Differences on the effect of heat waves on mortality by sociodemographic and urban landscape characteristics Yihan Xu, 1,2 Payam Dadvand, 1,3 Jose Barrera-Gómez, 1,3 Claudio Sartini, 4 Marc Marí-Dell’Olmo, 3,5 Carme Borrell, 2,3,5 Mercè Medina-Ramón, 1,3 Jordi Sunyer, 1,2,3,6 Xavier Basagaña 1,3 ▸ Additional material is published online only. To view please visit the journal online (http://dx.doi.org/10.1136/jech- 2012-201899). 1 Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Catalonia, Spain 2 Department of Experimental and Health Sciences, Universitat Pompeu Fabra, Barcelona, Catalonia, Spain 3 CIBER Epidemiologia y Salud Publica (CIBERESP), Barcelona, Catalonia, Spain 4 Department of Primary Care and Population Health, UCL Medical School, London, UK 5 Agència de Salut Pública de Barcelona, Barcelona, Catalonia, Spain 6 IMIM (Hospital del Mar Research Institute), Barcelona, Catalonia, Spain Correspondence to Dr Xavier Basagaña, Centre for Research in Environmental Epidemiology (CREAL), Doctor Aiguader 88, Barcelona, Catalonia 08003, Spain; xbasagana@creal.cat Received 31 August 2012 Revised 18 January 2013 Accepted 4 February 2013 Published Online First 26 February 2013 To cite: Xu Y, Dadvand P, Barrera-Gómez J, et al. J Epidemiol Community Health 2013;67:519–525. ABSTRACT Background Mortality increases during heat waves have been reported worldwide. The magnitude of these increases can vary within regions according to sociodemographic and urban landscape characteristics. The objectives of this study were to explore this variation and its determinants, and to identify the most heat-vulnerable areas by mapping heat vulnerability. Methods We conducted a time-stratified case-crossover analysis using daily mortality in the Barcelona metropolitan area during the warm seasons of 1999–2006. Temperature data on the date of death were assigned to each individual, which were assigned to their census tract of residence. Eight census tract-level variables on socioeconomic or built environment characteristics were obtained from the census. Residence surrounding greenness was obtained from satellite data. The relative risk (RR) of mortality after three consecutive hot days (defined as those exceeding the 95th percentile of maximum temperature) was calculated via conditional logistic regression. Effect modification was examined by including interaction terms. Results Analyses were based on 52 806 deaths. The effect of three consecutive hot days was a 30% increase in all-cause mortality (RR=1.30, 95% CI 1.24 to 1.38). Heterogeneity of this effect was observed across census tracts. The effect of heat on mortality was higher in the census tracts with a large percentage of old buildings (RR=1.21, 95% CI 1.00 to 1.46), manual workers (RR=1.25, 95% CI 0.96 to 1.64) and residents perceiving little surrounding greenness (RR=1.29, 95% CI 1.01 to 1.65). After three consecutive hot days, mortality doubled in the most heat-vulnerable census tracts. Conclusions Sociodemographic and urban landscape characteristics are associated to mortality risk during heat waves and are useful to build heat vulnerability maps. INTRODUCTION Adverse health impacts of high ambient tempera- tures and heat waves have been consistently reported worldwide. 12 Prolonged, extremely high ambient temperatures may result in important increases in mortality and morbidity. For instance, the death toll attributable to heat exceeded 70 000 in Europe during the summer of 2003. 3 Heat waves are projected to be more frequent and severe in terms of intensity, duration and geographical extent. 4 Besides, population ageing, which has emerged as a major demographic trend, especially in developed countries, will lead to a growing susceptible population. 5 Thus, the health effects of heat waves constitute an important public health problem that may be aggravated in the future. Some studies have further examined how socio- demographic and built environment characteristics modify heat-induced mortality and have identified a number of high-risk subpopulations. Elders (≥65 years), especially those with pre-existing medical conditions, are among the most vulner- able. 16 Women have been reported to be more vul- nerable than men in some studies, although their higher longevity could explain part of the differ- ence. 6–8 The findings for socioeconomic status (SES), however, are less conclusive. While some studies reported an increase in heat-related mortal- ity for individuals with low SES, 8–12 others found no effect modification. 13 14 Higher air-conditioning ownership, which is also associated to higher SES, has also been reported as a protecting factor against heat-related mortality. 15–17 Apart from the sociode- mographic characteristics, some studies have also drawn attention to the role that urban landscape characteristics play during heat episodes. Higher surrounding greenness and lower surface tempera- ture are suggested to reduce heat-related mortality in previous studies. 18–22 The different results for SES in different popula- tions and the variation in characteristics of urban landscape from place to place suggest that heat vul- nerability may be highly contextual and cannot be extrapolated. In addition, local studies mapping the most heat-vulnerable populations can be helpful to provide targeted preventive interventions. 23 Although several studies have examined the excess mortality during heat waves in Spain, none of them have explored its variations by sociodemographic and urban landscape characteristics. 7 8 24–26 The aim of this study was to explore the effect modification of the heat-mortality association by sociodemographic and urban landscape characteris- tics across the Barcelona metropolitan area. In add- ition, we aimed to identify the most heat-vulnerable areas by mapping heat vulnerability. METHODS Study design and population We conducted a time-stratified case-crossover ana- lysis using daily mortality in the Barcelona metro- politan area from 1999 to 2006 during the warm seasons, which we defined as the period between 15 May and 15 October. 7 The Barcelona Xu Y, et al. J Epidemiol Community Health 2013;67:519–525. doi:10.1136/jech-2012-201899 519 Research report