Targeting the spatial context of risk factors associated with heat-related mortality via multiscale geographically weighted regression

Extreme heat events appear to be a major cause to weather-related human morality in much of the world. The association between heat stress and public health is recognized as a complex interplay of multifaceted factors. Effective policy-making and action plans require a better knowledge of where and which of those factors should be targeted for intervention. However, little research has separated the underlying scales of effect of key components or taken into account geographical context in an analysis of those factors, which could lead to misguided policy actions in heat health risk reduction. In a case study of Hong Kong, we use the most recent multi-scale geographically weighted regression (MGWR) methodology to narrow this gap. We find that via MGWR, a combination of global and local processes could produce a better fit for the risk of heat-related mortality. Explanatory variables can be divided into three groups: global variables (such as age, educational attainment, and socioeconomic status), intermediate variables that vary on a relatively small scale (such as work environment, place of birth, and language), and local variables (i.e. thermal environment, low income). These findings suggest the need for targeting spatial context to multi-dimensional factors associated with heat-related mortality and highlight the hierarchical policy-making processes and site-specific action plans.