Air pollution and daily mortality in Rome, Italy.
OBJECTIVES: To assess the relation between several daily indicators of air pollution (particulates and gases) and daily mortality in the metropolitan area of Rome and in the central part of the city. METHODS: Time series analysis. The associations between daily concentrations of pollutants (particles, SO2, NO2, CO, O3) recorded by five fixed monitors and daily total mortality in the period from January 1992 to June 1995 were evaluated. The analysis included examination of the pollution effect on mortality by place of residence within the metropolitan area, by season, age, place of death (in and out a hospital), and cause of death (cardiovascular and respiratory disease). The Poisson model included loses smooth functions of the day of study, mean temperature, mean humidity, and indicator variables for day of the week and holidays. RESULTS: The mean daily number of deaths was 56.9 (44.8 among people > or = 65 years old). A mean of 36.3 deaths occurred in the city centre; 37.3 deaths a day were recorded in a hospital. Total mortality was significantly associated with a 10 micrograms/m3 increase in particles (0.4%) on that day (log 0), and with a 10 micrograms/m3 increase in NO2 at lag 1 (0.3%) and lag 2 (0.4%) (1 and 2 days before, respectively). The effect of particles (lag 0) and of NO2 (lag 2) on total mortality was higher among those living in the city centre (0.7% and 0.5%, respectively). The risk estimates were higher in the warmer season (1.0% and 1.1%, respectively), whereas no difference was found for those dying in or out of the hospital. The effect of particles was robust to a sensitivity analysis and to the inclusion of NO2 in the regression model. CONCLUSIONS: Increase in particulates and NO2, generated by the same mobile combustion sources, is associated with a short term increase in mortality in Rome. The effect is more evident among residents in the city centre, where the levels of exposure to pollutants recorded by fixed monitors are probably more reliable indicators of personal exposure.