Examining Statistical Rationale Underpinning Predictor Variables of Heat-Health Outcomes

Keywords: Heat Vulnerability, Land Surface Temperature, Ambient Temperature, Principal Component Analysis, Spatial Regression
Session Type: Virtual Paper Abstract
Day: Monday
Session Start / End Time: 2/28/2022 05:20 PM (Eastern Time (US & Canada)) - 2/28/2022 06:40 PM (Eastern Time (US & Canada))
Room: Virtual 77

Abstract

Heat hazards are accentuated by the triad of the Urban Heat Island (UHI), greenhouse-gas-induced climate change, and population dynamics with potentially devastating impacts on human health. Understanding the cumulative influence of this triad on health entails characterizing human exposure, sensitivity, and adaptive capacity, which currently lacks a robust conceptual framework yet is crucial for policymaking towards adaptation and mitigation. This commentary examines the methodological rationale for the derivation of heat vulnerability indices (HVIs) as predictor variables for heat-health outcomes. Decision criteria involving variable selection, additive and reductive techniques, weighting approaches, data transformations, scale choices, GIScience challenges, and spatial visualization options are evaluated. The implications of the statistical choices pursued and the underlying assumptions regarding datasets used are explored. In a case study of Maricopa County, Arizona, death reports for 2018-2020 offer a benchmark for evaluating the reliability of heat studies. Findings indicate that the nature and availability of socioeconomic, biophysical, and health outcomes datasets primarily influence methodological choices. Further, the threshold temperature, beyond which health outcomes accelerate, is indeterminate given disparate exposure indices. Reporting of heat-health outcomes should delineate residency and exposure sites, thus requiring different outdoors and indoors exposure indices. Creating synergies between HVI construction, reference data (e.g., census data), policymaking, and heat-health outcomes reporting are imperative in calibrating and validating HVIs. When assessing various heat-health outcomes, new studies should incorporate contextual effects of heat, personal heat load variables, and harmonize spatial scale with administrative scale while comprehensively characterizing exposure, adaptive capacity, and sensitivity independently and as composites.