The number that actually matters
Most people watching a heat forecast look at tomorrow's high. That's the wrong number.
The statistic that determines whether a heat wave kills is the overnight low. The human body uses night to recover — to bring its core temperature down, to give overworked organs rest from hours of thermal stress. When nighttime temperatures stay above 80°F, that recovery can't happen. Heat stress accumulates across days. By the third or fourth night of a dome like this one, bodies that could have managed a single afternoon at 102°F begin to fail quietly — in apartments, in darkness, while the city assumes the worst is over.
Epidemiologists who study heat events watch the overnight low the way storm trackers watch wind speed. It's the variable that separates a dangerous heat wave from a deadly one. This week, not one night across the eastern United States will fall below that threshold.
Overnight Lows — Heat Dome, June 29 to July 4, 2026
Every bar that extends past the dashed line is a night the body cannot cool itself. All six nights extend past it.
Why heat deaths disappear from the record
Heat kills indirectly. It overwhelms the body's ability to regulate temperature, which places strain on the cardiovascular system, the kidneys, the respiratory system. When those systems fail, that is what gets written on the death certificate — heart failure, acute renal failure, respiratory arrest. The underlying cause, the 95-degree apartment that never cooled below 85°F overnight for four days, rarely appears.
This is not a clerical oversight. It reflects a genuine uncertainty in medical causation: a person with preexisting heart disease dies during a heat wave. Did the heat kill them? Medical examiners, trained to identify proximate cause, record what the body did, not what the air temperature was.
The result is a systematic disappearance of heat from the death record. When researchers go back through death certificates and cross-reference them with heat event dates — comparing excess mortality against expected mortality for the same period — they find numbers between five and ten times what the official heat death count shows. The CDC itself has acknowledged the problem. The methodology has been critiqued in peer-reviewed journals since the 1990s. And yet the official count has barely changed.
"People may present to the Emergency Department with a heart attack, stroke, asthma attack, or kidney failure — and heat is often not recorded in medical records as the underlying cause."
PBS NewsHour · "Inconsistent methods for counting U.S. heat deaths," 2026If hurricanes had a tenfold undercount, there would be congressional hearings. If tornadoes were killing ten times their recorded number, the Federal Emergency Management Agency would face pressure to explain why. Heat's tenfold undercount is a footnote in academic literature, because heat leaves no footage.
The disasters we respond to are the ones we can photograph.
A hurricane produces before-and-after images. A tornado leaves a scar across the landscape. Flood footage has a visual grammar that news organizations know how to cover — cars floating, people on rooftops, the single dramatic moment of catastrophe. These images drive emergency declarations, federal appropriations, infrastructure investment.
Heat kills in a way that produces none of this. It kills one person at a time, usually indoors, usually at night, usually an elderly person who lives alone. There is no moment of catastrophe — there is only a body that stopped working, a death certificate that says cardiac arrest, and a summer that looks, from the outside, as if it ended without incident.
This is not accidental. The political economy of disaster response runs on imagery. FEMA funding formulas respond to declared disasters; disaster declarations respond to visible events; visible events are events with footage. Heat produces no footage. And so heat receives, per death, a fraction of the federal attention allocated to disasters that do.
The consequence is infrastructure. Cities in hurricane corridors have building codes, evacuation systems, early warning networks, and designated shelters built over decades of pressure from a visible threat. Most American cities have no serious heat emergency infrastructure — no mandatory cooling center network, no heat-specific early warning system, no requirement that landlords maintain maximum indoor temperatures. Heat has not generated the political pressure to build these things, because heat's deaths do not accumulate in a way that can be photographed.
What the investigation found
On July 14, 1995, Chicago logged a daily high of 106°F and an overnight low that never fell below 84°F. Over five days, 739 people died. The city's official death toll at the time was 17.
When the sociologist Eric Klinenberg later investigated what had happened, he found that most victims had died alone, in apartments, at night, with their windows closed. Many had been afraid to open windows for fear of crime. Many had no one who knew to check on them.
The defining variable, Klinenberg found, was not poverty or age or access to air conditioning. It was social isolation. The people who died were the people who were already invisible — to neighbors, to family, to the public health system that might have counted them. Heat found them precisely where no one else was looking.
Tonight, across 28 states, it will not cool down.