Who France Can Find in a Heat Wave

After the 2003 heat wave, France asked every town to keep a list of residents to check on when the weather turns dangerous. This summer, for the first time, the government adopted rules for putting people on that list without waiting for them to ask. Twenty-three years in, what has being listed actually guaranteed anyone?

Who noticed in August 2003

In August 2003, the first sign that someone in a Paris apartment was in trouble was usually a small one. A knock went unanswered. A phone kept ringing. Someone who was normally out and about in the neighborhood stopped showing up. Sometimes there was a smell in the stairwell.

We know this because two forensic doctors, Dominique Lecomte and Dominique de Penanster, went back through the files of 452 Paris residents whose deaths that month had been referred to the city's forensic institute. They read police reports and medical exams, and they talked with relatives, neighbors and social workers. The files kept describing stifling heat indoors, in some apartments between 97 and 104 degrees Fahrenheit. In 364 cases, the doctors could tell who had first raised the alarm:

• Friends or neighbors: 43%
• Family: 24%
• Doctors, nurses or social workers: 19%
• Building caretakers: 14%

Those caretakers, called gardiens, are a fixture of older Paris apartment buildings. They sort the mail, take out the bins, and usually know who comes and goes. Together with the friends and neighbors, they account for more than half of all the alarms, raised by people with no official job of checking on anyone. They checked anyway, though because the study covers only deaths, everyone they found had already died.

Of the 383 people whose living situation was known, 337 lived alone. The authors were careful with that number. Living alone, they wrote, did not mean having nobody.

The study has limits. It covers only deaths referred to the forensic institute, a small slice of the toll. Many bodies were found so late that doctors could no longer confirm heat as the cause. Across France, researchers estimated 14,802 more deaths between August 1 and 20 than in a typical year. That figure, called "excess deaths," measures how far deaths climbed above normal. It is not a count of death certificates that name heat.

Still, the study gets at something governments rarely ask after a disaster, which is who was actually paying attention.

The burial at Thiais

On September 3, 2003, 57 people were buried side by side at a cemetery in Thiais, just south of Paris. All of them had died in the heat, and no one had come forward to claim them. According to the Paris city council's record, each white stone carried a name. The Associated Press reported that an official read the names aloud while strangers left flowers. A TV news archive listing names President Jacques Chirac, Health Minister Jean-François Mattei and Paris Mayor Bertrand Delanoë among those present. (The Guardian counted 66 burials; the council says 57. I could not resolve the difference.)

Several hundred bodies had waited in Paris morgues for weeks while officials searched for relatives. Most were eventually claimed. These were the ones left over.

It would be easy to call them forgotten, but the record doesn't support that. Their names were known. Nothing in the documents shows that all of them lacked family or had lost touch with public services. What the city didn't have, during those August afternoons, was any way of knowing that these particular people were in danger. Knowing who someone is and knowing they need help right now are separate things, and in the years since, France has tried to connect them with a list.

"Our workers didn't go to those people"

That winter, France's National Assembly held hearings on what had gone wrong. The transcripts include testimony from the organizations that send aides into older people's homes, and they are some of the most candid documents in the whole record.

Emmanuel Verny, who led a national federation of home-help services, said his workers stepped up their visits in August, but mostly to people who were already clients. Older people who had been healthy enough to manage on their own could decline without anyone noticing, because, as he put it, "nos intervenants ne passaient pas chez ces personnes": our workers didn't go to those people.

Danièle Dumas, who led another home-care network, described a sheet kept in each client's home listing family members, aides and neighbors to contact. The practice predated the heat wave, she said; afterward, it was made easier for workers to find.

Delanoë testified that during the crisis the city contacted more than 8,500 older residents, working from records it already had: people with emergency alarm pendants, people getting meal delivery, people receiving home help. Caretakers in city-run senior housing checked on 4,000 residents twice a day.

These are people describing their own work, and no one independently checked who it saved. But their accounts line up. In an emergency, you call the people whose numbers you already have. An older man who had never needed an aide, a meal or an alarm pendant showed up in none of those files. So Parliament set out to build a file that would include him.

A list built on asking

The law signed June 30, 2004 did two main things. It required an emergency plan in every département (roughly France's version of a county), run by the prefect, the national government's representative there. And it required mayors to collect the name, age and address of residents "qui en ont fait la demande," meaning those who asked to be included. When the emergency plan was triggered, the list would be used to "organiser un contact périodique," to arrange regular contact.

A relative or neighbor could sign someone up, as long as that person didn't object. The list was confidential, and misusing it was a crime. As written, the law put a name in the register only when someone asked, either the person or someone acting for them. The state did not pull names from records it already held.

A decree that September filled in the details. Eligible people were those 65 and older living at home, those over 60 judged unable to work, and adults with disabilities living at home who received certain benefits. You could sign yourself up by letter, phone or email. Signing up someone else had to be done in writing. Within eight days, the mayor had to notify the person, explain that silence counted as consent, and tell them they could leave at any time.

The law says contact will happen. It does not say how often, what happens when no one answers, how fast someone has to show up, or whether anyone will offer a ride somewhere cooler. Each town decides that for itself.

The senator's question

In December 2004, six months after the law passed, Senator René-Pierre Signé put the obvious problem on the record. Mayors were having trouble getting the word out, and he doubted the lists would be reliable. His main worry was this:

In other words, the people most at risk are often the least likely to sign up.

The government replied in March 2005. It reminded mayors of their duty to inform residents and pointed out that family members or others could register someone on their behalf, adding: "la demande doit être faite par écrit." The request must be in writing.

That answer can read like a shrug, since it pointed to an option already in the law. I think it's fairer to look at the constraint the government was working within. The system rested on consent, in a country with strong protections around personal records. A written request leaves a record of who asked. The person being registered still had to be told, and could still refuse.

But look at who that answer handed the work to. Reaching the people least likely to sign up fell back on friends, neighbors and relatives, the same people who had done most of the noticing in 2003, and now they were being asked to do it on paper.

What a call looks like

Paris's list is among the best documented in France. A 2023 account from the city walks through how it works. People register by phone. Staff record contact details, whether the person lives alone, whether they have an alarm pendant, and the entry code for the building's street door, which matters a great deal in a city full of locked courtyards. One staff member, Vincent Merdjan, says some callers need reassurance about why the city wants to know all this.

When a heat wave is declared, phone staff begin calling people on the list within the first 48 hours. They go through a set of questions and give each person a risk score. Francky Lanimarac, who runs the call center, says a high score sets off medical or social follow-up. A city notice from the same year adds that city doctors and volunteer doctors handle urgent cases, and a social worker may visit.

What Paris doesn't publish is what happened after the calls: how many were made, how many were answered, how many led to a visit, and what those visits found. Even the size of the list is fuzzy. The same 2023 page says 9,600 in its headline and about 10,650 in the body.

Fréjus, a town on the Mediterranean coast, publishes the piece Paris leaves out, which is what it does when no one answers. Staff call registered residents regularly. If someone doesn't pick up, the town contacts the person named on their form, whether a neighbor, relative, friend or social worker, and may call the fire department. That's one town's procedure, and it doesn't tell us how often it worked.

Any program that moves people through stages can be counted at each one, and the count at the start rarely tells you much about the count at the end. A heat list has its own stages: eligible, registered, called, answered, helped. France's public record is strong on registration and thins out quickly after that.

Who ended up on the list

Paris's list grew from about 8,000 names in 2019 to somewhere around 10,000 to 10,500 in recent years. A 2021 city inspection report, written when the count was near 10,000, considered that low. Paris has 170,000 residents over 75, and 84,000 of them live alone. That comparison leaves out people between 65 and 74 and adults with disabilities, who are also eligible, so the real shortfall is bigger. The inspectors suggested asking people for permission to add them when they apply for APA, a benefit that helps older people pay for care at home. That suggestion will come up again.

Nationally, the numbers are softer. France's national audit office, the Cour des comptes, wrote in a 2024 report that fewer than 10% of vulnerable people were on the lists, without saying how it defined "vulnerable." It noted that towns struggled because registration was voluntary and because many re-asked residents every year. Senate notes from the same year put the figure at 10% to 15%. A 2019 survey by the national public health agency also found low registration, and officials worried that the people easiest to sign up were the ones already well informed or already surrounded by others. Few towns answered that survey, so it may not represent the country.

As far as the public record shows, nobody follows people through all of those stages and on to whatever happened next.

That gap matters for the most-cited evidence that France's heat response got better. A 2008 study estimated about 2,065 excess deaths in the July 2006 heat wave, compared with roughly 6,452 that would have been expected under pre-2003 patterns. The authors listed possible reasons: less vulnerability in the population, more public awareness, preventive measures, and the new warning system. They couldn't isolate what the list contributed. The study shows that far fewer people died than the old pattern predicted. It can't say why, or whether being on a list made a difference for any one person.

Heat deaths have continued. For summer 2023, the health agency estimated 5,167 heat-attributable deaths between June 1 and September 15, using a different method from the 2003 estimate, so the two figures can't be compared directly.

The 2026 rules add names from benefit files

This summer, France changed how names get onto the list.

A 2024 law allowed benefit records to feed the registers and widened what they could be used for: telling people about services, noticing when someone is losing independence, and reducing isolation. The working details came in a decree dated July 3, 2026. Three groups will now be added automatically unless they object:

• People receiving APA, the home-care allowance for older people
• Adults receiving PCH, a payment that covers disability-related support
• People receiving help with daily independence from regional pension funds

The first group is the one the Paris inspectors had in mind. In 2021 they proposed asking APA applicants whether they wanted to be added. The decree adds them unless they say no. For people in these three groups, getting on the list no longer starts with a request, which was the premise the 2004 law was built on.

AAH, France's main disability income benefit, isn't among the automatic sources. Adults with disabilities living at home have been eligible for the register since 2004, but for someone whose only benefit is AAH, the ways in are still the old ones: sign up, or have someone else sign you up in writing.

New recipients will find an explanation in their approval letter and have two months to object before their details go to town hall. Current recipients must be notified "Au plus tard le 31 janvier 2027," no later than January 31, 2027. That deadline is for sending letters, not for finishing the lists. Someone notified at the end of January still has two months to opt out, so their name could reach a mayor in spring at the earliest. The decree doesn't say when the lists have to be complete. The mayor then has to write to confirm and explain how to leave.

What gets shared is limited: name, contact details, which benefit the person receives, and any legal guardian. No diagnoses are included, and no assessment of how dependent someone is. France's data protection authority, the CNIL, cautioned that even the type of benefit reveals health information and needs tight safeguards. Once a name arrives, mayors may ask whether the person's home stays cool, which services come by, when they'll be away, and who to call in an emergency.

In an August 4 answer to a parliamentary question, the government described the goal as identifying and supporting vulnerable people living at home without waiting for each of them to apply. Twenty-one years after Senator Signé asked who would actually sign up, that is, in effect, the government's reply: the state will do some of the signing up itself.

For a lot of people who would never have filled out a form, that's a real change. The new names, though, come from files that exist because someone already applied for help at home. It's the same boundary Verny described to Parliament in 2004, when he said his workers didn't reach people who had never been clients. Someone who has never applied for APA has no file to send. And once a name reaches town hall, the law still promises only "regular contact," carried out however each town decides. The decree changes who is on the list but adds nothing about what happens between the list and someone's front door.

What Houston and Portland recorded

Texas, under a very different legal system, has tried to write that last step into law.

The state runs STEAR, a free, voluntary registry shared with local emergency planners. The state says plainly that local use varies and that signing up guarantees no particular service. But a Texas law that took effect in June 2021, four months after the February freeze, goes further for people designated "medically fragile." During extended power outages and declared disasters, the state has to arrange wellness checks: automated calls and texts, then a personal phone call, then, if that call goes unanswered, an in-person visit as soon as practicable and within 24 hours of the triggering event. On paper, that's a firmer commitment to reaching the door than anything in French law.

That raises the Texas version of France's question: how does a registrant end up with that designation? The sections I relied on spell out what the state owes designated registrants. In what I reviewed, they don't settle who decides that a person qualifies or how that person finds out. As with Paris's missing outcome numbers, the step I most wanted to see is not public.

After a windstorm knocked out power in Houston in May 2024, the city published how its STEAR checks went. There were 2,907 people on the list. Automated messages confirmed 1,346 were safe; 1,430 weren't reached at first. After 48 hours of phone calls, staff completed 765 in-person checks, and 455 people were still listed as unreachable at the end. The numbers don't fully add up, and they say nothing about lives saved. But every stage is there, from list to call to answer to knock, which is more than Paris has made public.

In Portland, Oregon, a county review of 2021 found 72 deaths officially attributed to overheating, 69 of them tied to the late-June "heat dome." Seventy-eight percent of those who died were 60 or older. Of the 68 who had stable housing, at least 48 lived alone, and for 11 more it wasn't known. Only ten case files mentioned air conditioning; in at least seven, the unit was unplugged or broken. The review doesn't say whether any of them were on a list, or whether anyone tried to reach them.

Who knows where you live

In all the French documents I drew on for this story, the people on the list never speak for themselves. We hear from federation heads, a mayor, call-center managers, inspectors, auditors and ministers. The closest a registrant comes is secondhand, in Merdjan's remark that some callers want to know why the city is asking.

I keep coming back to that remark. I was 13 in October 2003, living in Fruitvale, in East Oakland, when a fire came close enough that the adults around me started organizing. (It was twelve years after the hills firestorm most people picture when they hear about an Oakland fire.) The people who made sure undocumented families on our block weren't left behind weren't working from any government roster. They knew who lived in which unit, who had a car, and whose grandmother couldn't manage the stairs. Some of those families had good reasons to stay off official lists.

The French record holds smaller traces of the same hesitation: callers who need reassurance, a data authority warning that the name of a benefit reveals something about a person's health, and a new system that only works as long as people choose not to object. Whether names go on by request or by default, a list can only reach people willing to be known to whoever keeps it.

For 23 years, the French state has been trying to write down the kind of attention that found people in 2003. Dumas's network kept a sheet of relatives' and neighbors' numbers in each client's home. The 2004 rules let a relative or neighbor register someone, in writing. Paris records the code to the building's street door. In Fréjus, when a registered resident doesn't answer, staff call the person named on the form, who is in effect being asked to go over and knock. That is what happened in 2003, when, for everyone in the forensic study, the knock came too late. French law has never required that kind of attention; it has tried to enlist it, with a name and a phone number on file ahead of time. Whether the knock now arrives in time, the public record doesn't say.

If you live somewhere that gets dangerously hot, or where the power goes out in summer, France's experience leaves a few questions worth carrying around:

• If the power went out tonight, which list would you be on, and what is whoever keeps it actually obligated to do?
• Who in your building isn't on any list?
• On the forms you've filled out, whose name is in the box marked "emergency contact," and do they know it's there?