7
Final Thoughts:
Ideas for the Future
As an introduction to a discussion on future directions, Teutsch noted that the workshop had already addressed what is needed to strengthen the field of modeling. “We have talked about the data issues, about how to enhance communications, how to help decision makers use these model differently, and how to capitalize on technology,” Teutsch said. He added that the field needs more people, particularly those who can help potential users access data and the models that are being developed. “We clearly need communicators to build bridges so that modeling is not just an academic exercise,” he said. At the same time, he continued, the field needs to build systems and incentives that link the clinical care system with community-based activities and policies that should drive both health improvements and savings.
Kindig agreed that if the value of modeling is to be fully realized, there are many areas that need attention, including at the research level where more studies are needed to understand the value and cost-effectiveness of interventions so that those data can be used to produce more reliable models. “That is an under-invested part of our academic research world, so there is a place for government and foundations to step up,” Kindig said. He also proposed that the government consider creating a model such as the one it uses for calculating the gross domestic product as a metric for national health outcomes.
Raymond Baxter from Kaiser Permanente said that one issue that needs to be considered is how to catalyze the culture change that will be required across many areas of health care and public health if stake-
holders are to accept modeling, to understand how and what it can be used for, and, eventually, to embrace it. “I suspect that modeling needs a public relations agent,” Baxter said. He agreed with Bobby Milstein and Pasky Pascual that the complexity of issues that modeling can address means that the community needs to move from mental models to formal models, even though he believes that most potential users of models still think they can integrate the growing amount of data and do these analyses in their heads. “Most of the people in this room are here because we are excited by models, we see the potential of how they can be used,” Baxter said. “But the challenge is that there is a variety of professions where the understanding of how to use models is simply not there, and it results in a kind of dismissal of models as black boxes.” Teutsch agreed with this last statement and noted that even good science gets dismissed by people who know how to use scientific uncertainty to cast doubt on topics where the science is actually compelling, such as the dangers of smoking, acid rain, and climate change.
Continuing on that theme, John Auerbach said that how to gain acceptance of modeling as a core tool is an important discussion to have in the context of other national discussions about what foundational public health is at the local, state, and federal levels. He said that people working at the community level need technical assistance and support in understanding how to do modeling and how to use it to shape discussions about public health and to add value to the models. “We need to build awareness that models can work,” Auerbach said, adding, “I would not have a clue on how to find someone who had expertise in modeling at a local level, and so I might find someone who was not good at it, and that might make for a bad experience.” The modeling community needs to provide guidance on how to find the right collaborators and the right model that can be plugged into a community organizing process, Auerbach said, and it needs to provide training and technical assistance.
Pascual said that while the centralized government approach and the distributed community-based approach to modeling go hand in hand, he does not believe that government should try to solve the health care sector ’s modeling problems. He also said that the modeling community is adept at writing computer code that can go out to the cloud, find data, and “scrape” it down into the model, and it is also good at writing code that can parse data, or convert data into a form that one particular model can use. Because of these two abilities, he said, he does not believe that a centralized approach to data gathering and model making is necessary. “If we just subscribe to the notion that you should describe your data, make it machine readable, and make it available, there is so much to be gained,” Pascual said.
Michael Weisberg said that as a modeler who works outside of the health community, he wanted the workshop participants to be aware of what members of other modeling communities, such as the social sciences modeling community, have done over the past couple of years that has been effective in making models more accessible. “They have developed a standard format for the presentation of the model so that people who do not choose to take time to understand the computational or mathematical detail have a sense of the assumptions in the model, the intended scope of the model, and the data that it relies on,” Weisberg said. He suggested that the roundtable could think about the needs of policy makers, physicians, and public health officials in terms of the inputs, outputs, design, and scope of models and set standards for characterizing models in those terms. He also expressed hope that the roundtable would continue to stress the importance of thinking of models as tools that can support policy decisions, not make them.
The final comment from the discussion came from a workshop participant who said that it seemed to him that the discussions of the day hinted at trying to reinvent community health planning while losing sight of the community stakeholders in the process. He called on the government to play a crucial role in ensuring that different actors in the system have input into the process, something that he did not hear happening in the modeling examples that were presented.
REFLECTIONS ON THE DAY1
In the workshop’s final discussion, roundtable members were given the opportunity to reflect on the key messages that they heard during the day’s proceedings and to offer any comments they had that were relevant to the topic of how modeling can inform strategies to improve population health. George Isham started the discussion with the comment that, with regard to government policy on the formation of accountable care organizations and the shift in payment models from fee-for-service, information and analytics are two of the capabilities that are most commonly identified as needs of health care systems. Isham noted that there are operational models that have a long track record in some of the leading health care organizations, such as Kaiser Permanente and HealthPartners, and he wondered what lessons can be learned from those operational models that can be applied to the larger task of improving health in populations and in communities larger than health care systems.
___________________
1Unless otherwise noted, the comments noted in this chapter were made by members of the Roundtable on Population Health Improvement, and the statements are not endorsed or verified by the Institute of Medicine.
James Knickman said that one encouraging piece of information he heard is that models may be able to simulate some of the changes in population health that may occur over a long time frame and, by doing so, address the important challenge associated with how long it takes to accumulate actual evidence of change produced by an intervention. While acknowledging the limitations of models compared to research-derived evidence, he said that models may represent the only way for society to reach decisions to invest in long-term interventions. He also said that models can be useful tools for presenting information clearly to decision makers.
Mary Pittman first commented that the ReThink Health model has improved and become more refined each time she hears a presentation about it. She then noted that an important theme that she heard in her working group was the need to humanize data in order for it to have an impact on policy making. “You can have harmonized data and not know how to apply it in a way that it will be a good input into a model,” she said, “or it could be data that is not relevant to the policy at hand.” It is also important to humanize data, she said, in order to get community inputs and to ensure that a model is accounting for the populations that need to be served, both by the data and the policy being modeled.
Catherine Baase reflected on the fact that each of the roundtable’s workshops she has attended has identified issues that have a big impact on population health, but the question that continues to hang over these workshops is how to amalgamate these issues and distill them into a path forward. One of her takeaways from the workshop, she said, is that modeling could be the vehicle or tool that can help with this necessary amalgamation and distillation and perhaps address one of the greatest challenges facing the roundtable, which is to create a path forward out of all of the information these workshops have produced. Yes, she said, modeling is a tool to support decision making, but it can also be a tool that is useful for convening and for shared learning.
Raymond Baxter agreed with Baase’s ideas that modeling could be an important vehicle for integrating ideas and that it could also help with communication, given that the roundtable has often heard how difficult it is to explain what population health is. One of the benefits of a model is that by examining its assumptions and tradeoffs, it is possible to see how taking one action plays itself out in other ways, Baxter said.
In response to a question from Isham about how modeling might be useful to employers interested in improving productivity, Baase said that the business sectors uses models all the time in that way and that they are very effective tools. Modeling is also a language that is very well understood in the private sector, and it could serve as a bridge between the private and public sectors and the community at large. Baxter added that
modeling can help any type of organization, private or public, understand the varied impacts it has on health by the way in which they do their business. Kaiser Permanente, for example, is in the process of measuring what it calls its Total Health Impact, which is the effect of everything that the organization does, including the pensions and wages it offers, the toxics that are used in the delivery of health care, the energy that it uses, and so on. Modeling is playing an important part of this effort and will provide Kaiser with a better picture of how its actions as an organization affect the health of the communities in which it operates.
The last comment came from Sanne Magnan, who remarked that one of the objectives for the workshop was to show how modeling could help decision makers, which the workshop did accomplish, but the presentations also highlighted how helpful modeling can be in creating conversations between the decision makers and communities, or, as she put it, between the treetops and the grass roots. This idea was reinforced throughout the workshop, she said, and it points to how important it is to engage and empower communities and respect the role that they have to play in population health.
After thanking the planning committee and the Institute of Medicine staff for putting together a workshop that generated thoughtful discussions and information that will be useful to the roundtable, Isham adjourned the workshop.
This page intentionally left blank.
