Australia recorded one of the world’s highest rates of temperature-related deaths from heart disease, lung disease, mental illness and nervous system conditions, with cold weather doing most of the damage, a 10-country study has found.

The research, published in eClinicalMedicine [link here] analysed 46 million deaths across 1117 locations in Australia, Brazil, Canada, Chile, Mexico, New Zealand, the Philippines, South Korea, Taiwan and Thailand between 2000 and 2019.

It found that 2.03 million deaths over the study period could be attributed to non-optimum temperatures, equating to 4.38% of all deaths and 620 years of life lost per 100,000 population.

Cold weather accounted for the lion’s share of that burden, contributing to 1.55 million deaths and 463 years of life lost per 100,000 people. Heat was responsible for 0.47 million deaths and 157 years of life lost per 100,000.

“For most causes of death, moderate cold contributed the largest share of mortality,” the authors wrote, with extreme cold and extreme heat making smaller contributions.

Australia’s burden

Chile recorded the highest overall attributable fraction at 6.25%, followed by Canada at 6.00% and Australia at 5.75%.

But when the researchers broke the numbers down by cause of death, Australia stood out for chronic disease.

“Australia experienced the highest temperature-related mortality burden for most chronic diseases, highlighting the need for improved housing insulation and effective heating strategies to mitigate cold-related risks, as well as implementing urban cooling strategies and heatwave early warning systems to reduce heat-related mortality,” the authors wrote.

The three leading causes of temperature-related death across all 10 countries were cardiovascular disease (0.69 million deaths), respiratory disease (0.22 million) and neoplasms (0.15 million).

Mental health and nervous system hit hardest

By cause of death, mental disorders carried the highest attributable fraction at 6.53%, followed by nervous system disease at 6.40% and cardiovascular disease at 5.71%. Neoplasms had the lowest attributable fraction, at 1.71%.

Mental disorders also recorded the largest relative risk under extreme cold of any cause examined, at 1.095, followed by renal disease at 1.079 and cardiovascular disease at 1.074.

Under extreme heat, nervous system disease carried the highest relative risk at 1.118, ahead of infectious disease at 1.080 and respiratory disease at 1.074.

The authors pointed to a biological explanation linking temperature extremes to brain and mental health conditions.

“For mental disorders and nervous system disorders, thermoregulation relies on intact hypothalamic and autonomic function, which may be compromised in neurodegenerative and psychiatric conditions,” they wrote, adding that some psychotropic medications “interfere with thermoregulatory and cardiovascular responses”.

They also noted a possible psychological pathway. “Cold exposure may increase stress levels (e.g., due to discomfort or worry about debt and affordability), potentially causing psychological changes that impair immune, cardiovascular, and hormonal functions,” the authors wrote.

Three distinct patterns

The study identified three exposure-response curves. Most causes of death followed an inverse J-shape, with cold posing far greater risk than heat. All-cause mortality and respiratory disease followed a U-shape, with risk rising at both ends of the temperature scale. Injury, external causes and infectious disease followed a J-shape, with heat the bigger driver.

For injury and external causes, heat was responsible for almost all the temperature-attributable burden, with an attributable fraction of 3.12% compared with essentially no risk from cold. Moderate heat alone accounted for 2.59 percentage points of that figure.

For infectious disease, moderate heat carried the highest attributable fraction at 1.29%, followed by extreme heat at 0.70%.

The authors linked the heat-related infectious disease finding to mosquito-borne illness. “Heat exposure contributed to the majority of the mortality,” for infectious disease, they wrote, “potentially linked to favourable breeding conditions (e.g., rainfall and high humidity levels) during hot seasons, which prolongs vector lifespans and enhances viral replication rates.”

For injury and external causes, the authors invoked what they called the “temperature-aggression theory”, which “posits that high temperatures act as environmental stressors that increase aggression and violent behaviour by causing discomfort, sleep deprivation, and heightened sexual arousal”.

Men, women and age groups differ

The study found clinically meaningful differences by sex and age.

Men had significantly higher relative risks under extreme cold than women for all-cause, cardiovascular, respiratory and digestive mortality. Women, in turn, had significantly higher relative risks under extreme heat for all-cause and cardiovascular mortality.

People aged 60 and over had higher relative risks under extreme cold than younger people for every cause examined except respiratory disease, and showed particularly steep risk increases for nervous system disease and neoplasms.

By contrast, people aged 0–59 had a significantly higher relative risk under extreme heat for injury and external causes of death than older Australians, a finding the authors linked to “greater outdoor exposure, risk-taking behaviours, and occupational hazards” among younger people.

When the deaths happen

The timing of temperature-related deaths also differed sharply between heat and cold.

At extreme heat, mortality risk peaked on the day of exposure itself, with nervous system disease and mental disorders showing the sharpest spikes, before falling to zero within two to five days.

Cold-related deaths took longer to emerge and persisted for far longer. “At extreme cold temperatures (8.3°C on average), mortality risks peaked on lag day 3–5 and gradually declined until lag day 21,” the authors wrote, although cold-related deaths from neoplasms had a shorter lag of up to 10 days.

How it compares with other estimates

The 4.38% overall attributable fraction sits within the range of earlier multi-country work but below some global estimates. The authors noted that a 2015 study led by Professor Antonio Gasparrini reported country-specific attributable fractions ranging from 3.37% in Thailand to 11.00% in China using a similar framework, while a separate analysis by Zhao and colleagues put the global figure at 9.43%.

“For countries included in both analyses, our estimates were broadly similar, suggesting methodological consistency,” the authors wrote, attributing the gap with the global estimate to differences in geographic coverage, population structure and exposure weighting.

The authors acknowledged their study relied on grid-level temperature data rather than individual exposure measurements, “potentially leading to exposure misclassification and underestimation of the true effects”, nor could they adjust for individual-level factors such as comorbidities, socioeconomic status or behaviour.

Africa and the Middle East were not represented in the dataset, and low-income settings were limited, the authors added.