Summer forecast for COVID-19: Some relief, but not enough

As spring turns to summer, will the warmer weather slow the spread of COVID-19, as many have hoped?

It’s a difficult question to answer, because research on the impact of factors like temperature and humidity on the virus’s transmission rate has so far been mixed and inconclusive. And that has left policymakers with fewer tools to help manage the crisis as the seasons change. 

A new study co-authored by an MIT Sloan professor may help. It suggests that summer weather impedes the transmission of the virus, although not enough to curtail the epidemic, and only under certain conditions.

The study shows that warmer, more humid weather with medium levels of ultraviolet index could dampen transmission rates by 30% to 40%. However, when multiple relevant factors are included, weather impact is modest and not enough to stop transmission of the coronavirus absent other interventions.

Hazhir Rahmandad,an associate professor of system dynamics at MIT Sloan, and colleagues from Harvard, UConn, and Virginia Tech built a statistical estimation model that used data on COVID-19 transmission, location, weather, pollution, population density, and other information from more than 3,700 locations around the world between mid-December 2019 and mid-April 2020.

One of the biggest challenges in figuring the impact of weather — and a potential reason why previous research has produced somewhat inconsistent results — is detection delay; that is, the time between individuals’ exposure to the virus and their inclusion in official case counts. That period can range from five to 17 days, making it difficult to pinpoint the exact day, and thus the specific weather, when a person contracted the virus, the researchers wrote.

Correcting for this delay was an important part of finding more reliable estimates for the relationship between weather and COVID-19 transmission, Rahmandad said.

Assessing the estimated relationships was difficult because the true relationships were unknown. So the team validated its statistical method using a separate simulation model, and used that simulation to generate synthetic data of multiple epidemic spreads that had known weather-transmission relationships.

Applying their estimation method on this synthetic data allowed the team to determine if the method was capable of finding the relationship that was built into the simulation. Iterating on this process, they fine-tuned the statistical method to give reliable estimates under a range of conditions expected to apply to real world COVID-19 data.

Heat, humidity, and . . .  moonlight?

The researchers then used the best identified estimation method to find how different weather variables impacted COVID-19 transmission, while controlling for variations in location-specific factors, from population density and cultural norms to differences in public policy and social distancing. Finally, they used these estimates to project the relative weather-related risk of COVID-19 transmission across the world and in large cities.

They found that transmission rates may fall by 30% to 40% in warmer, humid weather. Specifically:

• As the temperature rises above 77 degrees Fahrenheit, transmission falls at a rate of about 2.0% per degree Fahrenheit.
• Modest increases in ultraviolet index (up to a UV index of 7) are associated with lower transmission rates, but beyond that the effect reverses.
• As humidity rises, the transmission rate falls, but to a smaller degree.
• The closer to sea level, the higher the transmission rate, also to a smaller degree.
• Precipitation, wind speed, and concentrations of ozone and sulfur dioxide, all had positive but small associations with higher transmission.
• Interestingly, among several other factors included in regressions (from sunlight to snowfall, particulate matter and nitrogen dioxide pollutants), moonlight showed up as possibly meaningful. A 1% increase in moonlight illumination was associated with a 0.25% decrease in transmission. The researchers put that information in an appendix, rather than the paper’s main findings, because they had no explanation for it. “It’s weird. We tried to explain it away, but it seems robust,” said Rahmandad. “So maybe someone should look into that more.”

Reproduction numbers matter

Relatively, the magnitude of the effect of temperature on transmission is modest.

Here’s why: To stop an epidemic’s spread by “flattening the curve,” a society must keep what’s called the reproduction number — the number of people infected by one person carrying the virus — to 1 or lower. For COVID-19, the estimated reproduction number is between 2 and 3, when limiting factors like social distancing are absent. Rahmandad and colleagues calculated the impact of temperature on the reproduction number to not exceed 0.7 in most places. Doing the math (2 x 0.7 or 3 x 0.7) does not produce a number less than 1.

“Even though high temperatures and humidity can moderately reduce the transmission rates of coronavirus, the pandemic is not likely to diminish solely due to summer weather,” said Rahmandad.

However, this understanding of how weather impacts transmission has important implications for managing the pandemic because it can inform public policy decisions about whether, when, and where to loosen policies like social distancing, Rahmandad said. So in the summertime, a warm, humid state, such as Florida, might be able to relax restrictions more than warm, dry Arizona. And Arizona might have fewer restrictions than chillier Maine.

But Rahmandad advised caution if governments do choose to loosen restrictions. “The downside is that people can get used to that new level of [higher] activity in the summer, then when fall comes you reverse that or otherwise you will see increasing spread,” he said. “So it’s a double-edged sword.”

The biggest factor in the team’s research that influenced the reproduction number was a given locale’s public policy response to the epidemic, he says. “Once the epidemic is known in a location, on average, people bring it down by 4% per day by reducing their social interactions and maintaining physical distancing,” Rahmandad said. “That effect is a much stronger predictor of transmission rates than weather.”

“Policymakers and the public should remain vigilant in their responses to the health emergency, rather than assuming that the summer climate naturally prevents transmission,” Rahmandad said. “At best, weather plays only a secondary role in the control of the pandemic.”

Further reading: Will hot weather kill the coronavirus where you live?