At the beginning of the pandemic, risky behavior dropped about 40% in Vermont, considerably below the rest of the Northeast, where risky behavior dropped about 30%.
The findings are from a new study by Beth Israel Deaconess Medical Center in Boston, which used GPS data from cellphones to track foot traffic at high-risk businesses across the Northeast, both before and during the pandemic. It found that Vermont was at a significant advantage for slowing the spread of the COVID-19 virus in the early days of the pandemic.
The GPS data identified two key factors — visitors per square foot and average duration of visits — as markers for where transmission of the virus was probably high, places such as gyms, salons, restaurants and bars.
During April, across the Northeast, traffic to those kinds of locations dropped significantly, according to the study, though it has been steadily increasing since the end of April, as businesses have reopened.
On average, traffic at high-risk businesses has already increased between 10 and 20 percentage points, and is now approaching pre-pandemic levels, the study shows.
However, Vermont’s levels of high-risk traffic patterns have continued to be notably lower than in nearby states.
“I definitely think population density is a really important factor that matters, both for how much people are frequenting businesses and how crowded businesses are,” said Ashley O’Donoghue, the researcher who headed the study. “And the trends, between Massachustts and Vermont especially, have been very different.”
O’Donoghue said her research didn’t go in-depth into why certain states might have had more risky traffic patterns than others. But she says key factors likely included public policy, people’s adherence and attitudes toward regulations and physical characteristics of each state.
Mike Pieciak, commissioner of the Department of Financial Regulation, said a big reason Vermont has been able to keep its numbers low — even after the period of time studied by the researchers — is that Vermont took the virus so seriously in the beginning. When the economy began to slowly reopen, Vermont was starting from a much safer point than some of its neighbors.
“The policy measures that we had in place were more significant than probably any other state in New England,” he said.
Pieciak said the study reinforced all the things the state has been learning from mobility data since March. He said O’Donoghue’s research was able to dig more deeply into what places are higher risk, with less square footage, or places that are known to be bars or restaurants, but still told the same story as the state’s internal data.
Another key factor in Vermont’s low numbers, Pieciak said, is that Vermonters live in small towns and small cities to a much higher degree than other New Englanders. He said when officials tell people to wear masks and stay home — not for themselves, but for their communities — Vermonters are probably a little bit more familiar with the people they’re looking out for.
“I think that was an easier message to get across to Vermonters who have that collective starting point for how we live in our state,” he said.
O’Donoghue said the kinds of public policy decisions that Pieciak and others have been making are why this kind of research is so important.
“Obviously, this is something that none of us have had to deal with before, and people have had to make a lot of decisions,” she said. “Is 25% capacity safe? Is 50% safe? These are really difficult questions that people are trying to figure out, and having the data that people are then able to link to actual transmission is really useful.”
The study used anonymized GPS data from people’s smartphones. O’Donoghue said it’s something any cellphone user can opt into, by turning on location services on their phone.
She said only about 10% of people activate location services, but even that small slice of the population creates enough data to discern broad trends.
The study used location data from January to June of 2020, though O’Donoghue said traffic patterns continue to be tracked. In recent months, she said, there’s been a worrying trend, with high-risk traffic patterns steadily rising.
“Later in the summer, as some people began to experience ‘pandemic fatigue,’ people became a little less risk-averse,” she said. “Someone who had once gotten takeout might have decided to eat in a restaurant, that kind of thing.”
The research, O’Donoghue said, can’t paint an entire picture of the virus’s transmission. She said factors like mask adherence and ventilation that can’t be seen through this data can make a big difference in the virus’s spread.
However, she said, the hope is that their work will give other scientists and public policy leaders a key piece of information about the virus spread.
“This is the first step toward figuring out how people are behaving,” O’Donoghue said. “And if we’re going to pull back on some reopening measures, we want to know what worked and what didn’t.”