While stay-at-home orders have become a part of daily life, people continue to struggle to shelter in place — and for different reasons. Limited access to doctors, public parks and recreation facilities, and food insecurity are just a few examples, according to Nabarun Dasgupta, an epidemiologist at the University of North Carolina at Chapel Hill.
In a recent study, Dasgupta analyzed location data from 65 million mobile devices to understand the public health impact of social distancing at a county level across the United States. The idea for the project evolved from a New York Times article featuring an interactive map that used the same data to show how well people of different political parties practiced social distancing measures.
Dasgupta and his team hypothesized that healthier and wealthier counties would have greater success in adopting social distancing procedures. They based this on the “healthy user effect,” an accepted research hypothesis that suggests that people who have the means to participate in preventative health behavior – like getting a flu shot or not smoking – are more likely to participate in others.
For the study, they compared variations in social distancing intensity to health care, economic, structural, and demographic factors. Overall, people within counties that successfully sheltered in place had higher incomes, more public spaces for physical activity, better access to food, and more primary care providers — notably, 50 percent more — than people in counties with poor compliance of shelter-in-place measures.
“It’s no surprise that privilege allows some people to shelter in place more easily,” Dasgupta says. “But we didn’t expect to see so much variation across these metrics. It seems like we’re missing part of the picture if we’re telling people to stay at home, but not paying attention to how good or bad their lives and circumstances might be.”
Dasgupta has also looked at mobile location data specific to North Carolina, where human movement has increased — a trend that has been the focus of national media coverage, according to Dasgupta.
“If you analyze this data as someone who lives here and understands what the underlying county structures are, you come away with a different interpretation,” he says.
For example, within the first few weeks of the outbreak within North Carolina, reports stated that a handful of counties including Watauga, Orange, and Guilford showed sudden increases in vehicle movement. Not only did it appear that North Carolina was not slowing down enough, it suggested some places were actually moving more — and this put people into panic mode, according to Dasgupta.
Those counties, though, all house major universities. And this was during spring break, when students were told to come back to campus, pack up their belongings, and then head home for the rest of the semester. By the end of the month, Governor Cooper issued an official stay-at-home order.
“What was skewing the North Carolina data was people actually being compliant with the stay-at-home guidance by getting off campus,” Dasgupta says. “You don’t see that kind of local color and context when you look at data dashboards.”
Mobile data analysis like this could help identify gaps in access to treatments or vaccines for COVID-19 once they are developed.
Going forward, Dasgupta and his team are working to make this information accessible to anyone, for free. This kind of data can be used for a variety of things, Dasgupta points out, from changes in air pollution levels to traffic movement.
Nabarun Dasgupta is a senior research scientist at the UNC Injury Prevention Research Center.