Journal of Economic Behavior & Organization · forthcoming
Cold weather reduces emergency department visits in Mexico by more than heat increases them.
Daily records from the universe of public hospitals in Mexico, 2008 to 2021, show three things.
- Cold days cut emergency department visits by up to 9.1 percent on the day itself.
- Accounting for the following 30 days, the reduction from a cold day reaches 16 percent.
- Hot days raise visits by up to 3.7 percent, and the effect does not accumulate the way cold does.
Unlike mortality, the demand for emergency care responds to temperature almost linearly.
Luis Sarmiento, Francesco Pietro Colelli and Filippo Pavanello
What to do with this
Eight things that follow from the estimates and matter for planning emergency services, four for cold days and four for hot ones.
Cold days
Cold kills, but it does not fill emergency rooms.
Mortality has a well-documented U-shaped relationship with temperature: both cold and hot days raise deaths, in Mexico as elsewhere (Cohen and Dechezleprêtre, 2022). Emergency visits do not. An additional cold day reduces them by 9.1 percent on the day and by 16 percent accounting for the following 30 days, while an additional hot day raises them by up to 3.7 percent.
The cold surprise →The cold effect is not over on the day.
An additional day below 10°C reduces visits by 9.1 percent on the day itself and by 16 percent once the following 30 days are accounted for; most of the difference builds up over the first two weeks.
The response →Cold empties the emergency room, except for respiratory cases.
Accounting for 30 days, an additional cold day reduces visits in every diagnostic chapter but one, from 11 percent for endocrine and metabolic conditions to 83 percent for infectious and parasitic diseases. Respiratory visits are the exception: they fall 7.3 percent on the day and rise 34 percent cumulatively, consistent with infections spreading after cold exposure.
Conditions →The young stay away longest; the old barely change.
Accounting for 30 days after a cold day, visits fall 23 percent for ages 12 to 20 and 19 percent for ages 20 to 40, against a statistically insignificant 1.4 percent for the over-80s. Same-day reductions run from 7 to 12 percent across ages; the gap opens in the weeks that follow.
Who →
Hot days
Heat is a same-day effect.
An additional day above 30°C raises visits by 3.7 percent on the day and by 2.7 percent once the following 30 days are accounted for. The cumulative effect is slightly smaller than the same-day one, a sign that some of the visits are brought forward rather than added.
The response →Heat hits the young.
On a hot day, visits rise 4 to 5 percent for people under 40 and do not rise for those over 60; for the over-80s they fall 2 percent. Children, adolescents and young adults are the groups to plan around.
Who →Heat brings bites, bugs and spoiled food.
On a day above 30°C, same-day visits for contact with venomous animals and plants rise 16 percent, gastroenteritis 11 percent and foodborne intoxication 8 percent. Dengue responds to warm rather than hot days: 7 percent on a day at 25 to 30°C, building over the following month. These are large shares of small numbers; venomous contacts average 1.2 visits per 100,000 people per day.
Conditions →Equal visit counts do not mean equal bed demand.
Around 2050, roughly 18 percent of the projected excess visits attributable to heat require hospitalization, against 10 percent of those attributable to cold and a historical rate of 13 percent across all visits. Heat adds severe cases; the cold-related change is concentrated in low-severity ones.
Midcentury →
What this is
A companion to the paper, built for people who plan and staff emergency services rather than for referees. Every figure is an estimate from the published paper, shown with its confidence interval and downloadable as the numbers behind it. Nothing here is new analysis.
For the method and robustness checks, see the paper →Cite the paper
Luis Sarmiento, Francesco Pietro Colelli and Filippo Pavanello. Emergency department visits and temperature: Evidence from Mexico. Journal of Economic Behavior & Organization.
Data and code
The full replication package, including the code that produced every estimate on this site, and the archived data.
github.com/FPavanello/tmp_er_admissionsdoi:10.5281/zenodo.21873500