Integrating Research and Practice: Health System Leaders Working Toward High-Value Care: Workshop Summary (2015)

10

Continuous Learning as an Executive Agenda Priority

One of the goals of the workshop series was to explore the challenges and opportunities that health system leaders see with respect to creating a continuous learning environment within their institutions. Over the course of the workshops there were several sessions that dealt with the issue of continuous learning as a priority for health system executives. Jonathan Perlin, the president of clinical services and chief medical officer at HCA, described an example of an effort that was successful in integrating research and practice and that resulted in cost savings. A panel discussion, moderated by Michael McGinnis of the IOM, sought to identify and prioritize the key issues for health systems leadership in moving toward a system that more tightly integrates care and knowledge-generating activities. This panel, which also discussed whether a shared value proposition is the key to sustainability, consisted of Raymond Baxter, the senior vice president for community benefit research and health policy at Kaiser Permanente; David Labby, the chief medical officer at Health Share of Oregon; Patricia Smith, the president and CEO of the Alliance of Community Health Plans; Janice Nevin, the chief medical officer for the Christiana Care Health System; and Russ Waitman, an associate professor of internal medicine at the University of Kansas Medical Center.

A second panel discussion, moderated by Lewis Sandy, the executive vice president for clinical advancement at UnitedHealth Group, featured short comments by four health system leaders: Glenn Steele, Jr., the president and chief executive officer of Geisinger Health System; Ronald DePinho, the president of the MD Anderson Cancer Center; Rodney Hochman, the president and CEO of Providence Health and Services; and Steven Corwin, CEO of New York-Presbyterian Hospital. That panel, which continued the explorations of the first panel, identified further opportunities for making learning activities an executive-level priority. Both panel discussions were followed by an open discussion among the panel members and workshop participants.

BEYOND THE REDUCE MRSA TRIAL

To provide an object lesson in how system executives’ activities can address issues that are relevant not only to the national agenda but also to the success of their institutions, Jonathan Perlin discussed his perspective on the REDUCE MRSA trial that Susan Huang had previously described. He also used this example of a successful data-enabled trial to illustrate some of the organizational challenges to the conduct of pragmatic research within a health care organization and to highlight some potential solutions to those challenges.

From his perspective, Perlin said, the REDUCE MRSA trial aimed to tackle a major problem for health systems—the hospital-acquired infections

that affect about 4.5 percent of all hospitalized patients and that result in some 80,000 deaths annually. Approximately one-quarter of the patients infected and about one-quarter of the patients who succumb are infected with either MRSA or some other form of Staphylococcus infection. Prior to the conception of the REDUCE MRSA trial, HCA had already been using a procedure, modeled on what is done in European hospitals, that screened patients for MRSA and then isolated them if they were positive. With support from the CDC, and together with other academic partners, HCA tested the effectiveness of this approach against two other promising strategies: decolonizing MRSA-positive patients with a chlorhexidine antiseptic sponge bath and 5 days of antibiotic therapy delivered via nasal ointment versus universal decolonization of everyone prior to their entry into the intensive care unit.

Over 18 months the REDUCE MRSA team tested the three procedures in 74 ICUs at 43 hospitals using a cluster randomization design. Ultimately, some 75,000 patients received one of the three interventions, Perlin said, and the clear winner was universal decolonization, which not only reduced MRSA infections but reduced all bloodstream infections by all pathogens by 44 percent. Putting these numbers into terms that are germane to health services researchers, Perlin said that for every 99 patients who are treated, one bloodstream infection was avoided. “This occurred on top of every other best practice, and so it really set a new standard for reducing bloodstream infections,” he said. Addressing the bottom line, Perlin estimated that for every 1,000 patients admitted to the ICU, HCA saves $170,000, or a total of $19,720,000 per year in institutional enterprise benefit.

Perlin said noted that although this was obviously an important outcome, an equally important benefit of the REDUCE MRSA trial was that it took a mere 18 months to complete because it drew upon archived data for baseline results and involved 43 hospitals, each contributing data. Perlin estimated that had this study been done at one hospital, it would have taken 64 years to treat enough patients and accumulate enough data to gain the power needed to answer the research question. “So it didn’t take 1 hospital 64 years, it took 43 hospitals 18 months,” he said. Moreover, this study was not conducted in an academic unit but rather in standard hospital settings using routine health workers. “What we think is particularly powerful about this is that it answered real-world questions in real-world environments that we believe generalized to real-world situations,” Perlin said, adding that the total cost for the study was only $3 million. In contrast, the slightly smaller Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial cost $80 million. The REDUCE MRSA study was so effective, he added, because of the federal and academic partners that participated in the study and that enabled HCA to leverage its internal expertise in infection prevention. Those partnerships were important for

supporting the business case for this study, which in the end provided a powerful return on the investment that HCA made to support the study.

Perlin acknowledged that he and his fellow health care executives often worry more about costs than about the promise of benefits, and he counseled the workshop attendees to pick questions to study that provide both operational and financial opportunities. He also said that this type of research activity, when aligned with organization priorities, can amplify and accelerate ongoing quality-improvement activities. It is time, he said, for health system administrators to challenge their information technology infrastructure and to look for opportunities to collaborate to address deficiencies in organizational capabilities. For example, HCA selected its academic partner for the REDUCE MRSA trial for its expertise in administering large clinical trials and in patient consent issues. It may also be necessary to challenge a health system’s clinicians to embrace research that has the purpose of improving performance, which if successful will generate pride that is self-reinforcing.

Why participate in such studies? Perlin said that the fundamental reason is to contribute to addressing the big questions that concern the nation regarding health care. Another important reason is to improve the care of individual patients while improving the sustainability of the health care system. “You can answer meaningful questions and support quality improvement by fostering evidence-based research that is oriented to performance improvement,” Perlin said. Participating in such research also provides a “first adopter advantage for improvement innovations,” he said, and there is also a strong business case to make for supporting this type of research. Health care systems have already made significant investments in information technologies to meet the meaningful use criteria, and participating in research networks creates the opportunity to leverage that investment with those of other organizations and answer questions that can be best addressed using data from multiple organizations. This type of research, Perlin said, provides the opportunity to bridge the translation gap that separates knowledge and practice. “Here is the opportunity to create knowledge out of practice itself and create a learning health system.”

HCA is now designing a study to determine best practices for preventing and reducing surgical site infections associated with cardiac and orthopedic procedures. This study will involve 50 hospitals and 400,000 patients and be conducted in 1 year. Another study will aim to identify best practices for detecting and treating sepsis more effectively. “With 20 million patient encounters, with hundreds of thousands of ICU encounters, and with physiologic data, we believe that we can find early markers, create a standard definition, and perhaps look serendipitously for processes or treatment interventions that work more favorably,” Perlin said. “We also want to look at the relationship to improve antimicrobial stewardship.” A

third study, with partner Vanderbilt University, will look at how transitions of care should be optimized for safety.

Ultimately, the main reason that HCA participates in this kind of research is that “we believe we can improve health care, we can improve value, we can improve the sustainability, and we can improve the health system,” Perlin said. “We think it’s fun and enjoyable to be part of a learning health system.”

In response to a question from Steele about whether HCA is “monogamous with randomized controlled trials,” Perlin said that the answer was no and that the organization is open to other validated methods of conducting CER. Commenting on the methodological aspects of the trials, Richard Platt of the Harvard Pilgrim Health Care Institute, which was the main academic partner on the REDUCE MRSA trial, noted that the reason it was possible to do this study so effectively was that it relied solely on information that was collected as part of routine care and on data that was available in HCA’s clinical data warehouse. “The reason that the randomized trial gave such a solid answer,” Platt said, “is that it was built on the basis of 12 months of historical data that was exactly analogous to the data from the 18 months of the study. Had we not had that baseline data, the study in fact would have been substantially underpowered, even at 75,000 patients.”

Platt also mentioned that the NIH Collaboratory is working at a conceptual level to identify the problems associated with embedding pragmatic trials in clinical settings and to identify ways to deal with them, an effort that is complementary to PCORnet’s role in building the infrastructure to conduct such trials and use what the Collaboratory is developing. Eric Larsen, the workshop planning committee chair, added that one observation that the Collaboratory made is that there is a great deal of standardization and training needed before trials of this sort can be run effectively, a comment with which Perlin agreed.

PRIORITY OPPORTUNITIES FOR CEO LEADERSHIP TO MAKE A DIFFERENCE

The final session of the first workshop drew on the previous day and a half of presentations and discussions to identify and prioritize the key issues for health systems leadership in moving toward a system that more tightly integrates care and knowledge-generating activities. The panelists each delivered brief, prepared remarks on this topic and then entered into a period of discussion among all workshop attendees. As a prompt for the panelists’ remarks, session moderator Michael McGinnis of the IOM listed several questions for them to consider in a strategic effort to achieve a closer, sustained alignment of research and practice:

• What might be the core benefits that would appeal to a CEO?
• What key infrastructure elements are needed?
• What obstacles need to be addressed?
• How can patients and families be enlisted as active allies?
• What is the most important step that a CEO can take to make that happen?
• What perspective will be the most attractive to board support?
• What policy signals or actions are important for engaging purchasers and research funders in this effort?
• What might be some measures of progress by the year 2020?

Raymond Baxter, Kaiser Permanente

In his remarks, Baxter commented that researchers working alone will not be able to address the needs of the health care system and that finding solutions will require a central role for operational leaders, clinical leaders, and patients. He noted that Kaiser Permanente refreshed its research strategy some 4 years ago so that the organization could answer clinically meaningful questions more quickly and efficiently. This reorganization included creating the Center for Effectiveness in Safety Research (which is charged with building the infrastructure needed to integrate research and care), expanding the company’s biobanking operations to a national level, and most recently, setting up its data portal and joining PCORnet.

At the same time, a team of Kaiser Permanente researchers, acting at the instigation of Kaiser’s National Research Council, queried Kaiser’s chiefs of staff and other clinical leaders in the organization to find out what questions they thought most needed answering in order to meet the triple aim. In response to this request, Baxter received 342 questions, which he realized could be an operational agenda for relevant research in a learning health care organization. To his surprise, between 25 and 30 percent of the questions already had answers, though the evidence was not known or available to those who proposed those questions. “That’s disturbing in its own light,” he said, “but in retrospect not too surprising.” Some questions were not posed in a way that made them easy to answer, and in the case of others there was nobody with an interest in finding an answer. Eventually, Baxter and his colleagues settled on five questions that they thought would provide “quick hits,” but 18 months later, none of those quick hits had produced answers. “They did eventually,” Baxter said, “but it is not clear that any care transformation happened from the ability to answer those questions.”

What Kaiser has done since is to examine the full range of analytic capabilities in the organization, from business intelligence to care analytics, and it is now trying to array them in a continuum that can triage questions

so that the questions go to the right analytic unit—the unit that has the right tools, measures, data, and approaches and that can deliver the necessary degree of rigor and precision so that the person who asked the question can take action at the end of the day. “That’s a much more complex process,” Baxter said.

Turning to the questions that McGinnis had posed, Baxter said that CEOs do not want research—they want performance improvement, and they want it at speed. “CEOs are impatient with what we do and what we have done traditionally,” Baxter said. What CEOs can do with regard to infrastructure is to create the forum and the mechanisms by which clinicians and patients can pose questions and where researchers and analysts can work out approaches for answering those questions. “Creating that forum of people working together on that is essential,” Baxter said, “because the researchers cannot guess at what are the questions that operators and patients have in their minds. And operators and patients can’t always pose the question in a way that’s readily translated by researchers.” He added that “a CEO can create not only the mechanism but the culture that says that that’s important.” He noted that while CEOs are sometimes credited with more power than they have, the one thing they absolutely can do is change organizational culture so that everyone is encouraged to participate in research to answer questions that are important for the organization. “That goes against the tradition of investigator-initiated research,” Baxter said, “but I believe that paradigm is going to have to change significantly.”

As far as what CEOs can do to enlist patients and families as active allies, Baxter said that they can insist on greater engagement and model it in their own behavior. “The CEO who talks to individual patients, who reads the complaints of members and patients who are not well served, that’s the kind of CEO that can drive this kind of change in research as well,” Baxter said. Concerning what a board needs to hear, he said that the message that research can affect costs is persuasive, but the message that research can change and improve care is probably more so. “Unless we can organize our research and analytic capabilities in a way that has a demonstrated impact on improving care and improving health,” he said, “I’m not confident that the cost arguments will be effective.” The reason, he said, is that organizations have only a limited supply of intellectual capital that should not be spent on research that is not producing change for patients and members. The barometer for success, Baxter said in concluding his remarks, is the speed at which research results produce changes in care and outcome.

David Labby, HealthShare Oregon

Labby began his comments by noting that, 2 years earlier, Oregon had adopted the concept of using coordinated care organizations for its

Medicaid population. These coordinated care organizations act as a regional health authority that oversees all spending on physical, mental, and dental health for all of the Medicaid patients in a given region. Labby noted that the state cajoled competitive organizations to participate by threatening them with a 30 percent rate cut if they did not join. The carrot was that the state gave these coordinated care organizations the freedom to transform care in ways that best served their interests and the interests of patients at lower cost. The sense of urgency to avoid draconian budget cuts, Labby said, changed CEO behavior so that suddenly competitors were working together to create the structures needed to cooperate on a large scale.

The main obstacle to success, these CEOs said, was the lack of data needed to assess operations and effectiveness. However, within 1 year, Labby and his colleagues had produced a data system that could aggregate information from different health systems. That success created a second problem—how to use the aggregated data in a productive and appropriate manner. Oregon is looking now to hire a person who can help solve this latest problem, Labby said. He added that he thinks that CEOs do not know exactly what questions to ask other than that they know they want to know how to increase system efficiency. What appeals to CEOs, Labby said, is not the development of new interventions but rather answering questions that will help them manage a global budget and integrate mental and physical health in a way that benefits patients and cuts costs.

Reflecting on the workshop’s discussions, Labby said that it is clear that there is a great deal of knowledge that is not being used. “Do we need to generate more knowledge that we don’t use,” Labby asked, “or do we need to figure out how to use the knowledge we already have? I think, from a health system point of view, if we could mobilize what we know in an efficient and an effective way, we would be way further down the pike than if we keep investing in and creating new knowledge that we don’t know how to deploy.”

Labby commended the work that Intermountain Healthcare has done in building infrastructure and the extreme amount of intentionality shown by the Bellin system in the way it approaches research that benefits the organization. What CEOs need, Labby said, is help creating those same structures and organizing their operations with the same intentionality. He noted the pressure that CEOs are facing, given the huge changes that are occurring in health care’s current transitional state between the older volume-driven, fee-for-service models and the newer value-based, capitated fee-for-performance payment models. The old organizational and operational structures will not work in the new model, Labby said, and CEOs need help developing those new structures.

Turning to the subject of patients and families, Labby spoke from the perspective of the Medicaid population with a focus on the so-called high

utilizers. In reality, he said, these individuals are not high utilizers, but rather they are people who have been marginalized and traumatized for generations. “What they need is a trauma recovery program,” he said, “and so our medical home for that population serves as a trauma recovery program.” Because this population needs help connecting with peers in their community, Oregon created a certification program for peer wellness specialists who, once they have been certified, become part of the health care workforce.

Boards of directors are interested in costs and meeting contracting budgets, Labby said. Concerning policies, he suggested that PCORI can seize the current opportunity to develop a new research agenda that helps with health care transformation. Regarding measures of success, he proposed that health care spending being less than 16 percent of gross domestic product (GDP) would be a positive sign that transformation is occurring.

Patricia Smith, Alliance of Community Health Plans

Smith began her comments with the observation that the people attending the workshop speak a different language from those she talks to on a daily basis in her job as a lobbyist—something that she characterized as a major barrier. “If we can’t cross the communication bridge, I think we will have lost a big part of the battle,” she said, noting how important it is to be able to convey the importance of research to learning. She explained that her board is made up of the CEOs of health plans and that, as a result, she has given a great deal of thought to what matters most to them. “Learning clearly matters, but learning in an environment that delivers for the public, for communities, and for the nation, is what really matters,” she said.

One of the facts of life for the CEOs that she works with, Smith said, is that all of their operating funds come from premiums, and premiums are what purchasers care most about. They pay less attention to value. What that means for researchers, Smith said, is that the research community needs to show that their work has relevance to consumers, whether it is in better health, price, credibility, trust, or a combination. Smith reiterated the earlier message that what CEOs value is to fail forward fast, to create value at speed, and to do so in a way that meets the priorities of their organizations.

Regarding how CEOs can engage patients, Smith said that the issue of trust truly matters to patients. Those in the executive suite must translate the need for public trust into a strong governance model that respects learning. Smith concluded her remarks by saying that population health is improving communities and that communities are healthier when the health care systems that operate within those communities are healthier.

Janice Nevin, Christiana Care Health System

To provide some context for her remarks, Nevin said that Christiana Care Health Systems is a large regional, community-based academic health center with three campuses, two acute care hospitals, a rehabilitation hospital, a Visiting Nurse Association operation, multiple outpatient sites, and revenues approaching $2 billion. Christiana serves Delaware, where it is the dominant provider in Wilmington and the surrounding New Castle County with an 85 percent market share, as well as south New Jersey, southeastern Pennsylvania, and Cecil County, Maryland. Because of its market dominance in the Wilmington area, Christiana’s leadership sees itself as a public utility upon which the community depends and which the community expects to be sustainable. For that reason, Christiana sees research and education as critical factors in its ability to serve its community, Nevin said.

In 2011 Christiana created an entity called the Value Institute whose express purpose is to bring together quality improvement and operational excellence and marry them to health care delivery research. “The perspective of our CEO is that the Value Institute is a way for us to focus on value as our fundamental strategy for service,” Nevin said. “This is about more than just quality and cost, but about value grounded in the needs of our neighbors as they perceive them.” Among the institute’s initiatives are studies to identify reliable tools in the setting of oversedation and its role in sepsis and also refining anticoagulation algorithms for patients undergoing surgery. In addition, the Value Institute also has a $10 million Center for Medicare & Medicaid Innovation grant to study how to use new technology to collect and organize data in a way that allows clinicians to use the data to improve the care of patients with ischemic heart disease.

Turning to the questions that McGinnis posed, Nevin said that the core appeal to a CEO of closely aligning practice and research is what is embodied in the Value Institute. “If we can align practice and research,” she said, “there is potential for us to do better in delivering our core mission, and we’re going to invest in that work.” She added that engaging patients and families is a “must do” as part of this alignment and added that a message for CEOs should be to first define how their organizations define “patient-centered.” CEOs, she said, need to embrace the message that several panelists throughout the day had stated, which is to look at partnerships with patients and families as a core business strategy. “Then they can start to provide an infrastructure that not only gives patients a seat at the table but a voice at the table,” she said. “To me, that’s job one.”

McGinnis asked what such an infrastructure looks like, and Nevin replied that it starts with patient and family advisory councils but goes beyond that. Christiana, for example, is embedding patient advisors in every patient care unit and including them in the system’s operational committees.

These patient advisors have become active and important contributors to the successful operations of the entire system, she said. The next step is to bring payers on board as partners and to do a better job of helping patients and families understand outcomes in a way that enables them to make good decisions for their health.

Russ Waitman, University of Kansas Medical Center

Speaking from his perspective as the principal investigator for the Greater Plains Cooperative CDRN, Waitman said that two ways to appeal to a health care system CEO could be to present patient-centeredness as an altruistic goal and to sell it as a way to drive costs down. He said that his CDRN is not seeking funds from the CEOs of the institutions that belong to this CDRN and that this is a plus because, as he put it, “I don’t know that we have formulated a good value proposition yet.” He also thought that patient-centeredness would be a strong appeal to a health system board, particularly for those of nonprofit institutions.

Waitman’s suggestions for how CEOs can engage with patients and families and recruit them as active allies focused on forming a team with Clinical Science Translation Award community engagement groups. Such an alliance could bring community and research perspectives together with the health system perspective and create the potential for generating new ideas. Concerning measures of progress by 2020, Waitman said that PCORI has metrics that it is supposed to meet by 2017 and that these metrics—reduced practice variation, improving the life of patients and those in the community, reducing disparities and variability in underserved populations, and others—can serve as a starting point. Other measures to judge progress should include reductions in resource consumption, increases in the quality of life and longevity, and increased patient satisfaction. Performing these measurements requires data, and data require an infrastructure that captures data efficiently, at low cost, and in ways so that the data can be integrated with other data sources. This last issue—making sure that data can be integrated—is where much of the initial focus of PCORnet lies, Waitman noted. “Making the CEOs aware of what steps are happening in terms of data integration and data structures is going to be important,” he said.

Waitman noted that one of the biggest signs of success would be if barriers for data exchange were lowered. As an example, he cited the difficulty in getting data from the Social Security Administration regarding which patients included in a large database have died. He suggested that policies need to be developed with an eye on how data can be repurposed back to the health system. Thinking about barriers brought Waitman to his final point. Health care, he said, is fundamentally a reactionary business today, and part of the mind-set of this business sector is to be cautious when roll-

ing out new programs that may or may not work in a particular health care system. As a result, one good approach is to iterate rapidly while taking off small pieces of risk, rather than engaging in projects that would require a total reworking of a system. At the same time, he said, systems need to be able to incorporate “blockbuster” developments, such as the development of endoscopic surgery. Balancing these three things in an organization will be a major challenge for PCORnet going forward, he said, particularly with regard to whether to introduce the data analytics that PCORnet will enable slowly or in a way that is more intertwined with a health care system that is fundamentally a reactionary one.

CHALLENGES AND OPPORTUNITIES FOR ENABLING CONTINUOUS LEARNING IN HEALTH CARE ORGANIZATIONS

In order to stimulate discussion about the challenges and opportunities faced by organizations in enabling continuous learning, moderator Lewis Sandy posed four questions for the panelists, each of whom was a leader of a health system, to answer. Panelists each delivered brief, prepared comments on the topic and then engaged in a discussion with all workshop attendees on those questions:

• How does knowledge generation fit on your agenda?
• How does this idea of integration of research and practice fit the rapidly changing environment of policy, practice, and reimbursement?
• How do we speed up knowledge generation and get it installed into a learning health system?
• How do you see this agenda really being advanced in the real world that you live in every day?

Glenn Steele, Geisinger Health System

Geisinger Health System, as Glenn Steele described it, is a “Petri dish for innovation because of our structure and our culture and our demography.” Indeed, the organization’s most important strategic aim is fundamental innovation in how it provides and pays for care, and continuous innovation based on dynamic data feedback is the means by which the organization goes about realizing that aim. Steele said that the focus of the organization’s behavior change over the past 12 years has been both in the provider part of the organization and, for the 50 percent of its patients who are also insured by Geisinger’s health plan, in the payer part of the organization. Looking forward, though, Geisinger is now aspiring to create enabling technology beyond the EHR that will change the behavior of its patients and members. “Basically, what we’re attempting to do, which

relates to our business model,” Steele said, “is create value and either redistribute that value to the people who buy our care or keep some of it for ourselves and redistribute it for more innovation or expansion.”

Addressing what his organization needs from PCORnet, Steele said that he and his colleagues worry that much of its observational data and the approaches that it takes to research are not amenable to RCTs and that they are “extremely nervous” about the biases that are built into the results it gets from using observational data. They also worry that the things it discovers at Geisinger will not be applicable anywhere else, given that Geisinger has built scaling and generalizing results as part of its systemic strategy. To address these fears, Steele said that he wants Geisinger to work with PCORnet on studies that may or may not involve RCTs but whose findings would be scalable and generalizable.

Ronald DePinho, MD Anderson Cancer Center

MD Anderson is an unusual cancer center with the responsibility of delivering multidisciplinary research-driven care to a large number of patients from around the nation and the world. It maintains a national network of health care partners and leads a 29-sister-institution program in 23 countries. It has a large research enterprise that is integrated into its care pathway, and it drives a clinical trials enterprise that leads approximately one-third of FDA approvals in cancer treatments. As such, Ronald DePinho said, MD Anderson has a solid perspective on the cancer field that relates to the quality of care throughout the United States and the world. “There are significant knowledge and competency gaps,” he said, “as cancer is an extremely complicated disease, requires multidisciplinary care, and is very technology intensive and knowledge intensive with respect to how to apply, for example, genome profiling to the care of patients.”

When he arrived at MD Anderson 3 years ago, DePinho said, the institution had a homegrown EHR with 70 different transactional systems, all of which are now being standardized into an Epic-centered EHR. In addition, there were 50 different research databases and a number of independent tissue banks, each of which was useful but not integrated or aggregated in a way that enabled the system to learn as much as it could from what was happening in its clinical operations or its research laboratories. DePinho said that he considers the organization’s EHR as merely another transactional system with which to ingest the vast amount of information coming from its clinical care enterprise. The consequence of these different clinical and research systems being so disconnected was that a study aimed at determining the cost of each step in the care of a cancer patient required manual curation of data to understand the outcomes value and economics

of each of several dozen steps involved in caring for patients with head and neck cancer.

Today, MD Anderson is implementing one EHR system and an eResearch platform that are fully integrated with one another. The organization piloted these new systems with leukemia, for which it already had standardized most of the processes involved in diagnosing and treating each patient. In addition, DePinho said, the organization is in the process of including more than 1 million patients and their legacy data into the same big data warehouse to which powerful analytic tools such as IBM Watson can be applied. Such an effort has many challenges, he said, and among the greatest is ensuring data quality upon ingestion. The organization’s major focus now that it has created this big data environment is to have the right kind of analytics interface to produce continuous learning that will drive optimal patient management and acquisition of new knowledge.

DePinho commented that most of the workshops’ discussions had been provider-centric and that there is a great need and opportunity to create a health/wellness system that is consumer-centric. Toward that end, MD Anderson is developing mobile platforms, decision-support cognitive computing systems, and interchange systems in order to advance the paradigm of care to systems that can reach the individual out in the community. He also said in closing that there is no difference at MD Anderson between clinical care and research. “Two-thirds of patients do quite well with standard of care,” he said, “but for a third of cancer patients who fail standard of care treatment, their standard of care is clinical trials.” DePinho concluded, “Research-driven, multidisciplinary care of the patient is the standard of care for MD Anderson.”

Rodney Hochman, Providence Health and Services

Providence Health and Services, Rodney Hochman told the workshop, is the third or fourth largest nonprofit health care system in the country, with 4,500 physicians and 35 hospitals providing care in Alaska, California, Montana, Oregon, and Washington. Providence Health owns a Catholic High School in Burbank, California, which Hochman says is the best place to go to find out what 15- and 16-years-olds are thinking, and it is a leader in public housing in both Oregon and Washington, enabling it to more closely study the social determinants of health. Providence Health also has the single largest installation of the Epic EHR.

With his headquarters in Seattle, Hochman has been aggressive about hiring former Amazon employees who have a strong background in and appreciation for customer care. Providence’s head of innovation strategy, for example, had not worked in health care before joining the organization, yet his customer- and consumer-centric approach is bringing a fresh look

at how Providence tackles innovation and designs its digital platform to enable advances in health care that are patient-centric.

Hochman said that his goal for Providence is to treat a third more patients at a third less cost, and the only way to get there, he said, is to have a robust digital platform to enable the kind of research that these workshops have highlighted and discussed. To get to that goal, he and his colleagues have picked out five areas across the 35 system hospitals for which Providence will first examine best practices every 3 months and then involve relevant clinicians in redesigning care to reflect those practices. “We’ve taken the approach of putting the clinicians in charge of how we are redesigning care,” Hochman said. He added that his belief has gone from one that felt that standardization over scale is the way that health care will change to one that holds that innovation over scale will make a difference in how medicine is practiced. He concluded his comments by saying that the health system could make significant improvements if would take action on those things that existing evidence already shows are effective at improving care and reducing costs.

Steven Corwin, New York-Presbyterian Hospital

Steven Corwin said that he and his colleagues at New York-Presbyterian Hospital regard PCORnet as a strategic imperative because the current framework for taking cost out of the system is, from his perspective, a blunt instrument that uses price and utilization but little knowledge. Taking cost out of the system will not happen without moving toward population health and providing value, and both of those steps require knowledge of the sort that a learning health system can produce. In the absence of knowledge to refine the blunt measurement of cost and utilization, Corwin commented, he fears that the health care system will bifurcate into one that has hospitals that treat the “haves” and hospitals that treat the “have nots,” which would be problematic for the country as a whole. In his view, Corwin said, data platforms and EHRs by themselves are insufficient to provide the real-time data needed to improve workflows in a way that will change behaviors. No matter what the incentives are, physicians are not going to create uniformly structured notes from every patient interaction in a way that will record every single detail that needs to be studied. One solution, he said, would be to move toward systems that are able to extract information using natural language processing.

Another problem that Corwin described relates to organizational culture, particularly with regard to the trend to merge systems and getting everyone in an organization to buy into a culture that is dedicated to continuous learning and improvement. Another issue that concerns him involves what is known now versus what will be known in the future. “The

evidence of yesterday is not the evidence of tomorrow,” he said, “and to say that we should not be in a continuous learning environment, I think, is problematic. Just think about the way we used to take care of bleeding ulcers.” Finally, Corwin said in his concluding comments, privacy and security issues need to be addressed in a way that balances transparency and liability.

DISCUSSION

Cost and Payment

David Posch of the Vanderbilt University Hospital and Clinics started the discussion by offering some general concerns. He noted that for 50 years hospital systems have been living in a period of seemingly unending money that has now, at last, ended. “Money is shrinking now, and that is a fundamentally new phenomenon in health care,” he said. “We have not in our generation had to deal with that, and we don’t know how to deal with that reality.” In his organization, for example, volumes have been higher than ever, but revenues fell by $120 million in a single year, forcing him to make cuts immediately or risk the ire of his board and the system’s bondholders. What guides him are three things that he uses as his “true north.” “First,” he said, “the health system is fragmented from a patient’s perspective as they try to manipulate themselves through the system. Second, we fail to apply evidence every time to every patient, and, third, we don’t engage our patients and families effectively. Anything we do has to solve those three problems.”

Stephen Grossbart of Catholic Health Partners said that one of the big challenges that he sees is that quality departments are not adapted to using financial data and presenting it in a way that finance people understand. “So from a CEO perspective,” he said, “placing expectations on finance and quality to work collaboratively would be helpful.” He stated that every hospital system chief financial officer should be a quality champion, in contrast to the situation that he sees today, where the finance department does not understand the relationship between quality and savings.

Corwin said that he does not see how the nation can take cost out of the system unless the nation engages in population health more effectively, and accomplishing that task will take a long-term strategy that is currently lacking. “I’m sure that in the short run we can use PCORI and other things to take cost out of the head of hospital care, but unless hospitals think of themselves as beyond the four walls and their systems, ultimately I don’t think it will mean anything,” he said. He noted that his system and others are taking expensive populations and reducing the cost of caring for those populations by putting in place resources such as community health and

mental health workers. His system has seen admissions and emergency room visits drop by 20 to 30 percent as a result of taking this type of population health approach. Steele added that reductions of that scale should be the norm and that any system not realizing those types of reductions will be in trouble going forward.

Scott Armstrong of Group Health commented that the biggest impediment that he sees in reforming the health system is the speed at which the payment system transforms from the current fee-for-service basis. Steele replied that Medicare is already moving aggressively to a fee-for-performance basis, and even the residual fee-for-service payments are dropping per unit of work performed. “That’s a real motivation to look at extracting as much cost as possible,” he said, “even for that residual fee-for-service payment.” Steele said that Geisinger is already working in an environment in which half of its revenues are capitated, which he says is terrific in that it puts a true focus on the total cost of care. Hochman added that health care system CEOs need to “get over” the ambiguity of where the nation is headed regarding fee-for-service and population health and focus on doing what is best for patients, and costs will take care of themselves. He noted that a contract that Providence recently signed with Boeing included many provisions about quality, which he says gets to the same point—doing the right thing for patients.

Uma Kotagal from Cincinnati Children’s Hospital Medical Center asked if there are ways of reducing the transaction costs associated with operating a learning health system, noting that in her view that systems are still “clunky” and that integration across a health system is still not routine or streamlined, which leads to high transaction costs. Hochman replied that the learning network has to become more real-time. “You can’t come to a meeting every 6 months and figure out what people are doing,” he said. “It really has to be day-to-day.” Real-time feedback will both increase the speed of adoption and reduce transaction costs. “The learning that comes out of a network like this has to be in real time and has to be continuous,” Hochman said, “and we have to figure out how to do that.” Steven Lipstein from BJC Healthcare said that tapping into PCORnet will help reduce transaction costs by increasing the size of the database with which to conduct research and generate knowledge without having to resort to having to build a dataset from the ground up with each research project.

Jeffrey Grossman from the University of Wisconsin Medical Foundation reiterated earlier comments about the challenges of executing and getting desired outcomes on a system-wide basis. “I’m really fascinated by those organizations that have managed to connect their ideas with outcomes,” he said. “We’re very good, but we’re not where we need to be.” Grossman also spoke of the challenge from the perspective of a physicians group associated with investing significant funds to develop learning systems when dealing

with a health system that still operates on a fee-for-service mentality and of the need to move research beyond what happens within the confines of the health care system facilities into communities to study and address all of the external determinants of health at the population level. That issue, he said, gets to the question of what the business model is for health care systems. “Is our business really health care delivery,” he asked, “or is it the health of the populations for which we’re at risk?”

Lipstein also offered a suggestion for his fellow CEOs who are investing in patient-centered outcomes research infrastructure projects. “If you’re going to make an investment in patient-centered outcomes research infrastructure this year,” he said, “don’t budget the return on investment this year. This is not a quick activity no matter how sophisticated or efficient we become.” Lipstein also pointed to the need for sustainable investments in patient-centered outcomes research. He said funds for those investments can be carved out of money currently being put into new equipment and instrumentation as well as from marketing and advertising. “Carve a little bit out and put in place a sustainable investment in patient-centered outcomes research and expect it to position you for long-term success,” he advised his fellow CEOs, “because then you will be the creators of new knowledge, and that will be an advantage for you in your respective parts of the country.”

Randall O’Donnell from Children’s Mercy Hospital commented on the confusion between coverage and access to care. “Medicaid coverage does not equal access,” he said, “and the farther you happen to live away from a metropolitan center where there is a mission-driven organization that runs primary care centers, the less likely it is that you’re going to have access.” What is needed, O’Donnell said, is organized access through medical homes, adequate case management, and a global payment system, but what is happening in the real world is that states are growing frustrated with Medicaid and are turning to proprietary plans that will, he said, “stick with fee-for-service until the cows come home because they feel that they can still eke out just that extra penny of profit if they’re managing on a fee-for-service basis as opposed to locking in their profit by globally farming out the cost to a willing provider.” Selby said that part of the agenda for improving health systems should be to study new models of coverage, including access.

Disparities and the Safety Net

When asked to provide his thoughts on what could be done to make his job as a health system CEO easier, Joel Allison from the Baylor Health Care System said that one of his biggest challenges is figuring out how to allocate resources and capital to create a learning health system and engage in the research needed to improve quality and reduce costs while also hav-

ing to serve the system’s community in the role of a safety net provider. “How do you make sure that whatever you are implementing is going to improve the health not only of the individual but that whole population?” Allison asked. Forming better connections between clinical excellence and medical education research is one step, he said, but he said he worries about reimbursement for some of the initiatives his health care system is enacting that do reach out to the larger population. He added that the transition to a pay-for-performance system must happen faster if the goal is to move more quickly to a population health approach. Scott Hamlin from Cincinnati Children’s Hospital Medical Center agreed with that comment, while wondering where the next few hundred million dollars will come from to make that happen and to keep current efforts to develop a learning health system going during this transition.

Addressing the issue of whether the business is health care delivery or the health of the populations for which they bear risk, Lipstein said that the issue of how to compare performance between systems that treat predominantly affluent communities and those that treat largely disadvantaged populations is a real one that does affect the business model. “The way CMS has this laid out, you can win easily by avoiding disadvantaged populations,” he said, and although CMS believes this is not going to happen, it is in fact taking place. “If you just look at the distribution of readmission rate penalties across the United States, they’re not happening in Scottsdale, Arizona, and they are happening in Detroit, Michigan,” he said. McGinnis added that CMS is seeking to form a sounding board for issues on this population health dimension that he encouraged the CEOs to join, and Lipstein asked the IOM to join the National Quality Forum in recommending that PCORI put together a special committee on disparities.

Priorities and Challenges for PCORI and PCORnet

Paul Viviano from the University of California, San Diego, Health System said that one of the biggest challenges his organization faces is that its multiyear implementation of its EHR system has been, in his words, a disaster. “There’s no way to calculate the expense and the impact this has had on our health system,” he said. “Our faculty, they’re resentful and they’re angry and they don’t want to talk about the next phases of this even though they’re desperately needed. And so it has been a huge distraction for the organization in every conceivable way.” Another challenge, he said, lies in linking the data from more traditional research with the data that are being accumulated within the health care system through the course of proving clinical care. This is as much a problem of academic silos as it is about EHR vendors creating such links, he said.

One issue that Posch identified was how to scale solutions across what

are becoming ever larger health care systems. Instead of introducing a change in a single or even several hospitals, he now has to worry about scaling across 50 hospitals and 4,500 physicians. In terms of PCORI’s mission, what he sees as the most pressing issue is how to execute and implement today’s knowledge at scale because, as he put it, “I’ve got to make those cuts now.” He urged PCORI to study the science of execution at scale so that all of the discoveries that PCORnet and other initiatives will produce will enable him as a hospital CEO to realize savings and recreate a clinical enterprise intelligently.

Jonathan Tobin from the Rockefeller University Center for Clinical and Translational Research and Robert Dittus of the Vanderbilt University Medical Center both agreed with the suggestion that PCORI needs to fund studies on implementation. Dittus said that, in particular, PCORI could help identify the data that need to be collected to better understand the variables involved in implementation. “The science of implementation is in its infancy, much like clinical epidemiology was in the early 1980s, and we should build on the shoulders of the people who built that science,” Dittus said. Joe Selby said that PCORI is well aware of the need for implementation science and believes that it is within its mandate to fund such studies. He added, though, that one of PCORI’s critical strategies is to get research started correctly by asking the right questions from the start, which he said should make implementation go more smoothly.

Russell Rothman from the Vanderbilt University Medical Center said that he thinks there are great opportunities for synergy between the tools that the CDRNs are building and the help that health systems are requesting. He cited three specific examples, the first of which was the work aimed at developing methods for extracting clinical data from health systems in a way that can be used for reporting and accountability as well as looking at variations in care, inefficiencies in care, and opportunities for improvement. The second example was the work the CDRNs are doing to expand the capacity to collect novel data, particularly patient-reported data and outcomes, and the third was providing tools for real-time, evidence-based clinical decision support.

Corwin commented that the future of precision medicine based on personalized genomics will depend on complementary data collection and new analytical tools to handle what will eventually be enormous datasets of dissimilar data. “I feel extremely strongly that the EHR that we have today will not be what we’re using tomorrow,” he said, adding that it will take an enormous investment to develop the capacity to handle and use these big datasets. DePinho agreed completely with that assessment and said that PCORI provides an overarching framework for the community to come together to think about these challenges and opportunities. Regardless of the solution that is developed, the standardization that PCORnet is facilitating

will be key to creating systems that can ingest and process the torrent of information that is coming. Hochman added that he hopes that more effort is made to break down the barriers between all of the different disciplines that will need to work together to find solutions to the big-data problem.

Rita Redberg said that she is enthusiastic about the idea of PCORnet as a way to facilitate practical clinical trials because there are some things that will be impractical or impossible to study using the gold-standard randomized clinical trial. She was not as excited about clinical decision supports because, she said, they are often not as good in execution as they are in theory, often because physicians are already fatigued by all of the information in an EHR and more often than not ignore the messages that decision support systems provide in the EHR. She also questioned whether embedding appropriate-use criteria into the EHR will address issues of overuse and underuse, and she encouraged systems to start including harms as well as benefits in the data collected.

John Gallin from the NIH Clinical Center commented on issues of data ownership and data sharing and wondering if PCORI should develop guidelines and policies regarding data ownership. DePinho replied that projects such as the Human Genome Project and the Cancer Genome Project have already developed guidelines for data ownership and sharing that health care systems considering issues such as academic promotion and tenure decisions can apply. He added that getting academics to buy into those policies takes effort, which should not be underestimated, but that in the end the researchers involved in those two projects realized that they gained far more from being part of a network and sharing data than they would have if they had worked alone. Corwin, referring to a three-institution data-sharing agreement that involves Columbia University, Weill Cornell, and New York-Presbyterian, noted that while they are difficult to create, such agreements, when done right, are effective.

James Weinstein of the Dartmouth–Hitchcock Medical Center said that the entire health care system needs to be reinvented because it is unsustainable as is. He advised PCORI not to be “too precious,” and he admonished the industry to work more collaboratively to meet the challenges that it faces.

Grossman asked whether PCORI should expand its agenda to go beyond the health care delivery system. Selby responded that PCORI’s mandate as defined in its authorizing legislation is clearly focused on comparative CER, but it has become clear that clinical care cannot fail to take account of the socioeconomic circumstances, and as a result, PCORI’s portfolio is already extending into areas of population health.

Gallin said that he has a concern about the rapid recruitment of patients to participate in studies. What he wants, he said, is a smart database of patient populations that researchers could tap into to identify a patient

not only by diagnosis but also by phenotype and genotype in order to generate with precision a cohort of patients who would be available because of some spadework that had been done in advance. Selby responded that developing the ability to identify patients eligible for specific trials using the EHR and other data is exactly what PCORnet is supposed to enable. The challenge that he sees is identifying those trials that have the biggest potential impact so that they can be prioritized and not overwhelm individual health care systems. “Once a system is convinced that the right trials are getting done,” he said, “then I think the notion of building infrastructure and talking to the patient population about the fact that a trial or trials are under way becomes more embraceable.” Lucila Ohno-Machado from the University of California, San Diego, agreed with the idea of a master list of patients, but she suggested that it include information about a patient’s contact preferences. The technology for creating such a list exists, she said, and what is needed now is an effort to educate patients on the importance of research, both to themselves and for the health care system.

Tobin said that he believes there is promise in PCORnet’s creating code that smaller health systems could use to extract information from their EHRs that could be turned into meaningful analyses and be benchmarked against others. “I think this is going to be a by-product of all of the studies conducted under PCORnet,” he said. Tobin noted the importance of capturing and disseminating such software applications in a way that can be downloaded and implemented at low cost in practices where research may be less well integrated.

Lipstein had two suggestions for PCORI that came from his perspective as both a CEO and a member of PCORI’s board. The first suggestion was to identify PCORI-funded investigators who are building a career doing CER and track them longitudinally to show that they can get promoted and that they can receive awards and recognition in the academic community. “The idea is to begin to profile individuals and celebrate their success across the academic community,” Lipstein said. The second suggestion was for PCORI and the American Association of Medical Colleges to work together to teach methods of patient engagement that get patients to participate actively in improving their own health and becoming involved in research. He said that this type of effort would have the potential of addressing the fact that investigators who submitted wonderful grants to AHRQ or NIH have not been scoring well with PCORI because they do not understand patient engagement.

Robert Kaplan from AHRQ asked Selby if PCORI is thinking about how to spread the message about the value of research and learning to the majority of institutions that are not part of the academic medical center community. Selby responded that there are community-based delivery systems that do engage in research and that do place value on efforts to

develop a learning health system. PCORI’s goal is to use these forward-thinking systems as examples of how to improve care and reduce costs that can be shared with the rest of the health care system.

Implementation and Dissemination at Scale

DePinho stressed how important it is to implement widely the findings of research because the gap between new knowledge and standard of care in the community at the level of the nonspecialist is significant and growing because the torrent of new information makes it impossible to stay current. As an example, he said that when the FDA approves a new anticancer agent, it gets used immediately at major academic medical centers, but that on average it takes 7 years before that new information gets incorporated into routine care in community oncology practices. “So from my perspective,” he said, “we need to extend this envelope that we’re talking about, not just to what we’re doing at Columbia or MD Anderson and within health care systems, but how that does actually drive acquisition of knowledge and new ideas at the level of the patient and at the level of the primary care physician and the patient.”

Steele continued that train of thought by noting that as inefficient as the nation’s health care system is at taking a new drug and putting it into practice, it is even more difficult to change the business model for medical oncology as a whole, which is something that has to happen in order to reduce the total cost of care in a meaningful way. In his view, there are three ways to generalize and scale. One is to have the insurance companies associated with major health plans, such as Geisinger’s insurance company, work with other providers to try to reproduce the special relationship that exists when you have provider and insurer under the same organizational umbrella. The second approach is through consolidation, where organizations that engage in continuous learning and change take over those organizations that cannot figure out how to change and that do not have a sustainable business model. The third approach is to take as much intellectual property as possible out into the field as part of the organization’s business model, turning learning and innovation into what would essentially be a consulting business.

Hochman was not as pessimistic in his assessment of what smaller institutions are capable of accomplishing. One point that he stressed was that informed patients can be a route for getting new information into the hands of local physicians. “I think there’s going to be a time where a patient is going to know about the study that’s at MD Anderson before the doctor will,” he said. Corwin, responding to a question from Jerry Krishnan of the University of Illinois Hospital and Health Sciences System about how to set priorities for dissemination from large to small systems, said there is

an artificial divide between academic and nonacademic health systems. “I think that first and foremost we need to have established venues for communicating between these two systems,” Corwin said, “because I think that there are informal mechanisms but not formal mechanisms. I think there’s much more in common than what separates us, and I think it is important for us to understand what’s important at the academic center and the nonacademic health system, and this will not be successful unless we bring those large systems into the discussion in a tangible way.”

Wyatt Decker from the Mayo Clinic Arizona spoke of the need for speed in driving waste out of clinical practice and moving to a population health model. In his view, he said, most of this effort does not involve research but merely small tests of change and lean process redesign. In that respect, he said, he saw the workshops as playing a critical role in determining how to ensure that the research community remains relevant to the rapid evolution that is “necessary for the very survival of all of our own institutions.” As the CEO of an institution that is not a typical university, he said his view of research has been focused on meeting the unmet needs of patients. He acknowledged that what is typically translational or clinical research is taking place at far too slow of a pace, and he said he is encouraged by what PCORI is doing in this space.

Steve Allen of Nationwide Children’s Hospital in Columbus, Ohio, said that children need to be developing every day and that, as a result, the responsiveness of the system needs to be faster when dealing with a population that is made up predominantly of children so that interventions can be enacted more quickly. Grossman reiterated earlier comments about the challenges of executing and getting desired outcomes on a system-wide basis. “I’m really fascinated by those organizations that have managed to connect their ideas with outcomes,” Grossman said. “We’re very good, but we’re not where we need to be.”

Kirch stated that what concerns him most, as a former CEO and as someone who visits many institutions, is that there is no consensus on the pace of change that is needed. Honesty is needed in recognizing that it is a minority of institutions that are creating value through learning, he said. Most institutions are trying to drive cost out, but they are not linking that to value, and they are not viewing research as a useful tool. “So I think one of the challenges for PCORI and for the IOM is to not delude ourselves that the passionate and highly effective group you hear from in a meeting like this is by any means the majority,” he said. Another thing that he expressed concern about is the lack of support for this type of research from health systems’ governance structures that are still largely concerned with the next quarter’s profit margin.

Partnerships and Engagement

Holly Peay with the DuchenneConnect PPRN said that she supports the idea of engaging patients around the definition of value and sees that as an interesting way of framing patient engagement. Although it is important to consider that patient engagement taken to an extreme can be burdensome to both patient and health care system alike, she said, discussions about value can help better connect patients and health care systems to their communities.

Bray Patrick-Lake reinforced the idea that many speakers had stated during the workshop that the key to reforming the health care system so that it is both economically viable and patient-centered is to forge partnerships with patient groups. She noted that this is particularly important in making decisions such as those highlighted in the Choosing Wisely campaign, where the guidance is often that less care can be better. Eric Larson noted that this partnership approach is a unique and foundational principle for PCORI and that he did not think that PCORnet will be sustainable unless it is driven by questions that matter to both CEOs and patients.

Steele pointed out the disconnect that often exists between what patients understand and what physicians think patients understand. Geisinger is starting to address this gap, he said, by opening up progress notes to patients so that patients can see what their physicians or nurse practitioners have said about them. He sees this as an example of the beginnings of a fundamentally different relationship between the person who is getting care and the person who is delivering care. Although it is a small step, he sees it as a step in the right direction.

Sean Tunis from the Center for Medical Technology Policy suggested that the community might start encouraging patients to become more insistent that health care systems engage in learning activities. He also proposed that the National Committee for Quality Assurance could develop a certification for a learning system that meets certain standards.

Michael Dinneen of the U.S. Department of Defense said that the Department has had to redesign its health system over the past 12 years in order to address the casualties from the Iraq and Afghanistan wars and that part of this redesign involved creating the Defense Health Agency, which has the specific goals of serving as an integrator in a system of care and to incorporate advanced analytics, measurement, improvement, and coordination across the Department’s health system. He said that this effort has been informed by the experiences of Geisinger and Kaiser Permanente and that the Department’s financial resources may allow it to experiment in ways that other systems may not have the luxury of doing and therefore contribute to the overall effort.

DePinho said that MD Anderson is working with AT&T on issues involving connectivity between the health system and the consumer with

regard to things such as monitoring devices and interoperability at the level of the consumer. It is also working with IBM and its cognitive computing platform to manage the personal welfare of individual patients. As an example, he said that work is now ongoing to determine how many colonoscopies a person will need over the life course, given family history, genomics, diet, and other lifestyle factors. He said that he considers this effort as something that addresses personalized wellness. “What is exciting to us now is that we actually have all the component parts,” he said. DePinho added that the goal is to take the four or five factors that would have a profound impact on the overall health and well-being of each person and tailor the system’s monitoring and response capabilities based on those specific factors. He predicts this would have a profound impact on the health care system and health care economics. MD Anderson is currently piloting such a system across its network, including in areas in which there is no academic medical center.

John Warner from the UT Southwestern Medical Center summarized his views with two points. First, he said, achieving the necessary pace of change will require that hospitals invest in patient-centered outcomes research, which in turn will require systems to collaborate and share more effectively. Doing research collaboratively, he said, would make this a sustainable investment for many health care systems. The second point he made is that cost data need to be incorporated into the clinical data networks.

Jonathan Silverstein from NorthShore University HealthSystem informed the workshop that Epic, the EHR vendor, is creating a data research network that has a convergent direction with what is happening at PCORnet. He also said that 61 organizations have signed on to a set of principles regarding moving toward continuous learning in health which came from a series of meetings involving ONC and the IOM. He said he has seen that organizations that do not have a well-established research infrastructure want to participate in these efforts and benefit from the activities that are going on at an advanced level at other institutions. These smaller health systems want to know how to contribute their data and use the software being developed to extract information and model activities in their own systems.

The Health Care Industry Landscape

Darrell Kirch of the Association of American Medical Colleges commented that the current wave of consolidation among health systems for business reasons may not be leaving the time or energy to think about constructing systems in terms of what would create the biggest learning opportunities. Corwin agreed with that assessment, but he added that, with

consolidation, access to larger patient populations under the same organizational umbrella creates an opportunity to do more population health research and then share data across a large swath of the health care system. He also said that he thinks that hospital systems still give more weight to advancing the public good than to consolidation for consolidation’s sake, even given the economics of health care today.

Hochman agreed with Corwin and said that Providence Health is spending much more time thinking about what the architecture of the health care system should look like beyond the walls of the hospital. For example, Providence is creating a new division of population health that will coordinate learning and knowledge transfer across the entire system.

Steele said that Geisinger’s consolidation activities are being driven by a desire to move into areas in which its patients live so that not all care has to be administered in the same place that training and research occur. Geisinger is also being driven, Steele said, by the conviction that its model of being both a payer and provider creates a value proposition that competitors either cannot match or are not interested in matching.

Workforce Issues

In response to a question about whether the right type of investigators are involved in developing a learning health system, Steele said that there needs to be more work done on how to use observational data in a much more systematic way. RCTs, even of the sort that uses cluster randomization, are not going to answer 80 percent of the questions that need to be answered, he said, and there needs to be more methodology research to make use of observational data. DePinho said that MD Anderson realized that it had gaps in its expertise when it came to making the best use of its data and that is why it turned to IBM, AT&T, Google, and PricewaterhouseCoopers. He noted that this is the first time that these companies have worked together because they also recognize that they did not have all of the necessary bandwidth, knowledge, and capabilities needed for this effort. Yet, despite all of this assembled expertise, putting together all of the pieces is difficult, and DePinho estimated that it will take about 24 months to create and test the system that is coming together through this combined effort.

Concerning the barriers that he sees, Decker said that a lack of adequate analytics makes it difficult to get the clinical data needed at speed to transform practice and conduct research. He added that most of the research community that is involved in generating new knowledge to transform the health care system does not have the skill sets necessary to enable rapid transformation. He is attacking this last obstacle by hiring health economists and experts in big data, as opposed to bench-trained researchers, and he is providing discretionary funds to Mayo Clinic staff who are interested

in redesigning practice. “We’re really asking our scientific community to step up and help us grow,” he said, “and if they can do that, we may be able to help fund them or provide bridge funding. If they can’t, then we say, Your work may be important, but you’re going to have to look for other funding sources.” Another aspect of getting research aligned with transformation is overcoming the cultural issues surrounding academic reward and promotion in light of the kind of research that the health system expects from the research community. Responding to these comments, Selby said that he was thinking about a program that would take physicians into a residency program that would simultaneously provide training in the clinic and training toward an M.B.A. or master’s degree in public health. He added that he believes the CDRNs are locations for analytic expertise from which the community at large can draw support.

PCORI’s Sarah Greene commented on the fact that the nurses and information technology staff who will have to be active in embedding research into care do not have research support as one of their performance goals, and they do not have training in research. Kaplan responded that AHRQ is concerned about these types of workforce issues. Kirch commented that he does not think there is that much of a problem with regard to systems that recognize, reward, and promote researchers who are doing applied scholarship or the scholarship of care delivery. Most institutions have loosened their promotion systems to create tracks that will allow for this, he said. The bigger problem, he said, is the lack of people trained to do this kind of research. He said he believes that there is an opportunity to use what he called blended learning modalities and online learning to create learning communities that can produce a cadre of people who have at least some expertise in the science of care delivery.

Lisa Harris from Eskenazi Health, a safety net health care system serving the inner city of Indianapolis, noted that her system has been building an EHR for more than 40 years and has been using these data to improve almost every aspect of quality, safety, effectiveness, and efficiency of care. “We are thriving as a safety net health care system because of basically being a learning health care system over four decades,” she said, adding that her system is going to engage in an effort to use its experience to teach physicians nationwide how to use an EHR effectively. The challenge that she is facing as the CEO of Eskenazi Health is that 10 years ago the health system made the decision to take its homegrown EHR and transition to a vended system, but now that system is being sunsetted by the vendor. “My question is, Do we believe there is a collective opportunity to influence vendors to maybe take more of, if not a patient-centered approach, at least a health system–centered approach that I think would allow us to continue the things that we’re talking about collectively?”

This page intentionally left blank.