The Opportunities and Challenges of the MSSP ACO Program: A Report From the Field
Farzad Mostashari, MD, ScM, and Travis Broome, MPH
The Medicare Shared Savings Program (MSSP) was founded in 2010 to create a vehicle through which doctors in the Medicare program could be paid more for delivering better outcomes at a lower cost, commonly called accountable care organizations (ACOs). The MSSP was established as part of a larger bipartisan movement to shift the delivery of healthcare in the United States from a fee-for-service to a value-based system—a movement that was reaffirmed most recently with the passage of the Medicare Access and CHIP Reauthorization Act (MACRA) of 2015.
Whether ACOs have been successful in delivering value has been the subject of much debate and speculation.1,2 Missing from this discussion, however, is a look at the program from the frontlines and the voices of those running MSSP ACOs. We aim to fill that gap.
In June 2014, we launched Aledade, a company with the mission to help independent primary care physicians (PCPs) form and operate physician-led ACOs. We offer providers data analytics and user-friendly technology to promote seamless care, regulatory expertise, best practices shared by a national network of hundreds of doctors, and face-to-face practice transformation support. Since our start, Aledade has formed ACOs in New York, Delaware, Maryland, Arkansas, West Virginia, Tennessee, Mississippi, Florida, Louisiana, Virginia, and Kansas, which collectively care for more than 100,000 Medicare patients.
The official results from CMS on the 2015 performance year were recently released,3 and CMS reported $429 million in total program savings. An increasing proportion of ACOs have generated savings above their minimum savings rate each year, and success appears to increase with the number of years in the program: increasing from 21% of first year ACOs to 42% of ACOs that were in their fourth year.3
There were 2 Aledade-initiated ACOs in the 2015 class. Based on CMS public data,4 the “Aledade Primary Care ACO” (APC) (with practices in Maryland, New York, and Arkansas) was in the 98th percentile of quality scores across all 327 ACOs that began in 2012 to 2014,5 but essentially broke even, with 0% savings and 0% cost increase. The “Delaware ACO” quality scores were in the 88th percentile, with a calculated cost increase of 2.5%.
In this article, we seek to provide the first detailed description of an MSSP ACO’s actions and results to help increase understanding of the challenges and opportunities facing ACOs, particularly those comprised of a network of independent practices. We will first provide the context of what the ACO implemented, alongside the impact reflected in the CMS final reconciliation reports for 2015. Use of the monthly claims and claims line feeds for more detailed investigation of our results will be the subject of a future article. We will conclude with lessons learned for ACO implementers and policy makers.
Strategies Patient engagement. Attribution is the CMS designation of which patients “belong” to each ACO. In the MSSP track 1 program, attribution is retrospective and is based on which PCP delivered more primary care services for a patient than any other PCP. We worked with practices to increase their availability and access to patients and to actively reach out to patients who had not been seen recently for annual wellness visits. Our hypothesis is that more intensive primary care should reduce more expensive specialty care and hospitalizations. Total primary care services received by our patients increased slightly in the APC and Delaware ACOs compared with national fee-for-service trends (by 2% and 5%, respectively).
As we had hoped, our ACOs were able to increase the number of patients attributed to the ACO. Our ACOs were also able to minimize the rate of patient “churn,” where accountability for a patient’s care shifts between primary care practices. The APC and Delaware ACOs saw increases of 5% and 15%, respectively, in the number of patients attributed to the ACO over the course of the year (vs 2% for the average MSSP ACO). Additionally, retention was high, with 87% and 88%, respectively, of patients who were preliminarily attributed to the ACO, based on last year’s data, being continuously enrolled by the end of 2015.
Quality improvement. All of our practices are users of certified electronic health record (EHR) technology, which we optimized with visit templates and order sets to increase consistency in provision of preventive services, screenings, and immunizations, particularly in the context of annual wellness visits. Although ACOs can earn full credit in their first year for successful reporting of their quality scores regardless of accomplishment, we believe that a focus on preventive services—particularly for cardiovascular health—will pay dividends over time, decreasing our patients' morbidity and mortality and improving longer-term cost trends.6 Consequently, we achieved rates of aspirin use for patients with ischemic vascular disease of 87%, screening and follow-up for elevated blood pressure at 90%, tobacco use screening and cessation counseling at 93%, and pneumonia vaccination status rose to 76%; all were substantially higher than applicable benchmarks.
Avoid unnecessary emergency department (ED) visits. We worked with practices to broaden access to same-day scheduling for urgent visits, improve after-hours telephone triage, and educate patients about appropriate use of the ED. In a Consumer Assessment of Healthcare Providers and Systems survey (obtained via confidential reports given to each ACO by CMS), 94% of patients reported that they could “usually” or “always” get an appointment for urgent care as soon as they needed it, and 88% reported being able to get a medical question answered as soon as they needed it when they called after hours. Rates of ED visits that led to hospitalizations fell by 5% and 4% in the APC and Delaware ACOs, respectively, while they increased by 1% in other MSSP ACOs.
Improve care transitions. Wherever possible, we established real-time notifications of hospital discharges through integration with health information exchanges, direct hospital Health Level Seven International (HL7) feeds, and optical character recognition and natural language processing of fax notifications.7 We developed a cloud-based workflow tool and training and feedback for practices to call patients within 48 hours of discharge and to see them for an appointment within 7 to 14 days of discharge. Our APC and DE ACOs saw a decrease in 30-day all-cause readmissions of 13% and 15%, respectively, compared with national benchmarks. Taken together, these interventions achieved a reduction in acute hospitalization utilization rates of 9% and 2%, respectively, compared with our benchmark.
So far, our physician-led ACOs have successfully increased primary care utilization (and revenue), and decreased ED and hospital utilization and readmissions. However, we failed to earn a return on our investment of practice time and Aledade resources in the first year of the MSSP program. In the following section, we explain what happened.
The first year of this multi-year journey toward better quality and lower costs creates 2 fundamental challenges for all ACOs. First, you cannot do everything at once; prioritization and learning is a necessary part of the first year as an ACO gains experience. Second, an ACO must learn how the rules that govern the measurement of better quality and the measurement of lower costs affect their specific population and their specific physician practices.
Specialty costs. Independent physician-led ACOs do not have to contend with the challenges of “demand destruction” that stymie hospital-led and multi-specialty ACOs,8 but they also cannot be complacent toward specialist costs. They must pay particular attention to specialist practices that have been bought and reclassified as hospital outpatient settings with facility fees that can double the cost to Medicare (and patients) of procedures and visits to specialists.9 These trends have resulted in a 5% increase in hospital outpatient costs nationwide, and increases of 13% and 7%—representing a total cost increase of 2.7% and 1.1%—in the APC and Delaware ACOs, respectively, thus erasing gains made elsewhere.
The benchmark. The MSSP program contains a compromise that creates an inaccurate estimate of the “counterfactual”—what costs would have been in the absence of the ACO. Our commercial gain-share/risk-share contracts typically use regional trends to estimate expected costs, whereas the MSSP program uses national trends. This may have helped national uptake of the program in areas with lower cost and cost trends, but it provides an inaccurate assessment of true savings. Delaware, for example, experienced 5% annual Medicare cost growth from 2012 to 2014, while Medicare as a whole experienced essentially no cost growth at all. What in 2012 had been a low-cost area, by 2014 had higher-than-average annual per-capita Medicare expenditures of $9787.
If these regional trends were projected forward, our benchmark (absent risk adjustment) would have been $10,213; if national inflation in 2015 was applied, the benchmark would have been $10,022. However, the MSSP program approach (weighted 3-year average with national inflation) produced a 2015 benchmark of only $9627—or 1.7% below 2014 levels (Figure). In total, the benchmark “headwind” for Delaware created a huge swing of 4.5% for the 2015 performance period to overcome. This misalignment of regional versus national benchmarks eventually goes away as an ACO transitions to the more accurate regional benchmarking in its second and third contract periods (years 4-9), but it lengthens the time horizon for some ACOs to see financial success.
Hospital coding. Another consequence of using national inflation versus regional trends to update ACO benchmarks is the inability to adjust for local trends in hospital coding practices. Unlike Medicare Advantage or commercial health plans, Medicare does not conduct prospective utilization management of hospital coding, and hospital billing staff and consultants can use specialized software to adjust billing codes toward higher-reimbursed diagnosis-related groups to maintain and increase revenue. The uptake of these tools takes place in different hospitals at different times. The Delaware ACO reduced hospitalizations by 2%, but hospital costs actually increased by 4%—largely due to a shift in coding (eg, pneumonia increasingly coded as “sepsis” with no bloodstream organism identified10). During this same period, hospitalizations went up 3.6% nationally, while costs only went up 1.4%. If hospital coding intensity had remained constant, total costs would have been 2.8% and 3.7% lower in APC and Delaware ACOs, respectively.
Risk adjustment. Risk adjustment is necessary to account for differences in how sick an insured pool of patients is compared with other populations and over time. However, the MSSP contains a peculiar policy that is unlike other Medicare Advantage, Next Generation ACO, or commercial ACO contracts: risk adjustment in the MSSP can decrease the benchmark, but never increase it. The current CMS public use file does not permit examination of the impact of risk adjustment on projected benchmarks, but this policy is likely to cause a systematic bias toward lower benchmarks in the MSSP. Newly enrolled and newly attributed patients are likely to have low risk scores (which are counted), but any trends toward higher risk in the continuously enrolled population (that would, naturally, tend to get sicker and more expensive over time) are not counted. The downside of our aggressive strategy of “putting your arms around your patients” is the high proportion of newly enrolled patients in our ACOs, which magnified this effect. In the APC ACO, a weighted relative risk score of 0.986 translated into 1.4% ($1.5 million) of savings removed from the ACO ledgers.
Information flows and information blocking. Much of what we accomplished was based on a foundation of information from multiple sources: clinical data extracted from EHRs, claims data from CMS, and hospital admission-discharge-transfer (ADT) event notifications empowered physicians to prioritize high-risk patients implement workflows for better care transitions. Unfortunately, our ability to do so was delayed in many practices due to the inability or unwillingness of EHR vendors to provide practices with their own clinical data in standardized formats, and by hospitals’ refusal to share ADT data with local health information exchanges—or with the PCPs of the patients they are discharging.
What are the implications of our first-year experience for Aledade ACOs? We have measurably improved patient care through more intense, engaged, and informed primary care. Yet, we must critically examine our assumptions and take steps to improve our strategy and execution. Below are our conclusions:
Delivery transformation is hard—it takes time and commitment. Although we welcome the urgency imposed by the program design, we recognize that improved quality (eg, pneumonia vaccinations and cardiovascular risk reduction) may take years to bear fruit. Furthermore, the inertial drag of habits acquired through decades of fee-for-service medicine can be hard to overcome. It takes time and painstaking initial groundwork to understand and drive change when you are impacting an entire network for care. The experience of the MSSP and the Massachusetts Alternative Quality Contract reveals continued and growing savings from accountable care structures over time.11 Furthermore, we have seen large variability between the practices in our ACOs in terms of their ability and willingness to change practice workflows. ACO governance must include transparency and holding each other mutually accountable—up to, and including, ACO termination of practices that are not yet ready for the transformation required.
Care management must be funded. Our philosophy has been to eschew centralized telephonic care management in favor of the practices owning the care management process (with assistance in training, recruiting, and tools). However, it is not possible to simply demand more from the existing practice staff in most independent primary care practices; there must be a return on investment for practices to invest in care coordinators. In 2016, we are using chronic care management fees and the ACO Investment Model to make it possible for most of our ACO practices to begin this process. Ongoing care management fees from commercial gain-share contracts, enhanced chronic care management fees as proposed in the 2017 Physician Fee Schedule, and the Comprehensive Primary Care Plus program greatly increase the financial viability of this strategy.
Changing primary care practice is not enough. Accountable care centers on primary care, and independent primary care practices are best suited to lead the movement from volume to value. However, as we have seen, we need to reach beyond our primary care core to enlist key specialists and ancillary providers as allies and affiliates. The sequencing is important, and these partnerships must be based on transparent and objective claims-based utilization patterns, as well as the clinical consensus of our primary care practices. In 2016, we are creating tiers of specialists and launching an affiliates program for engagement and inclusion of high-value specialists.
Scale matters. The Aledade network now includes an additional 5 MSSP ACOs launched in 2016, and we have submitted applications for an additional 9 MSSP ACOs for 2017. We have also established several commercial gain-share agreements that extend the work of the ACOs to non-Medicare patients.12-14 This allows us to share investments in infrastructure and learnings across a broader network. A portfolio of ACOs also works to balance out regional and random variations. In addition, increasing local scale can bring positive network effects, where each additional practice benefits the existing members of the ACO through increased data and increased ability to influence the local healthcare ecosystem, including through contracts with commercial payers and downstream suppliers. Our Delaware ACO is growing from 9000 attributed Medicare lives to nearly 25,000 lives—over 20% of the Medicare primary care market in Delaware. This will be especially important as we move toward Advanced Alternative Payment Models (AAPMs) and 2-sided risk.
MSSP policies can be improved. Using regional inflation trends to calculate benchmarks will improve the fidelity and fairness of the MSSP. The most sophisticated academic analysis of the MSSP to date15 found systematic “underestimation of savings in the independent primary care groups (relative to the hospital-integrated groups) when savings were determined from comparisons with CMS benchmarks.”16 Aledade has the resources to continue to support our ACOs over the long term, but many ACOs that are high-value, and often hospital-independent, will not be able to continue in the program in markets with strong cost headwinds. Meanwhile, ACOs that have earned savings due to the peculiarities of the program may also be less likely to continue into 2-sided risk arrangements and AAPMs. To shorten the timeline to financial success for ACOs that achieve lower costs with better quality, CMS should continually strive to make the counterfactual approach the gold standard of ACO performance. Current policies make these adjustments over a period of 4 to 9 years, but not every ACO can wait that long, and we believe that no ACO should have to wait that long.
CMS should find ways to accommodate legitimate risk score increases and adjustment to the benchmark while protecting against excesses. The current CMS policy is in reaction to what was perceived as excessive risk coding in the Medicare Physician Group Practice Demonstration.17 However, the policy pendulum has swung too far in the other direction in the current ACO rules by not allowing modifications to the benchmark except in 1 direction unfavorable to ACOs.
Medicare Program Integrity and Recovery Audit Program organizations should collaborate more closely with ACOs on identifying and responding rapidly to emerging patterns and outliers in Medicare billing that may reflect overuse or misuse of Medicare resources. The combination of claims analytics and clinical insights into individual patients by ACO providers can be a powerful and untapped resource to reduce fraud, waste, and abuse in the Medicare Program.
State and federal regulators can encourage and enforce expectations that hospitals share ADT event notifications with patients’ primary care providers via health information exchanges where operational. For example, Florida’s 1115 Managed Medical Assistance Waiver Demonstration Program requires hospitals receiving “Low Income Pool” funds to participate in the Florida Event Notification program.18
Finally, current site-based reimbursement policies create artificial arbitrage opportunities and encourage hospital consolidation and lower market competition. Site-neutral payments, as advocated by the Medicare Payment Advisory Commission,19 should be expanded to existing arrangements, as well as the prospective limitations introduced by MACRA.
We have learned that, given the right support and incentives, independent primary care practices can embrace population health and practice redesign. These efforts can begin to bear fruit in improved patient access, quality of care, and appropriate utilization in the short term. We strongly believe that the benefits of the program to patients and the taxpayer are not limited to those ACOs that received shared savings distributions. However, lack of recognition of these contributions may stifle continued innovation and physician engagement with alternative payment models. Aledade is committed to navigating these challenges and we are committed to sharing our learning so that more independent physician-led ACOs can succeed in their mission to profitably deliver better care at lower cost. We also hope that policy makers and commercial payers continue to work to remove the unintended policy headwinds ACOs must presently overcome.
The authors first acknowledge the 84 independent PCPs who led the ACO efforts. They also acknowledge Dr Mark McClellan, Dr Bob Kocher, Edwin Miller, Mat Kendall, Laura Chmar, Dr Emily Maxson, Dan Bowles (executive director, Aledade Primary Care ACO), Ahmed Haque (executive director, Aledade Delaware ACO), and Alexander Wess for their editorial inputs and their contributions in guiding the ACOs.