Previous Page Table of Contents Next Page


4.3 Exposure assessment model, model parameters and assumptions

4.3.1 Introduction

The previous section compared and contrasted previous exposure assessments of S. Enteritidis in eggs. This section intends to describe a simple exposure assessment for the purpose of completing a hypothetical risk-characterization exercise. Results from the exposure assessment model described here were combined with the dose-response model described in Section 3.4 to yield the risk-characterization results for S. Enteritidis in eggs in Section 5.

The exposure model developed here combines and modifies the US SE RA and Health Canada exposure models described in the preceding section. Generally, where input types were similar, the Health Canada parameters were used. If an input type was missing in one model, then the other model’s parameters were used (e.g. the Health Canada model did not consider cooling constants, therefore these were specified by the US SE RA model). The exposure model structure was generally based on the US SE RA model.

It should be noted that this model is necessarily indicative of North American management practices, but it is not intended to reflect any specific country’s risk. The effects of different parameter settings have been evaluated within the context of this model (e.g. different flock prevalence levels, different storage times and temperature profiles). Such differences are intended to reflect a wide array of situations, some of which might be indicative of particular countries or regions. Nevertheless, there are undoubtedly exceptions in specific regions that render this a poor model for assessing risk. Despite such limitations, it is hoped that the general framework and analysis completed here is of some value to a country or region as they begin conducting their own risk assessment of S. Enteritidis in eggs.

4.3.2 Model overview

The general structure of the S. Enteritidis in shell eggs risk assessment is outlined in Figures 4.34a, b & c. The exposure assessment model consists of production, shell egg processing and distribution, and preparation modules.

The production module of the exposure assessment (Figure4.34.a) is concerned with predicting the fraction of contaminated eggs among the population of all eggs produced per unit time. This fraction is determined by considering the flock prevalence, the within-flock prevalence, and the fraction of eggs laid by infected hens that are contaminated with S. Enteritidis.

The shell egg processing and distribution module of the exposure assessment (Figure 4.34b) is concerned with predicting the amount of growth of S. Enteritidis in contaminated eggs due to storage and handling of eggs between the farm and retail or institutional storage. Growth within each step of this module is a function of the storage time, temperature and environment. Environment is reflected in the cooling constants (k) for each step. In contrast to the production module, which estimates a population fraction of contaminated eggs, this module simulates individual contaminated eggs.

The preparation module (Figure 4.34c) is concerned with the effects of egg storage, egg meal preparation (e.g. serving sizes, mixing of eggs together), and the effectiveness of cooking in reducing the amount of S. Enteritidis in contaminated eggs. As in the previous module, growth of S. Enteritidis during steps in this module is a function of storage time, temperature and the value of k. Furthermore, pooling practices influence the number of servings per contaminated egg, and product type and serving size influence the amount of S. Enteritidis per serving after cooking. This module also simulates individual contaminated eggs.

Figure 4.34a. Schematic diagram of production module.

4.3.3 Production

The production model is a simplification of the US SE RA and Health Canada models, and models in the following manner the likelihood that an egg is contaminated.

If a flock is infected, the fraction of S. Enteritidis-contaminated eggs among all eggs a flock produces (FEggs_Flock) depends on the fraction of hens that are S. Enteritidis-infected in that flock (FHen_Flock) and the fraction of eggs an infected hen lays that are S. Enteritidis-contaminated (FEggs_Hen). This is described as:

FEggs_Flock = FHen_Flock × FEggs_Hen

Equation 4.1

Among the population of all infected flocks in a region or country, the fraction of hens infected per flock varies. In other words, it is not true that every flock in a region or country contains exactly the same proportion of infected hens. Therefore, the input FHen_Flock should be represented by a variability distribution.

By definition, flock prevalence (Prev) describes the proportion of flocks for which FHen_Flock is >0%. If we know the flock prevalence, then we know that the fraction of flocks in which 0% of hens are infected (FHen_Flock = 0%) is 1-Prev. For example, if 60% of flocks are infected (Prev = 60%), then 40% of flocks are not infected and FHen_Flock is 0% for these flocks.

As a convention, one can represent a variable input in bold. If referring to a specific value from the variability distribution, the input will not be in bold. Therefore, FHen_Flock refers to a distribution and FHen_Flock refers to a particular value from that distribution that occurs with frequency f(FHen_Flock).

Mathematically, this convention means:

FHen_Flock = {FHen_Flocki, f(FHen_Flocki)}

Equation 4.2

where {} describes the set of all possible values of FHen_Flocki
and .

FEggs_Flock describes the flock-to-flock variability in egg contamination frequency for all infected flocks in a region or country. The expected value of this distribution is:

Equation 4.3

Equation 4.3 calculates the fraction of S. Enteritidis-contaminated eggs among all eggs produced in a region. Consequently, 1-EV[FEggs_Flock] equals the fraction of all eggs that contain zero S. Enteritidis (i.e. not contaminated) at the time of lay. Eggs that are not contaminated are not modelled further in this exposure assessment.

A flock is defined as a group of hens of similar ages that are housed and managed together. A farm may contain more than one flock if, for example, two poultry structures (e.g. buildings) exist on the farm and there is little commingling of the birds between the structures. In such a case, one flock on the farm might be affected with S. Enteritidis while the other is not.

Flock prevalence (Prev) is assumed to be a scalar value in this model, but three levels are evaluated: 5%, 25% and 50% (Table 4.20). Such a convention can be interpreted either as examining the effect of uncertainty about flock prevalence, or as examining the influence of different country or region situations.

Flock prevalence is the proportion of all flocks in a country or region that are infected. A flock is considered infected if S. Enteritidis exists in the flock or its environment. At any given time, there is a fixed proportion of flocks that are S. Enteritidis infected, but flock prevalence might theoretically vary according to season. For example, some flocks might only be infected during the summer. This could happen if the flocks became infected because of exposure to S. Enteritidis from, for example, migratory waterfowl that gained access to the flock. In this case, flock prevalence would theoretically increase in the summer, and be lower during the rest of the year. However, the available evidence suggests that most flocks are infected early in their production cycle and their likelihood of infection is independent of season. Furthermore, unless a flock manager specifically takes steps to rid the flock of S. Enteritidis it is unlikely that an infected flock will become non-infected during its lifetime. Therefore, the assumption that flock prevalence is constant across seasons seems reasonable.

Flock prevalence in a country might also vary from year to year. For example, it seems likely that S. Enteritidis flock prevalence in the United States of America was very low before the 1980s. Subsequently, S. Enteritidis became established in a substantial proportion of the United States of America commercial flocks and flock prevalence increased (although the lack of surveillance prior to recognition of the problem prohibits quantitative estimates). Survey evidence suggests that flock prevalence stabilized somewhat in the 1990s (Hogue et al., 1997). For the purposes of this risk characterization, it is assumed that a country is dealing with a S. Enteritidis problem that has stabilized and that control programmes are expected to commence in the near term. Nevertheless, if a country is in the early stages of an epidemic, it might be important to consider the future risk of illness for its population as the epidemic worsens and eventually stabilizes at higher endemic levels.

Table 4.20. Description of assumed production model inputs and parameters.

Production model inputs

Distribution

Parameters

Prev (Prevalence of infected flocks)

Uncertain scalar

5%, 25% or 50%

FHen_Flock (Percentage of infected hens within infected flocks)

Variable
(Lognormal)

Mean: 1.89%

S.D: 6.96%

FEggs_Hen (Prevalence of contaminated eggs from infected hens)

Uncertain scalar
(Beta distributed)

Alpha: 12

Beta: 1109

Within-flock prevalence (FHen_Flock) is the proportion of infected hens within infected flocks. Because evidence suggests that this proportion is not constant among infected flocks or even in the same infected flock across time, within-flock prevalence is a random variable in the model. A probability distribution was estimated for within-flock prevalence by statistical fitting to data cited in the USDA and Health Canada assessments (Hogue et al., 1997; Poppe et al., 1991). It was assumed that the Hogue et al. survey detected 76% of infected flocks. Therefore, the data were adjusted to indicate that 24% of infected flocks had within-flock prevalence levels less than this survey’s lowest observed prevalence (i.e. 0.33%). A statistical fitting software (BestFit®; Palisade Corp., Newfield NY) determined that a lognormal distribution best fitted the data (c2 = 0.66, P>0.90) (Table 4.20).

It was assumed that the fraction of contaminated eggs an infected hen lays (FEggs_Hen) is a scalar value. It is biologically plausible that this input varies during the period the hen is infected. Furthermore, the value is also likely to be influenced by the infecting strain of S. Enteritidis, the strain of hen and environmental and managerial factors. In the US SE RA, egg contamination frequency was modulated based on the class of flock (e.g. high or low prevalence, moulted or not moulted). In that risk assessment, within-flock prevalence was not explicitly modelled. Instead, empirical evidence concerning the proportion of contaminated eggs produced by infected flocks was used. In contrast, the exposure assessment model developed here for FAO/WHO explicitly includes the variability in within-flock prevalence, but assumes the frequency at which infected hens lay contaminated eggs is constant. Therefore fluctuations in egg contamination frequency between infected flocks - resulting from differences in S. Enteritidis strain, hen strain or environmental and managerial factors - are assumed to be reflected by the within-flock prevalence variability distribution (FHen_Flock). For example, when FHen_Flock is high, the egg contamination frequency from that type of flock is correspondingly high (and vice versa).

FEggs_Hen is derived from data cited in the Health Canada risk assessment, where 11 of 1119 eggs were found to be S. Enteritidis contaminated from naturally infected hens (Humphrey et al., 1989a) (Table 4.20).

The frequency at which infected hens lay contaminated eggs was compared with the US SE RA model inputs and outputs. The US SE RA model predicts that an average of 1 in 20 000 eggs produced in the United States of America is contaminated with S. Enteritidis. The average flock prevalence for that model was 37%. From this information and the within-flock prevalence described above, the frequency at which infected hens lay contaminated eggs was calculated using Equation 3. The answer, 0.7%, was the 11th percentile of a Beta distribution from the Humphrey et al. (1989a) data. Although those data may reflect a more virulent strain of S. Enteritidis than occurs in the United States of America, the US SE RA model results are reasonably consistent with the Humphrey et al. (1989a) results.

Figure 4.34b. Schematic diagram of shell egg processing and distribution module. k is cooling constant.

4.3.4 Shell egg processing and distribution

This part of the model combines the US SE RA and Health Canada inputs and structure, with most storage times and ambient temperatures based on the latter model. The US SE RA cooling constants (denoted as k) are used to model the transition of internal egg temperature given ambient temperature and time of storage, with input settings as shown in Table 4.21.

The values in Tables 4.21 and 4.22 arguably represent conditions in North America. The PERT distributions representing egg-to-egg variability in storage time, temperature and k values reflect the North American climate and local management practices. Other countries will have different ambient temperatures and times of storage. To examine the effect of the assumed variability distributions for time and temperature, the baseline parameter values in these default distributions were arbitrarily adjusted up and down by 10% and the adjusted distributions denoted as "elevated" and "reduced" time-temperature scenarios, respectively. These adjustments can be interpreted as effects of uncertainty about the true distributions or as different scenarios applicable to different countries or regions. In all simulations, the lowest temperature is truncated at 4.4°C to avoid excessive refrigeration or freezing of eggs.

Table 4.21. Shell egg processing inputs(1) used in the baseline scenario of the risk characterization exercise. These inputs are based on those used in the US SE RA and Health Canada models.

Inputs

Distribution

Number of S. Enteritidis per egg when laid

=ROUND(RiskTexpon(152,1,400) 0)

Initial temperature of egg (°C)

=37

Probability of yolk contamination

=RiskBeta(1,33)

Storage temperature before transportation (°C)

=IF(RiskBinomial(1,RiskUniform(0.9 0.95)), RiskPert(10,13,14),RiskPert(18,25,40))

Value of k

=RiskPert(0.0528 0.0800 0.1072)

Storage time before transportation (hours)

=RiskUniform(0,IF(RiskBinomial(1,RiskPert(0.6 0.7 0.8)), RiskUniform(56,84),RiskUniform(84,168)))

Temperature during transportation (°C)

=Storage temperature before transportation

Value of k

=RiskPert(0.0528 0.0800 0.1072)

Time for transportation (hours)

=RiskPert(0.5,2,8)

Storage temp. before processing (°C)

=RiskPert(11,13,14)

Value of k

=RiskPert(0.0528 0.0800 0.1072)

Storage time before processing (hours)

=RiskUniform(1,24)

Temperature addition at processing

=RiskNormal(5.6,0.56)

Temperature at processing (°C)

=RiskPert(15,20,25)

Value of k

=RiskPert(0.3300 0.5000 0.6700)

Time for processing (hours)

=RiskPert(0.1 0.2 0.5)

Storage temperature after processing (°C)

=RiskPert(11,13,14)

Value of k

=RiskPert(0.0053 0.0080 0.0107)

Storage time after processing (hours)

=RiskPert(12,48,168)

Transportation temperature (°C)

=RiskPert(7,10,32)

Value of k

=RiskPert(0.0660 0.1000 0.1340)

Transportation time (hours)

=RiskUniform(1, 6)

NOTES: PERT distribution has parameters RiskPert(minimum, most likely, maximum). Uniform distribution has parameters RiskUniform(minimum, maximum). Truncated exponential distribution has parameters RiskTexpon(mean, minimum, maximum). Beta distribution has parameters RiskBeta(number positive +1, number negative + 1). Binomial distribution has parameters RiskBinomial(number of samples, probability of positive).

Table 4.22. Shell egg storage distributions used in the baseline scenario of the risk characterization exercise. These inputs are based on those used in the US SE RA and Health Canada models.

Inputs

Distribution

Retail Storage

Retail storage time (hours)

=RiskTlognorm(7, 10, 1, 30)*24

Retail storage temperature (°C)

=RiskPert(4.4,7,12)

k value

=0.24

Home storage

Home storage time (hours)

=RiskUniform(0,RiskTlognorm(14,10,1,60)*24)

Home storage temperature (°C)

=IF(RiskBinomial(1,RiskPert(0.001 0.005 0.02)), RiskPert(15,20,25),RiskPert(4.4,7,12))

k value

=0.24

Institutional storage

Institutional storage time (hours)

=RiskUniform(12, 147)

Institutional storage temperature (°C)

=RiskPert(4.4,4.4,7)

k value

=0.24

Home pooling

Time post pooling (hours)

=RiskCumul(0,48,4 0.8)

Temperature post pooling (°C)

=RiskCumul(4.4,32,7 0.8)

Pool size

=ROUND(RiskPert(2,4,12) 0)

Institutional pooling

Time post pooling (hrs)

=RiskCumul(0,48,4 0.8)

Temperature post pooling (°C)

=RiskCumul(4.4,32,7 0.7)

Pool size

=ROUND(RiskUniform(6,48) 0)

Ingredient use

Home serving size

=RiskDiscrete({2,4,6,8,9,10},{0.0233 0.1938, 0.6047, 0.1473 0.0078 0.0233})

Institutional serving size

=ROUND(RiskUniform(6,48) 0)

NOTES: PERT distribution has parameters RiskPert(minimum, most likely, maximum). Uniform distribution has parameters RiskUniform(minimum, maximum). Binomial distribution has parameters RiskBinomial(number of samples, probability of positive). Truncated lognormal distribution has parameters RiskTlognorm(mean, stand deviation, minimum, maximum). Cumulative distribution has parameters RiskCumul(minimum, maximum, range of values, cumulative probabilities of each value in range). Discrete distribution has parameters RiskDiscrete(range of values, probability weight of each value in range).

One feature not examined in the earlier models is the possibility that some eggs might be contaminated in the yolk at lay. The US SE RA and Health Canada models assumed that all internally contaminated eggs were only contaminated in the albumen of the egg at the time of lay. More recent evidence makes it difficult to ignore the possibility of some yolk-contaminated eggs (Gast and Holt, 2000a, b).

Based on our understanding of the growth of S. Enteritidis in eggs, yolk-contaminated eggs theoretically would support immediate amplification of numbers of S. Enteritidis in eggs after lay. In contrast, albumen-contaminated eggs experience a lag phase during storage and processing until there is sufficient breakdown in the yolk membrane to allow access of S. Enteritidis organisms to critical yolk nutrients.

If all contaminated eggs were contaminated in the yolk, then one would expect enumeration of S. Enteritidis per egg to demonstrate very large numbers after just a few days of storage (even at room temperature). For example, according to some growth equations, one would expect 5.5 logs of growth after just one day at room temperature, yet the evidence from a limited number of naturally contaminated eggs suggests that none of 32 eggs stored up to 21 days at room temperature had levels of S. Enteritidis consistent with being yolk contaminated (Humphrey et al., 1991). Using this evidence in a beta distribution, it was calculated that 2.9% of contaminated eggs are contaminated in the yolk, on average. Nevertheless, this estimate is undoubtedly biased upwards. It assumes that before considering the data from the 32 eggs, we were uniformly uncertain about the prevalence of yolk-contaminated eggs. In other words, before consideration of the data, it was believed that this prevalence could be 0% with the same confidence that it could be 100%. Of course, the prior belief was truly more in favour of very low prevalences, but here it has been decided to ignore the effect of such prior beliefs. For a specific application, however, it is expected that a more informed input be used to estimate the prevalence of yolk-contaminated eggs in a particular region or country.

As shown in Figure 4.34b, this module is a series of steps during which S. Enteritidis can increase within a contaminated egg. The model used is described below.

Let Gi be the amount of growth during step i. Think of Gi as a multiplier of the organisms that were in a contaminated egg before step i. If there was no growth in the egg during step i, then Gi = 1.0 (or 0 logs of growth). If there was one log of growth during step i, then Gi = 10.

Mathematically, Gi can be represented as Gi = g(Ti,ti). In other words, growth in a step (e.g. storage before processing) is some function, g, of the temperature distribution, T, and the storage time distribution, t, for that step. The functional relationship is complex and involves the influence of storage time and temperature on yolk membrane breakdown time and the exponential growth rate (EGR) for S. Enteritidis in eggs. The algorithms for modelling these dependencies were discussed earlier.

The output of this module is a variability distribution for the number of S. Enteritidis in contaminated eggs. Let SE_egg be this variability distribution. Then

SE_egg = InitSE × G1 × G2 ×... × G6

Equation 4.4

where InitSE is the variability distribution for the initial number of S. Enteritidis in contaminated eggs at the time of lay, and G1 through G6 are the growth predicted to occur during the six steps of this module (i.e. from storage before transportation through to transportation after processing).

Using Monte Carlo methods, SE_egg can be estimated. This describes the variability in number of S. Enteritidis in contaminated eggs after shell egg processing and distribution.

Figure 4.34c. Schematic diagram of shell egg preparation module. k is cooling constant.

4.3.5 Egg products processing

The egg products processing stage is concerned with predicting the S. Enteritidis contamination in bulk liquid egg products before and after pasteurization. The US SE RA model was used to simulate the numbers of S. Enteritidis organisms remaining after pasteurization of 10 000-lb containers of whole liquid egg. Only S. Enteritidis contributed via internally contaminated eggs are considered in this risk characterization. As noted in Section 4.2.4, any modelling of Salmonella contamination from sources other than internally contaminated eggs is based on scant quantitative data. Furthermore, the available qualitative data on the occurrence of species or strains other than S. Enteritidis in bulk liquid egg does not explain the sources or transfer mechanisms involved.

The US SE RA did not originally consider the potential for growth of S. Enteritidis inside eggs prior to being sent for breaking and pasteurization. To model this growth, this report used the shell egg processing and distribution stage, but limited the total amount of time between lay and breaking. Consistent with United States of America data (Ebel, Hogue and Schlosser, 1999), it was assumed that 69%, 30% and 1% of eggs pasteurized were nest run, restricted and graded, respectively. Nest run eggs are not washed or graded before being sent to pasteurization. Restricted eggs are those washed and graded eggs that are found inappropriate for sale as shell eggs. These eggs include eggs with cracked shells, thin shells, eggs with internal blood spots, or eggs that are leaking their contents. Graded eggs are eggs that are deemed suitable for sale as shell eggs, but for some reason are re-routed to egg products plants (e.g. eggs returned from retail stores). Each of these types of eggs may be stored a variable amount of time before they are broken and pasteurized. Nest run, restricted and graded eggs are stored an average of 2, 5 and 11 days in the model.

4.3.6 Preparation and consumption

Generally, the inputs used to model preparation and consumption practices for eggs were those of the US SE RA model (Tables 4.22 and 4.23). Pathway probabilities were assumed constant for this analysis (Table 4.24).

At the beginning of the preparation module, contaminated eggs have some number of S. Enteritidis that is described as SE_egg. As shown in Figure 4.34c, the preparation module simulates each contaminated egg as it traverses one of several pathways to eventual consumption. A contaminated egg might go to retail (and eventually home) or institutional users. It might be pooled with other eggs or not be pooled. Growth can occur during any of the storage steps that a contaminated egg experiences. Growth is modelled as described for the storage and distribution module.

A contaminated egg might be served as an egg-based meal or as an ingredient. Therefore, the effect of cooking depends on which path it follows. The number of servings to which that egg contributes also depends on its pathway.

An output of the preparation module is a variability distribution for the number of S. Enteritidis per serving for each of the possible pathways, SE_servingj (where the subscript j refers to a specific pathway). For example, Monte Carlo methods are used to estimate SE_serving for the pathway in which eggs are consumed in the home, the eggs are pooled, the eggs are consumed in egg-containing dishes, and the eggs are thoroughly cooked. This is a variability distribution because the input to the preparation module, SE_egg, is a variability distribution, as are factors such as storage time and temperature, k values, and cooking effectiveness within the preparation module.

The penultimate output of the preparation module is the classic risk triplet, which describes the exposure risk, ER, for the population of egg consumers. This can be represented as

ER = {pathj,f(pathj),SE_servingj}

where the symbols {} represent the complete set of paths, pathj identifies a specific path, f(pathj) is the likelihood of that pathway among all possible pathways, and SE_servingj is the consequence of that path (i.e. the exposures resulting from contaminated eggs).

If one integrates across all the possible doses (i.e. SE_serving) to calculate their likelihoods within all the pathways, you derive the exposure distribution for the population (Expos). This is the ultimate output of the exposure assessment and is represented as

Expos = {dosei,f(dosei)}

which is a variability distribution for the dose of S. Enteritidis per serving ingested by the consuming population. This distribution is combined with the dose-response function to calculate the likelihood of illness, on a per serving basis, from S. Enteritidis in eggs. This integration occurs in the risk characterization exercise in Section 5.

Table 4.23. Shell egg cooking and post-cooking handling distributions used in the baseline scenario of the risk characterization exercise. These inputs are based on those used in the US SE RA and Health Canada models.

Inputs

Distribution

Home cooking - pooled

Fully cooked eggs (log reduction)

=RiskUniform(6,8)

Eggs cooked as ingredients (log reduction)

=RiskUniform(0,8)

Boiled (log reduction)

=RiskPert(0,1,7)

Fried (log reduction)

=RiskPert(0,4,7)

Scrambled (log reduction)

=RiskPert(0,6,7)

Home cooking - non-pooled

Fully cooked eggs (log reduction)

=RiskUniform(6,8)

Eggs cooked as ingredients (log reduction)

=RiskUniform(0,8)

Boiled (log reduction)

=RiskPert(0,1,7)

Fried (log reduction)

=RiskPert(0,4,7)

Scrambled (log reduction)

=RiskPert(0,6,7)

Institutional cooking - pooled

Fully cooked eggs (log reduction)

=RiskUniform(6,8)

Eggs cooked as ingredients (log reduction)

=RiskUniform(0,8)

Boiled (log reduction)

=RiskPert(0,1,7)

Fried (log reduction)

=RiskPert(0,4,7)

Scrambled (log reduction)

=RiskPert(0,6,7)

Institutional cooking - non-pooled

Fully cooked eggs (log reduction)

=RiskUniform(6,8)

Eggs cooked as ingredients (log reduction)

=RiskUniform(0,8)

Boiled (log reduction)

=RiskPert(0,1,7)

Fried (log reduction)

=RiskPert(0,4,7)

Scrambled (log reduction)

=RiskPert(0,6,7)

Post cooking storage

Home egg handling

Time (hours)

=RiskExpon(0.25)

Temp (°C)

=RiskUniform(4.4,32)

Home ingredient handling

Time (hours)

=RiskExpon(1)

Temp (°C)

=RiskUniform(4.4,32)

Institutional egg handling

Time (hours)

=RiskExpon(1)

Temp (°C)

=RiskUniform(4.4,32)

Institutional ingredient handling

Time (hours)

=RiskExpon(1)

Temp (°C)

=RiskUniform(4.4,32)

NOTES: Uniform distribution has parameters RiskUniform(minimum, maximum). PERT distribution has parameters RiskPert(minimum, most likely, maximum). RiskExpon distribution has a single parameter RiskExpon.

Table 4.24. Probabilities used in the baseline scenario of the risk characterization exercise. These inputs are based on those used in the US SE RA and Health Canada models.

Inputs

Probability

Egg goes to institutional consumer

=0.25

Home pooling

=0.02

Institutional pooling

=0.05

Home-pooled egg used as egg

=0.90

Home-pooled egg used as egg - undercooked

=0.33

Home-non-pooled egg used as egg

=0.90

Home-non-pooled egg used as egg - undercooked

=0.33

Institutional-pooled egg used as egg

=0.70

Institutional-pooled egg used as egg - undercooked

=0.33

Institutional-non-pooled egg used as egg

=0.90

Institutional-non-pooled egg used as egg - undercooked

=0.33

Home-pooled egg used as ingredient - not cooked

=0.02

Home-non-pooled egg used as ingredient - not cooked

=0.02

Institutional-pooled egg used as ingredient - not cooked

=0.30

Institutional-non-pooled egg used as ingredient - not cooked

=0.30

Cooking by boiling

=0.22

Cooking by frying

=0.49

Cooking by scrambling

=0.29


Previous Page Top of Page Next Page