What do thermal indices lack to help predict heat-related risk in elite Athletics?

Objectives This study aimed to characterise the frequency of exertional heat stroke (EHS), did-not-finish (DNF) and medical encounter (MED) outcomes in elite Athletics competitions, and to evaluate their association with environmental, individual and race-related variables. Particular attention was given to the wet bulb globe temperature (WBGT), a thermal index widely used by international federations to estimate athlete heat strain. Methods Data were collected from 4938 athletes of both sexes participating in 80 World Athletics races held between 2019 and 2024 in five disciplines: long distance track events, marathon, racewalking, cross-country and trail-running. Spearman rank coefficient correlations, Pearson χ2 tests and negative binomial (NB) regressions were used to explore associations between outcomes (EHS, DNF, MED) and environmental, individual and race-related variables. The predictive ability of these variables was assessed using leave-one-out cross-validation to identify the best-performing univariable and multivariable models. Results EHS frequency was 8/1000 on average and 16/1000 when the WBGT was above 28.0°C, independent of sex (p=0.297) or discipline (p=0.980). DNF and MED were higher in marathon, racewalking and trail-running than shorter disciplines (p<0.001). Among variables, the mean radiant temperature showed the strongest univariable association with EHS (R2=0.11, p<0.001), while WBGT association was weaker (R2=0.04, p=0.018). Predictive accuracy was limited, with the best model having an error of 0.5±0.7 EHS per race. Conclusion The risk of EHS is particularly high during elite endurance events in Athletics. While EHS was greater with higher WBGT and radiant temperatures, neither provided sufficient predictive accuracy for heat-related risk in elite endurance athletes. Future models should consider other factors such as heat acclimatisation and recent illness to enhance predictive ability.