Spatial Delineation of the Temperature-mortality Relationship During Extreme Hot Weather: a Case Study in Greater Vancouver, Canada

Climate change has increased the severity, intensity, and frequency of extreme hot weather events over the past century (Meehi and Tebaldi, 2004, Reid et al, 2009). Many of these events have been associated with excess mortality in urban areas, as documented in the United States (Semenza et al. 1996, Kaiser et al. 2001, Curriero et al. 2002), Europe (Filleul et al. 2006, Schifano et al. 2009), China (Huang et al, 2010), and Russia (Trenberth & Fasullo, 2012). Less severe impacts have also been documented in the Canadian cities of Montreal (Smargiassi et al. 2009), Toronto (Pengelly et al. 2007), and Vancouver (Kosatsky et al. 2012). Cities are at particular risk during extreme hot weather events due to the urban heat island (UHI) effect, which leads to intra-urban variability in heat exposure (Hart et al, 2009; Hawkins et al, 2004; Roth et al, 1989).

Although many studies have examined the temperature-mortality relationship, very few have examined spatial differences in the relationship across an urban area, which are likely driven by spatial variability in both heat exposure and population vulnerability. Furthermore, none have compared spatial differences in the widely-used land surface temperature (LST) with metrics that might better describe human health risks, such as apparent temperature, and few have examined the performance of heat vulnerability indices (Tomlinson et al, 2011).

Here we develop an approach that combines data from a single representative weather station with exposure and vulnerability maps in a temporally and spatially stratified case-crossover design. The objective is data-driven delineation of high risk areas via assessment of spatial differences in the temperature-mortality relationship across a city (Greene et al, 2011; Hattis et al, 2012; Hondula, 2012; Schaeffer et al, 2015; Smargiassi et al, 2009). We apply these methods to the metropolitan area of greater Vancouver, Canada, where an extreme hot weather event during the summer of 2009 was associated with 40% increase in mortality and more than 100 excess deaths over a 7-day period (Henderson & Kosatsky, 2012; Kosatsky et al, 2012). The specific objectives of our study are to: 1) assess spatial differences in the temperature-mortality relationship during extreme hot weather using maps of LST, air temperature, and apparent temperature; 2) assess spatial differences in the temperature-mortality relationship during extreme hot weather using maps of social vulnerability indicators; 3) evaluate whether areas at higher risk due to exposure and vulnerability have additive effects on the temperature-mortality relationship; and 4) describe how these methods can be used for emergency planning and public health protection during future extreme events.

Results showed that apparent temperature and air temperature maps were more useful than maps of LST for examining spatial differences in the temperature-mortality relationship for greater Vancouver. Measures of social and material deprivation were also associated with differences in the temperature-mortality relationship, with participation in the labor market being a key indicator for the study area. By combining maps of apparent temperature and labor participation, we were able to identify priority areas for emergency planning and public health protection.