Yet when they ran the numbers and looked at mortality in places affected by storms, they were initially perplexed by the results, Hsiang told me earlier this week. The numbers they came up with didn’t even make sense at first.
“It was months of us trying to understand what we were looking at,” Hsiang said. “And then once we realized what we were seeing, it was years of us checking our work to find what we missed.” Only when it was clear that their work resembled other American public health statistics — specifically, that the white-Black mortality mirrored what has been found in other studies — that the horrifying truth sunk in.
The finding: Hurricanes are hundreds of times deadlier than anyone has realized.
Their study, which was published on Wednesday in Nature, finds that the average hurricane kills 7,000 to 11,000 people after making landfall in the United States. These previously uncounted deaths happened not during a storm or in its immediate aftermath, but as a long, slow trickle of mortality that plagues a region long after the clouds have cleared and floods have abated.
In any one year, the number of storm-related deaths is not very high. And yet a wave of excess deaths is visible in population data for at least 15 years after a storm hits an area, they found.
“It lasts for so many years, and because there’s so many storms hitting so many states, once you add up, it becomes this enormous number,” Hsiang told me. When added together, hurricanes’ long-term death toll exceeds American combat deaths in all wars, combined. The number so dwarfs previous estimates that it suggests tropical cyclones alone are a major determinant of public health across the United States.
Kerry Emanuel, an MIT meteorology professor who studies climate change and hurricanes, told me that the results were “truly astounding” and “persuasive,” although he noted that he is not an expert in the statistical approach used in the paper.
“Summed over all hurricanes, this amounts to three to five percent of all deaths near the Atlantic coast,” he said. “I expect this result will prove controversial and will be followed up by many other studies of long-term mortality from natural disasters.”
The paper fits into a growing body of research on what others have called the hidden or invisible public health threat of environmental threats. For years, researchers have known that air pollution and heat waves, seemingly silent hazards, can in fact kill tens of thousands of people. Lately they have begun to apply the same techniques to other hazards, with outsized results.
Officially, Hurricane Maria killed 64 people when it struck Puerto Rico in 2017. But when researchers surveyed households across the island months after the storm, they found the death toll was closer to 4,600. (The territory’s government later revised the official figure to 2,975.) These deaths were caused not by the cyclone’s high winds or torrential floods, but rather by secondary effects of the storm’s destruction. Maria took out the island’s power grid and road networks, for instance, and preventing people with heart attacks and strokes from reaching the hospital in time.
That paper was written six months after Maria struck the island; this new hurricane paper considers a wider time horizon, finding that more than 80,000 Americans die each year as a result of a hurricane, whenever it occurred. Black people were disproportionately killed by the aftermath of hurricanes, at least partly because a larger share of the country’s Black population lives in storm-afflicted areas. About 37,000 white deaths each year are due to a prior tropical cyclone.
How could such storms cause such a long tail of deaths, affecting areas 10 or 15 years after they come ashore? The paper cannot answer those questions today. But Hsiang and Young hypothesize that hurricanes cause extreme economic distress, which can resonate for years or decades afterward. “If someone suffers a loss and can’t invest in their business, then it will have ramifications for their income long into the future,” Hsiang told me. “If someone is on a fixed income and their garage is destroyed, and they pull from their retirement funds to fix the garage, then eight years later when they face a big medical decision, they might choose” a cheaper or less effective form of treatment.
“When you talk to people, you hear stories like this,” Hsiang added. The time and money invested in dealing with the storm is often a “pure loss,” even if some of the damage ultimately gets reimbursed. “Even if you have insurance, that just means you already paid for it in some way,” he said.
Storms cause disruption in other ways. They can break up communities and social networks. (If children move away, for instance, their parent can face higher medical bills.) Hurricanes can also impose high costs on states, towns, and cities, which may then have to reduce or restrict other services as a result.
“When you think about how communities rebuild — local municipalities and states — they also play a lot of games with their budget” in the aftermath of a storm, Hsiang said. “If they spend a lot of money to rebuild a bridge or boardwalk somewhere, does that come out of some social program 10 years later? Or building a new NICU hospital?” That could explain why an infant — even one born 15 years after a storm struck a given area — could face a higher chance of death.
Young and Hsiang think that these economic drivers are most likely to be the big reason for the excess deaths — the effect is just too big and drawn out to make any other cause likely — but other possibilities exist, they recognize. Hurricanes could be deadly simply because they are highly stressful events. “We see an effect on cancer rates and also cardiovascular illness. Stress matters a lot to those,” Hsiang said. It’s also possible that hurricanes unleash contamination into the environment that then makes people sick. A flooded basement can become a breeding ground for mold. “There’s gas stations in every town. What chemicals come out when there’s flooding?” Hsiang wondered.
The paper may also help resolve a riddle in American public health. On average, Americans die earlier in the eastern half of the continental United States than in the western half. This effect is worst in the Gulf Coast and Southeast but persists to some degree in the Mid-Atlantic and Northeast.
The paper suggests that hurricanes may have something to do with this geographic phenomenon. For infants, people below the age of 44, and Black people of all ages, hurricanes may explain a large share but not all of the mortality gap.
