Which heatwave measure has higher predictive power to prevent health risks related to heat: EHF or GATO IV? Evidence from modelling Lisbon mortality data from 1980 to 2016

To prevent the risk associated with heat-related health, several countries and institutions have built heat-health warning systems (HHWS). An HHWS is designed to alert the general public and decision-makers about the danger of high temperature by triggering a series of actions that avoid adverse health outcomes. The comparison of the various HHWS is complicated because there is no universal quantitative definition to predict and define a heatwave. The slightest variability at the threshold of definition the heatwave can trigger considerable differences in the action plan, health service demand and the time the population at risk must prepare. The choice of the index influences the number of days of heatwaves and its characteristics, such as severity. Estimating the risk of mortality associated with heatwave is variable according to the indexes, and the selection of the threshold is essential to prevent the burdens of heat on public health. The aim is the comparison between two metrics to know, which has higher predictive power to prevent health risks related to heat. On the one hand, a new way of defining heatwaves that have generated high consensus worldwide - the Excess Heat Factor (EHF); on the other hand, the Generalized Accumulated Thermal Overload (GATO IV) – an opportunity to improve the existing Lisbon heatwaves surveillance system. Daily mortalities and air temperatures from 1980 to 2016 in Lisbon with both indexes are modelled using Generalized Linear Models, with the calculation of the predictive power of the models using ROC curves for two levels of mortality severity. It is concluded that for total mortality, both indexes were statistically significant. Though, for daily mortality in individuals with 65 years or older with all diseases of the circulatory and respiratory system, when considering both indexes together, GATO IV was the only index significantly predicting the impact of heatwaves on mortality. GATO IV metric seems to have the best statistical properties. Nevertheless, EHF also stands out as a good indicator to predict heat-related mortality in Lisbon.