ACOs using analytics, IT to manage out-of-network care
Accountable care organizations face a persistent conundrum. They're expected to reduce overall healthcare costs by coordinating care: minimizing ineffective or duplicative services and maximizing communication among the various members of each patient's care team. They may not make money unless they can coordinate effectively.
At the same time, under current federal ACO rules, providers have no leverage to keep their attributed ACO patients from seeking care outside their network, and they may not know until well after the fact. A June 2014 study of care patterns in 145 ACOs, published in the Journal of the American Medical Association, found that 66 percent of specialist visits took place outside the ACO's network.
The commonly used term is "patient leakage," as if a patch here and there would fix it. But it's a complicated issue, encompassing patient choice, physician referral patterns, and whether the ACO should offer certain services or refer them to out of network providers. And one midnight ride to the ER can thwart the best-laid plans. Here are some ways that ACOs are trying to understand and stem leakage.
Atrius Health, Newton, Mass., has learned how to make accountable care pay: while about half of its patients are covered under some form of risk contract, they account for 75 percent of its revenue, says deputy chief medical officer Joe Kimura, M.D. Atrius is a Medicare Pioneer ACO and also holds risk contracts with Blue Cross and other commercial insurers.
Atrius has 750 physicians in 35 specialties, but doesn't own hospitals or skilled nursing facilities. Any care delivered in those settings is by definition "out of network." Kimura prefers not to call it "leakage," because patients need those types of care sometimes, and going outside the network is appropriate and beneficial. He thinks in terms of "outside utilization," and classifies it into "managed" and "unmanaged." "Managed" is better.
"It's about trying to optimize value rather than driving the patients to particular providers," he says. "The value is not just the dollars associated with a particular event. There's a broader picture around potentially expensive procedures that could save you money downstream because you've treated something early."
Atrius uses homegrown analytic tools to translate claims data into information that clinicians can use to understand the impact of their referrals on their patients and the organization. Kimura says massaging the data is challenging, when a single clinical event can take up 100 line items on a claim.
Once his analytic tools have translated claims data into individual episodes of care and other measures that are meaningful to clinicians, Kimura can also use it to determine where Atrius should focus expansion efforts. "If we have enough volume with a particular provider, we should have that provider inside our network," Kimura says.
ACO management company Aledade, Bethesda, Md., manages patient leakage by creating "preferred provider partnerships" and encouraging its primary care physicians to refer accordingly, says ACO Initiatives Lead Laura Chmar.
Aledade works with 110 physician practices across 11 states, and its Medicare ACOs cover about 100,000 lives. Aledade's eight ACOs do not include hospitals or specialists in their networks. Its analytics platform, developed in house, enables Aledade to identify which specialists and hospitals its patients use, and the referral patterns of its physicians.
Using a combination of data from Medicare claims, health information exchanges, and electronic health record systems, the platform enables Aledade to identify the best specialists or the best points of care for certain procedures, and encourage its physicians to refer accordingly.
"When you show physicians the data on where their patients are going, they're shocked by it and see it as a missed opportunity," Chmar says. For example, primary care physicians in one ACO might refer their patients to 20 different cardiologists. Aledade may trim that list to four, taking into account existing referral patterns, which specialists offer the best value on routine services, whether an office can handle urgent care, and other differentiators. Aledade then works with the preferred specialists to make sure they can cover all patients' needs. One cardiology practice didn't offer urgent appointments, sending patients to the hospital instead. Aledade negotiated with the practice to expand same-day access.
The model helps cement relationships between primary care and specialty providers, and encourages specialists to make sure their patients are seeing a primary care physician regularly, Chmar says.
Aledade's analytics platform can flag danger signals, such as a dramatic uptick in specialist costs or a pattern of urgent care visits, to let primary care physicians know that a patient may be having new problems. "They can intervene and get the patient back on track," she says.
U.S. Medical Management, Troy, Mich., provides management services for the Visiting Physicians Association, which employs 200 primary care physicians and operates in 12 states on a 100 percent house call basis.
Through USMM, VPA has participated in several Medicare accountable care arrangements including Medicare Shared Savings and the Independence at Home demonstration. The VPA's home-bound patient population seems like it would defy leakage, but CIO David Vezina says patients most often slide out of their physician's oversight via ambulance.
"Our biggest leakage is when our patients go the emergency room and get admitted to the hospital," he says. "Because they have so many comorbidities, it's easy for a hospital to pick up a three-day DRG, and if we're not aware, the hospital will send them to rehab on discharge." VPA physicians may not find out about the stay until they visit a patient's house and find him or her not home.
Typically the patient won't be much help, says Joan Valentine, vice president of quality and clinical integration. "As far as they're concerned, their doctor is 'that nice man who comes to the house,' " she says. They also are challenged to remember the details of their often extremely complex conditions and medication regimens.
To make sure it knows when its patients are receiving hospital care, USMM is setting up admission-discharge-transfer (ADT) feeds from the health information exchanges in the areas where VPA has practices.
USMM provides monthly patient lists to the information exchanges, and receives real-time notifications of any admissions. A staff of quality managers uses that information to connect with hospital discharge planners. "We stick our noses into that discharge process to make sure the rounding physician connects back to the PCP and directs the patient back home," Valentine says.
The main issue with those ADT data feeds is lack of consistency, Vezina says. "Every time they send you a record, you get a tremendous amount of unusable data." He is currently in talks with a national claims clearinghouse that may be able to provide one real-time data feed for all the states where USMM is active, and also create a translation table to standardize the data coming from the hospitals and help filter out irrelevant information.
USMM also seeks to keep patients from entering the hospital in the first place, using a data warehouse and analytics from Health Catalyst that will be running by fall. Vezina plans to improve USMM's ability to identify which patients may need extra attention.
Currently, patients who have had two or more ER visits or hospital stays within the previous 60 days are assigned to a registered nurse who assesses them regularly and when necessary connects them with community support resources. Vezina expects to be able to target high-risk patients more accurately and also identify the most cost-effective ways to help them. "We think this will increase our shared savings by 50 percent, at least," he says.
Sharp Healthcare, San Diego, dropped out of the CMS Pioneer ACO program in 2014, citing an unworkable financial model. "We did fabulously from a utilization standpoint, but we barely broke even" because the program used national rather than regional benchmarks to measure performance, says Alison J. Fleury, senior vice president of business development and former CEO of the Pioneer ACO.
But the organization is still deeply involved in risk-based contracts, including Medicare Advantage and commercial ACO plans. About 60 percent of Sharp's patients are covered under some type of risk or shared savings arrangement, and 30 percent are under full risk contracts.
Sharp had all the IT systems and structures in place for accountable care before entering the Pioneer program, Fleury says, including a data warehouse that could track utilization throughout the San Diego area. The only incremental cost was salaries for case managers and nurses to coordinate care.”
Sharp, like other San Diego area providers, uses Experian's Passport Eligibility to verify patients' insurance eligibility, and the organization also receives notification when another provider is checking eligibility for a Sharp patient. Sharp has an department than manages all out-of-network activity, sending a nurse to the ER to help manage the care and get the patient transferred to an appropriate facility after being stabilized.
Fleury says part of the problem with the Pioneer program was how patients are allocated to ACOs: via claims data rather than patient choice. "They're not selecting to be with you,” she says. "They might already get 49 percent of their care somewhere else."
Even though Sharp did its best to provide exceptional care, patients were apt to keep seeing the out-of-network specialists they were used to. In contrast, Medicare Advantage and commercial ACO patients choose the plan and generally are required to stay in-network. With those populations, Fleury says, leakage is minimal and mostly associated with ER visits.
Duncan Gallagher, CFO of Allina Healthcare in Minneapolis, believes that the health system has to make the case to patients directly about the advantages of staying in network. And by extension, it also has to make the case to the clinicians who work with those patients.
Value-based care isn't growing as quickly as Gallagher expected. Total-cost-of-care contracts cover about 100,000 patients, but currently represent only about three percent of Allina's revenues. The health system, which includes 13 hospitals and 90 clinics and participates in the Pioneer ACO program, is actively seeking to grow its percentage of value-based care, and is adopting an analytics platform from HealthCatalyst to track utilization and referral activity. Previously, Allina has made do with a proxy for outside referrals, tracking activity and utilization within the system and comparing it with expected levels.
Once Allina can identify more precisely where patients are going, Gallagher says, it will be easier to figure out why they seek care elsewhere and persuade them to stay in-network. "We have an opportunity at the point of service to provide certainty about the next steps in their care, rather than just handing them a list of options and saying, 'Good luck!' " Even if patients assume all providers are of equal quality, they may be sold on factors like care coordination and having all their records in the same place.
Clinicians will need comprehensive provider directories and ways to close the feedback loop with specialists, as well as training to help them understand how to make the best referrals for both the patient and the organization.
Gallagher says a relatively simple pie chart can deliver a dramatic message about how many patients stay in network, how many go to Allina's partners, and how many end up with competitors. "It's not so much about the money that's leaking out of the system, but about patients who are getting sub-optimal care because we're not conveying the value proposition for staying in the system."