KEY SPEAKER POINTS
- Creating the space for learning and the will to change stems from understanding what is important to those affected and from the willingness to honor and integrate their perspectives. According to Sarah Greene, this willingness to integrate a variety of viewpoints is also the cornerstone of the continuously learning health system.
- “What we want to do is fail forward fast,” Thomas Graf said. “The concept is if you are not failing, you are not doing enough.” Graf added that what is important is to learn something from failure, to fail forward, and then keep iterating to the best design possible.
- To be useful, knowledge must be accessible to people who need it, it must be relevant to a specific market, and it must be of high quality and available at a reasonable cost, Trent Haywood said. He added that the health care system needs to consider how to structure itself in a way that does not lead to knowledge hoarding.
- Haywood said that there is no good model yet that addresses how systems get compensated for making investments to generate knowledge. “If people are making investments, and then we want people somehow to be altruistic and forgo their return
on this investment for the greater good, we need to some type of model that rewards that sacrifice that they are making,” he said.
- John Steiner said that the value proposition of integrating research with operational and clinical care is much broader than just what can be learned. Tools developed for research that eventually become part of operations also contribute to the value equation.
The goals of this session were to discuss a vision for a continuously learning health care system, to begin to articulate the value proposition for such a system, and to contemplate what it will take to create such a system. Session moderator Sarah Greene, senior program officer at PCORI, began this discussion with a proposal for how to create a value proposition for a continuously learning health care system. Three panelists—Thomas Graf, chief medical officer for population health and longitudinal care service lines at the Geisinger Health System; Rita Redberg, professor of medicine and director of women’s cardiovascular services at the University of California, San Francisco (UCSF); and Trent Haywood, chief medical officer for the Blue Cross Blue Shield Association—responded to Greene’s presentation with case examples to illustrate the opportunities to make a learning health care system the norm.
Greene began her presentation by pointing out that creating the space for learning and the will to change stems from understanding what is important to all who are affected by that change and from the willingness to honor and integrate the different perspectives of all those who are affected by that change. This willingness to integrate a variety of viewpoints is also the cornerstone of the continuously learning health system. Greene added that those health care systems that have demonstrated a great deal of success in creating a continuously learning health system have individuals in leadership roles who have a foot in both the research and system administration worlds and who can act as a bridge between different perspectives.
Referring to the “virtuous cycle” of a learning health care system (see Figure 3-1), Greene said that this cycle is based on the bidirectional flow of information between practice and research, and she characterized this process as “messy.” Proponents of the learning health care system, she said, need to be able to convince their colleagues that this virtuous but messy
FIGURE 3-1 The virtuous cycle of a learning health system.
SOURCE: Reprinted with permission from the Annals of Internal Medicine.
process of continuous learning is the model that can help achieve the triple aim of better care, lower costs, and improved health and to address other goals that are important to health system leaders.
The important elements of a value proposition, Greene explained, are focused on the needs of the customer; they include a promise of value to be delivered and a belief by the customer that value will be experienced, according to factors that are important to the customer. Putting the customer at the center of the value proposition means that a deep knowledge of the potential and current customer base is invaluable. “I would offer to you that we are not yet completely clear on who all of the customers of a learning health system are and what it is they want from this health system,” Greene said. This uncertainty may make it challenging to articulate a value proposition, she told the workshop attendees, but it would be a win for everybody if their work can increase clarity for decision makers and reduce the decision makers’ anxiety about the changes that are buffeting health care systems.
Greene then laid out a set of four questions to consider when creating a roadmap of customer and stakeholder, needs:
- What do stakeholders value intrinsically?
- What will they need to invest to create a learning system?
- What will they get in return for their investment?
- What are the potential risks and challenges in adopting an organization orientation to a learning health system?
To think about these questions, Greene offered a grid mapping these four questions against the needs of eight different stakeholders: employers/purchasers, health insurers/payers, health care system leaders, providers, researchers, patients, policy/government officials, and industry (see Figure 3-2). Employers, for example, are concerned about total cost of care and productive employees, whereas providers care about professional autonomy and their relationship with their patients. The patient, in turn, values a relationship with the provider and being able to live a life without the burden of illness. In short, defining the value proposition may require tailoring it to the specific stakeholders.
One question that Greene said she hoped the workshop would address was how to go from a compelling value proposition to conditions that support learning. “If we do get to a compelling value proposition out of the course of this meeting and the next meeting in June,” she asked, “how are we going to be sure that we can go home and have the conditions that support learning? It is not automatic. There need to be some sustaining features.” Some of those sustaining features can be found in three organization conditions that support learning: time allocated to exploration, discovery, and learning; a physical and social environment that enables one to be a “student”; and core values that appreciate learning in its own right and encourage curiosity, knowledge, and discovery. She also listed individual attributes that support learning, including humility, curiosity, self-awareness, a tolerance for ambiguity, openness, and vulnerability. This
FIGURE 3-2 An example of mapping stakeholder needs to create a value proposition.
SOURCE: Reprinted with permission from Sarah Greene.
last feature is perhaps the most difficult to create because so much of the ethos of the health care profession involves knowing exactly what needs to be done. “Learning requires vulnerability,” she said, but in health care “we don’t ever want to look weak in front of our colleagues.” She added that her research at Group Health proved the value of being vulnerable and being willing to fail in an environment that supports and is primed to learn from failure.
Greene also said that it is important to think about the many external market conditions that are putting pressure on health system leaders. These external factors include the Affordable Care Act, the increasing importance of big data and predictive analytics, the data release requirements of the Centers for Medicare & Medicaid Services (CMS), meaningful use and other new regulations, and payment reform. Meanwhile, efforts such as the Data for Good campaign being conducted by PatientsLikeMe are aimed at helping patients understand the value proposition for research. Although this may be the beginning of an inflection point for patient engagement in research, the receptivity to research demonstrated by health systems and clinicians may be affected by these other external variables.
In the end, Greene said, watchful waiting is not an option, and she ended her comments with a quote from Paul Keckley, the former executive director of Deloitte Health Solutions: “This environment rewards leaders and their organizations not paralyzed by its uncertainties, uninformed about its fast changing, nor fearful of the spotlight. Leaders . . . face unparalleled challenges and opportunities, but with them, pressures. Monitor closely the pulse of the industry . . . the stakes have never been higher.” Greene predicted that PCORnet is going to trigger such a change and said, “What I have seen already is that PCORnet is creating this climate of trust and humility from the outset and that it is purposely uniting the perspectives of CEOs, researchers, and patients to develop a compelling value proposition.”
To provide context for his comments, Thomas Graf began by explaining that the Geisinger Health System is a 1,000-physician medical group that owns 6 hospitals and a 400,000-member health plan. “The culture of physicians that drives Geisinger is really integral to a number of the elements of success and lays the groundwork for that success,” he said. He also noted that Geisinger is not a closed system, as only about 40 percent of the patients who see Geisinger physicians or who use Geisinger hospitals are members of the health plan, and only about 40 percent of the health plan members see Geisinger physicians or use Geisinger facilities. What these numbers mean in practice, he said, is that Geisinger has a special environment that allows it to try new approaches in the 40 percent of pa-
tients that the system both insures and treats. “We can go fast and take risks that other folks really cannot take, at least on the financial side,” Graf said. “But then we have an immediate scaling opportunity to take something that works to the other 60 percent of our patients and the other 60 percent of the providers that care for the members of the health plan.”
According to Graf, Geisinger has spent the past 10 to 15 years developing a culture of quality and innovation. “It began with creating the right infrastructure, the right space, and the right thought process, but also the demand for using quality as the lever to move physicians and nurses in a way that they had not been able to do before,” he said. Creating the right environment also involved creating positions for a chief innovation officer and a chief transformation officer to speed innovation and transformation. The proven models that Geisinger has become known for resulted from a culture of innovation and quality that diffused through the system’s clinical service lines, its community medicine department, and its frontline health professionals. “That is important,” Graf said, “because if we want to be successful and we want to maintain that value proposition over the long term, it has to be tied to things that intrinsically create value for the folks who are delivering that care.”
Geisinger’s ProvenHealth Navigator, its medical home transformation program, is one example of its success in creating a continuously learning health system. This program, which has now been put in place for 100 percent of the system’s patients and clinical practices, started as a series of pilots that were encouraged to fail. “I am not sure if failing often is necessarily the goal,” Graf said. “What we want to do is fail forward fast. The concept is if you are not failing, you are not doing enough.” What is important, he said, is to learn something from failure, to fail forward, and then keep iterating to the best design possible. In this case, Geisinger deployed its medical home pilots in 2006 and then rolled out the final system across 45 Geisinger sites and 45 sites that contract with its health plan but are not owned by the Geisinger system. The result was a dramatic improvement in quality across all the sites, including a 500 percent improvement in comprehensive diabetes care over a 5-year period and a 350 percent improvement in comprehensive preventive care for adults and children over a 3-year period (Maeng et al., 2012a) as well as a 7.5 percent reduction in costs over the first 3 years (Maeng et al., 2012b). Graf added that the results improved with the length of exposure to the innovation.
The bottom line, Graf said, “is that we have created a compelling model that creates better care for patients, which is a value proposition that we need to adhere to, and easier care for medical professionals.” He added that his one complaint about the triple aim is that it omits any mention of health professionals. “If we reduce total cost of care, we improve quality, and we improve patient experience, but we do it on the backs of the health
care professionals, it is not going to be sustainable,” Graf said. Instead, the goal should be to achieve what Geisinger calls the “triple aim plus,” which also includes improving the experience of medical professionals.
Achieving the triple aim plus requires taking an unreliable chaotic practice and reducing the variability of care by eliminating the 30 to 40 percent of what patients experience that adds no value to the patients. Automation plays a role, and so does better delegation across all members of the health care team, including doctors, nurses, front desk staff, and even patients. “If we can check ourselves into am airport post-9/11,” Graf said, “we can get patients engaged in their care, and guess what? They want to be participants.”
The final component for success is building evaluation into the organization’s culture so that continuous learning can take place. “That is what allowed both the rapid expansion and the ability to scale and generalize the program to organizations outside of Geisinger,” Graf said in closing. “That culture of quality embedded in daily life, providing value to patients that is patient-centered and that patients truly value, I think is what has allowed us to succeed.”
Rita Redberg began her comments by discussing the “Less Is More” feature that she launched with Deborah Grady at JAMA Internal Medicine (Grady and Redberg, 2010). The two editors were prompted to start “Less Is More” by the release in 2009 of the U.S. Preventive Services Task Force recommendation on mammography that suggested that there were more harms than benefits associated with mammograms for women ages 40 to 50. This report received a negative reception that Redberg attributed in part on the fact that the report did not emphasize the harms message and people assumed the recommendations were motivated by costs. “I am not sure that patients talk about or care about cost very much in our current system, but certainly, everyone cares about harms,” she said. “If you understood that you were going to have a much higher chance of being harmed than benefited from a test of any kind, I think you would think differently certainly about the test.”
The result was the launch of the “Less Is More” collection that highlights areas of health care that have no known benefit and potential harms, of which Redberg said there are quite a few. “If there is no benefit, we feel that a harm, no matter how small, is still going to tip the equation,” she said. She added that discussing harms, as well as whether there is a benefit or not associated with a particular intervention, is one approach to improve people’s ability to choose wisely, and she recommended that the importance of discussing with patients risks in the context of benefits be emphasized
in medical school and residency training. “Right now, we have a culture that embraces new things but without considering if it is an improvement and if the benefits outweigh the risks,” she said, adding that the medical profession needs to do a better job of informing itself about risks and then discussing those risks with patients.
Journals and continuing medical education are two avenues for helping providers learn to choose wisely. Quality measures are another, Redberg said, but she cautioned that using too many quality measures can overload physicians and interfere with their ability to care for patients. She said she believes that the use of “report cards” that provide feedback data on the usage of common testing has the potential to help physicians choose wisely without adding significant burden. Switzerland, she noted, now issues report cards that show doctors how they compare with other physicians on a variety of measures. Swiss officials, she said, believe this is an effective approach to improve care.
The American Board of Internal Medicine (ABIM) Foundation has launched the Choosing Wisely campaign which includes “top-five lists” for different medical specialties (Good Stewardship Working Group, 2011). The reason for creating these lists was to get physicians to start thinking about some of the things they do routinely that they should be doing less of going forward. The ABIM Foundation also partnered with Consumer Reports to get the word out to patients. The top-five list for internal medicine included such recommendations as “Use only generic statins when initiating lipid-lowering drug therapy.” The accompanying explanation said that statins are all equally effective in decreasing mortality, heart attacks, and strokes when the dose is titrated appropriately to reduce low-density lipoproteins (LDLs, or the “bad cholesterol”), and it suggested that switching to more expensive brand-name statins such as Lipitor (atorvastatin) or Crestor (rosuvastatin) should be done only if generic statins caused clinical reactions or did not achieve LDL cholesterol goals.
Redberg said that there are many reasons for overuse of medical care. One reason is a general enthusiasm for new technology, while another is defensive medicine. Misaligned incentives are an important factor as well, as is demand from patients. Redberg argued that many diagnostic tests are ordered just because it is easy to do. “You get more information,” she said, “but it is not always necessary, and even with what seems like the most innocent imaging test, things can go wrong.” What providers should do, she said, is to consider before ordering every test or treatment, no matter how mundane, what a test will tell them, what they would do differently given the results of a test, whether it will lead to a change in outcomes, and if the information from that test is needed to make that change. She argued that instead of ordering a test for “reassurance,” physicians can reassure their patients amply by talking to them (Redberg et al., 2011).
In her closing remarks, Redberg briefly described the Caring Wisely initiative being carried out by the recently created Center for Healthcare Value at UCSF. The premise for this initiative is that academic health centers have expertise and a tradition of innovation while at the same time having the credibility and integrity to participate in shaping controversial policies. Under the initiative, the center’s goal is to work with stakeholders at the state and national level to create a health care system that fosters high-value health care. The initiative, backed by departmental and hospital funds, tests staff-generated ideas that have the potential to provide better care while saving money. For example, when a literature review of the evidence suggested that nebulizers were not helping patients when used beyond 24 hours, pictures went up all over the UCSF hospital reminding physicians not to use nebulizers after 24 hours. “That has been a successful campaign,” Redberg said.
Knowledge is the critical component for creating a learning health care system, Trent Haywood said in his brief remarks, but to be useful knowledge must be accessible to people who need that knowledge, it must be relevant to a specific market, and it must be of high quality and available at reasonable cost. It is also important to consider whether particular knowledge is a public good and available to everyone or whether it is a commodity that can improve competitiveness. “I don’t think we are necessarily clear about when we are talking about this knowledge as being a public good and when we are talking about this knowledge as a private commodity where people are actually setting up infrastructure in place for financial incentives for people to actually compete in that particular setting,” he said. He added that the health care system as a whole needs to consider how to structure itself in a way that does not lead to knowledge hoarding.
Another question that needs thought is how individual health systems should invest in research whose purpose is to meet the triple aim. Haywood noted that this question has proven difficult to answer because “we know we are not satisfied with some of the research that we get.” Concerning value-based payments, Haywood said that his organization is supportive of using them to drive improvement across the board. The challenge is to disseminate the knowledge needed to determine value in the most effective manner. The Blue Cross Blue Shield Association’s Center for Clinical Practice spends time working with all of the 37 Blue Cross Blue Shield plans to identify leading practices with some measure of validation and to disseminate it quickly. Haywood concluded his comments by saying that this approach works in a closed system but that he is not sure this would be the case in an open competitive environment.
Greene, in her role as session moderator, took Haywood’s comments about knowledge and competitiveness and asked the panelists how they go about balancing the tension between transparency and competitiveness. Graf responded that it is important to understand the difference between transparency of knowledge and the ability to execute using knowledge. Geisinger, for example, generates little new groundbreaking research. Instead, it takes knowledge available in the literature and determines how to put it together and deliver it in a reliable way. Although Geisinger freely publishes the findings that result from this effort, it does expect to get paid if it helps other systems with implementation. “We are very free about the knowledge, but less so about the execution,” he said.
Transparency is important, Redberg said, because it enables others to replicate and validate findings and to avoid instances where incomplete data availability might lead to patients being harmed rather than helped. “We need to have more transparency and public availability so that people understand risks and benefits,” she said. “That can only be done in an open, data-transparent environment.” Haywood agreed with Redberg and commented that part of the tension reflects the fact that there is no good model yet that addresses how systems get compensated for making the necessary investments to generate knowledge. “If people are making individual investments, whether it be for their own individual institution or for other organizations, and then we want people somehow to be altruistic and forgo their investment for the greater good, then we need to some type of model that rewards that sacrifice that they are making,” he said.
Stephan Fihn of the Veterans Health Administration (VHA) asked Haywood if he had any ideas on how to value knowledge that does not have a market value as opposed to knowledge that directly improves the bottom line. An example would be knowledge that results in some process being improved but that does not meet its full potential in the eyes of stakeholders such as patients because it has not become widely disseminated. In that case, the knowledge only has value after it has been widely disseminated and incorporated into medical practice. Haywood suggested that reward does not have to be strictly monetary and that it could be in the form of social capital.
Fihn’s question prompted Robin Wittenstein of the Penn State Hershey Health System to ask the panelists if they had any ideas on how to develop knowledge markets so that learning can be disseminated more rapidly to the eclectic mix of people who have to be involved in researching, understanding, and then implementing new ideas. Redberg said that the use of journals has worked well as a way of disseminating knowledge, but Haywood commented that physicians do not have spare time in the current health
system model to go find information in the literature that could improve their practices. What has to happen, he said, is for knowledge transfer to become part of routine practice. Graf agreed with Haywood and said it is important not to place the responsibility for information gathering and dissemination on health care practitioners but rather to have the health care system do that part and instead reward practitioners for participating in the system. “The system is what reliably produces results,” Graf said. Geisinger found that feedback on performance is an effective reward. “It is important to make it as easy as possible for every person in the health care system to get the right answer,” he said.
Graf also said that enabling providers to get the right answer for 90 percent of their patients enables them to be more creative and customize the approach to take with the 10 percent of patients who need customized care. “Make the 90 percent automatic so that the doctor can focus on the remaining 10 percent,” he said.
Jerry Krishnan from the University of Illinois Hospital and Health Sciences System suggested that the value proposition for patients to participate in research still needs to be made clearer. He and his colleagues have been engaging patient groups and have found that they want to be involved in research once they understand what value it has for them. Graf noted in response that patients certainly value organizations that participate in research, in large part because of the reputation it gives that particular health care system. Whether a patient participates in research ultimately comes down to a matter of trust between the patient and the physician, and building that connection on an ongoing basis is critical. Krishnan also commented that as systems become known for their research and improved quality, it should give them a competitive advantage; thus an effort to understand how to place a value on that advantage could help support the infrastructure needed to generate new knowledge, he said.
Lorraine Johnson, a patient representative for PCORI, said that the question of how patients value research is context dependent and depends on balancing risk and reward for each patient. It is important, she said, to make sure that patients see and understand the benefits of the research in which they participate. Greene, playing devil’s advocate, wondered if hearing the term “learning health care system” might scare patients away. “I don’t want a health system that learns, I want one that knows,” she said. “I think we need to be thoughtful about this.”
David Grossman suggested that in the case of not-for-profit health systems, generating and disseminating knowledge could be seen as delivering the community benefit that such hospital must satisfy to keep their 501(c)(3) status. The same could be said, he added, for government health systems such as the VHA. In addition, being first to market in terms of publishing results could be an advantage for not-for-profit systems. Haywood
agreed with these comments but said that it will be necessary to develop some kind of metric to account for this type of activity.
John Steiner from Kaiser Permanente Colorado argued that the process of research itself, not just the improvement in practice, is a huge part of the value proposition. “In particular, I think one way in which research is helpful is in framing what I think of as well-formed questions,” he said. “When we work with operational leaders, oftentimes they have a general idea of what they want to learn. Our job is to help them get more precise about what exactly it is they want to learn.” The tools developed for research that eventually become part of operations also contribute to the value equation, he said, citing interactive voice response as a one-time research tool that is now largely an operational tool. “My general point is that the value proposition of integrating research with operational and clinical care is much broader than just what we learn,” he said. “How we learn brings a lot to the table.”
Addressing the issue of widespread application of new ideas, Lewis Sandy from UnitedHealth Group asked if there is any leverage in a network focused on evaluating complicated system interventions, with the patient-centered medical home (PCMH) as an example. In particular, he noted that although Geisinger’s PCMH intervention produced better results over time, Group Health’s PCMH did not produce better results until it was retooled, and a recently published study of PCMHs in Pennsylvania found only modest improvements in quality and no reduction in utilization over a 3-year period (Friedberg et al., 2014). These varied results, Sandy said, raised the question of whether every learning health system has to conduct its own evaluation of the PCMH concept. “If so, there is not a lot of leverage in terms of installing and deploying a better model of primary care,” he said.
Redberg replied that interventions such as the Randomized Evaluation of Decolonization versus Universal Clearance to Eliminate MRSA (REDUCE MRSA) trial that Susan Huang discussed in the workshop’s opening session are straightforward enough that they should be applicable across health care systems, but for a PCMH it is likely that each health system will have to develop and test its own unique implementation that takes into account local factors, patient mix, institutional culture, and other factors. “I think there can still be a learning experience from sharing,” she said, “but you do need more granular data for complicated interventions such as PCMH.” Graf agreed, but wondered if it would be possible to look across PCMH models and draw some conclusions about core elements that are critical for successful interventions. He noted, too, that large payers such as UnitedHealth have the opportunity to conduct such studies with large sample sizes. Haywood suggested that by looking at a granular level across systems, it should be possible to identify core elements for success.
In response to a final question from a workshop participant about whether grants that fund research should include an obligation for applying the knowledge gained from that research, Graf said that this was in fact happening, and he cited the recent CMS innovation awards that tie knowledge generation to implementation. “I think this concept of tying generation and dissemination is important,” he said. “The ability to do that will be the challenge.” Greene added that this would be a good goal but that the incentives for academia would have to change substantially. She put the challenge this way: “When is the Group Health faculty going to be incentivized and given tenure for getting the knowledge into practice versus getting the journal article published? I think we are a little ways away from that.”
Summarizing the discussion, Greene listed the many incentives that she heard during the session that could feed into the value proposition. These included social capital, reputation management, ease and speed of information gathering, making it simple and easy, monetary savings, better care, and lower anxiety. She noted that synergy itself is an incentive. “Having that collaborative capability in an organization creates the joy-in-work phenomenon that I think we would all value,” Greene said. She also noted that the research tools and analytical methods developed for this type of research can themselves benefit an organization’s operations. For patients, she said, aspects of altruism and knowing that they are “paying it forward” by participating in research that will benefit future patients could act as an incentive as well. She commented that one incentive she did not yet hear about was that research will generate better data that can, in turn, support quality improvement and other aspects of improving care.
This page intentionally left blank.