The investigation was built on the baseline survey of a large multi-center cohort spanning three Chinese provinces: Fujian, Hubei, and Yunnan. In total, 134,403 participants provided fasting blood glucose measurements and detailed personal information, allowing the researchers to classify each individual as having diabetes, prediabetes, or normal glucose metabolism. Rather than treating environmental exposure as a uniform background condition, the team estimated CO2 concentrations at a fine spatial resolution of 10 kilometers and heatwave frequency at 27 kilometers, anchored to each participant’s place of residence. This spatial epidemiological approach, which blends satellite-derived environmental data with individual health records, represents a significant methodological advance over earlier studies that relied on city-level or provincial averages.

The CO2 exposure estimates were derived using machine learning techniques, including a light gradient boosting machine model trained on data from the Orbiting Carbon Observatory 2 satellite, known as OCO-2. Heatwave exposure was characterized using the Universal Thermal Climate Index, a measure that accounts not only for air temperature but also for humidity, wind, and radiation, providing a more physiologically meaningful picture of heat stress than temperature alone. By averaging exposures over one- to four-year windows before the health assessment, the researchers captured the cumulative environmental conditions that each participant experienced in the period leading up to their clinical evaluation.

The statistical core of the study was a series of multilevel logistic regression models with random intercepts at the community level, a design that accounts for the fact that people living in the same neighborhood share environmental and socioeconomic conditions. After adjusting for a comprehensive set of individual-level covariates, including age, sex, education, smoking status, alcohol consumption, physical activity, diet, and other environmental pollutants such as fine particulate matter, nitrogen dioxide, ozone, and carbon monoxide, the results were striking. Elevated fasting blood glucose levels were associated with higher CO2 concentrations, with a regression coefficient of 0.09, and with increased heatwave frequency, with a coefficient of 0.22. More importantly, both exposures were linked to increased odds of diabetes itself: each increment in CO2 concentration corresponded to an odds ratio of 1.14, while increased heatwave frequency corresponded to an odds ratio of 1.48.

These effect estimates carry real public health weight when translated to population scale. An odds ratio of 1.48 for heatwaves means that people exposed to more frequent extreme heat events had nearly a fifty percent higher odds of diabetes compared with those in cooler conditions, after accounting for a long list of confounding factors. The CO2 association, though smaller per unit, is consequential precisely because atmospheric carbon dioxide is ubiquitous and rising everywhere; no one can opt out of exposure the way one might avoid a polluted street or contaminated water supply. The study’s authors argue that this ubiquity makes even modest per-person effects cumulatively enormous across billions of people living under warming, carbon-rich atmospheres.

One of the most intriguing aspects of the findings concerns the body mass index, or BMI, which the researchers tested as a potential mediator, a biological stepping stone through which environmental exposures might influence diabetes risk. Among participants who had experienced heatwave events, BMI mediated approximately 8.2 percent of the association between CO2 and diabetes, and about 2.9 percent of the association between CO2 and prediabetes. Among those who had not experienced heatwaves, BMI mediated roughly 11.5 percent of the CO2-prediabetes association. This pattern suggests that the pathway linking carbon dioxide to disordered glucose metabolism runs partly through weight gain or altered fat distribution, and that heat exposure modifies how strongly that pathway operates. While mediation percentages may appear modest, they represent a measurable biological signal in a field where environmental determinants of metabolic disease have been notoriously difficult to pin down.

The study also revealed that vulnerability to these climate-related exposures is not evenly distributed. The association between CO2 and diabetes was stronger among individuals with less than a high school education, those living in rural areas, and those residing in non-compact built environments, meaning sprawling, low-density settings where car dependence and limited access to services may compound environmental stressors. Meanwhile, the heatwave association was stronger among people consuming 400 grams or more of fruit and vegetables per day, a finding the authors note warrants further investigation but which may reflect complex interactions between diet, hydration, thermoregulation, and metabolic health. These subgroup patterns echo a broader theme in environmental health research: the people least equipped to adapt, whether through education, infrastructure, or urban design, often bear the greatest burden.

The researchers took extensive precautions to ensure the robustness of their results. They examined variance inflation factors to rule out problematic multicollinearity among the many covariates, conducted sensitivity analyses to test whether the associations held under alternative model specifications, and adjusted for a battery of co-pollutants to isolate the specific contribution of CO2 from that of conventional air pollution. The exposure models themselves were validated with root mean square error metrics, and the spatial patterns of CO2 and heatwave frequency across the three provinces were mapped to confirm that the exposure gradients were meaningful rather than artifacts of interpolation. Ethical approval was obtained from the relevant Chinese review boards, and all participants provided written informed consent in accordance with the Helsinki Declaration.

Why might carbon dioxide, a gas long considered biologically inert at ambient concentrations, influence glucose metabolism? The study does not settle this question, but it situates itself within a growing body of literature suggesting that elevated CO2 may affect human physiology through several plausible routes, including direct effects on plant food chemistry that alter nutritional quality, possible influences on human respiratory and acid-base balance, and indirect effects operating through the same climatic conditions that CO2 drives. The mediation findings pointing to BMI hint that metabolic regulation, energy balance, and possibly appetite or activity patterns may be part of the chain. Heatwaves, by contrast, have clearer mechanisms: extreme heat is known to induce physiological stress, dehydration, inflammation, and sleep disruption, all of which can impair glucose regulation and insulin sensitivity.

The authors emphasize that their study is observational and cross-sectional in its baseline design, meaning it captures associations rather than proving causation, and that residual confounding cannot be entirely excluded. Nevertheless, the consistency of the findings across multiple provinces, the individual-level exposure assessment, and the mediation analysis collectively strengthen the case that climate change is not merely an environmental or economic crisis but a metabolic one. The research team, led by Manyao Li and Lan Zhang of Wuhan University with senior authorship by Peng Jia, and conducted in collaboration with provincial centers for disease control in Hubei, Fujian, and Yunnan, argues that the work provides scientific evidence for climate change mitigation while advancing understanding of how planetary and personal health are intertwined. As CO2 concentrations continue their upward trajectory and heatwaves grow more frequent and intense, studies of this kind suggest that the global diabetes epidemic may be yet another front on which the cost of a warming world is being tallied, and that protecting vulnerable populations from heat and curbing emissions could yield measurable metabolic dividends for hundreds of millions of people.

Subject of Research: Associations of co-exposure to carbon dioxide and heatwaves with diabetes and prediabetes risk in China

Article Title: Co-exposure to CO2 and heatwaves and the risk for diabetes: a multi-center spatial epidemiological study in China

Article References: Li, M., Zhang, L., Shi, Y., Yang, S., Qin, K., Wang, Z., Li, C., Yu, B., Dai, S., Li, X., Shao, Y., Yang, S., Yin, C., & Jia, P. (2026). Co-exposure to CO2 and heatwaves and the risk for diabetes: a multi-center spatial epidemiological study in China. BMC Medicine. https://doi.org/10.1186/s12916-026-05197-w

Image Credits: AI Generated

DOI: 10.1186/s12916-026-05197-w

Keywords: climate change, carbon dioxide, heatwaves, diabetes, prediabetes, body mass index, spatial epidemiology, fasting blood glucose, China, public health, BMC Medicine, environmental health