Estimation of risk due to low doses of microorganisms: a comparison of alternative methodologies.

The log-normal, or log-probit, simple exponential and beta distributed effectiveness models were evaluated for their ability to describe experimental dose-response data for human exposure to waterborne bacteria and viruses. Each of the models was capable of describing at least some of the available data; however, the beta-distributed model appeared to be the most widely applicable. When used to extrapolate to extremely low exposure levels, divergent predictions are obtained for each of the three models. On the basis of this analysis, it is impossible to rule out the hypothesis that a single microorganism when ingested has the potential of inducing infection or disease.