How heat exposure is defined matters: evidence from multi-metric mortality estimates across Europe

Extreme heat is an escalating health threat, particularly as Europe warms faster than the global average. However, mortality estimates depend on how heat exposure is measured, because different metrics can capture different dimensions of heat, including intensity, duration, acclimatisation and combined thermal stress. We examined how alternative heat metrics influence estimates of heat-related mortality across five European sites representing diverse climate conditions: Heraklion (Greece), Bologna (Italy), Geneva (Switzerland), Augsburg (Germany) and Oulu (Finland). Using harmonised daily mortality and climate data for 2020–2023, we applied distributed lag non-linear models to estimate heat-mortality associations. Mortality data were population-based for all sites except Augsburg, where hospital-based mortality data were used. Continuous exposure-response functions were modelled for air temperature, UTCI and WBGT across the full observed exposure range. Binary heatwave indicators, including EHF, WSDI and percentile-based definitions, were analysed separately under the same framework. Results are presented first as cumulative relative risks (RRs) and then as attributable mortality burdens. RRs indicated stronger heat-mortality associations in Heraklion and Bologna, moderate associations in Geneva, and non-significant associations in Oulu or in the hospital-based Augsburg data. Attributable mortality burdens showed similar spatial patterns but also varied according to how heat exposure was defined and captured by different metrics. Older adults and women consistently emerged as the most vulnerable groups. Continuous models generally produced higher burden estimates than binary heatwave definitions, reflecting their ability to capture risk across the full range of heat exposure rather than during heatwave events. The findings demonstrate that the choice of heat exposure metric can influence mortality estimates, highlighting the importance of using multiple complementary heat metrics to characterise heat-related health risks. Reporting both RRs and attributable burdens can support fair comparisons across settings and help policymakers identify locally appropriate heat adaptation strategies. Not applicable.