Model-based prediction contrasts across 22 countries indicate small pooled differences in self-rated physical health associated with annual temperature metrics, alongside variation in direction and magnitude across individual nations. The research evaluated 273,031 pooled observations collected between 2023 and 2024 through the Global Flourishing Study, integrating the survey records with climate data from the Coupled Model Intercomparison Project Phase 6.
The investigation examined the relationships linking annual mean temperature and annual temperature standard deviation to overall physical health. In framing the inquiry, the authors noted that environmental conditions are relevant to physical health, but associations between annual temperature metrics and subjective physical health outcomes have remained insufficiently characterized across diverse populations.
To analyze the data, the researchers employed a unified two-stage Extreme Gradient Boosting framework to evaluate predicted physical health scores under simulated climate shifts. The analytical framework assessed predicted health scores under an annual mean temperature increase of 1 °C alongside an annual temperature standard deviation increase of 0.1 °C. The modeling design distinguished between direct contrasts, which held generated symptom and behavior probabilities fixed, and total contrasts, which updated those probabilities.
For an annual mean temperature increase of 1 °C, the pooled direct prediction contrast was 0.00488 health-score points, with a 95 percent bootstrap interval from −0.00219 to 0.00887, while the pooled total prediction contrast was 0.00547 health-score points, with an interval from −0.00333 to 0.00975. For an annual temperature standard deviation increase of 0.1 °C, the pooled direct contrast was −0.00155 health-score points, with an interval from −0.00455 to −0.00037, and the pooled total contrast was −0.00054 health-score points, with an interval from −0.00420 to 0.00157. Both pooled indirect-contrast intervals included zero.
Statistical uncertainty was evaluated using 100 country-stratified native-grid cluster bootstrap draws that refit both stages with fixed hyperparameters. The authors observed that country-level contrasts varied in direction and that the magnitude and direction of these local contrasts depended on the chosen temperature data source and analytical specification. The researchers concluded that the findings provide an empirical basis for further research examining how local environmental and social conditions shape associations between temperature metrics and physical health.
Subject of Research: Associations between model-derived annual temperature metrics and physical health across 22 countries
Article Title: Associations between model-derived annual temperature metrics and physical health across 22 countries
Article References: Li, C., Mi, J., Keeley, A. R., & Managi, S. (2026). Associations between model-derived annual temperature metrics and physical health across 22 countries. Archives of Public Health. https://doi.org/10.1186/s13690-026-02133-5
Image Credits: AI Generated
DOI: 10.1186/s13690-026-02133-5
Keywords: physical health, temperature metrics, Global Flourishing Study, CMIP6, XGBoost, climate data, bootstrap analysis, environmental health, Associations, between, model-derived, annual
News Source: Ophelia Keating. (November 6, 2026). Study Evaluates Annual Temperature Metrics and Physical Health Across 22 Countries. Scienmag.
