Uncertainty in Models for the Assessment of Risk in Drinking Water

Abstract

The nature of uncertainty in models for the assessment of risk in drinking water is considered with reference to the estimation of human health risks from exposure to Cryptosporidium. Two types of uncertainty in model parameters are distinguished, namely variability and certitude. It is shown how the relative significance of each of these types of uncertainty can be assessed, and in particular the importance of identifying critical interactions between these types of uncertainty. A simple heuristic framework for so doing is demonstrated. This framework has general applicability to model-based inputs to risk assessment in a wide range of contexts. In terms of the specific Cryptosporidium case study in the current paper, it is shown that by far the most critical parameter is that representing oocyst concentrations in water at the consumer's tap. This finding is of considerable significance for public policy.