Cover Close-Up: Medicaid ACOs Accounting for Challenges
Following is the full-length Cover Story from HDM's upcoming November print edition.
The much-ballyhooed model of care delivery known as the accountable care organization, or ACO, is moving front and center in the industry. In essence, an ACO attempts to supplant volume-based, fee-for-service care delivery with an outcomes-oriented approach that rewards prevention and care coordination—and in fact often penalizes those who fall short.
ACOs have emerged in many different flavors, in both the commercial and public sectors. In the latter, the Medicaid ACO is tackling one of the industry’s most difficult population groups—the poor and the underserved. “The Medicaid ACO is a great idea, but very hard to do,” says Lisa Suennen, co-founder of the Psilos Group, a venture capital firm which invests in health I.T. “It requires everyone to transform their way of doing business.”
And Medicaid recipients themselves—a diverse mix of young, old, disabled and low-income people who now number about 60 million nationwide—face challenges rarely imagined by the more affluent population, she adds. “If you are on Medicaid, you may not have easy access to providers in your community. If you’re living check-to-check, you are not thinking about jogging.”
Medicaid programs are administered state by state, with federal dollars in support. And faced with skyrocketing costs, several of the states are launching Medicaid ACO demonstration efforts to help not only lower outlays, but improve the health care received by some of their most vulnerable citizens. Following are profiles of a few of these efforts.
Camden Coalition: Group Visits Put EHR to Test
Formed in 2002, the Camden (N.J.) Coalition has a straightforward mission statement: “Improve the quality, capacity, and accessibility of the healthcare system for vulnerable populations in the City of Camden.” Its founder, Jeffrey Brenner, M.D., recently won a MacArthur “genius grant” of $625,000 in recognition of its efforts to revamp care delivery. The organization now spans some 50 staff members, three local hospitals, and 15 primary care practices, says Aaron Truchil, manager of research. “Our approach is to take small steps and then dig deep into practice redesign.”
The coalition launched a Medicaid ACO project with the state in 2011. While final regulatory details remain to be hammered out, the group will in essence take responsibility for the care delivery to about 30,000 local Medicaid recipients through the state’s fee-for-service benefit program. In addition, it is entering into an effort with a smaller group of Medicaid managed care patients with a commercial insurer, Truchil says. The ACO efforts will piggyback on I.T. infrastructure the coalition has been building for several years. “The challenge is getting hold of data,” Truchil says.
Toward that end, the coalition launched a local health information exchange in 2011. It is fed by the three local hospitals, receiving admission/discharge/transfer data, lab reports, radiology reports and discharge summaries from the hospitals. Primary care offices—which use a variety of EHR systems in addition to paper records—can access the database in read-only mode, Truchil says.
The main purpose of the HIE—which feeds a citywide, all-payer claims database also maintained by the coalition—was to establish the organization as an “honest broker” of health information. “It sets the precedent for us as a neutral party,” Truchil says, describing efforts to build a “platform of trust” as one of the most difficult components of building out the I.T. needed to support population health management. “It would be nice to have access to every data set out there,” he says.
But even a simple ADT feed can be a boost to local practices. The coalition’s daily reports going out to hospitals and physicians show which patients are hospitalized—a first step in planning post-discharge care coordination activities needed to avoid further episodes. Truchil says the next step will be to advise physicians electronically when their patients show up in the ED.
Area physicians involved in the coalition say the data exchange is valuable in the outpatient setting. Steve Kaufman, M.D., is one of them. Serving as deputy medical director at the Urban Health Institute (part of Cooper Health System, a coalition member), Kaufman is an endocrinologist who has carved out a niche serving diabetic patients. Once the Camden Coalition finalizes its reimbursement details with the state, Kaufman stands to be one of the beneficiaries of the shared savings model—in which the coalition will be rewarded for beating expected outlays among its Medicaid patients. But Kaufman—who also maintains a suburban, commercially funded practice—says he doesn’t discern among patients based on their insurance carrier.
The demographics of Camden make the city a health care challenge, he says. “This is a poor, minority population here,” he says. “People are unemployed or living close to the poverty line. And for any chronic illness, poverty makes it worse. Poverty is associated with worse outcomes and worse access.”
Kaufman’s professional life is devoted largely to finding solutions to that problem. At the Urban Health Institute, which targets predominately Medicaid patients, Kaufman has devised new treatment approaches for the underserved that stand conventional fee-for-service practice patterns on their head. The old model barely works in a fee-for-service practice with a commercially insured base, he contends. “The physicians move from room to room and at the end of the day have a pile of charts to complete,” he says.
In a notable about-face, the Institute has eschewed the one-on-one physician encounter. At the institute, Kaufman and a team of other caregivers, including medical assistants, nurses and social workers, see patients in a group setting with as many as 12 diabetic patients. Prior to the visit, the care team meets to discuss all the patients coming in that day, updating med lists if needed or pre-ordering tests in the EHR, from Epic. The total in-person visit is about 90 minutes, he says, with patients moving from station to station, first checking in, then delivering vital signs, discussing unmet care needs, and finishing up with Kaufman himself. As patients move through the chain, the various caregivers document their findings in Epic, which Cooper’s I.T. staff had to customize to adapt to the new model. Team members add their findings and suggestions in a special tab, which Kaufman reviews and signs off on at the end of the visit. “I am the only who can sign off,” he says. “If the team is in sync, and we have done good huddle work before the visit, I only have to ask a couple of questions and clarify any issues.”
Once patients have moved through the caregiver stations, they regroup and meet with the care team in a large conference room. “I take questions from the group,” Kaufman says. “People are learning from each other and there can be a real bonding among patients. It creates a community of caring.”
For the ACO, Kaufman helped determine many of the quality metrics that the state will use to assess progress. “I’m not worried about the ACO model,” he says. “When we get paid based on how many patients we see, it doesn’t make sense. We should be paid on how good a physician we are.”
The group visit also facilitates access for the poor people who come to the Urban Health Institute. “It is perfect for continued follow-up,” the physician says. “I can have someone come back in two weeks, whereas my individual patient calendar can be booked weeks in advance.”
The endocrinologist would like the see the HIE expand its reach. “We need tie-ins to more pharmacies and to adult medical day programs,” he says. “Patients come in and are not sure about all their meds. They say ‘My pharmacist knows.’ When you are dealing with an under-served population, having the ability to get information at your fingertips is vital. You don’t want to delay care and wait for a phone call.”
Other data challenges await as well. Truchil says the coalition is still negotiating with the state about whether it will turn over its claims files—which ostensibly would reveal a wealth of information outside the scope of the coalition’s current database fed by providers. “Without the entirety of the claims data, we will be operating blindly,” he says. “We are advocating for release of the data.”
Minnesota: Big Data
Minnesota’s Medicaid program is big—with nearly 800,000 enrollees in the state. And with a cohort of some 100,000 of them, the state is launching a three-year ACO demonstration project. According to Marie Zimmerman, health care policy director for the St. Paul-based Dept. of Human Services, the effort targets a focused group of long-term recipients and excludes dual-eligibles also in Medicare. The ACO is divided among six different participating provider organizations, each of which will have unique contracts. “There are many moving parts,” says Zimmerman.
Most of the organizations in the effort—which include Allina Health, a large IDN based in Minneapolis—will participate in a shared savings model. Providers who beat anticipated cost outlays for their attributed population—a number based on the previous year’s expenses paid by the state—will be eligible to get some of the money back. There’s a catch though: Providers must also meet certain quality metrics to keep their share of the savings, Zimmerman says. In addition, providers agree to take risk on their Medicaid patients, meaning if they exceed their budgets and fail to uphold quality metrics, they may have to return part of their Medicaid reimbursement. “The ACO will look at the total cost of care spanning inpatient, outpatient, physician, pharmacy and ED,” says Zimmerman. Other expenses pertaining to long-term care, dental and mental health services are optional components that none of the providers has chosen to include thus far.
The state does not have any firm expectations of how much of its $12 billion annual Medicaid outlay could be trimmed, Zimmerman say, but she says a 3 percent reduction is one projection. That was the amount the state forecast it might save in the effort when it applied for—and received—a $45 million innovation grant from the Centers for Medicare and Medicaid Innovation as part of the effort. The grant, says Zimmerman, will be used to hire a data analytics vendor to serve providers as they attempt to coordinate care and prevent unnecessary hospitalizations and ED visits.
There will be a lot of data to analyze. Drawing from its own data warehouse, the state is providing three kinds of monthly reports to providers in the ACO, Zimmerman says. First is a basic claims extract for the patients attributed to the given provider that shows utilization of inpatient, physician and pharmacy services. Second is a provider alert, which identifies enrollees who went to the ED or had an inpatient stay in the previous six months. The third report is a risk-adjusted, care management analysis that uses predictive analytics—from software developed originally by Johns Hopkins—to identify patients likely to enter the hospital and in need of increased care coordination.
As part of their agreement in entering the ACO, providers must attest that they can accept and use the data, but the requirement, Zimmerman says, “is pretty broad. Some providers will take administrative claims data and integrate it into their own data warehouse. Others are relying on the care management report more because they don’t have the analytics capacity to deal with individual claim files.”
The care management report, she acknowledges, can be unwieldy to work with. “It’s a large Excel file and is hard to sort and manage.” The state is hoping to use its grant money to figure out ways to streamline the analytics effort on behalf of the providers.
The state too faces its own challenges in parsing data to share among ACO providers. It must comply with federal laws governing the release of chemical dependency treatment data. “It is a manual process for us to sort it out,” Zimmerman says. “It gets very complicated. It is not just one set of providers who have legitimate access.”
The state is looking to expand the kinds of data it can share among its Medicaid providers. Right now, it is building out capacity in its benefits administration system to let providers know with which, if any, ACO a patient is involved. That information will be revealed when a patient shows up for services and the providers runs an eligibility check against the state’s enrollment database. “If a patient shows up in the ER, the hospital can use the system to connect the patient back with their primary care physician,” Zimmerman explains.
Memorial Healthcare: I.T. Integration Required
Although Memorial Healthcare’s shared savings arrangement with the state of Florida’s Medicaid program began some five years ago, upholding the ACO model of care delivery has not been a principal driver of its I.T. strategy. In 2009, the six-hospital, Hollywood-based public health system began standardizing on an integrated EHR platform, from Epic, which encompasses inpatient/outpatient and clinical/financial transactions. Replacing a homegrown mainframe system, the integrated system was driven by the then-looming ICD-10 code set (since postponed to launch in October 2014) more than by any accountable care arrangement, says Forest Blanton, senior vice president
But the enterprise Epic system—coupled with an outreach program to connect community physicians to the health system’s private HIE—has played right into the ACO effort, Blanton says. The health system serves as a medical home for children enrolled in the Medicaid program. As part of the deal, the state pays Memorial Healthcare a flat rate per child based on historical utilization patterns over the prior two years, and if the health system meets both financial and clinical metrics, it gets a portion of the savings. According to the state, the health system saved the Medicaid program $20 million for the care of 4,000 children over the last three years and improved quality scores as well.
Memorial incorporated multi-disciplinary team meetings including medical directors, nurses and social workers as part of the project. The team prioritizes what is needed to reduce utilization and improve care coordination.
Upholding the effort is a physician scorecard Memorial distributes among the medical staff, Blanton notes. The health system uses an analytics platform, from Verisk, as part of the care coordination program—which it undertook in conjunction with Premier, a hospital member organization and purchasing cooperative. Blanton says Memorial uses the analytics tool to analyze both Medicaid claims and its own self-insurance claims paid for some 23,000 employees. The data warehouse receives a database of medical and pharmacy claims from the state for its Medicaid population. “We can see the timing of diagnoses and tests,” the CIO says. “We can stratify the patients into those predicted to be the most expensive.”
For example, diabetic patients are monitored by the presence of labs, foot exams and other tests. Memorial circulates the care gap reports among primary care and other physicians, showing the per-member per-month costs for both individual physicians and their departments. The report goes to some 1,000 physicians—a mix of employed and community providers—who have joined the health system’s “clinically integrated physician-hospital organization,” says Blanton. The arrangement was launched in January of this year and participating physicians stand to benefit from shared savings and quality bonuses for upholding standard metrics.
Blanton hopes to build out the HIE to offer analytics reports through it to community physicians. Due to its outreach effort, about 35 group practices have adopted EHR technology with another 80 in the works this year. The health system subsidized the purchase of the technology via a Stark exemption, offering a handful of supported systems including eClinicalWorks and Greenway Medical Technologies. But there are limits to progress. “In our community, we have over 50 different EHR systems in place and on a practical basis, we can’t connect to all of them,” Blanton says.
For its 150 employed physicians, however, the health system is well-positioned to drive quality and uphold any future ACO contracts, the CIO says. The Epic system, for example, has a built-in alert mechanism with its order entry module. For certain tests, the system will look for identical tests performed within the previous 6-12 months, depending on clinical guidelines, and ask the physician if they need to be re-ordered. “That is the low-hanging fruit,” Blanton says.
University Hospitals: Extending Physicians
Based in Cleveland, University Hospitals’ pediatric facility is among some two dozen provider organizations splitting up $123 million in grant money from the Centers for Medicare and Medicaid Innovation—all aimed at reforming care delivery among seniors, low-income patients and their children.
University Hospitals Rainbow Babies and Children’s Hospital is using its $12.7 million grant to launch a pediatric ACO, says Andrew Hertz, M.D., medical director. The program will attempt to improve care among some 70,000 Medicaid patients 18 years and younger. It’s a shared savings arrangement, with the hospital standing to gain if it can hold down expected costs. “As we increase quality and decrease utilization, we will share in the savings,”
The effort is complicated by the fact that five Medicaid payers will be in play at once—various plans with which the state has contracted to administer its Medicaid outlay—which hit nearly $18 billion in 2011, according to the state. “Many providers contract with all five Medicaid payers, so we have to contract with all of them unless we want to divide the way they give care,” Hertz adds. In the initial effort, 150 pediatric physicians and advanced nurse practitioners will be involved—a mix of about 60 percent employed and 40 percent community-based, he says.
The sheer size of the effort qualifies it as a “big data” effort, says Ethan Cernin, director of the UH Rainbow Care Connection, the official name of the ACO. “We want to evaluate the total cost of care,” he says. “We want to decrease avoidable ED visits, facilitate more care coordination among children with complex conditions and increase access to behavioral health and other services.” Toward that end, the hospital is deploying a data analytics package, from the Advisory Board Company, that can parse out claims data the facility receives from the state—on behalf of the payers.
The database will offer insight into practice patterns and help identify care gaps, Cernin says. “We will identify high utilizers and also people not getting preventive care,” he explains. Eventually, access to the database will be granted to physicians in the program and other care managers. Cernin is banking on good old-fashioned competition to help drive performance improvement. “Enabling doctors to see how they stack up against their peers is a big behavior change,” he says.
There’s more to the effort, however, than just quality reporting. To help boost access to care, the hospital has installed telemedicine hubs in two inner-city community centers with a high proportion of ED users and low per-capita income, Cernin says. “We did a zip code analysis of who frequented the ED,” he says. Now, without an appointment, kids ages 3-18 years can meet virtually with one of the hospital’s on-call physicians through the telemedicine feed. The local community center is staffed by a medical assistant who takes vital signs and then connects an audio-video feed to the physician. Various medical devices, including a stethoscope and otoscope, are also connected, so the doctor can assess the patient from afar. “The physician can do a full physical,” Cernin says, noting the goal is to steer patients away from the ED.
In another effort, the hospital has identified a small group of patients who represent about half of the total Medicaid spend. “They are kids with cerebral palsy, seizure disorders and genetic syndromes,” explains Hertz, the project’s medical director. “They have multi-organ involvement, are wheelchair bound and non-verbal. Their use of home nursing services and hospitalization rate are high.”
The hospital is distributing iPads to these patients and their families. The devices will be equipped with a Web cam and be configured to initiate a spontaneous dialogue with caregivers on-call at the hospital. Cernin says upwards to 500 patients will receive the devices. “If they have questions about a rash, or their gastrostomy tube looks funny, they can send over a picture,” he says.
Health Share of Oregon: Driving P4P
Portland-based Health Share is among 15 “coordinated care organizations” the state’s Medicaid program is launching in hopes of creating a national role model for better care delivery and reduce costs. The stakes are high: CMS is investing a whopping $1.9 billion over the next five years in the state to help implement a pay-for-performance initiative with local community control over how the dollars are best spent. Health Share will administer the program on behalf of 165,000 area Medicaid recipients, says David Labby, M.D., chief medical officer. A number of health systems are participating, including Kaiser Permanente, Legacy Health and Oregon Health and Science University. These care delivery systems work with the seven different health plans that pay claims to them directly on behalf of the state. Although it doesn’t pay claims directly, Health Share will monitor the overall budget and help drive quality improvement efforts via analytics, says Daniel Dean, CIO.
This year, approximately 2 percent of the Medicaid outlay will be linked to various pay-for-performance measures, says Labby, adding that the proportion will grow as the program unfolds. The metrics will include chronic disease measurement as well as preventive screenings. “Medicaid delivers 53 percent of the babies born in the state,” he notes. “It is a large population which lacks resources and who often have multiple chronic conditions.”
Participating providers and health plans have a variety of population health management and risk management tools already, Dean says. For its part, Health Share has a separate data warehouse, which receives claims data from its seven health plan members in addition to enrollment data from the state. Initially, Health Share is generating reports on high utilizers, which it shares with its health plan partners. The self-developed dashboard offers a glimpse into patient utilization during the preceding year, he says, summarizing inpatient stays, ED visits, primary care physician engagement and mental health service visits. Dean says Health Share is taking a prudent approach when it comes to compiling population health data. “The health plans each have their own data warehouse and related analytics capacity. We have tried to enhance what they have. They lack a holistic view.”
For example, patients who wander away from their assigned provider, or who use dental and mental health services, can easily be missed. “Historically, physical health organizations have lacked insight into mental health organizations, and vice versa,” Dean says. “High utilizers require both kinds of services.”
Labby says the next big discussion will be around the creation of a community care management tool to help coordinate care among all the players. “If you talk to providers, they don’t want 10 dashboards, one for each payer in their commercial, Medicare and Medicaid mix. We have to figure out how to create a utility that works as broadly as possible.”