The first part of what may be the first epidemiologic text ever written begins like so: “Whoever wishes to investigate medicine properly, should proceed thus: in the first place to consider the seasons of the year.”
The book is On Airs, Waters, and Places, written by Hippocrates around 400 B.C. Two and a half millennia later, the Northern Hemisphere is staring down its coming season of the year with growing apprehension. America’s grimmest phase of the coronavirus pandemic so far occurred from November 2020 to February 2021. Now the calendar has turned to a new November, and even though the majority of Americans are fully vaccinated against COVID-19, cases are once again, horrifyingly, on the rise.
If Hippocrates was right, we could be doomed to repeat the sickness and death that defined last winter. To be fair, Hippocrates also thought that among the most important factors in anyone’s health was their balance of black and yellow bile. But evidence is piling up that COVID really is a seasonal disease, surging with the weather and the annual rhythms of human life. If that’s the case, then understanding those seasonal patterns could help us predict where the virus is headed next, and address its attacks in advance.
The seasonal-COVID hypothesis, and its promised benefits for pandemic planning, have been around nearly as long as the disease itself. Way back in February 2020, President Donald Trump predicted that in April, “when it gets a little warmer,” the coronavirus would “miraculously” disappear. That clearly didn’t happen, but evidence of seasonality, the thinking went, might show up in summer, when things got really warm. In fact, by mid-July, the country saw its highest case rates yet—and then the massive winter surge easily surpassed it. The virus came in waves, but the waves were hitting all year round.
COVID’s seasonality hadn’t been disproved; to know for sure, we’d need to wait and see what happened next. Now we have nearly two years’ worth of data—from eight full seasons of pandemic—to pick apart for clues, and may well be closing in on an answer. We also know something we didn’t in spring 2020: In all likelihood, we will not eradicate COVID. That makes it all the more important that we know how cases might ebb and flow with weather in the months and years to come.
Anyone who lives in a temperate climate has an intuitive understanding of seasonal disease. The most canonical example is the common cold—just look at what it’s named. But infecting more people in cold weather is far from the only once-a-year cycle a disease can settle into. Lyme disease peaks in the summer. Polio was historically a summer sickness. Even genital herpes tends to spike around the spring and summer in the United States. The same disease can also show different patterns in different places. Americans are used to a winter flu season, but in Bangladesh, flu cases spike during the monsoon season, which runs from May to September and is the warmest part of the year. As one public-health researcher argued in a 2018 paper, seasonal cycles “may be a ubiquitous feature of human infectious diseases.”
Some seasonal disease patterns are a result of how efficiently a particular pathogen invades our bodies in particular weather. Flu, for example, is much better at surviving and traveling between humans in dry air. Early in the pandemic, a group of researchers led by Tamma Carleton, an environmental economist now at UC Santa Barbara, checked to see how COVID fared in different weather conditions around the world. Their study didn’t find much of a role for temperature or humidity, but suggested that case rates would go up in a particular area during periods of lower UV exposure. Since then, the coronavirus has been shown to die off in the presence of UV rays with the same wavelength as sunlight. (That, in combination with airflow, could help explain why the virus tends to spread much less outdoors.)
But Carleton’s study also showed that the influence of sunlight was minimal in comparison with that of shifting human behavior. “How we interact with each other, where we interact with each other, changes so much with different climate conditions,” she told me. She suspects that her study picked up on both the direct virucidal effects of sunlight and the fact that people might be more inclined to gather inside when it’s crummy out. Both would contribute to the seasonality of COVID, she said, but, “I’m not sure I’m that hopeful in ever disentangling them.”
As Carleton and her colleagues did their work in the spring of 2020, they could look only at case rates over periods of weeks. Subsequent research would have access to many months’ worth of data. In July 2021, a team from the University of Pittsburgh put out a study (which has not yet been peer-reviewed) showing that differentiating between regions in North America reveals a much stronger seasonal pattern. “You don’t get a clear signal just from analyzing the United States as a whole,” Hawre Jalal, one of the authors of that study, told me. That could be because warmth doesn’t mean the same thing for all Americans. Those who live in cooler parts of the country can spend time outside more easily in July than January, while the opposite is true for residents of the hottest parts of the South. (No one has yet empirically proved a link between air-conditioning weather and indoor transmission of the virus.)
By sifting for seasonal patterns across individual states, Jalal and his collaborators found very robust results. They argue that the calendar of COVID in North America has already taken shape, in the form of three repeating waves like the ones that swept the continent in 2020: one starting in New England and eastern Canada in the spring, the second traveling north from Mexico over the summer, and the third emanating in all directions from the Dakotas during the fall. In keeping with that idea, their paper predicted a summer 2021 wave in the South, and a fall 2021 wave in the north-central states—which is more or less exactly what happened.
This three-peaked seasonality, if it’s real, would seem to make COVID an outlier, at least compared with single-season diseases like the flu. But if COVID really is driven more by seasonal changes than factors such as masking and vaccination rates, no community should expect to see a surge more than once a year. The disease would still behave like the flu on a local level, in the sense that each place would see one peak season every year—even while the country overall had three.
This pattern may sharpen in the next few years. David Fisman, an epidemiologist at the University of Toronto, told me that the patterning of past pandemics has tended to follow a sort of script: chaos, then seasonality, then less-destructive chaos. When a pandemic first arrives, virtually everyone on Earth is vulnerable, so the pathogen rips through populations like wildfire. Then, as more people develop immunity through vaccination or infection, the fire needs more help to find new fuel, and seasonal influences become more apparent. Finally, once the overwhelming majority of the population is immune, those same influences could become so subdued as to be invisible.
For a lot of diseases, Fisman said, the effective reproduction number—that is, the number of people to whom each infected person passes a disease, on average—hovers below one during the off-season. Then, the kids go back to school, or the deer-tick nymphs emerge into the world, or the humidity drops, and the disease suddenly has the upper hand. The reproduction number jumps above one for a few months, before dropping again. Transmissibility was elevated during the early months of the pandemic, and again during the Delta-variant wave, which could have pushed the country back toward the initial-chaos phase and blunted any seasonal influence on COVID. Maybe in the absence of Delta, we’d have realized that transmission is even more seasonal than it looks right now.
At this point, even initial skeptics agree that COVID rates are varying with the seasons. Ben Zaitchik, an Earth scientist at Johns Hopkins University who co-chairs the World Meteorological Organization’s COVID-19 Research Task Team, once found seasonality claims to be weak. In February, he co-wrote a review of 43 studies of the topic (including Carleton’s) from early in the pandemic. Researchers simply didn’t have enough data in the first several months of 2020 to find strong patterns, he told me. Testing was inconsistent. Many teams, unable to compare the cold and warm or rainy and dry seasons in particular places, compared one region’s cold with another’s heat—say, winter in Italy with summer in Australia—which doesn’t tell you much about what will happen once Italy gets hot and Australia gets cold. But the data have since improved enough that Zaitchik feels confident saying that weather influences COVID transmission in a statistically significant way.
He’s not as convinced that this influence matters for public health. “COVID-19 has proven beyond a doubt that it can create hugely deadly outbreaks anywhere in the world at any time of the year. And that’s still true,” Zaitchik said. Until we see the end of countercyclical outbreaks—until Montana stops having August surges, and Florida’s cases stay flat in February—arguing that seasonality is a dominant driver of the disease will be difficult. And if it isn’t yet the dominant pattern, staking a public-health response on it could backfire. “I think that a lot of responsible people in the decision-making space kind of say, ‘I don’t want to talk about seasonality now, because I’m not ready to, because I know that there are bigger risk factors to be taken into account,’” Zaitchik said. Telling northerners they can let their guard down in the summer, and southerners they can party like it’s 2019 over the winter, could have disastrous consequences.
At the same time, avoiding all discussion of seasonality could mean missing opportunities to fight COVID smarter, not harder. Donald Burke, one of Jalal’s co-authors, suggested that public-health officials could plan to deploy extra anti-COVID strategies in times when and places where the virus is at a disadvantage, given that beating back a disease is much easier when it’s not circulating widely. Jalal said that the United States could direct resources such as health-care workers and PPE to areas that are likely to see a wave before it arrives, rather than reacting to it once it’s already half-crashed.
If these sorts of ideas haven’t gotten much traction, Jalal said, it may be because some researchers are underestimating the importance of seasonality. He warns against concentrating too much on the global or national picture, where the many waves in several seasons make the pattern less obvious. Burke suggested that wishful thinking could also be to blame: “I think most people want to believe that we have more power over the course of the epidemic,” he said.
To acknowledge a strong seasonal influence might feel like admitting defeat: If Louisiana is going to face devastating case rates every summer, and Minnesota will fall prey to a winter surge like clockwork, how much can we really do? But a regular pattern doesn’t have to mean inevitable suffering. Pandemic-fighting policies can take strategic account of seasonality; they’ve done so before. “Having made these investigations, and knowing beforehand the seasons,” Hippocrates wrote, a doctor “must be acquainted with each particular, and must succeed in the preservation of health, and be by no means unsuccessful in the practice of his art.”