LAUSR

Wewere interested to read the paper by Liu and colleagues published in the in Environmental Pollution in the Mar 2016 (Liu et al., 2016). The authors purposed to examine the impact of metals exposure with diabetes risk among coke oven workers. The associations of 23 urinary metals levels with diabetes risk were evaluated separately using multivariable logistic regression. A significant associations were found between urinary copper and zinc with risk of diabetes (Liu et al., 2016). Although the study made noteworthy contribution to the area, somemethodological and statistical issues should be taken into account. The authors conducted separate logistic regression for each urinary metals, however, we think this type of analysis add nothing because in occupational setting, workers may be exposed simultaneously with many highly correlated multiple exposures. On other hand, These 23 heavy metals arise from a common source (occupational coke oven worker) and, thus, may share a common effect on diabetes risk. Thus, one of the most challenge in environmental and occupational epidemiology is analyzing the multiple exposures. Hierarchical Bayesian modeling allows the researcher to model multiple exposure measures while accounting for common effects, multicollinearity, measurement error and as well as misclassification (Hamra et al., 2014). Hence, we suggest the authors to re-analyze the study's data to clarify the true role of each urinary metals on diabetes risk.