Occurrence of antimicrobial resistant bacteria in healthy dogs and cats presented to private veterinary hospitals in southern Ontario: A preliminary study

Abstract

The prevalence and patterns of antimicrobial susceptibility of fecal Escherichia coli, Salmonella spp., extended β-lactamase producing E. coli (ESBL-E. coli), methicillin-resistant Staphylococcus aureus (MRSA), and methicillin-resistant Staphylococcus pseudintermedius (MRSP) were determined for healthy dogs (n = 188) and cats (n = 39) from veterinary hospitals in southern Ontario that had not had recent exposure to antimicrobials. The prevalence of antimicrobial resistance in E. coli was as follows: streptomycin (dogs — 17%, cats — 2%), ampicillin (dogs — 13%, cats — 4%), cephalothin (dogs — 13%, cats — < 1%), and tetracycline (dogs — 11%, cats — 2%). Eleven percent of dogs and 15% of cats had isolates that were resistant to at least 2 antimicrobials. Cephamycinase (CMY)-2 producing E. coli was cultured from 2 dogs. No Salmonella spp., ESBL-E. coli, MRSA, or MRSP isolates were recovered. The observed prevalence of resistance in commensal E. coli from this population was lower than that previously reported in companion animals, but a small percentage of dogs may be a reservoir for CMY-2 E. coli.

Introduction

In recent years, interest in antimicrobial resistance in companion animals has increased. This is due, in part, to an increasing number of reports of companion animals infected or colonized with clinically and epidemiologically important multiple-drug resistant organisms, such as methicillin-resistant Staphylococcus aureus (MRSA) (1,2). Most companion animal antimicrobial resistance studies have been derived from diagnostic laboratory submissions (3), zoonotic transmission case reports (4), or nonrepresentative populations of companion animals (5). Data derived from these sources are useful, but they provide, at best, only partial insight into the overall epidemiology of antimicrobial resistance in companion animals.

A comprehensive epidemiological study of antimicrobial resistance in animals should involve the investigation of 3 areas: veterinary pathogens, zoonotic pathogens, and indicator bacteria (6). The study of zoonotic pathogens, such as Salmonella spp. and MRSA, in companion animals provides information of special relevance to public health. Companion animals have been a reservoir of Salmonella spp. for humans through direct contact with pets, contact with feces from pets, preparation of raw meat and bones for pet consumption, and the handling of commercial pet treats (7,8). Similarly, pets have been colonized and infected with MRSA and have acted as reservoirs of infection for their human contacts (9).

Indicator commensal bacteria, such as E. coli, are useful for the study of antimicrobial resistance, because 1) they constitute part of the normal enteric microflora; 2) some of them may be opportunistic pathogens; 3) they can act as indicators of antimicrobial selection pressure; and 4) they may represent a reservoir of antimicrobial resistance genes for pathogenic or zoonotic bacteria (10). Examination of antimicrobial susceptibility patterns in commensal E. coli allows for comparison of resistance across different animal systems and antimicrobial-use practices (6). Although there has been a recent, modest increase in publications concerning antimicrobial resistance in companion animals and companion animal zoonotic bacteria, investigations of antimicrobial resistance have been centered mainly on human and food animal populations. Since companion animals share the home environment with humans, consume animal source food products, and occasionally require treatment with antimicrobials, understanding the baseline community prevalence of resistance in commensal E. coli and zoonotic pathogens, such as Salmonella spp. and MRSA, is an important component to the global understanding of the epidemiology of antimicrobial resistance in animals and humans. The objectives of this study were to establish the prevalence and antimicrobial susceptibility patterns of commensal E. coli, extended spectrum β-lactamase (ESBL) producing E. coli, Salmonella spp., MRSA, and methicillin-resistant Staphylococcus pseudintermedius (MRSP) by studying healthy dogs and cats in southern Ontario.

Materials and methods

Limited data from the literature (11) and information provided by experienced veterinary practitioners suggested that the prevalence of resistance in E. coli to most antimicrobials would be 5% to 10%. The required sample size, given a confidence level of 95%, to detect a prevalence of resistance between 5% and 10% was estimated as being between 139 and 292 animals, depending on the specific antimicrobial (12). In order to achieve the target sample size of 200 dogs and 200 cats, approximately 20 companion and mixed animal practices, providing 10 dogs and 10 cats each, were required for the study. Approval from the University of Guelph Animal Care Committee was obtained prior to the initiation of the study.

Clinic recruitment

Clinic recruitment started in May 2002. The study area comprised 2 regions in southern Ontario, an area within 100 km of Guelph and an area within 100 km of Belleville. All companion animal practices, as defined by the College of Veterinarians of Ontario (CVO), in the study areas were identified (n = 375); a subset (n = 66) was selected for the initial recruitment mailing, using stratified, systematic, random sampling with the Guelph and Belleville areas as the strata. All mixed animal veterinary practices (n = 65) in the study areas were also included in the initial mailing. For the purposes of this study, a mixed animal practice was defined as a facility licensed by the CVO as a companion animal office or a companion animal hospital with additional licenses for a food animal or equine hospital, or mobile. A letter describing the study objectives and the role of the practices in the study was mailed to the selected practices. The practices willing to consider participating were asked to respond by mail, facsimile, or telephone with a completed practice-demographic questionnaire. The questionnaire included questions on the size of the practice, the number of canine and feline patients, and the proportion of patients with urban, rural, and farm home environments. Nonresponders were contacted a 2nd time by telephone. Based on results of the demographic questionnaire, practices were purposively selected to provide a cross-section of companion (n = 11) and mixed animal practices (n = 7) in rural and urban regions. Participating clinics were visited on 1 to 3 occasions.

Animal enrollment

Dogs were enrolled, with owner consent, from among those presented on the day of the investigator’s attendance at the participating clinic for wellness appointments (routine physical examination, vaccinations), lameness evaluations, conditions that did not involve clinical signs of systemic illness or infection at the sampling sites (perineum, rectum), and elective surgical/ technical procedures. To limit stress and facilitate sample collection, cats were enrolled, with owner consent, from among those that were presented for elective surgical and technical procedures requiring sedation or general anaesthesia. In addition, only pets with current medical records at the participating clinics were included. As we wished to describe the baseline occurrence of antimicrobial resistance in animals in the absence of recent antimicrobial exposure or recent gastrointestinal perturbation, an antimicrobial prescription history was obtained from the medical record. Animals that had received any antimicrobials (oral, otic, ophthalmic, parenteral, topical, or rectal) in the preceding 6 wk, or had had episodes of vomiting or diarrhea within the preceding 7 d were excluded. Additionally, animal demographic and risk factor data (diet, other pets in the household, contact with livestock) were collected through a questionnaire answered by the owner or guardian of each animal. Data collection was completed by December 2002.

Sample collection

A rectal swab and a perineal swab were collected from each enrolled animal. Swabs were collected by using an aerobic bacterial culturette (BBL CultureSwab; Becton, Dickinson and Company, Franklin Lakes, New Jersey, USA). Swabs were stored in a portable cooler containing an ice pack and then refrigerated at 4°C until being processed within 72 h of collection.

Bacteriologic studies

The fecal swabs were cultured for E. coli, Salmonella spp., and ESBL-E. coli; the perineal swabs were cultured for MRSA and MRSI. Escherichia coli was isolated by streaking the rectal swabs onto MacConkey agar (Becton, Dickinson) and incubating at 37°C for a further 18–24 h. Five isolates per sample were selected and plated onto MacConkey agar and incubated at 37°C for 18–24 h. A colony from each plate was streaked onto Luria Bertiani (LB) agar (Becton, Dickinson) and incubated at 37°C for 18–24 h. To isolate ESBL-E. coli, rectal swabs were streaked onto MacConkey agar containing 2 μg/mL of cefotaxime and incubated at 37°C for 18–24 h. If there was no growth present after 24 h, the plates were incubated for an additional 24 h at 37°C. Up to 5 isolates were plated onto MacConkey agar for 18–24 h, followed by LB agar for 18–24 h. The identification of E. coli and ESBL-E. coli was verified by colony morphology on MacConkey agar and by biochemical tests, including the production of indole and no utilization of citrate as the sole carbon source. Further identification of ESBL-E. coli was made by using a standardized identification system (API 20 E; bioMérieux, Marcy l’Étoile, France).

Phenotypic confirmation of presumptive ESBL-E. coli was performed according to the guidelines of the Clinical Laboratory Standards Institute [CLSI, formerly National Committee for Clinical Laboratory Standards (NCCLS)], using disc diffusion methods (13,14) with the following antimicrobials: ceftazidime, ceftazidime-clavulanic acid, cefotaxime, and cefotaxime- clavulanic acid. In addition, a phenotypic confirmation test was performed on any isolates of commensal E. coli that had intermediate susceptibility or resistance to amoxicillin-clavulanic acid or cefotaxime (see susceptibility testing) on disc diffusion. Escherichia coli ATCC 25922 and K. pneumoniae ATCC 700603 strains were used as controls.

Salmonella spp. isolation was performed by placing the rectal swabs into 3.6 mL of buffered peptone water (BPW) (Becton, Dickinson) and incubating it at 37°C for 18–24 h. Then 0.1 mL of the incubated BPW was deposited on the periphery of a modified semisolid Rappaport Vassiliadis (MSRV) (Oxoid Canada, Nepean, Ontario) plate with 20 μg/mL of novobiocin and incubated at 42°C for 24–72 h. In case selective outgrowth of putative Salmonella spp. bacteria occurred, a loopful was taken from the outer edge of the migration and plated onto MacConkey agar and incubated at 37°C for 18–24 h. Thereafter, 8 nonlactose fermenting colonies per sample were plated onto LB agar and incubated at 37°C for 18–24 h. Colonies were identified as Salmonella spp. by morphology on triple sugar iron (TSI) (Becton, Dickinson) slants, lack of urease production on urea slants, and slide agglutination with polyvalent A1 and Vi antisera (Becton, Dickinson).

To isolate MRSA and MRSP, the perineal swabs were streaked directly onto mannitol salt agar containing 2 μg/mL of oxacillin (Oxoid Canada). In addition, the swabs were enriched (broth of 7.5% NaCl, 2.5 g/L yeast extract, 10 g/L tryptone, and 10 g/L mannitol) for 24 h at 35°C, prior to plating. Plates were incubated at 35°C for 24–48 h. Up to 5 colonies were plated onto sheep blood agar and incubated at 35°C for 18–24 h. Oxacillin resistance of colonies of catalase positive, coagulase positive, gram-positive cocci was confirmed by the Kirby-Bauer disc diffusion technique with a 1-μg oxacillin disc (Becton, Dickinson), following CLSI guidelines (13,14). Identification to the species level was made by using a standardized identification system (API STAPH, BioMerieux).

Susceptibility testing of E. coli isolates

Five E. coli isolates per animal and all presumptive ESBL-E. coli isolates were tested, using the Kirby-Bauer disc diffusion technique, for susceptibility to the following antimicrobials (13): ampicillin, amoxicillin-clavulanic acid, cefotaxime, cephalothin, chloramphenicol, ciprofloxacin, enrofloxacin, gentamicin, streptomycin, tetracycline, and trimethoprim-sulfamethoxazole (Becton, Dickinson). In addition, presumptive ESBL-E. coli were tested for susceptibility to aztreonam, cefoxitin, ceftazidime, ceftriaxone, and imipenem (Becton, Dickinson). Clinical Laboratory Standards Institute guidelines for disc diffusion susceptibility breakpoints were followed (14) and the E. coli ATCC 25922 strain was used as a control. Any E. coli isolate classified as intermediate or resistant, using Kirby-Bauer disc diffusion methodology, and all the presumptive ESBL E. coli isolates had minimum inhibitory concentration (MIC) values determined by using the broth microdilution method (Sensititre; Trek Diagnostics, Cleveland, Ohio, USA) and the National Antimicrobial Resistance Monitoring System (NARMS) gram-negative MIC plate format CMV6CNCD (Sensititre; Trek Diagnostics), which comprises amikacin, amoxicillin-clavulanic acid, ampicillin, apramycin, cefoxitin, ceftiofur, ceftriaxone, cephalothin, chloramphenicol, ciprofloxacin, gentamicin, imipenem, kanamycin, nalidixic acid, streptomycin, sulfa-methoxazole, tetracycline, and trimethoprim-sulfamethoxazole. The resistance breakpoints were those used by the Canadian Integrated Program for Antimicrobial Resistance (CIPARS), which are derived from CLSI/NCCLS breakpoints and NARMS (15).

Molecular epidemiology

Presumptive ESBL-E. coli isolates, as well as any E. coli isolates resistant to cefotaxime, cefoxitin, and ceftiofur, were selected for amplification, sequencing, and hybridization, using polymerase chain reaction (PCR), in order to test for the presence of CMY-2 β-lactamase. Polymerase chain reaction primers for the identification of the Citrobacter freundii bla_AmpC gene encoding the CMY-2 β-lactamase were used (16,17).

Plasmid DNA preparations were electrophoresed and Southern blots were made. A labelled probe was generated [PCR DIG (digoxigenin); Probe Synthesis Kit, Roche Canada, Mississauga, Ontario] and hybridizations with the probe were performed. The PCR mixtures were cleaned prior to DNA sequencing (MinElute PCR Purification Kit; Qiagen Canada, Mississauga, Ontario). The DNA sequences were determined by using a sequencing kit (DYEnamic ET terminator cycle; Amersham Pharmacia, Piscataway, New Jersey, USA) and a sequencing system (MegaBACE 500; Amersham Pharmacia) (16).

The DNA-plugs were prepared following standard protocols and digested, using the restriction enzyme Xba1 (New England Biolabs; Ipswich, Massachusetts, USA) according to the manufacturer’s recommendations. The electrophoresis was performed in a 1% agarose gel (PFGE agarose, Sigma, St. Louis, Missouri, USA) and 0.5× Tris-borate-EDTA buffer at 12°C with an electrophoresis unit (CHEF Mapper; BioRad, Hercules, California, USA). An electrical field of 6 V/cm with an angle of 120° and a linear ramping switch time starting at 1 s and ending with 40 s was applied to the gel for a period of 16 h (18).

Statistical analysis

Descriptive statistical analyses were performed by using statistical software packages (Microsoft Excel 2000; Microsoft Corporation 1985–1999, Troy, New York, USA; Intercooled STATA 7.0, Stata Corporation 1984–2001, College Station, Texas, USA). Differences in the prevalence of resistance between dogs and cats were tested by using Fisher’s exact test. A P-value of ≤ 0.05 was considered significant. For the analysis of patterns of antimicrobial susceptibility within individual animal samples, intermediate susceptibility was combined with resistance and patterns were determined by using this susceptible versus reduced susceptibility classification.

Results

Thirty-two of the 131 (24.4%) practices contacted in the initial mailing were interested in participating in the study, but 2 declined after initial contact. The 99 practices that did not respond to the survey were contacted again, but none agreed to participate. From the remaining 30 practices, 20 were purposively selected for enrollment in the study. Two clinics requested removal from the study prior to arrangement of a clinic visit. The study population was drawn from the remaining 18 clinics.

A total of 188 dogs and 39 cats were enrolled in the study; 129 dogs and 27 cats from companion animal practices (n = 11), 59 dogs and 12 cats from mixed animal practices (n = 7). The mean number of dogs enrolled per clinic was 9 (range: 5 to 15). The mean number of cats enrolled per clinic was 3 (range: 0 to 11). The median age of dogs was 4 y (range: 4 wk to 16 y) and cats was 3 y (range: 8 wk to 16 y). The most common breed of dog was mixed breed (n = 57, 30.3%); the remainder consisted of 40 different purebred breeds. Most cats were domestic short hair (n = 28, 71.8%); 10 (25.6%) were domestic long hair and 1 was a Scottish fold.

Five isolates of E. coli per animal were selected from each sample (total canine isolates, n = 940; total feline isolates, n = 195). Three presumptive canine ESBL-E. coli isolates were recovered from the initial ESBL screening procedures. No Salmonella spp., MRSA, or MRSP bacteria were recovered.

The frequency of resistance in the E. coli isolates to most antimicrobials was low. Resistance was generally less frequent in isolates from cats (Table 1), but the difference was significant only for streptomycin (P = 0.02). When disc diffusion methods were used, intermediate susceptibility or resistance was observed in at least 1 canine isolate to all antimicrobials tested by disc diffusion, except ciprofloxacin. Although there were no isolates classified as resistant to enrofloxacin or cefotaxime, 1 canine isolate had intermediate susceptibility to enrofloxacin, and 5 canine isolates (from 5 dogs) had intermediate susceptibility to cefotaxime. When broth microdilution methods were used, none of the tested canine isolates (n = 299) were resistant to amikacin, ceftriaxone, imipenem, or ciprofloxacin. Three feline isolates (from 3 cats) had intermediate susceptibility to cefotaxime. Additionally, no feline isolate was identified as being resistant to amoxicillin-clavulanic acid, cefoxitin, ceftiofur, chloramphenicol, ciprofloxacin, enrofloxacin, gentamicin, or trimethoprim-sulfamethoxazole.

Table 1.

Prevalence of resistance to individual antimicrobials in canine and feline commensal Escherichia coli isolates, and percentage of dogs and cats with E. coli isolates resistant to individual antimicrobials, when the Kirby-Bauer disc diffusion method was used

	Isolate level prevalence (%)b		Animal level prevalence (%)b
Antimicrobiala	Canine (n = 940)	Feline (n = 195)	Dog (n = 188)	Cat (n = 39)
AMC	1 (0.4–2)	0 (0–2)c	4 (2–8)	0 (0–9)c
AMP	7 (6–9)	10 (6–15)	14 (9–20)	15 (6–31)
CEF	4 (3–5)	0.5 (0.01–3)	13 (8–19)	3 (0.06–13)
CHL	0.4 (0.1–1)	0 (0–2)c	2 (0.5–5)	0 (0–9)c
CIP	0 (0–0.04)c	0 (0–2)c	0 (0–2)c	0 (0–9)c
CXT	0 (0–0.4)c	0 (0–2)c	0 (0–2)c	0 (0–9)c
ENR	0 (0–0.4)c	0 (0–2)c	0 (0–2)c	0 (0–9)c
GEN	1 (0.6–2)	0 (0–2)c	3 (1–7)	0 (0–9)c
STR	9 (7–12)	4 (1–7)	17 (12–23)	8 (2–21)
SXT	2 (1–3)	0 (0–2)c	5 (2–9)	0 (0–9)c
TET	5 (4–7)	4 (1–7)	11 (7–16)	8 (2–21)

^aAMC — amoxicillin-clavulanic acid, AMP — ampicillin, CEF — cephalothin, CHL — chloramphenicol, CIP — ciprofloxacin, CXT — cefotaxime, ENR — enrofloxacin, GEN — gentamicin, STR — streptomycin, SXT — trimethoprim-sulfamethoxazole, TET — tetracycline.

^b95% confidence interval given in parentheses; prevalence is for all isolates considered as a population.

^c97.5% one-sided confidence interval; prevalence is on a per animal basis; that is, prevalence of dogs and cats with 1 or more E. coli isolates resistant to a given antimicrobial.

When disc diffusion methods were used, 3 or more distinct patterns of antimicrobial susceptibility were observed among the 5 isolates tested per animal in 34% of cats and 25% of dogs; 3% of cats and 4% of dogs had 5 different patterns of antimicrobial susceptibility among the 5 isolates, suggesting 5 distinct isolates (Figure 1).

Figure 1.

Number of phenotypically distinct antimicrobial susceptibility patterns among 5 commensal Escherichia coli isolates recovered from individual dog and cat fecal samples when the Kirby-Bauer disc diffusion method was used.

With the Kirby-Bauer disc diffusion technique (11 antimicrobials in the panel), overall there were 17 different patterns of multiple antimicrobial resistance (≥ 2 antimicrobials) in E. coli from dogs and 2 in E. coli from cats. When broth microdilution was used (17 antimicrobials in the panel), overall there were 30 different antimicrobial resistance patterns in E. coli from dogs and 6 combinations in E. coli from cats. When the broth microdilution and disc diffusion results were combined, 11% of dogs and 15% of cats had isolates that were resistant to ≥ 2 antimicrobials. Two dogs had isolates that were resistant to 6 of 11 antimicrobials, using disc diffusion, and 8 of 18 antimicrobials, using broth microdilution (broth microdilution results shown in Figure 2). One cat had an isolate that was resistant to 6 of the 18 antimicrobials tested with broth microdilution (Figure 2).

Figure 2.

Percentage of dogs (n = 113) and cats (n = 39) with Escherichia coli isolates that were resistant to 2–8 antimicrobials when broth microdilution^a was used.

^a Susceptibility testing, using broth microdilution, was performed on presumptive ESBL E. coli isolates and commensal E. coli isolates that were classified as intermediate or resistant, using the Kirby-Bauer disc diffusion method.

The ESBL screening procedure identified 3 isolates from 2 dogs (1 from #48 and 2 from #142) (Table 2) as presumptive ESBL-E. coli. An isolate with the identical susceptibility pattern to the 2 isolates from dog #142 was also recovered from dog #142 using the non-selective E. coli isolation method. Further susceptibility testing revealed that these 4 isolates were resistant to cefoxitin and amoxicillin-clavulanic acid, but fully susceptible in vitro to aztreonam, ceftriaxone, ceftazidime, and imipenem. An additional 2 isolates from dogs #61 and #131, isolated using the non-selective E. coli isolation method, were identified as resistant to cefoxitin, amoxicillin-clavulanic acid, and other antimicrobials. The observed patterns of resistance to a β-lactam/β-lactamase inhibitor combination, cephamycin and a third generation cephalosporin, were not characteristic of ESBL-E. coli, but rather of CMY-2-producing E. coli.

Table 2.

Patterns of antimicrobial resistance in commensal Escherichia coli with cefoxitin resistance isolated from healthy dogs

Dog number	Signalment	Antimicrobial prescription history	Resistance patternc
61	12-week-old female Siberian husky	none	AMC, AMP, CEF, FOX
131	6-year-old female spayed Rottweiler	none	AMC, CEF, FOX
48a	7-year-old male neutered cocker spaniel	single injection of long acting penicillin 5 years prior to study	AMC, AMP, CEF, CXT, FOX, KAN, SMX, SXT
142b	4-week-old male west highland terrier	none	AMC, AMP, CEF, CXT, FOX, FUR STR, TET

^aE. coli isolated using the ESBL screening procedures.

^bThree isolates obtained from this dog, 2 isolates using ESBL screening procedures, 1 using the non-selective E. coli isolation procedure. All 3 had the same susceptibility pattern.

^cAMC — amoxicillin-clavulanic acid, AMP — ampicillin, CEF — cephalothin, CXT — cefotaxime, FOX — cefoxitin, FUR — ceftiofur, KAN — kanamycin, NAL — nalidixic-acid, SMX — sulfamethoxazole, STR — streptomycin, SXT — trimethoprim-sulfamethoxazole, TET — tetracycline.

Isolates from dogs #48 (1 isolate) and #142 (2 isolates from ESBL E. coli isolation method and 1 from the non-selective E. coli isolation method) had a large 72 Mda plasmid that contained the bla_CMY-2 gene. In addition, the PCR product of the isolate of dog #142 was consistent with the presence of a plasmid-mediated cmy-2 gene. The isolate from dog #131 did not have any plasmids, but a PCR product of 631 bp was detected, using primers for the cmy-2 gene.

Discussion

In general, the prevalence of resistance at the isolate level was low for all antimicrobials tested. Resistance was most prevalent to streptomycin, ampicillin, and tetracycline. Despite this general finding, the dogs in this study had E. coli isolates with resistance to ceftiofur (n = 3) and amoxicillin-clavulanic acid (n = 7), antimicrobials that are defined as being of “very high importance” to human health by the Canadian Veterinary Drug Directorate (19). The prevalence of resistance to individual antimicrobials was higher at the pet level than at the isolate level. Resistance evaluation is a more clinically and epidemiologically relevant assessment at the animal level than that at the isolate level, since the phenotypes of multiple strains within individual animals more closely reflect their potential as a reservoir of resistance and the potential health risk to the animal.

There was a higher than expected prevalence of multiple drug resistance, including that of CMY-2 producing E. coli isolated from 2 dogs. There are reports of cmy-2 genes in E. coli and Salmonella spp. from humans (10), food-animals (20–22), and companion animals (23). In each of these reports, the cmy-2 gene was encoded on a plasmid. However, plasmids were not detected in the isolate from dog #131, although the PCR product obtained from this isolate was consistent with the cmy-2 gene. This suggests that the strain may have had a chromosomally located ampC gene that was up-regulated as a result of mutations of the promoter and attenuator regions (24).

Some human E. coli isolates expressing the cmy-2 gene have been reported to be indistinguishable from those isolated from food animals (21,22). Each of these studies has alluded to the use of extended-spectrum cephalosporins in animals as a plausible contributing factor to the emergence of resistance mediated by this gene. Ceftiofur is a third generation cephalosporin and is licensed for use in cattle, swine, sheep, horses, turkey poults, and dogs in Canada. The use of ceftiofur has been hypothesized as a risk factor for resistance mediated by the cmy-2 gene in cattle and turkeys (20,21) and a potential risk factor for the development of nosocomial infections associated with CMY-2 E. coli (23). However, the dogs identified as carriers of E. coli with the cmy-2 gene in the present study had no recent history of antimicrobial exposure. Although the estimated prevalence in this study was low [1% of dogs, 95% confidence interval (0.1%, 3.8%)], detection of any isolates in possession of this gene from this population of healthy pets may be cause for concern, particularly when no obvious potential risk factor was identified. Bacteria possessing the cmy-2 gene are resistant to cephamycins and β-lactam inhibitor combinations; therefore, therapeutic options for treatment of infected individuals are somewhat limited. In addition to the clinical significance, these isolates could have a public health impact, if zoonotic transfer occurs.

This study had some limitations. Far fewer cats were enrolled in the study than originally planned and anticipated. To minimize the stress in cats, sampling was limited to cats requiring sedation or general anesthesia. The frequency of these procedures was lower than the “wellness” type of appointment used to collect the canine samples. In addition, cats are generally seen less frequently than dogs at veterinary clinics in Ontario (25). No animals in the study were identified as being carriers of Salmonella spp., MRSA, or MRSP. However, some factors relative to sampling may have contributed to the lack of recovery and possible underestimation of colonization with these organisms. Rectal swabs yielded only a small amount of fecal material from which isolation of generic E. coli, ESBL-E. coli, and Salmonella spp. was performed. Therefore, there may not have been sufficient fecal matter for optimal recovery of Salmonella spp. In addition, shedding of Salmonella spp. can be intermittent (26) and repeated sampling is often required to identify carriers.

Although rectal swabs yield a small amount of fecal material, this approach was selected to facilitate animal enrollment and sample collection. Feces could have been collected from the rectum digitally; however, this would likely have yielded results similar to those obtained from the rectal swabs, with varying volumes of fecal matter being obtained, and rectal swabs are less invasive. To obtain a larger amount of fecal matter, pet owners would have had to collect the fecal samples themselves in advance of, or after, the clinic visit, or some arrangement would have to have been made with the participating practice, which would have increased the burden of participation. We anticipated that the latter would have resulted in poor compliance from the participating practices, as their ability to invest substantive time into this type of field research is limited, and the former would have resulted in poor compliance and poor sample quality due to transport delays. However, consequent to the results of this study, we have adopted the owner collection of fecal samples in subsequent studies with some success (27).

Sampling the perineal region may have led to underestimation of the prevalence of MRSA and MRSP colonization. Although the perineal region has been identified as a location where humans carry MRSA, the nasal mucosa has a higher sensitivity for the recovery of MRSA (28). Allaker et al (29) found that the recovery of coagulase-positive Staphylococcus spp. (CPS) in dogs was greatest from the nasal mucosa (15/20 dogs); however, the perianal region also had a relatively high recovery rate for CPS (11/20 dogs). Other studies, on hospital-visitation dogs (30) and dogs presented to a veterinary teaching hospital (31) in southern Ontario, have found similar prevalences. These studies suggest that even in dog populations in southern Ontario that are potentially at higher than average risk of colonization or infection by MRSA, the current prevalence is very low, likely < 2%. Given the sample size of this study, MRSA could have been present in as many as 2% of dogs in the population and still been undetected in our sample [observed prevalence 0%, 95% CI (0, 2%)].

Despite these limitations, the data obtained from the dogs in this study provide baseline prevalence of resistance and patterns of resistance in dogs in southern Ontario. The information gathered is useful in understanding the epidemiology of antimicrobial resistance in companion animals. These pets represent a community reservoir of antimicrobial resistant E. coli and resistance genes, including the cmy-2 gene, that could pose a risk to animal and human health. The results provide a starting point for investigating the impact of antimicrobial resistance in companion animals and contributing information to the global understanding of the epidemiology of antimicrobial resistance.