Modelling transfer of Listeria monocytogenes during slicing of 'gravad' salmon

Transfer of a rifampicin-resistant mutant of Listeria monocytogenes from an inoculated slicing blade to slices of 'gravad' salmon (Salmo salar), and from inoculated salmon fillet to the slicing machine and subsequently to slices of uninoculated fillet was studied. The effect of slicing temperature (0 °C, 10 °C and room temperature), inoculum level (approx. 3, 5 and 8 log CFU/blade), and attachment time of inoculum to blade (10 min and 2.5 h) were investigated and predictive models of the transfer were produced. In the tests of transfer from inoculated blade (5.9-9.0 log CFU/blade) initially 2.5-5.3 log CFU/g was present on the slices, slowly decreasing to an overall average decrease of 1.6 ± 0.2 log CFU/g during slicing of 39 slices; the lowest reduction being 1.3 log CFU/g at 0 °C. In tests of transfer from contaminated salmon (7.6 ± 0.1 log CFU/fillet) to uninoculated blade and further to uninoculated salmon, the reduction in number of L. monocytogenes in slices was 1.5 log CFU/g during slicing of 39 slices. For example 5.3 ± 0.3 log CFU/g was transferred to second slice when the inoculum level was 8.4 ± 0.4 log CFU/blade, but clearly (p < 0.05) lower total number of L. monocytogenes were transferred to slices when the inoculum level was lower, the temperature was colder or the attachment time was longer. There was a progressive exponential reduction in the quantity of L. monocytogenes transferred and, based on statistical parameters, an exponential model (y = a * e(-x/b)) fit the data from different test conditions and was suitable for predicting an expected number of L. monocytogenes on the salmon slices. Based on the predicted values, the logarithmic reduction in number of L. monocytogenes in slices was highest at room temperature with an inoculum level of 8.4 ± 0.4 log CFU/blade (attachment time 10 min); the other test conditions differed significantly from this (p < 0.05). Despite statistically significant differences, in all test conditions the number of bacteria were predicted to reduce quite rapidly (i.e. after slicing of the fourth fillet) to < 1 log CFU/g, though this prediction was an extrapolation after 39 slices. The predictive models described herein can assist salmon processors and regulatory agencies in assessing cross-contamination from contaminated slicing machines to product and in designing risk management strategies. © 2007 Elsevier B.V. All rights reserved.