Abstract
Ambient temperature effect on mortality varies between places and populations, suggesting the existence of effect modifiers for this association. This study analyzes the influence of geographic, urban, and socioeconomic factors on the ambient temperature effect on non-accidental mortality in the general and older adults population of Brazilian metropolitan areas, and on that associated with circulatory, respiratory, and other mortality in older adults. Effects of this association were estimated for each group in 42 locations using a generalized additive model combined with the nonlinear distributed lag model. A meta-analysis was then performed to estimate the effects at the national and regional levels. Meta-regression determined the influence of effect modifiers. Estimated relative risks of the temperature-mortality association varied between locations in the Brazilian territory. Heat effects on non-accidental mortality at the national level were 1.09 (95%CI: 1.04-1.15) and 1.13 (95%CI: 1.07-1.20) for the General and Older Adult groups, respectively. Cold effects were 1.26 (95%CI: 1.21-1.32) and 1.30 (95%CI: 1.24-1.36) for the General and Older Adult groups, respectively. We observed a greater effect of cold than heat in both groups. For all causes of death, effects of heat and cold were greater in the Southeast and South Brazil. Amplitude of the mean temperature was the factor that best explained the heterogeneity between locations, followed by latitude, income and schooling. Hence, implementing adaptive measures to reduce the ambient temperature effects on mortality depends on the profile of each location.