The Autopsy / Case File · Environmental & Climate
Hundreds of Miles Downwind
No fire ever reached Manhattan in June 2023. The sky over it turned the color of a traffic cone anyway, and for days the air was among the most hazardous on the planet. Nobody there died in a wildfire. Some number of them, a growing body of research says, still died because of one.
A causal-modeling study published in Science Advances estimates wildfire smoke caused roughly 24,100 deaths a year across the contiguous United States between 2006 and 2020 — a toll driven almost entirely by fine particulate pollution, PM2.5, drifting far past the fire's own perimeter. A separate 2023 study in Science of the Total Environment modeled city-by-city downwind exposure and put New York's annual wildfire-smoke death toll at roughly 119, Atlanta's at 76, and Houston's at 65 — none of them fire-adjacent cities, all of them regularly inside a smoke plume that started hundreds of miles away. A longer-horizon PNAS analysis, using different methodology, estimated a lower baseline of about 720 annual deaths in 1996-2005 but projected the toll could climb past 71,000 a year by 2050 under continued warming. The estimates vary by method. None of them put the number near zero, and none of them require a body to be found anywhere near a burn scar.
Join the discussion →
Wildfire death tolls, as most people encounter them in news coverage, are counts of people who died in the burn zone — trapped by flame, overcome by heat, unable to evacuate in time. Those numbers are real and they're tracked closely by fire agencies. They are also, epidemiologically, a small fraction of what wildfires actually kill. The larger and far less visible toll comes from what a fire releases into the air long after — and far outside — the area it burns: fine particulate matter, PM2.5, small enough to lodge deep in lung tissue and pass into the bloodstream, capable of triggering cardiovascular and respiratory events in people who never saw a flame, never smelled smoke up close, and in many documented cases never even knew a fire was burning anywhere nearby.
A National Estimate, and the Method Behind It
The most comprehensive recent national estimate comes from a causal-modeling study published in Science Advances, which used a quasi-experimental design — comparing mortality patterns in areas exposed to varying levels of wildfire-specific PM2.5 against comparable unexposed periods and locations, rather than relying on simpler correlational methods — to isolate wildfire smoke's mortality signal from the broader, year-round burden of ambient air pollution. That approach put the average annual U.S. death toll attributable to wildfire smoke PM2.5 at approximately 24,100 people across the fifteen years from 2006 to 2020, a period covering the buildup toward what's now widely recognized as a more severe and longer wildfire season across much of the American West. The causal-modeling method matters here specifically because wildfire smoke doesn't arrive on its own — it mixes into whatever background air pollution a city already has — and separating the two requires exactly the kind of exposure-variation design the study used rather than a simple total-pollution correlation, which would risk attributing non-wildfire pollution deaths to fire smoke or vice versa.
A separate and methodologically distinct analysis, published in Science of the Total Environment in June 2023 by researchers at Cornell University, the University of Houston, and Nanjing University of Information Science and Technology, took a more geographically granular approach: quantifying wildfire-induced PM2.5's premature mortality and economic cost city by city across the contiguous U.S. That study's national range — 4,000 to 9,000 premature deaths and $36 to $82 billion in annual economic loss — sits below the Science Advances causal estimate, a gap the researchers themselves attribute partly to different exposure-response assumptions and study periods, underscoring that this is still an active, evolving area of environmental epidemiology rather than a single settled number. What the city-level breakdown adds that the national topline can't is specificity about where the burden actually lands: the study estimated roughly 119 premature deaths a year in New York City, 76 in Atlanta, and 65 in Houston — three major population centers with no significant wildfire activity of their own, absorbing a meaningful share of a mortality burden generated somewhere else entirely.
The June 2023 Event, as a Case Study in Real Time
The smoke event that made wildfire's downwind reach visible to millions of Americans who'd never thought about it happened in early June 2023, when smoke from an unusually intense wildfire season in Quebec drifted south and settled over the Northeast and Mid-Atlantic. New York City's air quality index reached levels classified as hazardous — among the worst readings any major world city recorded that week — and the smoke reached as far south as Washington, D.C., and Baltimore, hundreds of miles from the nearest active fire. It was, in the language epidemiologists use for this kind of event, a natural experiment: a sudden, sharp, well-documented spike in PM2.5 concentration across a densely populated region with no local fire source, giving researchers an unusually clean signal to study smoke's short-term health effects against baseline air quality in the same cities just days before and after.
The event didn't generate an immediate, comprehensive excess-mortality count in the way a heat wave's death toll sometimes gets tallied within weeks — smoke-attributable mortality, like most air pollution mortality, tends to be estimated through longer-term epidemiological modeling rather than real-time case counting, because most of the deaths it contributes to are cardiovascular and respiratory events that get coded to the underlying condition, not to the smoke that helped trigger it. That's precisely the mechanism driving the broader undercount problem in this category of death: a heart attack triggered by three days of hazardous PM2.5 exposure gets a cardiac cause of death on the certificate, with no field anywhere on it for "acute wildfire smoke exposure," even in cases where the timing lines up exactly.
The smoke that turned New York's sky orange in June 2023 started as a fire that killed almost no one directly. The mortality it caused afterward, hundreds of miles away, doesn't show up in any wildfire death toll — because officially, by the time it kills someone, it isn't wildfire smoke anymore. It's a heart attack.
Why the Estimates Don't Agree — and Why That's Not a Red Flag
The spread between the Science Advances, Science of the Total Environment, and PNAS estimates — 24,100 versus a 4,000-to-9,000 range versus a 1996-2005 baseline of roughly 720 — can read, at first glance, like a field that doesn't know its own numbers. In practice, the gap reflects real methodological differences that environmental epidemiologists treat as expected rather than alarming: different study periods (the PNAS baseline predates the more severe fire seasons of the 2010s and 2020s almost entirely), different geographic scope, and different statistical approaches to estimating the exposure-response relationship between a given concentration of wildfire-specific PM2.5 and excess mortality risk — a relationship researchers are still actively refining, including ongoing work on whether wildfire smoke's PM2.5 is more toxic per unit of mass than general ambient PM2.5, a question with a growing but not yet fully settled body of supporting evidence. What every recent estimate agrees on, despite the range, is direction and order of magnitude: this is a mortality category in the thousands to tens of thousands of American deaths annually, not the low hundreds, and every model that's incorporated recent, more severe fire seasons has revised its estimate upward, not down.
The projections compound that trend. The PNAS analysis, despite its comparatively low historical baseline, modeled wildfire PM2.5-related mortality climbing to somewhere between 1,300 and 2,300 annual cases by mid-century under continued climate warming — and the Science Advances research, extrapolating from its higher current baseline, projected a 2050 toll approaching 71,420 annual deaths under a high-warming scenario, a 73 percent increase over the 2011-2020 average the same study calculated. Different starting points, same trajectory: every model built on this data expects the smoke-mortality burden to keep growing as fire seasons intensify, independent of which specific number a given study reports for today.
A Federal Tracking Gap
Neither the EPA's AirNow system nor CDC's Environmental Public Health Tracking Network was built to produce a real-time, cause-specific wildfire-smoke death count the way NHTSA's FARS system tracks traffic fatalities or the CDC's own vital statistics system tracks heat deaths with a dedicated coding convention. AirNow tracks air quality in real time and issues public health advisories, and the Tracking Network compiles smoke-event and air-quality data for researchers, but translating a smoke event into an attributable death count still requires the kind of after-the-fact epidemiological modeling the Science Advances and Science of the Total Environment studies represent — modeling that takes months to years to produce, runs through independent academic teams rather than a standing federal surveillance program, and yields estimates that, as shown above, still don't fully converge. For a mortality category researchers increasingly agree kills more Americans annually than die in the fires themselves, there is no single federal agency currently positioned to publish an annual, authoritative number the way there is for almost every other major cause of death in this series.
And Then You're Dead
Nobody who dies of wildfire smoke five hundred miles downwind gets counted as a wildfire death. They get a cardiac or respiratory cause of death, filed the way any other version of that death would be, with nothing on the certificate connecting it to a fire they never saw burning. The smoke crossed state lines and time zones to reach them. The paperwork stopped noticing it the moment it did.
That's it. That's the whole thing.
Read about the CDC's heat-death undercount · or see every case file
Sources
- The estimated 24,100 average annual U.S. deaths attributable to wildfire smoke PM2.5, 2006-2020, and the causal quasi-experimental modeling methodology — "Wildfire smoke PM2.5 and mortality rate in the contiguous United States: A causal modeling study," Science Advances, 2024
- The 4,000-9,000 annual premature death range, $36-82 billion annual economic loss estimate, and city-level downwind mortality figures (New York ~119, Atlanta ~76, Houston ~65 deaths/year) — Pan S, Gan L, Jung J, Yu W, Roy A, Diao L, Jeon W, Souri AH, Gao HO, Choi Y, "Quantifying the premature mortality and economic loss from wildfire-induced PM2.5 in the contiguous U.S.," Science of The Total Environment, vol. 875, June 1, 2023, as reported by Cornell Chronicle and University of Houston News, June 2023
- The 1996-2005 baseline of approximately 720 annual wildfire PM2.5-attributable deaths and the 2050/2090 projections (1,300-2,300 and 1,900-2,300 cases respectively under RCP4.5-8.5 scenarios) — "Estimating PM2.5-related premature mortality and morbidity associated with future wildfire emissions in the western US," peer-reviewed climate/wildfire mortality projection literature, PMC
- The 73% increase (to roughly 71,420 annual deaths by 2050 under high-warming scenarios) relative to the 2011-2020 average — Science Advances wildfire smoke mortality causal-modeling study, projection component
- The June 2023 Quebec wildfire smoke event's hazardous air-quality readings across New York City, Washington D.C., and Baltimore — contemporaneous 2023 air-quality reporting and NOAA/EPA AirNow event documentation
- The absence of a dedicated, real-time federal wildfire-smoke mortality surveillance system comparable to NHTSA's FARS or CDC's heat-mortality coding conventions — EPA AirNow and CDC Environmental Public Health Tracking Network program scope documentation
- Photo: "Quebec Canada Wildfire Smoke Consumes New Jersey and New York City, June 7, 2023," Anthony Quintano, CC BY 2.0, via Wikimedia Commons
Comments
Loading…