Application of quantile regression in environmental epidemiology

Abstract

Abstract Introduction. Among many problems present in studies evaluating associations between health conditions and exposure to ambient air pollution, there is the correlation between environmental factors. These issues are usually resolved by providing a correlation matrix for the parameters of interest. Aim. To explore correlations between environmental factors. Material and methods. As sample data we use environmental factors presented in Milan mortality data (Italy, 1980-1989) and emergency department visits for asthma in Windsor (Canada, 2004-2010). Here, we propose to use a series of quantile regression evaluations to emphasize and identify dependency among environmental factors. Results. This presentation outlines an important role to investigate the potential correlations among ambient air pollutants, weather factors, and the values of the Canadian Air Quality Health Index (AQHI). In environmental epidemiology studies, these components are usually used in a common statistical model. Their correlations affect the values of the estimated relative risks, odds ratios or other estimated health effects. The presented approach examines associations among the factors as well as changes in correlations along quantiles. The examples used in this study explain various environmental phenomena; for example, the negative relationship between ambient ozone and nitrogen dioxide. Conclusions. By a consequence, this work can aid in further developing policies aimed at reducing the health impacts of air pollution as it allows to identify highly correlated factors in the constructed models.