Estimating the health effects of exposure to multi-pollutant mixture

Air pollution constitutes a major public health concern because of its ubiquity and of its potential health impact. Because individuals are exposed to many air pollutants at once that are highly correlated with each other, there is a need to consider the multi-pollutant exposure phenomenon. The characteristics of multiple pollutants that make statistical analysis of health-related effects of air pollution complex include the high correlation between pollutants prevents the use of standard statistical methods, the potential existence of interaction between pollutants, the common measurement errors, the importance of the number of pollutants to consider, and the potential nonlinear relationship between exposure and health.

We made a review of statistical methods either used in the literature to study the effect of multiple pollutants or identified as potentially applicable to this problem. We reported the results of investigations that applied such methods.

Eighteen publications have investigated the multi-pollutant effects, 5 on indoor pollution, 10 on outdoor pollution, and 3 on statistical methodology with application on outdoor pollution. Some other publications have only addressed statistical methodology.

The use of Hierarchical Bayesian approach, dimension reduction methods, clustering, recursive partitioning, and logic regression are some potential methods described. Methods that provide figures for risk assessments should be put forward in public health decisions.