Both heat and cold can kill.
But cold is far more deadly. For every death linked to heat, nine are tied to cold.
Will global warming make temperature less deadly?
The scientific paper published in the June 2021 issue of the journal Nature Climate Change was alarming. Between 1991 and 2018, the peer-reviewed study reported, more than one-third of deaths from heat exposure were linked to global warming. Hundreds of news outlets covered the findings. The message was clear: climate change is here, and it’s already killing people.
But that wasn’t all that was happening. A month later, the same research group, which is based out of the London School of Hygiene and Tropical Medicine but includes scientists from dozens of countries, released another peer-reviewed study that told a fuller, more complex story about the link between climate change, temperature and human mortality. The two papers’ authors were mostly the same, and they used similar data and statistical methods.
Published in Lancet Planetary Health, the second paper reported that between 2000 and 2019, annual deaths from heat exposure increased. But deaths from cold exposure, which were far more common, fell by an even larger amount.
All told, during those two decades the world warmed by about 0.9 degrees Fahrenheit, and some 650,000 fewer people died from temperature exposure.
Chart shows 4-year rolling averages of annual deaths
Source: Zhao et. al., 2021. Data courtesy of the authors.
While it was not covered widely in the press, the second study circulated on Twitter, where many people made some version of the same argument: if cold was deadlier than heat, and the planet was getting hotter, global warming might actually save lives.
But whose lives? Projections indicate milder temperatures may indeed spare people in the globe’s wealthy north, where it’s already colder and people can buy protection against the weather. Yet heat will punish people in warmer, less wealthy parts of the world, where each extra degree of temperature can kill and air conditioning will often remain a fantasy.
Like Goldilocks, our bodies like things just right. Too hot, and our bodies work to cool down, which puts pressure on our hearts and kidneys. Too cold, and our blood vessels constrict to keep our cores warm, slowing the flow of blood and oxygen to the heart. Extremes are dangerous, but unprotected exposure to even mild heat and cold over several days can be deadly.
In most places, the temperature is more often too cold than too hot, which helps explain why more than 90 percent of temperature-related deaths were from cold, according to the Lancet study. On every continent, cold deaths surpassed heat deaths.
Twitter chatter notwithstanding, the Lancet paper did not conclusively prove that global warming led to a drop in temperature-linked deaths. It relied on just 20 years of data, and scientists require at least 30 years of observations to draw conclusions about climate change. Plus, its estimates came with great uncertainty, so what looked like a drop in cold-related deaths could actually have been an increase.
Still, it is at least plausible that, as the paper’s authors put it: “The results indicate that global warming might slightly reduce net temperature-related deaths in the short term.”
What about the long term? A groundbreaking peer-reviewed study, published in November in Harvard’s Quarterly Journal of Economics, gives us a glimpse. In the study, a team of researchers projected how mortality from temperature would change in the future.
The worldwide temperature-linked mortality rate is projected to stay about the same, but you can see enormous geographic variation: colder, wealthier countries do well, while hotter, poorer countries suffer.
Globe shows RCP4.5 scenario. Antarctica left blank because it has no permanent human population.
Source: Climate Impact Lab via Human Climate Horizons. Method detailed in Carleton et. al., 2022.
The study’s method works like this: Imagine a future where the climate stayed as it was in 2015, and compare it to scenarios in which humans keep emitting greenhouse gases. The researchers used two emissions scenarios, one with very high emissions, which has academic interest but is unlikely to actually happen, and a medium emissions scenario, which climate scientists tell me is more likely, if a touch optimistic.
In the medium emissions scenario, Niger, one of the poorest and hottest countries in the world, is projected to suffer the largest increase in temperature-linked mortality, while cold, wealthy Finland sees the largest decrease. That pattern was common, said Michael Greenstone, a University of Chicago economist who co-authored the study.
Greenstone, who over a decade ago helped the Obama administration calculate the “social cost of carbon,” a measure of the damages caused by each ton of carbon dioxide released into the atmosphere, was struck by the unfairness the results implied. “Taking $100,000 from a poor person wouldn’t be equal and opposite to how you feel about giving $100,000 to Elon Musk,” he told me. “I think that’s the underbelly of the climate problem.”
The chart below shows the relationship between 199 countries’ average temperature and how their temperature-linked mortality is projected to change by the end of the century.
|GDP per capita|
Chart shows RCP4.5 scenario. Temperature averaged from 2011-2020. GDP per capita from 2020.
Source: Climate Impact Lab via Human Climate Horizons, U.N. Department of Economic and Social Affairs, World Bank data explorer and Climate Change Knowledge Portal. Method detailed in Carleton et. al., 2022.
What sets this work apart from similar studies is that its model accounts for humans’ evolutionary superpower: adaptation. Of the millions of species on planet Earth, only one of them can turn on the air conditioner.
“If you have any friends who live in Seattle or the Bay Area, they don’t really have air conditioning right now, but they’re all going to,” Greenstone said. “The idea that they were just going to sit there every year, wait for the temperature to come and cause misery and destruction, and not do anything about it, violates everything we understand about human behavior.”
Many studies that project future temperature deaths do not account for adaptation — not because they don’t believe humans will adapt, but because adaptation is hard to quantify. Those studies are likely to overestimate the number of temperature-linked deaths because the people in them act more like coral reefs than humans.
To account for adaptation, the researchers used a statistical technique that compares the behavior of people with similar incomes but who live in places with different climates. It helps answer questions like: how much would it cost Seattle to learn to handle heat waves the way Houston does now?
Kristie Ebi, an epidemiologist at the Center for Health and the Global Environment at the University of Washington who was not involved in the research, has spent a lot of time thinking about questions like that. “All heat related deaths are preventable, and so we should be preventing every death we can from the heat,” Ebi told me. “There’s lots of mechanisms to do so.”
When Seattle endured a record-breaking heat wave last summer, the county’s Regional Homeless Authority opened indoor cooling spaces, and the city’s public utility agency had already filled the reservoirs to prepare for higher water usage. Across the city, air conditioners are selling fast.
But Seattle is a high-tech city in a wealthy country. For much of the world, adapting to heat will be harder.
“Adaptation is not something that just kind of happens when we’re asleep by unicorns and elves. It takes policy and it takes money,” said Andrew Dessler, a professor of atmospheric sciences at Texas A&M University who was not involved in the research. “The people who are barely getting by today are not going to be able to adapt. They’re not gonna have the resources. And so what are they gonna do? Well, they’re going to suffer.”
Information on the data sources I used to produce in this article is available in this computational notebook. I have also published computational notebooks with data and code to produce the map and scatter plot. Be warned that the map notebook may take a long time to load.
You can use the code and data to produce your own analyses and charts — and to make sure mine are accurate. If you do, email me at firstname.lastname@example.org, and I might share your work in my next column.
Thanks to Simon Ducroquet for the illustrations of people at the top of the story.