The community environment is the underlying determinant of population health outcomes. A place-based community is a complex system comprised of multiple, interacting and interrelated systems and parts. Parks, education, transportation, the local economy, social ties, networks and community agencies, groups and organizations, for example, are all community-level systems that interact together and change each other over time. Brita Roy, MD, MPH, MHS explains, “Within the community environment lie the roots of both the causes and the solutions of many poor health outcomes. Therefore, if we want to efficiently improve population health, given the multi-system complexity and limited available resources within a community, we should focus our energies on improving those facets of the community that can make the most impact.”
It is not as simple, however, as just picking the biggest influencer on public health and fixing that factor. This is because changing one system in a community might be insufficient on its own or might result in unintended consequences due to feedback loops within the system. For instance, if a low-income community has high rates of obesity and it is observed that there is no access to healthy foods, a logical conclusion may be that opening a grocery store with healthy foods will fix the problem. However, studies have shown this ‘solution’ may be insufficient, because even though the grocery store is opened, residents do not purchase healthy food there because of other competing, influential factors, such as the high price of the food or cultural and social eating preferences and habits. The presence of the new grocery store might also lead to further development of higher income real estate and result in gentrification – a potential unintended consequence of this intervention. Understanding and anticipating the effectiveness of and consequences of community-based interventions are critical steps prior to implementation given the great investment of resources needed to instigate changes in low-resource settings.
Dr. Roy’s current project uses system dynamics modeling, a type of simulation that shows how things change over time, to investigate how communities can address intractable public health issues, like community gun violence in New Haven issue. In 2011, the murder rate in New Haven was 26.2 per 100,000 persons, out-pacing Washington D.C. and Chicago. Since that time, local community members have been working together to try to curb the rates of gun violence through multiple approaches. Rates decreased for a period of time, but are now rising again. Each death affects the immediate family and loved ones of each victim as well as entire communities. Dr. Roy is partnering with these communities to use system dynamics modeling to inform and maximize their efforts. If done well, this model should be able to simulate the dynamic relationships among many factors within New Haven that influence gun violence, and its multiple feedback loops and interactions to determine where to focus energy and resources.
Dr. Roy is going a step further with this project. To understand the complexities of the New Haven community systems, she is working with a group of community leaders and stakeholders, including representatives from education, community-based organizations, and law enforcement, who have been meeting monthly since the Fall of 2019 to think about the many different factors in the community that influence the number of people affected by gun violence. Some of the major factors the group thought impacted the number of people affected by gun violence were access to guns, education and income, the presence of gangs and peer pressure influences to join gangs, feelings of safety among those who live in New Haven neighborhoods, drug and alcohol use, mental health access, incarceration rates and resources provided for people released from prison, neighborhood cohesion, and community-police relations. Once the stakeholder group had this list of the many different things influencing gun violence, they drew a ‘causal loop diagram,’ or concept map, that shows how they believe one thing feeds another, and then feeds back on different factors. The group thought each of the listed factors was connected directly or indirectly to all the other factors.
The stakeholder engagement process is a critical part of the work. It makes the causal loop diagram more unique to the local community context and closer to reality than if a researcher designed it on their own. No single person can understand the complexity of the various community factors influencing gun violence or other intractable public health issues. Each stakeholder brings their own unique lens, and a diverse group is essential. As a result, the causal loop diagram they come up with together is much richer, more accurate, and more complete.
Now, Dr. Roy is using that “causal loop diagram” to build a system dynamics model, a mathematical computer simulation, where they can simulate different scenarios over time. Once the model is built, the next step is to fit it to available data. Rates of gun violence over the past 10 years (homicides and assaults in particular) were gathered from the New Haven Police Department. For the many other variables in the model, Dr. Roy will use data from various sources: the DataHaven Community Wellbeing Survey (DCWS), which every 3 years measures neighborhood cohesion and rates of drug and alcohol use; the YNHH System data for rates of gunshot injuries; the City of New Haven for high school graduation rates; and Bureau of Labor Statistics data on income. Unfortunately, there are no available data on certain factors like peer pressure. For those variables, Dr. Roy creates a 0-100 scale and asked her community partners to gauge “What level of peer pressure do you think currently exists?” Sensitivity analyses are performed to test the accuracy of these estimates and the effects of using different initial values on these arbitrary scales.
Once the model is fit, the goal is to determine, “If we were running at equilibrium and could then ‘implement interventions’ by changing one or more of the variables, how would the rate of gun violence be changed over time?” More precisely, the group can assess questions such as:
- If we do nothing, what will happen to rates of gun violence over the next 50 years?
- If we increase rates of funding for mentorship by $XX/year, what would happen?
- If we increase high school graduation rates by 5%, what would happen over time?
- What if we increased rates of funding to mentorship programs and improved high school graduation rates?
Based on the simulations, the group will have a better understanding as to which intervention or set of interventions have a higher likelihood of success and they will seek funding to implement and evaluate them.
The current project includes only stakeholders and data from New Haven. However, the model that is developed could be adapted by other communities to include their own unique characteristics, strengths and weaknesses related to gun violence and its causes. For instance, one idiosyncrasy in New Haven is that gang membership is more fluid and they are not strictly neighborhood-based, as in many other cities, even those in cities about the same size as New Haven.
Future projects utilizing the expertise of community stakeholder groups to generate causal loop diagrams will examine how community factors influence rate of physical activity in both New Haven and three communities in California, and how social media posts influence behaviors related to COVID-19 prevention, which is not restricted to a place-based community. Dr. Roy cautions, “Using system dynamics modeling is not a prediction tool, but a way to compare different scenarios”. It is a method to help pick the best strategies to make change, and its richness lies in its ability to look at both intended and unintended consequences. It is also not a panacea, but it could help community leaders and policy makers make better decisions. Dr. Roy recalls a famous quote in this field that says, “All models are wrong, but some are useful.”