(95% CI, 1.00–1.13) and 4% (95% CI, 1.00–1.08) higher odds of asthma and rhinitis, respectively. Although these studies evaluated the combined effects of multi-pollutants, their approaches did not account for the relative effects of individual pollutants on the phenotype of interest, that is, each pollutant was not weighted depending on its relative effect size. Our study aimed to obtain a more precise relative effect size of each pollutant on each lipid outcome by estimating the weights (regression coefficients) from a randomly split training dataset and then computed ERS in an independent validation dataset.
