Trends in Salmonella Infantis human illness incidence and chicken carcass prevalence in the United States; 1996-2019.

The incidence of human illness due to Salmonella Infantis reported to Foodborne Diseases Active Surveillance Network and the prevalence of Infantis on chicken carcasses reported by the United States Department of Agriculture Food Safety and Inspection Service have increased significantly in the past decade. However, the trends do not appear coincident, as would be expected if the increased prevalence in chicken led to the increase in the incidence of human illness. Salmonella Infantis incidence and prevalence trends are analyzed using penalized B-spline methods for generalized additive regression models. The association between the two time series is analyzed using time-lagged rank-order cross-correlation. Geographic variations in reported incidence and trends are also explored. The increase in human incidence of Salmonella Infantis began circa 2011. The increase in chicken carcass prevalence began circa 2015. A 4-year lag on chicken carcass prevalence maximizes the rank-order cross-correlation with the incidence of illness. While chicken consumption undoubtedly contributes to the incidence of human illness due to Salmonella Infantis, the initial increase in reported illness was likely due to one or more other transmission pathways. Other potential transmission pathways include non-chicken foodborne, waterborne, person-to-person, animal contact, and environmental.

Trends in reported illness due to poultry- and nonpoultry associated Salmonella serotypes; United States 1996-2019.

Retrospective review is a key to designing effective food safety measures. Despite the reported reduction of Salmonella prevalence in poultry products, there has not been a concomitant reduction of the overall incidence of Salmonella illnesses reported to the US Foodborne Diseases Active Surveillance Network (FoodNet) since 1996. However, there have been significant annual trends among Salmonella serotypes. This analysis examines trends in the reported incidence of illness due to poultry- and nonpoultry associated Salmonella serotypes. Overall, the findings indicate declining trends in illness due to the poultry-associated serotypes and increasing trends in illness due to Salmonella serotypes not associated with poultry.

Trends in Reported Foodborne Illness in the United States; 1996-2013.

Retrospective review is a key to designing effective food safety measures. The analysis examines trends in the reported incidence of illness due to bacterial pathogens commonly transmitted by food in the United States during 1996-2013 with and without specifying a model form for trend. The findings indicate early declines in reported incidence followed by a period of no significant trend for Campylobacter, Listeria, Shiga toxin-producing Escherichia coli O157, and Yersinia. The results are inconclusive about whether there is no trend or an increasing trend for Salmonella. While Shigella exhibits a continuous decline, Vibrio exhibits a continuous increase. Overall, the findings indicate a lack of evidence for continuous reduction in illness due to bacterial pathogens commonly transmitted by food in the United States during 1996-2013.

Risk-Based Sampling: I Don't Want to Weight in Vain.

Recently, there has been considerable interest in developing risk-based sampling for food safety and animal and plant health for efficient allocation of inspection and surveillance resources. The problem of risk-based sampling allocation presents a challenge similar to financial portfolio analysis. Markowitz (1952) laid the foundation for modern portfolio theory based on mean-variance optimization. However, a persistent challenge in implementing portfolio optimization is the problem of estimation error, leading to false "optimal" portfolios and unstable asset weights. In some cases, portfolio diversification based on simple heuristics (e.g., equal allocation) has better out-of-sample performance than complex portfolio optimization methods due to estimation uncertainty. Even for portfolios with a modest number of assets, the estimation window required for true optimization may imply an implausibly long stationary period. The implications for risk-based sampling are illustrated by a simple simulation model of lot inspection for a small, heterogeneous group of producers.

Optimal Food Safety Sampling Under a Budget Constraint.

Much of the literature regarding food safety sampling plans implicitly assumes that all lots entering commerce are tested. In practice, however, only a fraction of lots may be tested due to a budget constraint. In such a case, there is a tradeoff between the number of lots tested and the number of samples per lot. To illustrate this tradeoff, a simple model is presented in which the optimal number of samples per lot depends on the prevalence of sample units that do not conform to microbiological specifications and the relative costs of sampling a lot and of drawing and testing a sample unit from a lot. The assumed objective is to maximize the number of nonconforming lots that are rejected subject to a food safety sampling budget constraint. If the ratio of the cost per lot to the cost per sample unit is substantial, the optimal number of samples per lot increases as prevalence decreases. However, if the ratio of the cost per lot to the cost per sample unit is sufficiently small, the optimal number of samples per lot reduces to one (i.e., simple random sampling), regardless of prevalence. In practice, the cost per sample unit may be large relative to the cost per lot due to the expense of laboratory testing and other factors. Designing effective compliance assurance measures depends on economic, legal, and other factors in addition to microbiology and statistics.

The economic efficiency of sampling size: the case of beef trim revisited.

A recent paper by Ferrier and Buzby provides a framework for selecting the sample size when testing a lot of beef trim for Escherichia coli O157:H7 that equates the averted costs of recalls and health damages from contaminated meats sold to consumers with the increased costs of testing while allowing for uncertainty about the underlying prevalence of contamination. Ferrier and Buzby conclude that the optimal sample size is larger than the current sample size. However, Ferrier and Buzby's optimization model has a number of errors, and their simulations failed to consider available evidence about the likelihood of the scenarios explored under the model. After correctly modeling microbial prevalence as dependent on portion size and selecting model inputs based on available evidence, the model suggests that the optimal sample size is zero under most plausible scenarios. It does not follow, however, that sampling beef trim for E. coli O157:H7, or food safety sampling more generally, should be abandoned. Sampling is not generally cost effective as a direct consumer safety control measure due to the extremely large sample sizes required to provide a high degree of confidence of detecting very low acceptable defect levels. Food safety verification sampling creates economic incentives for food producing firms to develop, implement, and maintain effective control measures that limit the probability and degree of noncompliance with regulatory limits or private contract specifications.

How to model a negligible probability under the WTO sanitary and phytosanitary agreement?

Since the 1997 EC--Hormones decision, World Trade Organization (WTO) Dispute Settlement Panels have wrestled with the question of what constitutes a negligible risk under the Sanitary and Phytosanitary Agreement. More recently, the 2010 WTO Australia--Apples Panel focused considerable attention on the appropriate quantitative model for a negligible probability in a risk assessment. The 2006 Australian Import Risk Analysis for Apples from New Zealand translated narrative probability statements into quantitative ranges. The uncertainty about a "negligible" probability was characterized as a uniform distribution with a minimum value of zero and a maximum value of 10(-6) . The Australia - Apples Panel found that the use of this distribution would tend to overestimate the likelihood of "negligible" events and indicated that a triangular distribution with a most probable value of zero and a maximum value of 10⁻6 would correct the bias. The Panel observed that the midpoint of the uniform distribution is 5 × 10⁻7 but did not consider that the triangular distribution has an expected value of 3.3 × 10⁻7. Therefore, if this triangular distribution is the appropriate correction, the magnitude of the bias found by the Panel appears modest. The Panel's detailed critique of the Australian risk assessment, and the conclusions of the WTO Appellate Body about the materiality of flaws found by the Panel, may have important implications for the standard of review for risk assessments under the WTO SPS Agreement.

Analyzing the power and error of Listeria monocytogenes growth challenge studies.

Domestic and international food safety policy developments have spurred interest in the design and interpretation of experimental growth challenge studies to determine whether ready-to-eat (RTE) foods are able to support growth of Listeria monocytogenes. Existing challenge study protocols and those under development differ markedly in terms of experimental design and the acceptance criteria under which a RTE food is determined not to support L. monocytogenes growth. Consequently, the protocols differ substantially with respect to the probability of incorrectly determining that growth occurs and the statistical power to detect growth if it does occur. Applying a fixed acceptance criteria exceedance value (e.g., less than a 0.5 log(10) or 1 log(10) increase) to distinguish real growth from quantitative measurement uncertainty over different experimental designs and/or measurement uncertainty values implies highly inconsistent type I error (alpha) probabilities. None of the L. monocytogenes growth challenge study designs currently being considered are likely to provide an F-test with alpha=0.05 and power >or=0.8 to detect a 1 log(10) increase in mean concentration over the entire range of measurement uncertainty values for enumeration of L. monocytogenes reported in food samples in a validation study of ISO Method 11290-2.

Modeling the frequency and duration of microbial contamination events.

The frequency and duration of microbial contamination events in the environment in which ready-to-eat (RTE) foods are exposed for processing and packaging is subject to uncertainty and variability. Variability, within-model parameter uncertainty, and uncertainty regarding model selection are formally considered in modeling the frequency and duration of such contamination events by Listeria species. The estimated duration of contamination events represents a case where variability dominates with relatively little uncertainty about parameter values or model form. The estimated frequency of contamination events represents a case where there is not only substantial variability but also considerable within-model parameter uncertainty, as well as some uncertainty regarding model selection. The Bayesian Information Criterion provides a formal way of taking into account model uncertainty.

Bayesian synthesis of a pathogen growth model: Listeria monocytogenes under competition.

The Bayesian synthesis method is applied to data from two studies of Listeria monocytogenes grown in broth monocultures to draw inferences about the joint distribution of two Baranyi growth model parameters-lag time and maximum specific growth rate. The resultant joint distribution is then combined with prior distributions for the initial and maximum pathogen density parameters under competitive growth conditions. Finally, the pathogen growth model is updated using the Sampling/Importance Resampling (SIR) algorithm with data on L. monocytogenes growth in competition with natural microflora in fish. Although the latter data provide no information on the stationary phase to directly estimate the maximum pathogen density parameter, combining them with relevant prior information provides a means to characterize L. monocytogenes growth in a food with mixed microbial populations. Based on a specified tolerance for L. monocytogenes growth, the updated model provides a storage time limit for fish held at 5 degrees C, pH 6.8, 43% CO(2), 57% N(2).

Quantifying the robustness of a broth-based Escherichia coli O157:H7 growth model in ground beef.

The robustness of a microbial growth model must be assessed before the model can be applied to new food matrices; therefore, a methodology for quantifying robustness was developed. A robustness index (RI) was computed as the ratio of the standard error of prediction to the standard error of calibration for a given model, where the standard error of calibration was defined as the root mean square error of the growth model against the data (log CFU per gram versus time) used to parameterize the model and the standard error of prediction was defined as the root mean square error of the model against an independent data set. This technique was used to evaluate the robustness of a broth-based model for aerobic growth of Escherichia coli 0157:H7 (in the U.S Department of Agriculture Agricultural Research Service Pathogen Modeling Program) in predicting growth in ground beef under different conditions. Comparison against previously published data (132 data sets with 1,178 total data points) from experiments in ground beef at various experimental conditions (4.8 to 45 degrees C and pH 5.5 to 5.9) yielded RI values ranging from 0.11 to 2.99. The estimated overall RI was 1.13. At temperatures between 15 and 40 degrees C, the RI was close to and smaller than 1, indicating that the growth model is relatively robust in that temperature range. However, the RI also was related (P < 0.05) to temperature. By quantifying the predictive accuracy relative to the expected accuracy, the RI could be a useful tool for comparing various models under different conditions.

Modeling the response of the Mediterranean fruit fly (Diptera:Tephritidae) to cold treatment.

Recent interceptions of live Mediterranean fruit fly, Ceratitis capitata (Wiedemann), larvae in fruit that had been cold-treated during transit from abroad led to a reevaluation of the scientific basis for the relevant regulatory treatment schedules. A time-temperature response surface model based on the original experimental data from 1916 was developed and evaluated based on subsequent experimental trials and recent surveillance data collected from shipping operations. The resultant model is reasonably robust and supports the conclusion that the previous treatment schedule falls short of the intended probit nine level of security. Given the vintage of the data, methodological inconsistencies among studies, and the potential consequences of new introductions, additional research is warranted. Quantitative analysis of the currently available data suggests that future studies regarding the efficacy of cold storage should focus on low-temperature, short-duration treatments, where uncertainty about performance appears greatest. The analysis of subsequent experiments also demonstrates that for cold treatment trials most often resulting in zero survivors, Bayesian statistical methods applied to a series of replicated trials of more manageable size offers a feasible alternative to conducting impracticably large trials.

A Model for Probabilistic Assessment of Phytosanitary Risk Reduction Measures.

Ideally, a phytosanitary performance standard would be defined as a probabilistic tolerance. For treatments such as solid wood pasteurization, this could be operationalized by stating with a specific degree of confidence that the treatment failure rate for a sentinel pest should be less than a defined level (e.g., X% confidence that the wood heat treatment failure rate for pest Y does not exceed Z%). This article illustrates a probabilistic approach to developing a phytosanitary performance standard, using heat treatment of the wood-inhabiting fungus Postia placenta as an example. The uncertainty about the proportion of wood blocks in which P. placenta survives after treatment is characterized by the Beta distribution, subject to the biological constraint that survival should decrease monotonically with increased time and temperature. Monte Carlo simulation techniques are then used to generate a probabilistic response surface relating proportion survival to treatment time and temperature. This modeling approach relaxes the parametric assumptions associated with traditional statistical methods for fitting response surfaces and is more flexible than conventional methods, resulting in a better fit to the observed data.