Skip to main content

Population, Transportation and Climate Impact A City’s Flu Season

The size and structure of cities affects how flu transmission rates respond to climate conditions. | Jordan Masse/Unsplashed via Oregon State University

If a city is densely populated and highly organized — including by efficient public transportation — its flu season will last longer, reports a new study. The city's makeup, the study says, can help the flu virus reach new hosts even early and late in the flu season, when climate conditions aren't as favorable for the virus' spread.

The results, published in the 5 October issue of Science, suggest that large, dense, well-organized cities endure longer periods of flu transmission while smaller cities experience shorter, but more explosive, spread.

The results have important public health consequences, said study co-author Cecile Viboud, an epidemiologist in the Division of International Epidemiology and Population Studies at the Fogarty International Center of the U.S. National Institutes of Health.

"In cities like Miami or New York, influenza epidemics are more diffuse, and the pressure on the healthcare system is more spread out during the year," she explained. "In contrast, in smaller cities, more intense outbreaks may contribute to overloading the healthcare system, especially around the peak of the epidemic and during severe influenza seasons."

"Our study provides a novel explanation for how focused epidemics are in U.S. cities in terms of underlying population mixing," said co-author Bryan Grenfell, professor of ecology, evolutionary biology and public affairs at Princeton University.

The report represents an important step towards improving outbreak predictions for influenza, which sickens millions in the U.S. annually, and kills tens of thousands. The 2017-2018 winter flu season in the U.S was particularly long in many cities, challenging regional healthcare systems.

Flu is not endemic year-round in the U.S. The virus is reintroduced every winter, which is why researchers use a global flu surveillance system to monitor it and to inform vaccine composition, developing a new flu vaccine annually. Flu season typically starts in November and ends in February or March, though there is notable variability between flu start time, particularly between cities.

"We and others have found that epidemics tend to have earlier onsets in southern U.S. cities," said Viboud, "then diffusing northwards, with occasional long-range jumps."

Various factors — including those related to individual immunity, social crowding, and shifting climatic conditions — have been proposed to explain the size and shape of seasonal influenza epidemics, but inferring the relative importance of each factor is challenging. Though previous studies have suggested a key role for climate, the same work has also revealed variations in flu season patterns from city to city that could not be explained by climate alone.

To better resolve this, study lead author Benjamin Dalziel, population biologist in the department of integrative biology and mathematics at Oregon State University, said it was critical to consider how "transmission efficiency" in urban centers — the way flu can spread easily because people are routinely tightly connected — interacts with a fluctuating climate.

He and his colleagues used U.S. census data that captured people's daytime and nighttime movements in 603 U.S. cities of varying size and transportation layout. They combined this information with weekly flu incidence data from the same locations, as represented by insurance data on U.S. patients visiting doctors' offices with influenza-like illness between 2002 and 2008.

Following their analyses, they report that influenza spreads differently in urban centers with pockets of high population density connected by organized movement like that on metros or highly used bus systems that brings people together often. In such metropolises, flu cases were more diffuse through the winter months — including early and late in the season when the weather is not optimal for flu transmission. By contrast, in smaller cities flu cases were more often tightly grouped in a short period during peak season.

The different patterns of seasonal flu epidemic the authors observed by geographic region — like the tendency for flu cases to be more tightly grouped in the southeast U.S., or to be more spread throughout the season in New York City and Los Angeles — persisted year after year, they say.

Hypothesizing that these patterns might be caused in part by responses to climate factors, the researchers evaluated the influence of a key weather metric called specific humidity. As it decreases, the moisture droplets that flu-infected people cough out remain viable for a longer time in the air outside their host. The researchers' modeling revealed how local humidity patterns in key cities contributed to differences in flu patterns there.

Humidity shifts played a bigger role in flu transmission in cities with smaller population sizes and less organized movement patterns, the authors said. In those locations, flu transmission due to high population density is lessened, Dalziel said, such that changes in climate that are good for flu spread, like decreased humidity, matter more. In these cities where humidity played a big role, flu cases tended to be grouped in peak season rather than diffuse over many months. These cities also experienced shifts in flu epidemic as humidity shifted, the authors said.

"By including urbanization as well as climate in our analysis," Dalziel said, "we were able to explain variation in epidemic flu patterns that could not be explained by urbanization or climate alone."

The researchers' approach highlights how mathematical transmission models can disentangle important drivers of influenza epidemics, which may not be obvious based on other methods like statistical analysis.

The work is "important for policy makers," writes Jacco Wallinga, a researcher at the Center for Infectious Disease Control at the National Institute for Public Health and the Environment in Bilthoven, Netherland, in a related Perspective. "[It] would indicate that metropolitan areas should focus on reducing influenza spread, whereas small towns should focus on reducing harm."

Viboud said the findings reported here eventually may inform vaccination efforts in geography-specific ways.

"Locations that experience earlier flu activity could be tempted to vaccinate earlier, if these patterns were systematic across years," she said. "However, this has to be balanced against waning of vaccine-induced immunity within the season. In other words, if you vaccinate too early, you lose effectiveness by the end of the season."

More immediately, Viboud said, "our work could help identify sentinel cities for influenza surveillance. Strengthening surveillance in cities that experience earlier flu cases in the season due to particular city structure or humidity conditions may help forecast the severity of epidemics on a regional or national scale, and guide interventions."

Dalziel would like to expand this work to look at the role of urbanization in global patterns of influenza circulation. "I would also to like to consider how city size and structure might interact with vaccination strategies, and if or how our results might be useful in developing a universal influenza vaccine," he added.

[Associated image: Frederik Ranninger/Flickr]