Course Corrections Are Needed to Better Address the Risk of Salmonellosis from Poultry Food

Thomas P. Oscar

doi:10.5772/intechopen.1011495

Abstract

Computer models that predict consumer exposure and response to Salmonella in poultry food are valuable tools for helping assess and manage this risk to public health. However, model predictions are only as good as the data and models used to make them. If the data and models are inaccurate, the model simulations and predictions will be erroneous, and the risk management decision will be wrong. In this study, a comparison is made between two different data collection and modeling methods for assessing the risk of salmonellosis from poultry food. First, are the probabilistic methods used in Quantitative Microbial Risk Assessment (QMRA). Second, are the rare event methods used in the Poultry Food Assess Risk Model (PFARM). Based on the comparison of these data collection and modeling methods and a case study, it was concluded that course corrections are needed to better assess and manage the risk of salmonellosis from poultry food. Namely, a shift away from the probabilistic methods of QMRA to the rare event methods of PFARM and a shift in focus from the pre-harvest and post-harvest sectors of the farm-to-table chain to the consumer sector where the most important risk factors for salmonellosis from poultry food are found.

Keywords

Salmonella
poultry food
risk assessment
public health
comparison
QMRA
PFARM

Author Information

Show +

Thomas P. Oscar *
- United States Department of Agriculture, Agricultural Research Service, Microbial and Chemical Food Safety Research Unit, University of Maryland Eastern Shore Worksite, Princess Anne, Maryland, USA

*Address all correspondence to: thomas.oscar@usda.gov

1. Introduction

A pathogen reduction, hazard analysis and critical control point, and performance standard (PR-HACCP-PS) program was implemented in the United States to reduce the risk of salmonellosis from poultry food [1]. This program has been successful at reducing Salmonella prevalence in carcass rinse aliquot samples at final product testing in the processing plant [2]. However, it has not been successful at reducing the risk or rate of foodborne salmonellosis. This has prompted a proposed expansion of the PR-HACCP-PS program to include new PS for Salmonella number and serotype [3].

When data collection and modeling methods for predicting the risk of salmonellosis from poultry food emerged in the 1990s there was a divergence of methods leading to an accepted approach (Quantitative Microbial Risk Assessment or QMRA) and an alternative approach (Poultry Food Assess Risk Model or PFARM). Here, a comparison of QMRA and PFARM data collection and modeling methods is made with a brief description of key studies. To reduce self-citation of PFARM studies, five recent studies [4, 5, 6, 7, 8] that review why and how the PFARM data collection and modeling methods were developed and used will be emphasized.

2. Comparison of QMRA and PFARM

The quality of a computer model simulation and prediction is only as good as the data and models on which it is based. Thus, if the data collection and modeling methods are not accurate, the model simulations and predictions will not be accurate, and the resulting course of action will not be the right one. After the comparison of the data collection and modeling methods used in QMRA and PFARM, a case study will be presented to illustrate the importance of the differences in these two approaches to assessing the risk of salmonellosis from poultry food.

2.1 Sampling method

Salmonella may be unattached, attached, or entrapped in the poultry food matrix [9, 10]. Therefore, the sampling method used must be able to detect a single cell of viable Salmonella regardless of how it is associated with the poultry food matrix, or the prevalence, number, and serotype data will not be accurate.

QMRA - Rinsing (chicken carcass and parts) and swabbing (turkey carcass and parts) of poultry food recovers some but not all the Salmonella [11]. This leads to an underestimation of prevalence, number, and serotype, and an inaccurate simulation and prediction of the risk of salmonellosis. PFARM - Whole sample enrichment (WSE) detects more if not all Salmonella on and in poultry food leading to more accurate prevalence, number, and serotype data for simulating and predicting the risk of salmonellosis [8]. Key QMRA study - Simmons et al. [12] reported that 13% of chicken carcasses were positive for Salmonella when rinse samples were used, whereas 38% were positive for Salmonella when WSE was used. Key PFARM study - Parveen et al. [13] reported that Salmonella prevalence was 88% in a compliant commercial processing plant (i.e. < 9.8%) when chicken carcasses were sampled using WSE instead of the carcass rinse aliquot method.

2.2 Salmonella prevalence

QMRA- Salmonella prevalence of poultry food is not expressed and simulated as a function of the size of sample analyzed resulting in an inaccurate simulation and prediction of the risk of salmonellosis. PFARM - Salmonella prevalence is expressed and simulated as a function of the size of sample analyzed resulting in a more accurate simulation and prediction of the risk of salmonellosis from poultry food. Key QMRA study - Surkiewicz et al. [14] reported that Salmonella in rinse samples of chicken carcasses was 4.9% when 10 mL was examined and 20.5% when 270 mL was examined. Thus, it has been known for some time, since 1969, that Salmonella prevalence changes in a nonlinear manner as a function of the size of sample analyzed, yet, in QMRA it is expressed and simulated without a denominator. Key PFARM study - Oscar [8] reported that Salmonella prevalence increases in a nonlinear manner as a function of the size of poultry food sample analyzed.

2.3 Microbiological independence

QMRA - The split sample method uses a different sample and a different size sample of poultry food for Salmonella prevalence (cultural isolation) and number (most probable number or MPN) leading to inaccurate data for simulating and predicting the risk of salmonellosis because the microbiological independence of the samples is not recognized. PFARM - A single sample of poultry food is used for Salmonella prevalence (cultural isolation) and number (growth units), which is determined by WSE, real-time quantitative polymerase chain reaction (qPCR). This leads to more accurate data for simulating and predicting the risk of salmonellosis. Key QMRA study - Bemrah et al. [15] used the split sample method for prevalence and number of Salmonella in turkey cordon bleu. However, only 2 of 36 samples that tested positive for Salmonella by cultural isolation, tested positive in the MPN assay because a different sample and a much smaller sample was used to determine Salmonella number. Key PFARM study - Oscar [8] used the one pathogen cell test to show why the QMRA split sample method is inaccurate and the PFARM single sample method (WSE-qPCR) is more accurate.

2.4 Location of final product testing

QMRA – Final product testing for Salmonella is in the poultry food processing plant leading to an inaccurate simulation and prediction of the risk of salmonellosis because the growth and spread of Salmonella in packaged poultry food after processing leads to unpredictable changes in Salmonella serotype prevalence and number among servings in the package [16]. PFARM - Final product testing for Salmonella is at the start of meal preparation to better capture effects of poultry food production, processing, distribution, storage, and interventions on Salmonella serotype prevalence and number leading to a more accurate simulation and prediction of the risk of salmonellosis [8].

2.5 Salmonella number

QMRA - Results of the MPN assay used in the split sample method are expressed and simulated on a per gram basis resulting in an inaccurate simulation and prediction of the risk of salmonellosis from poultry food because Salmonella number does not increase in a linear manner as a function of serving size, which is assumed in this method of expression. PFARM – Results of the WSE-qPCR method used in the single sample method are expressed and simulated as a function of the size of sample analyzed resulting in more accurate simulation and prediction of the risk of salmonellosis from poultry food because Salmonella number is predicted to increase in a non-linear manner as a function of serving size. Key QMRA study - De Man [17] reported that for a 3 replicate by 3 dilution MPN assay, the MPN per g ranges from 0.3 for a result of 0-0-1 to 110 for a result of 3-3-2. However, when results are expressed as a function of the size of sample analyzed, the range is 1 to 366 MPN per 3.33 g, which is more accurate. The MPN result of 0-0-0 is not ≤0.3/g. Rather, it is 0 per 3.33 g or negative for Salmonella. Key PFARM study - Oscar [8] reported that Salmonella number increases in a nonlinear manner as a function of the size of sample analyzed or simulated.

2.6 Growth potential

QMRA - The MPN method does not capture the growth potential of Salmonella, which is important for accurate simulation and prediction of the risk of salmonellosis from poultry food because it captures the physiological state of the pathogen and the impact of the native microflora on its growth in the poultry food matrix. Importantly, it does not capture effects of processes that injure Salmonella and reduce the risk of salmonellosis from poultry food. PFARM – The WSE-qPCR method is a growth-based assay that captures the growth potential of Salmonella in poultry food resulting in a more accurate simulation and prediction of the risk of salmonellosis. Key PFARM study - Oscar [8] used WSE-qPCR to enumerate Salmonella growth units in poultry food as a non-linear function of serving size.

2.7 Exposure assessment

QMRA - The probabilistic method overestimates consumer exposure to Salmonella because it only simulates contaminated grams of poultry food and assumes that Salmonella number increases in a linear manner as a function of serving size. PFARM - The rare event modeling method does not overestimate consumer exposure to Salmonella because it simulates both contaminated and noncontaminated servings and predicts the non-linear increase of Salmonella prevalence and number as a function of serving size. Key PFARM study - Oscar [5] provided 13 reasons why the QMRA probabilistic method provides an inaccurate prediction of consumer exposure to Salmonella from poultry food.

2.8 Dose-response assessment

QMRA - The probabilistic, sigmoidal, dose-response model [18] and its variants [3] do not explicitly simulate the Salmonella, poultry food meal, and consumer interaction or disease triangle resulting in an inaccurate simulation and prediction of the risk of salmonellosis. PFARM - The rare event, disease triangle, dose-response model [6] simulates risk of salmonellosis as a function of the zoonotic potential of Salmonella serotypes, the buffering capacity of the poultry food meal, and the health and immunity of the consumer. Key PFARM study - Oscar [19] reported that the dose-response curve is non-sigmoid in shape when the food is contaminated with two or more serotypes of Salmonella with different virulence and prevalence.

2.9 Risk prediction

QMRA - Predicts that a reduction of Salmonella in poultry food at final product testing will result in like reduction in the risk of salmonellosis [3], which is not consistent with the current situation in the poultry food industry. PFARM - Predicts that a reduction of Salmonella in poultry food at final product testing will not result in like reduction in the risk of salmonellosis, which is consistent with the current situation in the poultry food industry. Key PFARM study - Oscar [20] showed that there was no relationship between Salmonella levels at final product testing and risk of salmonellosis because of post-testing risk factors.

2.9.1 Risk prediction adjustment

QMRA - A correction factor is needed to align predictions of the risk of salmonellosis from poultry food with epidemiological data [3]. PFARM - No correction factor is needed to align predictions of the risk of salmonellosis from poultry food with epidemiological data [20]. Key PFARM study - Oscar [5] provided 13 reasons why the probabilistic modeling method used in QMRA makes inaccurate predictions of the risk of salmonellosis from poultry food.

2.10 Lot size

QMRA - Individual grams of Salmonella-contaminated poultry food is simulated from final product testing until consumption. At consumption they are multiplied by serving size and initial Salmonella prevalence resulting in an inaccurate simulation and prediction of the risk of salmonellosis. This is because lot size increases at consumption as a function of mean serving size and initial Salmonella prevalence. Thus, the lot size could differ among scenarios resulting in confounded comparisons. PFARM- Salmonella-contaminated and non-contaminated servings and Salmonella serotype prevalence and number (growth units) per size of sample analyzed are simulated from final product testing to consumption for more accurate prediction of the risk of salmonellosis because lot size at final product testing and at consumption is the same resulting in non-confounded comparisons among scenarios [5].

2.11 Predictive models

Predictive models for cross-contamination, growth, and thermal death of Salmonella in poultry food are used in QMRA and PFARM for exposure assessment. However, important differences exist for how predictive models are used to predict the risk of salmonellosis from poultry food.

2.11.1 Cross-contamination

QMRA - Salmonella transfer rates from poultry food to kitchen fomites and then to ready-to-eat (RTE) food are used to simulate cross-contamination of RTE food with Salmonella, which is simulated as probabilistic event [21]. This method does not consider Salmonella as a minority member of the native microflora of food. PFARM - Bacterial transfer rates from poultry food to kitchen fomites and then to RTE food are used to simulate cross-contamination of RTE food with native microflora including Salmonella, whose transfer is simulated as a discrete event using the rare event, line of transfer method [5]. This method simulates Salmonella as a minority member of the native microflora leading to more accurate simulation and prediction of the risk of salmonellosis because in nature, a cell of Salmonella does not have a probability of transfer, rather, it is transferred or not.

2.11.2 Growth

QMRA- Models for growth of Salmonella are developed using a mixture of serotypes inoculated at a high and nonecological level in poultry food without native microflora resulting in a less accurate simulation and prediction of the growth of Salmonella and risk of salmonellosis [22]. PFARM - Models for growth of Salmonella are developed using individual serotypes inoculated at a low and ecological level in poultry food with native microflora leading to a more accurate simulation and prediction of the growth of Salmonella and the risk of salmonellosis [4]. Key QMRA study - Kim et al. [3] stated that serotype-specific growth models are needed for accurate risk assessment. Key PFARM study - Oscar [23] developed a model for three serotypes of Salmonella on poultry food for use in risk assessment.

2.11.3 Thermal death

QMRA – Models for thermal death of Salmonella are developed with a mixture of serotypes and a high and nonecological number of the pathogen in poultry food without native microflora [24]. In addition, the models predict the probability of Salmonella survival, which is >0 for all grams simulated [25]. This leads to an overestimation of consumer exposure to Salmonella and an inaccurate simulation and prediction of the risk of salmonellosis from cooked poultry food [5]. PFARM – Models for thermal death of Salmonella are developed with individual serotypes of Salmonella in poultry food with native microflora [26]. In addition, a rare event, line of death method is used to simulate survival of individual cells of Salmonella as minority members of the native microflora of poultry food during cooking [5]. Thus, survival of a Salmonella cell is simulated as a discrete (survival or no survival) event resulting in most (> 97%) cooked servings having zero Salmonella at consumption and the rest having ≥1 Salmonella. This leads to a more accurate simulation of what occurs in nature and a more accurate prediction of the risk of salmonellosis from cooked poultry food [5].

2.11.4 Model validation

QMRA- Models for behavior of Salmonella in poultry food are validated using conventional statistical methods and the bias and accuracy factor method of Ross [27]. These methods have no criteria for test data, limited criteria for model performance, and no criteria for model validation. This leads to an incomplete, less accurate, and more biased validation of models. In turn, this could lead to less accurate and more biased simulation and prediction of the risk of salmonellosis from poultry food. PFARM - Models for behavior of Salmonella in poultry food are validated using the Acceptable Prediction Zones (APZ) method [4]. The APZ method has objective criteria for test data, model performance, and model validation resulting in more accurate and less biased models and more accurate simulation and prediction of the risk of salmonellosis from poultry food. Key PFARM study - Oscar [7] provided 12 reasons why the APZ method is better than conventional methods and the bias and accuracy factor method of Ross for validating predictive models.

2.12 Case study

The end game of a program like PR-HACCP-PS is the identification and removal of unsafe lots of poultry food from the marketplace. However, the use of less accurate data and models increases identification of safe lots of poultry food as unsafe and unsafe lots of poultry food as safe. Both situations harm public health. The first by shortages, inflation, malnutrition, lower immunity, and more disease. The second by increased hospitalization and death and loss of brand confidence resulting in economic damage. Thus, it is important to get it right.

Data for Salmonella prevalence, number (growth units), and serotypes of chicken gizzards [8] were collected over time in a production chain using PFARM methods. The data was used to compare QMRA and PFARM modeling methods for predicting consumer exposure to Salmonella [5]. It was found that QMRA predicted higher consumer exposure to Salmonella than PFARM using the same data and exposure pathways [5].

In this case study, the comparison was extended to predicting risk of salmonellosis (illness) from chicken gizzards. However, the data was repurposed. Instead of five scenarios for running windows of 60 samples over time in the production chain, the five time-period scenarios represented five, 1000 kg lots of chicken gizzards.

2.12.1 Risk of salmonellosis among lots of chicken gizzards

QMRA: The risk of salmonellosis (illness) differed (p ≤ 0.05) among lots of chicken gizzards (Figure 1A). It was highest for lots A and B, lowest for lot D, and intermediate and the same for lots C and E. PFARM: The risk of salmonellosis differed (p ≤ 0.05) among lots of chicken gizzards (Figure 1B). It was higher for lot E than for lot B; otherwise, it was the same for all lots. Compared to QMRA (Figure 1A), the risk of salmonellosis predicted by PFARM (Figure 1B) was lower and differed less among lots. The results for QMRA were like those for exposure [5].

Figure 1.
Risk of salmonellosis (illness) from chicken gizzard meals as predicted by: (A) the probabilistic method of QMRA; and (B) the rare event method of PFARM for the same data and risk pathways. Bars are means ± standard deviation of 20 replicate simulations. The lot size simulated was 1000 kg of chicken gizzards per replicate simulation. The average portion size was 168 g per meal. Thus, 5952 meals were simulated. Within a panel, bars with different letters differ at p ≤ 0.05 per one-way analysis of variance followed by Tukey’s multiple comparison test (Prism, version 10, GraphPad Software, Boston, MA). Abbreviations: QMRA = quantitative microbial risk assessment; and PFARM = poultry food assess risk model.

2.12.2 Risk of salmonellosis as a function of dose

QMRA- The probability of illness increased in an exponential manner as a function of the dose of Salmonella consumed from chicken gizzard meals (Figure 2A). This was because differences in zoonotic potential of Salmonella, buffering capacity of the meal, and consumer health and immunity were not explicitly simulated in the dose-response model. For the simulated replicate of lot E (Figure 2A), the probabilistic method used in the QMRA predicted 9–10 cases of salmonellosis per 1000 kg of chicken gizzards. In this case, all meals (n = 5952) had a Salmonella dose consumed that was >0. The meals with Salmonella doses >0 and < 1 were 5833, ≥ 1 and < 10 were 51, and ≥ 10 and < 10,000 were 68. However, the probabilistic modeling method used in QMRA does not provide an accurate simulation and prediction of what occurs in nature for the reasons stated above. In fact, it incorrectly mimicked consumer exposure to Salmonella [5] and resulted in a biased and overestimated prediction of the risk of salmonellosis. Moreover, it incorrectly simulated and predicted that a reduction in Salmonella dose consumed would result in a like reduction in the risk of salmonellosis, which is not what occurs in nature. Again, this occurred because the QMRA method did not explicitly simulate differences in zoonotic potential of Salmonella, buffering capacity of the meal, and consumer health and immunity.

Figure 2.
The relationship between *Salmonella* dose consumed from meals of chicken gizzards and (A) probability of illness or salmonellosis as predicted by QMRA and (B) consumer response as predicted by PFARM for a single replication from lot E. Abbreviations: QMRA = quantitative microbial risk assessment; and PFARM = poultry food assess risk model.

PFARM - In contrast to QMRA, PFARM predicted that consumer response did not increase as a function of the Salmonella dose consumed from 1 to 100 growth units (Figure 2B). In fact, the consumer who ate the meal with 30 growth units had no response, whereas some consumers who ate meals with lower doses became infected (fecal shedding or antibody response with no symptoms of salmonellosis) with Salmonella. In contrast to QMRA, which predicted 9–10 cases of salmonellosis from lot E, PFARM only predicted one case (Figure 2B). In addition, PFARM predicted that most meals (5675 of 5952) resulted in no exposure to Salmonella and of those that resulted in exposure, most (275 of 277) had Salmonella doses from 1 to 10 growth units, whereas only two had a Salmonella dose >10 growth units. Thus, compared to QMRA (Figure 2A), PFARM (Figure 2B) predicted much less exposure to Salmonella and much less risk of salmonellosis. One important difference was the PFARM simulated the zoonotic potential of Salmonella serotypes, the buffering capacity of the meal, and the health and immunity of the consumer in its dose-response model, whereas QMRA did not [6].

2.13 Lessons learned from COVID-19 pandemic

A lesson learned from the coronavirus disease 2019 (COVID-19) pandemic was that zoonotic diseases can become endemic at a baseline level that fluctuates over time in response to changes in the pathogen, environment, and host interaction or disease triangle [28]. This same dynamic may be at play in salmonellosis from poultry food leading to the possibility that salmonellosis from poultry food may be at an endemic level that is fluctuating over time in response to changes in the disease triangle. Thus, it might be difficult to reduce foodborne salmonellosis further because a reduction in consumer exposure to Salmonella could be countered by a decrease in consumer immunity to salmonellosis. In other words, interventions like a PR-HACCP-PS program that reduce Salmonella at final product testing may have the unintended consequence of decreasing consumer immunity to salmonellosis [29] and help explain why PR-HACCP-PS has not reduced the risk of salmonellosis from poultry food.

Another lesson learned from the Covid-19 pandemic was that acquired immunity to the pathogen (severe acute respiratory syndrome coronavirus 2 or SARS-CoV-2) from vaccination or recent illness, reduces the risk of hospitalization and death [30]. Thus, in PFARM, the risk endpoint is severe salmonellosis (hospitalization, death) instead of salmonellosis (illness, hospitalization, death) because illness is viewed as important for maintaining consumer immunity to salmonellosis and reducing the risk of hospitalization and death.

Finally, a risk-to-risk comparison of the COVID-19 pandemic to salmonellosis from chicken indicates that it would take 1931 years for salmonellosis from chicken to cause the same number of hospitalizations as the COVID-19 pandemic and 18,073 years for salmonellosis from chicken to cause the same number of deaths as the COVID-19 pandemic [31, 32]. Thus, compared to COVID-19, salmonellosis from chicken poses a much lower risk to public health.

3. Conclusions

The comparison of QMRA and PFARM methods indicates that course corrections are needed to better address the risk of salmonellosis from poultry food. First, a shift from the probabilistic data collection and modeling methods used in QMRA to the rare event data collection and modeling methods used in PFARM is needed to better simulate and predict the risk of salmonellosis from poultry food. Second, a shift in focus from the pre-harvest and post-harvest sectors in QMRA to the consumer sector in PFARM is needed because the most important risk factors for salmonellosis are in the consumer sector. This would involve the application of PFARM data collection and modeling methods in the consumer sector to better identify lots of unsafe poultry food, if they exist. The PFARM methods could also be used to test the hypothesis that salmonellosis from poultry food is at a baseline level that cannot be reduced further.

Acknowledgments

The author thanks Vijay Juneja of the United States Department of Agriculture, Agricultural Research Service for his help in publishing this work. Mention of trade names or commercial products is solely for providing specific information and does not imply recommendation or endorsement by the United States Department of Agriculture, which is an equal opportunity provider and employer. The opinions expressed are those of the author and do not represent an official position of the United States Department of Agriculture.

Conflict of interest

The author declares no conflict of interest.

References

1. McNamara AM. Generic HACCP application in broiler slaughter and processing. Journal of Food Protection. 1997;60:579-604
2. Williams MS, Ebel ED, Saini G, Nyirabahizi E. Changes in Salmonella contamination in meat and poultry since the introduction of the pathogen reduction and hazard analysis and critical control point rule. Journal of Food Protection. 2020;83(10):1707-1717. DOI: 10.4315/JFP-20-126
3. Kim M, Barnett-Neefs C, Chavez RA, Kealey E, Wiedmann M, Stasiewicz MJ. Risk assessment predicts most of the salmonellosis risk in raw chicken parts is concentrated in those few products with high levels of high-virulence serotypes of Salmonella. Journal of Food Protection. 2024;87(7):100304. DOI: 10.1016/j.jfp.2024.100304
4. Oscar TP. Construction and validation of tertiary models for predicting growth of Salmonella Infantis in chicken liver during a processing chain deviation. Frontiers in Sustainable Food Systems. 2025;9:1565247. DOI: 10.3389/fsufs.2025.1565247
5. Oscar TP. Poultry food assess risk model for Salmonella and chicken gizzards: III. Dose consumed step. Journal of Food Protection. 2024;87(4):100242. DOI: 10.1016/j.jfp.2024.100242
6. Oscar TP. Poultry food assess risk model for Salmonella and chicken gizzards: II. Illness dose step. Journal of Food Protection. 2023;86(6):100091. DOI: 10.1016/j.jfp.2023.100091
7. Oscar TP. Validation software tool (ValT) for predictive microbiology based on the acceptable prediction zones method. International Journal of Food Science and Technology. 2020;55(7):2802-2812. DOI: 10.1111/ijfs.14534
8. Oscar TP. Poultry food assess risk model for Salmonella and chicken gizzards: I. Initial contamination. Journal of Food Protection. 2023;86(2):100036. DOI: 10.1016/j.jfp.2022.100036
9. Thomas CJ, McMeekin TA. Attachment of Salmonella spp. to chicken muscle surfaces. Applied and Environmental Microbiology. 1981;42(1):130-134. DOI: 10.1128/aem.42.1.130-134.1981
10. Thomas CJ, McMeekin TA. Contamination of broiler carcass skin during commercial processing procedures: An electron microscopic study. Applied and Environmental Microbiology. 1980;40(1):133-144. DOI: 10.1128/aem.40.1.133-144.1980
11. Lillard HS. Comparison of sampling methods and implications for bacterial decontamination of poultry carcasses by rinsing. Journal of Food Protection. 1988;51(5):405-408. DOI: 10.4315/0362-028X-51.5.405
12. Simmons M, Fletcher DL, Berrang ME, Cason JA. Comparison of sampling methods for the detection of Salmonella on whole broiler carcasses purchased from retail outlets. Journal of Food Protection. 2003;66(10):1768-1770. DOI: 10.4315/0362-028x-66.10.1768
13. Parveen S, Taabodi M, Schwarz JG, Oscar TP, Harter-Dennis J, White DG. Prevalence and antimicrobial resistance of Salmonella recovered from processed poultry. Journal of Food Protection. 2007;70(11):2466-2472. DOI: 10.4315/0362-028x-70.11.2466
14. Surkiewicz BF, Johnston RW, Moran AB, Krumm GW. A bacteriological survey of chicken eviscerating plants. Food Technology. 1969;23(August):80-85
15. Bemrah N, Bergis H, Colmin C, Beaufort A, Millemann Y, Dufour B, et al. Quantitative risk assessment of human salmonellosis from the consumption of a Turkey product in collective catering establishments. International Journal of Food Microbiology. 2003;80(1):17-30. DOI: 10.1016/s0168-1605(02)00145-9
16. Oscar TP. Risk of salmonellosis from chicken parts prepared from whole chickens sold in flow pack wrappers and subjected to temperature abuse. Journal of Food Protection. 2017;80(9):1496-1505. DOI: 10.4315/0362-028X.JFP-17-097
17. De Man JC. MPN tables, corrected. European Journal of Applied Microbiology and Biotechnology. 1983;17:301-305
18. WHO FAO. Risk Assessments of Salmonella in Eggs and Broiler Chickens. Microbiological Risk Assessment Series 1. Geneva, Switzerland: World Health Organization and Food and Agricultural Organization; 2002. p. 34
19. Oscar TP. Dose-response model for 13 strains of Salmonella. Risk Analysis. 2004;24(1):41-49. DOI: 10.1111/j.0272-4332.2004.00410.x
20. Oscar TP. A quantitative risk assessment model for Salmonella and whole chickens. International Journal of Food Microbiology. 2004;93(2):231-247. DOI: 10.1016/j.ijfoodmicro.2003.12.002
21. Smadi H, Sargeant JM. Quantitative risk assessment of human salmonellosis in Canadian broiler chicken breast from retail to consumption. Risk Analysis. 2013;33(2):232-248. DOI: 10.1111/j.1539-6924.2012.01841.x
22. Juneja VK, Valenzuela Melendres M, Huang L, Gumudavelli V, Subbiah J, Thippareddi H. Modeling the effect of temperature on growth of Salmonella in chicken. Food Microbiology. 2007;24(4):328-335. DOI: 10.1016/j.fm.2006.08.004
23. Oscar TP. General regression neural network and Monte Carlo simulation model for survival and growth of Salmonella on raw chicken skin as a function of serotype, temperature, and time for use in risk assessment. Journal of Food Protection. 2009;72(10):2078-2087. DOI: 10.4315/0362-028x-72.10.2078
24. Murphy RY, Duncan LK, Johnson ER, Davis MD, Smith JN. Thermal inactivation D- and z-values of Salmonella serotypes and Listeria innocua in chicken patties, chicken tenders, franks, beef patties, and blended beef and Turkey patties. Journal of Food Protection. 2002;65(1):53-60. DOI: 10.4315/0362-028x-65.1.53
25. Stathas L, Aspridou Z, Koutsoumanis K. Quantitative microbial risk assessment of Salmonella in fresh chicken patties. Food Research International. 2024;178:113960. DOI: 10.1016/j.foodres.2024.113960
26. Oscar TP. Modeling the effect of inoculum size on the thermal inactivation of Salmonella Typhimurium to elimination in ground chicken thigh meat. American Journal of Food Science and Technology. 2017;5(4):135-142. DOI: 10.12691/ajfst-5-4-3
27. Ross T. Indices for performance evaluation of predictive models in food microbiology. Journal of Applied Bacteriology. 1996;81(5):501-508. DOI: 10.1111/j.1365-2672.1996.tb03539.x
28. Halaji M, Heiat M, Faraji N, Ranjbar R. Epidemiology of COVID-19: An updated review. Journal of Research in Medical Sciences: The Official Journal of Isfahan University of Medical Sciences. 2021;26:82. DOI: 10.4103/jrms.JRMS_506_20
29. McCullough NB, Eisele CW. Experimental human salmonellosis: II. Immunity studies following experimental illness with Salmonella meleagridis and Salmonella anatum. Journal of Immunology. 1951;66(5):595-608
30. Araf Y, Akter F, Tang YD, Fatemi R, Parvez MSA, Zheng C, et al. Omicron variant of SARS-CoV-2: Genomics, transmissibility, and responses to current COVID-19 vaccines. Journal of Medical Virology. 2022;94(5):1825-1832. DOI: 10.1002/jmv.27588
31. CDC. COVID Data Tracker. 2023 August 29. 2023. Available from: https://covid.cdc.gov/covid-data-tracker/#datatracker-home
32. Anonymous. National Enteric Disease Surveillance: Salmonella Annual Report, 2016. 2018. Available from: https://www.cdc.gov/nationalsurveillance/pdfs/2016-Salmonella-report-508.pdf

[1] 1. McNamara AM. Generic HACCP application in broiler slaughter and processing. Journal of Food Protection. 1997;60:579-604

[2] 2. Williams MS, Ebel ED, Saini G, Nyirabahizi E. Changes in Salmonella contamination in meat and poultry since the introduction of the pathogen reduction and hazard analysis and critical control point rule. Journal of Food Protection. 2020;83(10):1707-1717. DOI: 10.4315/JFP-20-126

[3] 3. Kim M, Barnett-Neefs C, Chavez RA, Kealey E, Wiedmann M, Stasiewicz MJ. Risk assessment predicts most of the salmonellosis risk in raw chicken parts is concentrated in those few products with high levels of high-virulence serotypes of Salmonella. Journal of Food Protection. 2024;87(7):100304. DOI: 10.1016/j.jfp.2024.100304

[4] 4. Oscar TP. Construction and validation of tertiary models for predicting growth of Salmonella Infantis in chicken liver during a processing chain deviation. Frontiers in Sustainable Food Systems. 2025;9:1565247. DOI: 10.3389/fsufs.2025.1565247

[5] 5. Oscar TP. Poultry food assess risk model for Salmonella and chicken gizzards: III. Dose consumed step. Journal of Food Protection. 2024;87(4):100242. DOI: 10.1016/j.jfp.2024.100242

[6] 6. Oscar TP. Poultry food assess risk model for Salmonella and chicken gizzards: II. Illness dose step. Journal of Food Protection. 2023;86(6):100091. DOI: 10.1016/j.jfp.2023.100091

[7] 7. Oscar TP. Validation software tool (ValT) for predictive microbiology based on the acceptable prediction zones method. International Journal of Food Science and Technology. 2020;55(7):2802-2812. DOI: 10.1111/ijfs.14534

[8] 8. Oscar TP. Poultry food assess risk model for Salmonella and chicken gizzards: I. Initial contamination. Journal of Food Protection. 2023;86(2):100036. DOI: 10.1016/j.jfp.2022.100036

[9] 9. Thomas CJ, McMeekin TA. Attachment of Salmonella spp. to chicken muscle surfaces. Applied and Environmental Microbiology. 1981;42(1):130-134. DOI: 10.1128/aem.42.1.130-134.1981

[10] 10. Thomas CJ, McMeekin TA. Contamination of broiler carcass skin during commercial processing procedures: An electron microscopic study. Applied and Environmental Microbiology. 1980;40(1):133-144. DOI: 10.1128/aem.40.1.133-144.1980

[11] 11. Lillard HS. Comparison of sampling methods and implications for bacterial decontamination of poultry carcasses by rinsing. Journal of Food Protection. 1988;51(5):405-408. DOI: 10.4315/0362-028X-51.5.405

[12] 12. Simmons M, Fletcher DL, Berrang ME, Cason JA. Comparison of sampling methods for the detection of Salmonella on whole broiler carcasses purchased from retail outlets. Journal of Food Protection. 2003;66(10):1768-1770. DOI: 10.4315/0362-028x-66.10.1768

[13] 13. Parveen S, Taabodi M, Schwarz JG, Oscar TP, Harter-Dennis J, White DG. Prevalence and antimicrobial resistance of Salmonella recovered from processed poultry. Journal of Food Protection. 2007;70(11):2466-2472. DOI: 10.4315/0362-028x-70.11.2466

[14] 14. Surkiewicz BF, Johnston RW, Moran AB, Krumm GW. A bacteriological survey of chicken eviscerating plants. Food Technology. 1969;23(August):80-85

[15] 15. Bemrah N, Bergis H, Colmin C, Beaufort A, Millemann Y, Dufour B, et al. Quantitative risk assessment of human salmonellosis from the consumption of a Turkey product in collective catering establishments. International Journal of Food Microbiology. 2003;80(1):17-30. DOI: 10.1016/s0168-1605(02)00145-9

[16] 16. Oscar TP. Risk of salmonellosis from chicken parts prepared from whole chickens sold in flow pack wrappers and subjected to temperature abuse. Journal of Food Protection. 2017;80(9):1496-1505. DOI: 10.4315/0362-028X.JFP-17-097

[17] 17. De Man JC. MPN tables, corrected. European Journal of Applied Microbiology and Biotechnology. 1983;17:301-305

[18] 18. WHO FAO. Risk Assessments of Salmonella in Eggs and Broiler Chickens. Microbiological Risk Assessment Series 1. Geneva, Switzerland: World Health Organization and Food and Agricultural Organization; 2002. p. 34

[19] 19. Oscar TP. Dose-response model for 13 strains of Salmonella. Risk Analysis. 2004;24(1):41-49. DOI: 10.1111/j.0272-4332.2004.00410.x

[20] 20. Oscar TP. A quantitative risk assessment model for Salmonella and whole chickens. International Journal of Food Microbiology. 2004;93(2):231-247. DOI: 10.1016/j.ijfoodmicro.2003.12.002

[21] 21. Smadi H, Sargeant JM. Quantitative risk assessment of human salmonellosis in Canadian broiler chicken breast from retail to consumption. Risk Analysis. 2013;33(2):232-248. DOI: 10.1111/j.1539-6924.2012.01841.x

[22] 22. Juneja VK, Valenzuela Melendres M, Huang L, Gumudavelli V, Subbiah J, Thippareddi H. Modeling the effect of temperature on growth of Salmonella in chicken. Food Microbiology. 2007;24(4):328-335. DOI: 10.1016/j.fm.2006.08.004

[23] 23. Oscar TP. General regression neural network and Monte Carlo simulation model for survival and growth of Salmonella on raw chicken skin as a function of serotype, temperature, and time for use in risk assessment. Journal of Food Protection. 2009;72(10):2078-2087. DOI: 10.4315/0362-028x-72.10.2078

[24] 24. Murphy RY, Duncan LK, Johnson ER, Davis MD, Smith JN. Thermal inactivation D- and z-values of Salmonella serotypes and Listeria innocua in chicken patties, chicken tenders, franks, beef patties, and blended beef and Turkey patties. Journal of Food Protection. 2002;65(1):53-60. DOI: 10.4315/0362-028x-65.1.53

[25] 25. Stathas L, Aspridou Z, Koutsoumanis K. Quantitative microbial risk assessment of Salmonella in fresh chicken patties. Food Research International. 2024;178:113960. DOI: 10.1016/j.foodres.2024.113960

[26] 26. Oscar TP. Modeling the effect of inoculum size on the thermal inactivation of Salmonella Typhimurium to elimination in ground chicken thigh meat. American Journal of Food Science and Technology. 2017;5(4):135-142. DOI: 10.12691/ajfst-5-4-3

[27] 27. Ross T. Indices for performance evaluation of predictive models in food microbiology. Journal of Applied Bacteriology. 1996;81(5):501-508. DOI: 10.1111/j.1365-2672.1996.tb03539.x

[28] 28. Halaji M, Heiat M, Faraji N, Ranjbar R. Epidemiology of COVID-19: An updated review. Journal of Research in Medical Sciences: The Official Journal of Isfahan University of Medical Sciences. 2021;26:82. DOI: 10.4103/jrms.JRMS_506_20

[29] 29. McCullough NB, Eisele CW. Experimental human salmonellosis: II. Immunity studies following experimental illness with Salmonella meleagridis and Salmonella anatum. Journal of Immunology. 1951;66(5):595-608

[30] 30. Araf Y, Akter F, Tang YD, Fatemi R, Parvez MSA, Zheng C, et al. Omicron variant of SARS-CoV-2: Genomics, transmissibility, and responses to current COVID-19 vaccines. Journal of Medical Virology. 2022;94(5):1825-1832. DOI: 10.1002/jmv.27588

[31] 31. CDC. COVID Data Tracker. 2023 August 29. 2023. Available from: https://covid.cdc.gov/covid-data-tracker/#datatracker-home

[32] 32. Anonymous. National Enteric Disease Surveillance: Salmonella Annual Report, 2016. 2018. Available from: https://www.cdc.gov/nationalsurveillance/pdfs/2016-Salmonella-report-508.pdf