Fig. 3.

Heat related mortality (OR, 95&CI and p-value for interaction) estimated from conditional logistic regression models used within a case-crossover framework. The error bars show odds of death in the best (grey) and worst (black) quartiles. Amongst the indicators of natural environment, the best quartiles show areas with higher urban vegetation or better living environments. In the indicators of built environment, the best quartiles are areas with lower indoor heat exposure, lower proportion of lone-occupier and socially rented households, and higher proportion of households owned outright. In the socio-economic variables, the best quartiles are areas with lower unemployment, lower income deprivation, lower proportion of non-native English speakers and higher levels of proportion of the population who perceive their health as ‘very good’.