Recently, several studies have examined multiple pollutants. Patel and colleagues adopted an approach widely used in analyzing high-throughput genotype data, genome-wide association study (GWAS), and proposed an Environment-Wide Association Study (EWAS) to examined wide ranges of environmental factors including toxic chemicals as well as nutrients in relation to type-2 diabetes [5], lipid profiles [6], blood pressure [7] and all-cause mortality [8] using data from the National Health and Nutrition Examination Survey (NHANES). This systematic approach avoided a potential bias from selective reporting of subsets of analyses, outcomes, and adjustments [6]. Another EWAS approach which examined 76 environmental and lifestyle factors in relation to metabolic syndrome was conducted in Sweden [9]. Although these EWAS studies have yielded intriguing results, the statistical analyses were still based on single pollutant approaches. Multi-pollutant models were not considered. Of note, unlike GWAS with millions of markers, current EWAS studies have a moderate number of exposures and are not really comprehensive or “ultra high-dimensional” in nature. Similarly, misclassification, measurement error, temporal variations, and incomplete exposure data are inherent challenges to an EWAS study that modern genotyping techniques have overcome in GWAS.
