Skip to Main content Skip to Navigation

Quantitative risk assessment model of human salmonellosis linked to the consumption of ground beef

Abstract : Salmonellae are one of the most important causes of foodborne illness associated with raw dairy products. The assessment of the real risk associated with the consumption of these products is needed and the most appropriate method to achieve this goal is the risk analysis process which links pathogens in food to the public health problem. The main aim of this thesis is to quantitatively assess the risk of human salmonellosis linked to the consumption of ground beef. A data gap that is routinely identified in risk assessment is the lack of quantitative data on pathogens contaminated food. Therefore, as a first objective of this thesis, we developed a rapid, sensitive and reliable method for the quantification of Salmonella in artificially contaminated bovine fecal samples. The method combines the principles of most-probable-number (MPN) method with a real-time PCR assay. With this developed assay (MPN-real-time PCR) low levels of Salmonella (1-5 CFU/mL) in fecal could be enumerated after 8 h of non-selective enrichment in buffered peptone water. All estimated MPN counts corresponded well to the estimated contamination level of Salmonella inoculated into fecal samples. In order to evaluate the utility of this developed quantification assay, our second objective was to apply it to naturally contaminated fecal samples collected from slaughterhouse located in Meaux, France weekly in February and March 2006 (an average of 40 samples per visit). 9.12% (27/296), and 34.62% (9/26) fecal and environmental samples, respectively, were found Salmonella-positive, with estimated MPN values or counts of Salmonella ranging from <1.8 - 1609 MPN/g of fecal samples. The mean of the log10 concentration of Salmonella is 0.6189 MPN/g with standrdard deviations of 2.7112 by using the censored regression approach. Counts were generally low, with the exception of 6 animals (>1400 MPN/g), while all the other 21 Salmonella positive animals had faeces with less than 80 MPN/g (from them 13 animals with MPN values <1.8 MPN/g. The prevalence of Salmonella showed no significant difference (p=1) between French (8.63%, 17/197) and Belgian cattle (10%, 10/99). Furthermore, neither the animals' area of origin (p=0.75), age (p=0.18), race (p=0.94), breed (p=0.23), or movement of the animal (p=0.89) had any impact on the prevalence of Salmonella. The application of the MPN-real-time PCR assay for quantifying Salmonella in fecal proved to be rapid, easy-to-perform and highly sensitive. In the assessment of potential risks associated with Salmonella in ground beef it was necessary to examine the ability of Salmonella to grow in ground beef. Therefore, we presented in this thesis as a third objective, model of a growth / no growth interface of Salmonella in ground beef. A growth/no growth data were modeled by logistic polynomial regression on the available growth data for Salmonella in ground beef in published papers and all the data related to ground beef in ComBase in order to lead us to an accurate description of the conditions that Salmonella can growth / no growth. The overall results indicate clearly that the temperature is the most important and the only factor significant in the study. There was no growth observed at temperature less than 10°C. Where as the temperature 10°C and 12°C are the only temperature; we did observe growth / no growth (sometimes with the same conditions). Even though pH and water activity are important factors for microbial growth, in our study they have no effect due to meat structure. Finally, a quantitative risk assessment of human salmonellosis linked to the consumption of ground beef is presented. Different distributions were assumed for the parameters of the model and a Monte Carlo simulation was used to model the process and to quantify the risk associated with the consumption of 100 g serving of ground beef patty. The expected percentage of ground beef patties with contamination greater than 5, 10 and 100 Salmonella cells were 29%, 17.1% and 0.02%, respectively at the time of consumption. The risk of salmonellosis per 100 g serving ranged from 0 to 2.33E-06 dependent on the type of cooking and the fat content. For 10 million servings of 100g, the expected number of cases of salmonellosis predicted by the model is in average of 11.04, 12.33 for fat content 7% and 24% respectively. The risk of salmonellosis was closed to zero when the 100 g serving ground beef patties consumed well done. The relative risk of getting salmonellosis from consumed the rare ground beef patties is 312, 61 times higher for fat content 7% and 24% respectively comparing to the consumption of well done patties. There are 35 batches with cases out of 2000 batches (1.8%). 15 of them have 2 cases or more (0.75%, 15/2000). Despite the limitations and the data gap, we demonstrated the benefit of risk assessment not only as a risk evaluation tool but also as a helping device in the decision-making and the risk management.
Document type :
Domain :
Complete list of metadatas
Contributor : Ecole Agroparistech <>
Submitted on : Tuesday, December 2, 2008 - 8:00:00 AM
Last modification on : Monday, October 19, 2020 - 11:07:52 AM
Long-term archiving on: : Saturday, November 26, 2016 - 7:21:03 PM


  • HAL Id : pastel-00003795, version 1



Abdunaser Dayhum. Quantitative risk assessment model of human salmonellosis linked to the consumption of ground beef. domain_other. AgroParisTech, 2008. English. ⟨NNT : 2008AGPT0036⟩. ⟨pastel-00003795⟩



Record views


Files downloads