Clues and uncertainties in the risk assessment of arsenic in drinking water.

On the basis of studies of the prevalence of skin cancer among users of As-rich well water in Taiwan, WHO experts recommended in 1984 a maximum As concentration of 50 microg/litre in drinking water. Since that time, a plethora of non-cancer as well as cancer effects has been observed in several other populations sustaining a chronic exposure to various As concentrations in drinking water. This prompted a revision of the standard and a provisional guideline of 10 microg/litre was recommended in 1993. While the uncertainty linked to the statistical inferences leading to the guideline are reduced by the fact that they are directly estimated from human data and result from extrapolations made relatively close to observed exposure levels, developed guideline depends strongly on the choice of the dose-response model (linear, quadratic, hockey-stick) and the accuracy of the exposure data. The potential exposure to As sources other than drinking water, dietary habits and genetic characteristics of the populations may also make more difficult the inference of a recommendation for As concentration in drinking water. Owing to the huge cost of strongly reducing the current As in water standard, many efforts are presently made to clarify the quantitative aspects of As-induced cancers, particularly at low dose levels. New data on the metabolism and carcinogenic mechanism of As in humans along with the results of epidemiological studies presently under way in several countries will help to reduce the uncertainty in the risk assessment of As.