Article Text

This article has a correction. Please see:

Download PDFPDF

Paper
Trends in antimicrobial resistance in equine bacterial isolates: 1999–2012
  1. I. C. Johns, BVSc DACVIM-LA and
  2. E.-L. Adams, BVetMed
  1. Department of Clinical Sciences and Services, Equine Referral Hospital, Royal Veterinary College, North Mymms, Hatfield, AL97TA, UK
  1. E-mail for correspondence: ijohns{at}rvc.ac.uk

Abstract

This study aimed to identify changing antimicrobial resistance patterns in isolates commonly obtained from equine clinical submissions. Laboratory records from 1999 to 2012 were searched for equine samples from which Escherichia coli or Streptococcus species was isolated. Susceptibility to enrofloxacin, ceftiofur, gentamicin, penicillin G, trimethoprim sulfamethoxazole (TMPS) and tetracyclines was noted. Isolates were divided into those identified between 1999 and 2004 (Early) and between 2007 and 2012 (Late). The proportion of isolates resistant to each antimicrobial and multiple drug-resistant (MDR) isolates (≥3 antimicrobial classes) was compared between time periods. There were 464 isolates identified (242 Early; 222 Late). A significant increase in the percentage of E coli isolates resistant to ceftiofur (7.3–22.7 per cent, P=0.002), gentamicin (28.5–53.9 per cent, P<0.001), tetracyclines (48.4–74.2 per cent, P=0.002) and MDR (26.6–49.4 per cent, P=0.007) was identified. There was a significant increase over time in the percentage of all streptococcal species resistant to enrofloxacin, ranging from 0 per cent (Early) up to 63 per cent (Late) depending on species. For Streptococcus zooepidemicus, resistance over time to tetracyclines and MDR increased. There was also a decrease in the proportion of S zooepidemicus resistant to TMPS over time. An increase in resistance over time of common equine pathogens to a number of commonly used antimicrobials supports the responsible use of antimicrobials.

  • Antimicrobials
  • equine
  • Antimicrobial resistance
View Full Text

Statistics from Altmetric.com

Introduction

Antimicrobial resistance (AMR) is considered one of the key developing health concerns globally (WHO 2012, Dept. of Health 2013). Antimicrobial use and the development of resistance in bacteria isolated from animals have received much attention in this respect, because of the potential for human exposure to resistant bacteria via the food chain, and also in the case of horses and other pets via close human contact (Angulo and others 2004, Mølgak 2004, Turnidge 2004, Dunowska and others 2006, Boerlin and Reid-Smith 2008, Ahmed and others 2010, Bryan and others 2010, Maddox and others 2011, Damborg and others 2012, Maddox and others 2012a, Maddox and others 2012b). The emergence of antimicrobial-resistant bacteria isolated from horses has the potential to limit the efficacy of treatment of infectious diseases resulting in increased morbidity and mortality as well as cost of treatment.

Antimicrobial-resistant bacteria have been increasingly reported in samples obtained from horses. The majority of these studies have described resistance patterns in equine commensal organisms such as faecal Escherichia coli and have identified factors such as previous treatment with antimicrobials, and previous or current hospitalisation as risk factors for the presence of resistant isolates (Salyers and others 2004, Dunowska and others 2006, Ahmed and others 2010, Bryan and others 2010, Maddox and others 2011, Damborg and others 2012, Johns and others 2012, Maddox and others 2012a, Maddox and others 2012b, Williams and others 2013). Commensal organisms are commonly used in studies investigating AMR as they have the potential to also be pathogens, but more importantly they can serve as a reservoir of mobile genetic resistance elements which can be transferred to pathogens (Salyers and others 2004). AMR trends in bacteria isolated from a variety of equine clinical samples, including foals with sepsis, endometrial samples from mares and a wide range of samples from various body systems have been reported (Ensink and others 1993, Clark and others 2008, Russell and others 2008, Davis and others 2013, Theelen and others 2014). These data are perhaps more relevant to equine practitioners, as the information can be used to guide empirical antimicrobial therapy, allowing for informed decisions regarding appropriate antimicrobial choice to be made in clinical practice. Additionally, as antimicrobial use is considered the biggest driver for the development of AMR, if an increasing rate of resistance to a particular antimicrobial is identified, then strategies such as limiting use of the particular antimicrobial may be implemented in an attempt to minimise the progression of that resistance (Bronzwaer and others 2002, Kollef and Micek 2005).

When a horse with a suspected bacterial infection is examined, obtaining samples for culture and antimicrobial susceptibility, and making subsequent prescribing decisions based on those results is considered ‘gold standard’ (Morley and others 2005). However, in many clinical situations this is not practical, and in the case of life-threatening bacterial infections, a delay in instituting antimicrobial therapy until results are available would not be appropriate. As a result, empirical antimicrobial prescribing based on anticipated bacterial isolates and susceptibility patterns is a common practice (Hughes and others 2013). Knowledge of anticipated antimicrobial efficacy and potential AMR is an essential part of equine practice. This knowledge can be gained from the veterinary literature, although resistance patterns which are not up to date or relevant to the geographical area in which the samples were obtained may not be useful, as tremendous variability in the AMR patterns in different geographical regions, as well as marked changes over time have been reported (Ko and Hsueh 2009). Ongoing monitoring of resistance patterns in bacterial isolates is vital to ensure that empirical antimicrobial therapy is evidence-based and current, and also to track trends in resistance. Reports of antimicrobial susceptibility patterns in bacteria that have been recently isolated in the local geographical area are rarely available for equine veterinary surgeons to access, but may be the most accurate method of assessing anticipated antimicrobial efficacy. Using this information to practice evidence-based antimicrobial stewardship is a key step in attempting to delay the development of AMR (Bowen 2013).

The current study reports AMR patterns of equine origin E coli and β-haemolytic streptococcal species (BHS) submitted to a diagnostic laboratory in the UK between 1999 and 2012. E coli and BHS have been reported as the most commonly isolated Gram-negative and Gram-positive bacterial isolates from equine submissions, representing up to 57 per cent of isolates (Clark and others 2008, Russell and others 2008, Davis and others 2013). As such, evaluation of susceptibility patterns in these two indicator organisms can be used as a reference point for monitoring changes in AMR. Isolates cultured from samples obtained from horses in both a hospitalised and ambulatory setting were included to investigate whether hospitalisation was a risk factor for the presence of resistant isolates, and antimicrobial treatment before sampling was noted where the information was available.

The aims of the current study were to determine the rate of resistance to a panel of commonly used antimicrobials in equine E coli and BHS isolates from clinical samples submitted to a diagnostic laboratory in south-eastern UK and to determine whether the rates of AMR have changed over a 14-year period.

Materials and methods

The electronic database for the Royal Veterinary College Diagnostic Laboratory was searched for equine submissions between August 1999 and April 2012, from which either E coli and/or BHS was cultured. The Diagnostic Laboratory accepts samples submitted from both the Equine Referral Hospital and from samples submitted by ambulatory practitioners in the local area. Isolates were further classified according to their origin as respiratory, skin/wounds, gastrointestinal or other, to whether they were obtained from a hospitalised horse or not and whether they were sampled during the first six years of the study period (1999–2004; ‘Early’) or the last six years (2007–2012; ‘Late’). Two distinct time periods ‘bookending’ the period over which records were available were chosen to allow comparison over time. Only isolates sampled during these two time periods were included in the analysis. Clinical information, in particular, antimicrobial administration before sampling, was recorded when available.

Specimens were cultured for aerobic bacteria by plating each sample onto 5 per cent blood agar and MacConkey's agar and in samples submitted from the respiratory tract, chocolate agar (Oxoid, UK). From 2007, samples which the microbiologist anticipated culture of a Staphylococcus species (e.g. samples from the respiratory tract) were additionally plated onto plates containing nalidixic acid and colistin sulfate (Staph/Strep Selective Supplement Oxoid). The plates were incubated aerobically at 37°C and examined for growth at 24 hours and 48 hours. Chocolate agar plates were incubated at 37°C in 5–10 per cent CO2. Samples submitted in blood culture bottles were incubated for 30 minutes at 37°C, and then attached to a growth indicator device (Oxoid Signal Blood Culture System). The system was then incubated at 37°C overnight and assessed for growth. A subculture was taken following 24 hours of incubation, and plated onto 5 per cent sheep agar and MacConkey agar and incubated at 37°C under both aerobic and anaerobic conditions. Identification of E coli and β-haemolytic Streptococcus species was made following morphological assessment, Gram staining and biochemical tests which were performed according to standard microbiological practice (Barrow and Feltham 1993). Streptococcal isolates were further categorised into Streptococcus equi subspecies zooepidemicus (S zooepidemicus), S equi subspecies equi (S equi), Streptococcus dysgalactia subspecies equisimilis (S equisimilis) and unidentified β-haemolytic Streptococcus species (UBHS).

Antimicrobial susceptibility was assessed using the Kirby-Bauer disc diffusion method, and the zones of growth inhibition were evaluated according to the contemporary Clinical and Laboratory Standards Institute (CLSI; previously, NCCLS: National Committee for Clinical Laboratory Standards). IsoSensitest agar supplemented with 5 per cent lysed sheep's blood was used, apart from during 2003 and 2004 when Mueller-Hinton agar was used. During the study period, laboratory protocols were updated according to the CLSI/NCCLS guidelines, with the only major change as noted. Susceptibility to enrofloxacin, ceftiofur, gentamicin, penicillin and trimethoprim-sulfamethoxazole (TMPS) was assessed for all isolates. Additionally, susceptibility to oxytetracycline, doxycycline or both was routinely included for both E coli and BHS from March 2002. Susceptibility to both drugs was evaluated, and then combined as ‘tetracycline’ susceptibility, where resistance was defined if either doxycycline or oxytetracycline resistance (or both) was identified. Isolates were reported as susceptible to an antimicrobial if the diameter of the zone of inhibition was greater than the breakpoint for the drug, based on CLSI guidelines (NCCLS 1999, NCCLS 2004, CLSI 2008). Isolates with intermediate susceptibility were classified as resistant. The presence of multiple drug resistance (MDR) defined as resistance to three or more antimicrobial drug classes was determined (Maddox and others 2012a, Maddox and others 2012b). Due to inherent resistance, penicillin resistance in E coli isolates was excluded in the analysis of MDR, and only isolates in which susceptibility to either doxycycline or tetracycline (or both) had been evaluated were included for MDR evaluation. Oxytetracycline and doxycycline resistance were combined for the purposes of assessing MDR, and thus an isolate was considered MDR if it was resistant to three or more of the following: enrofloxacin, gentamicin, ceftiofur, penicillin (BHS only), TMPS or tetracyclines. The proportion of isolates obtained during the Early and Late periods and from hospitalised versus non-hospitalised horses was compared using the chi-squared goodness-of-fit. The proportion of each bacterial species resistant to each antimicrobial was compared between the Early and Late periods and between samples obtained from hospitalised horses and non-hospitalised horses using the chi-squared test for independence or the Fisher's exact test, as appropriate. The proportion of MDR isolates was similarly compared between isolates obtained in the Early versus Late period and from hospitalised versus non-hospitalised horses. Results were considered significant if P<0.05.

Results

A total of 212 E coli and 252 BHS (n=464) isolates were cultured from equine specimens submitted in the two time periods of the study (Table 1). Of the 252 BHS isolates, 69.8 per cent (176/252) were S zooepidemicus, 10.7 per cent (27/252) were S equi, 10.7 per cent (27/252) were S equisimilis and the remainder (22/252; 8.7 per cent) were UBHS. The majority of isolates (56.9 per cent) were obtained from the respiratory tract, including tracheal washes (68.6 per cent of respiratory samples), nasal or nasopharyngeal swabs (21.6 per cent), with the remainder from guttural pouch lavages and sinuses. Isolates from skin/wounds, which included incisional sites (73.4 per cent of skin/wounds), skin scrapes and skin biopsies, as well as traumatic wounds, comprised 20.3 per cent of all isolates. A small proportion of isolates were obtained from the gastrointestinal tract (3.7 per cent; peritoneal fluid, rectal biopsy, oesophagus) and other sites (19.0 per cent; including the reproductive tract, urinary tract and isolates from which the body system could not be determined).

TABLE 1:

Effect of Hospital 1's purchase of physician practices on prices and spending in ZIP codes A–D

Antimicrobial administration history was available for 187/464 (40.3 per cent) of horses from which isolates were obtained. Of these, 107 (57.2 per cent) had received antimicrobials. Although the proportion of isolates obtained from horses treated with antimicrobials was higher in the Late period, this was not statistically significant (47/91 51.6 per cent of Early samples versus 60/96 62.5 per cent of Late samples; P=0.177).

The total number of isolates was evenly divided between the Early (242/464; 52.2 per cent) and the Late periods (222/464; 47.8 per cent P=0.353). Significantly more isolates were obtained from non-hospitalised horses (272/464; 58.6 per cent) compared with hospitalised horses (192/464, 41.4 per cent; P<0.001). There was no significant difference in the proportion of samples in the Early period submitted from hospitalised horses (91/242; 37.6 per cent) compared with the Late period (101/222; 54.9 per cent; P=0.103). A significantly greater proportion of isolates obtained from hospitalised horses were E coli (114/192 59.4 per cent) compared with BHS (78/192 40.6 per cent; P<0.001). Similarly, a significantly greater proportions of isolates obtained from non-hospitalised horses were BHS (174/272; 64 per cent; P<0.001).

AMR in E coli isolates

Proportionally more E coli isolates were included from the Early period compared with the Late period (123/212 Early; 58.0 per cent; P=0.020). There was a significant increase over time in the proportion of isolates resistant to gentamicin, ceftiofur, doxycycline, tetracycline and MDR, but no increase in resistance to enrofloxacin, TMPS or oxytetracycline (Table 2). The proportion of isolates obtained from hospitalised horses was significantly different between the Early and Late periods, with more isolates obtained from hospitalised horses in the Late period (65.2 per cent) compared with the Early period (45.5 per cent; P=0.007). When comparing rates of AMR for E coli in samples obtained from hospitalised horses with samples obtained from non-hospitalised horses, significantly more resistant isolates were obtained from hospitalised horses for all antimicrobials tested, apart from ceftiofur and TMPS. The proportion of E coli displaying MDR was also significantly greater in samples obtained from horses that were hospitalised (Table 3). When samples obtained from hospitalised horses were compared between the Early and Late periods, there was a significant increase over time in the proportion of isolates resistant to gentamicin, ceftiofur, doxycycline and tetracyclines (Table 4). In non-hospitalised horses, a significant increase in the proportion of MDR E coli was identified between the Early and the Late periods (5/36, 13.9 per cent Early; 12/31, 38.7 per cent Late P=0.041). A variety of patterns of resistance were identified in MDR isolates: 9/61 (14.8 per cent of MDR; 4.2 per cent of all E coli) isolates were resistant to all tested drugs; 19/61 (31.1 per cent) showed susceptibility only to ceftiofur, 8/61 (13.1 per cent) showed susceptibility only to enrofloxacin, 7/61 (11 per cent) showed susceptibility only to enrofloxacin and ceftiofur, with the remainder showing mixed patterns of susceptibility.

TABLE 2:

Proportion of Escherichia coli isolates resistant to listed antimicrobials, and change over time. Tetracycline resistance was determined as resistance to doxycycline, oxytetracycline or both

TABLE 3:

Comparison of the percentage of Escherichia coli isolates resistant to various antimicrobials in samples obtained from hospitalised horses and non-hospitalised horses

TABLE 4:

Proportion of Escherichia coli isolates obtained from hospitalised and non-hospitalised horses resistant to tested antimicrobials

AMR in β-haemolytic Streptococcus species

β-haemolytic Streptococcus species isolates were evenly distributed between the Early (119/252; 47.2 per cent) and Late (133/252; 44.8 per cent) periods (P=0.378).

The majority of BHS were S zooepidemicus (176/252; 69.8 per cent), with the remainder evenly distributed between S equi (27/252; 10.7 per cent), S equisimilis (27/252; 10.7 per cent) and UBHS (22/252; 8.7 per cent) (Table 1).

All BHS species showed a significant increase over time in the proportion of isolates resistant to enrofloxacin (Table 5). There was also a significant increase in the proportion of S equi isolates resistant to gentamicin. S zooepidemicus isolates, additionally, showed an increased proportion resistant to oxytetracycline, tetracyclines and MDR, and a decrease over time in the proportion resistant to TMPS. The MDR S zooepidemicus isolates displayed a variety of resistance patterns, with the two most common being resistance to enrofloxacin, gentamicin and tetracyclines (12/32; 37.5 per cent of MDR isolates) or enrofloxacin, gentamicin and tetracyclines (6/32; 18.8 per cent). For the other BHS species, no distinct patterns of resistance predominated within the MDR isolates.

TABLE 5:

Proportion of streptococcal isolates resistant to various antimicrobials

The majority of all BHS isolates were obtained from non-hospitalised horses (Table 1), but the proportion obtained from hospitalised versus non-hospitalised horses was not different between the Early and the Late period (hospitalised: 29.4 per cent 35/119 Early; 32.3 per cent 43/133 Late. P=0.717). There were no significant differences in susceptibility to each antimicrobial tested between samples obtained from hospitalised and non-hospitalised horses apart from S zooepidemicus, where a greater proportion (6/58; 10.7 per cent) of hospitalised samples were resistant to penicillin (compared with 2/118; 1.7 per cent. P=0.027; Table 6). Looking at temporal changes in samples obtained from hospitalised horses, there was a significant increase over time in the proportion of S zooepidemicus isolates resistant to enrofloxacin, oxytetracycline and tetracyclines, and a significant decrease in the proportion of isolates resistant to TMPS (Table 7). Similar trends were identified in non-hospitalised horses with the addition of an increased proportion of isolates resistant to doxycycline and MDR. For all other BHS species, no changes in resistance pattern over time between hospitalised and non-hospitalised horses were identified.

TABLE 6:

Comparison of the percentage of Streptococcus zooepidemicus isolates resistant to various antimicrobials in samples obtained from hospitalised horses and non-hospitalised horses

TABLE 7:

Proportion of Streptococcus zooepidemicus isolates obtained from hospitalised and non-hospitalised horses resistant to tested antimicrobials

Discussion

The significant increase in the proportion of isolates resistant to commonly used antimicrobials, as well as resistance to antimicrobials designated as prioritised critically important by WHO highlights the need for ongoing monitoring of AMR and the importance of responsible antimicrobial use.

β-haemolytic Streptococcus species

Despite its widespread use, the majority of BHS isolates remain sensitive to penicillin, and its recommendation for use as a first-line antimicrobial for the treatment of respiratory infections in horses, where it is a commonly isolated pathogen, is supported by the findings in this study (Clark and others 2008, BEVA 2012). Although resistance to ceftiofur remained low throughout the study period, use of this third-generation cephalosporin, a WHO-designated prioritised critically important antimicrobial as a first-line treatment in horses should be discouraged, especially with the significant increase in resistance of E coli to the drug identified in this study and the increasing reports of extended-spectrum β-lactamase (ESBL) production in equine isolates (Dolejska and others 2011, Damborg and others 2012, Johns and others 2012).

The significant increase in the proportion of all BHS isolates resistant to the fluoroquinolone and enrofloxacin is of major concern. Fluoroquinolones are prioritised antimicrobials of critical importance by WHO, and as such, they should not be used as first-line therapy (Morley and others 2005, BEVA 2012, WHO 2012). Antimicrobial use is considered the biggest driver of AMR, and although a recent survey of antimicrobial-prescribing practices in the UK (where enrofloxacin is not authorised for use in horses) suggest that it is not frequently used, especially in first-opinion practice, it is possible that increased use of the drug has driven this increase in resistance (Hughes and others 2013). Insufficient data were available in the current study to determine whether previous treatment with enrofloxacin was associated with subsequent isolation of an enrofloxacin-resistant isolate. In a recent report, treatment with the drug was associated with a significantly higher risk of identifying an enrofloxacin-resistant faecal E coli in horses, suggesting a link between antimicrobial use and subsequent resistance (Johns and others 2012). A low rate of resistance to tetracyclines was identified in the Early period in the current study for all streptococcal species (0–4.2 per cent). This low rate of resistance has previously been reported in a small number of BHS from equine submissions in the Netherlands between 1988 and 1991 (Ensink and others 1993), although a report from the USA in 1998 identified higher rates of resistance in S zooepidemicus (29 per cent), highlighting the possibility of geographical variation in patterns of resistance (Wilson 2001). From the relatively low rate of resistance in the Early period, a dramatic increase in the proportion of all BHS (apart from S equi) resistant to tetracyclines was identified between the two time periods in the current study (38–50 per cent resistance in the Late period). Similar high rates of tetracycline resistance have been identified in recent studies from North America with rates of 35.3–56 per cent in BHS isolates excluding S equi (Clark and others 2008, Erol and others 2012). In agreement with other studies, variable rates of resistance to tetracycline within the BHS group were identified in the current study, with very low rates of resistance in S equi isolates and higher rates in S zooepidemicus and other BHS, respectively (Erol and others 2012). By contrast with the current study, however, no significant increase over time in resistance rates for S zooepidemicus alone, as compared with all BHS, was identified over a 10-year period from 2000 to 2010 (Erol and others 2012). S equi isolates remained uniformly sensitive to tetracyclines, consistent with the findings of Erol and others (2012). Gentamicin resistance rates for all BHS were high in the current study (51.9–74.2 per cent) and did not change over time, apart from S equi where there was a significant increase in the proportion of isolates resistant to gentamicin in the Late period. Previous reports have typically identified lower resistance rates of BHS to gentamicin (3.9–19 per cent) (Davis and others 2006, Clark and others 2008, Erol and others 2012, Theelen and others 2014). The reason for the discrepancy is unknown, but may be explained by antimicrobial administration before sampling, differing antimicrobial prescribing practices and geographical differences. Although gentamicin would not typically be selected as an empirical antimicrobial choice for a suspected streptococcal infection, the high rate of resistance to this antimicrobial in the current study, and the differences between studies, emphasises possible regional differences and the need for antimicrobial use policies based on local antimicrobial susceptibility patterns (BEVA 2012).

Potentiated sulfonamide combinations have traditionally been considered to have excellent antimicrobial activity against S zooepidemicus, with low resistance rates reported (Ensink and others 1993, Feary and others 2005). However, more recent studies have suggested that resistance in BHS to TMPS is widespread, with more than 60 per cent of S zooepidemicus isolates resistant in one study (Erol and others 2012). In the current study, a significant decrease in resistance of S zooepidemicus to TMPS (83.5–14.4 per cent) was identified between the Early and the Late periods, a finding not replicated in the other streptococcal species. Feary and others (2005) identified a falsely high rate of resistance of S zooepidemicus due to incorrect laboratory practices as a possible explanation for an increase in resistance over time (Feary and others 2005). In that study, using a quality control organism that was inappropriate for media containing whole blood resulted in a higher proportion of isolates being read as resistant due to indistinct zones of inhibition. Although it is possible that a similar error could explain the change in the resistant rates to TMPS, laboratory practice at the (name to be supplied upon acceptance) used lysed sheep's blood added to IsoSensitest agar, to avoid the presence of para-aminobenzoic acid or thymidine in whole blood-based media, with the use of test isolate recommended as a sensitive indicator of thymidine levels (E faecalis ATCC 29212).

A decrease in the proportion of isolates resistant to an antimicrobial is rarely reported. A recent report from Denmark reported a significant decrease in the percentage of extended-spectrum cephalosporinase-producing E coli in pigs following a voluntary ban on the use of cephalosporins in pork production, suggesting that decreased antimicrobial use can contribute to decreased rates of AMR (Agerso and Aarestrup 2013). Without data to support a decreased use of TMPS, it is impossible to know whether this change in resistance pattern is associated with a change in antimicrobial prescribing practice, or other factors including changing laboratory methods, changing submission trends or other reasons. Although an increased proportion of BHS isolates were considered sensitive to TMPS in vitro in the current study, this may not relate to an increased efficacy in vivo, as the drug has been shown to have relatively poor efficacy in tissue cage models of purulent infections induced by BHS (Ensink and others 2003, Ensink and others 2005).

While resistance to a number of antimicrobials has been reported in BHS, MDR is rarely reported. In the current study, 20.4 per cent of BHS isolates were classed as MDR, based on resistance to three or more antimicrobial classes. Within the BHS species evaluated, MDR varied with higher rates reported for S zooepidemicus (18.8 per cent) and S equisimilis (46.2 per cent), and lower rates for S equi and UBHS (<10 per cent). As mentioned previously, a higher-than-expected proportion of all BHS were resistant to gentamicin, which is likely to have contributed to this high proportion of MDR isolates. Indeed, if gentamicin is removed from the drug classes used for evaluation of MDR, then only 5.2 per cent (13/230) of BHS isolates would be considered MDR.

E coli

The current study reports an increased resistance to ceftiofur, gentamicin and tetracyclines over a 14-year period, as well as an increased proportion of isolates with MDR. High rates of resistance to all tested drugs were identified, as well as MDR in 40 per cent of isolates. This effect was most evident in samples obtained from hospitalised horses, with only an increase in MDR over time identified in samples obtained from non-hospitalised horses.

Resistance of E coli to ceftiofur increased from 7.3 per cent in the Early period to 22.7 per cent in the Late period. Ceftiofur is a third-generation cephalosporin licensed in the UK for the treatment of horses with bacterial respiratory disease associated with Streptococcus species (including S zooepidemicus and S equi), Staphylococcus species and/or Pasteurella species. (NOAH 2014). Resistance to third-generation and fourth-generation cephalosporins, in particular, through production of extended-spectrum β-lactamases, has been recognised as an increasing problem in equine E coli isolates (Dolejska and others 2011, Damborg and others 2012, Johns and others 2012). Treatment with β-lactam antimicrobials, in particular, third-generation and fourth-generation cephalosporins, is a recognised risk factor for the identification of ESBL-producing isolates in both human beings and horses (Ofner-Agostini and others 2009, Damborg and others 2012). As such, it is possible that the increasing resistance to ceftiofur in E coli isolates in the current study may be linked to increasing use of the drug over the study time period, although without antimicrobial prescribing information, this remains speculative. The genes which code for ESBL production are most commonly located on integrons, transposons and/or plasmids, which may also contain resistance genes to other unrelated antimicrobials (Schwaber and others 2005, Dolejska and others 2011). Treatment with antimicrobials other than cephalosporins, such as fluoroquinolones, can thus co-select for cephalosporin resistance (Johns and others 2012). In one study, the odds of ceftiofur resistance being identified was 14.7 times more likely in horses being treated with enrofloxacin, thus supporting this concept (Johns and others 2012).

Although there was no significant increase in the rate of enrofloxacin resistance in E coli isolates over time, approximately 25 per cent of all isolates (33.8 per cent in hospitalised patients) were resistant to enrofloxacin. Similarly, high rates of enrofloxacin resistance have been recently reported in E coli from uterine cultures although there was a variation in the year-to-year resistance, from as low as 3 per cent of isolates up to 22 per cent (Davis and others 2013). Treatment with antimicrobials, as well as hospitalisation, are considered risk factors for fluoroquinolone resistance (Johns and others 2012, Maddox and others 2012a, Williams and others 2013). In the current study, a larger proportion (67.4 per cent) of E coli isolates were obtained from hospitalised horses, which may explain the relatively high resistance rates.

This retrospective study has several important limitations, in particular, the lack of clinical information in the majority of cases. Without knowledge of why samples were being obtained (e.g. refractory infections v samples obtained before treatment) and whether the reasons for sampling changed over time, it is impossible to know whether the resistance changes identified here are reflective of a true increase in AMR or merely a reflection of an increase in sampling of refractory infections. For those horses in which clinical information was available, there was no difference in the proportion treated with antimicrobials between the Early and Late periods suggesting, at least for these isolates, that previous administration of antimicrobials cannot explain the increased resistance rates. Samples were analysed from both hospitalised and non-hospitalised horses in an attempt to determine whether antimicrobial-resistant isolates were present in community-acquired infections and whether this differed from rates in hospitalised animals. Although the majority of BHS isolates were obtained from non-hospitalised horses, the proportion of hospitalised versus non-hospitalised did not change between the Early and Late periods, suggesting that a valid comparison over time could be made. By contrast, for E coli, more isolates were obtained from hospitalised horses, which were over-represented in the Late period. As such, it is perhaps most appropriate to independently compare isolates obtained from hospitalised versus non-hospitalised horses to determine temporal trends; these findings confirmed that in this study, the change in resistance to antimicrobials, such as gentamicin and tetracyclines, was predominately noted in specimens from non-hospitalised horses. Although antimicrobial use is a recognised risk factor for the isolation of resistant isolates, this information was only available for 40 per cent of isolates, precluding meaningful analysis of this as a risk factor in the current study. Finally, it is important to recognise that breakpoints are subject to revision by the CLSI or British Society for Antimicrobial Chemotherapy over time and, therefore, could be the primary cause of differences in percentages of bacteria reported as being susceptible when studies performed over a prolonged time period are compared.

It is also difficult to make direct recommendations regarding antimicrobial use from the data provided in the current study. Although the source of the isolates was reported, the clinical presentation was unknown for the majority of cases, limiting the ability to apply the information directly to clinical cases. However, in many cases, the bacteria isolated from a sample can be predicted, either based on the clinical presentation or, in some cases, by cytological examination and Gram staining. The information can then be applied, pending microbiological culture and susceptibility results.

Conclusions

The findings in this study suggest that AMR to most commonly used antimicrobials in equine practice has increased over time, with high rates of resistance in some cases, including in samples obtained from horses treated outside a hospital setting. This emphasises the need for ongoing monitoring of AMR, as well as highlighting the need for responsible antimicrobial stewardship. As antimicrobial use is considered the biggest driver of the development of AMR, modifications to how and when to use antimicrobials are needed. The usage of antimicrobial use policies, such as those suggested in the British Equine Veterinary Association document ProtectME provide useful guidelines on responsible antimicrobial use for equine practitioners. Further research is required to investigate methods by which the rate of AMR development can be slowed and how empirical antimicrobial choices can be optimised.

Acknowledgments

The authors would like to thank Peter Dron, Dr Ruby Chang and the technical staff at the Diagnostic Laboratory, Royal Veterinary College for their assistance in the preparation of this manuscript.

References

View Abstract

Footnotes

  • Provenance: Not commissioned; externally peer reviewed

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Correction
    BMJ Publishing Group Limited