Poultry food assess risk model for salmonella and chicken gizzards: III. Dose consumed step

Authors: Oscar, Thomas

Submitted to: Journal of Food Protection
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/9/2024
Publication Date: 2/14/2024
Citation: Oscar, T.P. 2024. Poultry food assess risk model for salmonella and chicken gizzards: III. Dose consumed step. Journal of Food Protection. 87(4):100242. https://doi.org/10.1016/j.jfp.2024.100242.
DOI: https://doi.org/10.1016/j.jfp.2024.100242

Interpretive Summary: Salmonella bacteria that contaminate poultry food are a leading cause of illness from this pathogen in the United States and throughout the world. Quantitative Microbial Risk Assessment (QMRA) is used to assess consumer exposure to Salmonella bacteria that contaminate poultry food. However, QMRA uses sampling, diagnostic test, and diagnosis methods that provide a biased and inaccurate prediction of consumer exposure to Salmonella bacteria from poultry food. Therefore, the Poultry Food Assess Risk Model or PFARM was developed to better assess consumer exposure to Salmonella bacteria that contaminate poultry food. The PFARM uses sampling, diagnostic test, and diagnosis methods that address problems in QMRA and thus, provide a better prediction of consumer exposure to Salmonella bacteria that contaminate poultry food. In the present study, the QMRA and PFARM were compared and it was found that QMRA overestimated consumer exposure to Salmonella from chicken gizzards. The reasons why QMRA overestimated consumer exposure to Salmonella from chicken gizzards were discussed.

Technical Abstract: The Dose Consumed (DC) step of the Poultry Food Assess Risk Model (PFARM) for Salmonella and chicken gizzards (CGs) was presented and compared to the Exposure Assessment step of Quantitative Microbial Risk Assessment (QMRA). The specific objectives were: 1) to demonstrate the DC step of PFARM for Salmonella and CGs; 2) to compare Salmonella DC from cooked CGs and cross-contaminated lettuce (Lcc); 3) to determine if Salmonella DC changed over time in a production chain; and 4) to compare PFARM and QMRA predictions of Salmonella DC. The PFARM and QMRA were created in Excel and simulated with @Risk. Salmonella prevalence (Pr) and number (N) data (n = 100) for CGs (56 g) and scenario analysis were used to address objectives 1, 2, and 4, whereas running windows of 60 consecutive CG samples and scenario analysis were used to address objective 3. A lot size of 1,000 kg of CGs was simulated. Mean portion size was 168 g resulting in simulation of 5,952 meals per lot. Of these, 3.69 ± 0.32% and 0.49 ± 0.07% resulted in Salmonella DC of = 1 per meal from cooked CGs and Lcc, respectively. However, Salmonella DC per lot from cooked CGs (272 ± 27; mean ± SD) was less (P = 0.05) than from Lcc (6,050 ± 4,929). The Salmonella DC per lot from Lcc was highly variable and uncertain because it was driven by a few high Salmonella DC of > 1,000 per meal from Lcc that was temperature abused. Over time in the production chain, Salmonella Pr per 5,952 meals and Salmonella DC per lot changed (P = 0.05) but the patterns differed. The QMRA predicted higher (P = 0.05) Salmonella DC per meal than PFARM. In part, this was because QMRA only simulated contaminated grams, whereas PFARM simulated contaminated and non-contaminated servings. However, other factors, which are discussed, also contributed to overestimation of Salmonella DC by QMRA.