On August 25, the Centers for Medicare and Medicaid Services (CMS) released financial and quality performance data for its accountable care organization (ACO) programs, including results from the third performance year (PY3) of the Pioneer ACO Program and second performance year (PY2) for the Medicare Shared Savings Program (MSSP). CMS reported that the 20 Pioneer ACOs and the 333 MSSP ACOs significantly improved quality of care in most measured dimensions and generated more than $411 million in net program savings in 2014.
A total of 97 ACOs had actual spending far enough below their benchmark, and did well enough on measured quality, to qualify for more than $400 million in shared savings payments. The overall reduction in spending has appropriately been characterized as modest, and many steps have been proposed by a wide range of commenters that could lead to more savings in the programs, some of which are being implemented by CMS. In this post, we provide further analysis of the latest Pioneer and MSSP results, and then discuss some implications for the future of Medicare’s accountable-care programs.
Overview Of Pioneer ACO PY3 Results
The Pioneer ACO Program is a Medicare demonstration for organizations with substantial experience in managing population health that aim to transition from Medicare fee-for-service (FFS) payment to being paid at least in part on a population basis; they bear “downside” repayment risk and are eligible for “upside” shared savings. The program began with 33 participants in 2012, declining to 23 organizations by the end of 2013 and 20 organizations by the end of 2014. An additional participant has since left the program, bringing the current number to 19.
The remaining Pioneer organizations have mostly achieved savings: as Exhibit 1 shows, three-quarters of ACOs in the Pioneer Program (15/20) were able to reduce their costs relative to their financial benchmark in 2014, with more than half (11/20) reducing costs significantly enough to earn shared savings. However, five ACOs had spending higher than their benchmarks, and three of those had high enough spending relative to their benchmarks that they were subject to a repayment penalty.
Exhibit 1. Pioneer Financial Results In Performance Year 3
Pioneer ACOs are continuing to improve the quality of care they deliver to beneficiaries, with improvement on 28 of 33 quality measures and an average improvement of 3.6 percent across all quality measures compared to Performance Year 2 (PY2). Particularly large improvements occurred in medication reconciliation (70 percent to 84 percent); screening for clinical depression and follow-up planning (50 percent to 60 percent); and qualification for an electronic health record incentive payment (77 percent to 86 percent).
As Exhibit 2 shows, and as CMS highlighted in its announcement, the average savings per ACO and quality score are increasing over time for the Pioneer ACOs, leading to an increase in total savings per year despite the decline in participating ACOs.
Exhibit 2. Average Savings And Quality Scores For Pioneer ACOs, Performance Year 1 To Performance Year 3
Exhibit 3 shows the distribution of Pioneer ACO financial performance behind the total savings relative to benchmarks in Exhibit 2, highlighting that the losses of ACOs in the program have declined with the departure of some Pioneer ACOs, while the savings relative to benchmark have actually increased slightly. (The corresponding data underneath each bar represents the number of ACOs that generated savings, earned shared savings, generated losses, and owed repayment to CMS.) These results suggest that the significant possibility of shared losses relative to benchmarks was the driving factor for some ACOs leaving the program, with the remaining ACOs increasingly likely to achieve savings relative to benchmarks.
Exhibit 3. Distribution Of Pioneer ACO Savings And Losses By Year, Millions Of Dollars
Source: Author analysis of publicly available MSSP financial data
While real risk of losses appears to be the motivation for many of the organizations to leave the Pioneer program, many remain committed to a shift from FFS to accountable care payment. They have chosen other options that involve either less risk or different rules for calculating savings.
Some have transitioned into the MSSP where they face only the potential for upside shared savings — an example is the University of Michigan launching the Physicians Organization of Michigan (POM) ACO. Others such as Sharp Healthcare have shifted their efforts to bearing more financial risk under the different benchmark structure in the Medicare Advantage (MA) program. Dartmouth-Hitchcock Medical Center, one of the Pioneer ACOs that had shared losses, recently announced that it would be departing the Pioneer program while remaining committed to improving population-based care delivery and payment reform to support it. We return to potential future developments in payment policies for Medicare ACOs below.
Overview Of MSSP PY2 Results
CMS reported that the MSSP increased total savings for Medicare in performance year 2 to $465 million, compared to $383 million in 2013. In contrast to the Pioneer program, the average savings per ACO declined while the total number of participating organizations increased substantially. Measured quality continued to improve in most dimensions: MSSP ACOs reporting in both 2013 and 2014 improved on 27 of 33 quality measures, including patients’ ratings of clinicians’ communication; beneficiaries’ rating of their doctor; screening for tobacco use and cessation; screening for high blood pressure; and Electronic Health Record use.
Exhibit 4 summarizes the financial results at the ACO level. Although 92 ACOs reduced their spending enough to qualify for shared savings, six of these ACOs failed to satisfactorily meet quality reporting requirements and consequently left shared savings on the table ranging from $900,000 to $5 million. The other 86 ACOs that qualified for shared savings received payments of more than $341 million, ranging from $660,000 to $22.7 million with an average of approximately $4 million per ACO.
Only three ACOs participated in the two-sided risk track in 2014, and all had lower spending than their benchmarks. Two of them saved enough to qualify for shared savings.
Exhibit 4. MSSP Financial Results in Performance Year 3
The number and share of ACOs reducing costs in PY2 increased slightly from PY1, from 118/220 (53.6 percent) in 2013 to 181/333 (54.4 percent) in 2014. Likewise, the number and percentage of ACOs reducing costs enough to earn shared savings increased in PY2, from 52/220 (23.6 percent) in 2013 to 86/333 (25.8 percent) in 2014. Thus, to date, more than half of ACOs in the MSSP have been able to reduce costs relative to their benchmark, and approximately one-quarter have been eligible for shared savings.
CMS reported that ACOs entering the program in 2012 or 2013 were more likely to generate shared savings by the second year (37 percent) than ACOs entering the program in 2014 (27 percent). This suggests that more ACOs are learning how to succeed in the program over time.
No ACOs facing downside risk in Track 2 of the MSSP had significant enough losses to have to pay back CMS in performance year 2. However, many MSSP ACOs do not appear to be on a course toward shared savings even after the second year. The low attrition for these ACOs, in contrast to the Pioneer program, may be because of the absence of mandated downside risk in the program to date; while upside-only risk encourages more participation, it also encourages more ACOs to try out and remain in the program without a clear expectation of savings. However, under new rules for 2016 more ACOs will be required to move to downside risk after their first six years in the program.
Exhibit 5 summarizes the results in terms of savings and losses. While total program savings are increasing, as just noted the Exhibit also reflects the addition and persistence of a significant number of ACOs in the MSSP who do not appear to be improving costs relative to their benchmarks. (The corresponding data underneath each bar represents the number of ACOs that generated savings, earned shared savings, generated losses, and owed repayment to CMS.)
Exhibit 5. Distribution Of MSSP ACO Savings And Losses By Year, Millions Of Dollars
Source: Author analysis of publicly available MSSP financial data
As Exhibit 6 shows, there continues to be almost no correlation between overall quality scores and savings among the MSSP ACOs, as we previously noted in reviewing the first-year results. While some ACOs are succeeding in improving quality and lowering costs, most organizations have not yet found a clear path to doing both at the same time. There is no clear or automatic relationship between higher quality and lower spending; achieving both is not so easy. These results reinforce the importance of an improved understanding what does predict better-quality and lower-cost performance. We return to that issue below.
Exhibit 6. Overall Quality Performance Score vs. Savings Rate For MSSP ACOs
Note: Exhibit does not include any 2014 MSSP starters (Under Pay For Reporting), nor does it include eight 2013/2014 ACOs failing to fully report quality scores.
Declining Average Shared Savings In The MSSP And The Modest Role Of ‘Quality Gates’
Unlike the Pioneer Program in which the average shared savings per ACO has increased year to year, average shared savings for MSSP ACOs declined from the first to the second years, from nearly $6 million in 2013 to $4 million in 2014.
The sharing rate for ACOs in PY2 and beyond is reduced as quality scores drop below 90 percent, and some have claimed that these “quality gates” are substantially impacting shared savings. However, the PY2 results do not suggest that the reduction in year 2 savings per ACO is due to this feature of the program. ACOs would obviously get more savings if Medicare set a higher shared savings rate, but those that are sharing in savings have not been losing much because of their quality scores.
Of the 64 ACOs in track one that qualified for shared savings—eligible for up to 50 percent of the total savings—the sharing rate ranged from 34.84 percent to 49 percent, with an average of 43.7 percent. The two two-sided risk ACOs that qualified for shared savings—eligible for up to 60 percent of savings—had final sharing rates of 54.7 percent and 56 percent. Altogether, ACOs earned 88 percent of their potential shared savings. Thus, other factors are primarily responsible for the decrease in shared savings per ACO.
One explanation is the change in the mix of participating ACOs, as a large number of additional ACOs entered the program in 2014. ACOs that had been in the program longer, joining in 2012 or 2013, were more likely to earn shared savings in PY2 than PY1 (65 to 52), and average shared savings per ACO were larger for the ACOs that had participated in the program for more than a year ($4.2 million vs. $3.2 million). Indeed, 30.3 percent of ACOs (65/214) with a year or more of experience earned shared savings, relative to only 17.6 percent (21/119) of ACOs in their first year of program participation. Furthermore, ACOs with a year or more experience were more likely to have a net reduction in spending (57.9 percent vs. 47.9 percent) and had higher average savings across all participants ($1.3 million vs. $40,200). However, while a growing share of ACOs are achieving savings and improving quality over time, not all are doing so: of the 52 ACOs that earned shared savings in PY1, 37 (71 percent) continued to achieve shared savings in PY2, and many ACOs did not share in savings in either year.
That said, many of the top ACO financial performers through 2013 also did well in 2014, as Exhibit 7 shows. In particular, Winchester Community ACO reduced spending relative to its benchmark by the greatest percentage in both years. An additional four ACOs (Integral Healthcare, Paradigm ACO, Rio Grande Valley ACO Health Providers, and Accountable Care Options) were top 10 percent savers in both years, and four other ACOs were in the top 20 percent of savers in both years. In other words, half of top 20 percent savers remained in the top 20 percent for both years.
Further, all top 20 percent ACO savers in 2014 had earned some level of shared savings in 2013, and only two ACOs beginning in 2014 were in the top 20 savers in their first year. Conversely, only one ACO that was a top 20 percent saver in 2013 had increased costs in 2014 (Catholic Medical Partners-Accountable Care IPA). Exhibit 7 below shows the change in percent savings from PY1 to PY2 for all of the top 20 shared savers in PY1.
Exhibit 7. Change In Savings For Top 20 Percent MSSP Savers From 2013
What Marks Successful MSSP ACOs?
It’s more than higher benchmarks. Thus, while it is premature to draw too many conclusions from data spanning only two years, the early results suggest that some ACOs have developed capabilities to achieve savings consistently over time. What are some of the characteristics associated with success?
In their recent blog post, David Introcaso and Gary Berger noted that physician-based ACOs performed better than hospital-based ACOs, and those with rural health clinics or federally qualified health centers performed even better. Going along with these findings, while there are many ACOs that have succeeded and many that have not at every size level, smaller ACOs continue to perform somewhat better on average than their larger counterparts. Exhibit 8 shows that ACOs with fewer than 6,500 attributed patients averaged savings of 1.5 percent, compared to just 0.50 percent savings for ACOs with over 20,000 attributed patients. Although this is not a statistically significant difference (p=0.24), this does suggest that smaller ACOs have early advantages over their larger counterparts in being able to achieve savings in the MSSP.
Exhibit 8. Average Savings Rate By Number Of Beneficiaries
Introcaso and Berger also report that having a high financial benchmark is the “primary” determinant of ACO success at reducing costs and earning shared savings. ACOs earning shared savings had a 9 percent higher financial benchmark on average, and as Exhibit 9 shows, there is a highly significant positive relationship (p=0.0003) between financial benchmark and percent savings.
At the same time, an ACO’s per-capita benchmark accounts for only a small share of actual savings or losses (e.g., r2 =0.0364 in a linear model). The large variation in performance not explained by the benchmark level suggests that other factors may have much more of an impact, as we also found in our analysis of last year’s results. For example, the ACO with the largest per capita benchmark ($21,545.73), Akira Heath of Fresno, also had the greatest financial loss relative to the benchmark (-13.39 percent). Likewise, Winchester Community ACO, which had the greatest savings rate (16.03 percent), had a benchmark of $10,904, only slightly higher than the average across all participants ($10,825).
Exhibit 9. Relationship Between MSSP ACO Financial Benchmark and Savings
Although financial benchmarks matter, additional research—including more comprehensive quantitative evaluations as well as case evaluations—is needed to illuminate what accounts for the much better performance of particular ACOs, and how additional ACOs can improve more rapidly from year to year.
Some Policy Implications
It’s still early, but the ACO results to date have implications both for health care organizations seeking to improve care through the Medicare ACO program, and for Medicare’s further approach to the program. Most importantly, the results suggest that it is possible for ACOs to succeed, and many are succeeding in both Medicare programs, but the path to success is not obvious or easy to replicate at this point.
Creating A Clearer Path To Success
A better understanding of how to succeed is needed; while it is possible to do many things that may lead to improvements in quality or reductions in spending, doing so in a way that translates into both improved quality and significant spending reductions within a year or two is not straightforward. Achieving a better understanding of how to accomplish this is among the aims of a range of collaborations that aim to facilitate shared learning from the growing range of ACO experiences.
Further policy steps are also needed to increase the predictability of success in the program and to help organizations assess the business case for making the needed investments in care reform to achieve that success. CMS has finalized new regulations with a range of changes to its ACO programs that will take effect in 2016. Among the changes are steps intended to give ACOs more predictability about their benchmarks, greater flexibility on financial terms such as minimum savings rate (MSR) and minimum loss rate (MLR)—the amounts by which spending must fall below or above benchmarks before ACOs share gains or losses—and additional tools to improve care, especially in ACOs that take on downside financial risk.
We and others have proposed additional ACO reforms. For example, success rates were higher in the programs (Pioneer ACOs and the currently very small MSSP Track 2) that involve downside risk, and new CMS policy steps are intended to encourage more ACOs to shift further away from fee-for-service payment. Organizations taking on downside risk will not only receive a higher rate of shared savings; they will also be offered new business models that depend less on fee-for-service payments, enabling them to redirect more resources to steps that improve care and lower costs and which are not financially sustainable under traditional payment models. With the introduction of another risk-bearing track in the MSSP and the forthcoming Next Generation ACO program from the Center for Medicare and Medicaid Innovation, Medicare will provide more of a pathway to bearing risk.
However, without better predictability about the likelihood of success, it is understandable that some providers are reluctant to move further away from fee-for-service payments. The transition to greater risk can be rendered more predictable through better and faster supporting data, better insights about what successful organizations are doing right, and more alignment in program design across payers. Indeed, many private insurers are implementing more significant shifts away from fee-for-service, accompanied by greater support and predictability in achieving success.
In addition, as noted by Introcaso and Berger, ACOs need as much predictability as possible about how their ability to succeed as a Medicare ACO will be affected by upcoming CMS payment reforms, such as the Comprehensive Care for Joint Replacement model and potential upcoming alternative payment models for physicians. Depending on how they are implemented, such reforms can be synergistic with ACO reforms, or they may compete with or undermine them.
Reforming Medicare ACO Benchmark Methods
As the Medicare ACO programs mature, how to set an ACO’s benchmark level and how to share savings and risk also need to change. Since the Pioneer and MSSP programs started as voluntary reform programs, CMS has been understandably reluctant to use benchmarks other than those based on past performance, lest organizations opt in based simply on a calculation that they can get paid more without changing their behavior. As the program grows and as ACOs achieve better quality and spending results than non-participating organizations, however, it is critical for them to have a reward for doing so.
One option now available to effective ACOs that are concerned about their benchmark level is to shift to the Medicare Advantage program, but this is not viable for most organizations without considerable experience in managing financial risk. CMS has noted that the Next Generation ACO pilot program and future rulemaking will provide a clearer path for ACOs away from their historical performance levels and toward regional or other benchmarks, and CMS has also noted that more support is coming for organizations willing to take on more financial risk.
Progressing Toward Two-Sided Risk
Providing more shared savings for organizations that take on two-sided risk, as CMS has started to do, would also likely increase the ACO program’s impact over time. For example, 87 one-sided risk ACOs had losses in excess of 2 percent relative to their benchmark, and Medicare would have achieved substantially more net savings in the MSSP if all of the participating ACOs rather than just a few had been in Track 2 facing “downside risk.”
Of course, ACOs that are uncertain about how to succeed in reforming would not have chosen to participate in such circumstances. It makes sense to give ACOs a significant period of time to try to develop confidence in their abilities to move away from FFS without facing additional downside financial risk. At the same time, a long-term one-sided risk program, especially for larger organizations, may limit the incentives for organizations to shift from FFS; it may also limit the amount of savings that CMS can share with ACOs while still having confidence that the overall program reduces spending. More overall savings will likely be possible if the upside-only ACO option does not remain a place to stay for all but the smallest ACOs, particularly if the path to success—with regard to both CMS policies and what ACOs need to do—becomes clearer.
Summing Up
The latest Medicare ACO results in the Pioneer and MSSP programs provide some grounds for optimism, but also reinforce certain concerns. On the one hand, the quality of care delivered to beneficiaries through these ACO programs continues to trend positively, and the Medicare program as well as beneficiaries and health care organizations are also getting some savings. On the other hand, savings remain relatively modest especially in the MSSP, with only a subset of ACOs earning significant savings even after two years.
With further refinements in ACO programs in both the public and private sectors, ACOs are on track to remain a key component of health care delivery and payment reform. The growing experience should enable a better understanding of the steps that ACOs and policymakers can take to achieve better health care and lower costs.
No comments:
Post a Comment