Impact of humidity on heat-related hospitalization risk on a global scale: a multicounty time-series study

Although the impact of temperature on mortality is well documented, the global burden of temperature-related hospitalization remains underexplored. Additionally, the epidemiological literature contains contradictory evidence regarding the role of humidity in heat-related mortality. We aim to provide novel insights into vulnerability to heat and contribute to clarifying the role of humidity using a large multi-location hospitalization dataset.

We collected daily data on all-cause and cause-specific emergency hospital admissions from more than 209 locations in 33 countries in the Multi-Country Multi-City (MCC) network. We assess the risk of hospitalization associated with heat using multiple heat stress indicators, including daily air temperature, wet-bulb temperature, and apparent temperature. We calculate daily time series of heat-stress indices for each location using hourly climate variables from the ERA5-Land reanalysis dataset. We estimate city-specific associations using time-series regression with distributed lag non-linear models (DLNM) and pool the results using multivariate meta-regression. We then employ a generalized random forest to identify vulnerability profiles based on area-level factors (e.g., poverty, green space) and individual-level factors.

Preliminary results from Switzerland revealed distinct risk patterns by heat-stress indices and cause-specific admissions. We reported the relative risk (RR) at the 99th percentile of the temperature distribution compared to the minimum hospitalization temperature, along with 95% confidence intervals (95% CI). Regarding daily air temperature (T2m), we observed a protective association with cardiovascular hospitalization across all cities, particularly in Basel (RR 0.71; 95% CI 0.53-0.96) and Zurich (0.78; 0.61-0.99). However, when assessing wet-bulb temperature (Twb), this pattern reversed in Lausanne (1.13; 0.8-1.6) and Lugano (1.01; 0.68-1.5), suggesting a potential increased risk. For genitourinary causes, both metrics indicated increased risks in Lugano and Geneva. However, in Geneva, the risk decreased from 1.73 (1.04-2.88) with T2m to 1.64 (0.99-2.73) with Twb. In the next steps, we will replicate the analysis across more than 209 locations and examine how factors such as green space and individual characteristics modify the association between hospitalization risk and heat, humidity, and heat stress.

This research will provide a comprehensive global evaluation of the risk of hospitalizations associated with heat stress and assess the role of humidity. Our study can help improve understanding of how humidity affects temperature-related health risks and identify vulnerability profiles across different countries.