Risk-based evaluation of Escherichia coli monitoring data from undisinfected drinking water

Drinking water regulations in the United States and elsewhere are based on the occurrence of fecal indicator bacteria. Though not meeting all the criteria of an ideal indicator, nonpathogenic strains of Escherichia coli (E. coli) are used worldwide as an indicator of potential fecal contamination for drinking water and for distribution systems. This is, in part, because real illnesses are related to human pathogens, such as E. coli O157:H7, whose presence may be predicted better by E. coli than by total coliform bacteria. Our objective was to estimate the number of E. coli O157:H7 illnesses attributable to drinking water exposures in the United States and the feasible relationships between positive occurrences of the indicator bacteria E. coli and E. coli O157:H7 in drinking water. Results of the modeling indicate that in undisinfected drinking water systems, the ratio of bacterial indicator E. coli positives to E. coli O157:H7 organisms is estimated to be between 6:1 and 90:1 with few model parameters accounting for the vast majority of the uncertainty. These results provide context for considering the potential public health implications of a positive E. coli result from routine monitoring of undisinfected drinking water.