The team assembled an extraordinary dataset. From inpatient discharge records spanning 2019 to 2024, they identified 2,598,149 cardiovascular admission events among 8,876,140 older adults, generating more than 11.4 million case-day and control-day records for analysis. Each participant was linked to a geocoded primary healthcare institution, which served as a proxy for where the person lived and was exposed to the weather. Environmental exposures were then assigned using ERA5-Land, a high-resolution land-surface reanalysis product developed by the European Centre for Medium-Range Weather Forecasts, which reconstructs daily mean air temperature, dew point, relative humidity, and fine particulate matter concentrations at fine spatial scales.

The study design was a time-stratified case-crossover analysis, an approach that is particularly well suited to transient triggers of acute events. In this framework, each person who experienced a cardiovascular admission serves as their own control: the weather on the day of admission is compared with the weather on nearby days of the same day of the week within the same calendar month, when that same individual was not hospitalized. This design automatically strips out confounding from stable personal characteristics such as age, sex, smoking history, and chronic disease status, because each person is compared only with themselves. It also neutralizes long-term seasonal trends, since comparisons are confined to narrow time windows.

To capture how temperature effects unfold over time, the researchers employed distributed lag non-linear models, a statistical framework that simultaneously estimates how the association between exposure and outcome varies with the intensity of the exposure and with the delay between exposure and response. The analysis examined cumulative associations over lag days 0 through 14, meaning the combined effect of temperature exposure over a two-week window following any given day. This is critical for cardiovascular outcomes, which can respond to cold exposure days or even weeks after the fact, as cold stress gradually promotes vasoconstriction, elevated blood pressure, increased blood viscosity, and the rupture of vulnerable atherosclerotic plaques.

The results reveal a strikingly non-linear relationship between temperature and cardiovascular risk. The fitted cumulative exposure-response curve reached its minimum at 28.5 degrees Celsius, the temperature at which admission risk was lowest. Relative to that optimum, the odds ratio for cardiovascular admission was 1.032 at minus 3.6 degrees Celsius, 1.064 at 0.2 degrees Celsius, and 1.015 at 32.0 degrees Celsius. But the most dramatic finding was at the other end of the mild range: the fitted maximum of the curve occurred at approximately 14.2 degrees Celsius, where the odds ratio climbed to 1.194. In other words, cool but not frigid temperatures in the mid-teens were associated with the highest relative increase in cardiovascular admissions among this population of older adults.

The lag structure of these associations differed markedly between temperature extremes. At 0.2 degrees Celsius, near freezing, the elevated risk emerged after approximately two days and persisted through lag day 14, a slow-burning effect consistent with the cumulative physiological burden of sustained cold exposure. At 32.0 degrees Celsius, the association was smaller and more immediate, appearing quickly and then fading, which aligns with the known acute cardiovascular stress of extreme heat, including dehydration, increased cardiac output, and heightened coagulability. This temporal asymmetry has practical implications: cold-related cardiovascular risk is a prolonged hazard that builds over days, while heat-related risk is a sharp spike concentrated in the hours and days immediately following exposure.

Beyond absolute temperature, the study broke new ground by examining directional temperature change, meaning whether the temperature was rising or falling over a short window, independent of the absolute level. After adjusting for concurrent mean temperature, relative humidity, and fine particulate matter, the researchers found that warming trends within the lag 0 to 3-day window were independently associated with cardiovascular admissions, with an odds ratio of 1.030 at a temperature rise of 8.5 degrees Celsius. This suggests that rapid upward swings in temperature, even from a comfortable baseline, impose an additional cardiovascular burden that cannot be explained by the temperature level alone. The authors are careful to characterize this finding as hypothesis-generating rather than definitive, noting that the effect size is small and requires confirmation in other populations and climates.

The physiological plausibility of these findings rests on well-established cardiovascular mechanisms. Cold exposure triggers peripheral vasoconstriction, raising blood pressure and increasing cardiac workload. It also promotes hemoconcentration and increases fibrinogen levels, tipping the blood toward a prothrombotic state. Sudden temperature changes may add an additional layer of stress by disrupting autonomic regulation of the cardiovascular system, particularly in older adults whose baroreflex sensitivity and thermoregulatory capacity are already diminished. Rapid warming, paradoxically, may trigger vasodilation and changes in blood pressure that destabilize vulnerable plaques or provoke arrhythmias in susceptible individuals.

The scale of the dataset lends considerable weight to the conclusions. With nearly 8.9 million older adults and more than 2.5 million cardiovascular admission events, the study has statistical power that few prior investigations of temperature and cardiovascular risk can match. The use of geocoded primary healthcare institutions as exposure proxies, rather than broad regional averages, reduces exposure misclassification. The linkage to elderly health management rosters, part of China’s Essential Public Health Services program, ensures that the cohort represents the full population of older adults enrolled in primary care across Henan Province rather than a selected sample.

Nevertheless, the authors are candid about the limitations of their work. Because the study estimated relative odds rather than absolute admission risk, the clinical and public health significance of the observed associations remains uncertain. The odds ratios, while statistically robust, are modest in magnitude, and the population-level burden of temperature-attributable cardiovascular admissions was not quantified. The directional warming finding, in particular, requires replication before it can inform clinical guidance or early warning systems. Still, the study adds to a growing body of evidence that climate variability, not just extreme heat or cold, poses measurable risks to cardiovascular health in aging populations, and it underscores the need for weather-sensitive public health strategies that account for both the level and the pace of temperature change.

Subject of Research: Short-term ambient temperature and directional temperature change and cardiovascular hospital admissions among older adults

Article Title: Associations of short-term ambient temperature and directional temperature change with cardiovascular admissions among older adults: a time-stratified case-crossover study

Article References: Zhao, Q., Guo, X., Dai, N., Zhao, Z., Ma, M., Wan, H., Zhao, Y., Miao, Y., Fan, J., Jin, T., Cao, T., Sang, Y., Ma, Z., Wu, J., Guo, Z., & Liu, R. (2026). Associations of short-term ambient temperature and directional temperature change with cardiovascular admissions among older adults: a time-stratified case-crossover study. BMC Medicine. https://doi.org/10.1186/s12916-026-05322-9

Image Credits: AI Generated

DOI: 10.1186/s12916-026-05322-9

Keywords: ambient temperature, cardiovascular disease, hospital admissions, older adults, case-crossover study, distributed lag non-linear models, temperature change, climate change, heat stress, cold exposure, public health, China