On the use of different climate counterfactuals in health impact attribution

New health attribution studies are emerging rapidly alongside advances in climate attribution methodologies. The increasingly complex research questions and the need for rapid estimates of climate change's contribution to recent observed events are driving the development of advanced frameworks in health impact attribution.  These contributions combine advanced methods in climate epidemiology with the latest developments in climate trends and extreme weather event attribution to estimate climate-related health impacts attributed to anthropogenic climate change. However, integrating methods and data from these two fields is not straightforward, as methodological assumptions and limitations may not always align. One key element is the definition of the counterfactual scenario. Currently, counterfactual climates can be generated using a broad range of methodologies and under different assumptions, potentially leading to contradictory findings. This contribution aims to highlight the main methodological considerations when combining epidemiological data and methods with counterfactual climate data. Using data from the city of Zurich (Switzerland) as a testbed, we compare heat-mortality estimates across different sets of counterfactual scenarios generated by various methodologies. For example, counterfactual data that mimics day-to-day variation in the actual climate, suitable for assessing single events, would require vulnerability estimates and daily observed mortality at the time of the event. While GCM-based simulated counterfactuals (e.g., DAMIP), appropriate for measuring the anthropogenic signal in impacts over long periods, can be combined with average vulnerability estimates for the whole period or for specific subperiods to capture adaptation. This contribution will identify the key methodological elements that must be aligned and help guide the researchers in defining the study design and selecting the data for their impact attribution assessments.