Final Report to the Department of the Environment, Transport and the Regions


June 1998

Executive Summary

The overall objective of this report is to develop a risk assessment model for Cryptosporidium in drinking water supplies in the UK. Microbiological risk assessment (MRA) is the emerging methodology to predict the risks to drinking water consumers from the small numbers of pathogens breaking through into drinking water supplies. In the US, MRA has been applied to derive microbial standards for drinking water and to determine the amount of removal required by drinking water treatment.

MRA models work by describing the daily pathogen exposures to drinking water consumers and then translating those exposures into a risk by way of a dose-response curve. Limited dose-response data from human volunteer studies are available for C. parvum. The fundamental information needed to develop a risk assessment model for Cryptosporidium regards the risk from exposure to a dose of just a single oocyst. At present, this risk can only be estimated by low dose extrapolation of the dose response curve. This prediction varies hugely depending on the mathematical form chosen for the curve. Studies on the fundamental mechanism of pathogenesis may shed light on which dose-response curve is most appropriate and in particular whether there is some sort of threshold dose for oocysts.

Computer simulations presented here demonstrate that a threshold dose for oocysts would make the predictions of MRA critically dependent on the assumed statistical distribution of oocysts in drinking water. Intensive sampling studies in the previous contract (DoE Ref No:. EPG 1/9/82) showed that drinking water treatment removed some 94 - 98% of spores but increased the spatial association of those remaining. This could provide a mechanism whereby some consumers are exposed to higher doses of pathogen than others through drinking water. Further experiments performed in this contract confirmed those findings and also showed differences in the behaviour of different species of micro-organisms during treatment. In particular, the spatial distribution of bacteria (coliforms and plate counts) was little affected compared to the spores. Disappointingly, the key experiments performed with Cryptosporidium oocysts were inconclusive because of the very low numbers recovered in the filtered waters. However, this confirmed that when the doses are low, the distribution of oocysts is unimportant.

It is concluded that the predictions of MRA can depend critically on the assumed statistical distribution of oocysts, and always depend on the choice of dose-response. Spatial variation in pathogen counts is an important consideration in designing sampling programmes. Indeed, a programme that missed the rare high count samples in drinking water would cause the risk to human health to be underestimated.

Copies of this report may be available as an Acrobat pdf download under the 'Pre 2000 Reports' heading on the DWI website.