In the final session of the workshop, keynote speaker Joe McCannon, a co-founder of the Billions Institute,1 shared his perspective on expanding population health, including advice on successfully getting from start to scale.
There are several prerequisites that must be in place before considering going to scale in any area, McCannon began. First, there must be promising prototypes or a promising evidence base that can be built upon. There are various examples of successful prototypes that offer some confidence that it will be possible to have an impact on population health at scale, he said, and some of them were discussed at this workshop. Second, there needs to be attention from influential leaders and stakeholders at national and local levels; many leaders in population health were in attendance at the workshop, he noted. Third, there needs to be a “conducive context.” By this, he meant that the implementation of the Affordable Care Act has stimulated a health environment that is conducive to change. Beyond government, there has also been a notable increase in venture capitalism and changes in patterns of investing by banks and universities. For example, he said, in 2014 digital health fund-
ing broke previous records, exceeding the total for 2013 in the first half of the year (Rock Health, 2014). The significant energy and attention in this area suggests a conducive environment for spread and scale, he said. He expressed confidence that there is a strong enough evidence base to begin and that it will be possible to continue to learn and refine the science going forward.
With the prerequisites in place, the question is how to seize the moment in population health. There are case examples from many different sectors that might be relevant to scaling impact in population health across the United States (e.g., infectious disease, public health, patient safety, corrections, homelessness, sex trafficking). Drawing on his work in and study of these sectors, McCannon focused his keynote remarks on the elements that take an initiative beyond typical to truly exceptional.
This type of change is very hard, McCannon acknowledged. He listed a variety of reasons why is it so difficult to take a sound initiative that has worked locally to a larger scale and to spread it effectively. There is a very crowded marketplace of ideas, he said, and the sheer volume of information and ideas is a barrier. Another barrier to change is what McCannon called “the myth of natural diffusion.” There is little evidence that simply putting something out there in the literature or the public domain will result in uptake because of its merit or intrinsic value. Other challenges that undermine change are conflicting values, inertia and the need to attend to business as usual, resignation and apathy, competition, and fear. Fear is the enemy of all change, McCannon said.
Typical Versus Exceptional Initiatives
McCannon presented ten attributes and behaviors of typical initiatives, and he contrasted those to the comparable attributes and behaviors of exceptional initiatives (see Table 6-1). Typical initiatives are not the result of bad intentions, he said, but more the result of the inertia that was noted as a barrier above. Exceptional initiatives stand out and have a very different feel or energy to them, he said.
Strategy Development Versus Starting
A typical initiative generally involves comprehensive strategy development, McCannon elaborated. It is a natural tendency when addressing complex problems to want to take time and consider all possible directions and outcomes in order to try to solve the problem. In contrast, exceptional initiatives have a bias toward starting—not despite complexity, but because of complexity, he said. McCannon referred to the work of Asupos
TABLE 6-1 Attributes of Typical Versus Exceptional Initiatives
|Comprehensive strategy development||Bias toward starting|
|Emphasis on consensus||Consensus kills|
|General goals for expansion||Explicit, time-bound aims|
|Design for success||Design for success and scale|
|Broad knowledge of audience||Detailed audience segmentation|
|One stimulant||Many stimulants|
|One teaching method||Many learning methods|
|Summative evaluation is the priority||Formative evaluation (daily data) is the priority|
|Management gives approval||Management removes barriers|
SOURCE: McCannon presentation, December 4, 2014.
and colleagues at the Aspen Institute,2 which suggests that the existence of complexity actually means that excessive strategy is wasteful—perhaps even absurd, McCannon added. Engaging with the world is the only way to know what will work and when for each context, he said. Modeling or network mapping can provide clues, but engaging with the environment is necessary. Complexity also means there is no silver bullet solution. One characteristic of initiatives that really succeed is a bias toward getting started, he said.
Another characteristic of typical initiatives is an emphasis on consensus and working to ensure that all stakeholders are in agreement. In reality, McCannon said, consensus is a very complex process, and he observed that the initiatives that succeed take the view that “consensus kills.” Consensus on aim is needed, but the process of trying to achieve perfect consensus, particularly on smaller decisions going forward, is actually damaging to progress.
Goals for Expansion
Another characteristic of initiatives that do not succeed is vague goals for expansion, McCannon said. Successful initiatives have explicit,
2 See http://www.aspeninstitute.org/sites/default/files/content/docs/pubs/Complexity_and_Community_Change.pdf (accessed February 20, 2015).
time-bound aims and concrete ideas about what the initiative seeks to accomplish. He quoted Donald Berwick of the Institute for Healthcare Improvement (IHI) who, in reference to the IHI 100,000 Lives Campaign,3 said “Some is not a number. Soon is not a time.” The campaign set out to accomplish a defined goal by a certain date. Explicit goals are determined by understanding what full scale looks like, McCannon said, and also with the understanding that full scale is not achieved in one move. For any phase of an expansion (moving from prototype to pilot to scale), a rate of expansion of five times to ten times is a reasonable expectation (what McCannon referred to as the “Rule of 5× to 10×”). For example, the 100,000 Homes Campaign started in about 20 cities, and the target for the expansion phase was 200 cities. A goal of 2,000 cities would have been unreasonable, he said. Another example is the Millennium Development Goals, which are eight very explicit goals to be achieved by 2015. There has been remarkable progress globally, particularly in certain regions of the world, on these goals, he added.
Design for Success and Scale
Initiatives tend to struggle when it comes to scaling because they design only for success, McCannon said. There is resource-heavy investment to ensure success at all costs, but this does not account for the need to reduce marginal costs and introduce economies of scale over time as the initiative expands. A better model is designing both for success and for scale from the outset, he said. Citing the work of Everett Rogers on the diffusion of innovations (Rogers, 1995), McCannon said that the attributes of an idea that facilitate adoption are relative advantage, simplicity, compatibility with people’s values and beliefs, trial-ability, and observability. An idea that is trial-able and observable, he explained, is one that people can test and experience and see its benefits in the near term. As an example of simplicity, he noted that a draft guide addressing methicillin-resistant Staphylococcus aureus as part of the IHI 100,000 Lives Campaign was initially about 140 pages long. In the interest of scale and making it simple enough to actually be used, it was reduced to about one-third of that size.4
There are also infrastructure requirements to consider when designing for success and scale. Human resources, financial resources, physical space, equipment and supplies, data collection, technology, logistics,
3 100,000 Lives was the IHI national patient safety campaign to avoid unnecessary deaths in U.S. hospitals. See http://www.ihi.org/engage/initiatives/completed/5millionlivescampaign/documents/overview%20of%20the%20100K%20campaign.pdf (accessed February 20, 2015).
4 This is a corrected figure from what McCannon said (12 pages) at the workshop. He inadvertently confused different guides.
and oversight are all critical to think about in the early design phase, McCannon said.
Understanding the Audience
In a typical initiative, the people carrying out the initiative have a broad knowledge of their audience, that is, the people at whom the initiative is aimed. Rogers’ diffusion of innovation curve illustrates how, for any innovation, a given population will distribute into a bell-shaped curve with regard to how the members adopt the given idea or innovation. In an exceptional initiative, McCannon explained, there is also a detailed audience segmentation by, for example, geography (country, state, region, district), readiness (experienced, intermediate, novice), profession (e.g., administrator, doctor, nurse, community health worker), or type of facility (primary, secondary, tertiary). It is important to understand the population that the initiative is intended to reach, or the “customers.”
A stimulus5 is an incentive or driving force for change. In a typical initiative, there tends to be one stimulus, McCannon said. As mentioned earlier, a common incentive is payment. In contrast, exceptional initiatives employ many different stimuli to drive change. Stimuli can be positive, negative, or anywhere on the spectrum in between. Examples include emotional connection, recognition, sense-making, empowerment, collaboration, enjoyment, evidence base, payment, transparency, regulation, and punishment. McCannon recommended an 80/20 balance, with 80 percent of incentives on the positive end of the scale, and 20 percent toward the negative. Negative stimuli, such as regulation or punishment, are appropriate where there are cases of negligence or sabotage, he said.
Teaching Versus Learning
A typical pitfall in initiatives is relying too much on one teaching method, or relying too much on didactics in general, under the assumption that simply providing the information leads to change. Some weaker strategies for spread are papers, pamphlets, courses, websites, or conferences. There is a place for these methods, McCannon said, but learning methods are more appropriate for spreading change. Learning methods essentially democratize the change process, empowering people to make the innovation work for them in their environment or circumstance. There
5 McCannon used the term “stimulant” during the presentation, but likely meant “stimulus.”
are numerous such methods, McCannon said, and he highlighted some that had been discussed in the workshop, including extension agents, the IHI Breakthrough Series Collaborative Model, the campaign model, grassroots organizing, wave sequence (wedge and spread), and parallel process (broad and deep). In response to a question, McCannon clarified that the extension agent concept has its origins in the U.S. Department of Agriculture. The extension agent travels from site to site across a remote geographic area to bring ideas, collect problems and challenges, and serve as a connection.
The core principles of any successful learning method are hands-on application and a rhythm or tempo. People must be testing new ideas, seeing their results, assessing their progress, understanding the data for their population, and making adjustments on a daily basis, he said.
Replication Versus Adaptation
As discussed by McGahan (see Chapter 2) and others throughout the workshop, spread and scale are not simply replication. Exceptional initiatives focus on adaptation and are able to improvise to follow the theme, regardless of surprises or setbacks. This is true not just at the local level, McCannon said, but also at the level of a movement or a large-scale change initiative. The patient safety movement, for example, has been able to adapt to and take advantage of the opportunities presented by world events and emerging trends (McCannon and Perla, 2009).
In a typical initiative, especially a heavily funded initiative, a summative evaluation is often the priority. One reason for this, McCannon explained, is the need for attribution, as funders may need to establish the value of their investments. However, a summative evaluation is a complement to a formative evaluation. Starting with a formative evaluation (of daily data) as the priority is a hallmark of a successful large-scale change initiative, he said. Successful improvement relies not just on data, but on timely data that can be used to make adjustments on a frequent basis. An allowance for local adaptation and an appreciation of local context are made impossible by a summative design that is too restrictive, he said (Langley et al., 2009; Pawson and Tilley, 1997).
A typical initiative that struggles to get to scale and to have impact at scale often has systems where management gives approval. In the initia-
tives that are successful at scale, management places priority on removing barriers. As an example, McCannon described two contrasting scenarios. In the typical scenario, district representatives submit reports to the central office, and the central office rewards the timely submission of data. Occasionally, the central office reviews data and ranks performance, and underperformers are called in. A common byproduct of this approach is that many district representatives are tempted to falsify their data. In the alternative scenario, which correlates with better results, senior officials visit districts and facilities on a rotating basis. They spend 25 percent of their time reviewing progress together with the people in the districts and facilities, sitting on the same side of the table as the representatives. They spend the balance of their time identifying specific barriers that the leadership will remove by the next visit and identifying new tests that local owners will run. Being successful at having impact at scale means spreading culture and values, McCannon said. The culture in the first scenario is one of fear, with limited learning in the culture. The second scenario is one where participants are invested in the outcome and are solving problems together as a team.
The essence of leading a successful large-scale change initiative is keeping the process free of fear, he concluded, so that people can test, fail, experiment, adjust, be transparent about problems, and overcome obstacles as rapidly as possible to constantly make the intervention better. McCannon noted that Rebecca Solnit’s book, A Paradise Built in Hell (2009), describes the profound teamwork and fear-free environments that emerge in times of crisis.
During the brief discussion that followed, participants reflected on getting started and having time-specific goals and on summative versus formative evaluation. Participants also discussed the concept of exceptional initiatives as a learning system and reiterated the issue of misalignment between the payment system and population health as a barrier to scale and spread.
In considering the need to get started rather than spending time developing comprehensive strategies, George Isham of HealthPartners recalled the examples and lessons from the panel on tobacco control regarding the spread of ineffective initiatives. McCannon agreed that there can be big miscalculations in developing large strategies, and he suggested that this supports the wisdom of getting started, but starting small. This is not to say that there is not time for deliberative thought and design, he said, but one should set a short timeframe for when the initiative will begin (e.g., 6 months).
Isham concurred with McCannon’s comments on the need for time-specific goals. He noted that the Institute of Medicine consensus studies have made time-specific recommendations for improving population health. He cited the first recommendation in the report For the Public’s Health: Investing in a Healthier Future, which recommends that the secretary of health and human services set targets for life expectancy in the United States to be achieved by 2030.6
Paul Jellinek of Isaacs/Jellinek suggested that a rigorous summative evaluation of the prototype can pay huge dividends in terms of the subsequent rollout. Compelling cost–benefit or cost-effectiveness data can help secure financing going forward. Formative evaluation is then more appropriate for the project rollout. It is a sequential process, he said. McCannon agreed, but added that people sometimes confuse a summative evaluation with randomized controlled trials, and there are many other valuable forms of summative evaluation that may allow for greater appreciation of the texture and the context of the innovation.
David Kindig of the University of Wisconsin pointed out that the components of exceptional innovations outlined by McCannon form what Donald Berwick of IHI has referred to as a “learning system.” Berwick has also observed that in the most effective initiatives, there is someone in charge to manage the learning system. This work is so deeply multi-sectorial that often there is no one accountable for the outcome. Kindig asked how a diffuse-accountability, multi-sectorial system can still perform in these exceptional ways. McCannon responded that there does need to be an entity or organization (or representatives from multiple organizations) that will be responsible for the learning system. A learning system supplies people with data that allow them to change and improve themselves or else gives them the ability to collect those data and make change themselves. There is a surveillance function that is designed to see what is happening around the system and that is able to identify what is good, distill it, repackage it, and redistribute it very quickly. The learning system does not catalog or create databases; it focuses on tacit knowledge rather than explicit knowledge. A learning system that works is created and managed intentionally by a core group of people, he said.
Debbie Chang of Nemours raised the issue of misalignment between the payment system and population health, which was discussed by the first panel (see Chapter 3), and asked how that barrier to spread and scale
6 “Recommendation 1: The Secretary of the Department of Health and Human Services should adopt an interim explicit life expectancy target, establish data systems for a permanent health-adjusted life expectancy target, and establish a specific per capita health expenditure target to be achieved by 2030. Reaching these targets should engage all health system stakeholders in actions intended to achieve parity with averages among comparable nations on healthy life expectancy and per capita health expenditures” (IOM, 2012, p. 4).
might be overcome. McCannon responded that there is now a critical mass of lives covered under value-based or population-based payment models to serve as demonstrations. In some states, there is innovative work going on with Medicaid, and there are some private payers that are closely following these models and conducting small tests with their own populations. McCannon suggested that demonstrating the success of these models will lead more private payers to follow.
This page intentionally left blank.