Part of the data from this study was presented at the 101st Annual General Meeting of the American Society for Microbiology, Orlando, FL, USA, May 2001.
Dr C.M. Logue, Department of Veterinary and Microbiological Sciences, The Great Plains Institute of Food Safety, 130A Van Es Hall, North Dakota State University, Fargo, ND 58105, USA (e-mail: Catherine.Logue@ndsu.nodak.edu).
Aims: To determine the incidence of antimicrobial-resistant Salmonella spp. on processed poultry (turkey) at Midwestern poultry plants.
Methods and Results: Two participating plants were visited at monthly intervals for a period of 1 year. Surface swabs were obtained from carcasses at two selected points on the production line, pre- and post-chill. In addition, samples of the chill water from chill tanks were also examined. Isolation and detection of Salmonella spp. from carcass swabs and chill water was carried out using standard enrichment techniques. Immunomagnetic separation was used to enhance the recovery of the pathogen. Salmonella isolates recovered were identified, serotyped and their antimicrobial resistance profiles determined using the National Antimicrobial Resistance Monitoring System. Results from the study indicated that the overall incidence of Salmonella was approx. 16·7%, with a greater incidence of the pathogen observed on pre-chill than post-chill carcasses. Salmonella isolates recovered displayed resistance to an average of four different antimicrobials. Approximately 15 different serotypes of Salmonella spp. were recovered, with Salmonella serotype Agona, Salmonella serotype Hadar, Salmonella serotype Heidelberg and Salmonella serotype Senftenberg being the most common.
Conclusions: The incidence of Salmonella spp. was relatively low and isolates recovered showed significant degrees of antimicrobial resistance. Factors such as the processing plant examined, the season and farms that were presenting animals for processing influenced the incidence of the pathogen.
Significance and Impact of the Study: Differences were observed in the serotypes of Salmonella recovered and the types of antimicrobial resistance found at the two plants. The study suggests that the use of antimicrobials at the farm level influences the creation of an environment that promotes the selection of antimicrobial-resistant Salmonella spp. The incidence, isolation and detection of Salmonella spp. on processed poultry are discussed.
Salmonella is one of the most common pathogens implicated in foodborne illness in the USA. It has been estimated to cause approx. 1·4 million illnesses annually (Mead et al. 1999). Worldwide, Salmonella has been consistently implicated in human illness through the consumption of a wide variety of processed food and agricultural products contaminated with the pathogen. The organism has been isolated from a range of foods in almost every country in which it has been investigated. Typical foodstuffs from which Salmonella has been isolated include meat, swine, poultry, poultry products and dairy products (Manie et al. 1998; Cloak et al. 1999; Duffy et al. 1999; Gebreyes et al. 2000; Rajashekara et al. 2000). A considerable number of human salmonellosis outbreaks have been associated with meat consumption, in particular contaminated or undercooked meats, of which poultry (turkey and chicken) has been implicated as a vehicle (Bryan and Doyle 1995).
The theory that antimicrobial resistance selection can occur on the farm and be transmitted has been investigated by a number of researchers (Bren 2001; Gustafson and Bowen 1997). These studies have shown that antimicrobial resistance can be encoded chromosomally or on a transmissible plasmid, resulting in two potential methods of transfer. If resistance is chromosomally mediated, drug resistance is transferred to daughter progeny, while plasmid-associated resistance may be transferred to another species by means of R factor plasmid vectors (Alcaide and Garay 1984; Piddock 1996).
Salmonella contamination of animals such as poultry has been demonstrated to occur frequently at the farm level (Bryan and Doyle 1995; Pearson et al. 1996; Manie et al. 1998; Rajashekara et al. 2000). As a consequence, transfer of the pathogen among animals within the same herd/flock can occur when stress levels are elevated. This has been demonstrated during transport where faecal shedding resulted in an increased pathogen load (Whyte et al. 2001). Such events can result in animals being presented for processing with considerably greater contamination levels and serving as a significant source of cross contamination for other animals and the processing environment. Studies have demonstrated that chill water and the chilling process can be a significant source of pathogen contamination contributing to cross contamination between carcasses during chilling (Izat et al. 1989). As a consequence, a small number of contaminated carcasses may have an impact in spreading contamination. Similar studies have demonstrated that other procedures used during processing, such as handling, may also contribute to cross contamination between carcasses (Beery et al. 1988). In recent years, workers have suggested that processing conditions may play a role in promoting/ influencing the selection of pathogens during processing; however, this phenomenon is not fully understood (Lisle et al. 1998; Rowe et al. 1998; Norwood and Gilmour 2000).
The purpose of the current study was to determine the incidence of Salmonella spp. on turkey presented for processing at two participating Midwestern poultry plants, determine the prominent serotypes of Salmonella among isolates recovered and evaluate isolated strains for the levels and types of antimicrobial resistance.
Materials and methods
Two Midwestern poultry plants, designated A and B, were selected for the study. Both plants are full-time operations processing turkeys to a finished product, all carcasses being processed to pieces and primals. Plant A, the smaller of the two, had a processing rate of approx. 800 carcasses h−1. All carcasses were chilled in batch chillers to an internal carcass temperature of 4°C, taking approx. 4 h. Plant B, the larger of the two, had a processing rate of approx. 8000 carcasses h−1. Chilling was carried out in a continuous chill tank where carcasses were moved through the tank by means of an auger, taking 4 h to reach a core temperature of 4°C.
Both plants were randomly visited at monthly intervals for a period of 1 year. Sampling of carcasses was generally carried out during the morning production run to ensure adequate time to process samples on return to the laboratory. In most cases, carcasses being processed on a sampling day were from a single farm or producer; it was not possible, however, to sample the same carcass at both sampling sites because of logistics and time constraints. Carcass sampling was carried out at two points on the production line, at pre-chill following the last visual inspection, just prior to loading the carcass into the chill tank and post-chill swabs obtained immediately after removal of the carcasses from the chill tank as they were being rehung onto the line for further processing.
All carcass sampling was carried out using a non-invasive/non-destructive procedure originally devised by Lasta et al. (1992) for beef. Briefly, a sterile plastic stomacher bag (VWR Scientific, Minneapolis, MN, USA) was inverted over the operator's hand. A sterile, moistened swab was grasped in the covered hand and used to swab the carcass surface in a downward motion from front to back including the wings and legs, ensuring that as much surface area as possible was swabbed. When complete, the bag was reverted to its original position with the swab inside. A total of 50 swabs from pre-chill carcasses and 50 swabs from post-chill carcasses was obtained at each visit using this method. In addition, a 100-ml sample of the chill water was removed from the chill tank or collected at the tank outflow pipe into a sterile bottle. All swabs and samples collected were placed in designated iceboxes for transport to the laboratory.
On arrival at the laboratory, all samples were logged and recorded. A volume (100 ml) of Maximum Recovery Diluent (CM733; Oxoid) was added to each stomacher bag and the samples homogenized in a stomacher (IUL Masticator, Torrent de Lestadella, Barcelona, Spain) for 2 min. Aliquots (10·0 ml) of the resulting suspension were removed and added to 10·0 ml of double-strength buffered peptone water (218105; Difco, Becton and Dickinson, Sparks, MD, USA) and incubated at 37°C for 18–24 h. Following incubation, a 1·0-ml aliquot of the enrichment culture was removed and added to a sterile 1·5-ml centrifuge tube containing 20 µl of anti-Salmonella immunomagnetic beads (Dynal, Oslo, Norway). The suspension was subjected to immunomagnetic separation and washed according to the manufacturer's instructions (Dynal). After the final wash, the beads were resuspended in 100 µl of sterile phosphate-buffered saline (E404; Amresco, Solon, OH, USA) containing 0·1% Tween 20 (EM Science, Gibbstown, NJ, USA) and mixed using a vortex shaker (Vortex Genie; Scientific Industries, Bohemia, NY, USA). The resulting suspension was added to 9·9 ml of Rappaport–Vassiliadis broth (218581; Difco) and incubated at 42°C for 24 h. Following incubation, the enrichment culture was streaked out on Brilliant Green agar (Modified) (MBGA; 218801; Difco) and Mannitol, Lysine, Crystal Violet, Brilliant Green agar (MLCB; CM 783; Oxoid) for the isolation of Salmonella spp. All plates were incubated at 37°C for 18–24 h and examined for suspect Salmonella of typical colonial morphology: on MBGA, white-coloured colonies against a bright pink background and, on MLCB, purple-coloured colonies with a black centre. Suspect isolates from these media were streaked out on Tryptone Soy agar (236950; Difco) for purification and incubated at 37°C for 18–24 h. Following purification, all isolates were subjected to initial screening tests using the Gram reaction, Triple Sugar Iron agar (TSI; 226540; Difco), Urea broth (CM71, SR20; Oxoid) and Lysine Decarboxylase broth (CM308; Oxoid). All biochemical tests were incubated at 37°C for 18–24 h. Positive control cultures of Salmonella species (subgenus 2,1,3 no. 588 from the North Dakota State University (NDSU) Veterinary Diagnostic Laboratory) were used to ensure that all media were performing to acceptable standards.
All isolates of typical characteristics (Gram negative; TSI (Alkaline/Acid, H2S +/–, gas +/–); lysine + and Urease –) were identified using the Sensititre Microplate System (AP 80; Trek Diagnostics, Westlake, OH, USA) according to the manufacturer's instructions, with incubation of the inoculated plate at 37°C for 18–24 h. Isolates identified as Salmonella spp. were serotyped by the National Diagnostic Services Laboratory (Ames, IA, USA).
Salmonella isolates were also examined for antimicrobial resistance characteristics using the National Antimicrobial Resistance Monitoring System (NARMS) (CMV 5CNCD; 2000 panel; Trek Diagnostics) according to Food and Drug Administration (FDA) and National Committee for Clinical Laboratory Standards (NCCLS) recommendations. This panel uses a 96-well microtitre plate to test the antimicrobial resistance of suspect strains to 17 different antimicrobials commonly associated with animal health over a range of concentrations. Antimicrobials tested include amoxicillin/clavulanic acid (concentration range 0·5–32 µg l−1), ampicillin (2–32 µg ml−1), ceftiofur (0·5–16 µg ml−1), cephalothin (1–32 µg ml−1), chloramphenicol (4–32 µg ml−1), gentamicin (0·25–16 µg ml−1), cefoxitin (4–32 µg ml−1), tetracycline (4–32 µg ml−1), trimethoprim/sulfamethoxazole (0·12–4 µg ml−1), amikacin (4–32 µg ml−1), ceftriaxone (0·25–64 µg ml−1), ciprofloxacin (0·015–4 µg ml−1), nalidixic acid (4–256 µg ml−1), sulfamethoxazole (128–512 µg ml−1), streptomycin (32–256 µg ml−1), kanamycin (16–64 µg ml−1) and apramycin (2–32 µg ml−1). Inoculation of the NARMS panels was carried out according to the manufacturer's instructions (Trek Diagnostics). Positive control cultures (NCCLS-specified strains) of Escherichia coli (ATCC 25922 and ATCC 35218), Staphylococcus aureus (ATCC 29213), Enterococcus faecalis (ATCC 2912) and Pseudomonas aeruginosa (ATCC 27853) with known antimicrobial resistance profiles were used to ensure the efficacy of the drug panel. The control strains were used to test each new batch of plates for screening drug resistance. Isolates were defined as resistant according to the FDA- and NCCLS-recommended breakpoints (NCCLS 2001) for each specific antimicrobial. Breakpoints were defined as the minimum drug concentration above which growth of the test isolate should not occur. Strains displaying growth at drug concentrations above the breakpoint were considered resistant. The resistance levels (breakpoints) for the 17 antimicrobials examined were defined as follows: amoxicillin/clavulanic acid, ≥ 32/16 µg ml−1; ampicillin, ≥ 32 µg ml−1; ceftiofur, ≥ 8 µg ml−1; cephalothin, ≥ 32 µg ml−1; chloramphenicol, ≥ 32 µg ml−1; gentamicin, ≥ 16 µg ml−1; cefoxitin, ≥ 32 µg ml−1; tetracycline, ≥ 16 µg ml−1; trimethoprim/sulfamethoxazole, ≥ 4/76 µg ml−1; ciprofloxacin, ≥ 4 µg ml−1; nalidixic acid, ≥ 32 µg ml−1; sulfamethoxazole, ≥ 512 µg ml−1; streptomycin, ≥ 64 µg ml−1; kanamycin, ≥ 64 µg ml−1; apramycin, ≥ 32 µg ml−1; ceftriaxone, ≥ 64 µg ml−1 and amikacin, ≥ 64 µg ml−1. All isolates were identified as sensitive or resistant and the data recorded.
Table 1 shows the incidence of Salmonella spp. on poultry presented for processing at the two plants in the study. Of 2411 samples obtained, 402 (16·7%) were positive for Salmonella spp. There was little difference in the overall incidence of the pathogen detected at individual plants, with an overall isolation rate of 16·3% for plant A and 17·1% for plant B. One significant observation, however, was that a greater number of chill water samples and post-chill carcasses were positive at plant B suggesting an association between carcass and water contamination.
Table 1. The incidence of Salmonella spp. on poultry presented for processing at two Midwestern poultry plants
Plant A indicated that they did not use chlorinated water in their chilling regime and relied on the quality of water supplied from the city (estimated to have a chlorine concentration of approx. 5 mg l−1). In contrast, Plant B stated that they hyperchlorinated the chill water to a concentration of 20 mg l−1. In both cases, none of the claims could be adequately established.
Figures 1 and 2 show the incidence of Salmonella spp. on carcasses from both plants over the period of the study. There was a degree of seasonality in the incidence of the pathogen in that Salmonella was more frequently recovered from carcasses in the spring/summer months than in the autumn/winter season.
Figure 3 shows the number of Salmonella isolates recovered and the numbers of antimicrobials to which the isolates were resistant. In general, most Salmonella isolates tested were resistant to an average of four antimicrobials. It was noted, however, that the number of isolates displaying resistance to multiple antimicrobials was greater among isolates recovered from plant B. One isolate was found to be resistant to nine different antimicrobials. When this particular isolate was re-examined to confirm the level of antimicrobial resistance observed, the strain only displayed resistance to six antimicrobials. Almost all isolates recovered displayed significant resistance to one antimicrobial, streptomycin, with most isolates displaying resistances at a level of 32 µg ml−1 and higher (Fig. 4).
Table 2 shows the percentage resistance observed among isolates recovered against the antimicrobial tested for the two plants sampled. Overall, resistance to eight different antimicrobials was observed for Salmonella isolates recovered from plant A vs 13 from plant B. Of greater concern, however, was the range of antimicrobial concentrations to which Salmonella strains were resistant, this was particularly evident for streptomycin and tetracycline (Figs 4 and 5).
Table 2. Isolates of Salmonella recovered and their types of antimicrobial resistance
Plant A (n=195)
Plant B (n=207)
*Indicates the breakpoint of the antimicrobial (µg ml−1).
Overall, 15 different serotypes of Salmonella spp. were recovered from the two plants in the study and four serotypes were common among the isolates recovered, Salm. Agona, Salm. Hadar, Salm. Heidelberg and Salm. Senftenberg (Table 3). The serotypes recovered were found on both pre- and post-chill carcasses.
Table 3. Serotypes of Salmonella isolates recovered from both plants
NT, Multiple serotypes were present and could not be serotyped.
Data from this study indicate that Salmonella contamination of poultry is common, with overall isolation rates of 16·3 and 17·1% observed for the two plants over the period of the study. These are relatively similar to the numbers reported by some workers but significantly lower than others. Lammerding et al. (1988) reported a Salmonella isolation rate of 61% on retail poultry, while UK studies reported a 79% contamination rate (Roberts 1982) and US studies by Silliker (1982) and Engel (1987) reported contamination rates in the range of 30–50% on refrigerated and frozen broilers. In further studies, Izat et al. (1989, 1991) reported a pre-chill carcass contamination rate of 16–100% on poultry and a retail contamination rate of 17–50%. These numbers contrast slightly with work carried out by others worldwide who report a range of retail contamination rates (Al-Rajab et al. 1986; Bokanyi et al. 1990; Cloak et al. 1999; Duffy et al. 1999).
In previous studies of poultry contaminated with foodborne pathogens, a number of factors have been implicated as contributing to carcass contamination. These have included feed and water, the situation at rearing and hatcheries (Pearson et al. 1996), the farm and stress associated with transport (Whyte et al. 2001) and general contamination that occurs during processing, such as that associated with chilling and other handling procedures (Beery et al. 1988).
Feed has been implicated as a significant source of Salmonella spp. (McChesney et al. 1995) and may, therefore, be a contributor to contamination of birds and, subsequently, the flock. Studies have demonstrated that increased pathogen loads in faecal shedding may occur during times of stress, such as high population housing or transport; as a consequence, cross contamination between carcasses may result in increased contamination loads and greater risk of contamination of the final product (Izat et al. 1989).
Post-chill levels of Salmonella isolated in the current study were significantly lower than pre-chill, indicating that the chilling process can significantly reduce the levels of the pathogen. This may be due in part to hyperchlorination of chill water, a common practice in one of the plants studied. Some studies have indicated that chlorination can have a significant impact in reducing the numbers of contaminants while others suggest that chlorination may have the opposite effect by influencing the selection of a particular pathogen due to the emergence of resistant strains (Lisle et al. 1998; Rowe et al. 1998; Norwood and Gilmour 2000). Infection caused by such organisms may become more difficult to treat in the future if chlorine resistance is a factor in promoting the selection of particular bacterial species that carry additional antimicrobial resistance.
Serotypes of Salmonella recovered from poultry in this study (Salm. Heidelberg, Salm. Hadar and Salm. Agona) have previously been implicated as causes of human illness. In the latest CDC (2000) annual summary of Salmonella, strains similar to those isolated in the current study are frequently listed in the ‘top 20’ of human clinical isolates, non-human isolates and non-clinical, non-human isolates reported to the Centers for Disease Control and the National Veterinary Services Laboratory (NVSL).
Of greater importance was the incidence of antimicrobial resistance in strains of Salmonella recovered. Overall, the levels of resistance observed were generally below the breakpoints (or there was a low incidence of resistant strains) for antimicrobials such as amoxicillin/clavulanic acid, ceftiofur, cephalothin, chloramphenicol, cefoxitin, trimethoprim/sulfamethoxazole, ciprofloxacin, nalidixic acid, apramycin, ceftriaxone and amikacin. Significant resistance levels above the breakpoints were recorded for antimicrobials such as ampicillin, gentamicin, tetracycline, sulfamethoxazole, streptomycin and kanamycin. It was noted that approx. 50% of all isolates displayed tetracycline resistance and 35% were streptomycin resistant. Higher levels of resistance were noted by Threlfall et al. (1997) for both streptomycin and tetracycline (97% resistance) and by Manie et al. (1998) where 97% of isolates were tetracycline resistant and 86% streptomycin resistant. Such high levels may be attributed to the use of subtherapeutic doses of antimicrobials, such as streptomycin and tetracycline, in feeds (Facinelli et al. 1991). Similar studies have indicated that antibiotic-resistant strains can occur in farm animals as part of their normal flora or as pathogens and that this is, generally, a consequence of prescribing practices and the availability of antimicrobials in animals and human health (MAFF 1998). Hatha and Lakshmanaperumalasamy (1995) have suggested that isolates displaying resistance to two or more antimicrobials tend to originate from high resistant sources such as commercial poultry farms where antimicrobials are often used. Antimicrobials in feed are also considered responsible for the emergence of drug-resistant bacteria (O'Brien et al. 1982; Cohen and Tauxe 1986). The current study isolated Salmonella spp. resistant to streptomycin and chloramphenicol, neither of which is approved for feed use (McChesney et al. 1995; Muirhead 1999). However, antimicrobials such as tetracycline, virginiamycin, lincomycin, tylosin, bambermycin, penicillin, sulphonamides and bacitracin are commonly used in poultry and other food animal feeds as additives to maintain general health and prevent illness as a consequence of bacterial infection (Muirhead 1999). The results of this study suggest that feed may be one factor influencing the creation of antimicrobial-resistant Salmonella in turkey but that its incidence may be influenced by the particular farm practice. Further work, however, is required to determine the influence of other factors, such as genetic transfer and additional contamination sources, that may contribute to antimicrobial resistance and its persistence in foodstuffs and meat intended for human consumption. Studies are ongoing to evaluate the role of virulence factors and other characteristics associated with Salmonella spp. that may contribute to antimicrobial resistance and the potential risk associated with human illness.
The authors gratefully acknowledge the financial support of the NDSU Grant in Aid Program and ND EPSCoR and the co-operation provided by the management and staff of plants A and B. The authors also acknowledge the assistance of the many undergraduates who helped with this project, Melissa Casteel, Ryan Hinnenkamp, Murray Leraas, Christine Oliver, John Passman, Tiffany Priebe, Joey Rexine, Mark Sundrud and Ann Littlefield.