The investigation, led by Dian Wang of the First Affiliated Hospital of Zhejiang Chinese Medical University and colleagues, is what epidemiologists call an ecological study: rather than tracking individuals, it examines population-level patterns across entire countries. The team analyzed data from 174 countries and regions over a 31-year period, from 1990 to 2021, making it one of the largest attempts to date to characterize the relationship between heat exposure and infertility at a global scale. Infertility already affects an estimated 10 to 15 percent of couples of reproductive age worldwide, and fertility rates in many countries have fallen below replacement levels, raising concerns about demographic and socioeconomic consequences.

What sets this study apart technically is its choice of exposure metric. Most previous research on temperature and reproduction has relied on ambient air temperature or simplified thermal indices as proxies for what the human body actually experiences. The researchers instead used the Universal Thermal Climate Index, or UTCI, a physiologically grounded measure developed in biometeorology that integrates air temperature, humidity, wind, and radiation into a single equivalent temperature under standardized reference conditions. In other words, UTCI estimates the air temperature at which a person would experience the same physiological strain as under the actual, combined meteorological conditions. The index is considered robust across different climates, seasons, and spatial scales, from microenvironments to global assessments, and is already used in public meteorological services and health warning systems.

To quantify exposure, the team drew on ERA5-HEAT, the first global historical gridded dataset built on a human biometeorological index, produced by the European Centre for Medium-Range Weather Forecasts and available through the Copernicus Climate Data Store. The dataset provides hourly UTCI estimates at a resolution of 0.25 by 0.25 degrees of latitude and longitude. For each grid cell and each year, the researchers counted the number of days on which the daily maximum UTCI fell into four established heat stress categories: moderate, corresponding to UTCI values of 26 to 32 degrees Celsius; strong, from 32 to 38 degrees; very strong, from 38 to 46 degrees; and extreme, above 46 degrees. These counts were then averaged within national boundaries to produce country-level annual exposure estimates.

Infertility prevalence data came from the Global Burden of Disease study, which models age-standardized prevalence rates, or ASPRs, for male and female infertility as separate outcomes using a Bayesian meta-regression framework that integrates population-based reproductive surveys, mainly among married or cohabiting women. Because such surveys cannot directly identify the responsible partner, the GBD framework generates sex-specific estimates informed by epidemiological evidence and prior attribution patterns. The researchers also adjusted their statistical models for the Socio-demographic Index, a composite measure of income, education, and fertility, as well as for ambient air pollution exposures including particulate matter, nitrogen dioxide, and ozone.

The statistical machinery was correspondingly elaborate. The team applied Geodetector, a spatial analysis method, to test whether the number of heat-stress days explained geographic variation in infertility prevalence, finding modest but statistically significant spatial associations for both sexes. They then trained an XGBoost machine-learning model on 5,568 country-year observations and interpreted it using SHapley Additive exPlanations, or SHAP, which assigns each predictor an additive contribution to the model’s output. Heat-stress days ranked among the relatively more important variables explaining variation in infertility prevalence for both men and women. The core analysis used linear mixed-effects models with country as a random intercept, adjusted for the Socio-demographic Index, the three air pollutants, and calendar year, with variance inflation factors ranging from 1.56 to 2.91, indicating that multicollinearity was not a major concern.

The headline results are striking. Each additional day of very strong heat stress was associated with an increase in male infertility ASPR of 1.292 per 100,000 population, and each additional day of extreme heat stress with an increase of 1.654 per 100,000. For women, the corresponding figures were 2.126 and 7.133 per 100,000, meaning that a single extra day of extreme heat stress per year was associated with a female infertility burden more than four times larger than for men. Moderate heat stress was positively associated with female infertility but not significantly with male infertility, while strong heat stress showed no significant association with either sex. The researchers also constructed a weighted heat-stress score combining very strong and extreme heat stress days, using a cross-fitting strategy in which countries were randomly partitioned into ten subsets to avoid overfitting. Each one-unit increase in this score was associated with increases of 0.337 and 0.909 per 100,000 in male and female infertility prevalence, respectively.

The geographic pattern of the findings adds a troubling dimension of inequity. Globally, the average annual number of days with UTCI at or above 26 degrees was 239.84, but exposure varied enormously: countries in the low Socio-demographic Index group averaged 326.99 heat-stress days per year, compared with just 64.80 days in the high-SDI group. Heat-stress days were most frequent in tropical and subtropical regions, with Africa averaging 323.69 days and low-latitude countries 327.85 days, while Europe averaged only 64.09 days. Subgroup analyses showed that the associations between the heat-stress score and infertility were significant in low and low-middle SDI regions, in Africa, North America, and South America, and at low and middle latitudes. The authors suggest that higher-SDI regions generally have better access to cooling infrastructure, healthcare, and protective resources, while populations in lower-SDI areas may face greater challenges in accessing such measures. Interestingly, the strongest association appeared in low-middle rather than low SDI regions, possibly because infections account for a larger share of the infertility burden in the lowest-SDI settings, diluting the relative contribution of heat.

The results align with a growing body of biological and epidemiological evidence. Studies have reported that exposure to high ambient temperatures and heatwaves is associated with poorer sperm quality, including lower sperm counts, reduced motility, and higher prevalence of abnormal morphology, likely reflecting disruption of spermatogenesis, a process known to be highly temperature-sensitive. For women, higher ambient temperatures have been linked to reduced antral follicle counts and diminished ovarian reserve, potentially affecting follicular development and ovarian function. Heat stress has also been associated with abnormalities in thyroid-stimulating hormone levels in women, suggesting endocrine pathways may contribute to the reproductive effects of heat. A nationwide pre-conception cohort study involving over 14 million Chinese couples planning a pregnancy found that heat stress exposure was associated with a longer time to pregnancy, providing individual-level support for the population patterns observed here. The authors also note that women in certain contexts may face greater heat and dehydration burdens due to gender-specific patterns of daily activity, such as unpaid household labor and water-fetching responsibilities, which may help explain the stronger associations observed among women.

The authors are careful to emphasize the limitations inherent in their design. As an ecological study, the findings describe associations at the population level and cannot establish causation for any individual. Residual confounding cannot be fully excluded, and factors such as access to assisted reproductive technology and occupational heat exposure were not available for adjustment. Infertility estimates derive from modeled epidemiological evidence rather than individual clinical assessments, and national-level averaging of heat exposure may not capture spatial variation within countries. Nevertheless, the associations remained directionally stable across a battery of sensitivity analyses, including two-way fixed-effects models with Driscoll-Kraay standard errors, log-transformed outcomes, additional adjustment for population density, life expectancy, urbanization, and smoking prevalence, and lag analyses testing delayed effects. With the frequency and intensity of very strong and extreme heat stress projected to increase under ongoing climate change, the authors argue that reproductive health deserves a place in climate adaptation strategies and heat-protection planning, particularly in the low-income, low-latitude regions where the burden appears greatest.

Subject of Research: Environmental health

Article Title: Association between infertility prevalence and exposure to heat stress: an ecological study based on the Universal Thermal Climate Index

Article References: Wang, D., Qian, J., Zhang, X., Mo, C., Chen, Y., Si, Y., Wang, Y., He, J., & Ma, J. (2026). Association between infertility prevalence and exposure to heat stress: an ecological study based on the Universal Thermal Climate Index. Frontiers in Public Health, 14, Article 1948152. https://doi.org/10.3389/fpubh.2026.1948152

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1948152

Keywords: heat stress, infertility, climate change, Universal Thermal Climate Index, male infertility, female infertility, Global Burden of Disease, public health, ecological study, reproductive health, Association, between