8

ProvenCare Chronic

By the time 74-year-old Arthur went to his first appointment with a Geisinger endocrinologist in 2009, he had had angioplasty twice to open blocked coronary arteries. His hemoglobin A1C was 9.6 percent when he began cardiac rehab, met with a registered dietitian and certified diabetes educator, and started to limit his carbohydrates to no more than 45 grams per meal. He followed his prescribed medication regimen and worked hard to manage his blood glucose levels, and in September 2010 underwent coronary artery bypass grafting (CABG) with placement of an implantable cardioverter defibrillator.

Arthur began growing vegetables, first for himself and his wife, then for family and friends. His garden became so prolific that he brought fresh vegetables to doctor visits during growing season each year. He had 31 appointments with a diabetes educator or endocrinologist from 2009 to 2015, approximately one visit every three months. He lost 18 pounds, maintains his low-density lipoprotein (LDL) or “bad” cholesterol at 72, and has persistent total cholesterol elevations in the 200 to 300 range, despite a combination of atorvastatin, Fenofibrate, and fish oil. Arthur has maintained his hemoglobin A1C at or below 7 percent from 2011 to 2015 and his blood pressure at 124/58 with medication. Now age 80, he enjoys an active retirement and visits from his grandchildren and plans to continue his lifestyle changes for many years to come. Despite the challenges of his disease, our ProvenCare Chronic Diabetes program has enabled him to maintain an active and enjoyable lifestyle.

The 2006 expansion of our value reengineering portfolio from ProvenCare Acute to ProvenCare Chronic made sense for several reasons. After the success of elective heart surgery and interventional cardiology acute care reengineering and the significant amount of external validation, in both the academic arena and the popular media, our hoped-for flywheel effect occurred dramatically. Additional Geisinger service lines and discipline-based areas of the organization wanted to get in on what they could do for their patients to attack both total cost of care and suboptimal outcome issues. Our success also improved our ability to recruit extraordinarily bright people to join the Geisinger family and further our innovation machine.

Many hospital-based service lines, including our most innovative first service line, community practice, began to contribute their own ideas about fundamental reengineering of care for the most prevalent diseases in their patients. This was welcome for two reasons.

First, it demonstrated that our major strategic goal of fundamental innovation could in fact be disseminated into the various discipline-based and multidisciplinary service lines. Not only was there top-down demand to achieve a common high-level strategic aim, there was bottom-up demand regarding goals for individual caregiving entities. The entire effort could not have been done without combined top-down strategic discussion and agreement and a bottom-up ability to define specific goals that were compelling to our people who were actually taking care of the patients. In addition, affirmation in both professional pride of purpose and total compensation was uniform and aligned throughout the entire organization. The top-down strategic insistence plus the bottom-up buy-in to individual provider-led patient care reengineering was the winning combination for getting everyone throughout the organization incentivized and energized to the Geisinger concept of a healthcare innovation engine.

The second reason we welcomed the enthusiasm for reengineering the management of prevalent chronic diseases was that almost every acute care episode was a window into a much larger, ongoing chronic disease management problem. Quite simply, doing an effective coronary artery bypass or placing a stent for a clogged coronary artery relieved the immediate problem, but did not change the overall challenge of long-term outcome in patients with systemic vascular disease. The interventional surgery was not a reset button, and it did not change the ultimate biology that caused the blockage. Only the combination of the effective intervention plus a fundamental reengineering of the patient’s and the doctor’s approach to the chronic disease would ultimately expand life and functionality.

Because ProvenCare Chronic would require close cooperation between primary care physicians (PCPs) and specialists, we sought assistance from our community practice service line (primary care) leaders to help identify the specific chronic disease we would tackle first.

COLLABORATING ON CHANGE

As with ProvenCare CABG, we wanted to start our chronic disease reengineering effort with a high-impact, high-probability winning result. There is strong incidence of diabetes in the Geisinger service area, nearly one million adults age 18 and older, according to the Pennsylvania Department of Health.¹ Our community practice doctors were caring for approximately 30,000 type 2 diabetes patients, and we had just recruited a full complement of excellent endocrinologists at our two hospital hubs.

Type 2 diabetes was challenging because it involved a number of departments and caregivers not typically collaborating to benefit patients, including endocrinologists, PCPs, pharmacists, nutritionists, general internists, and nurses, among others.

Our starting point was to entice the hospital-based specialists to open their hospital-based clinic schedules to diabetic patients in crisis. But our aspiration for ProvenCare chronic disease management went well beyond simply opening up schedules and being responsive, because it’s unacceptable that patients must travel from wherever they are and from whomever is taking care of them to see hospital-based specialists.

The endpoint for ProvenCare Chronic reengineering for care of all chronic conditions was to identify ahead of time the patients at most risk for medical crisis and fundamentally change our care for them before they go into crisis. We wanted to get as much collaborative best practice care to patients near where they live, with the entire provider group (and at Geisinger, the payer as well) committing to achieve a common metric of all-or-none best practice bundle-of-care measures delivered to the patient in the community setting. Some of these best practices are taken from the discipline-based evidence and consensus process led by the specialty societies, and some are decided upon internally as part of the default best practice socialization process.

The reengineering approach to chronic disease care requires a fundamentally different interaction between specialists and PCPs. From the beginning of the reengineering effort, we insisted on bringing care to patients with extraordinarily difficult type 2 diabetes management, rather than demanding that they come to us. Instead of opening up schedules for these patients to be seen when necessary at the endocrinology-based clinics typically near our hub hospitals, we systematically took our endocrinology expertise out to the community practice offices. This fostered interaction between endocrinologists and our PCPs when patients with type 2 diabetes were in or approaching a crisis.

The only way bundled best practice works, and to some extent it’s used as a forcing function, is if data from payers is used to stratify which chronic disease patients need the most intense care. Initial redesign for all type 2 diabetes patients would have been useless and incredibly costly. For the most fragile diabetes patients, for example, our initial goal was to hone in on those patients requiring the most intense care and to meld the rapidly changing specialty knowledge of the endocrinologist with the access, general management, and credibility in the community-based practitioner. This was a superb way of creating more patient-centric care delivery without sacrificing the expertise that prior to our bundled best practice and care reengineering demanded that patients physically move from their community practice-based interaction to the specialists in or near the hospital hubs.

We took a similar approach to congestive heart failure, involving hospital-based cardiology specialists and community practitioners in a way that enabled a significant amount of caregiving for the most difficult patients to be provided in the community practices near where patients lived, as opposed to simply opening up schedules and demanding that patients and their families travel to hospital-based hubs. Almost 80 percent of the patients normally referred for specialist visits could be cared for much more efficiently by having specialists available to the PCPs in our 55 community practice sites. Our outcome metrics are decreased acute care needs, decreased frequency of secondary disease consequence, and decreased cost of care over time, the ultimate increased value outcome.

There’s another compelling reason for specialist and PCP collaboration in reengineered prevalent chronic disease care. Without making the most expert opinion available to frontline caregivers and caring for both healthy and sick patients, we could not feel confident that the best care was delivered in the most convenient way to our sickest patients. Working together toward this goal was directly correlated with the 20 percent innovation-related compensation targets for the specialists as well as the PCPs.

PERFORMANCE MEASURE SET

The ProvenCare approach to diabetes management is a team-based model of care that uses the ProvenCare methodology to help practitioners manage type 1 and type 2 diabetes patients in the primary care setting. The three-pronged approach combines work flow improvement, information technology (IT) optimization, and performance measurement. The system helps caregivers proactively manage their patient population’s compliance with a set of nationally recognized performance measures. Based on these measures, providers can pursue appropriate chronic condition management for their patients. Specifically, the diabetes management system of care includes:

•   An all-or-none set of 14 measures for diabetes that tracks patient compliance to evidence-based guidelines. (See Figure 8.1.) The measures provide a consistent way to manage the diabetes patient’s health based on best practice care, and all measures are required in the all-or-none measure set.²

FIGURE 8.1   Diabetes Patient Compliance Measures

•   Clinical process redesign to eliminate, automate, delegate, incorporate, and activate.

•   Clinical decision support through the electronic health record (EHR) at clinic nurse and provider levels (evidence-based alerts and health management reminders).

•   Patient-specific strategies using registry report data.

•   Activation strategies such as patient letters and e-mail communication via secure patient portals.

An all-or-nothing measures set raises the performance bar by more closely reflecting the interests and desires of patients, fostering a systems approach to achieving all goals, and providing a more sensitive scale for assessing improvements. Both patients and physicians want to either slow disease progression or prevent the consequences of additional diseases that might be avoided by more optimal treatment. We presupposed this could happen only if all the known best practices for a given condition were achieved every time for every patient. So we committed to a best practice bundle even though there could be either medical or practical issues mitigating optimal achievement for individual components of the bundle.

Not all patients will achieve each measure; for example, not all will quit smoking. The set of measures offers real-time feedback regarding progress by the patient and in the population. The measures also attempt to stratify the type 2 diabetes patients most at risk and to enable much more proactive input from the endocrinologist in addition to the PCP. Finally, the measures also seek to include patients and their families in a self-care partnering arrangement to achieve the best possible outcomes.

We included patients and families in each of the care delivery reengineering processes, redesigning the care pathways, delineating new responsibilities for providers, patients, and their families in jointly defined accountability to achieve optimal outcome, and fundamentally reframing the relationship between the caregiver and patient. Even giving patients and their families access to our progress notes was a fundamental realignment. Finding out how often the patients and their families did not understand or agree with what was documented in their progress notes was eye-opening. Setting a new baseline of mutual understanding and agreement was an important starting point in optimizing chronic disease management.

Our initial approach was to use our EHR, Epic, employed across the entire Geisinger system, to embed the provider prompts and feedback enabling behavior. We now are working on bolt-on and content embedding applications that would enable connections to Epic, Cerner, and Athena Health.

In the beginning, our PCPs and endocrinologists committed to achieve nine best practice goals for the type 2 diabetes patient population. The first few years focused essentially on the usual surrogate markers, such as hemoglobin A1C, microalbumin, pneumococcal vaccination, LDL, blood pressure, and so on. We eventually included 14 best practice measures.

To the aggravation of most community practice leaders, whenever there was a year-over-year improvement in the process or surrogate panel, particularly since it would always meet or beat the innovation requirements for the performance part of compensation, Dr. Steele would ask, “So what?” He was interested in the actual long-term benefit to diabetes patients included in the improved best practice bundle.

Remarkably, it took only three years of this fundamentally changed set of practice incentives and practice enablers to show that the answer to “So what?” meant that there were 306 prevented heart attacks compared to what would have been expected; likewise 141 prevented strokes and 166 prevented cases of retinopathy, simply by having the patients cared for within this bundled best practice value reengineering change.³ In addition to the patient benefit, bundled best practices significantly decreased the total cost of care. Value was increased by both improving quality and lowering costs.⁴ (See Figure 8.2.)

FIGURE 8.2   Diabetes Bundle Exposure Impact on Total Medical Cost of Care ($ per Member per Month)

The bundled best practice sets then were expanded from the initial 30,000 diabetes patients to almost 20,000 patients with coronary artery disease and to more than 260,000 patients who were placed under a series of care best practices for prevention purposes. The specific preventive care metrics depended upon whether the patients were young, middle-aged, or old. Most important was the commitment of PCPs to consider everything known in the literature about prevention as a “must-do” for their patients, with rational compartmentalization regarding what was appropriate for various age groups, lifestyles, and behaviors.

For the type 2 diabetes patients, the most important ramification of our care change was diminishing long-term disease consequences. The economic benefit of decreasing the need for hospital care and treatment for diabetes-related diseases went straight to our insurance company’s bottom line. As usual, a financial deal was made between Geisinger as provider and Geisinger as an insurance company, but similar to ProvenCare Acute, the care reengineering was expanded to include all of our type 2 diabetes patients, no matter who insured them.

Based on the ProvenCare Chronic diabetes outcomes, we experienced the following benefits:

•   More efficient care processes were created.

•   Patients were identified as to when they were likely to need additional care.

•   Providers were empowered to carry out their own transformational change as they gained experience and knowledge. Doctors strive to be at the top of a performance list, and when they are in the middle or lower quadrants, there is automatic pressure to improve. At Geisinger, there actually was a best practice competition between community practice sites and among individual providers. We try to understand what is being done better in one group and transmit it to other groups not performing quite as well.

•   Patient outcomes were improved with individual measures of care such as influenza vaccination rates, hemoglobin A1C at goal, and LDL at goal. Hemoglobin A1C at goal increased 45 percent over a seven-year period for a study population of 25,000 people, and LDL at goal increased 18 percent over the same period in the same population, even after establishing more strenuous goals. Most important, there also were reduced rates of stroke, myocardial infarction, and retinopathy in the same population.⁵ Most important, these intermediate and performance metrics subsequently were shown to link to better diabetes-related disease outcomes, less need for acute care hospitalizations, and longer, more functional lives.

•   Compliance increased across all measures within the set. For the nine original measures (percentage of influenza vaccination, percentage of pneumococcal vaccination, percentage of microalbumin result, percentage of hemoglobin A1C measured and at goal, percentage of LDL measured and at goal, percentage of blood pressure less than 130/80, and percentage of documented nonsmokers), compliance in the study population of 25,000 increased from 2.4 percent to 14.5 percent over a seven-year period. Within the first year of implementation, compliance went from 2.4 percent to 7.2 percent.⁶

•   IT was used more fully to reinforce the new roles of practice site staff.

For all of the IT enabling that was part of the care reengineering, the key was to change who did what and how the care was actually delivered to patients. The transactional EHR, the functional content added to it, and the analytics that came from the claims and clinical data were useful only in direct linkage to changing the entire care pathway. Both at Geisinger and in the literature, it was obvious that chaos would ensue if the care pathways were not changed at the same point in time for the increased IT usage.

Further, IT both enabled and reinforced the changing roles of the care team and the changed care pathway. The enabling technology and the new pathway had to be easier than what was done previously, because change would not occur if the pathways were more complex than the ones already in use.

For example, when the suggested new pathway for autism didn’t fit this criterion, we accepted it as a failure and did not adopt it. On the other hand, we created and adopted a very successful new care pathway regarding the use of erythropoietin (EPO) in patients with anemia associated with chronic renal disease. The new approach was adopted only when the transactional EHR-enabled best practice algorithm for EPO treatment could be applied to pharmacists and pharmacy techs and withdrawn as a responsibility of the nephrologist. Only through the new approach could significant benefit be shown in tightening indications (for example, using EPO only when iron would not be equally beneficial) and in increasing the efficiency of the actual EPO treatment through algorithm use transacted by techs and supervised by pharmacists and doctors. There are many additional examples of how the best practice bundle was systemized throughout our entire community practice and endocrinology specialists. It could not have been done without strengthening the fundamentally changed care pathway with enabling technology.

PROVIDER PROCESS

To identify a patient with type 1 or type 2 diabetes for inclusion in the diabetes bundle best practice measurement set and to trigger future alerts, the provider must select a diabetes diagnosis and add it to the patient’s list of health problems in the EHR. When the diagnosis is entered on the patient’s problem list, the patient is automatically in the diabetes registry. The registry for diabetes management lists all patients in the practice who meet the measure set criteria.

The diabetes diagnoses offer caregivers the opportunity to select a specific diagnosis that matches the current state of the patient’s condition and allows for patient-specific goal setting. This provides the additional information to ensure accurate measurement.

For example, a typical diabetes diagnosis such as “diabetes mellitus without mention of complication, type 2 or unspecified type, not stated as uncontrolled,” will have specific ProvenCare diagnosis options that map to the same root International Classification of Diseases code. Examples of ProvenCare best practice codes include: diabetes type 2, goal A1C < 7; diabetes type 2, goal A1C < 8; diabetes type 2, goal A1C < 9; diabetes type 2, goal A1C to be determined; and diabetes type 2, goal symptom management.⁷ The specificity of these ProvenCare codes allows the care team to track the patient’s diabetes measure progress and ensures that all staff are aware of the goals for the patient and are focusing appropriately on the problem. In addition, these specific codes are used in the diabetes set of measures reports.

The presence of a diabetes diagnosis in the patient’s list of health problems will trigger health management reminders, with the EHR system automatically posting the patient-care activities for the evidence-based protocols. Activities can be reflected as due for care, overdue for care, or care completed. Completion of the activity is captured based on information contained in the EHR or other health tracking tools.

The provider can view a summary report for a particular patient before entering the examination room. This report provides an update of the relevant information for treating diabetes based on care protocols and assists the provider in preparing for the actions that should occur during the particular office visit. The diabetes summary report provides the following information on the patient:

•   Allergies

•   Current medication list

•   Body mass index

•   Social history

•   Blood pressure, pulse, height, and weight from the past two office visits (if available)

•   Diabetes labs for the past three results over two years

•   Most recent immunizations/injections

•   Summary of patient care activities, indicating via a symbol what is late, due, due soon, or on hold

The provider will address any alerts displayed for the patient. Patients and their family members share information to assist the provider in making complex diabetes care decisions based on combining information from the EHR and/or other clinical systems such as lab values, patient care activities, and the diabetes diagnosis. The provider reviews each alert, selects the appropriate care action, then accepts the alert to satisfy the action. How these specific activities are satisfied is outlined in Figure 8.3.⁸

FIGURE 8.3   EHR Alerts

CLINICAL ORDERS

At the end of the office visit, the provider reviews and signs the orders the nurse has noted as pending during the patient rooming process. The provider has the option to sign all the orders at one time, edit an order, or remove orders that are unnecessary based upon information captured during the office visit. If not already completed, the provider determines what care is due based on the diabetes protocols. The provider reviews the patient’s health problems to decide if the diabetes diagnosis is still applicable and adjusts the treatment as necessary. New problems are added to the patient’s problems list in the EHR. For subsequent visits, administrative staff can inform patients that they are overdue for certain care and at the visit prepare orders for the provider. Any appropriate clinical staff members, who also view alerts for best practices, can perform procedures such as diabetic foot screenings.

Performance data for the diabetes management set of measures is displayed in multiple management reports to aid operational and clinical staff in monitoring and addressing performance on a monthly basis. Practice site directors, operations managers, clinic staff, and providers access appropriate reports for the site they are responsible for at the level of detail needed. We obtain individual patient and individual site feedback in near real time and use the variation in performance to determine why one site or one individual is doing better than another. This must be part of the socialization process for provider behavior change to occur.

PATIENT OUTREACH

It’s important for diabetes patients to become active partners in their care with the caregivers they see regularly for ongoing diabetes care and other health issues as well, since most diabetes patients have multiple issues. In addition to our clinicians developing such partnerships with their patients during their office visits, Geisinger uses patient self-management and regular chronic disease “communications” (both letters and e-mails) to encourage patient involvement. The self-management messages explain the patient’s current diabetes condition and offer suggestions for clinical care. They encourage the patient to become a member of the care team. Much of this outbound communication has been enabled by our Epic patient portal and the recent systemwide rollout of progress notes being available to patients.⁹

The chronic disease communications are designed as targeted outreach to encourage patients to seek care by scheduling an appointment. The communications are automatically generated monthly to patients who meet the following criteria: older than age 65; PCP within the Geisinger system; diabetes diagnosis present on the patient list of health problems or a diabetes diagnosis used more than four times at an office visit; no appointment scheduled with a PCP in the next four months; no chronic disease management visit scheduled in the next four months; did not receive a chronic disease management phone call; and did not receive a chronic disease letter in the past six months.

ACHIEVEMENT

As reported in the American Journal of Managed Care, a study of claims data for Geisinger Health Plan (GHP) members meeting the criteria for a diagnosis of diabetes found a “significantly lower risk of macrovascular and microvascular disease end points in the first three years of a diabetes system of care that included an all-or-none bundled measure compared with primary care without this intervention.… Perhaps the most notable finding is the apparent early impact of the care model. The findings suggest an impact in the first three years with the possibility that a reduction in risk began to emerge after the first year.”¹⁰

Another study published in this journal utilized GHP claims data for patients exposed to our diabetes system of care who met the Healthcare Effectiveness Data and Information Set criteria for diabetes and had two or more diabetes-related encounters prior to 2006. This group was compared to a second group of patients from 2006 to 2013 who were not exposed to ProvenCare Chronic. The study found that, “Over the study period, the total medical cost saving associated with bundled best practice exposure was approximately 6.9 percent. The main source of the savings was reduction in inpatient facility cost, which showed approximately 28.7 percent savings over the study period. During the first year of the bundled best practice exposure, however, there were significant increases in outpatient (13 percent) and professional (9.7 percent) costs.”¹¹

There were two reasons why costs were higher at the beginning. First, before any steady state was achieved in many cases, patients generally were seen more frequently either at the community practice offices, in their homes, or in skilled nursing facilities to ensure that everything was done to achieve the bundled best practice. Second, a significant amount of the benefit in achieving the best practice goals for these bundles came through improved medication adherence, which was viewed to be a worthwhile trade-off. If pharmacology costs went up but the consequence was significantly decreased need for emergency room visits, office visits, and ultimately hospitalizations, the net gain both in terms of quality of outcome for patients and decreased total cost of care was extraordinarily worthwhile. But there was a lag in the decreased hospitalization benefit until after some period of increased pharmacologic adherence was achieved. The overall benefit in quality outcome and decreased total cost of care was a twofold value increase in our diabetes population.

PATIENT CASE STUDIES

Creating better outcomes for patients was the key to energizing our doctors and team to develop and implement the reengineering innovations. The following patients presented to our PCPs with extremely poor diabetes control and consequent high risk of developing diabetes complications. By working with diabetes management clinic pharmacists and other members of the care team, the necessary medication and lifestyle adjustments were made to improve care over a relatively short period of time. As a result, none of these patients had their disease progress to nephropathy, retinopathy, neuropathy, or vascular disease.

Candice is a 34-year-old patient referred to one of our clinics for diabetes management and education. She presented with a baseline hemoglobin A1C of 11.9 and no previous education about diabetes care. She was not tolerating her only diabetes medication, experiencing stomach upset. Our physicians and pharmacists worked with Candice to switch to an extended-release version of the medication and slowly increased the dose to a tolerable and effective level. Working with our team, Candice was able to develop a meal plan and exercise routine to fit her lifestyle. After six months, her hemoglobin A1C improved to 6.5 and was at goal. Despite being at high risk for diabetes complications at a young age, Candice changed the trajectory of her health by partnering with our team.

Marie is a 44-year-old patient referred to a diabetes management clinic for disease management and education. She presented with a baseline hemoglobin A1C of 11.1 and no previous education about diabetes care. She was on Lantus insulin and glimepiride, but admitted that she was not compliant with the medications because she felt defeated by her diabetes and had gained weight since starting them. She had been on metformin in the past, but the medication was discontinued because she could not tolerate the nausea and intestinal distress. Our physicians and pharmacy team worked closely with Marie to adjust her medications. The pharmacists replaced glimepiride with Victoza and added the extended-release version of metformin, slowly titrating the dose based on Marie’s tolerance. Over the next 10 months, we worked with Marie to make dietary improvements and continued to adjust her medications. Her diabetes control improved significantly, with her hemoglobin A1C decreasing to 6.6. Not only was her diabetes better controlled, she required less insulin than at baseline and was working toward continued weight loss in an effort to become less dependent on medications to maintain her health.

Matthew is a 35-year-old patient who came to one of our diabetes clinics for disease management and education. He was recently diagnosed with type 1 diabetes and had a hemoglobin A1C of 14.1. He was overwhelmed by his diagnosis, as he had just started a family of his own and suddenly life as he knew it was changing. Our physicians and pharmacists started him on intensive insulin therapy and followed him weekly to make necessary adjustments to his dosing. Our nutritionists and pharmacists provided Matthew with a thorough education of his disease state, including carbohydrate counting, exercise, sick-day rules, and self-care principles. Working with all members of the interdisciplinary primary care team, Matthew was empowered to control his diabetes by making adjustments specifically tailored to his lifestyle. After just three months, his diabetes was significantly improved with his hemoglobin A1C down to 5.5. More important, Matthew had gained an understanding of the active role he plays in his diabetes care and felt confident knowing he could now maintain his health and avoid the many complications of this disease.

LESSONS LEARNED

•   It’s possible to apply default best practice to how chronic disease is managed.

•   Provider-led, technology-enabled commitment to a bundle of best practices for diabetes begins to change medical outcomes in one year.

•   As medical outcomes improve, total cost of care decreases.

•   Payers, PCPs, and hospital-based endocrinologists must work together to improve where and how care is provided.

•   Patients and their families are key partners in redesigning and receiving care.

•   Innovation at the highest level of institutional strategy must be transacted by providers energized to help patients to better long-term outcomes.

•   Success in care reengineering creates a flywheel effect.

•   Socialization of fundamental care redesign must be consistent and consistently affirmed throughout the organization.