Is the midnight-to-midnight average concentration of pollutants an appropriate exposure index for a daily mortality study?