“Making the Numbers Work”: Henry Ford Leaders Explain ACO Critical Success Factors, Step by Step
What is it like to actually build an accountable care organization (ACO) from the ground up? And what kinds of calculations go into the decision to participate in any of the available federal ACOs sponsored by the Centers for Medicare and Medicaid Services (CMS). Leaders from the Henry Ford Health System in Detroit shared a wealth of insights on Monday morning, during the Population Health Symposium, one of numerous pre-conference symposia held at the Sands Convention Center in Las Vegas, as part of the annual HIMSS Conference.
Bruce K. Muma, M.D., CEO of the Henry Ford Physician Network (HFPN), and Matt Hussman, the HFPN’s director of analytics, shared insights with their audience around how they and their colleagues ultimately made the decision to participate in the Next Generation (Next Gen) ACO program under CMS.
The Henry Ford Physician Network is a 2,000-plus network of physicians, some of them employed by Henry Ford Health System, and others independent. It overlaps with the Henry Ford Medical group, a 1,300-employed physician and scientist group within HFHS.
Despite the large size of the Henry Ford Health System—the integrated system encompasses eight hospitals, 200 care sites, 30,000 employees, a durable medical equipment company, home healthcare, and pharmacy, and a service area with 1 million residents of southeast Michigan—“Despite the fact that we’re a really large health system, we really struggled over the decision to take on two-sided risk,” Dr. Muma told the audience on Monday.
“Henry Ford was an early adopter of value-based concepts,” Dr. Muma said. “We formed our first ACO in 2010, a year after the ACA [Affordable Care Act] was passed. The early intent was to create an ACO that would be attractive to commercial payers from a narrow-network standpoint. We had reached a size of 40,000 covered lives over a period of four years; but to be honest, in the overall scheme of things, considering the size of our service population, that didn’t end up being that huge a deal,” he said.
“Still,” Muma continued, “when the Pioneer ACO model came along, we looked at that. But we chose not to pursue it, because of the financial model. The formula wasn’t conducive to success; the shared savings threshold looked pretty steep to us.” In fact, he said, “Probably the biggest reason we didn’t do Pioneer was the adverse impact of the IME and DSH payment pass-through,” he said, referring to the federal Indirect Medical Education and Disproportionate-Share Hospital Adjustments to Medicare Inpatient Payment Rates under the Medicare program. “Those pass-through payments were not taken out of the formula, so we would automatically be designated as a ‘highly inefficient performer’” under that system. Further, he reported, “We opted not to pursue the MSSP [Medicare Shared Savings Program] opportunity in 2014 and 2015, because the same pass-through problem persisted.”
But, “Finally, when the Next Gen model came along and the pass-through problem was solved, we chose to go forward,” Muma explained, and on Dec. 21, 2016, HFPN signed a two-sided risk contract covering 21,000 attributed Medicare beneficiaries, in the Next Gen program.
“We really struggled with the decision” to go forward, Muma said. “The biggest risk was a $35 million downside. Even as a $6 billion [in annual health system revenues] system, sending CMS a check for $35 million would have taken a really big bite out of our contribution.” What’s more, Henry Ford Health System is a safety-net health system that serves an aging, underserved, inner-city population, he noted. But, he said, the opportunity to leverage strategic partnerships in the region, and to learn how to be successful with value-based care, was appealing enough to overcome those hesitations, he said.
Moving forward—analytically
In making the momentous decision to join the Next Generation ACO program, Muma, Hussman, and their colleagues knew from the start that they needed to engage in some scenario-based strategic thinking around the risk/reward calculations involved. So, Muma said, “We did an intensive sensitivity analysis, looking at how CMS might view our level of efficiency” as a participant in the Next Gen program. “We looked at best-case, moderate or average, and worst-case efficiency scenarios. In the best-case scenario, we found that we were looking at the possibility $11 million-shared savings check. On the other hand, if we let healthcare spending go up 3 percent—which is simply the national rate of annual medical inflation, we’d have to write a check to CMS for $4 million. The middle-of-the-road scenario gave us a $2.2 million gain. In other words, those calculations were based on a decreased total spend of 3 percent; an increased total spend of 2 percent; or staying at the same rate of spending.”
Meanwhile, Muma said, “We realized that this Next Gen ACO contract really would be a tipping point for Henry Ford Health System. As one of the speakers earlier this morning stated, if a health system has 30 percent or more of their volume based on risk, that’s a tipping point, and this took us from the mid-20s to about 33 percent. So that was a tipping point.”
Not only was it necessary to engage in a great deal of scenario-driven analysis to calculate the potential for success in the Next Gen ACO program, once the decision was made to go forward, it was vital to engage all the stakeholders in the organization, Muma said. As a result, he said, “We did more than 40 presentations in the first six months, talking to governing councils, boards, medical staff committees, corporate groups, business units—in the first six months—and that was important.”
Meanwhile, Muma said, “Another key challenge was engaging the beneficiaries. Unfortunately, despite, intensive effort in that area, we didn’t really make a difference” in terms of engaging beneficiaries to the point of broadly changing their behaviors. “We still struggle with beneficiary engagement,” he conceded, “though we agree it’s one of the most important things you can do” in achieving ACO program success.
Defining and eliminating waste—on a broad scale
Early on, Muma, Hussman, and their colleagues came to the conclusion that eliminating a variety of types of waste would be essential to success in the Next Gen program. They concluded that it would be important to identify and attack “hot spots of waste.”
In practice, Muma said, “It was pretty simple” to focus on the waste issue. “We looked for the places where we had the biggest piles of low-value care, or waste. Looking at our data, and benchmark data that we used, from a variety of sources, including from the Premier ACO Collaborative, which HFPN joined shortly after joining the Next Gen ACO program, “we identified three main hot spots,” he said.
The first was the high- and rising-risk populations”—typically, frail people with multiple chronic conditions, or cancer. The biggest opportunity with those populations, Muma said, “is changing the site of care from admissions, ED visits, or long stays in SNFs”—skilled nursing facilities. The second area of opportunity was around the high variation of medical decision-making within acute care. And the third was around post-acute care and transitional care. “Here,” he said, “it meant looking at variation in medical decision-making” in terms of managing post-acute care. “We didn’t own any SNFs, so we were spending money outside our system that wasn’t coming back in,” he said.
And the actions taken? Several. “First,” Muma said, “we built programs for the top 5 percent of patients in terms of prospectively determined risk. Then, we drilled down on acute episodes. We found that the decision to admit a patient from the ED to the hospital involved a three-to-four-fold variation in doctor decision-making. There were also huge variations in referrals to specialists.” In terms of transfers to SNFs, he said, “We found that our average length of stay in SNFs was 27 days. Looking at national benchmark data,” he said, they learned that other ACOs had reduced that average to 14 days; thus, there was great opportunity there.
As a result of all of those initiatives, Muma reported, the organization saved $5 million, in an overall spend that was $244 million in the first year, down from HFPN’s previous annual spending of $248.9 million; in other words, $3.9 million in net shared savings (about 2 percent), or $16.48 per member per month.
And what were the specific programs that came out of this? There were several, in fact:
- Post-Acute Care (PAC) Surveillance. “We stole the idea from other organizations that had done this,” Muma said. “We hired two case managers, had them call the SNFs, ask for a care plan, and assist patients in getting back to their primary care physicians. And that alone resulted in about a 10-percent reduction in SNF length of stay.
- Emergency Department Disposition Support (EDS): “This program was designed to help ED doctors make the right decisions—ED doctors are most fearful of sending a patient home with risks,” Muma explained. “Our ED chair [at the flagship Henry Ford Hospital] said, give us tools that will help reassure us that a particular patient will be safe if they’re sent home” rather than admitted. “So we developed a toolbox for them, and implemented that at our main hospital. That has resulted in a 7-percent reduction in admissions to the main hospital; and we’re hoping to replicate that at our other hospitals,” as that program is rolled out at the system’s other facilities.
- Comprehensive Care Clinics (ambulatory intensive care units. “We’ve opened two of these—they were functioning like ambulatory ICUs; they involve two physicians armed with a team of support, including nurses, pharmacists, etc., who run IV fluids, oxygen, etc., on patients. We enrolled about 500 patients, from that top 5-percent category [of high utilizers]—and in the first year, we saw about a 25-percent reduction in their PMPM [per member per month] costs.”
- Case Management Integration: equipping those case managers with the highest-risk patients in the ACO
- Clinical Decision Support: We adopted the Choosing Wisely/Referring Wisely program early on, and we essentially implemented a program purchased from a spinoff of Cedars-Sinai—they’ve spent a lot of time translating Choosing Wisely recommendations into clinical decision support alerts.” The result? $500,000 in reductions in claims expenses. In that, he said, “We worked with the chairman in their departments to identify the top four or five inappropriate decisions, and we asked them to develop recommendations for avoiding referrals; we want to develop e-consult capabilities to support this, so that that will create a kind of economy consult for PCPs; specialists are overburdened with volume, and this will help both types of physicians.” That program is set to roll out this year.
Data analytics—an essential component to ACO success
Data analytics has been essential to facilitating all of this work, Muma noted. “We realized early on that we needed to leverage our EMR/Data/analytics platform, HELIOS. We needed to understand clinical and claims data, and build risk prediction/benchmarking, and performance dashboards.”
One major operational challenge? “We also started to conduct ROI analysis. And that proved to be the most difficult part—we had to convince senior leaders to invest in us, hire staff, and develop models, and we found that we could reduce claims expense, but the finance people said, claims expense is just revenue reduction, and unless you can reduce variable or fixed costs, we can’t count this” as financial gain. “We still struggle with that. We’ve come to a common understanding, and our finance people understand that somebody else is going to have to do the ‘backfill.’”
With regard to some of the nitty-gritty of the analytics work, Hussman said, “You have to start with your eligibility or membership files first. A key enabler,” he noted, “is being able to look at external claims, and to see true performance of at-risk contracts. We started being able to take external claims and plug into analytics and see claims,” he said. And, he added, “We spend a lot of time bringing in that external claims data. We’re a bit different, where we’re looking to create the final claim.”
Importantly, Hussman noted, “CMS data does not look like commercial data; they look completely different. And data from two different commercial payers can look totally different, too. And you need to normalize all the different types of data,” in order to make the analyses work.
Key lessons learned
What have been the key lessons learned in all this work so far? There are several, Muma told the audience. First, he said, “Perfect data is not required; there’s lots of waste out there! You just need to start the discussion.” Meanwhile, “Internal EDW analytic capability is a required element in managing financial risk.” What’s more, “Building an executive dashboard is very important, to get executive buy-in.”
In addition, Muma said, “Aggregation and analysis of claims/EMR data is hard—building trust with physicians is even harder; they’re still struggling with the idea that variation in medical practice is form of waste.” And, with regard to waste, he urged his audience to “Focus on the large buckets of waste, and on supporting providers to do the right thing, as opposed to imposing controls.” Empowering clinicians, he said, “is the most effective strategy; an example of that was working with our ED director—he said, do something to help us send the patients home.” Meanwhile, in all this, he said, “The biggest challenge remains how you translate value-based initiatives into traditional ROI models” in calculating financial and operational success in the new healthcare.
Critical success factors? “It’s impossible to over-communicate the vision and expected challenges,” Muma said. In addition, “You need to engage your clinical leaders as care process owners, and inspire them to create better models of care.” What’s more, “Multidisciplinary teams are vital for creating value across the horizontal continuum of care.” And, when it comes to the data and analytics, “you need to build reliable metrics and dashboards to demonstrate value in population health programs in real time, as much as possible,” and importantly, “You need to engage your organization’s finance team to fully understand and be able to measure the impact of value-based care programs.”