9

ProvenHealth Navigator: Geisinger’s Advanced Medical Home

Like many elderly patients, Robert had multiple chronic conditions including diabetes, lung disease, and heart failure. His heart, functioning only between 12 and 14 percent, was his main problem, but he was a Geisinger Gold Medicare HMO member and participant in our ProvenHealth Navigator® (PHN) advanced medical home program, which allowed him to stay as healthy as he could and out of the hospital for as long as possible. And for that, he was grateful.¹

To help patients like Robert, we embed nurse care managers, who are employed by our insurance company, in the primary care office, where they become part of the patient-care team. The care manager’s job is to focus on the sickest patients in the practice, such as those with congestive heart failure or diabetes, and ensure they are taking prescribed medications appropriately, keeping appointments, and following up with preventive measures. The goal is to help these patients maintain health and avoid repeated hospitalizations.

Our pioneering concept of the embedded care manager as concierge caregiver for the sickest patients is the foundation of Geisinger’s version of advanced medical home. We designed this concierge care based on payer data that showed us which patients needed the most hands-on care. The embedded care manager concept came from our conviction about providing such care physically, as opposed to what has been shown in many studies, both anecdotal and formal clinical trials, that telephonic or distant care management does not work. We decided that this physical interaction with our embedded nurses as care managers was an absolute necessity for our sickest 150 or so patients per community practice, and it enables our physicians to do a different task than they were doing prior to stratification and segmentation of care. They are freed from a focus on increasing patient volume across all severity stratifications into something much more manageable.

In our version of advanced medical home, we’ve developed a sophisticated combination of technology and people. While our redesign of care may be technology-enabled, it is based on our view that a long-term human relationship between the patient, the patient’s family, and a care manager (the healthcare quarterback paying attention to all the details) is imperative for success.

For example, when Robert stepped on a Bluetooth-enabled scale at home, his weight was transmitted to his doctor’s office, where care manager Anita McCole noticed the slightest increase. Well aware of Robert’s medical condition, she called to ensure he was OK. When he mentioned weakness in his legs, Anita was able to facilitate physical therapy to build strength. From their multiple conversations over time, Anita and Robert developed a rapport, and he was comfortable talking with her about his health issues.

In addition to managing chronic illness, our nurse care managers ensure that patients are safe in their homes, have the necessary transportation to get to their appointments, are eating well and taking medications as prescribed, and are complying overall with their care plans. The care managers confirm that their patients schedule tests and procedures and receive their flu, pneumonia, and shingles vaccines. In essence, the care manager becomes the patient’s partner and ally to connect the patient with the healthcare team. Keeping patients as healthy as possible saves money by decreasing the need for expensive hospitalizations, but most importantly it is beneficial to our patients and their families.

Results are what matter, and we are pleased with PHN’s ability to improve care while reducing hospital admissions. (See Figure 9.1.)

FIGURE 9.1   Admission and Readmission Metrics

In addition, our advanced medical home program has demonstrated improvement in the risk of heart attack, stroke, and retinopathy in individuals with diabetes. Our three-year results for 25,000 patients found that PHN prevented 305 myocardial infarctions, 140 strokes, and 166 cases of retinopathy. While emergency department visits remained flat, acute care admissions decreased 27.5 percent and all cause 30-day readmissions decreased 34 percent. Further, 72 percent of patients say quality of care improved when they worked with a care manager.

PHN was developed as part of Geisinger’s response to the national problem of not having enough primary care physicians (PCPs) available to meet patient demand, especially with the aging of the baby boom generation and the increase in patient volume associated with the 2010 Patient Protection and Affordable Care Act. PHN has three major components: primary care redesign, population health management, and the medical neighborhood.

PRIMARY CARE REDESIGN

The typical response to the physician shortage, attempting to train and hire more PCPs and to pay them more, is totally inadequate. Simply hiring more is impossible, because there aren’t enough at the present time, and it will take 10 to 15 years for any significant increase because of training latency. Higher pay redistributes rather than solves the problem, creating market imbalance somewhere else.

Redefining the role of the PCP is another inadequate response. In various parts of the United States, we’ve seen the specialty-based disciplines of obstetrics/gynecology, cardiology, medical oncology, and even general surgery claim that chronic disease management for a certain component of patients is best done by the specialist. Such claims may have credibility, but this doesn’t solve the problem for most patients, who need a team captain for the multitude of specialty- and subspecialty-linked medical problems generally associated with the increasing aggregation of chronic diseases of aging, such as the fragile diabetes patient or the extreme congestive heart failure patient with hypertension and reactive depression.

Geisinger fundamentally reengineered the primary care process, relocating as much of the patient’s care as possible into our community-based practices and changing the relationship between our community-based primary physician team and our specialists, who more often than not are located in hospital-centric clinics. We did this by utilizing payer side data based on previous claims to identify patients who needed more hands-on care, stratifying the tasks of the community practices, adding the embedded nurse manager, and enabling that nurse to be the concierge care and triage caregiver for a group of the sickest patients in each practice. The other patients, based on decreased past utilization and expected decreased future utilization, were assigned to other members of the team for specific care. A good example is our algorithm-driven approach to managing hypertension run by a pharmacy tech, as opposed to a nonspecific accountability for helping to optimize blood pressure control that typically resided with the PCP. In the latter scenario, nothing usually was accomplished in between the patient’s yearly doctor visits, and the hypertension remained a continuing problem. We also encourage and expect our nurse managers to go outside the doctors’ offices as necessary, into patient homes or skilled nursing facilities where patients with the highest utilization often reside.

Four main components undergirded our primary care redesign: a PCP-led team delivering care, with all members of the team functioning at the top of their licenses; enhanced access for patients and their families; services guided by patient needs and preferences; and significantly enhanced patient and family involvement in caregiving outside of doctors’ offices. Again, the payer data and stratification of patient needs is essential to redesigning not only the care itself, but also who provides it. And the expectation that our concierge care commando nurses leave the offices and visit the highest-need patients in person helps the at-home caregivers become more involved, for instance, in the daily monitoring of weight and other appropriate tasks.

Our primary care redesign was matrixed with “all-or-none” bundling of care measures for patients with prevalent chronic diseases, similar to what we used to reengineer care of type 2 diabetes patients in creating ProvenCare Chronic. We agree with Donald M. Berwick, a leading advocate of high-quality healthcare and former administrator of the Centers for Medicare and Medicaid Services, who supported the all-or-none bundle commitment because it more closely reflects the interests and desires of patients, fosters a systems approach to achieving goals, and provides a more sensitive scale for assessing improvements.²

In addition to activating patients and their families to become partners, we also concentrated on eliminating, automating, delegating, and incorporating what was easiest into the normal patient flow and provider caregiving. In short, we combined the redesign of the primary care based on specific payer data with the stratification of assignments to the various team members. Plus, we committed to achieving all the known best practices, as socialized by our PCPs and hospital-based specialists and subspecialists, for the most optimal outcomes for patients with type 2 diabetes, coronary artery disease, congestive heart failure, and other chronic conditions.

It was this combination of our commitment to individual high-prevalence chronic disease optimization plus our primary-care reengineering that led to Geisinger’s overall decrease in hospitalization per thousand. The bundled best practice plus PHN as an integrated force changed the cost of care in two ways: getting better outcomes for chronic disease patients and reorganizing how care is provided to these patients so they are better cared for and able to avoid those all-too-frequent weekend visits to the emergency room. This ultimately decreases their need for hospitalization.

We charged everyone to go for out-of-the-box transformational change, rather than incremental redesign. To do so, we:

•   Asked outrageous questions

•   Made outlandish suggestions for consideration

•   Became comfortable taking risks

•   Anticipated, managed, and promoted emotional connections

•   Celebrated successes and learned from failures

At the same time, we watched carefully and didn’t readily accept the familiar reasons people use to resist change. This included making sure that routine needs were handled and communicated via the electronic health record (EHR) prior to the physician seeing the patient so the visit with the doctor could focus on solving problems as opposed to simply gathering data. It also included fundamental changes, such as the patients being brought into the examining rooms by team members not involved in actual caregiving. When these team members “room” the patients, doctors and nurses can spend their time and effort appropriately solving issues to benefit patients.

Our 20 percent of compensation based on achieving care transformation goals directly linked to top strategic innovation commitments also was doing something different than before. These goals were not related to relative value units, panel size, or the other usual fee-for-service volume-based productivity units. The providers’ performance in caring for the entire universe of patients in their given practices, as well as in the overall community practice service line, is fed back almost in real time to the payer. Analysis on the payer side produces a bell-shaped curve representing how individual providers vary in their use of resources and in patient outcomes, particularly related to hospital admissions and readmissions.

Use of this two-way data flow comes into play in determining best practice. We can see who is doing the best job with type 2 diabetes, coronary artery disease, or congestive heart failure patients, or with those patients who have multiple chronic diseases. We also can see where the best job is being done among our community practice sites. The obvious systemwide commitment, particularly among the leadership and community practice, is to scale and generalize from the individual physicians and the individual practices doing the best in terms of high quality and low cost.

The initial test for community practice reengineering at Geisinger involved our sweet spot: the overlap between members of our commercial, Medicare Advantage, and Medicaid managed care insurance plans and the patients cared for by Geisinger and nonemployed panel providers in Geisinger-owned hospitals. (See Figure 9.2.)

FIGURE 9.2   Sweet Spot for Partnership and Innovation

The structural and cultural aspects of the overlap between payer and provider were fundamental in enabling the significant behavior changes necessary for both providers and patients. The two-way change in data flow and in the way that providers and payers worked together to modify the processes of care went significantly beyond simply altering how the insurance company paid the providers. Individual tasks for general internists, PCPs, nurses, nurse practitioners, physician assistants, and pharmacists all changed. Interactions also changed between those who were providing care in community settings and our specialists, who were for the most part located in the hospital-centric clinical specialty locations. An explicitly different interface between the specialists, who are most often hospital-based clinicians, and the community practitioners was required to achieve bundled best practice for patients. This was linked to the Geisinger commitment to provide care as much as possible close to where the patients live. This is in distinct contrast to simply asking the specialists to open up their hospital-based clinic scheduling so patients and their families could travel to the hub. About 75 to 80 percent of patient issues can be handled efficiently simply by having immediate open access to specialists to answer questions either electronically or by phone, obviating the physical interaction that normally means patients going to see the specialists.

In addition, all of the best practice algorithms, particularly for the bundled best practice, were socialized by having specialists work with the PCPs. For example, PCPs and endocrinologists collaborated to determine the metrics to be achieved for all type 2 diabetes patients; similarly, cardiologists worked with PCPs in determining the bundled best practice for coronary artery disease and congestive heart failure. This fundamental interaction was socialized to obtain the bundled best practice algorithms for each prevalent chronic disease and to attain buy-in from both primary care and specialist physicians.

Having a health plan employee interacting effectively as part of the care team seemed revolutionary at the start of PHN. Our experience has shown that employees paid by the insurance company function well as part of the team in the doctors’ offices. The health plan representatives not only are directly responsible for managing high-utilizing patients, they are key enablers of smooth data flow between payer and provider.

We started with two beta sites, one in a Geisinger community practice in Lewistown and a second at our community practice office in Lewisburg. While the names of these two towns are similar, demographically they represent opposite ends of the socioeconomic strata. Only after we showed that we could obtain good results at both ends of the spectrum did we scale PHN throughout our entire system. In addition to learning how we could provide care and achieve results given such varying demographics, we also found that Lewisburg was close enough to the main Geisinger Medical Center hub that there was an interesting tension in determining which patients should be referred to this major hub and which should go to a non-Geisinger, but very good, community hospital within a mile of our Lewisburg practice. Lewistown, on the other hand, was more than 45 miles away from any Geisinger hub, and the only nearby hospital was a non-Geisinger facility that was, for most of the time until it joined the Geisinger family, relatively restricted in terms of resources, both human and capital. It was an interesting set of experiments with an overall commitment to keep patients as close to home as possible, even if they had significant health issues. What we learned, in essence, was that it could be done at both of these places with an early result of significant decreases in hospitalization needs and excellent patient and doctor satisfaction.

Internal scaling consisted of 42 Geisinger-owned primary care practices, 40 non-Geisinger-owned practices that were heavily reimbursed through Geisinger insurance products, and private practices that used Geisinger-owned hospitals when acute care was needed. Non-Geisinger primary care practices in California, Illinois, Maine, New York, West Virginia, Virginia, and Wisconsin have undergone similar successful PHN reengineering efforts.³

We learned as we scaled that unless we kept attentive to the data flow, both from payer to provider and provider to payer, and looked at variations in care almost on a real-time basis, there could be recidivism in either hitting the optimal metrics in the bundled best practice commitment for high-prevalence chronic disease or in the metrics of hospitalization per thousand. Recidivism was likely the default, which required our active participation to avoid. Another important lesson was that we could scale out to nonemployed, non-Geisinger community practices as long as those practices had an adequate volume of Geisinger insurance patients to justify getting the data into those practices and capturing the attention of the non-Geisinger practitioners. Although we had no direct leverage over their total compensation, we could add quality bonuses based on getting the same kind of population health benefit, which amounted to an increase of 15 percent or greater to their total compensation. This was certainly sufficient for them to do the same kind of PHN redesign and make the same kind of commitment to bundled best practice that we were able to achieve with a much greater and more direct leverage among employed Geisinger caregivers in our own community practices.

POPULATION HEALTH MANAGEMENT

Population health management, the second major component of PHN, involves identifying, segmenting, and risk-stratifying populations of our patients and insurance plan members by analyzing data provided by our insurance operations as close to real time as possible. Chronic disease and both primary and secondary preventive care are enhanced by clinical decision support communicated through the EHR. Gaps in care and the appropriate interventions are discovered and transmitted in real time to the provider team and also to patients and their families. We consider the EHR to be an important member of the team, but only as an enabler, not as the primary solution.

Effective population health management is founded on the ability to stratify patients with different risks based on past utilization: patients considered basically to be well; those considered at risk, with one or two chronic diseases; and chronic and complex patients with a multitude of chronic diseases and significant history of multiple acute care admissions. (See Figure 9.3.) The latter group is the chief focus of the PHN care managers.

FIGURE 9.3   Care Approach by Patient Risk Status

Our embedded care managers are tasked with understanding past and managing concurrent utilization. Most of the time, the care managers are registered nurses. We specify the individual chiefly responsible for frequent follow-up with each patient and his or her family. Care managers are given variable caseloads, with approximately 300 at-risk patients and 125 to 150 complex chronic disease aggregate patients assigned to each care manager.

Although employed by our insurance company, the care managers work as members of the community practice team, providing information from the insurance company that is modified for immediate use by the entire team. The care managers are chiefly responsible for everything that happens to their caseload of patients, and everything is triaged through these managers.

Most of the time, this embedded care management entails daily interaction between some member of the provider team and the patient and family. It often means linking the primary care manager with the appropriate specialists, either physically or by phone, to address acute access issues, always to be coordinated by the care managers. They often are out of the office, directly interacting with patients and their families in their homes, skilled nursing facilities, or wherever their patients are receiving care.

Care managers do condition screenings, monitor symptoms, assess the patients’ or their families’ ability to manage the medication regimen, and catalyze as much patient activation and engagement as possible. In addition, the care managers are tasked with closing all gaps in care, particularly with high-prevalence chronic diseases such as diabetes, asthma, hypertension, osteoporosis, coronary artery disease, chronic obstructive pulmonary disease, congestive heart failure, and reactive depression.

We also employ advanced care management, reserved for patients identified through predictive modeling (done on the payer side) to be at highest risk for acute care utilization. Most of the data is from medical claims and pharmacy; however, a significant amount of concurrent data also comes from the EHR and through our data warehousing and provider-side analytics. Targeted populations most often include the prevalent chronic disease aggregates: cancer, end-stage renal disease, high-risk pregnancy, special populations such as those with multiple sclerosis, cerebral palsy, and cystic fibrosis, and in general the frail elderly. The embedded advanced care managers also are asked to assess the social and behavioral issues associated with the medical diagnoses aggregations. We do this to better understand the link between physical and psychological gaps in care and to more effectively work with patients, their families, and social supports in determining how to create real behavior change in the caregiver/patient partnership.

Patients being discharged or transferred from acute care facilities are an additional target population across all of our risk stratifications and represent a particular challenge if they do not already have a PCP in our system.

The care management solution for non-Geisinger providers entails either training and onboarding care managers for the provider’s system or creating care management outsourcing solutions embedded as a turnkey operation. (See Figure 9.4.)

FIGURE 9.4   Our Approach to Advanced Care Management

THE MEDICAL NEIGHBORHOOD

The medical neighborhood, the final core component in PHN, is an attempt to create a 360-degree care system including skilled nursing facilities, the acute care hospital before transition into the ambulatory setting, home health, and pharmacy. It involves defining resource utilization differences between employed and nonemployed physicians, selective specialty referrals, a systematic process attempting to create efficient transitions of care processes, and integration with community services.

The key is to create a fundamentally different relationship between the hospital-based specialists and the community practitioner team located near where the patients and their families live. The other equally important change is to create a relationship between the community practitioners and an enhanced care model that includes skilled nursing facilities and non-doctor’s-office social resources. (See Figure 9.5.)

FIGURE 9.5   Optimizing the Primary Care Physician and Specialist Connection

Two basic concepts are important. The first is that specialists and PCPs work together to determine the algorithms and the commitments to bundle best practice for patients with prevalent chronic diseases. Second, in the event of a diabetes or a congestive heart failure patient in crisis, most of the time health systems considering themselves responsive would simply figure out how to open up the daily schedules of the hospital-based specialists and subspecialists so patients could come to them and be seen the same day. Geisinger didn’t think that aspiration was good enough, and what we did was have the specialists and subspecialists available 24/7 to take either phone calls or electronic communication from the PCPs. Most often this solved the issue. Only about 15 to 25 percent of the time was there still a residual need for patients to come in to be seen by the specialist or subspecialist.

In addition to keeping patients out of the hospital emergency department unless there is a true emergency, preempting chronic disease management issues that lead to emergency situations, and providing care in our doctors’ offices, we also include the patients’ home settings in their overall care. We create an effective medical neighborhood to further develop the continuum of care by getting to the kitchen tables of patients who have four or five chronic diseases and take 15 to 20 medications daily. We visit patients in skilled nursing facilities and intervene before they experience a 5- to 10-pound weight gain and are transferred to the local hospital emergency department to handle their fluid retention.

DRIVING SUSTAINABLE OUTCOMES

Performance metrics are straightforward, with admissions per 1,000 and reduced readmission rates our primary endpoints. Metrics focused first on patient and clinician satisfaction, then on the cost of care before and after reengineering. Decreased acute hospital utilization was the first sign of success. Specific quality metrics addressing particular high-prevalence chronic disease outcomes improved. And we looked closely at how the reengineering could help bridge the movement from fee-for-service to pay-for-value as the dominant form of reimbursement transformation.

In our experience, success in scaling for both the Geisinger and non-Geisinger nonemployed physicians almost always has been obtained within a year, with a significant decrease in total cost of care based chiefly on decreased acute care days per 1,000 patients. Additional extraordinarily important outcome metrics include patient and physician satisfaction and improvement in chronic disease-specific process and outcome metrics. (See Figure 9.6.)

FIGURE 9.6   Effective Redesign and Care Coordination Delivers Rapid Impact

The most important benefit from the patient standpoint, in addition to a satisfying and effective relationship with the physicians in the newly reengineered community practices, is the effect on disease outcome. Some 99 percent of our patients believe working with a care manager is good, and 79 percent think the care they receive is better. For the type 2 diabetes patients who were involved in the reengineering, in fewer than three years, significant numbers of heart attacks, strokes, and retinopathy cases were prevented when compared to the practices before reengineering or to practices that had not been reengineered. (See Figure 9.7.)

FIGURE 9.7   ProvenCare Chronic Disease Value-Driven Care Outcome Improvements

In scaling to non-Geisinger practices in non-Geisinger markets, admissions, readmissions, and emergency department visits all were decreased significantly and sustainably within a year. Cost-of-care reductions obviously were affected by the hospital-centric finance officers’ tendency to increase price per unit as volume decreased. Nevertheless, there was total reduction in cost of care per patient in a number of these scaling exercises. Finally, extension into the pioneer accountable care organizations was extraordinarily gratifying and affirmed. The value in decreasing hospital admissions was the primary benefit to both patients and their families, as well as to the financial total cost of care endpoint. We believe that a huge amount of our PHN redesign and our bundled best practice beneficial effect on chronic disease patients was, in fact, the seed for attempting to recapitulate a payer-provider interaction like the Geisinger fiduciary structure in many other types of payer-provider relationships throughout the country, most predominantly the Centers for Medicare and Medicaid Services Pioneer Accountable Care Organization (ACO) model and other ACOs.

Our most obvious gratification was not just that we were a model for this redesign, but also the fact that we had shown that we could get significantly better outcomes with a population of patients. The huge decreases in the need for hospitalization and rehospitalization were proof of the fact that quality and cost do relate (and that usually they are inversely related) so higher quality results in lower cost. Incidentally, our PHN redesign is also a win for doctors. Some 86 percent of our physicians believe they provide more comprehensive care with our advanced medical home, 82 percent believe timelier information is available regarding patients’ transitions of care, and 93 percent would recommend advanced medical home to other PCPs.

LEADERSHIP ISSUES

Leadership teamwork between the payer and provider sides of Geisinger was key to PHN success. This was a transformational relationship in which both the payer and provider asked how quality and value could be improved for their mutual constituency and was significantly more than simply changing the payment incentives from insurer to provider. It started with a strategic discussion involving clinical and payer leaders defining the single highest cost group of patients in the ambulatory setting. The assumption was that these almost always were those patients with the least successful outcomes. Once the high cost/poor outcome cohort was defined, leadership on both sides of the organization came to consensus on what would be considered an optimal outcome. Payer side analytics as well as the clinical enterprise healthcare data warehousing and analytic capabilities were employed in this exercise. How the caregiving could be redesigned for different patient groups with different severities of disease, different disease and living needs, and different utilization patterns was a fundamental benefit of payer and provider leadership working together to the benefit of their mutual constituency. This fundamentally different relationship and working partnership was never generalized to any of the non-Geisinger payers within our market areas.

Sustainability of our payer/provider sweet spot may come under stress as overall Geisinger leadership throughout the organization evolves, leaders assume additional operational duties, and clinical and insurance markets become more stringent. From our scaling experiments outside the traditional Geisinger market into Delaware, Maine, and West Virginia, clinical enterprise commitment without sustaining commitment from a dominant payer in the volume-to-value reimbursement transition has demonstrated that most positive outcomes are not sustainable long-term, despite early success.

LESSONS LEARNED

•   Provide dedicated care managers enabled by both claims and clinical data.

•   Implement a best practice team with proper staff allocation and be willing to change what people do.

•   Be sure to have data up front.

•   Provide training for all involved in advanced medical home.

•   Accept that it’s not just reengineering; be in it for the long haul.

•   Build a strong infrastructure with guidelines for accountability.

•   Pay for better patient outcomes, not filling hospital beds.

•   Define outcome by individual provider and by each community practice group.

•   Spread what you learn from the most successful to the least successful.

•   Enable continuous innovation with some room for failure.

•   ProvenHealth Navigator’s success simultaneously means better health outcomes and lower total costs of care.