Likelihood Ratio Analysis of Skin Cancer Prevalence Associated with Arsenic in Drinking Water in the USA

Abstract

Several published studies have examined skin-cancer prevalence in regions of the USA that have concentrations of arsenic above the USA drinking-water standard. Those studies with adequate exposure and health outcome data did not report any skin cancer cases. Because the USA's arsenic concentrations are relatively low compared to some other countries and the study populations are small, the absence of reported skin-cancer cases could be due to an absence of risk in USA populations or random variability from a predicted risk. We used the current EPA arsenic cancer slope factor (CSF) model, which is derived from Taiwanese populations exposed to high levels of arsenic, to predict the expected number of skin cancers. We then conducted a likelihood ratio analysis, which showed that a null hypothesis (no additional skin cancer risk from arsenic) was approximately 2.2 times more likely than the hypothesis that ingested arsenic causes the predicted rate of skin cancers. Although based on small numbers, our analysis of USA populations indicates that the CSF derived from arsenic exposure in Taiwanese populations may be an overestimate when applied to USA populations.