provides caregivers a financial incentive to keep patients healthy. To keep those caregivers from reducing expenditures simply by skimping on care, CMS has created 33 quality measures for Medicare ACOs, and private insurers have created their own metrics for commercial ACOs.1 During their first year of operation, providers need only report the percentage of patients that meet each target. Subsequently, providers that want to keep sharing in “savings” must demonstrate that they’re meeting targets on each metric for ever increasing percentages of their patients.

The initial word on ACO performance came last summer, when CMS reported the first-year results for 32 organizations that participated in Medicare’s “Pioneer” program.2 All of the Pioneers had begun experimenting with the coordinated care model before CMS created its programs—an experience that made those groups an ideal model to prove the benefit of an ACO—and most of the Pioneers came through.

Costs for the more than 669,000 beneficiaries aligned to Pioneer ACOs grew less than half as much in 2012 as costs for similar patients with traditional care. Nearly half of the Pioneers beat cost targets enough to share in “savings” that totaled $87.6 million, and all but 7 of the 32 saved some money.

Better still, the Pioneer ACOs outperformed fee-for-service organizations in all 15 quality measures for which direct comparison was possible. Among diabetics, for example, the median rate of cholesterol control was 57% for ACO patients but only 48% for others.

Early this year, however, CMS released the first numbers from its larger group of ACOs— the Medicare Shared Savings Program—and the picture was decidedly mixed.3 Only 54 of the 141 ACOs in that program managed to keep costs below their targets, and only 29 of them kept costs low enough to share in savings. Most of the participating ACOs actually billed more for patient care than a typical fee-for-service provider. CMS also released data on how the ACOs in the Shared Savings Program performed on 5 quality measures, 4 of which gauged diabetes care: aspirin use, tobacco non-use, glycated hemoglobin (A1C) less than 8%, and blood pressure less than 140/90 mm Hg.4

Overall, ACO performance exceeded comparable numbers from fee-for-service providers, and on average, the groups in the program met each target for 65% to 75% of their patients. The gaps between the best performers and the worst performers, however, were dramatic. An ACO in Wisconsin called Bellin-ThedaCare Healthcare Partners reported that 84% of its patients kept their A1C levels under 8%. Accountable Care Coalition of Maryland, on the other hand, reported that only 24% of its patients could maintain that target. Success rates on the blood pressure target ranged from 33% to 88%.

Success rates for tobacco nonuse were similar, but the variations in aspirin use were even more extreme. Several plans reported that all of their diabetic patients used aspirin, while several others reported that less than 30% did. Those numbers may overstate the actual size of the gaps somewhat. Some of the lowest-scoring ACOs have announced that failure to collect data for some patients (and, in one case, failure to appropriately tabulate percentages5) deflated their scores. Still, the true gaps were substantial.

“The disparities from the Shared Savings plan make sense when you consider the huge variety among the ACOs that are participating,” said Brett Erhardt, a director with a research and consulting firm called The Advisory Board. “Some of them had been experimenting with coordinated care for years, but most had little to no experience with it. They were dipping their toes in the water and testing different concepts. Big gaps were to be expected across the board. Gaps were particularly natural on the quality measures because the numbers came from the program’s first year, when ACOs have a reporting requirement but no performance targets. They were, quite rationally, focused on other matters,” he added.

Another likely explanation for at least some of the performance gaps—or possibly even the majority of them—is variance among the patient populations of different ACOs. Aggregate differences in age, education, income, and other demographic factors correlate strongly with differing propensities to keep using tobacco, quit using medications, or otherwise confound efforts to measure caregiver quality.6

Some caregivers think it’s unfair to judge them on measures not completely under their control, and CMS is evaluating those complaints—looking for measures that everyone can support. Indeed, the agency has spent years working to settle disagreements, align conflicting goals, and build something approaching a consensus for its method of evaluating ACO performance.

The current system, then, is an attempt at compromise: a compromise between those who support fee-forservice and those who support population-based payments, a compromise between those who believe doctors should be measured by what they do and those who believe they should be measured by the outcomes they partially produce, and a compromise among countless

other warring ideals.

Indeed, the diabetes measures illustrate the difficulty of the task. Even now, after years of work, the ADA takes issue with 1 of the 4 official measures. CMS wants caregivers to get A1C levels below 8% for all patients, but the ADA thinks a significant minority of patients fare just as well with higher levels. Still, CMS has succeeded in creating a surprising amount of consensus by

proving itself willing to consider complaints and tweak its system. For example, the agency originally proposed twice as many quality measures, but slashed the list following objections over duplicate metrics and excessive reporting costs.

Despite several years of work, it’s a work in progress. Just this past July, CMS added 4 new quality measures, eliminated an existing measure, and tweaked 2 others.7 Additional consensus-building efforts will probably result in further changes to the quality measures and may even lead CMS to adjust its performance targets on those measures. Although the agency factors patient demographics into its cost targets for each ACO, it currently does not consider demographics in setting performance targets. All ACOs must meet the same performance targets to share in savings.

Some observers expect CMS to concede the effect of demographics and begin customizing quality targets, while others expect it to eliminate targets entirely and demand a yearly improvement from each ACO instead. CMS has not announced any plans on additional data-sharing on ACO performance, but the information that it does provide is likely to be at least 6 months old, simply because it takes at least that long for all claims to be processed and any savings to be tabulated.

Everyone recognizes the complexity of gathering considerable information, analyzing it, and making it public in anything resembling a timely manner; but many still think that if the ACO experiment is to provide maximum benefits to patients and taxpayers, CMS must share more. The president’s own Council of Advisors on Science and Technology believes that sharing information ranks among the 6 most important strategies for improving healthcare.8 “Communicating the lessons learned can help those starting system-improvement efforts.”

With little hard data on hand, observers must cobble together anecdotal evidence and independent research to combine with the 2 releases from CMS to estimate the general strengths and weaknesses of existing ACOs, and to surmise the difference between the best and worst performers. The evidence they see suggests that nearly all ACOs, even the newest among them, have achieved the most important initial goal: identifying the neediest patients, the ones who end up in hospital several times a year because they cannot manage their conditions.

Research suggests that this small group of patients consumes much of the money spent on chronic illness each year, and thus represents much of the opportunity for improving health and cutting costs. The potential for significant improvement certainly exists. One study of a coordinated care program launched more than a decade ago by the Veterans Health Administration found that it reduced the hospitalization rate of diabetics by 50% and reduced the average stay for people who were hospitalized by 3 days.9

Significant results of this order have generated widespread enthusiasm for coordinated care programs, but many experimental programs have failed to justify that enthusiasm. A large trial in Australia, for example, managed to produce moderate improvements to patient health, but overall expenditures rose significantly.10

For many new ACOs, though, the biggest initial challenges seem to be technological. Coordinated care only works when information flows among all the caregivers on the team, even when all those caregivers work at different practices, with different initial software, and with different initial work flows. “Most caregivers have struggled for years to make technology improve communication and increase efficiency at individual practices. Figuring out best practices for connecting multiple practices with health information exchanges and population health management software is clearly a work in progress,” said David B. Muhlestein, PhD, director of research at Leavitt Partners, LLC.

ACOs made up entirely of caregivers who work for a single hospital or network of hospitals naturally enjoy the advantage of compatible software, and that’s not their only advantage. Large organizations have larger capital budgets; some experience coordinating care between different types of doctors and clear chains of command to settle disputes that arise. Hospital ACOs may also enjoy another advantage over ACOs made up of smaller independent practices: a greater ability to get patients to buy into the team concept.

“Patients have always thought about choosing individual caregivers rather than care teams. They find one doctor for this and one doctor for that, and they get comfortable with those doctors, and they won’t change easily,” said Muhlestein.

“Caregivers who work for ACOs will need to change that mind-set by explaining why teams can provide patients such better care than unconnected individuals. Getting patients to buy into the concept will improve everyone’s numbers, and improving numbers will get more patients to buy in. It’s a positive cycle, but one that will take years to play out,” Muhlestein added.References

1. ACO Shared Savings Program Quality Measures. CMS website. http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/Downloads/ACO-Shared-Savings-Program-Quality-Measures.pdf. Accessed August 22, 2014.

2. Pioneer Accountable Care Organizations succeed in improving care, lowering costs [press release]. Baltimore, MD: CMS Newsroom Center; July 16, 2013. http://www.cms.gov/Newsroom/MediaReleaseDatabase/Press-Releases/2013-Press-Releases-Items/2013-07-16.html.

3. Performance year 1 interim results for ACOs that started in April and July 2012. CMS website. http://www.cms.gov/Medicare/Medicare-Feefor-Service-Payment/sharedsavingsprogram/Downloads/PY1-InterimResultsTable.pdf. Accessed August 22, 2014.

4. ACO quality data. Data.Medicare.gov website. https://data.medicare.gov/Physician-Compare/Accountable-Care-Organization-ACO-Quality-Data/ytf2-4ept. Accessed August 22, 2014.

5. Rau J. Medicare data show wide differences in ACOs’ patient care. Kaiser Health News website. http://www.kaiserhealthnews.org/Stories/2014/February/21/Medicare-Data-Shows-Wide-Differences-In-ACOs-Patient-Care.aspx. Published February 21, 2014. Accessed August 22, 2014.

6. Socioeconomic position and health among persons with diabetes mellitus: a conceptual framework and review of the literature. Epidemiol Rev. 2004;26(1):63-77.

7. Sullivan K. CMS proposes additional quality measures for ACOs for 2015. The American Journal of Managed Care website. http://www.ajmc.com/focus-of-the-week/0714/CMS-Proposes-Additional-ACO-Quality-Measures-for-2015. Published July 7, 2014. Accessed August 22, 2014.

8. Cassel CK, Saunders RS. Engineering a better health care system: a report from the President’s Council of Advisors on Science and Technology [published online July 14, 2014]. JAMA.doi:10.1001/jama.2014.8906.

9. Chumbler NR, Neurgaard B, Kobb R, Ryan P, Qin H, Joo Y. Evaluation of a care coordination/home-telehealth program for veterans with diabetes. Eval Health Prof. 2005;28(4):464-478.

10. Battersby MW. Health reform through coordinated care: SA HealthPlus. BMJ. 2005;330(7492):662-665.