We did not consider pollutant-pollutant or pollutant-nutrient interactions when important individual pollutants were selected. Some pollutants may interact and have synergy. A well-known example is cigarette smoking and asbestos on lung cancer [52], [53]. On the other hand, beneficial nutrients may mitigate toxic effects of pollutants. For example, people with higher intake of antioxidant vitamins, B-vitamins (folate and vitamin B12) or omega-3 fatty acids had lower effects of air pollution [54]–[56]. Conventional statistical approach that includes cross-product terms of two interacting factors may have low power and therefore effect estimates would be unstable. A recent study by Sun et al. [10] proposed statistical strategies to examine multi-pollutants and their interactions using a two-stage model. Other dimension reduction techniques may also work for estimating risk models when a large number of pollutants and their interactions exist. A planned future study accounting for pollutant-pollutant and pollutant-nutrient interactions is expected to improve the model prediction, and therefore, potentially the utility of ERS.
