A Multisite-Multivariate AQI and the Determination of New Threshold Values for Health Risk Categories

Assessing air pollution and determining its impact on human health have been the focus of many studies. Various air quality indices (AQIs) have been proposed and calculated on observed data: given that pollution is a phenomenon described jointly by several chemical variables, AQIs should combine the various pollutants, characterized by different orders of magnitude and units of measure, in such a way to provide a simple and comparable indicator properly describing air quality. A further problem arises when we need to aggregate pollution data measured at different monitoring stations located over a certain geographical area. In this case, we have a three-dimensional array containing daily data on diverse pollutants at multiple sites, and the aggregation procedure can be performed either by first reducing the space dimension, or by first managing the variety of the pollutants. When considering the reduction of data dimensions, principal component analysis (PCA) is a candidate statistical technique: three-way PCA can be applied symmetrically or asymmetrically, allowing an easier interpretation of the data structure. When we apply this technique on pollution data, we obtain a spatial synthetic aggregate AQI capable of avoiding the problems of ambiguity and eclipsicity (Swamee PK, Tyagi A, J Air Waste Manag Assoc J 49:88–91, 1999) in order to be reliable. Aggregation combining different pollutants at multiple sites leads to an AQI that will take on values measuring the conjoint effect of the pollutants over a given area, and therefore higher than the values observed for the single pollutant. Official environmental agencies indicate health risk categories for any single pollutant, but not for a possible combination of them, for which values representing new thresholds have to be generated by the same technique used for aggregating the observed data. The aim of the present paper is the suggestion of a procedure for the determination of proper health risk categories for the overall aggregated AQI. A reliable AQI is of utmost importance for studying associations with respiratory and cardiovascular diseases, as well as with Covid-19 spread.