Proceedings of a Workshop
Integrating Systems and Sectors Toward Obesity Solutions
Proceedings of a Workshop—in Brief
The Roundtable on Obesity Solutions of the Health and Medicine Division of the National Academies of Sciences, Engineering, and Medicine held a virtual public workshop, Integrating Systems and Sectors Toward Obesity Solutions, on April 6, 2020 (Part I), and June 30, 2020 (Part II). The workshop explored complex systems and contributing factors that can influence obesity, and shared real-life examples of applying systems thinking and systems science approaches to addressing obesity and population health and well-being. In Part I, speakers provided an overview of systems science theories and approaches and their application. In Part II, speakers discussed complex systems in society that have the potential to shape the public’s health and considered opportunities for systems change with regard to obesity solutions. Specifically, the workshop explored how factors such as power dynamics, structural racism, relationships, resources, place-based issues, policy, and political will affect systems that can influence obesity, as well as how these factors can impact communications and cross-sector collaboration to address obesity.
This Proceedings of a Workshop—in Brief highlights the presentations and discussions that occurred at the workshop and is not intended to provide a comprehensive summary of information shared during the workshop.1 The information summarized here reflects the knowledge and opinions of individual workshop participants and should not be seen as a consensus of the workshop participants, the Roundtable on Obesity Solutions, or the National Academies.
INTRODUCTION AND OVERVIEW OF SYSTEMS THEORIES AND APPROACHES AND THEIR APPLICATION
Part I of the workshop featured a session that provided background information on complex systems theories and approaches and their applications. Christina Economos, co-founder and director of ChildObesity180 and professor and New Balance Chair in Childhood Nutrition at the Gerald J. and Dorothy R. Friedman School of Nutrition Science and Policy at Tufts University, shared definitions for key terms used throughout the workshop (see Box 1).
Ross Hammond, Betty Bofinger Brown Associate Professor at Washington University in St. Louis, presented an historical overview of systems science, including its advantages and potential applications to public health and obesity research and intervention. He began by underscoring that the complex, multifaceted nature of many public health problems, including the obesity epidemic, make them well suited for examination with systems science approaches.
Hammond described four characteristics of complex systems, such as those that drive obesity. First, many different factors interacting across multiple scales affect relevant behaviors and outcomes, resulting in a deeply interconnected system. He pointed out that an isolated focus on a single part of such a system risks missing many other
1 The workshop agenda, presentations, and other materials are available at https://www.nationalacademies.org/event/04-06-2020/integrating-systems-and-sectors-toward-obesity-solutions-part-1 (accessed August 26, 2020) and https://www.nationalacademies.org/event/06-30-2020/integrating-systems-and-sectors-toward-obesity-solutions-part-2 (accessed August 26, 2020).
important factors and linkages. A second characteristic is the inclusion of the population being studied. These multiple heterogeneous actors have different incentives, information, and network connections, Hammond said. He explained that these interconnections are important because an intervention may unintentionally affect certain actors. Because of actors’ adaptability, their short-term and long-term responses to environmental and policy changes may differ. A third characteristic is the importance of both the context and timing of exposures as people move through their environments. Potentially important factors within the structure of their lived experiences are concealed, he suggested, if measures of exposures are averaged across jurisdictions. Lastly, Hammond highlighted the dynamic nature of complex systems and the potential importance of the timing and the sequence of exposures and interventions as they relate to the development of obesity and other chronic disease outcomes across the lifespan.
Turning to the implications of applying systems science approaches to obesity, Hammond emphasized that the problem’s complexity is not likely to be captured by standard tools focused on single drivers. Furthermore, he continued, a growing consensus supports solutions that are broad enough to address a variety of contributing factors but also are sufficiently tailored to specific contexts and adaptive over time. To support these points, Hammond referenced a series of reports2 endorsing the use of complex systems thinking and approaches to inform solutions for obesity and other public health issues. These approaches help leverage or at least address the problem’s complexity, he explained, and also come with the challenge of cultivating collaboration across disciplines and sectors.
Hammond stressed that stakeholders examine systems from a holistic perspective and search for coordinated solutions and points of leverage beyond traditional arenas of health. He explained that the field of public health has advanced from using systems thinking to adopting specific systems science tools and methods. Hammond expounded on quantitative tools, explaining that they embrace complexity and seek to understand mechanisms that drive outcomes, but differ with regard to the perspective from which they seek to understand processes and outcomes, the relative emphasis they place on data inputs versus theoretical inputs, the training they require, and their field of origin.
Hammond briefly reviewed the history of complex systems science, pointing out that many of its methods date back to the 1950s or earlier and emerged from different fields. Public health first used these methods to inform the control and management of infectious diseases, he recounted, and they gained traction thanks to research investment that resulted in the creation of the large Models of Infectious Disease Agent Study (MIDAS) network in 2003. According to Hammond, complex systems approaches have since been applied to other public health topics for three key purposes: understanding the etiology of outcomes of interest to inform intervention targets; retrospectively deducing the key influences on intervention outcomes; and conducting prospective modeling to forecast potential outcomes of
2 IOM (Institute of Medicine). 2012. Accelerating progress in obesity prevention: Solving the weight of the nation. Washington, DC: The National Academies Press. https://doi.org/10.17226/13275; IOM and NRC (National Research Council). 2015. A framework for assessing effects of the food system. Washington, DC: The National Academies Press. https://doi.org/10.17226/18846; IOM. 2015. Assessing the use of agent-based models for tobacco regulation. Washington, DC: The National Academies Press. https://doi.org/10.17226/19018; NASEM (National Academies of Sciences, Engineering, and Medicine). 2016. Assessing prevalence and trends in obesity: Navigating the evidence. Washington, DC: The National Academies Press. https://doi.org/10.17226/23505; Secretary’s Advisory Committee for Healthy People 2030. 2018. Issue briefs to inform development and implementation of Healthy People 2030. Washington, DC: U.S. Department of Health and Human Services. https://www.healthypeople.gov/sites/default/files/HP2030_Committee-Combined-Issue%20Briefs_2019-508c.pdf (accessed September 9, 2020).
different policy and intervention options that might be challenging, cost prohibitive, or unethical to study with real-world experiments.
Hammond shifted his exposition of history to describe how obesity research has been informed by systems science approaches, typically systems dynamics modeling,3 social network models,4 and agent-based modeling.5 He remarked that from 2009 to the present, the literature has evolved from explaining why systems science approaches are apt for studying obesity and how they can help examine its etiology to offering solutions. This evolution is the result of investment, training, and collaboration, Hammond maintained, and he noted that in the future, systems science approaches can inform the sustainable, tailored, effective implementation of multifaceted or whole-of-community interventions. Hammond acknowledged that there are inherent limitations to using these tools, including resources, training, and expertise to use the models and evaluate their effectiveness.
Sandro Galea, dean and Robert A. Knox Professor at the Boston University School of Public Health, discussed using systems thinking to understand population health. The fundamental point, he asserted, is that population health and complex challenges, such as obesity, cannot be broadly understood without taking a systems science approach.
Galea said that population health is often approached from a deterministic perspective, which he explained is characterized by the assumption that individuals are relatively interchangeable. In reality, he pointed out, substituting one person for another could have vastly different effects in a given situation because populations are heterogeneous. They are comprised of markedly diverse individuals who have specific social network structures and connections, he elaborated, and the same inputs do not always produce the same outcomes because populations evolve and adapt over time.
Galea shifted to expound on population health’s compatibility with systems thinking by explaining three foundational principles of population health science. The first is that population health manifests as a continuum. Because a simple dichotomy of healthy versus unhealthy does not exist, one must recognize the importance of understanding the full spectrum of health within and across populations, he said, describing the range of body mass index (BMI) values within a given population as an example of this point. Cutoffs for being overweight (BMI ≥ 25 kg/m2) or having obesity (BMI ≥ 30 kg/m2) mask the full scale of the distribution of weights and BMIs throughout the population, he pointed out, as well as the dynamics of how they are shifting.
The second principle is that small changes in ubiquitous causes may result in more substantial changes in the health of populations than larger changes in rarer causes. Galea used a metaphor of a goldfish in a fishbowl to illustrate this point. If the goldfish wants to be healthy, he said, it might make a variety of behavioral changes such as swimming daily laps and regulating its intake of fish food. But if the water in the bowl—which is ubiquitous around the goldfish—is not changed, Galea continued, modifications in the goldfish’s behavior will matter little because the unclean water will cause it to die. With regard to obesity, Galea said that a ubiquitous cause that can undermine individual behavioral changes is an unhealthy food environment. Differences in food environments can help explain heterogeneity in the prevalence of obesity across adjoining communities, he explained, and he called for the application of systems science approaches to understand both the individual drivers and the ubiquitous forces that affect population health.
Galea highlighted a third principle: the magnitude of the effect of an exposure on disease is dependent on the prevalence of the factors interacting with that exposure. As an example, he explained that it is impossible to determine the percentage of risk that one’s genes (the exposure) contribute to obesity (the disease outcome) unless the environment (the interacting factor) that creates the condition for the disease is known. This principle forms a fundamental argument for the application of systems thinking to obesity etiology, Galea asserted, because it accepts that multiple co-occurring factors determine obesity, and that the relationship between one cause and an outcome can be understood only if the other causes are understood.
Galea moved on to the final portion of his presentation, a discussion of complex systems and counterfactual thinking. He described the concept of a counterfactual by comparing two universes: an observed universe and a parallel universe that is identical to the observed universe except for a single variable and the outcome. The observation of
3Systems dynamics uses informal and formal models with computer simulation to uncover and understand endogenous sources of complex system behavior. Source: Luke, D. A., and K. A. Stamatakis. 2012. Systems science methods in public health: Dynamics, networks, and agents. Annual Review of Public Health 33:357–376.
4Social network models measure and analyze relationships and flows among actors, including people, organizations, and other information processing entities. Source: Luke, D. A., and K. A. Stamatakis. 2012. Systems science methods in public health: Dynamics, networks, and agents. Annual Review of Public Health 33:357–376.
5Agent-based modeling uses computer simulations to examine how elements of a system (agents) behave as a function of their interactions with each other and their environment. Source: Luke, D. A., and K. A. Stamatakis. 2012. Systems science methods in public health: Dynamics, networks, and agents. Annual Review of Public Health 33:357–376.
a different outcome based on the manipulation of only one variable allows a researcher to determine that the manipulated variable has a causal effect on the outcome, Galea explained, and instills confidence in selecting that variable as the intervention target. It is difficult to mimic such parallel universes, Galea admitted, and he suggested that systems science approaches, such as agent-based modeling, can complement other observational and experimental approaches that help simulate the counterfactual.
Finally, Galea urged participants to address population health issues with the mindset of “as simple as possible, but not simpler,” a quote attributed to Albert Einstein. Population health epidemics such as obesity are driven by a complex set of forces, he stressed, and he called for systems thinking to identify key levers for intervention that will achieve the maximal impact on both improving overall population health and narrowing disparities.
Douglas Luke, professor and director of the Ph.D. program in public health sciences at Washington University in St. Louis, provided an overview of applications of systems science approaches. Luke described three insights from his group’s application of systems science approaches to the study of chronic disease topics. The first is that systems science approaches facilitate exploring the role of context in chronic disease processes and interventions, which are shaped by social, economic, physical, temporal, and political/historical contexts, among others. Luke urged that frameworks and models capture these contexts so that stakeholders can understand how interventions are embedded within them. To illustrate the importance of context, Luke described the evolution of epidemiological models that predict the time course of infectious disease outbreaks. Traditional models that ignored social structures and assumed random mixing have been replaced by network graphics, he explained, which illustrate the nonrandom social and physical contexts of disease transmission.
Luke’s second insight was that systems science approaches can help discover causal mechanisms that help explain how or why something works. He described his research group’s Tobacco Town initiative, which uses agent-based modeling as a policy laboratory to explore potential effects—based on computational models—of various tobacco retail policies across contexts and populations. Tobacco Town features computer-based, virtual towns as simulation models to identify interactions between the tobacco retail environment and tobacco purchase and use behaviors.
Luke reviewed examples of Tobacco Town in action, reporting that it confirmed the nonlinear relationship between retailer density and the time and monetary costs of procuring tobacco products. Density reduction may need to reach a threshold, he elaborated, before behavioral effects are observed. The patterns of these effects differ based on the urbanicity and income levels of modeled towns, he added, an observation implying that policies have different potential for affecting disparities and behaviors in different types of communities.
Luke shared a heat map visualization of the potential effects of two types of tobacco retailer policies as an example of how agent-based modeling can help explore underlying mechanisms. Luke pointed out that although both policies had similar effects on reducing retailer density, one of them had a bigger impact on proximity because it created a much higher average distance between residents and retailers. He highlighted this example’s mechanism of proximity as critical for community policy development and intervention.
Luke’s third insight was that complex systems approaches can produce tools that are particularly relevant and useful for community stakeholders and policy makers. He recounted an application of the Tobacco Town model to communities in Minnesota, which found that effects of policies to reduce tobacco retailer density were not “one size fits all.” The effects varied by contexts, such as a community’s urbanicity and income level, and disparities in density diminished across communities as policies became stronger. The Tobacco Town model has also produced an interactive Tobacco Swamp dashboard, Luke added, to help policy makers explore the effects of various policies in different cities.
COMPLEX SYSTEMS IN SOCIETY AND THE CONTEXT FOR OBESITY
Part II of the workshop began with a session featuring four speakers who explored complex systems in society that provide context for obesity and have potential to shape population health and well-being.
Chandra Ford, associate professor of community health sciences and founding director of the Center for the Study of Racism, Social Justice & Health at the University of California, Los Angeles, discussed power dynamics, structural racism, and systems science approaches from the lens of critical race theory (CRT).
Ford drew a parallel between systems science approaches and CRT, which she explained as a set of intellectual ideas, principles, and approaches to identify, understand, and undo the root causes of race, racism, and power as they operate and drive societal inequities in the post–civil rights era.
She highlighted three founding critical race theorists: Derrick Bell, who asserted that to address the less perceptible forms of racism, their mechanisms must first be made explicit through racism-conscious research; Kimberlé Crenshaw, who coined the term intersectionality; and Cheryl Harris, who promulgated the concept of whiteness as property. Ford also mentioned Public Health Critical Race Praxis, which she described as an offshoot of CRT that provides a model for applying CRT to the health sciences.
Ford suggested three ways that CRT can help obesity-focused systems science research incorporate the primacy of racialization. Her first suggestion was to identify and include racism in mental models and analyses. Substantial evidence links racism to health, Ford maintained, and systems dynamics models that include appropriate measures of racism could illuminate how structural racism contributes to obesity. Ford also urged measuring race and ethnicity as social constructs. Her second suggestion was to address racialized power dynamics. She encouraged power sharing and promoted community-based participatory research as an approach to fostering equitable partnerships between researchers and communities. Ford also cautioned against using sources of big data that reinforce racial or ethnic marginalization. Her third suggestion was to address the social construction of knowledge. Although the scientific method enhances the reliability of empirical findings, she maintained that it does not necessarily eliminate the influence of racial bias. She suggested that researchers engage in “critical reflexivity” to assess how they may apply inadvertent subjectivity to their work. According to Ford, anecdotal evidence suggests that raising people’s awareness of their biases and increasing the amount of time they have to respond to them may reduce the inadvertent effects of implicit biases.
Ford concluded by affirming that systems science approaches can illuminate racism-related dimensions to obesity morbidity and mortality, and that CRT can help systems science approaches identify underlying racial drivers of obesity and inequities.
Kayla de la Haye, assistant professor of preventive medicine at the University of Southern California, discussed social networks and relationships. She indicated that it is challenging for people to sustain lifestyle behavior modification because they are not embedded in social and environmental structures that support healthful habits over the long term. de la Haye suggested that much potential exists to leverage social networks when taking a systems science approach to solutions for obesity. She defined these as structures comprised of social actors and their interrelationships, such as local networks of family, friends, and community contacts that share kinship, as well as global networks of community members, stakeholders, and decision makers who collaborate to shape the structural features of lived environments.
de la Haye next described social network analysis, a systems science methodology that uses theoretical and analytical frameworks to study emergent patterns of actors and relationships in a network. Social network analysis also examines the impact of social structures on individual and group outcomes, referencing research indicating that adults and children with obesity tend to cluster in social networks and that having social connections with obesity increase a person’s obesity risk over time. de la Haye highlighted three drivers that have been linked to these patterns.
The first driver is homophily, she began, a phenomenon whereby people tend to select or form social ties with others who have similar characteristics, including ethnicity, social economic status, or health risks and inequalities. Propinquity, she went on, refers to people connecting to others through shared social and physical spaces in homes, communities, and organizations, where structural and environmental influences are similar. Weight-based stigma is a third driver, she said, explaining that people with obesity are often socially rejected by peers who do not have obesity, resulting in the formation of more social connections among individuals with similar weight status and the marginalization of those who are overweight. In addition to these three drivers, de la Haye reported that evidence also supports that social ties directly influence people’s weight norms and weight-related behaviors through a number of social influence mechanisms, which can lead to similarities in the risks for being overweight among family and friends.
Next, de la Haye shared two intervention approaches to leverage or change social networks as part of obesity solutions. One is segmentation, which targets at-risk social groups (e.g., family or peer groups) rather than individuals, as promoters of positive social influence, norms, and support for healthy habits. A second is alteration, she continued, in which an intervention tries to build new adjacent health networks that can increase a person’s access to positive social influences, norms, and support.
Lastly, de la Haye highlighted the promise of novel data sources, such as smartphone-captured big data and network information visible on social media, for providing insights into complex social systems and how they affect obesity solutions. She cautioned that certain voices, relationships, and social phenomena are likely to be privileged or hidden in different data sources.
Ana Diez Roux, dean and Distinguished University Professor of Epidemiology in the Dornsife School of Public Health at Drexel University, discussed the impact of neighborhood characteristics on health. She emphasized that neighborhoods are important contexts for physical and social exposures, making neighborhood differences potentially important contributors to health inequities.
Diez Roux said that identifying the causal effects of neighborhoods is complex because many factors, such as neighborhood features and individual attributes of the residents, interact and evolve over time to drive neighborhood differences in health. According to Diez Roux, one advantage of using systems science approaches to examine neighborhood effects on health is their utility for better understanding the bidirectional person–environment relations that occur in neighborhoods. Other advantages, she continued, are the ability of systems science approaches to parse
interactions among people, interactions and interrelations between physical and social environments, and spatial patterning or segregation of individual and environmental characteristics.
Diez Roux highlighted research called agent-based modeling that applied a systems science methodology to explore how residential segregation by income can create disparities in diet even if differences in food preference or price do not exist, and found that changing food preferences is not enough to eliminate disparities. The modeling exercise compelled the team to consider processes through which dietary disparities arise, she added, which generated ideas for new data collection and analyses.
Diez Roux suggested that the most valuable aspect of a systems science approach may be that it prompts users to rethink their research questions. This can transform a relatively limited question (e.g., “are neighborhood characteristics independently associated with health after accounting for individual-level socioeconomic status?”) into a more nuanced, policy-relevant question (e.g., “to what extent [and under what conditions] could residential segregation generate and reinforce health disparities by race?”). Finally, she submitted that stakeholders can gain a more complete understanding of population health by combining systems science approaches and using them to quadrangulate across different methodologies, such as agent-based modeling and group model building.
Tiffany Powell-Wiley, Earl Stadtman Tenure-Track Investigator at the National Institutes of Health and chief of the Social Determinants of Obesity and Cardiovascular Risk Laboratory, discussed the intersection of policy and systems science approaches in the context of social determinants of physical activity and obesity. Powell-Wiley provided examples of her work on the relationship between cardiometabolic risk and neighborhood social environments. These have included community engagement, epidemiologic studies, systems science approaches (including agent-based modeling, which uses computer simulation to study complex systems), and translational science and intervention trials.
Powell-Wiley described how these examples played out in her team’s work with the Washington, DC, Cardiovascular Health and Obesity Collaborative, which she said was built on engaging communities through an advisory board of multidisciplinary church and community leaders. The board provided input for research project development and design, she explained, such as the collaborative’s first project, a cardiovascular health and needs assessment that was conducted with community-based participatory research principles. The assessment identified the feasibility of using mobile health technology to develop a physical activity intervention, she reported, and also indicated that community members perceived crime and limited safety as barriers to physical activity.
After the team reviewed data from the Multi-Ethnic Study of Atherosclerosis (MESA) cohort study in an attempt to extend the findings of the community-engaged work, Powell-Wiley said they found that a greater decrease in neighborhood- and individual-level safety over time was associated with a greater increase in adiposity, but a clear relationship did not exist between objective crime and adiposity. This illuminated the concept of perceived safety as a potential intervention target, she underscored, which could both increase physical activity and address residents’ concerns about crime and safety.
Powell-Wiley next described the intervention’s use of agent-based modeling to simulate the impact of crime on physical activity and obesity among African American women from 80,000 households in areas of the city with the highest prevalence of obesity. Powell-Wiley reported that as an individual’s propensity to exercise increased, reductions in crime that increased accessibility to physical activity locations were associated with a greater decrease in prevalence of obesity. She suggested targeting crime through urban renewal policies to improve perceived safety in resource-limited urban communities, and shared an example of a health equity–focused urban renewal process that engaged community residents and research partners in a community economic development planning process. Powell-Wiley shared that in the future, her research group will develop and use agent-based models to test multilevel mobile health interventions to promote physical activity, and potentially also test how crime may limit the intervention.
IMPACTING COMPLEX SYSTEMS THAT CAN INFLUENCE OBESITY
Part II’s second session included four speakers who explored how complex systems may influence obesity and who considered opportunities for systems change as they relate to obesity solutions.
Leah Frerichs, assistant professor in the Department of Health Policy and Management at the University of North Carolina at Chapel Hill, discussed engaged and participatory systems thinking as a way to integrate systems science approaches and community-based participatory research. She outlined synergies that exist between engaged and participatory research and systems thinking and systems science approaches, such as shared emphases on holistic understanding and social ecological frameworks, and she explained that the two approaches can be combined in a variety of ways.
As the continuums of engaged and participatory systems thinking and community involvement flow from the initial community outreach (where she said researchers retain most of the power) to shared leadership (where she explained that researchers and community members become more equitable partners as they co-develop processes for ongoing engagement), Frerichs pointed out that a gradual increase occurs in the level of community involvement, impact, trust, and communication flow. She described two examples to illustrate this continuum.
Frerichs first described an effort that engaged community health care providers to understand clinical processes for colorectal cancer screening. The resulting process flow diagrams help identify health-related decision points, responsible parties for various processes, and potential gaps or bottlenecks. The formative research that led to the diagrams also informed a micro-simulation model that has helped forecast the potential impact of health coverage changes and suggested evidence-based, cost-effective interventions.
In a second example, Frerichs described a shared leadership approach that engaged high school students in a rural community toward understanding complex systems influences on physical activity. The students helped define research questions and establish study protocols, she said, and took gradually increasing responsibility for planning meetings and facilitating activities. The project linked five storytelling elements with concepts from agent-based modeling, an exercise that Frerichs said made the modeling components more relatable for the students: conflict (the health issue, i.e., physical activity), characters (agents), character development (model properties and rules), setting (environment), and plot (model simulation). She said the final model structure will be used to forecast the relative impact of different solutions to increase physical activity.
To conclude her presentation, Frerichs expressed hope for future engaged systems thinking strategies to focus on deeper, higher-impact leverage points, such as changing mindsets, which she said have strong potential to create sustainable change and better address inequities.
Matt Kasman, assistant research director at the Brookings Institution Center on Social Dynamics and Policy, discussed the use of systems science approaches to explore connections between education and health. He offered three categories into which these connections can be placed and shared examples of systems science research in each category.
Kasman explained that the first category—direct relationships—includes the ways that school environments and school setting interactions can impact student health, such as access to healthy school meals and appropriate physical activity options. Kasman highlighted the Childhood Obesity Modeling for Prevention and Community Transformation (COMPACT) collaboration, explaining that one of its key insights is to illuminate ways that community members in different settings, including educational institutions, can work together to prevent childhood obesity.
The second category is indirect connections, Kasman continued, which refers to the ways that educational outcomes relate to later health outcomes. A strong body of evidence indicates that college attendance and completion can influence later employment, place of residence, and social connections, which Kasman said are factors that can have important health implications. He referenced a systems science approach using an agent-based model of college enrollment that informs policies and programs to make access to higher education more equitable and reduce persistent attendance gaps.
Kasman described the third category, causal influences of health on educational outcomes, as a nascent area for research. Because clear linkages exist between education and health, he explained, causal mechanisms in the opposite direction may represent important feedback loops. He suggested that systems science approaches could help illuminate key dynamics, disentangle interconnected causal mechanisms, and identify promising intervention points to improve health outcomes and reduce disparities.
Kasman ended by suggesting three actions to facilitate the use of systems science approaches for examining relationships between health and educational activities and outcomes: build audience familiarity and enthusiasm for using systems science approaches; increase capacity for systems science research such as by developing tutorials and promoting mentoring relationships between experienced and new researchers; and open lines of communication among systems science modelers, researchers, policy makers, intervention experts, and practitioners.
Eric Hekler, director of the Center for Wireless & Population Health Systems and associate professor in the Department of Family Medicine and Public Health at the University of California, San Diego, spoke about the use of systems science approaches to foster behavioral change.
Hekler emphasized the complexity of trying to help people live healthy lives by listing three dimensions to consider when advocating for behavioral change: the importance of contextual factors, the dynamic nature of these factors, and the idiosyncratic nature of how different factors influence an individual. He then described three lessons from his evolving thinking about advancing systems science approaches so that they correspond to the complexity of human behavioral change. Hekler introduced the notion of humanistic systems science that acts less like a yardstick and more like a Global Positioning System in that it is attuned to context in order to determine how to modify goals under different circumstances.
First, he espoused a focus on “verbs” instead of “nouns”; for example, by focusing on dynamics, interconnectivity, and flows within and across systems instead of interventions, levels, and outcomes. To illustrate this point, Hekler gave an overview of control systems engineering, which he described as a large, pervasive field focused on dynamic decision making in complex environments. Control systems focus on the dynamics of each individual system involved in a process, he explained, and use a technique called system identification to build dynamic models based on data from an actual person or other unit of interest. The models feed into simulations of how a person will respond in future
scenarios, he continued, and can account for changing contexts and honor interpersonal differences as they determine the best decisions over time in order to reach a goal.
Hekler cautioned that an unintended consequence of building control algorithms is that they can manipulate people, which led to his second lesson: Whoever defines success and categories has the power. Defining success is better left to the people being served by systems science tools, he maintained, and he urged opening innovation pathways for nontraditional researchers and experts to contribute to these tools.
The third lesson, Hekler continued, is that training in humanistic systems science is important and must balance mathematics, computation, modeling, and algorithm development with the cultivation of appropriate mindsets, processes, and skills. Noting that scientists tend to hold systematic biases and privileges, he shared his personal experience with overcoming what he called “confident ignorance and emotional blindness.” Hekler espoused the qualities of curiosity, humility, and compassion as key ingredients for recognizing one’s biases and building awareness of others’ lived experiences.
Erin Hennessy, assistant professor at the Gerald J. and Dorothy R. Friedman School of Nutrition Science and Policy at Tufts University, discussed the use of systems thinking and systems science approaches to advance community-level obesity prevention. She began by describing how an iceberg model can help explain systems thinking and systems change because it illustrates how an outcome, such as obesity, manifests visibly but is strongly influenced by underlying trends (e.g., poor diet, inactivity, and stress), structures (e.g., policies, practices, resource flows), relationships (e.g., relationships and connections, and power dynamics), and mental models or deeply held beliefs, assumptions, and operating styles that influence thoughts and actions. Therefore, Hennessy affirmed, the structural, relational, and mental model elements are promising leverage points for effecting systems change.
Hennessy described four obesity-related opportunities for structural, relational, and transformative change in communities: (1) taking whole-of-community approaches, which are implemented holistically throughout an entire community and target multilevel influences and behaviors through policy, practice, and resource flows; (2) leveraging community coalitions, which are groups of leaders and stakeholders from diverse organizations, settings, and sectors working collectively on a common objective; (3) sharing and shifting mental models; and (4) integrating systems science approaches (e.g., community-based systems dynamics or group model building) to engage communities in the process of understanding and changing systems, which can greatly expand research questions and yield novel insights.
Hennessy described how the Shape Up Somerville initiative leveraged each of these opportunities in its whole-of-community, community-based participatory research approach to affect structural and relational changes to prevent obesity. She reported that the intervention’s policy and environmental changes, which were informed by a systems science approach (a qualitative systems mapping exercise) and diffused through a community coalition, resulted in a significant decrease in BMI z-score among children living in the intervention community. She noted that broader efforts are under way to generate empirical evidence on the contributions of coalitions to whole-of-community interventions in improving child obesity outcomes.
EXAMPLES FROM THE FIELD: APPLYING SYSTEMS THINKING TO POPULATION HEALTH ISSUES
Part II’s third session featured three speakers who shared field examples of research that applies systems thinking to address obesity and population health.
John Jakicic, director of the Healthy Lifestyle Institute and chair of the Department of Health and Physical Activity at the University of Pittsburgh, discussed examples of integrating systems and sectors to promote physical activity. Jakicic shared examples of his team’s translation of learnings from a controlled research environment into an intervention that aims to increase physical activity among faculty and staff at the University of Pittsburgh, which employs around 15,000 adults. The team began by surveying members of the university community about supports for physical activity. Although much of the feedback revealed a desire for more exercise facilities and equipment, he reported, targeting exercise is markedly different from promoting a physically active lifestyle. Furthermore, he reported that only a small percentage of adults at the university are regular fitness center users, but that those who avoid exercise facilities indicated a willingness to engage in other forms of activity.
Jakicic said the team discovered that a key element of building a community engaged in physical activity was to make people feel like they are part of something along with people who are just like them. This led to the birth of the Be Fit Pitt initiative, he continued, which attracted people who are not regular exercisers but who want to move more as they view activity from a healthier lifestyle perspective. The initiative introduces new activity options and programming every 8 to 12 weeks to creatively activate the university community, an evidence-based tactic that Jakicic said helped the initiative take root and grow. Over time, the initiative has introduced remote activity opportunities such as live-streamed or recorded video activities, he added, which satisfies participants’ desire to engage in physical activity without feeling the intimidation of being visible.
Jakicic recounted that as the team expanded the initiative into the broader Pittsburgh community, they discovered that a key strategy was to make activity accessible for people at different levels of fitness and self-efficacy, such as by providing relatable models and a variety of activity options to cater to people’s varying fitness levels. He ended his presentation with a note about community engagement centers, where university representatives host community discussions to understand community needs and assets.
Bonnie Spring, professor of preventive medicine, psychology, psychiatry, and public health at Northwestern University and director of the Center for Behavior and Health within the university’s Institute for Public Health and Medicine, discussed the design of multilevel interventions to treat obesity, which she said produce larger and longer-lasting effects than interventions that target only one level of the social ecological model. It is important, Spring added, to understand multilevel systems synergies and constraints on implementing an obesity intervention, and to continually optimize such interventions for both effectiveness and efficiency.
Spring described how her team intervened to modify the Diabetes Prevention Program (DPP), an intensive, multicomponent behavioral treatment program to prevent or delay type 2 diabetes, to target additional dimensions of the social ecological model beyond the individual level. The initial DPP treatment package involved 24 one-on-one, in-person treatment sessions led by health professional counselors, Spring explained, but now it is common for group sessions to be led by trained lay counselors for a much lower cost. Spring elaborated that her team’s intervention was remote and targeted the individual level by promoting app-based, self-monitoring of daily weight and dietary intake and provided 24 instead of 12 individual coaching calls. It targeted the interpersonal level by pairing participants with support buddies, she continued, the organizational level by sending weight loss progress reports to participants’ primary care providers, and the environmental and policy level by offering 1 week of portion-controlled meal replacements and then recommendations for meal replacements in subsequent weeks. Ultimately, Spring said the intervention sought to discover which treatment components contributed most to 6-month weight loss and at what cost.
Spring reported that the intervention component that significantly increased weight loss was buddy support, an effect that was enhanced when an individual’s weight loss progress report was sent to a primary care provider. She explained how the study’s results data are useful for decision making about which intervention components are essential to achieve meaningful weight loss, particularly when limited-resource situations call for best value solutions.
With regard to scaling the DPP into community settings, Spring called for linkages that can promote both scale and reach by spanning the dimensions of care delivery, community services, and family and individual engagement and empowerment.
Steve Allender, professor of population health and founding director of the Global Obesity Centre at Deakin University, shared lessons from community-based obesity prevention trials in Australia. Allender provided an overview of the interventions, explaining that an early step is to engage a catalyst in each community, which is a person who is well connected and motivated to do the work. Researchers work with the catalyst to engage local leadership using systems science methodologies such as systems dynamics and group model building, he continued, to build a shared understanding of the complexity of the causes and outcomes of people being overweight or having obesity in each community. According to Allender, the researchers’ role is to provide the evidence base and help the community understand how it might inform community action, and then support the community’s momentum for change by building capacity, providing tools and resources, and providing monitoring, evaluation, and feedback.
Allender displayed a causal loop diagram constructed by community stakeholders as part of a whole-of-systems approach to address children who are overweight or have obesity in Campbelltown, Australia. The diagram is color coded to distinguish subsystems of interest such as physical activity, nutrition, education, and social determinants of health. One benefit of the causal loop diagram is that community stakeholders can begin to see where they fit into a response to obesity, Allender observed, and another benefit is that the diagram can be used to track actions corresponding to the topics that it illustrates. Allender highlighted Systems Thinking in Community Knowledge Exchange (STICKE), a software his group invented to help communities facilitate the process of creating causal loop diagrams and tracking actors and actions. He showed a portion of the Campbelltown causal loop diagram that visualized actors, actions, and the connections between each.
On average, the trials resulted in a 4 percent reduction in the prevalence of overweight and obesity over the first 2 years, Allender reported, along with significant improvement in health-related quality of life among children in the intervention communities. By using a range of systems science approaches, combining methodologies, and collaborating, researchers are able to better understand the complexity of childhood obesity, Allender said in closing.
Nicolaas Pronk, president of HealthPartners Institute, chief science officer at HealthPartners, Inc., and chair of the Roundtable on Obesity Solutions, ended the workshop with a preview of the roundtable’s September 2020 workshop. He indicated that, although both qualitative (e.g., mapping) and quantitative (e.g., modeling) systems science approaches are important, the workshop will focus on modeling various approaches that could guide future obesity research and action and discuss support structures for those approaches. ◆◆◆
DISCLAIMER: This Proceedings of a Workshop—in Brief was prepared by Emily A. Callahan as a factual summary of what occurred at the workshop. The statements made are those of the rapporteur or individual workshop participants and do not necessarily represent the views of all workshop participants, the planning committee, or the National Academies of Sciences, Engineering, and Medicine.
*The National Academies of Sciences, Engineering, and Medicine’s planning committees are solely responsible for organizing the workshop, identifying topics, and choosing speakers. The responsibility for this published Proceedings of a Workshop—in Brief rests with the rapporteur and the institution.
REVIEWERS: To ensure that it meets institutional standards for quality and objectivity, this Proceedings of a Workshop—in Brief was reviewed by Don Bradley, Duke University School of Medicine, and Barbara Schneeman, University of California, Davis. Lauren Shern, National Academies of Sciences, Engineering, and Medicine, served as the review coordinator.
SPONSORS: This workshop was partially supported by the Academy of Nutrition and Dietetics; Alliance for a Healthier Generation; American Academy of Pediatrics; American College of Sports Medicine; American Council on Exercise; American Society for Nutrition; Banner Health; Bipartisan Policy Center; Blue Shield of California Foundation; BlueCross BlueShield of North Carolina Foundation; The California Endowment; General Mills, Inc.; Greater Rochester Health Foundation; Intermountain Healthcare; The JPB Foundation; The Kresge Foundation; Mars, Inc.; National Recreation and Park Association; Nemours; Novo Nordisk; Obesity Action Coalition; The Obesity Society; Partnership for a Healthier America; Reinvestment Fund; Robert Wood Johnson Foundation; SHAPE America; Society of Behavioral Medicine; Wake Forest Baptist Medical Center; Walmart; WW International, Inc.; and YMCA of the USA.
For additional information regarding the workshop, visit nationalacademies.org/our-work/roundtable-on-obesity-solutions.
Suggested citation: National Academies of Sciences, Engineering, and Medicine. 2020. Integrating systems and sectors toward obesity prevention: Proceedings of a workshop—in brief. Washington, DC: The National Academies Press. https://doi.org/10.17226/25936.
Health and Medicine Division
Copyright 2020 by the National Academy of Sciences. All rights reserved.