|Home | About | Journals | Submit | Contact Us | Français|
Twenty-eight research dogs were enrolled to determine the prevalence of salmonellae shedding after consumption of 1 Salmonella-contaminated commercial raw food diet meal. Sixteen dogs were exposed to Salmonella-contaminated commercial raw food diets and 12 to Salmonella-free commercial raw food diets. Seven of the exposed dogs shed salmonellae 1–7 days after consumption of Salmonella-contaminated raw food diets. None of the dogs fed Salmonella-free diets shed salmonellae. No clinical signs were observed in either group. Five of the 7 dogs shed the same serotypes as those recovered from food samples used for feeding. Results showed the same serotypes and antimicrobial resistance pattern in 2 of the 7 shedders. Dogs fed Salmonella-contaminated raw food diets can shed salmonellae and may, therefore, be a source of environmental contamination potentially leading to human or animal illness.
Risque d’excrétion de salmonelles chez des chiens soumis à un régime alimentaire commercial composé d’aliments crus contaminés aux salmonelles. Vingt-huit chiens de recherche ont été utilisés pour déterminer la prévalence de l’excrétion de salmonelles après la consommation d’une ration de nourriture commerciale crue contaminée aux salmonelles. Seize chiens ont reçu une ration contaminée aux salmonelles et 12 une ration exempte de salmonelles. Sept des chiens ayant consommé la ration contaminée ont excrété des salmonelles de 1 à 7 jours plus tard alors qu’aucun des chiens ayant reçu la ration non contaminée n’en a excrétées. Aucun signe clinique n’a été observé tant dans un groupe que dans l’autre. Cinq des 7 chiens ont excrété les mêmes sérotypes que ceux retrouvés dans les aliments contaminés. Les résultats ont montré les mêmes sérotypes et les mêmes profils de résistance antimicrobienne chez 2 des 7 excréteurs. Les chiens nourris d’aliments crus contaminés aux salmonelles peuvent excréter des salmonelles et peuvent ainsi constituer une source de contamination environnementale conduisant potentiellement à des maladies humaines ou animales.
(Traduit par Docteur André Blouin)
Veterinarians and public health officials have recognized shedding salmonellae by dogs as a possible source of Salmonella infection for dog owners and their communities (1–3). The prevalence of salmonellae in dogs in the community is not well established, as dogs can be asymptomatic carriers, capable of shedding the organism without exhibiting signs of illness. The prevalence of Salmonella isolation from clinically healthy and hospitalized dogs has been estimated to be between 1% and 35% (4,5). Clinical salmonellosis is rare in dogs, but clinical signs include fever (40°C–41.1°C), anorexia, diarrhea, bloody diarrhea, abdominal pain, and abortion (5,6). Asymptomatic dogs can shed salmonellae for 6 wk or more, continuously during the 1st week, and then intermittently (5,6).
There are various potential sources of salmonellae for dogs, including unprocessed or raw dog food or pet treats of animal origin. Raw meat obtained from rendering plants and used to feed dogs can be contaminated with salmonellae and has been associated with canine salmonellosis (7–9). Raw meat is the main ingredient in “raw food diets,” a relatively new and increasingly popular type of dietary trend in dogs. In addition to raw meat, these diets contain vegetables, grains, and fruit; they may also include ground bones and are served raw to dogs as their main meal (10). Raw food diets can be purchased frozen at pet stores and some veterinary clinics, or they can be homemade by following recipes found on the Internet (11,12) or in books (13–17).
Regardless of the possible benefits of raw food diets claimed by advocates, dogs that consume them are at some risk of Salmonella infection. Raw meat used for the production of these diets can originate from several sources, including human-food grade processing plants, rendering plants, and products no longer deemed suitable for human consumption (18). As these diets do not undergo any type of heat processing or sterilization, existing bacteria and parasites can be present at the time of consumption.
Joffe and Schlesinger (19) enrolled 20 client-owned dogs at a veterinary clinic to investigate the risk of Salmonella infections in dogs consuming raw chicken diets. Ten dogs with previous exposure to raw food were fed homemade raw food diets and 10 dogs were fed commercial dry dog food. Eight (80%) of the 10 homemade diets tested positive for salmonellae. Three of the 10 dogs (30%) fed these diets had salmonellae organisms isolated from samples of their feces. None of the dogs fed commercial diets shed Salmonella in their feces.
The purpose of this study was to further investigate the health risks that commercial raw food diets pose to the dogs consuming these products and to their owners. The study was linked to a retail survey (20) of raw food diets that was conducted in 3 Canadian cities to assess the prevalence of salmonellae in frozen, commercially available, raw food diets. The objectives of the study were to determine if dogs shed salmonellae after consuming a Salmonella-contaminated commercial raw food diet, and if so, to determine how long it would take them to stop shedding after a single exposure. A further objective was to determine whether or not the serotype and resistance phenotype found in the contaminated raw food diet matched the serotype and resistance phenotype of the salmonellae isolated from the dogs’ feces.
Commercially available raw food diets were sampled in Mississauga and Guelph, Ontario, and in Calgary, Alberta, as part of a study assessing the prevalence of Salmonella in these products (20). Frozen samples were forwarded to the Canadian Research Institute for Food Safety (CRIFS) laboratory at the University of Guelph for testing. One half of each raw food diet sample was tested for salmonellae and the other half of the sample was kept frozen. Selection of the samples used in this study was based on culture results.
Three methods were used in parallel format for the isolation of salmonellae from commercially available raw food diets (20). The 1st method followed the FDA Bacteriological Analytical Manual (21). The 2nd and 3rd methods were derived specifically for the study in consultation with Salmonella experts at the Ontario Veterinary College, University of Guelph and the Laboratory for Foodborne Zoonoses, Public Health Agency of Canada (LFZ-Guelph), Guelph. All Salmonella isolates cultured from the raw food diet samples were submitted to the LFZ-Guelph for serotyping and for antimicrobial susceptibility testing. Antimicrobial susceptibility testing was conducted by using broth microdilution (Sensititre®; Trek Diagnostics, Westlake, Ohio, USA) and a panel of 16 antimicrobials (amikacin, amoxicillin/clavulanic acid, ampicillin, cefoxitin, ceftriaxone, ceftiofur, cephalothin, chloramphenicol, ciprofloxacin, gentamicin, kanamycin, nalidixic acid, streptomycin, sulfamethoxazole, tetracycline, and trimethoprim/sulphamethoxazole). The resistance breakpoints were those used by the Canadian Integrated Program for Antimicrobial Resistance Surveillance, which are derived from Clinical Laboratory Standards Institute (CLSI, formerly [NCCLS]) breakpoints.
A sample size was calculated by using a statistical package (MiniTab Release B; Mini Tab, State College, Pennsylvania, USA), based on the highest available prevalence estimate for salmonellae in dogs, 35%. In order to detect a difference in Salmonella status between dogs fed contaminated raw food diets and those fed noncontaminated raw food, the total sample size required, with Type I and Type II error rates set at 5% and 20%, respectively, was 46 dogs, 23 exposed and 23 unexposed.
Purpose bred beagle dogs were acquired from a reputable commercial source. Five dogs, 1 more than needed, were randomly selected prior to each individual trial period. Only dogs that had stool samples that tested negative for salmonellae for 3 consecutive days were eligible for enrollment. Four dogs, if all 5 were negative, were randomly selected for inclusion. Information on age, sex, spay/neuter status, and prior antimicrobial usage was collected for each dog.
An Animal Utilization Protocol was reviewed and approved by the Animal Care Committee, University of Guelph. This committee follows the guidelines established by the Canadian Council on Animal Care.
The 4 dogs for each trial run were housed in an isolation unit in individual kennels. Groups consisted of either 3 exposed and 1 unexposed, or 2 exposed and 2 unexposed dogs.
On the 1st 2 d, dogs were fed their normal commercial dog food to allow an acclimation period to the isolation unit. On the 3rd d, the exposed group was fed Salmonella-positive raw food diets, while the unexposed group was fed Salmonella-free raw food diets. All dogs were fed Salmonella-free raw food diets on the 4th and 5th d. From the 6th d until the end of their observation period, all 4 dogs were fed their normal commercial dog food. All commercial dog food products fed during the feeding trial were tested for Salmonella contamination throughout the study.
Dogs remained in their kennels at the isolation unit until 8 consecutive Salmonella-negative fecal samples had been obtained, at which point they were released from the study and returned to the Central Animal Facility. Each dog was socialized individually for 20 min in the morning and 10 min in the afternoon while in the isolation unit. Unexposed dogs were socialized prior to exposed dogs to minimize environmental contamination. A plastic guard was placed on the bottom kennels to prevent nose-to-nose contact between dogs in kennels and the dog in socialization. During socialization, dogs were taken out of their kennels to play, fecal samples were collected, kennels were cleaned, and the health status of each dog was observed and recorded on a health-monitoring form. Information on dog attitude, amount of food intake, presence of vomit, and appearance of feces was noted. After each individual animal had been socialized, floors were cleaned with a diluted mixture of a quaternary detergent/disinfectant (Ascend®; Huntington Professional Products; St. Paul, Minnesota, USA) and left to dry for 10 min. Animal handlers wore full protective clothing. Protective gowns and gloves were changed and boots disinfected between dogs.
From each fecal sample, 10 g of fresh feces was transferred into a sterile stomacher bag filled with 90 mL of buffered peptone water (BPW). Contents were emulsified and incubated at 37°C for 18 to 24 h. A portion (0.1mL) of the BPW was transferred into 10 mL of Rappaport Vassiliadis (RV) broth and incubated at 42°C for 24 h. Following incubation, a loopful of RV broth was plated onto 3 selective media: xylose lysine tergitol 4 (XLT4) agar, brilliant green sulfa (BGS) agar, and bismuth sulfite (BS) agar. The selective media plates were then incubated at 37°C for 18 to 24 h. After incubation, the plates were examined for colonies suggestive of salmonellae. If there were any colonies present, at least 2 typical colonies, as well as 2 atypical colonies, from each selected plate were plated onto MacConkey agar and incubated at 37°C for 18 to 24 h.
From each MacConkey plate, presumptive salmonellae were plated onto nutrient agar plates (NAP) and incubated at 37°C for 18 to 24 h. From these, biochemical testing was conducted by using triple sugar iron (TSI) agar and urea slants. Isolates were incubated at 37°C for 18 to 24 h. If TSI slants were positive for salmonellae and urea slants were negative, indole testing followed. If the indole test was negative, then agglutination was performed by using Salmonella O Antiserum Poly A-I and Vi. If results from the agglutination test were positive, a loopful of growth was inoculated onto tripticasesoy agar (TSA) slants.
All Salmonella isolates were forwarded to the LFZ-Guelph for serotyping and antimicrobial susceptibility testing, as described above for the raw food samples.
Comparison of the risk of Salmonella isolation from the feces of dogs in the exposed and unexposed groups was conducted by using a 2-tailed Fisher’s exact test (22) run on 2 statistical programs (DISTRIB; William Sears, Department of Population Medicine, Ontario Veterinary College, University of Guelph, and SAS V8.2; SAS Institute, Cary, North Carolina, USA). A correction factor of 0.5 was added to all cells in the 2 × 2 table in order to estimate the relative risk when any of the cells were zero (23).
None of the 50 dogs that were screened tested positive for salmonellae prior to enrolment in the feeding trial. Because of welfare concerns associated with having dogs housed in isolation for extended periods of time, and because the objective of the study was to detect a significant difference in Salmonella shedding status, not to produce a stable estimate of that difference, Salmonella prevalence results were assessed periodically, with a decision to stop the trial early based on achievement of a significant difference of P < 0.05. The feeding trial was terminated early with 28 dogs enrolled in the study from February 18, 2004, to July 8, 2004. This included 16 exposed and 12 nonexposed dogs in 7 groups of 4.
The control group averaged 21 mo of age (range: 9–31 mo) and consisted of 8 males and 4 females. The exposed group averaged 20 mo (range: 9–33 mo) and consisted of 9 males and 7 females.
Of the 16 dogs exposed to Salmonella-contaminated commercial raw food diets, 7 (44% with 95% CI: 21%–69%) shed salmonellae in their feces. None (0/12) of the control dogs shed salmonellae. The exposed group’s shedding rate (44%) was significantly different (RR ≥ 11.4; P = 0.01) from the zero shedding rate found in the unexposed group. The dogs in the exposed group started to shed salmonellae 1 to 7 d after consuming Salmonella-positive raw food diets. Total days shedding ranged from 1 to 11 d, with a mean of 3.9 d. Five of 7 dogs had more than 1 d of shedding and days between shedding ranged from 1 to 9 d (Table 1). None of the 7 dogs that shed salmonellae had any prior antimicrobial treatment. None of the exposed dogs developed diarrhea or showed any clinical signs.
Table 1 shows the serotypes and antimicrobial susceptibility patterns of salmonellae recovered from exposed dogs compared with the Salmonella isolates obtained from the raw food diets they consumed. Five of 7 dogs shed the same serotypes as found in their food. Of these 5, 2 shed only the serotype found in its food, 1 had 2 serotypes in its food but only 1 in its feces, 1 had 1 serotype in its food but shed 3 serotypes, and 1 had 2 serotypes in its food but shed 3 serotypes. Two dogs shed a different serotype from that identified in its food. One of the dogs was fed a raw food diet containing S. Thompson, and then shed S. Typhimurium, while another dog was fed S. Typhimurium, and then shed S. Thompson.
In most cases, where serotypes matched between food and feces, the antimicrobial susceptibility pattern also matched. In dog Devon, the food isolate (S. Heidelberg) was resistant to ampicillin alone, but the fecal isolate was resistant to ampicillin and cephalothin. Dog O’Henry was exposed to food containing S. Heidelberg with 3 patterns and shed S. Heidelberg with the same patterns plus 2 other patterns, and dog Dundas exposed to a susceptible strain of S. Agona shed that strain plus a susceptible strain of S. Tennessee and an S. Heidelberg strain resistant to 11 antimicrobials.
Our results showed that Salmonella infections tend to be non-clinical and that shedding can occur for almost 2 wk following a single meal in dogs not previously fed raw food diets. Although no dogs became ill, they were easily colonized, if only transiently, and therefore at risk of developing salmonellosis.
People who feed dogs commercial raw food diets should be aware that the products are often contaminated with salmonellae (24) and that dogs consuming them have a high probability of shedding the organism. The shedding rate (44%) in this study was higher than that (30%) observed by Joffe and Schlesinger (19) in dogs fed homemade raw food diets. This difference may have been due to the differences in length of time that dogs were exposed to raw food diets in the 2 studies. The difference may also have resulted if the total load of salmonellae present in the commercial diets was greater than that in the homemade diets. Determining this would have required a quantitative measure of salmonellae organisms in the 2 types of diet, which was not done in either study. However, one might expect the microbial load to be lower in the commercial diets, since they are frozen, a process that should kill some of the salmonellae (25). Finally, the difference may simply have been a sample size artefact. Sample sizes in both trials were relatively small, and in this study, the 95% confidence interval for the shedding rate (21%–69%) included the 30% found by Joffe and Schlesinger.
In a previous study, the prevalence of Salmonella-contaminated raw food diets in Calgary was 37% (20). Dogs in Calgary that were fed these diets during the same time, therefore, could have been exposed to salmonellae, on average, once every 3 feedings. Our results indicated that dogs can shed salmonellae shortly after ingestion of a single contaminated meal, with shedding lasting up to 1.5 wk. The results of Joffe and Schlesinger’s trial indicated that this shedding can be prolonged in dogs fed the raw food diets over a longer period of time.
Clinical salmonellosis in dogs as a result of raw food diet consumption has not been reported in the literature; however, Siver et al (24) recently reported 2 cases of septicemic salmonellosis in cats fed raw food diets. Salmonellae were isolated from several organs in both cats, and the isolates obtained from the cats were identical to those isolated from raw beef that had been incorporated into the diet of 1 of the cats. Salmonella Newport originating from the meat used in their meals was found to be the cause of death for both cats.
Salmonella Heidelberg was the most common serotype found in positive diets, most of which contained chicken as the main ingredient (Table 1). Salmonella Heidelberg was also the most common serotype isolated from the feces of dogs in this study. Results of a retail surveillance conducted in Canada in 2003, showed S. Heidelberg to be the most prevalent serovar (73%) in 16% of chickens purchased from retail stores and markets, followed by S. Kentucky (11%) (26). In Canada, S. Heidelberg was the most common cause of human salmonellosis, accounting for 26% of all Salmonella isolates obtained from human cases through enhanced passive surveillance in 2003. Salmonella Typhimurium was the second most common serotype (25%) associated with human salmonellosis, followed by S. Enteritidis (15%) (26). Public health officials should, therefore, ask about exposure to raw food diets and dogs that consume these products when interviewing possible human cases of salmonellosis.
Several isolates recovered from both the food samples and the canine fecal samples were found to have different serotypes and to carry different antimicrobial susceptibility patterns as compared with the isolates from the corresponding raw food samples. Two dogs were found to carry serotypes not recovered from the raw food diet they received during the feeding trial. Both of these dogs were in the same feeding trial group of 4 animals and received their contaminated portion of raw food diet on the same morning at the same time. A possible explanation for the apparent “cross-over” in serotypes that the 2 dogs shed in their feces was that documentation of which diet was fed to which dog was mixed up. Review of the data sheets yielded no evidence of a recording error and all of the remaining fecal shedding results were consistent with the documented dietary exposures.
Additional serotypes could have been present in the raw food diets and not recovered despite the use of 3 methods and 3 selective media in parallel. Some dogs in the research colony from which our trial dogs were sourced were apparently given pig ear dog treats sometime prior to the feeding trial. Salmonellae have been recovered from pig ear dog treats in the past (27,28); however, recent retail sampling suggests the prevalence of Salmonella contamination of pig ear treats sold in Ontario is low (20). Three days of negative tests used for screening dogs may not have been sufficient to completely rule out pretrial Salmonella carriage. However, if prior exposure to pig ears was the true source of Salmonella serotypes for dogs in our trial, we would have expected some of the control dogs also to have shed during the trial, which did not occur. The regular, heat processed commercial food that was given to the trial dogs as their normal diet was not a likely source, because all samples tested during the feeding trial were negative for salmonellae.
Some Salmonella serotypes were more likely to be shed than others. The majority of dogs that were fed raw food diets contaminated with S. Heidelberg and S. Infantis had Salmonella isolates recovered from their fecal samples. However, dogs fed raw food diets contaminated with S. Hadar and S. Meleagridis, and some fed S. Infantis, S. Thompson, and S. Heidelberg, did not shed salmonellae during the feeding trial period. It is possible that there really is a difference in the likelihood of shedding related to serotype, which could be related to differences in factors like inoculum size, survival in the frozen raw food diet matrix, or ability to colonize the canine gut. It is also possible that some of the Salmonella-negative dogs were false negatives, either because the sampling frequency (once a day) was inadequate or because of methodological issues.
Antimicrobial resistance was similar in isolates recovered from the raw food and the fecal isolates of the Salmonella-positive dogs. One strain recovered from dog O’Henry was resistant to ampicillin, cephalothin, cefoxitin, and ceftiofur, which may indicate that this isolate was an extended spectrum β-lactamase producer. The source of the extra resistance is unclear; however, transfer of genes from other bacteria, such as Escherichia coli is a possibility. Testing of generic E. coli was also performed in both food and canine fecal samples (data not shown). Generic E. coli isolated from the food fed to O’Henry included resistance to all 3 additional antimicrobials (data not shown). Resistance genes can be transferred by viruses or taken up by bacteria after another bacterium dies and releases its contents into the environment (29). Although these mechanisms are possible, transfer of resistance through conjugation is more likely (30). Further investigation of the raw food samples and canine fecal samples from this study will include an examination of the antimicrobial susceptibility pattern of E. coli isolates. If there is concordance between E. coli and Salmonella isolates, molecular techniques will be used to investigate the possibility of exchange of antimicrobial resistance genes between E. coli and Salmonella in the dogs’ intestinal tracts.
To date, this is the 1st study conducted to investigate the prevalence and antimicrobial resistance patterns of salmonellae in dogs after consumption of Salmonella-contaminated commercial raw food diets and the 1st study investigating the shedding in dogs previously unexposed to raw food diets. It complements and expands on the earlier study of homemade raw food diets by Joffe and Schlesinger (19). As commercial raw food diets are relatively new, one of the main unknowns is the number of dogs currently consuming these products. Without these data, it is difficult to estimate the risk to human and canine populations. However, it is clear that dogs that eat Salmonella-contaminated raw food diets can shed the bacteria, increasing the risk of salmonellosis for dogs and dog owners. The diets are a potential source of salmonellae resistant, in some cases, to multiple antimicrobials, and to antimicrobials of importance in human medicine, including third generation cephalosporins. Furthermore, there is evidence, albeit unproven, of acquisition of new resistance by salmonellae in the canine gut.
As the main objective of this study was to determine whether or not dogs consuming Salmonella-contaminated raw food diets would shed salmonellae, dogs enrolled in the study were not followed after their 8th-consecutive Salmonella-negative fecal sample. It is unknown how long Salmonella-positive dogs can go between episodes of shedding. Anecdotally, 5 d of consecutive negative tests has been thought sufficient to assume a negative status. However, 1 dog at the end of the study was kept an extra day in isolation because she could not be relocated on the scheduled day. A Salmonella isolate was recovered on that day, 9 d after the previous positive fecal sample. Future studies should include a longer surveillance period to verify Salmonella-negative status.
Since this study was an experimental trial with laboratory beagles, results may not be completely indicative of what would be experienced with owned dogs of various breeds. Determining that the raw food diet was the main source of salmonellae shedding in owned dogs would be more complicated, as they could be exposed to several other possible sources of salmonellae, including other animals; other food items, including treats; the environment; and their owners. There is also the need to investigate the effects of multiple exposures to raw food diets in dogs and to compare shedding in naïve dogs exposed to a single meal with those exposed to multiple meals to determine long-term effects.
To minimize the risk of canine and human salmonellosis, regulations governing the manufacture and sale of commercial raw food diets should be established and enforced. Better labels should be placed on packages containing commercial raw food diets warning dog owners of the probability of the products being contaminated with harmful bacteria. Organizations that take dogs to seniors’ homes, long-term care hospitals, or other hospital areas for therapeutic visits should not feed these diets, because of the risks posed by bacterial pathogens and antimicrobial resistant bacteria. Dog owners, especially those at higher risk of developing acute salmonellosis, should be warned about the risks of handling raw food diets.
We thank Nicol Janecko, Heather Lim, and all the laboratory staff for the help they provided with the microbiological analysis of the samples and socialization of the dogs. We also thank Dr. Anne Muckle, Betty Wilkie, Linda Cole, Andrea Desruisseau, Ketna Mistry and Abigail Crocker at the Laboratory for Foodborne Zoonoses, Public Health Agency of Canada for the serotyping and antimicrobial susceptibility testing conducted on the Salmonella isolates. CVJ
Foodborne, Waterborne and Zoonotic Infections Division (Finley, Aramini) and Laboratory for Foodborne Zoonoses (Reid-Smith), Public Health Agency of Canada, 160 Research Lane Unit 206, Guelph, Ontario N1G 5B2.
Reprints will not be available from the authors.
Dr. Ribble’s current address is Faculty of Veterinary Medicine, University of Calgary, G380, 3330 Hospital Drive, NW, Calgary, Alberta T2N 4N1.
Funding for this research was obtained from the Ontario Veterinary Pet Trust, Public Health Agency of Canada.