Multidrug-resistant bacteria isolated from surgical site of dogs, surgeon's hands and operating room in a veterinary teaching hospital in Brazil

The aim of this study was to evaluate the prevalence and antimicrobial resistance prole of Gram-positive cocci and Gram-negative bacilli isolated from the surgical environment. All samples were collected during the intraoperative period of clean/clean-contaminated (G1) and contaminated (G2) surgery. A total of 150 samples were collected from the surgical wound in the beginning (n = 30) and end (n = 30) of the procedure, surgeon’s hands before (n = 30) and after (n = 30) antisepsis and the surgical environment (n = 30). Forty-three isolates with morphological and biochemical characteristics of Staphylococcus spp. and 13 of Gram-negative bacilli were obtained. Coagulase-negative staphylococci (85.71% [18/21]), coagulase-positive staphylococci (9.52% [2/21]) and Pseudomonas spp. (47.52% [1/21]) in G1, and coagulase-negative staphylococci (40% [14/35]), coagulase-positive staphylococci (20% [7/35]), Proteus spp. (17.14% [6/35]), E. coli (8.57% [3/35]), Pseudomonas spp. (2.86% [1/35]) and Salmonella spp. (2.86 [1/35]) in G2 were more frequently isolated, and a high incidence of multidrug resistance was observed in coagulase-negative staphylococci (87.5% [28/32]), coagulase-positive staphylococci (100% [11/11]) and Gram-negative bacilli (76.92% [10/13]). Methicillin-resistant Staphylococcus spp. accounted for 83.72% (36/43) of the Staphylococcus strains. Gram-negative bacilli cefotaxime-resistance constituted 81.82% (9/11) and imipenem resistance constituted 53.85% (7/13). The high rate of resistance of commensal bacteria found in our study is worrying. Coagulase-negative staphylococci are community pathogens related to nosocomial infections in human and veterinary hospitals, their presence in healthy patients and in veterinary professionals represent an important source of infection in the one health context. Continuous surveillance and application of antimicrobial stewardship programs are essential in the ght against this threat.


Introduction
The infection caused by multidrug-resistant bacteria (MDR) presents a main challenge in the one health context because it increases morbidity, mortality and the costs related to healthcare (Suthar et al. 2014;Schwarz et al. 2016;McEwen and Collignon, 2018). The zoonotic potential of most MDR bacterial species related to serious infection highlight veterinary medicine as a relevant axis in the ght against to dissemination of these pathogens (Guardabassi et al. 2004a;Guardabassi et al. 2004b;Brusselaers et al. 2011;Suthar et al. 2014;Madec et al. 2017;McEwen and Collignon 2018;Grӧnthal et al. 2018;Shoen et al. 2019).
Methicillin-resistant Staphylococcus (MRS), extended-spectrum β-lactamases (ESBL) Enterobacteriaceae species, Pseudomonas aeruginosa and Acinetobacter baumannii are commonly cause of surgical site infections in pets and humans. These bacterial groups are the most prevalent microorganism acquired during the hospitalisation in the intensive care unit in human and veterinary medicine (Gandol -Decristophoris et al. 2012;Turk et al. 2014;Andrade et al. 2016;Walther et al. 2016;Chaudhary et al. 2017; Kaspar et al. 2018;McEwen and Collignon 2018;Shoen et al. 2019).
One of the main risk factors to post-surgical infection in dogs are the bacterial colonisation of skin, mucous and gastrointestinal tract by MDR bacteria and the presence of these pathogens in the operating room (Chauveaux 2015;Andrade et al. 2016;Burgess 2019). Furthermore, the introduction of these strains in the hospital environment poses a big threat to health care professionals, which become carriers and disseminators of MDR bacteria (Paul et al. 2011;Rodrigues et al. 2017;Kaspar et al. 2018).
Therefore, the aim of this epidemiologic study was to evaluate the incidence and antimicrobial resistance pro le of Gram-positive cocci and Gram-negative bacilli present in the super cial surgical site from dogs, in surgeon's hands and in the surgical environment during the intraoperative period.

Patient and surgical team preparation
The inducing and maintaining anaesthesia protocol were performed individually by small animal anaesthesiology care according to the American Society of Anaesthesiologists classi cation (Mayhew 2019). Cefazolin sodium (25mg/kg [IV]) was administered between 30 and 60 minutes before surgical incision. Hand's antisepsis was performed with 2% chlorhexidine solution or 2% povidone iodine.
Hair removal by clipping was initiated in the surgical preparation room and nalized after induction in the operating room. Surgical site antisepsis was performed with 2% chlorhexidine solution, 70% isopropyl alcohol and 0,5% alcoholic chlorhexidine solution.

Sample collection
Specimens were collected using a dry sterile cotton-tipped swab and transported in a sterile tube with 0,1% peptone salt solution. Sixty samples were collected from the hands surface of the main surgeon immediately before (C1 [n=30]) and after (C2 [n=30]) antisepsis. Swabs were rubbed in a circular motion from the wrists to the ngertips (back and palm of the hand), this procedure was repeated three times toward each nger by the same people with sterile gloves in all collections.
Immediately after the skin incision (M1 [n=30]) and previously to cutaneous closure (M2 [n=30]), the swab was rubbed in the super cial surgical site (skin and subcutaneous tissue) from the patient. The specimens of the surgical site were double collected. A Petri plate with Brain Heart Infusion agar (BHI; Oxoid®, United Kingdom) was positioned nearly and in the same height to the surgery table during the procedure.
Bacteriological culture Immediately after the procedure, each sample was spread onto MacConkey agar (Oxoid®, United Kingdom) and 5% bovine blood agar (Oxoid®, United Kingdom) and, together with the BHI environment plate, were incubated at 37ºC for 24h, under aerobic conditions. Three morphologically distinct colonies were collected from each plate. Each strain was classi ed by cellular morphology, Gram staining, presence or absence of spores, and biochemical techniques (Barrow and Feltham 1993).
Catalase test was performed to identify bacterial genera of Staphylococcus spp. and Streptococcus spp. Coagulase test was performed with lyophilized rabbit plasma (Coagu-plasma®, Laborclin, Brazil) for the Staphylococcus spp. strains. Gram-negative strains were spread in Triplice Sugar Iron agar (TSI; Oxoid®, United Kingdom) and incubated at 35ºC for 18-24h, under aerobic conditions. Urease test was performed for the Salmonella spp. and Proteus spp. strains and citrate tests were performed for E. coli and Klebsiella spp. strains (Koneman et al. 2001).

Antimicrobial susceptibility tests
The antimicrobial susceptibility test was performed and inhibition zone diameters were measured following CLSI standard guideline (CLSI 2013;CLSI 2015). The antimicrobials disks (DME®, Brazil) used in the test are described in Table 1. Clindamycin and erythromycin antimicrobial disks were placed at a distance of 15mm to perform inducible clindamycin resistance testing (D-test). Table 1 Drugs used in the antimicrobial susceptibility test Legend: *The italic classes were used to classi ed multidrug resistance.
The proportion of resistance was obtained by analysing the speci c resistance of each identi ed pathogen to every antimicrobial tested (% R). The intermediate results were considered to be resistant to the data analyses since the outcome of treatment for these cases directly depended on the antimicrobial dose and the site of infection (CLSI 2015).
The phenotypic presence of the mecA gene (MRS) was assigned to the pathogens which exhibited oxacillin resistance (Bemis et al. 2009;Wu et al. 2016;Naccache et al. 2019). The phenotypic presence of the ermA gene (MLS B phenotype) was assigned to Staphylococcus spp. strains which exhibited erythromycin resistance and a D-shape zone around the clindamycin in the area between the two disks (Fiebelkorn et al. 2003). MDR was con rmed when a strain was resistant to at least 1 drug belonging to 3 or more antimicrobial classes (Magiorakos et al. 2012;Basak et al. 2016).

Statistical analysis
A descriptive analysis of the infection sites, pathogens, and resistance proportion was performed. The chi-square test was used to compare bacterial resistance rate. All analyses were performed using R ® software 3.3.0 (R Foundation for Statistical Computing, Austria). The signi cance threshold was set at 0.05.
Statistical difference in the different moments of collect was observed only in the proportion of strains obtained from surgical sites in M1 (p=0.007). The bacteria growth in both groups was described in Table  2. Table 2 Number and proportion of strains obtained from environment, surgeon's hands and super cial surgical site Legend: C1=surgeon's hands before antisepsis, C2=surgeon's hand after antisepsis, M1=initial moment, M2= nal moment. Signi cance level was set at p<0.05.
Contaminated surgeries showed a high number of Gram-positive cocci (70%) and Gram-negative bacilli (50%), which is expected due to the higher rate of bacterial contamination of the super cial surgical site in these procedures (Nelson 2011;Turk et al. 2014;Chaudhary et al. 2017). Gobbo et al. (2017) and Menezes et al. (2020) reported the same pathogens and high proportion of antimicrobial resistance in samples of surgical environment and clinical infections in a veterinary teaching hospital in São Paulo, Brazil. Staphylococcus spp. was the most frequent species reported in other veterinary teaching hospitals in other states in Brazil (Fernando et al. 2015;Murta et al. 2015). Although the pathogens were similar in these studies, the difference in the bacterial prevalence, mainly between different states, underscores the local and temporal dynamics of the bacterial epidemiology.
The bacterial species observed in this study were the most prevalent related to the surgical site infections in dogs in Brazil and other countries (Suthar et al. 2014;Fernando et al. 2015;Murta et al. 2015;Verwilghen andAmeet 2015, Andrade et al. 2016). Staphylococcus spp. are commensal pathogens from skin and mucosa of humans and animals (Brusselaers et al. 2011;Paul et al. 2011;Paul et al. 2012), what justi es the high prevalence of these microorganisms in samples of the surgical site and surgeon's hands in both groups (Fig. 1).
We observed more CoNS strains (32/43 [74.42%]) than CoPS (11/43 [25.58%]), and CoNS was most prevalent than CoPS in G1 (p < 0.05). This fact may be justi ed due CoNS strains, mainly S. epidermidis, are opportunistic pathogens frequently found in dog's and human's skin and usually is not related to infectious in health patient (Kern and Perrenten 2013;Wu et al. 2016;Shoen et al. 2019). S. pseudointermedius and S. aureus are frequently reported in cutaneous infections in dogs, what is usually observed in super cial surgical site in the contaminated surgery (Guardabassi et al. 2004a;Howe and Boothe 2006;Hillier et al. 2014;Singh and Weese 2017).
MRS strains present resistance to all β-lactam drugs due to the presence of the mecA gene (Bemis et al. 2009;Wu et al. 2016;Naccache et al. 2019). In the present study, oxacillin-resistance was most prevalent in CoNS (90%) than CoPS (64%), which is concerning because these species have become potential pathogens related to sepsis in intensive care units (ICU). Furthermore, these strains are associated with bio lm formation and can cause post-surgical infections until one year after the procedure (Becker et  CoNS in humans' hospitals, which was also observed in our study (Fig. 2). The high proportion of oxacillin-resistance in our strains could lead to surgical antimicrobial prophylaxis failure, which is usually performed with rst and second-generation of cephalosporins (Boothe and Boothe 2015;Singh and Weese 2017).
Three of 37 (8.11%) Staphylococcus spp. strains, that were resistant to erythromycin, presented inducible clindamycin resistance (D-test). This fact can occur due to the presence of erm gene, which encodes resistance to macrolides, lincosamides and streptogramin B, called the MLS B phenotype (Fernandes et al. 2007). The proportion of strain positive to D-test in our study was more than other reports in North America (Faires et al. 2009;Gold and Lawhon 2013) and less than health workers in North Africa (Mahmoud et al. 2015).
Fluoroquinolone resistance (Fig. 2) was between 54-91%, highlighting Gram-negative bacilli resistance to enro oxacin (91%). This drug is usually used to treat serious infections caused by enterobacteria strains and Pseudomonas spp. in veterinary medicine (Papich 2013;Feng et al. 2019;Pang et al. 2019). Your empiric administration could contribute to the resistance to other antimicrobials of the same class used to treat human infections, as cipro oxacin (Barrasa et al. 2000;Papich 2013). In this study, we observed similar proportions of resistance between the drugs of this class (Fig. 2).
The gastrointestinal tract of dogs and humans are colonized by bacteria from the Enterobacteriaceae family. The presence of these pathogens in the hospital environment and skin may indicate a failure in cleaning and disinfection of these surfaces (Cinquepalm et al. 2013;Suthar et al. 2014;Drzewiecka 2016). In this study, we observed that in G2 the strains of these pathogens were larger (13/35 [37.14%]) than in G1, in which no strain was obtained. The most frequent species were Proteus spp. (6/10 [60%]) and E. coli (3/10 [30%]).
The presence of these strains, mainly Proteus spp., in the surgical site and surgeon's hands is probably due to cross-infection and could suggest a failure in antisepsis techniques (Drzewiecka 2016). Chlorhexidine compounds are more effective against Gram-positive than Gram-negative strains, which may justify the persistence of enterobacteria after biocide application (Karpinski and Szkaradkiewicz 2015;Belo et al. 2018).
ESBL production is the mainly mechanism of resistance in Enterobacteriaceae family, which results in penicillin, cephalosporin, and aztreonam hydrolysis. β -lactamase inhibitors, such as clavulanic acid, sulbactam, and tazobactam could inactivate your action, and these strains are usually sensitive to carbapenem class. The cefotaxime-resistance could be used to detect the phenotypic production of this enzyme (CLSI 2015;Laudy et al. 2017). Gram-negative bacilli obtained in our study presented high prevalence (82%) of cefotaxime-resistance.
Carbapenemase-production is another important resistance mechanism of Enterobacteriaceae. These strains presented resistance to all drug of the β-lactam class (Seibert et al. 2014). The high prevalence of imipenem-resistance (54%), a carbapenem class-drug, in our study is worrying because this drug is used to treat serious infections in human medicine. Furthermore, the proportion of MDR strains were high in our strains of CoNS (88%), CoPS (100%) and Gram-negative bacilli (77%), which could mean high probability in treatment failure (Madec et al. 2017).
The presence of MDR strains in healthy patients and in the environment is frequently reported in the literature (Chah et al. 2013;Priyantha et al. 2016, Dupouy et al. 2019Ortega-Paredes et al. 2019), as well as the possibility of transmission to the veterinarians (Paul et al. 2011;Rodrigues et al. 2017). Thus, the patient, health professional and environment's colonisation for these strains, highlighting MRS, ESBL Enterobacteriaceae and carbapenemase-producing Enterobacteriaceae, could represent a high risk to human and veterinary patient due therapy failure, and to public health because they are important silent sources of infection (Guardabassi et al. 2004a;Guardabassi et al. 2004b, Becker et al. 2014Suthar et al. 2014). Furthermore, these strains need, urgently, development for news drugs to treat e ciently these infections (McEwen and Collignon 2018).
This study has some limitations. Although morphological and biochemical tests are not the most sensitive alternative for diagnosing species and bacterial strains, they are nancially viable and technically simpler than molecular tests, as a polymerase chain reaction, 16S rRNA gene sequencing, and pulsed-eld gel electrophoresis (Ribot et al. 2006;Woo et al. 2008;Srinivasan et al. 2015). The results obtained in the bacteriological culture and antimicrobial susceptibility tests are important in clinical care and they are essential in an epidemiological database for the development of antimicrobial stewardship program and in epidemiological surveillance in the one health context.

Conclusion
This study identi ed a high rate of methicillin-resistant CoNS (90%), Gram-negative bacilli resistant to cefotaxime (82%) and MDR strains (87.50%). Furthermore, we observed a high resistance proportion in drugs used to treat serious infections in humans, as teicoplanin (37.21%) and imipenem (54%). These high rates of resistance are worrying because these strains were collected in skin surfaces without infectious, and the pathogens described here are frequently reported in human and veterinary surgical site infection and intensive care units. Beyond the high potential for therapy failure due to infections for MDR bacteria, we observed the veterinary surgical environment as potential reservoir and disseminator of these pathogens, mainly of community species, as MRS.
Finally, this study highlights the need of control and prevention in MDR strains transmission in healthcare, that address the topic according to the One Health concept and included veterinary medicine as an important cornerstone in the ght against this threat.  Tables   Table 1 Drugs used in the antimicrobial susceptibility test Legend: *The italic classes were used to classi ed multidrug resistance.