Opens in a new tab

Heart failure has a weather problem. Health systems need to plan for it

A nationwide US study covering more than 1.5 million heart-failure hospitalizations found that cooler ambient temperatures were associated with higher admission risk among adults aged 65 and older, with substantial differences between climate regions. The findings do not prove that cold weather directly causes decompensation, and the timing of exposure mattered considerably in statistical models.

November 18, 2026
Editorial
Colder weather was associated with more heart-failure hospitalizations among older US adults, but the size of the association differed substantially between climate regions.DimaBerlin / Shutterstock.com

IPM Take

Cardiology has become increasingly sophisticated at predicting which patient will deteriorate.

It is much less accustomed to asking what the weather forecast will do to the ward.

A nationwide analysis of US Medicare data now adds substantial evidence that ambient temperature belongs in that conversation. Across more than 1.5 million heart-failure hospitalizations between 2007 and 2022, cooler temperatures were associated with increased hospitalization among adults aged 65 and older. The relationship was not uniform across the country, and some of the largest associations appeared in western and north-central regions rather than simply in the places with the coldest winters.

That matters.

Climate-health policy is often dominated by extreme heat, understandably so as global temperatures rise and heat waves intensify. But cardiovascular adaptation cannot become synonymous with heat adaptation.

For heart failure, cold may remain a major operational risk.

The study does not mean that every drop in temperature directly triggers decompensation. Exposure was estimated from residential ZIP codes rather than individual living conditions, acute infections and daily heating behaviour were not measured, and the association changed depending on how long researchers allowed temperature exposure to influence subsequent hospitalization.

But those limitations do not make weather irrelevant.

They make adaptation more complicated.

The policy lesson is not simply “warn heart-failure patients when it gets cold”. It is that climatic vulnerability depends on housing, heating, acclimatisation, geography, age, comorbidities and local infrastructure.

A patient living through 5°C weather in a poorly insulated home may experience a very different physiological exposure from someone experiencing the same outdoor temperature in a heated building.

If environmental conditions are contributing to predictable surges in heart-failure care, then warming assistance, housing quality, seasonal surveillance and hospital-capacity planning start to look less like social policy adjacent to medicine and more like cardiovascular prevention itself.

Executive Summary

A study published in JAMA Network Open examined the association between daily ambient temperature and heart-failure hospitalization among older US adults using a 50% national sample of fee-for-service Medicare beneficiaries.

Researchers identified 1,517,756 heart-failure hospitalizations among 996,312 patients between January 2007 and December 2022.

Patients were aged 65 years or older, with an average age of 78 years, and approximately 60% were women.

The study used a time-stratified case-crossover design. For each hospitalization, researchers compared temperature exposure around the event with non-event days for the same patient matched by weekday, month and year. This approach reduces confounding by characteristics that remain stable within an individual over time.

Daily temperature was estimated at residential ZIP-code level using high-resolution climate data.

For colder temperatures, the primary analysis examined cumulative effects across 14 days, using 75°F, approximately 23.9°C, as the reference temperature.

Nationally, temperature at the 25th percentile was associated with 16% higher cumulative odds of heart-failure hospitalization compared with 75°F. At the 5th percentile, the cumulative odds were approximately 12% higher.

Regional variation was substantial.

At each region’s fifth-percentile temperature, some of the largest associations were observed in:

West North Central: OR 1.62
West: OR 1.40
Northwest: OR 1.34
Southwest: OR 1.34

The relationship was therefore not simply a map of the coldest US regions.

Temperatures above 75°F were not meaningfully associated with increased heart-failure hospitalization nationally. The authors caution that this should not be interpreted as evidence that heat is harmless to people with heart failure or that heat does not affect cardiovascular mortality.

A major methodological caveat is the timing of the cold effect. When researchers examined only same-day or two-day temperature exposure, colder conditions were generally associated with fewer hospitalizations. The increased association emerged in models incorporating longer, seven-, 14- and 21-day periods.

The findings therefore support a delayed association between colder conditions and heart-failure admissions rather than a simple immediate trigger.

Why it matters

  • HTA bodies: Environmental interventions rarely fit conventional technology-assessment frameworks, but measures such as home heating support, weatherisation and targeted warning systems may affect healthcare utilisation among high-risk cardiovascular populations. Evaluation should increasingly consider avoided admissions and system-level benefits when assessing climate-resilience interventions.
  • Payers: If cold-weather periods predict greater heart-failure utilisation, insurers and integrated health systems could target high-risk patients with medication review, remote monitoring, early symptom outreach and support addressing heating insecurity before expected seasonal surges. The key challenge will be proving which interventions reduce admissions rather than merely identifying higher-risk weather.
  • Industry / innovation partners: Heart-failure remote-monitoring systems could integrate environmental exposure data with weight, symptoms, blood pressure, heart rate and other clinical information. But adding weather to an algorithm is useful only if it improves prediction or changes care. The next opportunity is prospective validation of climate-informed risk tools, not another dashboard displaying temperature beside patient data.

A weather forecast rarely appears on the cardiology problem list.

Perhaps it should.

Heart failure is unusually sensitive to changes in physiology that alter vascular resistance, fluid balance, heart rate and myocardial workload. A patient who is clinically stable under one set of conditions can decompensate when the cardiovascular system is placed under additional stress.

Temperature is one such stressor.

A new nationwide US study provides one of the largest analyses yet of how that environmental exposure relates specifically to heart-failure hospitalizations.

Researchers examined Medicare claims from 2007 through 2022 and identified more than 1.5 million acute heart-failure admissions among almost one million people aged 65 and older.

The scale matters.

Previous studies have linked temperature with cardiovascular mortality and heart-failure outcomes in individual countries and cities. What has been less clear is whether the association behaves similarly across the enormous climatic diversity of the United States.

It does not.

Nationally, cooler ambient temperatures were associated with greater odds of heart-failure hospitalization over the subsequent two weeks.

At temperatures corresponding to the 25th percentile of the national distribution, cumulative hospitalization odds were 16% higher than at the study’s 75°F reference temperature. At the fifth percentile, the increase was 12%.

The apparent decrease from moderate to more extreme cold is a reminder that environmental exposure-response relationships are not always linear and that percentile temperatures differ substantially across regions.

The regional results are arguably more important than the national average.

At fifth-percentile temperatures, the West North Central region showed a 62% higher odds of hospitalization relative to 75°F, while the West showed a 40% increase and both the Northwest and Southwest showed increases of around 34%.

That pattern complicates an intuitive assumption that the strongest effect should occur in the coldest part of the country.

It may instead reflect adaptation.

A population accustomed to prolonged winter conditions may have better heating infrastructure, housing insulation, clothing behaviour and public-health preparedness than a population experiencing less frequent but still physiologically important cold.

The same outdoor temperature therefore does not create the same exposure everywhere.

That distinction is central to climate-health policy.

Meteorology measures what happens outside.

Health outcomes depend partly on what happens between the weather station and the patient’s body.

Housing is part of that pathway.

So is heating affordability.

So are social isolation, mobility, neighbourhood infrastructure, access to shelter, income and the ability to modify behaviour during adverse conditions.

The study cannot measure all of those factors directly.

Its geographic variation suggests they matter.

Cold itself has several plausible cardiovascular effects.

Peripheral vasoconstriction helps conserve body heat, but it also raises systemic vascular resistance. For someone with impaired cardiac function, that can increase afterload and force the heart to work against greater pressure.

Cold exposure also activates sympathetic and neurohormonal pathways, can affect heart rate and blood pressure, and has been linked with changes in blood viscosity, thrombosis and inflammation.

Several of those mechanisms could plausibly contribute to heart-failure decompensation.

The timing observed in the study is particularly interesting.

This was not primarily a same-day phenomenon.

The researchers’ main model allowed cold exposure to influence hospitalization over 14 days. When they shortened the window to the same day or two days, the relationship changed direction and cooler weather was generally associated with fewer admissions.

When longer windows were used, the higher hospitalization risk emerged again.

That finding deserves more attention than a simple headline saying “cold increases heart-failure admissions”.

The statistical estimate depends on assumptions about how long environmental exposure takes to influence clinical deterioration.

A delayed effect is biologically plausible.

Cold-related vasoconstriction, blood-pressure changes, inflammatory effects, respiratory infection, changes in activity or gradual fluid accumulation could contribute to decompensation over several days rather than producing an immediate emergency.

But the lag sensitivity also means the result is not a simple immutable risk coefficient.

Different modelling choices matter.

That is one reason the study should be treated as strong observational evidence rather than proof of a direct causal mechanism.

Another limitation is how exposure was measured.

Researchers linked patients’ residential ZIP codes with 4-kilometre-resolution estimates of outdoor temperature.

That is far more detailed than relying on one citywide weather station.

It still does not establish what temperature an individual actually experienced.

An older adult may spend almost the entire day indoors.

Two neighbours can have identical ZIP-code exposure and radically different indoor temperatures if one has reliable heating and the other cannot afford to run it.

This is where environmental epidemiology begins to overlap with social policy.

If indoor exposure is what ultimately matters physiologically, energy insecurity becomes part of cardiovascular vulnerability.

A prescription cannot warm an apartment.

The authors themselves identify interventions such as warming centres, shelters, winter-clothing programmes, home weatherisation and heating assistance as potential components of climate adaptation.

These measures rarely appear in conventional discussions of heart-failure therapeutics.

Yet if they prevent physiological stress that precipitates decompensation, the boundary between social intervention and medical intervention becomes less useful.

That does not mean this study proves heating subsidies prevent heart-failure admissions.

It does not.

Prospective evaluation would be needed to demonstrate that specific programmes change outcomes.

The study identifies where such interventions might plausibly be tested.

A parallel 2026 nationwide analysis from Sweden strengthens the broader signal.

Using roughly 482,000 heart-failure hospitalizations, Swedish researchers found that cold spells and lower ambient temperatures were associated with increased risk of admission several days later.

Unlike the US study, the Swedish analysis also detected a small increase associated with higher temperatures during the warm season.

That difference is important.

Climate-health evidence is geographically dependent.

The US result should therefore not be turned into a universal statement that heat does not worsen heart failure.

The United States has widespread air-conditioning access compared with many countries. Behavioural adaptation, housing construction, humidity and baseline climate can all alter the relationship between outdoor heat and physiological exposure.

The US study’s highest observed temperatures also do not represent every extreme heat condition that populations may experience.

And hospitalization is only one outcome.

People exposed to extreme heat may die before reaching hospital, seek care through other pathways or experience cardiovascular effects that are not coded as acute heart-failure admissions.

The authors explicitly warn against extrapolating the absence of increased hospitalization to heart-failure mortality.

This distinction matters politically because climate debate can produce an unhelpful competition between heat and cold.

A warming planet does not abolish cold-weather cardiovascular risk.

Climate change alters distributions, variability and extremes rather than simply making every day warmer by the same amount.

Health systems have to prepare for the exposures that actually affect their populations.

The American Heart Association’s 2026 scientific statement on nonoptimal temperature makes that point broadly.

Both heat and cold have been associated with myocardial infarction, stroke, heart-failure decompensation, arrhythmias and sudden cardiac death. Vulnerability is shaped by age, chronic disease, socioeconomic conditions, housing, geography and the built environment.

The scientific statement calls for temperature-related cardiovascular risk to move more clearly into clinical practice, health-system preparedness and policy.

That is a significant conceptual shift.

Traditional cardiovascular prevention focuses overwhelmingly on characteristics attached to the patient.

Blood pressure.

LDL cholesterol.

Smoking.

Diabetes.

Weight.

Kidney function.

Ambient temperature is different because clinicians cannot lower it with a prescription.

The exposure is external, shared and often determined by infrastructure.

This forces cardiovascular prevention beyond the consulting room.

The new US findings also suggest that intervention should be region-specific.

The same national cold-weather alert threshold may have very different relevance in Minnesota, Arizona and California.

Public-health warning systems typically define extreme weather using local climatology for precisely this reason.

Clinical systems may eventually need to become equally local.

Hospitals already use seasonal models to anticipate influenza and respiratory demand.

Heart-failure services could potentially incorporate environmental forecasts when planning winter capacity or identifying periods when vulnerable patients may require additional support.

A health system might use upcoming temperature changes to trigger targeted outreach to older patients with recent heart-failure admissions, reinforce monitoring of daily weight and congestion symptoms, check medication availability or identify patients at risk of inadequate home heating.

That remains a hypothesis.

The study did not test such a programme.

But healthcare already acts on risk predictions considerably less intuitive than tomorrow’s weather.

The barrier may therefore be organisational rather than technological.

Environmental data are abundant.

The harder question is what clinicians should do when the forecast changes.

This is where health systems should resist the temptation to build climate-alert systems before demonstrating that the alerts improve outcomes.

A weather warning that generates another stream of electronic notifications without a defined intervention will contribute little beyond alert fatigue.

Climate-informed care needs an action attached to the signal.

That action may differ according to the patient.

Someone with advanced heart failure living alone in poorly heated housing may need a very different intervention from a well-supported patient living in a climate-controlled environment.

Precision medicine usually evokes genetics and molecular biomarkers.

Environmental precision may matter too.

The study also reveals why equity cannot be separated from climate adaptation.

Older adults with limited income may have less ability to heat or cool their homes.

People experiencing homelessness have even less control over exposure.

Patients living in poor-quality housing may face drafts, dampness and temperature instability that are invisible to outdoor meteorological datasets.

Rural populations may have longer journeys to emergency care.

Individuals with mobility limitations may find warming centres theoretically available but practically inaccessible.

The same forecast does not create the same risk.

That means a climate-informed cardiovascular policy based solely on telling patients to “stay warm” would reproduce the familiar mistake of treating structural constraints as individual behaviour.

The finding that nearly one in five heart-failure patients in the study was dually eligible for Medicare and Medicaid adds relevance to this discussion, although the study did not establish that socioeconomic disadvantage explained regional temperature associations.

The evidence remains observational.

Claims data also lacked important clinical information, including heart-failure phenotype, ejection fraction, cause of heart failure and symptom class.

The study therefore cannot tell whether patients with reduced or preserved ejection fraction respond differently to cold, or whether some clinical phenotypes are substantially more vulnerable than others.

Rapidly changing confounders were also not fully measured.

Respiratory viral infections are particularly important because they are seasonal, can precipitate heart-failure decompensation and may coincide with colder temperatures.

Daily heating and air-conditioning use were also unavailable.

Researchers adjusted for humidity, fine-particle air pollution and holidays, while the case-crossover design controlled for many stable individual characteristics.

Residual confounding nevertheless remains possible.

And the population itself was specific.

These were older fee-for-service Medicare beneficiaries.

The results should not automatically be applied to younger adults or other insurance populations.

But those caveats do not erase the consistency of the broader literature.

The US analysis now sits alongside evidence from Sweden, Japan, Spain and multinational cardiovascular mortality studies suggesting that temperature affects cardiovascular outcomes and that the magnitude and timing vary substantially by geography.

That should change how the field talks about environmental cardiovascular risk.

There is no single universal “dangerous temperature”.

There is no single climate response curve.

And a national average can hide the communities where the risk is largest.

This is precisely why climate adaptation has to become local healthcare policy.

For heart failure, the operational question is increasingly straightforward.

If hospitals can anticipate periods when environmental conditions are associated with more admissions, should they continue treating those surges as unpredictable?

The answer will require intervention trials, health-services research and better individual exposure measurement.

But waiting for perfect causal certainty before measuring climate-related healthcare demand would miss the direction in which the evidence is moving.

Cardiology already accepts that physiology changes with environment.

Health policy now has to decide what to do with that knowledge.

The weather forecast will never replace an echocardiogram. But if colder weeks reliably bring more older adults with heart failure through hospital doors, treating temperature as somebody else’s policy problem is becoming harder to defend.

Source & Evidence