Cellphone Data Link 80 Percent Of Early Covid Infections To Restaurants, Gyms

Analysis of cellphone data cross-referenced with Covid infection data show restaurants were four times as risky as other venues - and that those in lower income neighborhoods were worse because they were more crowded than in other areas. JL

Benedict Carey reports in the New York Times:

Restaurants, gyms, cafes and other crowded indoor venues accounted for some 8 in 10 new infections in the early months of the U.S. coronavirus epidemic. “Restaurants were by far the riskiest places, four times riskier than gyms and coffee shops, followed by hotels” in terms of new infections. The study, which used cellphone mobility data from 10 U.S. cities from March to May, provides an explanation for why many low-income neighborhoods were hardest hit. The public venues in those communities were more crowded than in more affluent ones, and residents were more mobile on average, likely because of work demands

Restaurants, gyms, cafes and other crowded indoor venues accounted for some 8 in 10 new infections in the early months of the U.S. coronavirus epidemic, according to a new analysis that could help officials around the world now considering curfews, partial lockdowns and other measures in response to renewed outbreaks.

The study, which used cellphone mobility data from 10 U.S. cities from March to May, also provides an explanation for why many low-income neighborhoods were hardest hit. The public venues in those communities were more crowded than in more affluent ones, and residents were more mobile on average, likely because of work demands, the authors said in the research published in the journal Nature on Tuesday.

The data came from the metro areas of Atlanta, Chicago, Dallas, Houston, Los Angeles, Miami, New York, Philadelphia, San Francisco and Washington, D.C.

Infectious disease models had provided similar estimates of the risk posed by crowded indoor spaces, going back to February; all such models are subject to uncertainties, due largely to unforeseen changes in community behavior. The new analysis provides more precise estimates for how much each kind of venue contributed to urban outbreaks, by tracking hourly movements and taking into account the reductions in mobility from lockdown restrictions or other changes that occurred during those first crucial months. It did not model infection in schools or office workplaces.

“Restaurants were by far the riskiest places, about four times riskier than gyms and coffee shops, followed by hotels” in terms of new infections, said Jure Leskovec, a computer scientist at Stanford University and senior author of the new report, in a conference call with reporters. The study was a collaboration between scientists at Stanford, Northwestern University, Microsoft Research and the Chan Zuckerberg Biohub.

Public officials across Europe and in parts of the United States, including Gov. Phil Murphy of New Jersey, have begun to institute partial closures of restaurants and bars, or limited indoor hours, as new infections have surged in recent weeks. In New York City, a spike in virus cases threatens the city’s recovery and could mean “a lot more restrictions,” Mayor Bill de Blasio said on Monday.

These measures are especially important in lower income areas, the new study suggests. Infections exploded in many such communities last spring, and the new model provides one likely explanation: Local venues tend to be more crowded than elsewhere.

The researchers looked closely at grocery stores, to understand differences between high and low income communities. In eight of the ten cities, transmission rates were twice as high in low as in higher income areas. The mobility data pointed at one reason: Grocers in low-income neighborhoods had almost 60 percent more people per square foot; shoppers tended to stay there longer as well.

And residents are apparently less able to shelter at home.

“We think a big reason for that is that essential workers had to be on the job, they weren’t working from home,” said Serina Chang, a co-author also at Stanford.

In the analysis, the research team mapped the hourly mobility of some 98 million people to and from indoor public spaces, like grocery stores, churches, hotels and bars. It calculated the traffic to each venue over the course of a day, how long people stayed on average, and the place’s square footage. Given a background infection rate, the researchers then ran the model forward — “hit play,” said Dr. Chang, and watched how infections spread and where, using standard infectious disease assumptions.

The estimates lined up well with what actually happened in those cities — a crucial reality check, since from March 1 to May 2, communities’ behavior changed drastically, because of stay-at-home orders.

In Chicago, for instance, new infections occurring at just 10 percent of indoor venues accounted for 85 percent of the predicted infections. Reopening just full-service restaurants, the analysis found, would have resulted in an additional 600,000 new infections by the end of May.

By focusing on indoor public venues, the researchers could also model the impact of partial restrictions. Limiting restaurant occupancy to one-fifth of capacity, for example, would reduce new infections there by 80 percent, while preserving some 60 percent of customers.

“These are important trade-offs,” Dr. Leskovec said. “Our work highlights that it does not have to be all or nothing,” when implementing restrictions.