Comparison of Risk Predicted by Multiple Norovirus Dose–Response Models and Implications for Quantitative Microbial Risk Assessment

Abstract

The application of quantitative microbial risk assessments (QMRAs) to understand and mitigate risks associated with norovirus is increasingly common as there is a high frequency of outbreaks worldwide. A key component of QMRA is the dose-response analysis, which is the mathematical characterization of the association between dose and outcome. For Norovirus, multiple dose-response models are available that assume either a disaggregated or an aggregated intake dose. This work reviewed the dose-response models currently used in QMRA, and compared predicted risks from waterborne exposures (recreational and drinking) using all available dose-response models. The results found that the majority of published QMRAs of norovirus use the F hypergeometric dose-response model with α = 0.04, β = 0.055. This dose-response model predicted relatively high risk estimates compared to other dose-response models for doses in the range of 1-1,000 genomic equivalent copies. The difference in predicted risk among dose-response models was largest for small doses, which has implications for drinking water QMRAs where the concentration of norovirus is low. Based on the review, a set of best practices was proposed to encourage the careful consideration and reporting of important assumptions in the selection and use of dose-response models in QMRA of norovirus. Finally, in the absence of one best norovirus dose-response model, multiple models should be used to provide a range of predicted outcomes for probability of infection.

Drivers of Microbial Risk for Direct Potable Reuse and de Facto Reuse Treatment Schemes: The Impacts of Source Water Quality and Blending

“…This is due to the fact that the Fractional Poisson disaggregated dose response model will predict higher risks at low doses compared to other available norovirus dose response models; risk curves for scenarios modeled by van Abel et al [60] using the disaggregated model show a flatter curve when the dose is below 100 gene copies compared to scenarios modeled using the aggregated model (see original reference Figure 1). It is worth noting that several studies have documented virus aggregation in natural waters [41,44,66].…”

“…RO had the most significant impact on risk reduction for trains (2) through (4) while MF and NF had the most significant impact on removal in train (5). However, it is noted that the benefit of RO removal is less influential for disaggregated norovirus risk estimates due to the caveats related to the norovirus-disaggregated dose response model described above [60]. A sensitivity analysis is shown in Table 5 for the DPR scenarios and supports that RO is an influential predictor of protozoan and virus risks (Spearman rank correlation coefficient ranging from −0.28 to –0.44 for Cryptosporidium , −0.07 to –0.32 for norovirus) but is less influential for Salmonella removal (−0.04 to –0.15).…”

“…In some cases, surrogates were used (such as MS2 or E. coli ); while this practice is appropriate for QMRA, additional research is needed to quantify the removal/inactivation of actual pathogens under full-scale conditions. In particular, the aggregation state of norovirus can have a large impact on risk estimates [60]. Additionally, the removal and inactivation of norovirus by various unit processes may be different depending on its aggregation state as has been noted in some bench-scale studies indicating the importance of solution chemistry [68].…”

“…The daily probability of infection for each pathogen ref was computed using dose response models provided in Equation (8) ( Cryptosporidium ), (9) ( Salmonella ), and (10) (norovirus):$P_{inf,daily,Cr}=1-e^{-r{D}_{Cr}}$where r is the exponential dose response parameter [9]; $P_{inf,daily,Sa}=1-{(1+\frac{D_{Sa}}{\beta })}^{-\alpha }$where α and β are parameters of the Beta-Poisson dose response model [9]; $P_{inf,daily,No}=P(1-e^{\frac{-D_{No}}{\mu }})$where Equation (10) describes the fractional Poisson model [59] and P and µ are parameters of the model describing the perfectly susceptible fraction of secretor status positive (Se+) individuals and mean aggregate size, respectively. Both aggregated and disaggregated forms of the norovirus dose response model are considered in order to provide a range of predicted outcomes as the selection of norovirus dose response model has been shown to yield meaningful differences in risk estimates at low norovirus doses [60]. Exposure and dose response parameters are summarized in Table 4.…”

Quantitative Risk Assessment of Norovirus Transmission in Food Establishments: Evaluating the Impact of Intervention Strategies and Food Employee Behavior on the Risk Associated with Norovirus in Foods

“…We took into account preexisting immunity of negative secretors (nonsusceptible population due to a lack of soluble blood group antigens that are believed to interact with the virus) (34) but did not include immunity associated with prior episodes of norovirus infection or the fact that genetic susceptibility factors of different norovirus strains may differ from what has already been described for the prototype virus. (32,42) Actually, the accuracy and applicability of this doseresponse model is still debated. (42)(43)(44)(45) Atmar et al (43) suggested that the 50% human infectious dose for norovirus could be higher, i.e., 2,800 GEC NoV for Se+ individuals.…”