Skip To Content
BuzzFeed News Home Reporting To You

Utilizamos cookies, próprios e de terceiros, que o reconhecem e identificam como um usuário único, para garantir a melhor experiência de navegação, personalizar conteúdo e anúncios, e melhorar o desempenho do nosso site e serviços. Esses Cookies nos permitem coletar alguns dados pessoais sobre você, como sua ID exclusiva atribuída ao seu dispositivo, endereço de IP, tipo de dispositivo e navegador, conteúdos visualizados ou outras ações realizadas usando nossos serviços, país e idioma selecionados, entre outros. Para saber mais sobre nossa política de cookies, acesse link.

Caso não concorde com o uso cookies dessa forma, você deverá ajustar as configurações de seu navegador ou deixar de acessar o nosso site e serviços. Ao continuar com a navegação em nosso site, você aceita o uso de cookies.

Ricardo Tomás for BuzzFeed News

We Don’t Know How Many People Are Killed By Extreme Weather. This Means Even More People Will Die.

No one is accurately measuring how many people are dying from climate change and extreme weather. This is a fatal mistake that takes pressure off political leaders.

Posted on December 28, 2021, at 2:36 p.m. ET

The number of people who die from extreme weather in the US is being alarmingly undercounted. Without accurate knowledge of the true loss of human life, it will be hard to know how to save more people when the next disaster strikes. Families of the uncounted victims will struggle to get benefits, political leaders will not feel the pressure to better prepare, and people will not realize how lethal climate change really is in the US, right now.

As global warming brings more intense weather extremes, the need for more reliable data is literally a matter of life and death. Here’s how and why the data is so inaccurate.

Take the unprecedented heat wave that hit the Pacific Northwest in late June and early July. According to data downloaded from the National Oceanic and Atmospheric Administration’s Storm Events Database, supposed to be a key US resource on the impacts of extreme weather, only seven people died in Washington state from excessive heat.

During this time, emergency services were overwhelmed by calls for help, and coroners later recorded that dozens of people had died from heat-related illnesses; it's absurd to suggest that the death toll was in single figures. Having combed through death certificates, state authorities eventually put the number of people who perished in Washington at 119 between June 26 and July 2. Yet the toll for the state in NOAA’s database remained unchanged (the database was down at the time of publication).

Even more troubling is that the state’s toll of 119 is also a massive undercount. This is according to a BuzzFeed News analysis that examines how many more people die after a given event, comparing it to the number predicted from long-term and seasonal trends. This approach, known as an “excess deaths” analysis, is the most comprehensive method for estimating hard-to-tally death counts — it has been used, for instance, to calculate the full global toll from the COVID-19 pandemic.

For the Pacific Northwest heat wave, our excess deaths analysis represents the most complete accounting yet. In Washington state, it reveals that the heat wave likely killed more than 440 people there in the week ending July 3. And it suggests that another 230 or so people were killed that same week in Oregon — again more than that state’s list of 96 recorded heat wave deaths.

The Pacific Northwest heat wave is just one example of how official sources are failing to account for the full toll from extreme weather. “I think it’s critically important that we know how destructive these events are, both in terms of capital losses and human losses — and we don’t,” John Mutter, an environmental and social scientist at Columbia University who studies the impacts of natural disasters, told BuzzFeed News. “We could be off by a long way.” Experts agree that there is currently no accurate national database accounting for the lives lost to extreme weather.

With climate change already increasing the frequency or severity of deadly weather extremes, this official ignorance is a big and growing problem. “It’s extremely important, because you can’t manage something that you don’t measure,” Susan Cutter, who heads the Hazards and Vulnerability Research Institute at the University of South Carolina, told BuzzFeed News.

Solving this problem will require an overhaul of how deaths are recorded and sophisticated analytical methods to model the relationships between extreme weather and deaths. Right now, however, that isn’t a top priority for most state and federal agencies. By failing to account for the full toll so they can learn how to reduce it, government officials are missing chances to save hundreds or thousands of people from avoidable deaths, overheated or frozen in their homes or on the streets, or caught up in the chaos of a devastating storm.


In the US, the two main official sources of information on deaths related to extreme weather are NOAA’s Storm Event Database and the CDC, which is the main federal agency responsible for mortality statistics. On a global scale, severe weather deaths are tracked in an international database, EM-DAT, which covers a wide range of disasters from earthquakes to transport accidents, and is the World Meteorological Organization’s go-to source for information on deaths from weather and climate extremes.

Each source has data for 50 states plus Washington, DC; NOAA and EM-DAT also include data for Puerto Rico.

Peter Aldhous / BuzzFeed News / Via CDC National Center for Health Statistics / NOAA Storm Events Database / EM-DAT

This chart, derived from NOAA’s Storm Events Database, EM-DAT, and underlying causes of death recorded by the CDC from death certificates, shows how each source gives a different picture of the annual death toll in the US from extreme weather.

“We have a problem here,” Debarati Guha-Sapir of the Catholic University of Louvain in Belgium, formerly director of its Centre for Research on the Epidemiology of Disasters, which runs EM-DAT, told BuzzFeed News. “If the data is misleading, it can bring us a whole lot of grief, because conclusions can be wrong.”

Much of the discrepancy in the leading databases arises from differences in their methods for tracking weather-related deaths. NOAA’s Storm Events Database starts from recorded weather events and then assesses the death toll mostly from numbers provided by local officials or reported in the media. EM-DAT takes a similar event-based approach — in its case looking at events in which 10 or more people died, 100 or more people were affected, or where a state of emergency was officially declared.

The other leading approach, used by the CDC, is to compile counts from the causes listed on death certificates — possible because the International Classification of Diseases (ICD) has a series of codes that relate to extreme weather, including “Exposure to excessive natural heat” and “Victim of flooding.”

“We have a problem here.”

The CDC’s numbers for extreme weather deaths are much higher. This is mainly due to its more complete accounting for deaths attributed to excessive natural cold or heat, which averaged 733 and 410 per year respectively from 2000 to 2020. Many people die from overheating or cold outside of extreme heat waves or cold snaps, so are missed by event-based methods.

But as the system works right now, death certificates are also missing many fatalities from weather extremes. The CDC gets its data from the states, which in turn compile information from local coroners and medical examiners. And how they fill in death certificates is far from standardized.

What gets recorded on a death certificate also typically misses the indirect ways in which storms and other severe weather events can kill, including deaths from traffic accidents as people evacuate, electrocution from downed power lines, or heart attacks from overexertion.

The NOAA event-based database does try to account for those indirect deaths, as does EM-DAT. But the impacts of a storm or other extreme weather ripple out in ways that are hard to measure. “Displaced elderly people are very vulnerable,” said Mutter. “They lose track of what their medications are.”

NOAA’s failure so far to recognize the full extent of what happened in Washington state during the summer heat wave shows how inadequate event-based recording can be. A NOAA spokesperson noted that the CDC “maintains the official government records on cause of death” and said that NOAA’s numbers for an event that happened almost six months ago were still “preliminary.”

“The verification process for the heat wave in late June/early July is ongoing,” the NOAA spokesperson said by email.

“If you don’t capture the damage, there is no pressure you bring to bear.”

Deaths can also accumulate over much longer periods than are typically considered by officials accounting for the impacts of a storm or other extreme weather. The most extreme example happened in Puerto Rico after Hurricane Maria hit in September 2017. As cremations took place in the wake of the storm without being included in the official death count, nobody familiar with the situation on the ground believed the US territory’s official tally of just 64 deaths.

The full toll was revealed only by comparing the number of deaths that occurred in previous years with the much higher numbers over the weeks that followed the storm, as Puerto Rico was hit by widespread power outages and disruptions to its healthcare system. In December 2017, the New York Times estimated that around 1,000 people more than expected had died in the six weeks after Maria made landfall. Finally, Puerto Rico’s governor commissioned a study from George Washington University, which estimated the storm-related death toll out to February 2018, coming up with a figure of 2,975.

In reality, none of the official sources are hitting the right numbers. “I tell students and others to never trust a death toll, no matter who is estimating it,” Mutter said.

As emergency managers and public health experts try to convince political leaders of the urgency of action to counter climate change and to reduce the impact of the severe weather it is brewing, the data gaps are becoming more pressing. “If you don’t capture the damage, there is no pressure you bring to bear,” Guha-Sapir said.

This year has shone a particularly harsh light on the deficiencies in official accounting for the toll from extreme weather. According to state and local counts, the winter storm that hit Texas in February and the Pacific Northwest heat wave were the two deadliest extreme weather events in the US this year — followed by the tornadoes that devastated parts of the South and Midwest over the night of Dec. 10–11 and Hurricane Ida and its remnants.

But it took excess deaths analyses to reveal how bad the Texas freeze and Northwest heat wave really were. In May, BuzzFeed News estimated the true toll in Texas in February, which was hundreds more than the official count. Similar to methods used to investigate the aftermath of Hurricane Maria in Puerto Rico, we trained statistical models on mortality data that the CDC collects from individual states, which were then used to estimate the number of expected deaths in any week in any state. These models were reviewed by three independent experts in excess deaths analysis.

We then compared the models’ predictions to the number of deaths that were actually recorded, after subtracting deaths attributed to COVID-19. (The models were trained on data from 2015 to 2019, before the pandemic hit.)

In May, when the Texas Department of State Health Services was recording 151 winter storm deaths, we estimated that just over 700 more people than expected had died in Texas during the week of the storm and worst power outages. The state’s tally has since increased to 210. As more death certificate data has been transferred from Texas to the CDC, our analysis now estimates the excess death toll in the week ending Feb. 20 at more than 750.

Our estimates for the Pacific Northwest heat wave, which suggest that around 670 people likely died as a result of the excessive heat across Washington and Oregon in the week ending July 3, were calculated in exactly the same way.

The BuzzFeed News analysis of these events also underlined the shortcomings of relying on the underlying cause of deaths recorded on death certificates, as the spikes we identified were driven in large part by deaths attributed to causes other than exposure to cold or heat. Instead, many deaths triggered by the Texas freeze and by the heat wave in Washington were recorded as deaths from underlying medical conditions — in particular, cardiovascular disease and diabetes.

That makes sense, as people who are already medically vulnerable are more likely to die if exposed to excessive cold or heat. But it means that the CDC’s counts of deaths from extreme weather based on ICD codes are missing an important part of the picture. “Underlying cause is certainly not always going to have the complete information,” Robert Anderson, chief of mortality statistics at the CDC, told BuzzFeed News.

In some states and counties, recording deaths is the responsibility of coroners who may have no specialist medical training. What’s needed, experts told BuzzFeed News, is greater standardization in death reporting, including noting additional factors that contributed to someone’s death as well as an underlying cause.


Clearly, more sophisticated approaches are needed to track the true toll of extreme weather and climate change on human lives. “We have to modernize. We have to find new methods,” said Guha-Sapir. That includes excess deaths analysis, she added.

Other insights can come from epidemiological studies that examine in detail the relationship between the number of recorded deaths in a given place and time, and extreme weather there.

Where that has been done, estimates for deaths associated with extreme heat exceed the numbers recorded by the CDC from death certificates. In April 2020, researchers led by Kate Weinberger, an environmental epidemiologist at the University of British Columbia, estimated that heat accounted for more than 5,600 deaths per year from 1997 to 2006 across some 300 counties. The area studied covered more than 60% of the US population.

A national study of trends in heat deaths published in December 2020 from researchers led by Scott Sheridan, a climatologist at Kent State University in Ohio, used similar methods to show that progress in reducing heat deaths was stalling. This came after decades in decline, possibly due to the widespread adoption of air conditioning and better awareness of the risks. Among men ages 45 to 64 across parts of the South and Southwest, heat death rates were actually increasing.

For wildfire smoke, one of the biggest single causes of weather- and climate-related deaths, epidemiological studies provide the only way of counting the dead.

Studies that have modeled the long-term effects of breathing air laden with tiny particles from wildfire smoke indicate that thousands of Americans die every year as a result. Given the trend for larger and more intense wildfires driven by climate change, the number is forecast to double by the end of the century, according to a 2018 study from researchers led by Bonne Ford, an atmospheric scientist at Colorado State University. This increasing toll from wildfire smoke is expected to partly offset public health gains from reducing other forms of air pollution.

But you won’t find deaths from wildfire smoke listed in any official database. “When people talked about the costs of wildfires, it was all about the cost of fighting fires, damage to property, and direct deaths from the fire,” Ford told BuzzFeed News. “That’s the motivation for why we started doing these studies.”

Knowing how many people are dying from extreme weather is only the start. To reduce the toll, you also need to know which groups of the population are most vulnerable, and have a plan to protect them at times when the risks are most severe. In general, the burden falls most heavily on the less wealthy and on racial minorities: A study published in October from the CDC, based on ICD codes from death certificates, found that the death rate from natural disasters for Black people was 1.87 times times higher than for white people, while for Native Americans the rate was 7.34 times higher.

The details vary from place to place, however. In Multnomah County, Oregon, which includes the city of Portland, the people hardest hit during this year’s heat wave were older, living alone, on higher floors of buildings, and with no air conditioning, according to the county’s public health director, Jessica Guernsey.

But in Maricopa County in Arizona, home to some 4.5 million people who live in the Phoenix metropolitan area, and where almost every day in the summer brings life-threatening heat, the risk profile is very different. There, the typical victim is a man without regular shelter in his 40s or 50s with an addiction to drugs or alcohol. Often people die from exposure to heat after overdosing, said David Hondula, who in September was appointed to head the city of Phoenix’s new Heat Response & Mitigation Office. “It’s that story over and over and over again,” Hondula told BuzzFeed News.

The Maricopa County Department of Public Health is one of the most thorough in the nation at recording deaths from extreme heat. Yet the toll there has been steadily rising. In 2020, Maricopa County recorded 323 heat-associated deaths, an increase of more than 60% from 2019, and 15 times the number recorded in 2001. If reducing heat mortality is the “barometer for success,” Hondula said, “we are failing.”

Given the link between these deaths and substance abuse, Hondula’s office is now considering measures that — without the data at its disposal — might seem unconnected to the dangers of excessive heat. One idea is to equip volunteers with the Phoenix Community Emergency Response Team with Narcan, the emergency treatment for opioid overdoses.

If cities and states across the nation are to reduce the death toll from extreme weather, they will need to overhaul their methods for counting the dead, develop better systems to recognize when emergency plans need to be activated, and have a comprehensive approach to reducing the risks for the most vulnerable people.

“There’s a whole suite of hazard mitigation that can come into play,” said Cutter. “We need better preparedness. We need better warning systems. We need to do some interventions such as adding more greenery to cities.” The last measure can help reduce the urban heat islands that can turn poorer neighborhoods into death traps in the summer as asphalt and concrete absorb the sun’s heat.

As the Pacific Northwest heat wave showed, responding to the reality of climate change may mean dealing with events that exceed anything that has come before. Kristie Ebi, an epidemiologist at the University of Washington who studies health risks from climate change, has advocated for “stress testing,” in which local health agencies, emergency responders, and other stakeholders conduct war game–like exercises to consider how to respond to hypothetical extreme weather outside of their previous experience.

But there are formidable structural and political obstacles to overcome. “I’ve never believed numbers from a state agency. They always have reasons to exaggerate or diminish,” Mutter of Columbia University said. When it comes to getting aid from the federal government, states have a vested interest in stressing how hard they were hit by a disaster. But when the death toll is largely due to failures in a state’s own infrastructure or planning, the incentive works the other way.

Deaths from extreme cold in Texas in February are a case in point. Several states were hit by the winter storm that swept through the region around Valentine’s Day weekend. But only Texas experienced a large spike in deaths, after its power grid failed. This came a decade after the state had been warned by federal officials that its power infrastructure was vulnerable to cold, after narrowly escaping a similar disaster in 2011.

The CDC, meanwhile, has been working with the National Association of Medical Examiners to try and improve the recording of deaths following natural disasters, and has online training materials that encourage coroners, medical examiners, and physicians to take a broader view of what gets recorded as a disaster-related death. But with responsibility for recording deaths lying with local and state authorities, the CDC is limited to an advisory role — it has no direct power to change how deaths are recorded at the local level.

“We know that extreme weather is a matter of life or death.”

And while several federal agencies have offices focusing on climate and health, there has so far been no concerted effort across the US government to better account for the lives lost to extreme weather. “It’s nobody’s responsibility,” Ebi said. The Federal Emergency Management Agency is charged with making sure the US is prepared to withstand and respond to major disasters, but has “not really been part of the picture,” Ebi added.

“We know that extreme weather is a matter of life or death,” Jaclyn Rothenberg, FEMA’s director of public affairs, told BuzzFeed News by email. “As emergency management continues to evolve to meet the demands of our changing climate, it will continue to necessitate us to become more advanced in the ways we understand and react to extreme weather.”

After a year of successive weather disasters, which have exposed both the fragility of US infrastructure and the nation’s inability to count the lives lost, experts say there should be an urgent focus on improving data gathering and acting on that information.

Right now, they warn, local, state, and federal agencies are in the dark, flying into a major storm, as climate change threatens an even more hazardous future.

“Disasters are windows of opportunity to make change,” Ebi said. But the tragedy is that officials often only start planning for weather extremes after the bodies start piling up. “Until it’s in their face, it tends to not happen.” ●

Zahra Hirji contributed reporting for this story.