Salmonella prevalence and diversity are impacted by sampling methodology

Introduction:

Multiple factors can influence both Salmonella prevalence and diversity of Salmonella serogroups detected from both pre- and post-harvest samples. Sampling methodologies vary widely when sampling the environment, individual live birds, carcasses, and parts. Furthermore, protocols for analyzing the samples following collection can also impact both prevalence and serogroup diversity. With more than 2,400 serotypes, it is reasonable to expect that different Salmonella serotypes will have differing abilities to grow and compete in different environments such as dust vs. litter vs. ceca. Even different isolates of the same serotype can have a greater or lesser ability to colonize birds. Differing abilities of certain serogroups to grow during cultivation may impact which serogroups are recovered. However, those serogroups recovered may not be representative of the serogroups present in the original sample.

Environmental Sampling

Many different production house environmental sampling methods for the recovery of Salmonella on the farm have been studied. Environmental sampling methods can be divided into two categories. Direct component methods incorporate removal and analysis of samples such as feces, litter, or dust. Surface sampling methods rely on contact between the environment and the moistened sampling media such as conventional drag swabs, "socks", shoe covers, and ISODS (intermittently stepped on drag swabs).

In a study comparing direct sampling methods (feces and litter) and surface sampling methods (drag swabs = gauze squares and socks = elastic bandage), sampling with socks yielded the highest percentage of samples positive for Salmonella in both challenged pens with high Salmonella levels and the adjacent non-challenged pens with lower levels of Salmonella (Buhr, et al., 2007). Surface sampling methods including drag swabs, ISODS, and socks were further compared. It was demonstrated that the ISODS and sock methods that incorporated a greater contact pressure with the litter surface from using a foot to step on the sampling material were more sensitive for the recovery of low levels of Salmonella in comparison to conventional drag swabs, which are simply dragged across the litter surface (Buhr, et al., 2007).

When boot swabs (similar to socks), drag swabs, litter, and dust were compared in various types of turkey flocks, the boot swabs and dust had the highest percentages of Salmonella detection followed by litter samples. Conventional drag swabs had the lowest percentage of positive samples (Mueller-Doblies, et al., 2009). When a turkey house was sampled with boot swabs, dust, and feces, both the boot swab and dust samples were again more sensitive than individual feces samples (Arnold et al., 2009). Overall, when taking house environmental samples for the detection of Salmonella, sampling methods that include stepping on the moist sampling material while in contact with the litter are more sensitive than conventional drag swab sampling of the litter surface.

In addition to environmental samples, flock status can also be determined through individual bird sampling. These types of samples can either be non-destructive or destructive, and include cloacal swabs, individual bird feces, and intestinal ceca. In studies utilizing non-destructive methods of cloacal swabs and fecal samples, it was demonstrated that fecal samples were much more sensitive for the detection of Salmonella than cloacal swabs (Ayachi, et al., 2010; Garcia, et al., 2011). When using a destructive sampling approach, sampling of the ceca (vs. other intestinal segments or organs) resulted in the most positive samples (Barrow, et al., 1988). Sampling in a turkey production facility comparing ceca (destructive method) and environmental swab samples (non-destructive) demonstrated that sampling ceca was three times more sensitive than environmental swabs (Sanad et al., 2016). When sampling individual birds, ceca or fecal samples are the most sensitive in the detection of Salmonella because the higher resident levels of Salmonella than on the cloaca epithelial surface.

In addition to what samples are taken, when the samples are taken can also influence the ability to detect Salmonella. During broiler growout (low Salmonella challenge) for both ISODS and ceca sampling methods, Salmonella can be detected at the highest frequency at about 3 weeks of age, after which detection steadily declines from 3 to 6 weeks of age (Wilson, et al., 2016). A similar trend is seen in turkeys where the greatest number of samples was positive for Salmonella during rearing at 14 days of age (Morris, et al., 2015). The frequency of positive samples decreased after 14 days through 20 weeks. 

Feed withdrawal can also influence the ability to detect Salmonella. When broilers are subjected to feed withdrawal, the crop empties within 6 hours and the pH within the crop increases to near neutral leading to the growth of enteropathogens such as Salmonella when present (Hinton, et al., 2002). In a study where Salmonella negative broilers were placed on Salmonella positive litter during feed withdrawal, Salmonella was frequently detected (48%) from crops and on two occasions (5%) in the ceca after a 12 hour period (Buhr, et al. 2017).

In addition to feed withdrawal prior to processing, restriction feeding of breeders may also impact Salmonella proliferation and persistence. In a study comparing every-day and skip-a-day feeding programs with broiler breeder pullets, pullets that were fed every day had a lower prevalence and shorter persistence of Salmonella than pullets that were fed with a skip-a-day program (Wilson, et al., 2013). Emptying of the crop every other day, allowing for an increase in crop pH and litter picking may contribute to a higher prevalence and persistence of Salmonella positive ceca.

When culturing for Salmonella detection, they type of selective media used can significantly impact the observed prevalence. In a study comparing multiple combinations of enrichment broths and plating medias, the selective enrichment broths impacted detection differences more than the solid medias tested. Incubation in Rappaport Vassiliadis with soy peptone was found to be preferable to incubation in selenite cysteine or tetrathionate with novobiocin (Schönenbrücher, et al., 2008). This study used a variety of minced meat products and edible offal for analysis. Preferred media for detection may depend on sample type and matrix. 

Testing of pooled neck skins or a whole carcass rinse are the most common methods when sampling carcasses following processing for meeting regulatory standards. Another carcass sampling method sometimes used for research purposes is whole carcass enrichment. Whole carcass enrichment entails incubating the entire carcasses in the enrichment media for 24 h prior to sample analysis. In a research study where carcasses were sampled following immersion chilling, sampling by whole carcass enrichment yielded the greatest Salmonella prevalence, followed by neck skin, then the whole carcass rinse method (Bourassa, et al. 2015).

During turkey and broiler processing the prevalence of Salmonella on carcasses changes between each step of the process. The levels of Enterobacteriaceae, of which Salmonella is a member, start high on the feathers and skin when the birds arrive at the processing plant. Levels are decreased during scalding but increase during defeathering. After defeathering, process controls including mechanical and chemical interventions continue to decrease levels of bacteria on the surface of the carcasses (Berrang and Dickens, 2000).

Multiple sampling methods including rinse, excision (skin or muscle), swab, neck skin, and whole carcass enrichment have been evaluated for the detection of Salmonella on carcasses. When sampling turkey carcasses, methods that required sampling of a larger surface area of the carcasses (whole carcass swabbing and whole carcass rinse) were more sensitive in the detection of Salmonella than one or two site swab sampling (McEvoy et al., 2005). When neck skin and whole carcass enrichment Salmonella sampling methods following air chilling were compared in broilers, both methods were found to be equivalent (Bourassa, et al., 2015). However, following immersion chilling, whole carcass enrichment was the most sensitive method, neck skin less sensitive, and whole carcass rinse the least sensitive.

Although sampling methods vary in sensitivity, each method has a minimal level of detection. The most sensitive method (whole carcass enrichment) requires approximately 8-10 cells being present on the carcass to yield a positive result following 24 hours enrichment incubation (Cox and Blankenship, 1975). Therefore, any product sampled can never really be declared “Salmonella-Free”. The absence of a positive result from a carcass or 325 g of comminuted product does not indicate that there is no Salmonella present. Sampling for Salmonella in processing plants is essential for evaluating process control for the minimization of contamination. Use of the phrase “Zero Tolerance” may lead consumers to believe that there is no Salmonella on raw product and may give a false sense of security.

When the detection of Salmonella serogroups was analyzed following three different carcass sampling methods (neck skin, whole carcass rinse, and whole carcass enrichment) the greatest number of serogroups was detected with whole carcass enrichment (four serogroups) followed by neck skin sampling (two serogroups), followed by whole carcass rinse (one serogroup) (Bourassa et al. 2015). It was hypothesized that the greater serogroup detection was due to the incubation of the carcass or skin within the rinse media. Incubation in the media would have allowed for Salmonella that was tightly attached to the carcass to be detected while the whole carcass rinse sample would only contain those Salmonella that were more easily rinsed off during sampling.

Sample enrichment and the use of selective media are commonly used when culturing for the detection of Salmonella. When multiple serotypes are present in a selective media, the serotypes will grow at differing rates, therefore, impacting the likelihood of detecting one serotype over another (Harvey and Price, 1976). In a study assessing Salmonella Newport, Enteritidis, and two Typhimurium strains, it was determined that the probability of detecting one serotype was not directly related to the initial number of cells in the selective media (Singer, et al. 2009). For example, when equal counts of Newport and Enteritidis were inoculated into a broth and allowed to grow, 92% of the Salmonella recovered were Newport. In a study comparing growth of groups of four Salmonella serotypes, levels of growth between serotypes also differed based on growth media (Gorski, et al., 2012).

In addition to serotype competition and variable growth rates, the number of colonies selected per plate for analysis can also impact the resulting determinations of which serotypes are present. For example, if a serotype 1 and a serotype 2 in a mixed culture grew in such a way that 90% of the Salmonella present upon plating were of serotype 1 and only 10% of the Salmonella present were of serotype 2, then if only 1 colonies was picked for serotype identification, there would be a 90% chance that only serotype 1 would be detected. If 3 colonies were picked, there is a 73% chance of selecting only serotype 1, an 8% chance of selecting both, and a 0.1% chance of selecting only serotype 2. When ground chicken meat samples were analyzed by two different methods (FDA or ISO) the ISO method resulted in a greater diversity of serogroups than the FDA method (Temelli, et al. 2010). Within each method, the different media also yielded differing serogroups.

References:

Arnold, M.E., D. Mueller-Doblies, J.J. Carrique-Mas, and R.H. Davies. 2009. The estimation of pooled-ample sensitivity for detection of Salmonella in turkey flocks. J. Appl. Microbiol. 107:936-943.

Ayachi, A. N. Alloui, O. Bennoune, and A. Kassah-Iaouar. 2010. Survey of Salmonella serovars in broilers and laying breeding reproducers in Eastern Algeria. J. Infect. Dev. Ctries. 4:103-106.

Barrow, P.A., J.M. Simpson, and M.A. Lovell. 1988. Intestinal colonization in the chicken by food-poisoning Salmonella serotypes; microbial characteristics associated with fecal excretion. Avian Pathol. 17:571-588.

Berrang, M.E. and J.A. Dickens. 2000. Presence and level of Campylobacter spp. on broiler carcasses throughout the processing plant. J. Appl. Poult. Res. 9:43-47.

Bourassa, D.V., J.M. Holmes, J.A. Cason, N.A. Cox, L.L. Rigsby, and R.J. Buhr. 2015. Prevalence and serogroup diversity of Salmonella for broiler neck skin, whole carcass rinse, and whole carcass enrichment sampling methodologies following air or immersion chilling. J. Food Prot. 78:1938-1944.

Buhr, R.J., D.V. Bourassa, A. Hinton, Jr., B.D. Fairchild, and C.W. Ritz. 2017. Impact of litter Salmonella status during feed withdrawal on Salmonella recovery from the broiler crop and ceca. Poult. Sci. 96:4361-4369.

Buhr, R.J., L.J. Richardson, J.A. Cason, N.A. Cox, and B.D. Fairchild. 2007. Comparison of four sampling methods for the detection of Salmonella in broiler litter. Poult. Sci. 86:21-25.

Cox, N.A. and L.C. Blankenship. 1975. Comparison of rinse sampling methods for detection of salmonellae on eviscerated broiler carcasses. J. Food Sci. 40:1333-1334.

Garcia, C., J.M. Soriano, V. Benitez, and P. Catala-Gregori. 2011. Assessment of Salmonella spp. in feces, cloacal swabs, and eggs (eggshell and content separately) from a laying hen farm. Poult. Sci. 90:1581-1585.

Gorski, L. 2012. Selective enrichment media bias the types of Salmonella enterica strains isolated from mixed strain cultures and complex enrichment broths. PLoS ONE 7:e34722.

Harvey, R.W. and T.H. Price. 1976. The examination of samples infected with multiple Salmonella serotypes. J. Hyg. (Lond.) 65:423-434.

Hinton, Jr., A., R.J. Buhr, and K.D. Ingram. 2002. Carbohydrate-based cocktails that decrease the population of Salmonella and Campylobacter in the crop of broiler chickens subjected to feed withdrawal. Poult. Sci. 81:780-784.

McEvoy, J.M., C.W. Nde, J.S. Sherwood, and C.M. Logue. 2005. An evaluation of sampling methods for the detection of Escherichia coli and Salmonella on turkey carcasses. J. Food Prot. 68:34-39.

Morris, V.K., J.J. Carrique-Mas, D. Mueller-Doblies, R.H. Davies, A.D. Wales, and V.M. Allen. 2015. A longitudinal observational study of Salmonella shedding patterns by commercial turkeys during rearing and fattening, showing limitations of some control measures. British Poult. Sci. 56:48-57.

Mueller-Doblies, D., A.R. Sayers, J.J. Carrique-Mas, and R.H. Davies. 2009. Comparison of sampling methods to detect Salmonella infection of turkey flocks. J. Appl. Microbiol. 107:635-645.

Sanad, Y.M., K. Johnson, S.H. Park, J. Han, J. Deck, S.L. Foley, B. Kenney, S. Ricke, and R. Nayak. 2016. Molecular characterization of Salmonella enterica serovars isolated from a turkey production facility in the absence of selective antimicrobial pressure. Foodborne Pathog. Dis. 13:80-87.

Schönenbrücher, V., E.T. Mallinson, and M. Bülte. 2008. A comparison of standard cultural methods for the detection of foodborne Salmonella species including three new chromogenic plating media. Int. J. Food Microbiol. 123:61-66.

Singer, R.S., A.E. Mayer, T.E. Hanson, and R.E. Issacson. 2009. Do microbial interations and cultivation media decrease the accuracy of Salmonella surveillance systems and outbreat investigations? J. Food. Prot. 72:707-713.

Temelli, S., A. Eyigor, and K.T. Carli. 2010. Salmonella serogroup detection in poultry meat samples by examining multiple colonies from selective plates of two standard culture methods. Foodborne Path. Dis. 7:1229-1234.

Wilson, K.M., B. McLendon, D.V. Bourassa, E. Montiel, J.L. Wilson, N.A. Cox, and R.J. Buhr. 2013. Impact of skip-a-day and every-day feeding programs on the colonization of Salmonella Typhimurium in broiler breeder pullets. Poult. Sci. 92(S1):P272. (Abstract)

Wilson, K.M., D.V. Bourassa, A.J. Davis, M.E. Freeman, and R.J. Buhr. 2016. The addition of charcoals to broiler diets did not alter the recovery of Salmonella Typhimurium during grow-out. Poult. Sci. 95:694-704.