7

ProvenCare Acute: Taking It to the Next Step

At age 63 and as CEO of Geisinger for seven years, Glenn Steele believed he was in good health. Although he was working 24/7 and under a good deal of self-imposed stress, he was running several miles a day, maintaining strength and flexibility, and skiing regularly. All of Dr. Steele’s male family members died of heart or peripheral vascular disease at a median age of 50, but they all had smoked cigarettes or cigars, never modified their predominantly meat-based protein diet, and believed that deep fried meant great taste. In addition, they all had vigorous and stressful professional or business careers. It had been 40 years since his father was among the first generation of patients having the groundbreaking coronary artery bypass graft (CABG) procedure for occlusive coronary artery disease.

Annual wellness exams and cardiac stress tests were all normal, but Dr. Steele saw his general internist and a cardiologist after noting a significant change in his resting pulse rate. Nothing else—no pain, no change in exercise tolerance—just a pulse rate of 85 when it previously had been 60. In retrospect, there were some brief unexplained episodes of shortness of breath, as well as some decrease in speed and intensity when skiing during the past few years.

Dr. Steele’s coronary angiogram showed occlusions not easily corrected by angioplasty or stent placement. Shopping for the desired opinion and a less invasive approach would have been possible, but the combination of internist and cardiologist opinion and Internet literature search all pointed to the preferred therapeutic approach, open heart surgery.

As a physician and healthcare leader, Dr. Steele had extensive access to publicly available and private data on cardiac surgery outcomes at all the best places and on the world’s best cardiac surgeons. He was well known enough that he would have been treated as a VIP (not necessarily a good thing in getting the best care) no matter where he decided to have his care.

Why did he choose to stay in his own system? With comparable individual outcome numbers for the best institutions and the best individual practitioners, what differentiated one institution from another in this superb group? It was how explicit the system was in ensuring that it actually functioned as a system.

What was the evidence, from the time of diagnosis through rehabilitation and into secondary prevention of an ongoing chronic disease, that everyone involved was incentivized, enabled, and motivated to work together to achieve an optimal outcome? What was the evidence that the system not only was committed to but actually had achieved continuous improvement, not just in the setting of a formalized, randomized clinical trial but all the time? What was the evidence that the fundamentally asymmetric relationship between caregiver and patient had begun to change to achieve a true therapeutic partnership, so that even if a bad outcome occurred, the patient would understand that both partners had worked together to minimize the chances of complication and, in the case of Geisinger’s ProvenCare, that the caregiver would fix the problem at no extra charge?

All of these components critical to achieving the highest probability of best outcome were built into Geisinger’s design and transaction of ProvenCare. The ProvenCare method is a structured approach to provide quality care at a reduced price for acute hospital-based surgical procedures and for select chronic diseases managed in an ambulatory setting. ProvenCare incorporates current evidence-based best practice elements into the work flow to reduce unwarranted variation in indications for a test or procedure and in the delivery of care processes. The goal is to extract the 35 to 40 percent of services and costs that do not produce benefit or may actually harm those served.

Return on investment in ProvenCare includes (1) improved clinical outcomes; (2) increased efficiency in resource use and patient throughput; (3) reduced total cost of care; (4) enhanced patient and provider satisfaction; and (5) program differentiation. We assumed that doing this care design reengineering would give us a market share advantage because of the unlocked value produced.

GETTING STARTED

To establish a baseline before care reengineering, the easiest way to learn how much unjustified variation is built into any routine caregiving is to ask a simple question at a key site of care. For hospitalized patient interventions, the recovery room is one of the most critical sites. Determinants of good outcome following most major surgeries are prevention of infection and blood clots that migrate from the leg veins to the lungs.

To determine the degree of care variation following surgery, ask the recovery nurse, “What’s the antibiotic protocol being used? Or what’s the blood-clot prevention protocol?” If the nurse asks you to wait a moment to check who the patient’s doctor is, that is all you need to know! This is straightforward confirmation of no consensus driving best practice and of continued “seat of the pants” non-evidence-based healthcare delivery. Providing care is similar to every other complicated task in life; the more unjustified variation is involved in the process, the more cost will be incurred and the higher the probability of a bad outcome.

Consider the following attributes of typical acute care in America today:

•   Uncertain appropriateness

•   Limited patient engagement

•   Unreliable compliance with evidence-based guidelines

•   Lack of accountability for outcomes and quality

•   Incomplete communication across the continuum of care

•   À la carte payment for services

•   Perverse incentive with more payment for more complications

•   Stunning geographic disparities in frequency of care

•   Widely variable outcomes not explained by severity stratification

•   Often an inverse relationship between quality of care and cost of care

Compare this to the attributes of a high-performance health system, where U.S. healthcare desires to be:

•   Designed to achieve high-quality, safe care

•   Access to care for all people

•   Efficient, high-value care

•   Capacity to continuously improve

How do we get from where we are to where we want to be? The following classic approaches are no longer reasonable or effective: overdependence on diligence and hard work; benchmarking to the mean (which ensures mediocrity); permissive clinical autonomy; and inadequate use of human factors knowledge in reliability science (understanding human error).

ProvenCare is a formalized process initiated to document appropriateness of care; establish evidence- or consensus-based best practices for all high-frequency services; reliably deliver default best practices by redesign of complex clinical systems that embeds new behaviors into everyday patient flow; activate patients and their families and engage them in the care processes as symmetrical partners with caregivers; and provide a packaged price for the episode of care, including a so-called “warranty” that transfers the risk for the financial effects of preventable complications to the caregiver.

There are six core components to our ProvenCare Acute program (reengineering of hospital-based or hospital-associated episodes of care): (1) documenting the appropriateness of care; (2) establishing all key elements of evidence-based or consensus-based best practices; (3) socializing and making routine the complex clinical systems that embed default best practices into new provider behaviors for everyday patient flow (the new care processes must be easier than what’s being replaced); and (4) activating patients and families so they are fully engaged in the care process redesign and implementation (caregivers and patients working together to achieve optimal outcomes). The last of the ProvenCare components comprise the product packaging: (5) negotiating a single bundled price for the entire episode of care with the payer or the actual buyer of the care, in the context of an employer self-insured buyer; and (6) transferring risk for the financial effects of preventable complications to the provider and to the health system via the bundled payment.

There are seven stages in the implementation of ProvenCare Acute:

1.   Start-up

2.   Literature review, best practice elements identification

3.   Current state

4.   Process redesign and electronic health record (EHR) tool development

5.   Database and report build

6.   Soft go-live

7.   Go-live

Best practice elements are derived from evidence-based literature or from the consensus of the in-house multidisciplinary team. A given best practice is included only if there is 100 percent provider agreement to include it. At any time, a member of the multidisciplinary team can request changes in the best practice elements.

Performance measure sets embedded into the team’s work flow, through the EHR or bolt-on EHR applications, and all-or-none compliance for each patient is monitored in real time. For example, the appropriate antibiotics begun at the appropriate time preoperatively and stopped at the appropriate time postoperatively.

The structure of the team is critical to socializing the behavior change, and the weekly time commitment for the various team members must be made explicit for the reengineering to succeed. Consider the weekly time commitment various team members must make for the reengineering to succeed. (See Figure 7.1.)

FIGURE 7.1   Reengineering Implementation Time Commitments

It’s important to understand operational complexity in the current care pathway before redesigning the process. Two preexisting beliefs represent the most frequent objections to change. The first is, “We do not have unjustified variation here. Our providers are driven by common goals and shared value systems, and we come to consensus on a regular basis.” The second is, “We already have informally established our best practice consensus, and we are certain it is uniformly applied by our superbly trained, highly productive specialists, hence our good outcomes.”

Inventorying the actual breadths and depths of variation in performance and the lack of knowledge of this variation begins the behavior change process. For example, when beginning our inventory of the indications for surgery and best practice elements throughout the treatment process, our seven cardiac surgeons couldn’t imagine there was significant variation. That was until we discovered all the differences in antibiotic usage and timing of antibiotic treatment and the variety of opinions about routinely obtaining carotid ultrasound before elective CABG and the benefits of routine continuous blood glucose monitoring. It was eye-opening when even these closely knit, incredibly collegial cardiac surgeons understood how they individually varied and what they each thought was the default best practice for their own patients. Process redesign for the defined care pathway begins with eliminating non-value-added work. We prefer work that can be performed using technology and, whenever possible, work is delegated to trained nonphysician staff. For example, nurse specialists are assigned to follow the patients’ continuous blood glucose and begin educating patients at the end of the inpatient stay to attempt to influence patient behavior for those found to be prediabetic. New work flows are incorporated into the practice and hardwired with reminders through either EHR tools or bolt-on apps to enhance reliability and efficiency of care.

Most important, patients and their families are activated as essential partners in the redesign and implementation processes. Specifically, patients and families were involved in the group that constructed the best practice algorithms, and potential patients are asked to commit to a “patient compact” under which they are part of the best practice default monitoring process and agree to stop smoking cigarettes or cigars prior to surgery. Patients above their ideal body weight commit to altering their diet and activity levels, both before and after elective surgery.

Databases are created that permit real-time monitoring of all-or-none compliance with the best practice elements in each patient, including financial metrics allowing cost analyses before and after redesign and providing immediate justification for individual provider deviation from a given best practice. Largely to defend against the accusation of forcing “cookbook medicine,” we allow provider variation from default best practice as long as the provider documents the variation and justifies it to his or her professional colleagues. Most individualized and non-evidence- or consensus-based variations in key elements are a function of learned behavior or idiosyncratic beliefs held by the provider. Neither of these justifications stand up to peer scrutiny.

It took 18 months from the time we defined our first ProvenCare design process through a soft go-live and ultimately a fully automated process implementation. The time was condensed as we added more care episodes to the redesign portfolio and as success with increasing quality and decreasing total costs of care became obvious to the clinical leaders. Throughout the effort, administrative and clinical commitment to attacking unnecessary, redundant, or hurtful care became almost a religious belief in striving for higher quality and has remained the most important success factor. Otherwise, too many other priorities would have taken precedence.

BETTING ON IMPROVED OUTCOME

The bundled payment package became a financial bet that care redesign would significantly decrease the cost structure while increasing quality outcomes by reducing complications. We looked at all complications during the 90 days after a given acute care episode and calculated our total costs for caring for these bad events, regardless of whether they were causally related to the original treatment episode. We simply negotiated a discount into the single price for the initial episode, including the costs of tests, interventions, the pharmacy, doctors, the hospital, and the post-acute phases of the care pathway. To articulate the single price, we needed to know the pre-redesign cost of care for the episode and either know the individual key element costs or assume that by removing unnecessary key elements the postdesign costs would be significantly less. The bet on how much improvement would occur from diminished complications was simply a negotiation. We halved the cost number that was ascribed to historical complication rates and the cost for treating those complications. Thus the care redesign had to improve the outcome by a factor of two for us to break even.

Negotiations with non-Geisinger payers have been based on our knowledge of how low our price could go and still allow us to make a margin, assuming our care redesign would be as effective in increasing quality and decreasing complications as it was during the beta test. The other aspect of negotiations with non-Geisinger payers was how low a price they needed to take the risk of incentivizing their members or employees to travel to a distant center of excellence for purportedly higher-quality, lower-cost care.

Another important aspect of our so-called “warranty” in a single-priced package was its simplicity. As providers, we were taking financial risk for the quality of our caregiving and, over time, acknowledging that the unlocked value in the reengineering process itself allowed for some of the value to be kept for the system’s sustainable business model, some of it to go directly back to the buyer of care, and some to go to the men and women who had changed their behaviors, leading to better quality at lower cost for our patients.

Another critical success factor in implementing the ProvenCare redesign process was linking our system’s top-down strategic commitment to innovation to the bottom-up individual service line commitments to ProvenCare goal setting. We used a reliable, consistent set of individual compensation metrics as an engine for affirming success in improving both quality and cost outcomes. We revised our professional compensation plan at the beginning of the care redesign era to link 20 percent of our providers’ total compensation to strategic goals only and not to relative value unit (RVU) type of productivity. We chose the 20 percent arbitrarily, and the non-RVU component easily could have been a higher percentage or significantly less. The importance was that a significant portion of our clinical leadership’s and the actual caregivers’ financial well-being was now tied to accomplishing the redesign of care pathways and sharing the benefit with patients and with the buyers of care.

ProvenCare would not have happened at all or would have been an evanescent innovation at best without this combination of leadership commitment to redesign, an almost religious belief in unlocking value through care process improvement, the professional pride of purpose in our clinicians seeing significant quality benefit coming to their patients, and redistribution of some of the unlocked value to those paying for the care.

EARLY SUCCESS

Fortunately, the first of our hospital-associated value reengineering projects, ProvenCare CABG, turned out to be a home run in every way. We chose this as our initial high-volume, high-impact bundled episode of care because we thought it had the greatest probability of quick success. Most importantly, we had motivated clinicians who already had established enthusiastic and effective clinical leadership throughout Geisinger in both cardiac surgery and cardiology. Our outcomes for elective CABG were already excellent, as defined by the external benchmarking Pennsylvania Health Care Cost Containment Council database, and robust systems collecting outcome data already existed both within Geisinger as well as nationally.

Significant volumes of care and an understanding of adverse outcomes, both for the patients and the financial aspects to the system, allowed rapid assessment of impact on quality and cost. Most of the analysis, either evidence-based or consensus-based, of what should be done every time for every patient from the diagnosis of coronary artery disease through rehabilitation had already been processed by the cardiology community and the cardiac surgery disciplines. What had never been done was a systematic care pathway reprogramming, enabling all of the best practices to be the default for every patient who was diagnosed with coronary artery disease and proposed to be a candidate for CABG. So ProvenCare Acute really began as a sociology experiment.

One of the most important aspects of systemizing care is documenting the appropriateness of the proposed care intervention in the first place. Once again, CABG was an excellent beta test, because there were unambiguous, authoritative criteria of absolute indications for intervention (Class I appropriate); ambiguous indications but still legitimate circumstances (Class IIa and IIb criteria); and unambiguously non-indicated intervention (Class III).¹ So the process at the beginning formally could exclude individual patients when there was evidence-based or consensus-based lack of indication for the procedure, obviously the easiest way to eliminate unnecessary or harmful cost.

The next important step is establishing the team to do the hard work. Our teams include a clinical champion, typically the leader of the discipline or service line; an operations leader, the administrative partner of the clinical champion; all physicians who participate in the actual episode of care; all nursing staff and their surrogate leadership who participate in the care; EHR staff, who determine where the behavior changes are embedded, either directly or through bolt-on applications; a reporting analyst; and a project facilitator.

The process begins with establishing the best practices; identifying the population, disease, and procedures to be reengineered, as well as when the episode begins and ends; selecting which hospitals are involved (initially at Geisinger, a combination of owned and nonowned); deciding whether the reengineering includes the post-acute care rehabilitative process; and determining how much responsibility is taken for some or all complications over how long a period of time following discharge from the acute intervention.

Literature then is searched for guidelines and outcome metrics from all appropriate authoritative sources. For example, CABG sources included the American College of Cardiology, Society of Thoracic Surgeons, Agency for Healthcare Research and Quality, and American College of Surgeons National Surgical Quality Improvement Program. Straw man guidelines are documented if guidelines from the literature search are not readily available. An advocate is assigned to each guideline believed to be key in impacting patient outcome. Validation is done through critique of the literature, advice of experts, and a consensus-driven process among all members of the team. It is critical to translate from generalities in the guidelines to specific ProvenCare behavior: What behavior needs to be accomplished? When in the care episode does it need to be accomplished? Who is accountable for the behavior? What is the process to track the care in real time? What are the opt-out provisions for justified variation from the default best practices?

The review team established its own commentary on the American College of Cardiology Foundation/American Heart Association Task Force guidelines for CABG indications. The surgeons on the review team vetted and validated each of 12 Class I and Class IIa guidelines, translated 40 verifiable, actionable behaviors with clear-cut accountability and timeline definitions, and achieved unanimity and complete buy-in to default best practice to be accomplished every time for every patient. The recommendations that became key default best practices in version one of ProvenCare CABG are found in Figure 7.2.

FIGURE 7.2   ProvenCare CABG Recommendations

But did ProvenCare CABG really work? There was only one way to know, as far as Dr. Steele was concerned: He had to try it. He was patient number 86.

The prehospital physical and educational preparation, the hospital acute care redesign, and the default best practices, plus his motivation to be out of the acute care setting as quickly as possible, resulted in a two-and-a-half-day hospital stay. So Dr. Steele felt very busy most of the time as an inpatient. When the transitions of care nurse visited him on the way into the post-acute setting, he was pretty well exhausted. And when told that continuous blood glucose monitoring indicated probable prediabetes, he finally gave in to stimulus overload and responded flippantly that at his age, pre-anything was a victory!

There were 144 separate entries into his EHR, all audited and legitimate, an incredible biopsy into the number of caregivers uniformly intent on achieving a perfect outcome. Everything observed during Dr. Steele’s care journey and documented as results afterward would be used to improve the next patients’ care process. Knowing that many fellow Geisinger family members had seen him in his medically neediest moment, and in a skimpy hospital gown, was a small price to pay for being a beneficiary not only of individual commitment and brilliance but of an entire system’s commitment to excellence and continual improvement. Dr. Steele was fully functional and back to work in one month. Clinical outcomes and reliability and financial outcome information is found in Figures 7.3 and 7.4.

FIGURE 7.3   Clinical Outcomes Pre- vs. Post-ProvenCare CABG Protocols

FIGURE 7.4   CABG Reliability and Financial Outcomes

GROWING THE PROVENCARE PORTFOLIO

With the success of ProvenCare CABG, other service lines at Geisinger were eager to adopt the ProvenCare approach for their particular service. We decided to move forward next with hip surgery. Two separate, but connected, value reengineering approaches were applied to high-volume hip surgery. One patient cohort was the group undergoing elective hip replacement, and the second was hip fracture, most often in elderly patients, the so-called “fragile hip.” For these two conditions, the composition of work groups was mostly similar for establishing baseline variation in care processes and developing consensus on default best practices for ensuring optimal care during each part of the patient flow. Since elective hip starts in the outpatient clinic, an important aspect of default best practice starts with a decision to operate in the first place. Since care for fragile hip almost always begins in the emergency room after the hip fracture, the default best practice begins after the indication for surgery. Both of these pathways, however, include default best practice all the way through complete functional rehabilitation. (See Figures 7.5 and 7.6.)

FIGURE 7.5   ProvenCare Acute Total Hip High-Level Flow

FIGURE 7.6   Fragile Hip High-Level Flow

To reiterate, common principles apply to all ProvenCare Acute process redesign projects: eliminating non-value-added work; automating work as much as possible; delegating work to nonphysicians whenever possible so physicians have more time for direct patient care; incorporating new work flows into the practice; hardwiring with EHR tools or bolt-on applications to enhance reliability and efficiency of care so the new behavior becomes easier than the old behavior; and activating the patients and their families so they become partners in striving for perfect outcomes.

The starting point for ProvenCare Hip and ProvenCare Fragile Hip was different than ProvenCare CABG because, at that time, uniform agreement on indications for elective hip replacement surgery through how and where the post-acute rehabilitation should be accomplished was nowhere to be found in any of the discipline-based consensus arenas. Even national registries collecting a standard set of success metrics post-hip and post-knee replacement were not established until the past few years, to some extent catalyzed by the ProvenCare CABG results. In addition, the extraordinarily close connection between device manufacturers and their sales organizations has led to many orthopedists in a number of systems (not Geisinger) to prefer specific devices because of financial self-interest, not necessarily because of what is a consensus best practice for their patients.

As a result, socializing the best practice default in hip took significantly more committed clinical and administrative leadership, particularly at the start of the process. The consensus was that determining default best practice was much more of an internal process than simply modifying and applying off-the-shelf, already available discipline-based rules regarding who should have surgery, what devices should be used, and how and where all the key best practice elements were required every time for every patient. Even before the present orthopedic registries were initiated, the work was accomplished and the benefit to patient quality and patient cost outcomes was apparent.

Uptake on the all-or-none process changes was similar to what we experienced with ProvenCare CABG, but with optimal performance that never reached 100 percent physician compliance. This was undoubtedly a function of the internal socialization process that until recently was without significant external validation or help from discipline-based off-the-shelf outcome metrics or best practice evidence or consensus.

Reengineering sustainability has come most importantly from sharing the improved patient outcomes, which affirms professional pride of purpose. In addition, the benefit in lowering the total cost of care has enabled additional throughput at lower cost per patient, potentially increasing market share, and enabling redistribution of some part of the unlocked value back to those who provided the improved care processes. This was accomplished as a direct add-on to the non-RVU total compensation, the 20 percent linked to achieving individual and service line components in the overall system’s strategic commitment to innovation. Our Geisinger data on ProvenCare Hip clinical and financial outcomes is found in Figures 7.7 and 7.8.

FIGURE 7.7   Clinical Outcomes: Hip

FIGURE 7.8   Financial Results: Hip

Expansion of the initial ProvenCare hospital-based redesign pathways now includes a wide spectrum of high-volume interventions and high-volume hospital-associated care pathways. Our ProvenCare portfolio includes ProvenCare Autism, ProvenCare Bariatric Surgery, ProvenCare Cellulitis, ProvenCare Chronic Obstructive Pulmonary Disease (COPD), ProvenCare Coronary Artery Bypass Graft (CABG), ProvenCare CNS Mets, ProvenCare Epilepsy, ProvenCare Fragile Hip Fracture, ProvenCare Heart Failure, ProvenCare Hepatitis C, ProvenCare Hysterectomy, ProvenCare Inflammatory Bowel, ProvenCare Lung Cancer (Commission on Cancer Collaborative), ProvenCare Lumbar Spine, ProvenCare Migraine, ProvenCare Multiple Sclerosis, ProvenCare Percutaneous Coronary Intervention, ProvenCare Perinatal, ProvenCare Psoriasis, ProvenCare Rectal Cancer, ProvenCare Rheumatoid Arthritis, ProvenCare Total Hip, and ProvenCare Total Knee.

SCALING AND GENERALIZING WITH PROVENCARE LUNG CANCER

The success of ProvenCare Lung Cancer is especially compelling, because of huge variation and noncompliance nationally with the most important staging and preoperative requirements for potentially resectable lung cancers of specific histologic types. Patients were not receiving generally agreed-upon evidence-based care, as defined by a number of national discipline-based and cancer-based organizations.

Since 2010, Geisinger’s ProvenCare template for surgical treatment of lung cancer patients has been piloted by a dozen hospitals of different sizes and models and with more than 2,000 patients through the Commission on Cancer’s ProvenCare Lung Cancer Collaborative.² The study involves patients with non-small cell lung cancer who are candidates for lung resections.

Participating hospitals provide treatment according to Geisinger’s ProvenCare Lung Cancer care pathway, containing 38 standardized elements that cover everything from antibiotic administration and pain management to mediastinoscopy and lymph node sampling. Results show compliance with the 38 elements rising to nearly 90 percent from below 40 percent when the study began and similar to what we experienced at Geisinger.

Researchers now are utilizing the findings, amassed in the Society of Thoracic Surgery National Database, to determine five-year clinical oncological outcomes of these patients, including staging accuracy, interoperate lymph node yield, and pneumonia and other respiratory complication rates before, during, and after the ProvenCare journey. The collaborative expects to see improvement in outcomes as a result of institutions following the ProvenCare pathway.

A new phase of research is looking at all stages of lung cancer. The program will now involve multiple departments at eight hospitals, including medical and radiation oncology, and encompass 53 elements as opposed to the 38 we started with, everything from diagnosis to staging, treatment, survivorship/palliative care, and end of life care related to lung cancer.

At the very least, we have demonstrated that the application of the process goes well beyond a single committed institution and is motivated by clinical leadership and pride of accomplishing real improvement in patient care, since none of the non-Geisinger institutions have the advantage of a payer/provider partnership that built in a pricing component as a part of the effort.

LESSONS LEARNED

•   Continuous improvement is possible in routine clinical practice.

•   Default best practices can improve quality and lower costs.

•   Unjustified variation can be mitigated without resorting to “cookbook” medicine.

•   Scaling to other institutions is promising.

•   Professional pride of purpose and clinical leadership are the key success factors.