A severe heat wave leaves nothing to photograph. No structure falls, no road washes out, and the damage assessment forms sitting in your emergency manager’s drawer have no line item that fits, which is part of why an event that kills several hundred people in a week can pass without a federal declaration, a task force, or a single unified command post. This piece covers how heat deaths get counted and why the counts disagree, what the well studied events in Chicago, western Europe and British Columbia found about who dies and where they are found, what cooling centers actually accomplish, and how to get heat into a plan your agency already maintains.

A hazard that damages nothing

Almost every mechanism American emergency management uses to recognize a disaster is keyed to physical damage. Local damage assessment teams count destroyed and major-damaged dwellings, the state rolls those counts up, and the request for a major disaster declaration under the Stafford Act is built around what a governor can document in structures, public infrastructure and response costs. A heat wave produces none of that, because the buildings are intact and the roads are open, so the paperwork that normally converts a bad week into a declared incident has nothing to grip.

The deaths are also dispersed in a way that defeats the usual triggers. Instead of arriving in one place at one moment, they accumulate over five or six days across a metropolitan area, one apartment at a time, mostly among people who were already elderly or chronically ill and whose deaths will be certified individually by a medical examiner or coroner working from circumstances rather than from a scene. Nobody stands in front of a smoking pile, no incident is named on the day it happens, and the total only becomes visible weeks later when somebody adds it up.

The consequence is a real gap in authority and funding. Emergency protective measures are the Public Assistance category that would cover things like opening and staffing cooling facilities if a declaration were in place, but a declaration for heat alone has been rare, and in 2024 a coalition of environmental and labor organizations petitioned FEMA to treat extreme heat and wildfire smoke as qualifying events under the Stafford Act. Check FEMA’s current declarations record and the current Public Assistance Program and Policy Guide for where that stands rather than relying on this summary, and check your own state code for what a governor’s or a county chairman’s emergency declaration can unlock locally, because in most heat events the state and local declaration is the only one you are going to get.

Chicago, July 1995: 465 certified and 739 excess

Chicago was hit in the middle of July 1995 by a hot, humid air mass that produced a reading of 106 degrees Fahrenheit at Midway Airport on 13 July, with heat index values well above that and overnight lows that stayed in the eighties for several consecutive nights. Power failures in parts of the city took air conditioning away from people who had it, water pressure dropped in places where hydrants had been opened, and emergency departments began going on bypass as ambulance volume climbed. The city’s initial public posture, documented in Eric Klinenberg’s Heat Wave: A Social Autopsy of Disaster in Chicago (University of Chicago Press, 2002), was to question whether the deaths being reported were really heat deaths at all.

The Cook County Medical Examiner’s Office certified 465 heat-related deaths for the week of 14 to 20 July 1995. The office’s storage capacity was exhausted, refrigerated trailers were brought in to hold bodies, and families waited days for autopsies and releases. Separately, Whitman and colleagues published an analysis in the American Journal of Public Health in 1997 that compared observed deaths in Chicago during the same week against the number that would have been expected from prior years, and estimated roughly 739 excess deaths. Those two figures disagree by more than half, and both of them are correct for what they measure.

The same analysis reported that death rates were highest among older residents and among Black residents of the city, a pattern that reappears in later heat events elsewhere. What made Chicago the reference case is not the size of the toll but the fact that it was studied hard afterward, by a medical examiner who kept careful records, by federal epidemiologists who ran a case-control study of who died and why, and by a sociologist who went back through the neighborhoods. Most American heat events get none of that attention, which means the absence of a documented toll in your own jurisdiction is at least as likely to reflect an absence of investigation as an absence of deaths.

Excess mortality and why two honest numbers disagree

Certified counts come from death certificates. A medical examiner or coroner decides that heat caused or contributed to a particular death, usually working from a measured body temperature at or around the time of death, the condition of the body, the ambient temperature in the room where the person was found, and the exclusion of other causes. The National Association of Medical Examiners has published criteria for this determination, and jurisdictions apply them unevenly. A frail 84-year-old with heart disease who dies on the fourth day of a heat wave in an apartment at 95 degrees may be certified as a heat death in one county and as a cardiac death in the county next door, and if the body is not found for a week the room temperature at the time of death is no longer knowable.

Excess mortality takes a different approach and does not require anyone to decide the cause of an individual death. Analysts model the number of deaths expected in a population during a given period based on prior years, adjusted for trend and season, then subtract that expectation from the deaths actually observed. The result captures the cardiovascular and respiratory deaths that heat precipitates without ever appearing on the certificate, which is why excess figures are almost always larger than certified figures. The method is sensitive to choices, including which baseline years are used, how long a window after the heat is included, and whether a seasonal adjustment is applied, so two competent teams working the same event can publish different numbers.

The other argument that runs through this literature concerns mortality displacement, sometimes called harvesting, meaning the question of how many of the people who died in the heat would have died within weeks anyway. The way to settle it is to look at whether the weeks after the heat show a deficit of deaths, and in the larger studied events the deficit accounts for only part of the excess. When you publish a number for your own jurisdiction, publish the method with it, because a reader who sees 465 in one document and 739 in another and no explanation will conclude that somebody is lying.

Europe 2003, and the response France built afterward

Western Europe was hit by an extended heat wave in the first three weeks of August 2003, and the estimates of what it cost vary more than for any other event discussed here. Robine and colleagues published an analysis in Comptes Rendus Biologies in 2008 under the title “Death toll exceeded 70,000 in Europe during the summer of 2003,” covering sixteen countries across a long summer window. Estimates circulated in 2003 and 2004 were considerably lower, on the order of 35,000 across a smaller set of countries, and the two figures disagree because they cover different countries, different date ranges and different baselines rather than because one side miscounted. For France specifically, an analysis prepared for the French government by researchers at INSERM put the excess at roughly 14,800 deaths for the first three weeks of August 2003.

France in August is a country on holiday, and the timing mattered. Hospital staffing was at summer levels, many physicians were away, families were out of the cities, and a large share of those who died were elderly people living alone in apartments under zinc roofs in Paris and other cities. Mortuary capacity in the Paris region was exceeded, and refrigerated facilities outside the city were used to hold bodies while families were located. Public attention arrived roughly a week into the event, and the director general of health resigned afterward.

What followed is the most complete national heat response system anyone has built. France established a national heat plan with graduated alert levels tied to a joint meteorological and health warning system, with thresholds set regionally rather than by a single national temperature, and it imposed a legal obligation on mayors to maintain a register of isolated elderly and disabled residents who consent to be listed so they can be contacted during an alert. Hospitals plan for summer surge capacity, and nursing homes were required to provide a cooled room. Verify the current French requirements with French authorities rather than with this summary, because the plan has been revised repeatedly since 2004, but the design principle travels well: the alert has to identify named individuals somebody is responsible for calling.

Who died, and where they were found

Semenza and colleagues published a case-control study of the Chicago deaths in the New England Journal of Medicine in 1996, comparing people who died of heat with living neighbors of similar age and location. Living alone, not leaving home daily, being confined to bed, and lacking access to transportation were associated with higher risk. A working air conditioner in the home was the strongest protective factor, and the other protective associations were having social contact, having transportation available, and visiting a cooled place during the heat. That is an individual-level finding from a controlled comparison, and it is the most directly actionable piece of evidence in the whole subject.

Klinenberg’s book took the question to the neighborhood level, comparing adjacent Chicago community areas with similar poverty and similar proportions of elderly residents but sharply different death rates, and argued that differences in street life, occupied storefronts, and whether older people had reasons and safe routes to leave the apartment explained part of the gap. He also documented people who kept windows closed and refused to open doors to strangers because they were afraid of crime. His neighborhood explanation is a sociological interpretation and has been argued about by other scholars, so I present it as his analysis rather than as a settled finding, while the individual risk factors from the case-control study stand on their own.

The pattern repeated in British Columbia during the heat dome of late June 2021, when the village of Lytton reached 49.6 degrees Celsius on 29 June and burned the following day. The BC Coroners Service identified 619 heat-related deaths between 25 June and 1 July 2021, and a death review panel reported in June 2022 that the overwhelming majority of those people died indoors in private residences without adequate cooling, that most were older adults, that many lived alone, and that chronic health conditions were common among them. Oregon and Washington each reported roughly a hundred heat deaths from the same weather system, with counts revised as investigations closed, so use each state health authority’s current published figure rather than the number that was in the news that week.

Cooling centers and who actually reaches them

Opening cooling centers is the standard American local response, and it is worth doing, but the record of the studied events shows that the people who died were not the people who used them. Reaching a cooling center requires knowing it exists, being physically able to leave the residence, having a way to travel there, and being willing to leave a home, a pet and a set of belongings behind. Those requirements filter out very nearly the exact population that the epidemiology identifies as highest risk, which is the frail, isolated older adult without a car and without anyone checking in.

Hours are the second structural problem. Many jurisdictions run cooling centers during business hours in libraries, senior centers and recreation buildings, and close them in the late afternoon. The human body relies on cooler nights to shed accumulated heat, and heat events with high overnight minimum temperatures are consistently the deadly ones, so a facility that closes at five in the afternoon does not address the exposure that is actually killing people. Phoenix and the surrounding Maricopa County jurisdictions have pushed toward longer hours and overnight capacity through a coordinated heat relief network of hydration, respite and cooling sites run by nonprofits and municipalities, which is a useful model to read before you write your own plan.

Because a working air conditioner is the strongest protective factor in the Chicago study, the interventions that put or keep cooling inside somebody’s home deserve at least as much of your planning attention as the centers do. That means utility disconnection policy, which Arizona regulators tightened with a summer shutoff moratorium after a widely reported death following a disconnection, weatherization and window air conditioner distribution programs, and rules in assisted living and licensed care facilities requiring cooled space and backup power. Check your own state utility commission and your state licensing agency for what currently applies, since these rules vary by state and change.

What heat does to the response system itself

Heat waves generate EMS surge that arrives in a specific shape: a climbing baseline of calls across days rather than a spike in an hour, weighted toward cardiac, respiratory and altered mental status complaints in older patients, with heat exhaustion and heat stroke calls from outdoor workers layered on top during the afternoons. Emergency departments fill, offload delays lengthen, units stack at hospitals, and the system loses unit hours to wall time at exactly the point when call volume is rising. Any jurisdiction whose surge plan assumes a short, sharp peak should look at whether it holds up on day four.

Medical examiner and coroner capacity fails early and quietly. Cook County ran out of room in 1995, the Paris region ran out in 2003, and the BC Coroners Service had to stand up a review process afterward. Storage capacity, transport contracts, additional investigators, and a plan for family notification when bodies are discovered days after death are all things to settle in writing before the week they are needed, because the alternative is refrigerated trailers arranged over the phone while the news cameras watch.

Responders are also at risk. Structural firefighting gear is a vapor barrier, and rehab requirements built around cold-weather assumptions do not fit a July working fire at 98 degrees with 60 percent humidity. The other exposure worth planning for is the combination of a heat wave with an extended power outage, which Brian Stone Jr. and colleagues at Georgia Tech modeled in Environmental Science and Technology in 2021 for several American cities, estimating that a multi-day blackout during a heat wave would push a very large share of the population of a hot-climate city into heat illness requiring care. Those are modeled estimates rather than counts, and I cite them as a scenario worth exercising rather than as a prediction, because the coupling of grid failure and heat is the plausible path from several hundred deaths to several thousand.

Putting heat into a plan you already have

Heat belongs in the hazard identification and risk assessment that sits at the front of your local hazard mitigation plan, and this is not a paperwork point. Eligibility for federal mitigation funding generally depends on the hazard appearing in an approved plan, which is what makes tree canopy, cool roof, weatherization and community resilience hub projects arguable as mitigation rather than as general municipal improvement. Federal mitigation grant programs have changed substantially in the last two years, so ask your state hazard mitigation officer what is currently open and what heat projects have actually been funded in your state rather than budgeting against a program name you remember.

The operational half of the work is a trigger-based annex tied to products your agency already receives. Find out from your local National Weather Service forecast office what criteria they use for heat advisories and warnings in your counties, how the CDC and NWS HeatRisk forecast product displays your area, and who at your agency receives those products at two in the morning on a Saturday. Then write down what happens at each level: who opens what, who calls whom, which registry of at-risk residents gets worked, who notifies licensed care facilities, and what the public message says about overnight cooling rather than daytime cooling.

Occupational heat rules are a live area to track. OSHA published a proposed federal heat injury and illness prevention rule in 2024 and the rulemaking has continued since, while California, Oregon, Washington, Maryland, Minnesota and Colorado have adopted state-level requirements of varying scope, so verify what applies to your own employees and to the contractors working on your jurisdiction’s projects with your state plan office or with OSHA directly. Several cities, including Miami-Dade County and Phoenix, have created dedicated heat officer positions or offices since 2021, which is worth knowing about mostly because it tells you somebody at those jurisdictions will answer the phone if you call to ask how they built their program.

What to do at your agency

• Have your EMS operations chief or QA officer pull dispatch and patient care data for the last three Junes, Julies and Augusts, count heat-related complaints and total call volume by day, and identify which day of a multi-day heat event the system actually started to strain.
• Ask your county coroner or medical examiner, in writing, how a heat-related death is certified in your jurisdiction, how many were certified in each of the last three years, and how many bodies the office can hold before it needs outside refrigeration.
• Obtain the current cooling center list from whoever maintains it, compare the posted closing time against the forecast overnight low for the hottest night of last summer, and give the mismatch to your emergency manager in a one-page memo.
• Add one item to an existing local emergency planning committee or public safety coordinating meeting agenda: who maintains a list of at-risk residents in this jurisdiction, what authority covers it, and who is responsible for calling the people on it during a heat alert.
• Have your training officer confirm that the department rehab SOG names a measurable trigger for mandatory rehab in hot weather, that the measurement is wet bulb globe temperature rather than shade heat index, and that a working WBGT meter is actually on an apparatus.
• Call your local National Weather Service forecast office and ask for the heat advisory and warning criteria used for your counties, then verify with your communications supervisor that those products reach a duty officer after hours and not just an unmonitored inbox.
• Write one paragraph into the existing mass care or public health annex of your emergency operations plan stating who opens cooling facilities, what time they close, how transportation is provided, and whether pets and personal belongings are accepted.

Takeaways

• Heat produces no damaged structures, which means the damage assessment and declaration machinery that converts other hazards into declared disasters has nothing to measure, and local and state declarations are usually the only ones a jurisdiction will get.
• The Cook County Medical Examiner certified 465 heat-related deaths for the week of 14 to 20 July 1995, while Whitman and colleagues estimated roughly 739 excess deaths for the same week in the American Journal of Public Health in 1997, and the two numbers differ because they measure different things.
• Certified counts depend on the investigative practice of the office doing the certifying, and excess mortality estimates depend on baseline and window choices, so every published figure needs its method, its date range and its source attached.
• Estimates of the 2003 western European toll range from roughly 35,000 in early tallies to more than 70,000 in the 2008 analysis by Robine and colleagues in Comptes Rendus Biologies, with the difference driven by countries covered, dates and baseline rather than by error, and the French national excess for early August 2003 was put at roughly 14,800 in an INSERM analysis prepared for the government.
• The case-control study by Semenza and colleagues in the New England Journal of Medicine in 1996 found that living alone, not leaving home daily, being bed-confined and lacking transportation raised risk, while a working air conditioner, social contact and access to a cooled place were protective.
• The BC Coroners Service identified 619 heat-related deaths in British Columbia between 25 June and 1 July 2021, and its 2022 death review panel reported that the great majority of those people died indoors in private residences without adequate cooling, most of them older adults and many living alone.
• Cooling centers serve people who can travel and who are willing to leave home, which excludes much of the population the epidemiology identifies as highest risk, and centers that close in the late afternoon do not address the high overnight minimum temperatures that drive mortality.
• Getting heat named as a hazard in your local hazard mitigation plan is what makes canopy, cool roof and weatherization projects arguable for mitigation funding, and your state hazard mitigation officer is the person who knows what is currently open.