Studies reported increased mortality during heat waves worldwide. While these studies
used daily central site temperature as the exposure variable, temperature is spatially
variable, and housing and landscape characteristics affect exposure. To investigate
acute and chronic effects of temperature on mortality we developed a spatio-temporal
model for temperature in Eastern Massachusetts for 2000 to 2004 and applied it to
a case-crossover study. We obtained daily temperature measurements from four different
data sources. Land use, elevation and census data were obtained. The prediction model
for daily temperature at an address included daily intercepts and smooth functions
of space, season-specific smooth functions of space, distance to coastline, elevation
and land use. Mortality data were obtained from Massachusetts Department of Public
Health. Residences of the 125,409 out-of-hospital-deaths were geocoded. A case-crossover
analysis was conducted with estimated temperatures at place of residence as exposure
variable. Piecewise linear splines of maximum temperature with knots at 4 and 27°C
allowed for non-linearity. We compared our results using temperature at Logan Airport
as the exposure variable. Each 1°C increase at temperatures between 4 and 27°C was
associated with an increased mortality risk of 0.3% (95% CI 0.1–0.4%) for temperature
at residence as well as at Boston Logan Airport. An increase in daily maximum temperature
at place of residence when above 27°C increased the risk of mortality by 1.6% (95%
CI 1.0–2.1%) per 1°C. Compared to this the association of mortality with maximum temperature
at Logan Airport was somewhat smaller. These effects were higher for subjects living
at residences with high location-specific average temperature level. The results suggest
that the observed associations between hot temperatures and mortality may be underestimated
when using central site monitoring data. Furthermore living in an area with hot average
temperature may increase the risk of mortality associated with hot temperatures.