PROBABILISTIC RISK ASSESSMENT OF GROUNDWATER ARSENIC

Risks to human health from groundwater arsenic hazard may be assessed directly through medical surveys or calculated through consideration of a combination of source, pathway and receptor models. Here we present the outline of a probabilistic risk assessment model used to calculate groundwater arsenic-attributable detrimental health impacts in one of the world's more severely impacted regions, West Bengal. Ways in which the basic model can be used to inform implementation of remediation strategies are noted and key uncertainties in the model highlighted. The importance of water and rice as exposure routes was found to vary widely, with rice becoming the major source of exposure in areas where arsenic mitigation has reduced drinking water arsenic to much below 50 ppb - we note that mitigation of drinking water supplies does little to reduce exposure through rice, other than to reduce the arsenic loading arising from cooking in arsenic-bearing water. Key model uncertainties include: (i) interpolation of groundwater arsenic concentrations in regions where the distribution is highly heterogeneous; (ii) lack of detailed knowledge of dietary habits and, in particular variations between different sub-populations; (iii) relatively poor dose-response relationships, especially for arsenic concentrations of less than 100 ppb in drinking water for many sequela for which arsenic exposure is a significant risk factor; (iv) lack of detailed quantitative knowledge of factors such as nutrition, genetic polymorphisms, abundance/presence of environmental mutagens. These all represent areas for productive future research. Future changes, as a result of human activities, in the distribution of arsenic hazard in groundwater and in agricultural soils is discussed. Lastly, the issue of risk substitution arising from some arsenic mitigation technologies has been considered through simple Quantitative Microbial risk Analysis (QMRA) - our study of a limited range of mitigation options suggests that pathogen-attributable risk were overwhelmingly related to end-of-use conditions, rather than to any particular arsenic mitigation technology, although our results cannot be universally extended to other areas and technologies.