Occupational and environmental regulatory standards are usually determined on the basis of data from high-dose exposures, either in humans or animals. Risk assessment models are applied to identify the level below which the risks are considered as acceptable. Even when the basis is human data, the validity of extrapolation is often questionable because the number of cases actually observed in the low-dose region is usually small, if any. Validation of the risk assessment model by using data on populations with low-dose exposures is desirable, but the lack of study power is the major concern in most cases. A meta-analysis combining data from more than one study may solve the problem, especially when a model overestimates risks associated with low-dose exposures. A maximum contaminant level (MCL) of arsenic in drinking water at 0.05 mg/l published by the US Environmental Protection Agency (EPA) was used as an example to demonstrate such a scenario. Because this MCL was derived by using data from a study on skin cancer in Taiwan, a validation was conducted by using data on Taiwanese with low-dose exposures. In comparison with the number of cases observed in four studies, the model was more likely to be invalid than to be valid at exposure levels below 0.17 ppm and overestimated the number of cases (11.08 vs. 5). Whereas the EPA has published a new MCL recently on the basis of new risk assessments on urinary bladder and lung cancers, re-visiting the validity of the old standard still provides insights for validating regulatory standards in the future.
All Science Journal Classification (ASJC) codes
- Public Health, Environmental and Occupational Health