Incorporating method recovery uncertainties in stochastic estimates of raw water protozoan concentrations for QMRA

The impact of incorporating recovery data on protozoan concentration estimates was investigated for Cryptosporidium and Giardia using a large dataset (n=99) of [oo]cyst assay results with paired recovery estimates. Stochastic [oo]cyst concentration was estimated using three approaches: I-no availability/consideration of recovery, II-limited recovery data, where sample recovery was considered as an independent random variable, and III-every [oo]cyst assay result was adjusted for a concurrently derived recovery estimate. Critically, Approach I underestimated [oo]cyst concentrations by about 100% compared to Approaches II and III, which were similar. The impact of dataset size on statistical uncertainty about the concentration estimate for Approach II was investigated; little improvement in parameter uncertainty was achieved beyond n=20. It is suggested that recovery data be incorporated into source water concentration estimates, especially when used to infer health risks to consumers, so as not to underestimate the risk. Where none is available, conservatively low recoveries should be assumed. When designing monitoring programmes, recovery data should be collected as a pair with [oo]cyst count data for an initial period at least, so that site-specific relationships between those parameters may be ascertained and incorporated into source water concentration estimates.