The multi-year-counterfactual design: a proposed approach to identifying specific risk factors for extreme heat event mortality using administrative health data

Extreme heat events (EHEs) can result in large increases in population mortality. Factors such as older age, social isolation, and cardiovascular disease are widely accepted as leading to increased risk of EHE death. However, these factors are associated with increased mortality risk at any given time, and it remains unclear which are significantly more present among those who die during EHEs. The epidemiologic designs available to study EHE mortality have different strengths and limitations that may affect findings. For example, administrative case-only and case-control designs were used to assess the impacts of the deadly 2021 EHE in British Columbia, Canada (BC), with some important discrepancies between results. This study proposes a multi-year-counterfactual (MYC) design for isolating risk factors specific to mortality during a discrete EHE. The MYC case-only and case-control designs combine analyses during the year of the EHE with the same analyses in years with no EHE, to examine the change in risk associated with the EHE. Here, both MYC designs are applied to the 2021 event in BC. The MYC case-control study demonstrated that only schizophrenia and Plan C (an indicator of income assistance) were risk factors associated with EHE-specific mortality. The risk associated with all other factors during the EHE was similar to their baseline risk in previous years. The MYC case-only study found schizophrenia and Plan C were associated with the largest EHE-specific risks, but ischemic stroke, chronic obstructive pulmonary disease, and asthma were also associated with smaller EHE-specific risks. Follow-up analyses demonstrated that findings were more robust for schizophrenia and Plan C than other potential risk factors. These results highlight how epidemiologic design can affect understanding of population mortality during EHEs. The proposed MYC approach provides a framework for separating EHE-specific risk factors from more general risk factors for mortality using administrative health data.