The first panel of the workshop, moderated by Wynne Norton, an assistant professor in the Department of Health Behavior at the School of Public Health of the University of Alabama at Birmingham, presented examples of approaches to spread and scale from the health sector. M. Rashad Massoud, the director of the U.S. Agency for International Development (USAID) Applying Science to Strengthen and Improve Systems (ASSIST) Project and senior vice president of the Quality and Performance Institute at University Research Co., LLC, discussed the ASSIST Project. Steven Kelder, a co-director of the Coordinated Approach to Child Health (CATCH) and professor of epidemiology at the Michael & Susan Dell Center for Healthy Living of the University of Texas School of Public Health discussed CATCH, which is focused on preventing obesity and promoting healthier lifestyles. Darshak Sanghavi, the director of the population and preventive health models group at the Center for Medicare & Medicaid Innovation (CMMI) at the Centers for Medicare & Medicaid Services (CMS), described two population and preventive health models that CMMI is exploring. (Brief background information on the case examples, including how speakers understand spread and scale in the context of their own work, was submitted by the panelists prior to the workshop and is available in Appendix C.)
The USAID ASSIST Project is part of the U.S. foreign assistance program aimed at improving health at scale, Massoud said. ASSIST is the fifth in a series of projects and has worked in 28 countries to date to strengthen their capacity and improve care. The USAID ASSIST Project is working with multiple partners, including more than 230 governments and implementing partners, more than 4,400 facilities, more than 900 communities, and more than 2,500 quality improvement teams reaching more than 96 million people in the areas served. The project is working to address global health priorities, focusing on technical areas such as HIV; tuberculosis; maternal, newborn, and child health; community health; health workforce development; noncommunicable diseases; and others, depending on the geographic area. Massoud highlighted the work that ASSIST is doing with the Ministry of Health and Family Welfare in India as an example. ASSIST works in 263 facilities, with a quality improvement team in each facility, and makes 12,000 to 14,000 deliveries per month, 30 percent of the total delivers in 27 high-priority districts (USAID, 2014).
There is no single best approach to scaling up, Massoud said, and a variety of methods have been used (Massoud et al., 2006, 2010). A mainstay approach is the collaborative improvement methodology developed by the Institute for Healthcare Improvement. ASSIST also uses extension agents heavily, deploying staff to the facilities. Another approach is wave sequence methodology, in which champions act as spread agents from the starting point to the remainder of the system. The majority of the scale up efforts, however, are hybrid models, taking an adaptive approach and catering to the particular setting at that particular time. In response to a question, Massoud said that the model used is often chosen by the different countries based on their situation, with guidance from ASSIST.
Massoud elaborated on the wave sequence methodology, which he said is an approach used when not everyone can be reached all at once (Massoud and Mensah-Abrampah, 2014). Starting with the full geographic area that ASSIST wants to cover, the team identifies a main center or central hub that has some sort of distribution network throughout the region (e.g., district, province). Each of these regions will probably have its own centers, and there will be many facilities in the regions where care is being delivered. The approach then takes a slice or a wedge of the population in each of the regions of interest, captures all of the different levels of care
in that system as part of the demonstration project. Once they have successful improvements that are determined to be scalable they are spread from that slice to the remainder of the slices of the system by the initial developers and champions of the improvements, with the support from the ASSIST Project.
CATCH is focused on preventing childhood obesity and promoting healthier lifestyles among children. The approach is based on the ASCD (formerly the Association for Supervision and Curriculum Development) Whole Child Initiative1 and the Whole School, Whole Community, Whole Child model, developed by ASCD and the U.S. Centers for Disease Control and Prevention (CDC)2 (see Figure 3-1). Healthy children attend school more frequently, Kelder noted, and, furthermore, research supports a relationship between physical fitness and academic achievement.
The key elements of the CATCH School Health Model include physical education (including professional development for teachers), nutrition services, classroom education, family education, preschool and afterschool programs, and physical activity breaks. Kelder described several challenges, such as the fact that child nutrition services are often under contract and it can be difficult to modify the food offerings. In the classroom, there is often not enough time in the day for health education. In addition, schools are not held accountable for health education. He added that physical education class provides most elementary and middle school children with about 20 minutes of moderate to vigorous physical activity every other day (about 40 minutes total per week), and there is almost no physical activity in high school unless students participate in athletics.
From the educational perspective, Kelder said, the desired student outcomes of the CATCH program are academic progress, achievement, and success; positive social and emotional development; high attendance; and parent and community support. The desired outcomes for staff include providing engaging and rigorous instruction, a high commitment to improvement, positive morale, and high attendance. Kelder emphasized that these desired outcomes were developed from meetings with school superintendent groups and that if a program is to operate within the value system of a school, it has to be oriented toward both what the students and staff need. Kelder highlighted the importance of the diffusion of innovation (Rogers, 1995) and the value of identifying program champions for taking programs to scale.
FIGURE 3-1 ASCD and CDC Whole School, Whole Community, Whole Child model.
SOURCE: ASCD, 2014.
The CATCH Global Foundation
The CATCH Global Foundation, established in 2014, is a 501(c)(3) organization devoted to improving children’s health worldwide by developing, disseminating, and sustaining the CATCH platform in collaboration with researchers at the University of Texas School of Public Health, Kelder said. The foundation links underserved schools and communities to the resources necessary to create and sustain healthy change for future generations. In closing, Kelder noted the value of social media for outreach and dissemination in the face of limited resources.3
3 In Appendix C, Kelder elaborates that the CATCH program of promoting healthy eating and physical activity is spread and adapted for use in afterschool programs, YMCA, and
The total annual federal spending for Medicare and Medicaid is more than $700 billion. More than 54 million Americans receive services that are covered by Medicare, and 70 million receive services covered by Medicaid. Sanghavi said that scale in this context means that to treat the whole person we need to pay for the whole person, which means transitioning away from fee-for-service medicine to population-based payments. In this regard, CMS is exploring innovative payment and service delivery models (e.g., value-based payments). Currently, approximately 20 percent of all payments are value based, and Sanghavi said that a critical mass of payers implementing value-based payments is needed before most organizations will invest in programs and services that will lead to improvements in population and community-based health.
Community and population-based health interventions should be as inclusive and generalizable as possible, Sanghavi continued. The tendency is to focus a program on a segment of the population (e.g., an economic or a geographic segment). Part of scalability is having broad incentive structures so that all people will buy into and support the intervention.
Sanghavi described two broad population and preventive health models that his group at CMMI is exploring. Despite the evidence, it is difficult to make the case for prevention to payers, Sanghavi said. One approach is to focus on robust analytics to predict risk and then develop ways to pay for reductions in aggregate risk. The model that CMMI is exploring involves calculating individual risk and then incentivizing people to lower that risk (e.g., blood pressure, cholesterol levels, smoking). This is not a new idea, Sanghavi acknowledged, but doing it at scale is new, especially at the scale of CMS. This is a very different way for CMS to think about payment, he said.
The second broad model Sanghavi described is an accountable health community. From the payer perspective, investing in community health requires demonstrating that the innovation substantially improves quality or reduces costs. A three-track model is being explored for use on a national scale. In the first track, which is low touch and high volume, patients can be provided with a list of services that could help them. The medium-touch, medium-volume track provides the information about services and also a connection to a person whose job it is to follow up and make sure they connect with those services. The third track provides the
parks and recreation programs. The CATCH program serves as a good example of how strategies of spread, scale, and sustainability may become inseparable in practice.
information and personal follow up, and also invests in creating durable linkages among the people delivering the services.4
These models have internal controls built in so that the total cost of care over time can be assessed. In this way it can be possible to demonstrate the value of investing in preventive services and to pursue innovative financing strategies, whether at the state, national, or other levels, Sanghavi concluded.
Norton, the session’s moderator, asked panelists to comment further on several topics raised in the presentations, including partnerships, barriers, and the evaluation of impact, as well as their thoughts on moving forward with spread and scale.
Norton noted that in all of the examples there was a need for relationships and partnerships in bringing a practice or program to scale. Kelder said that CATCH started by developing local partnerships with the larger school districts through the diffusion-of-innovation model, finding those people who were interested in and passionate about the topical area for which CATCH had solutions. Later, CATCH partnered with the Texas State Department of Education to align the program with the state educational objectives and garner approval from the State Board of Education to allow any school in Texas to implement the program. CATCH also partnered with the Texas Department of Health to obtain funding for professional development for health education and also with the Department of Agriculture, which has responsibility for the food served at the schools. In summary, the researchers had to step out of the university, meet the elected officials, and find the champions throughout the state who were willing to accept the innovations that CATCH had to offer. Other districts around the state and the country then began calling for information about the program.
Massoud said that the initial conversation that ASSIST has with foreign governments is about which issues are most important to them. An outsider can make improvements on a small scale, but larger-scale sustainable change has to come from within, he said. ASSIST engages governments in a partnership, working with them to develop capability
4 In Appendix C, Sanghavi notes that CMMI uses evaluation, learn and diffusion strategies, and “public accountability of results of pilot programs” to support spread and scale strategies.
and infrastructure so that they will ultimately take over and lead when the program goes to scale. This type of partnership is one where the exit strategy of the ASSIST Project is part of the plan from day one.
CMS is a big payer, but it is still only one payer, Sanghavi said, and it is not enough to move the market if only CMS endorses a particular idea. A core challenge is how to get private payers to catch up to and join CMS in paying for innovation. He noted that truly innovative health care centers are struggling to find sustainable funding, falling into the gap where the private payers are not yet paying.
Panelists highlighted a variety of barriers that are encountered when going to scale and maintaining sustainability.
Sanghavi said that it is often the communities that already have durable, highly invested institutions that are the ones applying for CMS funding for innovative health care programs. There are very large areas of the country without innovative care solutions or a focus on community health. There is a divide between the haves and the have-nots in terms of the sophistication of the health systems. CMS tries to be cognizant of what it can offer to those communities and what it can do to allow them to participate in health care innovation.
Kelder added that after the recent recession and cuts in state budgets, many school personnel lost their jobs, personal development days were taken away, and training programs for school health specialists were cut. Researchers who are designing, developing, and evaluating programs and creating the evidence base are struggling to monetize these programs or to get them implemented. Faculty are not well versed in how to deal with intellectual property issues, he said. CATCH was able to find a commercial partner to print and market materials and supplies, but, he added, the marketplace can be both a facilitator and a barrier. Having a commercial partner made CATCH ineligible for funding because many institutions will not fund for-profit organizations.
Change in staff at the leadership level as well as in health care delivery institutions is a particular challenge, Massoud said, and new champions must be continually identified and developed. Another obstacle is the pervasive notion among leadership that all that is necessary is to replicate a successful pilot program over and over, in a linear fashion. Building on the experience and using good spread methods allows for scale at a much faster rate and much lower cost than simple linear replication.
Evaluation of Impact
Norton raised the issue of the evaluation of impact and population health outcomes. CATCH has done a number of studies to determine efficacy, Kelder said. For example, CATCH is in place in all of the middle schools in the city of Dallas. CATCH was able to use the existing fitness standard testing done at the schools (e.g., obesity rates and fitness levels) to show that schools that were higher implementers experienced a stronger effect than schools that were lower implementers. Kelder highlighted the value of finding existing public data sources when programs go to scale, because the collection of original data is often unfeasible. Staff at the school district level can usually provide information about the evaluation methods that they use to meet state standards. Kelder added that Texas—like many other states—requires schools to fill out a campus improvement plan annually. The campus improvement plans often do not have a health component. CATCH has written standards and disseminated them to the school and district administrations.
In evaluating impact, CMS generally focuses on the reduction in total cost or on in improvement in quality, Sanghavi said. A challenge is that an evaluation can suggest a correlation between an initiative and an outcome, but not causation. Conclusions drawn are often colored by the agenda of the evaluator. A community health advocate, for example, might suggest that costs went down because of the intervention. Another challenge is the design of the intervention, Sanghavi said. If the intervention is not designed to actually answer a question, it is highly unlikely that the data will be very persuasive. The gold standard in evaluation is to conduct studies in which the interventions are randomized, but that is very difficult to do in practice, he noted. He added that CMS is exploring a cardiovascular risk reduction model, and, if it is done, it will likely be a randomized study.
Massoud added that when conducting an evaluation of a scale up, one wishes to determine not just whether the desired result was achieved, but also how long it took and how much it cost relative to the demonstration project and also relative to other alternatives.
Norton asked the panelists to summarize approaches that others could use moving forward with spread and scale. In response, Kelder said that when working with schools, it is important to talk with the state agencies, especially the state board of education, to find out what they are already doing and if the proposed program elements will work with their current structure. Linking the program priorities with both federal and private philanthropic interests and missions is also helpful, as is engag-
ing professional organizations. Kelder reiterated the value of finding the innovators and the early adopters to serve as champions for the diffusion of the innovation.
Massoud suggested starting with the end in mind when developing interventions, defining what to achieve, and then designing how to get there. Start with something small and deliberate at the demonstration level. He also suggested having an upfront agreement with the leadership that the key staff for the demonstration project will be allowed to participate in taking the program to full scale (releasing them from current obligations as needed). He also reiterated the need for a deliberate transition scheme, so that the program will ultimately be handed over to the local leadership.
Sanghavi concurred and added that large health systems should try to engage the private-payer partners at a very early stage, making sure that they have input and are participating in the design of the intervention. Building on the prior discussion of evaluation, he also suggested developing an independent and well-thought-out evaluation strategy prior to the intervention.
During the open discussion that followed, George Isham of HealthPartners observed that each speaker had provided a very situationally dependent view of spread, scale, and impact. He asked whether the evidence base is complete and whether it is distinct or if each could learn from the other examples. Massoud responded that the evidence base is far from complete and may never be complete. For example, there is much to learn about the rate of spread or adoption. Kelder agreed and added that in his area, for example, there are programs that are known to work with middle school children that do not work with preschoolers. The problems are different, the solutions are different, and the personnel are different, he said. There are always new ways to improve programs and get outside of the silos, he said, noting that in his case, he needs to interact with pediatricians, school nurses, economists, and the state government. If the ultimate goal is to have healthier individuals living in healthier communities, Sanghavi said, then spread, scale, and impact are multilayered issues. CMS can look at one or two parts of that—for example, the payment incentives. What is needed overall is a rigorous system of professionalism, education, and community engagement.
Participants then commented on partnerships, funding, evaluation, and prevention as they relate to the spread and scale of population health. Finally, panelists offered their advice on priorities for the roundtable moving forward.
Partnerships and Shared Responsibility
One participant raised the issue of the “edges between systems,” or where the responsibilities of one program end and where those of other programs begin. Sanghavi said that a positive side of division is that people are often more invested in dealing with a problem if they feel as if they own it. He also raised the issue of controlling “leakage” when the walls between systems are removed. For example, if a payer tries to reduce emergency room visits by giving free air conditioners to patients who were repeat visitors for heat concerns or by giving all chronically homeless people housing, it would become difficult to draw the line for who should get free air conditioners or housing from that payer.
One path forward might be what Sanghavi referred to as “virtual braided funding.” If different organizations (e.g., health, social services, corrections, housing, and welfare) consider the trends for where they spend money, they can collectively fund shared interventions (e.g., investing in substance abuse treatment) and then determine if their individual costs were reduced over time. Sally Herndon of the North Carolina Division of Public Health shared another example of virtual braided funding. Together with the private sector, North Carolina has built support for smoke-free affordable housing by becoming the second state in the nation to provide tax credits for building new multi-unit housing that is smoke free.
Massoud agreed with the need to look at the bigger picture, and commented that the likelihood of achieving better health outcomes is even higher when interrelated efforts are combined. As an example, he noted that USAID’s work in caring for vulnerable children and families started with emergency relief for the orphans of the AIDS epidemic, and it now encompasses other health care, schooling, food and nutrition, economic household strengthening, and other elements.
Kelder said that CATCH tries to tailor its approaches to the local communities as much as possible, using their own value and belief systems as well as their prioritization of problems. He explained that he has a portfolio of projects, some created by him and some by other institutions, and he can make broad program recommendations to schools based on the problems they are interested in instead of being restricted to just the programs he has funding for.
Neal Kaufman of the University of California, Los Angeles, schools of medicine and public health, asked how to make partnerships between the private sector and universities more robust. Kelder responded by noting the importance of understanding intellectual property, especially licensing agreements and the payment of royalties for both nonprofit and for-profit institutions, when developing these relationships.
A question was raised about the role of accountable care organi-
zations (ACOs), and how to incentivize them to focus on population health. Sanghavi said that incentives are important, but that there are other elements to consider. One approach could be to capitate the payments in some way, creating a full-risk ACO. Another issue is attribution of patients—that is, assigning a provider in the ACO to be accountable for a patient’s overall care, both cost and quality, regardless of which providers deliver the care. Communication with patients in the network is also essential because many people do not even know that they are in an ACO or even what one is.
Pamela Russo of the Robert Wood Johnson Foundation asked the panel to comment on social impact bonds5 as a way to fund innovation. Sanghavi said that CMS is exploring the use of “pay for success.” One of the core issues is that social impact bonds are not a very attractive investment vehicle. Rather, they are more of a charitable venture. If the programs are great ideas that people are going to invest in, then social impact bonds may be helpful as bridge funding to buy time, but ultimately there has to be a rigorous political process in the background. Kelder agreed that a lot of this work is charitable, especially—because of state budget cuts—in his field of education. The private marketplace has not stepped in with the intellectual property because there is such a thin profit margin compared to, for example, drug treatment or many other treatments. It is important to find those charitable contributions and also to take the long route of asking the state agencies to do the right thing, he said.
Jeannette Noltenius of the National Latino Tobacco Control Network raised a concern about whether the evaluation of impact goes deep enough to see disparities. Will the right data exist to identify the impact and the cost of having large poor populations with multiple chronic diseases? Kelder agreed that more research is needed that demonstrates efficacy within certain groups. This can be a challenge for hospitals from a workforce perspective, but is important for prevention and screening to reach the populations who are at a disparate health risk. Sanghavi suggested that for large health systems and public health agencies, it may be more effective to find ways to improve care for the entire population, which
5 Social impact bonds are a “pay for success” funding model in which private investors fund public projects, receiving a return on investment only if the project successfully demonstrates improvement in social outcomes.
would help in eliminating disparities. While there may be approaches that could be used to target particular populations, the biggest yield could come from the public reporting of measurement and transparency, which would help ensure that all are receiving care.
Kaufman suggested that people care less about their lifespan and more about their “performance span,” that is, the years during which they have the ability to do the things they want to do. He suggested that there is a need to consider the risk of accumulating second or third chronic conditions. He added that money can be saved in the short term not only by preventing disease but its complications as well—for example, lowering the rate of sleep apnea in overweight and obese people, thereby reducing the need for and costs of continuous positive airway pressure machines. Kelder concurred, citing the problem of overweight children and the advent of bariatric or other surgical weight control techniques. It is better to prevent the child from becoming morbidly obese in the first place, he said.
Priorities for the Roundtable
Sanne Magnan of the Institute for Clinical Systems Improvement in Minnesota asked panelists to advise the roundtable on priorities, given the roundtable’s three basic goals: increasing life expectancy and other health outcomes, decreasing disparities in those outcomes, and decreasing health care expenditures and using the savings upstream. Kelder said that there are no quick and immediate solutions for decreasing disparities and improving outcomes and reducing health care expenditures. The work takes time. He suggested that increasing life expectancy and reducing health disparities should be priorities. Massoud suggested that a place to start would be looking at preventive interventions that will provide the most impact at the lowest cost. He also suggested targeting the people most affected in order to address the disparities issues. Breaking down the barriers among sectors is also essential for progress. Essentially, most people do not care what the average lifespan of Americans is; they care about their own lifespan. The first challenge then is how to make population health meaningful to the average person. Sustainability requires community buy-in. He added that cost effectiveness, quality-adjusted life years, and other such measures are strongly subject to bias. He suggested considering what is needed to achieve those three goals regardless of cost and then to talk about scalability.