Skip to main content
ARS Home » Research » Publications at this Location » Publication #243666

Title: Estimating removal rates of bacteria from poultry carcasses using two whole-carcass rinse volumes

item WILLIAMS, MICHAEL - Food Safety Inspection Service (FSIS)
item EBEL, ERIC - Food Safety Inspection Service (FSIS)
item GOLDEN, NEAL - Food Safety Inspection Service (FSIS)
item Berrang, Mark
item BAILEY, JOSEPH - Biomerieux, Inc
item HARTNETT, EMMA - DeCisionalysis Risk Consultants

Submitted to: International Journal of Food Microbiology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/2/2010
Publication Date: 5/15/2010
Citation: Williams, M., Ebel, E., Golden, N., Berrang, M.E., Bailey, J.S., Hartnett, E. 2010. Estimating removal rates of bacteria from poultry carcasses using two whole-carcass rinse volumes. International Journal of Food Microbiology. 139(3):140-146.

Interpretive Summary: Broiler carcasses may be contaminated with bacteria that can cause human food borne illness. It is important to regulatory officials to be able to sample carcasses and detect low numbers of dangerous bacteria. In the US, broiler carcasses are commonly sampled by placing the carcass into a bag with a specified volume of rinse liquid and performing a whole carcass rinse. The rinsate is then cultured for the bacteria of interest. There are two common rinse methods used in the US; the difference in methods is largely due to difference in volume of liquid used to rinse the carcass. Some laboratories rinse carcasses with 400 mL of liquid while others use 100 mL. It is unclear how the volume of rinse liquid affects the detection of bacteria from broiler carcasses. This study was undertaken to compare broiler carcass rinses conducted with 100 mL to those conducted using 400 mL of rinse liquid. Fully processed broiler carcasses were collected and rinsed in a commercial processing plant using either 100 or 400 mL of liquid. Rinse liquid was cultured for E. coli, Enterobacteriaceae and Campylobacter. Statistical analysis was conducted on the numbers of bacteria detected on carcasses rinsed with each volume of liquid. Statistical models suggest that under the conditions tested, 400 mL of rinse liquid is more effective to remove and allow recovery of E. coli and members of Enterobacteriaceae. Campylobacter, however, was detected more frequently in 100 mL rinsates than in 400 mL rinsates. This study demonstrates the utility of Bayesian model statistics to compare rinse volume for microbiological examination of broiler carcasses. Future studies with larger sample numbers are required to improve the accuracy of the estimates determined herein.

Technical Abstract: Rinse sampling is a common method for determining the level of microbial contamination on poultry carcasses. One of the advantages of rinse sampling, over other carcass sampling methods, is that the results can be used for both process control applications and to estimate the total microbial level on a carcass. The latter objective is possible because rinse sampling removes a portion of the bacteria from the entire carcass, whereas methods such as neck-skin sampling focus on a small area of the carcass where the level of contamination may not be representative of the entire carcass. Two recurring issues with rinse sampling are differences in sampling protocols and the difficulty of determining the proportion of bacteria removed during sampling. A situation arose where 300 rinse samples were collected using two different rinse fluid volumes (i.e., 100 and 400 mL). The original intent of the study was to demonstrate the similarity of the removal rates for the two methods, but summary statistics suggested substantial differences. A Bayesian model was constructed to estimate the removal rates for the two sampling methods as well as to estimate the parameters of distributions describing the carcass-level contamination across 3 days of processing. The results of the study suggest that nearly 7 times as many bacteria are removed from the carcass when using a 400 mL rinse sample than with a 100 mL rinse sample. While this estimate is subject to a rather large degree of uncertainty, the 95% Bayesian credible interval of (4.3, 10.7) still indicates a significant difference in the removal rates for the two sampling methods.