Tracking and modeling norovirus transmission during mechanical slicing of globe tomatoes

Recent epidemiological evidence indicates that preparation of fresh produce for use as ingredients in ready-to-eat food in commercial settings has been a significant source of the norovirus (NoV) infections in the U.S. This research investigated the dissemination of NoV from a single tomato to many others via the use of an 11-horizontal blade slicer commonly found in restaurants or sandwich shops. A total of eight trials were conducted. The source of contamination in each trial was a soak-inoculated, air-dried globe tomato containing ~8log10 murine norovirus (MNV). Each trial began by slicing a single un-inoculated tomato in the slicer, followed by slicing an inoculated tomato. This was then followed by slicing 9 to 27 un-inoculated tomatoes. A similar and constant hand pressure on the slicer was used in every trial. Three slices from each tomato were collected for virus elution, concentration, and extraction before RT-PCR detection of MNV. The change in MNV per sliced tomato was averaged over all eight trials, and two mathematical models were fit to the average data using a logarithmic model or a power model. Regression analysis determined that the equation that best fit the data was y=-0.903∗ln(x)+7.945, where y=log10 MNV per slicing and x=tomato slicing number. An acceptable fit (R(2)=0.913) was indicated. The MNV levels transferred (y) generally decreased as the number of tomatoes sliced (x) increased, with some exceptions. Infrequent but erratic transfers, where the MNV level of a subsequent tomato was higher than that of a preceding tomato, occurred in later transfer of some trials. In contrast, the first and second transfers of each trial were always shown to have sharply decreased levels of MNV from the inoculum. The MNV log10 reduction per slicing event changes throughout the process: with a predicted 0.63log10 reduction from tomato 1 to tomato 2 (76% reduction); a 0.07log10 reduction predicted from tomato 13 to tomato 14 (a 14% reduction); and 0.03log10 reduction predicted from tomato 27 to tomato 28 (a 7% reduction). Virus transfer is clearly variable even given the consistent slicing procedure used throughout each trial. This study illustrates the complex nature of risk prediction associated with NoV cross-contamination during food preparation in commercial establishments.