Longer commute times and crowded apartments came with higher rates of COVID-19 transmission at the height of the pandemic in New York City, according to a recent study from Cornell University. The researchers, whose disciplines merged architecture, public health and engineering, looked at a range of environmental factors that could be tied to the spread of the coronavirus in urban areas and found commutes and crowding had the strongest correlation.

Timur Dogan, senior study author and assistant professor of architecture at Cornell, said he and his team wanted to try to identify data that would “link how we design cities with a kind of pandemic resilience.”

Simply being on a train or bus has never been a great predictor of catching the virus. A September 2020 study found no clear connection between coronavirus spread and riding the New York City subway and other transit systems worldwide, noting that cases fell in the five boroughs even as ridership gradually increased. The New York Times pointed out in May 2020 that ridership had rebounded in cities around the world where the pandemic had ebbed without mass transit being linked to any superspreader events.

The Cornell study made a similar finding, but by looking at a wider variety of potential trends, it revealed a correlation between infection rates and the average number of minutes spent commuting on public transit.

The study relied on U.S. Census data on New Yorkers’ commute times published before the start of the pandemic. It then compared those commuter times against case rates in different parts of the city across five phases of the city’s first wave. Phase No. 1 was the initial outbreak from January 20th to March 22nd. Phases No. 2 (March 23rd to April 5th) and No. 3 (April 6th–June 7th) represented the start of the lockdown period. Phase No. 4 marked the original reopening from June 8th–July 19th, and Phase No. 5 ran from then to August 26th, “when the reopening process concluded and the city entered the ‘new normal,’” according to the study authors.

During Phase 3, each extra minute added onto the average commute time for a given neighborhood correlated with a 0.8 increase in the daily new case rate (the number of new cases per 100,000 people). Add 30 minutes, and infections go up by 24 cases per 100,000. Tack on 45 minutes, and the rise is 36 cases per 100,000.

During the other periods analyzed, there was still a relationship between a neighborhood’s average commute time and case rate, but it was less significant.

Dogan speculated that those who had long commutes on subways or buses before the pandemic may have been more likely to continue taking public transportation during the outbreak because they lacked another alternative. They might have then risked greater exposure to the virus by spending a significant amount of time in close proximity to others during their commutes, he said.

Nicholas Bloom, a professor of urban policy and planning at Hunter College who was not involved in the study, expressed some skepticism that long commutes actually drove transmission. He noted that commute time could simply correlate with people coming from areas outside of Manhattan—where other factors such as housing contributed to high infection rates.

A person wearing a face mask on a quiet subway train in New York City, July 30th, 2021.

On the topic of living arrangements, plenty of data show a high rate of COVID-19 transmission within households last year. But the Cornell study distinguishes between crowding, which refers to people sharing rooms within a household, and density, which is defined by the number of buildings and people living in a given geographic area. A higher density of residential buildings in a neighborhood was associated with a lower COVID-19 case rate during the period studied from March through August 2020.

Dogan said that could be because neighborhoods with a larger population often have more services, including access to health care and supermarkets. People living in those areas don’t need to travel as far for food, medicine and other essentials—potentially limiting their overall exposure, he added.

Household crowding, by contrast, came with a higher case rate. “Even in high-density urban environments, there’s not necessarily crowding happening if it’s well designed,” Dogan said.

The study identified the Bronx as an early COVID-19 hotspot that had a high number of occupants per room. The authors acknowledged that additional research would be needed to assess the impact of other variables such as income level and age. But other COVID-19 studies reached similar conclusions, including one that determined that cities in Brazil with a high level of overcrowding had an average of 461 cases per 100,000 people in the first six months of the pandemic, compared with a nationwide average of 88 cases per 100,000.

“Crowding was a pretty well-established problem, even before the pandemic, in terms of public welfare,” said Bloom. “If you go back 100 years, the problem was tuberculosis and the spread of disease in the old tenements.” He said that’s part of why non-family members were banned from sharing apartments in the city’s public housing when it was established in the early 1930s.

Overcrowding remains a growing burden for New York City. In 2018, about 290,000 of the city’s households were overcrowded, a 17% increase from a decade prior, according to an April 2020 report from the city comptroller. It cited this as a factor behind New York becoming ground zero for the pandemic.

It’s not an easily fixed problem, Bloom said: “To basically have the government go in and do occupancy control [in non-public housing] is very controversial and would, of course, accelerate the homelessness problem in the city.”

The Cornell study authors concluded that one solution to the transportation problem could be to design a more decentralized city where jobs are not concentrated in one area but instead spread across different neighborhoods so that people have shorter commutes to work.