Influence of set-up conditions of exposure indicators on the estimate of short-term associations between urban pollution and mortality.

In the past few years many studies on air pollution and health based on time series have been carried out. Yet, this approach does not assess exposure to air pollution at an individual level but it is based on ambient concentrations measured by air quality monitoring networks. Questions on the estimates of exposure to pollutants have been raised, in particular the fact that background measuring stations only have been considered in the set up of pollution indicators. To assess the impact of exposure indicator characteristics on the results of time series analysis, two series (black smoke and sulfur dioxide, respectively) of exposure indicators to urban air pollution were set up taking into account a growing part of proximity measures (industrial sources) available in the studied urban area (Le Havre, France). For each pollutant, indicators distributions were almost similar, especially for black smoke. Whatever the pollutant, the most obvious heterogeneity could be observed between the 100% background indicator and the indicator including the arithmetic mean for all the stations (50% background stations and 50% proximity stations). Then the sensitivity of the associations between mortality and air pollution to these indicators was studied. These indicators did not show statistically significant differences in the estimated excess risk. Yet, confidence intervals were more statistically significant as the contribution of proximity stations was more substantial, in particular for SO2. To conclude, the use of proximity measurements did not influence dramatically on the mean estimates of the association between air pollution and mortality indicators in Le Havre. Therefore it does not seem relevant to include the data provided by the proximity stations in the urban exposure indicators within the context of the epidemiology monitoring system.