Estimating probability distributions of bacterial concentrations in food based on data generated using the most probable number (MPN) method for use in risk assessment

Distribution-based maximum likelihood estimation methods are preferred for estimating Salmonella concentration in chicken when contamination data are highly left-censored

Salmonella is a common chicken-borne pathogen that causes human infections. Data below the detection limit, referred to as left-censored data, are frequently encountered in the detection of pathogens. The approach of handling the censored data was regarded to affect the estimation accuracy of microbial concentration. In this study, a set of Salmonella contamination data was collected from chilled chicken samples using the most probable number (MPN) method, which consisted of 90.42% (217/240) non-detect values. Two simulated datasets with fixed censoring degrees of 73.60% and 90.00% were generated based on the real-sampling Salmonella dataset for comparison. Three methodologies were applied for handling left-censored data: (i) substitution with different alternatives, (ii) the distribution-based maximum likelihood estimation (MLE) method, and (iii) the multiple imputation (MI) method. For each dataset, the negative binomial (NB) distribution-based MLE and zero-modified NB distribution-based MLE were preferable for highly censored data and resulted in the least root mean square error (RMSE). Replacing the censored data with half the limit of quantification was the next best method. The mean concentration of Salmonella monitoring data estimated by the NB-MLE and zero-modified NB-MLE methods was 0.68 MPN/g. This study provided an available statistical method for handling bacterial highly left-censored data.

Quantitative microbiological risk assessment of nontyphoidal Salmonella in ground pork in households in Chengdu, China

Foodborne disease caused by nontyphoidal Salmonella (NTS) is one of the most important food safety issues worldwide. The objectives of this study were to carry out microbial monitoring on the prevalence of NTS in commercial ground pork, investigate consumption patterns, and conduct a quantitative microbiological risk assessment (QMRA) that considers cross-contamination to determine the risk caused by consuming ground pork and ready-to-eat food contaminated during food handling in the kitchen in Chengdu, China. The food pathway of ground pork was simplified and assumed to be several units according to the actual situation and our survey data, which were collected from our research or references and substituted into the QMRA model for simulation. The results showed that the prevalence of NTS in ground pork purchased in Chengdu was 69.64% (95% confidence interval [CI], 60.2-78.0), with a mean contamination level of -0.164 log CFU/g. After general cooking, NTS in ground pork could be eliminated (contamination level of zero). The estimated probability of causing salmonellosis per day was 9.43E-06 (95% CI: 8.82E-06-1.00E-05), while the estimated salmonellosis cases per million people per year were 3442 (95% CI: 3218-3666). According to the sensitivity analysis, the occurrence of cross-contamination was the most important factor affecting the probability of salmonellosis. To reduce the risk of salmonellosis caused by NTS through ground pork consumption, reasonable hygiene prevention and control measures should be adopted during food preparation to reduce cross-contamination. This study provides valuable information for household cooking and food safety management in China.

Comparison of Statistical Methods for Identifying Risk Factors for Salmonella Contamination of Whole Chicken Carcasses

ABSTRACT

The complexity of the food system makes analyzing microbiological data from food studies challenging because many of the assumptions (e.g., linear relationship between independent and dependent variable and independence of observations) associated with common analytical approaches (e.g., analysis of variance) are violated. Repeated sampling within an establishment introduces longitudinal correlation that must be accounted for during analyses. In this study, statistical methods for clustered or correlated data were used to determine how correlation impacts conclusions and to compare how assumptions associated with statistical methods impact the appropriateness of these methods within the context of food safety. Risk factor analyses for Salmonella contamination of whole chicken carcasses were conducted as a case study with regulatory data collected by the U.S. Department of Agriculture Food Safety and Inspection Service between May 2015 and December 2019 from 203 regulated establishments. Three models, generalized estimating equation, random effects, and logistic, were fit to Salmonella presence or absence data with establishment demographics and inspection history included as potential covariates. Beta parameter estimates and their standard errors and odds ratios and their 95% confidence intervals were compared across models. Conclusions drawn from the three models differed with respect to geographic region, whether the chicken establishment also slaughters turkeys, and establishment noncompliance with 9 CFR §417.4 (hazard analysis critical control point system validation, verification, and reassessment) in the 84 days leading up to sample collection. The results of this study reveal the need to consider clustering and correlation when analyzing food microbiological data, provide context for selecting a statistical method, and suggest that generalized estimating equation and random effects models are preferrable over logistic regression when analyzing correlated food data. These results support a renewed focus on statistical methodology in food safety.

HIGHLIGHTS

The Prevalence and Epidemiology of Salmonella in Retail Raw Poultry Meat in China: A Systematic Review and Meta-Analysis

Foodborne disease caused by Salmonella is an important public health concern worldwide. Animal-based food, especially poultry meat, is the main source of human salmonellosis. The objective of this study was to evaluate the prevalence and epidemiology of Salmonella contamination in raw poultry meat commercialized in China. Following the principle of systematic review, 98 sets of prevalence data were extracted from 74 publications conducted in 21 Chinese provincial regions. The random-effect model was constructed for subgrouping analysis by meat category, preservation type, and geographical location. The prevalence levels differed from high to low among raw poultry meat, including chicken, 26.4% (95% CI: 22.4-30.8%); pigeon, 22.6% (95% CI: 18.2-27.8%); duck, 10.1% (95% CI: 5.3-18.2%); and other poultry meat, 15.4% (95% CI: 12.0-19.5%). Prevalence data on the preservation type revealed that chilled poultry meat might be more likely to experience cross-contamination than non-chilled poultry meat in China. The distribution map of Salmonella for raw poultry meat showed that a higher prevalence level was found in the Shaanxi, Henan, Sichuan, and Beijing regions. All subgroups possessed high amounts of heterogeneity (I² > 75%). The scientific data regarding the differences in prevalence levels between meat category, preservation method, and geographical region sources might be useful to improve specific interventions to effectively control the incidence of Salmonella in poultry meat.

BIKE: Dietary Exposure Model for Foodborne Microbiological and Chemical Hazards

BIKE is a Bayesian dietary exposure assessment model for microbiological and chemical hazards. A graphical user interface was developed for running the model and inspecting the results. It is based on connected Bayesian hierarchical models, utilizing OpenBUGS and R in tandem. According to occurrence and consumption data given as inputs, a specific BUGS code is automatically written for running the Bayesian model in the background. The user interface is based on shiny app. Chronic and acute exposures are estimated for chemical and microbiological hazards, respectively. Uncertainty and variability in exposures are visualized, and a few optional model structures can be used. Simulated synthetic data are provided with BIKE for an example, resembling real occurrence and consumption data. BIKE is open source and available from github.

Modelling biogenic amines in fish meat in Central Europe using censored distributions

The aim of the study is to model concentrations of selected biogenic amines in various fish species (Atlantic salmon, Atlantic cod, striped catfish) bought in retail stores in Central Europe. Since the data contains non-detectable values, statistical methods for left-censored values from the exponential and Weibull distributions are applied and used to evaluate and compare the amount of biogenic amines in fish samples. Moreover, a risk of exceeding certain limits of biogenic amine concentrations to protect human health is determined. There are relatively high concentrations of putrescine, cadaverine and histamine in almost all fish species. Moreover, there was a significant difference in mean concentrations (distributions of concentrations, respectively) of histamine, tyramine and spermidine among the species.

Probabilistic model for estimating Listeria monocytogenes concentration in cooked meat products from presence/absence data

A quantitative probabilistic model was developed to estimate the concentration of Listeria monocytogenes in cooked meat products based on presence/absence data and an assumed zero-inflated distribution, i.e. zero-inflated Poisson (ZIP) or zero-inflated Poisson lognormal (ZIPL) distribution. The performance of these two distributions was compared in two data sets (data set A and B), which represented L. monocytogenes prevalence and concentrations in cooked meat products. In this study, L. monocytogenes contamination data consisted of 4.23% (8/189) and 4.17% (5/120) non-zero counts for data set A and B, respectively. The contamination level of L. monocytogenes, determined by the most probable number (MPN) technique, ranged from 3 to 93 MPN/g among 13 positive samples. The goodness-of-fit test indicated that the ZIPL distribution was better than the simpler ZIP distribution, when L. monocytogenes contamination levels on positive cooked meat samples illustrated large heterogeneity. Results obtained from ZIPL distribution showed that the logarithmic mean value of L. monocytogenes positive samples was 1.5 log MPN/g (log σ = 0.4) for data set A and B. This study provides an alternative probabilistic method when only qualitative data is available in Quantitative microbial risk assessment (QMRA), in particular if pathogen concentrations consist of large numbers of zero counts and represent high variability.

Recognizing Structural Nonidentifiability: When Experiments Do Not Provide Information About Important Parameters and Misleading Models Can Still Have Great Fit

In the quest to model various phenomena, the foundational importance of parameter identifiability to sound statistical modeling may be less well appreciated than goodness of fit. Identifiability concerns the quality of objective information in data to facilitate estimation of a parameter, while nonidentifiability means there are parameters in a model about which the data provide little or no information. In purely empirical models where parsimonious good fit is the chief concern, nonidentifiability (or parameter redundancy) implies overparameterization of the model. In contrast, nonidentifiability implies underinformativeness of available data in mechanistically derived models where parameters are interpreted as having strong practical meaning. This study explores illustrative examples of structural nonidentifiability and its implications using mechanistically derived models (for repeated presence/absence analyses and dose-response of Escherichia coli O157:H7 and norovirus) drawn from quantitative microbial risk assessment. Following algebraic proof of nonidentifiability in these examples, profile likelihood analysis and Bayesian Markov Chain Monte Carlo with uniform priors are illustrated as tools to help detect model parameters that are not strongly identifiable. It is shown that identifiability should be considered during experimental design and ethics approval to ensure generated data can yield strong objective information about all mechanistic parameters of interest. When Bayesian methods are applied to a nonidentifiable model, the subjective prior effectively fabricates information about any parameters about which the data carry no objective information. Finally, structural nonidentifiability can lead to spurious models that fit data well but can yield severely flawed inferences and predictions when they are interpreted or used inappropriately.

A Quantitative Risk Assessment of Human Salmonellosis from Consumption of Walnuts in the United States

We assessed the risk of human salmonellosis from consumption of shelled walnuts in the United States and the impact of 0- to 5-log reduction treatments for Salmonella during processing. We established a baseline model with Salmonella contamination data from 2010 to 2013 surveys of walnuts from California operations to estimate baseline prevalence and levels of Salmonella during preshelling storage and typical walnut processing stages, considered U.S. consumption data, and applied an adapted dose-response model from the Food and Agriculture Organization and the World Health Organization to evaluate risk of illness per serving and per year. Our baseline model predicted 1 case of salmonellosis per 100 million servings (95% confidence interval [CI], 1 case per 3 million to 1 case per 2 billion servings) of walnuts untreated during processing and uncooked by consumers, resulting in an estimated 6 cases of salmonellosis per year (95% CI, <1 to 278 cases) in the United States. A minimum 3-log reduction treatment for Salmonella during processing of walnuts eaten alone or as an uncooked ingredient resulted in a mean risk of <1 case per year. We modeled the impact on risk per serving of three atypical situations in which the Salmonella levels were increased by 0.5 to 1.5 log CFU per unit pretreatment during processing at the float tank or during preshelling storage or posttreatment during partitioning into consumer packages. No change in risk was associated with the small increase in levels of Salmonella at the float tank, whereas an increase in risk was estimated for each of the other two atypical events. In a fourth scenario, we estimated the risk per serving associated with consumption of walnuts with Salmonella prevalence and levels from a 2014 to 2015 U.S. retail survey. Risk per serving estimates were two orders of magnitude larger than those of the baseline model without treatment. Further research is needed to determine whether this finding reflects variability in Salmonella contamination across the supply or a rare event affecting a portion of the supply.

Learning Something From Nothing: The Critical Importance of Rethinking Microbial Non-detects

Accurate estimation of microbial concentrations is necessary to inform many important environmental science and public health decisions and regulations. Critically, widespread misconceptions about laboratory-reported microbial non-detects have led to their erroneous description and handling as "censored" values. This ultimately compromises their interpretation and undermines efforts to describe and model microbial concentrations accurately. Herein, these misconceptions are dispelled by (1) discussing the critical differences between discrete microbial observations and continuous data acquired using analytical chemistry methodologies and (2) demonstrating the bias introduced by statistical approaches tailored for chemistry data and misapplied to discrete microbial data. Notably, these approaches especially preclude the accurate representation of low concentrations and those estimated using microbial methods with low or variable analytical recovery, which can be expected to result in non-detects. Techniques that account for the probabilistic relationship between observed data and underlying microbial concentrations have been widely demonstrated, and their necessity for handling non-detects (in a way which is consistent with the handling of positive observations) is underscored herein. Habitual reporting of raw microbial observations and sample sizes is proposed to facilitate accurate estimation and analysis of microbial concentrations.

A Quantitative Assessment of the Risk of Human Salmonellosis Arising from the Consumption of Pecans in the United States

A quantitative risk assessment was conducted to assess the risk of human salmonellosis acquired from consumption of pecans in the United States. The model considered the potential for Salmonella survival, growth, and recontamination of pecans from the sheller to the consumer, including steps such as immersion in water, drying, conditioning, cracking, partitioning, and storage. Five theoretical microbial reduction treatment levels (1 to 5 log CFU) were modeled. Data from the 2010 to 2013 surveys by the National Pecan Shellers Association were used for initial prevalence and contamination levels. The impacts of atypical situations in the pecan production system were also evaluated. Higher initial contamination levels, recontamination during processing, and a delay in drying postconditioning were the modeled atypical situations. The baseline model predicted a mean risk of salmonellosis in the United States from consumption of in-shell and shelled pecans processed by cold conditioning with no microbial reduction treatment and no further home cooking as 1 case per 775,193 servings (95% confidence interval [CI]: 1 case per 1,915,709 to 178,253 servings). This predicted risk per serving was estimated as a mean of 529 cases of salmonellosis per year (95% CI: 213 to 2,295 cases). Hot conditioning for shelled pecans and microbial reduction treatment of both shelled and in-shell pecans had a significant impact on the predicted mean risk of illness. Assuming 77% of the shelled pecans sold at retail (i.e., 80% of the retail supply) received hot conditioning, the mean estimated salmonellosis cases per year from consumption of in-shell and shelled pecans uncooked at home was 203 (95% CI: 81 to 882 cases) if no additional microbial reduction treatment were applied. The predicted risk of illness per serving was higher for all atypical situations modeled compared with the baseline model, and delay in drying had the greatest impact on risk.

A Quantitative Assessment of the Risk of Human Salmonellosis Arising from the Consumption of Almonds in the United States: The Impact of Preventive Treatment Levels

The presence of Salmonella on almonds continues to result in product-related outbreaks and recalls in the United States. In this study, the impact of microbial reduction treatment levels (1 to 5 log CFU) on the risk of human salmonellosis from the consumption of almond kernels in the United States was evaluated. An exposure model, including major steps in almond processing, was used to estimate prevalence and levels of contamination of Salmonella on almonds at the point of consumption. A Salmonella dose-response model and consumption data for almonds in the United States were used to assess risk of illness per serving and per year, quantifying variability and uncertainty separately. A 3-log reduction treatment resulted in a predicted mean risk of illness of two cases per year for almonds consumed as a core product not cooked at home (95% confidence interval [CI], one to four cases), one case per year for almonds consumed as an ingredient not cooked at home (95% CI, one to two cases), and less than one case per year for almonds consumed as an ingredient cooked at home (95% CI, 7 × 10-7 to 3 × 10-6 cases). A minimum 4-log reduction treatment resulted in an estimated mean risk of illness below one case per year in the United States. This study also includes an assessment of the risk of human salmonellosis as a result of an exceptional situation, which results in higher risk estimates compared with the baseline model. The exceptional situations modeled posttreatment resulted in estimates of mean risk that were not significantly affected by treatment level. Sensitivity analysis results showed initial Salmonella contamination level to be the factor with the most impact on risk per serving estimates, given a certain treatment level. The risk assessment also includes a simulation of the events that occurred in 2001. Treatment levels with a minimum 4-log microbial reduction would have been sufficient to prevent the outbreak cases. The uncertainty range in the estimates indicates that additional information is needed to make more precise predictions of this specific outbreak event.

Risk-based enteric pathogen reduction targets for non-potable and direct potable use of roof runoff, stormwater, and greywater

This paper presents risk-based enteric pathogen log reduction targets for non-potable and potable uses of a variety of alternative source waters (i.e., locally-collected greywater, roof runoff, and stormwater). A probabilistic Quantitative Microbial Risk Assessment (QMRA) was used to derive the pathogen log10 reduction targets (LRTs) that corresponded with an infection risk of either 10-4 per person per year (ppy) or 10-2 ppy. The QMRA accounted for variation in pathogen concentration and sporadic pathogen occurrence (when data were available) in source waters for reference pathogens in the genera Rotavirus, Mastadenovirus(human adenoviruses), Norovirus, Campylobacter, Salmonella, Giardia and Cryptosporidium. Non-potable uses included indoor use (for toilet flushing and clothes washing) with occasional accidental ingestion of treated non-potable water (or cross-connection with potable water), and unrestricted irrigation for outdoor use. Various exposure scenarios captured the uncertainty from key inputs, i.e., the pathogen concentration in source water; the volume of water ingested; and for the indoor use, the frequency of and the fraction of the population exposed to accidental ingestion. Both potable and non-potable uses required pathogen treatment for the selected waters and the LRT was generally greater for potable use than non-potable indoor use and unrestricted irrigation. The difference in treatment requirements among source waters was driven by the microbial quality of the water - both the density and occurrence of reference pathogens. Greywater from collection systems with 1000 people had the highest LRTs; however, those for greywater collected from a smaller population (~ 5 people), which have less frequent pathogen occurrences, were lower. Stormwater had highly variable microbial quality, which resulted in a range of possible treatment requirements. The microbial quality of roof runoff, and thus the resulting LRTs, remains uncertain due to lack of relevant pathogen data.

Quantification of Nonproteolytic Clostridium botulinum Spore Loads in Food Materials

We have produced data and developed analysis to build representations for the concentration of spores of nonproteolytic Clostridium botulinum in materials that are used during the manufacture of minimally processed chilled foods in the United Kingdom. Food materials are categorized into homogenous groups which include meat, fish, shellfish, cereals, fresh plant material, dairy liquid, dairy nonliquid, mushroom and fungi, and dried herbs and spices. Models are constructed in a Bayesian framework and represent a combination of information from a literature survey of spore loads from positive-control experiments that establish a detection limit and from dedicated microbiological tests for real food materials. The detection of nonproteolytic C. botulinum employed an optimized protocol that combines selective enrichment culture with multiplex PCR, and the majority of tests on food materials were negative. Posterior beliefs about spore loads center on a concentration range of 1 to 10 spores kg(-1). Posterior beliefs for larger spore loads were most significant for dried herbs and spices and were most sensitive to the detailed results from control experiments. Probability distributions for spore loads are represented in a convenient form that can be used for numerical analysis and risk assessments.

When Are Qualitative Testing Results Sufficient To Predict a Reduction in Illnesses in a Microbiological Food Safety Risk Assessment?

Process models that include the myriad pathways that pathogen-contaminated food may traverse before consumption and the dose-response function to relate exposure to likelihood of illness may represent a "gold standard" for quantitative microbial risk assessment. Nevertheless, simplifications that rely on measuring the change in contamination occurrence of a raw food at the end of production may provide reasonable approximations of the effects measured by a process model. In this study, we parameterized three process models representing different product-pathogen pairs (i.e., chicken-Salmonella, chicken-Campylobacter, and beef-E. coli O157:H7) to compare with predictions based on qualitative testing of the raw product before consideration of mixing, partitioning, growth, attenuation, or dose-response processes. The results reveal that reductions in prevalence generated from qualitative testing of raw finished product usually underestimate the reduction in likelihood of illness for a population of consumers. Qualitative microbial testing results depend on the test's limit of detection. The negative bias is greater for limits of detection that are closer to the center of the contamination distribution and becomes less as the limit of detection is moved further into the right tail of the distribution. Nevertheless, a positive bias can result when the limit of detection refers to very high contamination levels. Changes in these high levels translate to larger consumed doses for which the slope of the dose-response function is smaller compared with the larger slope associated with smaller doses. Consequently, in these cases, a proportional reduction in prevalence of contamination results in a less than proportional reduction in probability of illness. The magnitudes of the biases are generally less for nonscalar (versus scalar) adjustments to the distribution.

The impact of screening-test negative samples not enumerated by MPN

In microbiological surveys, false negative results in detection tests precluding the enumeration by MPN may occur. The objective of this study was to illustrate the impact of screening test failure on the probability distribution of Salmonella concentrations in pork using a Bayesian method. A total of 276 swab samples in four slaughter steps (69 samples in each slaughter step: after dehairing, after singeing, after evisceration, and before chilling) were screened for Salmonella and enumerated by the MPN method. Salmonella contamination data were fitted to a lognormal distribution by using a Bayesian model that uses the number of positive tubes at each dilution in an MPN analysis to estimate the parameters of the concentration distribution. With Salmonella paired data, three data sets were used for each slaughter step: one that includes the positives in the screening test only, a second one that includes false negative results from the screening, and a third that considers the entire data set. The relative sensitivity of the screening test was also calculated assuming as gold standard samples with confirmed Salmonella. Salmonella was confirmed by a reference laboratory in 29 samples either by screening or MPN method. The relative sensitivity of the screening test was 69% (CI 95%: 52%-85%). The data set that included enumerations from screen-negative samples (false negative results) tended to have higher μ̂ and smaller σ̂ in comparison with the data set that discards false negative results, suggesting that the lack of sensitivity of the screening test affects the distribution that describes the contamination across the population. Numerous surveys on fitting distribution methods of microbial censored data have been published and discuss source of bias due to fitting method. Results of this survey contribute with that discussion by illustrating another possible source of bias due to failure of the screening methods preceding the MPN.

Fitting a distribution to microbial counts: making sense of zeroes

The accurate estimation of true prevalence and concentration of microorganisms in foods is an important element of quantitative microbiological risk assessment (QMRA). This estimation is often based on microbial detection and enumeration data. Among such data are artificial zero counts, that originated by chance from contaminated food products. When these products are not differentiated from uncontaminated products that originate true zero counts, the estimates of true prevalence and concentration may be inaccurate. This inaccuracy is especially relevant in situations where highly pathogenic bacteria are involved and where growth can occur along the food pathway. Our aim was to develop a method that provides accurate estimates of concentration parameters and differentiates between artificial and true zeroes, thus also accurately estimating true prevalence. We first show the disadvantages of using a limit of quantification (LOQ) threshold for the analysis of microbial enumeration data. We show that, depending on the original distribution of concentrations and the LOQ value, it may be incorrect to treat artificial zeroes as censored below a quantification threshold. Next, a method is developed that estimates the true prevalence of contamination within a food lot and the parameters characterizing the within-lot distribution of concentrations, without assuming a LOQ, and using raw plate count data as an input. Counts resulting both from contaminated and uncontaminated sample units are analysed together. This procedure allows the estimation of the proportion of artificial zeroes among the total of zero counts, and therefore the estimation of true prevalence from enumeration results. We observe that this method yields best estimates of mean, standard deviation and prevalence at low true prevalence levels and low expected standard deviation. Furthermore, we conclude that the estimation of prevalence and the estimation of the distribution of concentrations are interrelated and therefore should be estimated simultaneously. We also conclude that one of the keys to an accurate characterization of the overall microbial contamination is the correct identification and separation of true and artificial zeroes. Our method for the analysis of quantitative microbial data shows a good performance in the estimation of true prevalence and the parameters of the distribution of concentrations, which indicates that it is a useful data analysis tool in the field of QMRA.

Impact of microbial count distributions on human health risk estimates

Quantitative microbiological risk assessment (QMRA) is influenced by the choice of the probability distribution used to describe pathogen concentrations, as this may eventually have a large effect on the distribution of doses at exposure. When fitting a probability distribution to microbial enumeration data, several factors may have an impact on the accuracy of that fit. Analysis of the best statistical fits of different distributions alone does not provide a clear indication of the impact in terms of risk estimates. Thus, in this study we focus on the impact of fitting microbial distributions on risk estimates, at two different concentration scenarios and at a range of prevalence levels. By using five different parametric distributions, we investigate whether different characteristics of a good fit are crucial for an accurate risk estimate. Among the factors studied are the importance of accounting for the Poisson randomness in counts, the difference between treating "true" zeroes as such or as censored below a limit of quantification (LOQ) and the importance of making the correct assumption about the underlying distribution of concentrations. By running a simulation experiment with zero-inflated Poisson-lognormal distributed data and an existing QMRA model from retail to consumer level, it was possible to assess the difference between expected risk and the risk estimated with using a lognormal, a zero-inflated lognormal, a Poisson-gamma, a zero-inflated Poisson-gamma and a zero-inflated Poisson-lognormal distribution. We show that the impact of the choice of different probability distributions to describe concentrations at retail on risk estimates is dependent both on concentration and prevalence levels. We also show that the use of an LOQ should be done consciously, especially when zero-inflation is not used. In general, zero-inflation does not necessarily improve the absolute risk estimation, but performance of zero-inflated distributions in QMRA tends to be more robust to changes in prevalence and concentration levels, and to the use of an LOQ to interpret zero values, compared to that of their non-zero-inflated counterparts.

FDA-iRISK--a comparative risk assessment system for evaluating and ranking food-hazard pairs: case studies on microbial hazards

Stakeholders in the system of food safety, in particular federal agencies, need evidence-based, transparent, and rigorous approaches to estimate and compare the risk of foodborne illness from microbial and chemical hazards and the public health impact of interventions. FDA-iRISK (referred to here as iRISK), a Web-based quantitative risk assessment system, was developed to meet this need. The modeling tool enables users to assess, compare, and rank the risks posed by multiple food-hazard pairs at all stages of the food supply system, from primary production, through manufacturing and processing, to retail distribution and, ultimately, to the consumer. Using standard data entry templates, built-in mathematical functions, and Monte Carlo simulation techniques, iRISK integrates data and assumptions from seven components: the food, the hazard, the population of consumers, process models describing the introduction and fate of the hazard up to the point of consumption, consumption patterns, dose-response curves, and health effects. Beyond risk ranking, iRISK enables users to estimate and compare the impact of interventions and control measures on public health risk. iRISK provides estimates of the impact of proposed interventions in various ways, including changes in the mean risk of illness and burden of disease metrics, such as losses in disability-adjusted life years. Case studies for Listeria monocytogenes and Salmonella were developed to demonstrate the application of iRISK for the estimation of risks and the impact of interventions for microbial hazards. iRISK was made available to the public at http://irisk.foodrisk.org in October 2012.