Where We Live: Neighborhoods

Where We Live: Neighborhoods

Untangling the multiple and complex ways that neighborhoods affect our health, for better and worse.

When a farmer's market opens in a neighborhood, things change. It's not just the sudden availability of fresh fruits and vegetables, although that;s important, says School of Public Health epidemiologist Ana Diez Roux. It's the way the farmer's market changes people's behavior. Suddenly there are new opportunities for social interaction, new connections between neighborhood residents, new ways for health information to be transmitted—and all of these things have physical- and mental-health consequences.

"All these connections between the social and physical environment potentiate each other and can be health-enhancing in ways you may not think about," says Diez Roux, who directs a multimillion dollar longitudinal study—the MESA Neighborhood Study—aimed at clarifying the complex and interconnected ways that built environments affect behaviors and other risk factors for chronic disease, especially cardiovascular disease. Funded by the National Heart Lung and Blood Institute of the National Institutes of Health as part of a larger parent study, the Multi-Ethnic Study of Atherosclerosis (MESA), the MESA Neighborhood Study is following approximately 6,800 people in six different U.S. locations: Chicago, Baltimore, Minneapolis/St. Paul, Los Angeles, New York City, and Forsythe County, North Carolina. The study began in 2000 and will continue until at least 2015.

Diez Roux says that previous studies of neighborhoods relied almost exclusively on census data, which "isn't very good because it doesn't tell you about factors that are important." By contrast, she and her colleagues in the MESA study are collecting information from a range of sources, including surveys, geographic information systems, and existing data on the location of different types of resources. They're looking at such factors as physical resources, social cohesion, air pollution, and environmental stressors. Past studies have also tended to be cross-sectional—that is, they focused on a particular place at a particular moment in time. But because the MESA study is longitudinal, it allows for greater complexity. "We can see whether something that changes in the environment is accompanied by a change in people's behavior or other aspects of their health," Diez Roux says.

In addition to traditional epidemiologic approaches, the MESA Neighborhood Study is employing novel methods from the field of complex systems in order to better understand the relationships between environments and chronic disease. This approach allows researchers to create dynamic—rather than statistical—models that demonstrate the intricate ways neighborhood and individuals affect each other over time and in turn affect human health.

The ultimate aim of the study is to foster awareness among both scientists and policymakers that changes within communities can have significant health consequences. This could lead both to health interventions and to changes in the way communities are created. "So when developers want to develop an area, for instance, they could be required to do an assessment of health consequences," Diez Roux says. "In order to do that, though, you need the kinds of data that we generate. Otherwise you're just speculating."

The MESA study could also have an impact on health care reform implementation. "If part of the goal of reform is to reduce cost by preventing chronic diseases," says Diez Roux, "we have to think seriously about what environmental factors we can act on.


For years, researchers studying chronic-disease prevention focused very much on individual-centered approaches such as health education, treatment, screening programs, and other interventions targeted at individuals. But more recently, the study of chronic disease has returned to the roots of epidemiology, "which are very environmental and holistic," says Diez Roux. She and her colleagues in the MESA Neighborhood Study are trying to understand which aspects of certain, mostly residential, environments may affect the distribution of cardiovascular risk factors. They're also studying the rate at which different risk factors develop over time, with the idea that if they can identify critical environmental factors, they can design more effective intervention strategies. Diez Roux says the team's research may also explain why many past strategies to prevent cardiovascular disease have not worked very well. "Changing people's behaviors is very difficult. One possibility is that many of these interventions didn't look at the context in which people were living."

Social Cohesion

Throughout the U.S., people are sorted into environments based on attributes of class, education, race, and ethnicity, and there are clear patterns associated with these attributes. When it comes to access to healthy foods and recreational facilities, for instance, and to safe streets and socially cohesive neighborhoods, there are stark differences among racial, ethnic, and socioeconomic groups. These differences, in turn, have an impact on health-related behaviors, including conditions like hypertension and insulin resistance—an early marker of diabetes. Recent findings show that summary measures of healthy food and physical environment scores are related to the incidence of Type 2 diabetes. By using survey data to describe differences in social cohesion across neighborhoods, the MESA team is endeavoring to understand such factors as social connections—or their absence—within neighborhoods, and how these affect health.


Whenever possible, members of the MESA team use GIS technology to determine whether sidewalks exist in a given location, and if so, the extent to which they exist, or what's called "street connectivity." They plot the mixes of land use in given areas. The chief predictors of walking activity are street connectivity and land-use mix. Diez Roux and her colleagues are also measuring access to public transportation and calculating the distance between people's homes and bus, subway, and train stations. The researchers then devise models to demonstrate how behaviors might change if public transportation were to be improved, or it were to become more expensive to own and drive a car. Would people walk more? Would cardiovascular disease rates plummet? They hope their models will help policymakers and city planners think not just about neighborhood resources—or their absence—but about the human processes and behaviors these engender.


A major component of the MESA Neighborhood Study is what researchers call "the local food environment," which they measure through survey questionnaires as well as data on the location of different types of stores and the kinds of food they offer. One MESA researcher audited all of the food stores in a series of Baltimore neighborhoods, using a systematic instrument to assess the availability of healthy foods. He then mapped the stores, each with a score indicating the availability of healthy foods, and overlay this with a map showing the distribution of race in the city. In wealthier, whiter suburbs, supermarkets had high availability scores, but inner-city neighborhoods told a different story. In addition, people who lived in areas with higher healthy food scores had better diets, even after accounting for other factors related to diet.

In order to better understand the dynamic processes through which environments affect health, the MESA team, led by epidemiologist and SPH alumna Amy Auchincloss of the Drexel University School of Public Health, has developed a dynamic "agent-based" model for food-related behavior, showing what can happen to people's shopping and buying behaviors when healthy foods are subsidized, or certain kinds of food stores are promoted in particular neighborhoods.


In order to measure neighborhood safety, Diez Roux and her team are using survey data and, "when we can get it," crime data. They're also taking daily measures of the stress hormone cortisol in neighborhood residents. Preliminary findings show that neighborhoods with higher levels of violence have a different daily cortisol profile than safer neighborhoods. Typically, says Diez Roux, cortisol "goes up when you wake up and gradually declines over the day. But our preliminary results suggest that violence is related to a less steep decline of cortisol." Higher levels of cortisol have been linked to several chronic health conditions, including diabetes and hypertension.