Spatial Variability of Heat-Related Mortality in Barcelona from 1992-2015: A Case Crossover Study Design.

Numerous studies have demonstrated the relationship between summer temperatures and increased heat-related deaths. Epidemiological analyses of the health effects of climate exposures usually rely on observations from the nearest weather station to assess exposure-response associations for geographically diverse populations. Urban climate models provide high-resolution spatial data that may potentially improve exposure estimates, but to date, they have not been extensively applied in epidemiological research. We investigated temperature-mortality relationships in the city of Barcelona, and whether estimates vary among districts. We considered georeferenced individual (natural) mortality data during the summer months (June–September) for the period 1992–2015. We extracted daily summer mean temperatures from a 100-m resolution simulation of the urban climate model (UrbClim). Summer hot days (above percentile 70) and reference (below percentile 30) temperatures were compared by using a conditional logistic regression model in a case crossover study design applied to all districts of Barcelona. Relative Risks (RR), and 95% Confidence Intervals (CI), of all-cause (natural) mortality and summer temperature were calculated for several population subgroups (age, sex and education level by districts). Hot days were associated with an increased risk of death (RR = 1.13; 95% CI = 1.10–1.16) and were significant in all population subgroups compared to the non-hot days. The risk ratio was higher among women (RR = 1.16; 95% CI= 1.12–1.21) and the elderly (RR = 1.18; 95% CI = 1.13–1.22). Individuals with primary education had similar risk (RR = 1.13; 95% CI = 1.08–1.18) than those without education (RR = 1.10; 95% CI= 1.05–1.15). Moreover, 6 out of 10 districts showed statistically significant associations, varying the risk ratio between 1.12 (95% CI = 1.03–1.21) in Sants-Montjuic and 1.25 (95% CI = 1.14–1.38) in Sant Andreu. Findings identified vulnerable districts and suggested new insights to public health policy makers on how to develop district-specific strategies to reduce risks.