Temporal trends in in vitro antimicrobial susceptibility patterns of bacteria isolated from foals with sepsis: 1979–2010

Authors


Summary

Reasons for performing the study

Monitoring the development of antimicrobial resistance is important for the rational selection of appropriate antimicrobial drugs to initiate treatment of foals with sepsis.

Objectives

To identify temporal trends in antimicrobial susceptibility patterns of bacteria isolated from foals with sepsis.

Study design

Retrospective review of medical records.

Methods

Foals aged <30 days with a diagnosis of sepsis, confirmed by culture of bacteria, were included. Susceptibility data, expressed as minimum inhibitory concentrations (MICs) (MIC50, MIC90, MIC range) and percent of isolates that were susceptible to a particular antimicrobial drug, were compared for bacteria isolated from foals during 3 different time periods: 1979–1990, 1991–1997 and 1998–2010. The Cochran-Armitage trend test and the Jonckheere-Terpstra test were used for statistical analysis.

Results

A total of 1091 bacterial isolates were cultured from 588 foals. Enterobacteriaceae, Actinobacillus spp. and β-haemolytic Streptococcus spp. showed a decrease in percent of isolates susceptible to gentamicin over time. Enterobacteriaceae, Actinobacillus spp. and β-haemolytic Streptococcus spp. showed an increase in MIC values for amikacin. Enterobacteriaceae showed a decrease in percent of isolates susceptible to ceftiofur. Enterococcus spp. and Pseudomonas spp. showed increased MIC values to ceftiofur. Enterobacteriaceae showed increased MIC values to ceftizoxime. Enterococcus spp. became more resistant to imipenem and showed increased MIC values to ticarcillin/clavulanic acid. In contrast, several trends in increased susceptibility were also seen.

Conclusions

Based on these in vitro results, the combination of amikacin and ampicillin remains an appropriate choice for initiating treatment of sepsis in foals while awaiting culture and susceptibility test results, although increasing development of resistance to amikacin was demonstrated. The decrease in in vitro activity of ceftiofur against Enterobacteriaceae is of concern. Similarly, the development of resistance of Enterococcus spp. to imipenem is an important finding that warrants monitoring in the future. Judicious use of antimicrobials is therefore crucial.

Introduction

Genes coding for antimicrobial resistance are present in bacteria cultured from horses [1, 2]. The emergence of bacteria that are resistant to antimicrobials could have significant health implications for horses, including foals, and must be monitored by susceptibility testing of bacteria isolated from appropriate samples collected from clinical patients, as well as from animals free from disease [1, 2]. Selection of antimicrobials for initial treatment is typically based on knowledge of susceptibility patterns of bacteria previously isolated from horses with the same or similar disease syndromes. Several factors, including hospitalisation and prior use of antimicrobials on the farm of origin, influence these susceptibility patterns [3-5]. Over time, trends of increasing or decreasing susceptibility of bacterial isolates to particular antimicrobial drugs can be observed. It is, therefore, necessary to regularly re-evaluate antimicrobial susceptibility profiles in order to provide the clinician with up-to-date information on which to base rational selection of antimicrobials for use in initial treatment protocols. This is particularly important when treating rapidly progressive life-threatening infections such as sepsis in foals.

Antimicrobial susceptibility patterns of bacterial isolates from foals with sepsis have been presented in several reports dating back to 1982 [6-12]. However, only 2 reports describe temporal trends in susceptibility patterns of bacterial isolates, and in only one of these studies are susceptibility data specified to the level of individual bacterial species [7, 12]. None reported susceptibility data in the form of minimum inhibitory concentrations (MICs).

This study represents an extension of work that has been ongoing at the University of California (UC) Davis for many years, the overall objective of which is to generate quantitative antimicrobial susceptibility data to guide rational selection of antimicrobial drugs for inclusion in treatment protocols for foals with sepsis. The goal of the current study was to document temporal trends in antimicrobial susceptibility patterns of bacteria isolated from foals diagnosed with sepsis.

Materials and methods

Case selection

The case records of foals ≤30 days of age presented to the William R. Pritchard Veterinary Medical Teaching Hospital (VMTH), UC Davis, USA, between January 1979 and December 2010, were reviewed. Data recorded in the medical record at admission and during hospitalisation of the foal were retrieved from the Veterinary Medical and Administrative Computing System (VMACS), as used by VMTH personnel at UC Davis, or from stored paper records if the data had not been entered into VMACS.

Records for those foals with a diagnosis of sepsis confirmed by culture of bacteria from blood or multiple internal organs were selected for further evaluation. Foals from which samples were collected by tracheal wash or by swabbing the umbilicus after surgical preparation, were also selected for further evaluation. Cases were included only if they showed clinical or laboratory signs of systemic sepsis (fever [>38.9°C], neutropenia or neutrophilia (<4.0 × 109/l or >12.0 × 109/l), increased band neutrophil count (>0.05 × 109/l), toxic changes in neutrophils, hyperfibrinogenaemia (>4.0 g/l), hypoglycaemia (<4.4 mmol/l), metabolic acidosis, scleral injection, petechiation, anterior uveitis, diarrhoea, respiratory distress or joint swelling. A total of 588 foals met the criteria for inclusion in the study.

Bacterial culture, identification and classification

Samples retrieved from foals with sepsis originated from several locations. For blood culture, up to 3 blood samples (5–10 ml each) were collected aseptically from the jugular or the cephalic vein, either by venepuncture or through an i.v. catheter. Culture of blood was performed using broth inoculation, with or without antibiotic resin (Trypticase Soy Broth)a, or by the lysis-centrifugation method (Isolator)b.

Foals that died or were subjected to euthanasia were given post mortem examination, during which samples from internal organs (e.g. liver, kidney, spleen, brain, body cavity or joint) were retrieved aseptically for bacteriological culture. All samples collected ante or post mortem were submitted to the VMTH Microbiology Diagnostic Laboratory (MDL) at UC Davis for bacterial culture and identification using conventional microbiological methods. All isolates, including those collected at necropsy, were saved as frozen stabilates at -80°C in skimmed milk or on Microbank beadsc and were available for later susceptibility testing.

Antimicrobial susceptibility testing

Susceptibility testing of isolates was performed using the microdilution Sensititred procedure, following Clinical Laboratory Standards Institute (CLSI) protocols [13, 14]. One bacterial colony was inoculated into brain-heart infusion broth and incubated for 4 h at 35°C. A small amount of this inoculated broth was then added to 0.85% NaCl solution to achieve a 0.5 McFarland Standard concentration, as measured using a nephelometer. A 10 μl sample of this suspension was then added to Mueller Hinton broth, and Sensititre plates (prior to 1988 plates were prepared manually) were inoculated with 100 μl of the Mueller Hinton broth in each well. The MIC was recorded as the lowest concentration of antimicrobial drug that inhibited visible growth of bacteria.

The following bacterial strains were run weekly as controls in accordance with the standard quality control procedures in place at the MDL: Staphylococcus aureus ATCC 29213, Enterococcus faecalis ATCC 29212, Escherichia coli ATCC 25922, E. coli ATCC 35218 and Pseudomonas aeruginosa ATCC 27853.

Sensititre plates were custom made for the UC Davis MDL. Bacterial isolates were tested for susceptibility to 15 antimicrobial drugs. An isolate was considered to be susceptible to a particular antimicrobial drug if its MIC value was less than or equal to the clinical MIC susceptibility breakpoint for that antimicrobial, as determined by the CLSI, occasionally modified by equine-specific interpretations based on research performed in horses [13, 14].

For bacteria isolated from samples collected ante mortem, antimicrobial susceptibility testing was performed at the time of the patient's visit. Because not all the antimicrobial drugs included in this study were available during the earlier time periods, it was necessary to retrieve some isolates from the lyophilised sample repository at a later date for susceptibility testing against these antimicrobials. For example, ceftiofur was not added to the panel of antimicrobials tested until 1988, and imipenem was not added until 1992. Bacteria isolated from samples collected at necropsy were all susceptibility tested retrospectively after re-culturing from the lyophilised repository and confirming the identity of the isolate.

When multiple isolates of the same bacterial species were retrieved from different samples or locations in a particular patient, they were considered to be the same isolate if their colony morphology, biochemical characteristics and antimicrobial susceptibility patterns were identical. Otherwise, isolates from different sites were considered to be different strains of the same bacterial species and both were included in the present study. Using the above criteria, a total of 1091 bacterial isolates were cultured from 588 foals.

Antimicrobial drugs

The following 15 antimicrobial drugs were analysed for activity against specific species of bacteria: amikacin, ampicillin, ceftiofur, ceftizoxime, cefalotin, chloramphenicol, enrofloxacin, erythromycin, gentamicin, imipenem, penicillin, rifampicin, tetracycline, ticarcillin/clavulanic acid and trimethoprim/sulfamethoxazole (TMS). The drugs included in this study that are designated as critically important antimicrobials by the World Health Organization were: ceftiofur, ceftizoxime, enrofloxacin, erythromycin and imipenem [15, 16].

Time periods for evaluation of temporal trends

The following 3 time periods were established in order to identify potentially significant temporal trends in the antimicrobial susceptibility profiles of bacterial isolates: 1979–1990, 1991–1997, and 1998–2010. These time periods were selected to take into account changes in approaches to antimicrobial use in neonatal foals at UC Davis during the 31 years of the study.

Data analysis

The Jonckheere-Terpstra test was used to detect significant temporal changes over time in MIC50 and MIC90, and the Cochran-Armitage trend test was used to detect significant temporal trends in the percentage of isolates susceptible (%S) to a particular antimicrobial over the 3 time periods. Analysis of these data was performed using commercial software (StatXact Version 9.0)e. Results were considered significant if P≤0.05.

Results

Cumulative susceptibility patterns of bacterial isolates over the entire timeframe of the study (1979–2010)

Escherichia coli and Klebsiella spp. isolates were highly susceptible (>90%) in vitro to amikacin, ceftiofur, ceftizoxime, enrofloxacin and imipenem. Enterobacter spp. isolates were highly susceptible in vitro to amikacin, enrofloxacin and imipenem and Salmonella spp. isolates were highly susceptible in vitro to amikacin, ceftizoxime, cefalotin, enrofloxacin and imipenem. More than 87.5% of Actinobacillus spp. isolates were susceptible to all antimicrobials tested, except for erythromycin and penicillin. Pseudomonas spp. isolates were found to be highly resistant to most of the antimicrobials tested, although amikacin (88.9%), ticarcillin/clavulanic acid (88.0%) and imipenem (82.6%) were active against >80% of the isolates.

With regard to Gram-positive isolates, more than 90% of β-haemolytic Streptococcus spp. isolates were susceptible in vitro to cefalotin, ceftizoxime, imipenem, ceftiofur, rifampicin, penicillin, chloramphenicol, ampicillin, erythromycin and TMS. More than 90% of Staphylococcus spp. isolates were susceptible to imipenem, amikacin, cefalotin and enrofloxacin. In addition, more than 90% of coagulase-negative Staphylococcus spp. isolates were susceptible to chloramphenicol and rifampicin. Enterococcus spp. isolates were highly resistant to most antimicrobials tested in the current study; none of the antimicrobials showed a susceptibility percentage of >90%. Ampicillin (72.4% susceptible) and chloramphenicol (75.7%) showed the highest level of in vitro activity against Enterococcus spp.

Cumulative susceptibility data on Gram-negative, Gram-positive and all isolates combined are presented in Table 1. Susceptibility data specified to the level of individual bacterial species are presented in Supporting Information Items S1–S12.

Table 1. Temporal and cumulative susceptibility of Gram-negative, Gram-positive and all isolates 1979–2010
Antimicrobial drugGram-negative isolates
1979–19901991–19971998–2010Total
%S (n)%S (n)%S (n)%S (n)
Amikacin98.5% (202)95.9% (271)96.3% (190)96.8% (663)
Ampicillin55.1% (225)59.4% (271)53.4% (193)56.3% (689)
Ceftiofur91.7% (156)94.1% (272)84.7% (190)90.6% (618)
Ceftizoxime91.0% (188)95.6% (272)91.7% (181)93.1% (641)
Cefalotin64.1% (209)48.2% (272)66.0% (156)57.8% (637)
Chloramphenicol74.0% (227)69.9% (276)75.3% (182)72.7% (685)
Enrofloxacin92.6% (150)97.1% (272)94.0% (182)95.0% (604)
ErythromycinNANANANA
Gentamicin82.4% (221)77.7% (264)71.4% (189)77.4% (674)
Imipenem99.3% (149)99.6% (247)98.1% (160)99.1% (556)
PenicillinNANANANA
RifampicinNANANANA
Tetracycline57.1% (217)62.8% (266)72.1% (183)63.5% (666)
Ticarcillin/clavulanic acid82.1% (151)79.1% (277)82.7% (196)80.9% (624)
Trimethoprim/sulfamethoxazole69.1% (220)70.6% (272)68.1% (191)69.4% (683)
Antimicrobial drugGram-positive isolates
1979–19901991–19971998–2010Total
%S (n)%S (n)%S (n)%S (n)
Amikacin54.8%* (42)45.8%* (59)39.7%* (58)45.9%* (159)
Ampicillin81.1% (53)76.1% (71)77.8% (90)78.0% (214)
Ceftiofur86.5%* (37)92.7%* (55)93.0%* (57)91.3%* (149)
Ceftizoxime90.0%* (40)94.8%* (58)92.9%* (56)92.9%* (154)
Cefalotin100%* (37)96.5%* (57)100%* (47)98.6%* (141)
Chloramphenicol80.3% (61)91.0% (78)89.2% (93)87.5% (232)
Enrofloxacin58.7% (46)59.7% (77)54.9% (91)57.5% (214)
Erythromycin61.8% (55)70.3% (74)63.4% (93)65.3% (222)
Gentamicin57.5%* (40)43.9%* (57)32.2%* (59)42.9%* (156)
Imipenem95.8% (48)91.2% (68)83.8% (80)89.3% (196)
Penicillin72.5%* (40)72.4%* (58)82.8%* (58)76.3%* (156)
Rifampicin76.5% (51)75.3% (77)69.8% (96)73.2% (224)
Tetracycline63.2% (57)63.2% (76)66.3% (95)64.5% (228)
Ticarcillin/clavulanic acidNANANANA
Trimethoprim/sulfamethoxazole90.5%* (42)87.9%* (58)84.5%* (58)87.3%* (158)
Antimicrobial drugAll isolates
1979–19901991–19971998–2010Total
%S (n)%S (n)%S (n)%S (n)
  1. n = number of isolates tested; %S = percentage of susceptible isolates; NA = not available; * = Enterococcus spp. isolates not included.
Amikacin91.0%* (244)87.0%* (330)83.1%* (248)87.0%* (822)
Ampicillin60.1% (278)62.9% (342)61.1% (283)61.5% (903)
Ceftiofur90.7%* (193)93.9%* (327)86.6%* (247)90.7%* (767)
Ceftizoxime90.8%* (228)95.5%* (330)92.0%* (237)93.1%* (795)
Cefalotin69.5%* (246)56.5%* (329)73.9%* (203)65.2%* (778)
Chloramphenicol75.3% (288)74.6% (354)80.0% (275)76.4% (917)
Enrofloxacin84.7% (196)88.8% (349)81.0% (273)85.2% (818)
ErythromycinNANANANA
Gentamicin78.5%* (261)71.7%* (321)62.1%* (248)71.0%* (830)
Imipenem98.5% (197)97.8% (315)93.3% (240)96.5% (752)
PenicillinNANANANA
RifampicinNANANANA
Tetracycline58.4% (274)62.9% (342)70.1% (278)63.8% (894)
Ticarcillin/clavulanic acidNANANANA
Trimethoprim/sulfamethoxazole72.5%* (262)73.6%* (330)71.9%* (249)72.8%* (841)

Temporal trends in susceptibility patterns of bacterial isolates

The numbers of isolates included in each period of time were as follows: 1979–1990: 328 isolates, 1991–1997: 415 isolates, 1998–2010: 348 isolates.

Two different temporal trends were analysed. The first was a significant difference in the percentage of isolates that were categorised as being susceptible or resistant to a particular antimicrobial drug based on the clinical MIC breakpoint for susceptibility. The second temporal trend was a significant change in MIC values over time, reflected in an increase or decrease in values for MIC50, MIC90 and MIC range limits. An increase in MIC values potentially indicates a trend towards the incremental development of resistance, but is not always reflected as a significant change in the percentage of isolates that are categorised as susceptible or resistant based on the clinical MIC breakpoint. A temporal trend of increased MIC values could potentially be regarded as an ‘early warning’ sign for the development of resistance [17]. The opposite can also be seen: a decrease in MIC values over time may indicate a trend towards the development of increased susceptibility.

Statistically significant temporal trends in increased MIC values and decreased percent of isolates classified as being susceptible to particular antimicrobial drugs can be found in Table 2. Enterobacteriaceae (and the subgroup Enterobacter spp.), Actinobacillus spp. and β-haemolytic Streptococcus spp. showed an increase in MIC values for amikacin. In the group β-haemolytic Streptococcus spp. this trend was reflected in a decrease in the percentage of susceptible isolates. For the other aminoglycoside, gentamicin, increased MIC values were found in the group of the Enterobacteriaceae (and the subgroup E. coli) and in Actinobacillus spp. A decreased percent of isolates in the group Enterobacteriaceae (and the subgroup Salmonella spp.), Actinobacillus spp. and β-haemolytic Streptococcus spp. were susceptible to gentamicin over time.

Table 2. Statistically significant temporal trends (P<0.05) in increased minimum inhibitory concentrations (MIC) values and decreased susceptibility of bacterial isolates to antimicrobial drugs
Antimicrobial drugIncrease in MIC valuesDecrease in percentage susceptible isolates
Amikacin

Enterobacteriaceae

Enterobacter spp.

Actinobacillus spp.

β-haemolytic Streptococcus spp.

β-haemolytic Streptococcus spp.
Ceftiofur

Enterobacteriaceae

Pseudomonas spp.

Enterococcus spp.

Enterobacteriaceae

Escherichia coli

Ceftizoxime

Enterobacteriaceae

E. coli

Salmonella spp.

 
Gentamicin

Enterobacteriaceae

E. coli

Actinobacillus spp.

Enterobacteriaceae

Salmonella spp.

Actinobacillus spp.

β-haemolytic Streptococcus spp.

Imipenem Enterococcus spp.
Ticarcillin/clavulanic acidEnterococcus spp. 

Concerning the cephalosporins, an increase in MIC values for ceftiofur was observed for Enterobacteriaceae, Pseudomonas spp. and Enterococcus spp. In the group of Enterobacteriaceae (and the subgroup E. coli), a decreased percentage of isolates were susceptible to ceftiofur over time. Increased MIC values for ceftizoxime were found for Enterobacteriaceae (and subgroups E. coli and Salmonella spp.). These increased MIC values were not, however, reflected in a decreased percentage of susceptible isolates in these groups. Enterococcus spp. showed a significant decrease in percent of isolates susceptible to imipenem over time and also showed increased MIC values for ticarcillin/clavulanic acid.

Statistically significant temporal trends in decreased MIC values and increased antimicrobial susceptibility can be found in Table 3. Enterobacteriaceae (and the subgroups E. coli and Salmonella spp.) showed an increase in percent of isolates susceptible to tetracycline over time. Actinobacillus spp. showed an increase in percent of isolates susceptible to ampicillin, penicillin and rifampicin and also showed a decrease in MIC values for erythromycin. Staphylococcus spp. (and its subgroup coagulase-positive Staphylococcus spp.) showed an increase in percent of isolates susceptible to chloramphenicol. Coagulase negative Staphylococcus spp. showed an increase in percent of isolates susceptible to ceftiofur and penicillin. Pseudomonas spp. showed an increase in percent of isolates susceptible to gentamicin and Klebsiella spp. showed a decrease in MIC values for ampicillin.

Table 3. Statistically significant temporal trends (P<0.05) in decreased minimum inhibitory concentrations (MIC) values and increased susceptibility of bacterial isolates to antimicrobial drugs
Antimicrobial drugDecrease in MIC-valuesIncrease in percentage susceptible isolates
AmpicillinKlebsiella spp.Actinobacillus spp.
Ceftiofur Coagulase-negative Staphylococcus spp.
Chloramphenicol 

Staphylococcus spp.

Coagulase-positive Staphylococcus spp.

ErythromycinActinobacillus spp. 
Gentamicin Pseudomonas spp.
PenicillinActinobacillus spp.

Actinobacillus spp.

Coagulase-negative Staphylococcus spp.

RifampicinActinobacillus spp.Actinobacillus spp.
Tetracycline

Enterobacteriaceae

Escherichia coli

Klebsiella spp.

Enterobacteriaceae

E. coli

Klebsiella spp.

Temporal susceptibility data on Gram-negative, Gram-positive and all isolates combined are presented in Table 1. The complete data on susceptibility of isolates in each time group, specified to the level of bacterial species and including the temporal differences, are presented in Supporting Information Items S1–S12 and can be accessed online.

Discussion

Some of the drugs evaluated in this study are classified as critically important antimicrobials by the World Health Organization, which means they are regarded as critically important to human health [15]. As clearly outlined in the antimicrobial stewardship policy of the Equine Veterinary Journal, these drugs should be reserved for cases where no other alternatives are effective and only after appropriate susceptibility testing or when evidence for their use in certain diseases is compelling [16].

Cumulative susceptibility of bacterial isolates 1979–2010

The finding that more than 90% of Enterobacteriaceae in this study were susceptible in vitro to amikacin confirms its utility as a first-choice antimicrobial for treating Gram-negative sepsis in foals. Enterobacteriaceae also showed a high level of susceptibility to enrofloxacin, imipenem and, to a lesser extent, ceftizoxime and ceftiofur, indicating that these drugs could be useful alternatives to amikacin under special circumstances in which the use of an aminoglycoside is contraindicated or the infecting organism is resistant to amikacin. Imipenem should be reserved to treat highly resistant infections in horses and should not be used as a first choice antimicrobial drug [18]. Enrofloxacin may induce arthropathy when used in neonatal foals; therefore, its use should be avoided unless other options are not feasible [19, 20]. Third-generation cephalosporin antimicrobials such as ceftiofur or ceftizoxime are therefore preferred for use in cases where administration of an aminoglycoside is contraindicated, such as in foals with azotaemia, dehydration or renal failure, or in those that are being treated concurrently with other nephrotoxic drugs.

Gram-positive organisms showed considerable differences in susceptibility patterns. As ampicillin is the drug most commonly used to treat Gram-positive sepsis in foals, it should be recognised that only β-haemolytic streptococci showed a high percent of isolates to be susceptible to this drug (94.6%: 1979–2010). Ampicillin is also one of the drugs with the highest level of in vitro susceptibility (72.4%: 1979–2010) against Enterococcus spp. Data on the susceptibility of Enterococcus spp. to penicillin are not presented because the breakpoint currently used for Enterococcus spp. was outside the range of concentrations tested during several of the years included in this study. Although Enterococcus spp. isolates have been reported to be highly susceptible in vitro to aminoglycosides, cephalosporins and TMS, it has also been reported that in vitro results for these antimicrobials often do not translate into effectiveness in vivo [13, 14]. Therefore, only MIC values, and not %S Enterococcus spp. isolates, are reported for these drugs in the Supporting Information Items. Enterococcus spp. are known to be intrinsically resistant to several antimicrobial drugs, and also readily accumulate mutations and exogenous genes that confer additional resistance through plasmids and transposons [21, 22]. The susceptibility pattern of Enterococcus spp. isolates is highly unpredictable and can pose a real therapeutic challenge for clinicians [21, 22]. Extensive use of antimicrobials imposes a strong selection pressure on Enterococcus spp. and other bacterial species and favours the selection of resistant strains [23].

When interpreting the results presented for β-haemolytic streptococci, one should keep in mind that despite in vitro activity of TMS against β-haemolytic streptococci, this drug has previously been shown to be ineffective in eradicating Streptococcus equi ssp. zooepidemicus in vivo in horses [24, 25].

Although several published studies have reported susceptibility data for bacteria isolated from foals with sepsis [6-12], few studies have presented susceptibility data for individual bacterial species [6, 9-12], and none have reported MIC values for these isolates. Additionally, the number of isolates tested in most of these studies was relatively low (Brewer and Koterba [10] n = 108; Marsh and Palmer [6] n = 203; Russell et al. [9] n = 124; Hollis et al. [11] n = 75), making meaningful comparisons with the results of our study difficult [6, 9-11]. Another factor that complicates comparison between studies is the method used to determine susceptibility – either disc diffusion techniques or a breakpoint inhibitory concentration system – and the breakpoints used. It is important to recognise that breakpoints are subject to revision by CLSI over time and therefore could be the primary cause of differences in percentages of bacteria reported as being susceptible when studies performed in different years and at different facilities are compared. Because the actual interpretive breakpoints used in the studies referenced above were not stated, it is not possible to determine whether this important factor influenced the results obtained [6, 9-11]. This underlines the importance of presenting MIC values in reports on antimicrobial susceptibility because it allows the data to be reinterpreted when recommended interpretive breakpoints for susceptibility change over time. The number of Gram-negative enteric bacteria included in the study by Sanchez et al. [12] was higher (n = 274), permitting a more meaningful comparison. Enterobacteriaceae isolated in our study appeared to be less susceptible to chloramphenicol (68.6% vs. 84.6%), gentamicin (75.3% vs. 92.1%), tetracycline (56% vs. 76.4%) and TMS (63.8% vs. 80.4%) than in the study reported by Sanchez et al. [12]. Percentages of Gram-negative enteric isolates that were susceptible to amikacin, ampicillin, ceftiofur, enrofloxacin and imipenem were similar in both studies. The differences between the results of the cited study and our study could reflect geographical or management factors or differences in usage of antimicrobial drugs that could affect the selection pressure for resistance. The year in which a study is completed is also likely to influence susceptibility results because temporal changes in susceptibility can be expected, as documented in the current study. Because our study and the one reported by Sanchez et al. [12] were conducted over a similar timeframe, this factor is unlikely to account for observed differences in results between the 2 studies.

Temporal trends in susceptibility patterns

Three time periods were established in order to identify potentially significant temporal trends in antimicrobial susceptibility profiles of bacterial isolates. Time periods were selected to take into account changes in approaches to antimicrobial use in neonatal foals at UC Davis during the 31 years of the study. The time periods selected were 1979–1990, 1991–1997 and 1998–2010. Prior to 1990, gentamicin was the aminoglycoside antimicrobial of choice for inclusion in treatment regimens for sepsis in foals. A change in the approach to initial antimicrobial therapy was made in 1990 based on the publication of a study by Wilson et al. [8] documenting that a substantially higher proportion of Enterobacteriaceae isolated from foals with sepsis were susceptible to amikacin than to gentamicin. In 1997, Madigan published an article in which he advocated the use of antibiotics prophylactically in foals that were born unobserved or had recognised risk factors [26]. This publication caused veterinarians in the field and at the VMTH to increase the prophylactic use of antibiotics in neonatal foals.

Changes in the commonly prescribed or first line antimicrobial drugs and dosing regimens used to treat foal sepsis at the VMTH over the years are outlined below:

1970s: Penicillin G (20,000–40,000 iu/kg bwt i.v. q. 6 h) and kanamycin (5 mg/kg bwt i.m. q. 8 h) or TMS (30 mg/kg bwt per os q. 12 h).

Late 1970s and 1980s: Gentamicin (2.2 mg/kg bwt i.v. q. 8 h or 3.3 mg/kg bwt i.v. q. 12 h) and penicillin G (20,000–40,000 iu/kg bwt i.v. q. 6 h) or ampicillin (20 mg/kg bwt i.v. q. 6–8 h).

1990s–present: Amikacin (7 mg/kg bwt i.v. q. 8 h or 10 mg/kg bwt i.v. q. 12 h until 1995, and 21–25 mg/kg bwt i.v. q. 24 h from 1995 until the present) and ampicillin (20 mg/kg bwt i.v. q. 6–8 h), or ceftiofur sodium (5–10 mg/kg bwt i.v. or i.m. q. 12 h).

The breakpoint inhibitory concentration system was used consistently throughout all years of our study, as were the breakpoints used to classify isolates as susceptible or resistant. Therefore, observed temporal changes in susceptibility data could not be ascribed to changes in methodology or to revisions of CLSI-recommended breakpoints over time. An additional feature of the current study was that susceptibility results were reported as quantitative MIC ranges, MIC50 and MIC90, as well as in the form of categorical (susceptible vs. resistant) data based on established breakpoints. There are several advantages to this approach [17, 27]. First, the development of bacterial resistance to a particular antimicrobial may be incremental and result in changes in MIC range, MIC50 and MIC90 for that bacterial species, but not a change in the percent of isolates classified as susceptible or resistant based on a particular breakpoint. These parameters may therefore be more sensitive early indicators of the development of resistance. Additionally, CLSI-recommended changes in MIC breakpoints over time do not influence raw MIC values, whereas they may influence calculated percentages of susceptible and resistant organisms. From a clinical standpoint, another important advantage of quantitative MIC data over qualitative susceptibility data is that the dose and dosing frequency for a particular antimicrobial can be adjusted, either up or down, to better customise the treatment protocol to the specific bacterial isolate from a particular case.

Several aspects of the design of this study could have influenced the results obtained. Susceptibility testing was not performed at the same time for all isolates included in this study. Some isolates were tested at the time of collection; others were analysed after storage as frozen stabilates at -80°C. Whereas storage is unlikely to have influenced antimicrobial susceptibility test results, the true impact is not known.

Prior administration of antimicrobial drugs and hospitalisation before sampling can influence susceptibility profiles of bacteria isolated from horses [5]. Not all isolates included in the current study originated from samples that were collected at the time of admission: several were from foals that had already been hospitalised for a variable period and others were isolated from samples collected at necropsy. The foals from which these isolates were cultured had typically, although not consistently, received antimicrobial treatment before the samples were obtained. Data on antimicrobial treatment before admission were not consistently available for all cases and could not therefore be taken into account. Whereas the inclusion of bacteria isolated from samples collected at necropsy from foals that died or were subjected to euthanasia may have influenced the overall susceptibility test results for the entire timeframe of the study, it is unlikely that inclusion of these isolates influenced the observed temporal trends in antimicrobial susceptibility. The proportion of isolates originating from necropsy samples was similar for all 3 timeframes (1979–1990: 29.6%; 1991–1997: 31.3%; 1998–2010: 20.4%).

It is not known whether the site of sample collection has any influence on susceptibility patterns of bacterial isolates. Such an analysis was judged to be infeasible in this study because of the low number of isolates from most sites.

The development of resistance to aminoglycosides was clearly evident over the years of this study. Decreased susceptibility of Enterobacteriaceae (and the subgroup Salmonella spp.), Actinobacillus spp. and β-haemolytic streptococci to gentamicin probably reflects the extensive use of gentamicin in foals and mature horses in California during the 1970s and 1980s, and has led clinicians in most parts of the USA to replace gentamicin with amikacin as the first-choice antimicrobial for treating foal sepsis while awaiting culture and susceptibility test results. It should be recognised, however, that clinicians in some other parts of the world continue to regard gentamicin as the drug of choice for treating foals with Gram-negative sepsis. Such an approach is rational as long as the susceptibility of Gram-negative bacterial isolates to gentamicin in the particular location is carefully monitored.

The increase in MIC values of Enterobacteriaceae (and the subgroup Enterobacter spp.), Actinobacillus spp. and β-haemolytic Streptococcus spp. for amikacin, the drug currently used most commonly to treat Gram-negative infections in foals in the USA, as detected in this study, is of concern. Although for Enterobacteriaceae (and the subgroup Enterobacter spp.) and Actinobacillus spp. the change in MIC values has not yet led to a significant reduction in the percentage of bacteria that are classified as susceptible to amikacin, this finding may be an ‘early warning’ sign for the development of resistance to amikacin [17]. The use of amikacin has expanded since the early 1990s, partially in response to a publication by Wilson et al. [8] reporting that a higher portion of Enterobacteriaceae were susceptible to amikacin than to gentamicin. The increased use of amikacin has presumably induced a selection pressure for resistance. Cross-resistance among aminoglycosides has been reported, although this is rarely seen for amikacin [28]. The number of amikacin-inactivating enzymes elaborated by bacteria is much lower than the number of inactivating enzymes for gentamicin, which may explain why the decrease in susceptibility of Gram-negative organisms to gentamicin occurs more rapidly than does the development of resistance to amikacin [28]. Ongoing monitoring of resistance will be important to determine whether amikacin will remain a reliable first-choice antimicrobial for treating foal sepsis in the future.

Owing to the nephrotoxic potential of aminoglycosides, third-generation cephalosporins are commonly used as an alternative antimicrobial treatment in foals affected by azotaemia, dehydration or renal failure. Ceftiofur is the most commonly used drug in this class. Enterobacteriaceae as a group (and the subgroup E. coli), showed a decrease in percent susceptible isolates to ceftiofur over time. Other bacteria, such as Pseudomonas spp. and Enterococcus spp., showed a significant increase in MIC values for ceftiofur, indicating that incremental development of resistance is occurring. This decrease in in vitro activity of ceftiofur is of concern and suggests that ceftiofur will potentially be less effective for treating foals with sepsis caused by Gram-negative enteric organisms than was the case in years past.

Of all the major species of Gram-positive bacteria, Enterococcus spp. showed the most significant temporal trends in both MIC values and percentage of susceptible isolates, demonstrating the development of resistance. Particularly noteworthy is the finding that Enterococcus spp. showed a decrease in percent of isolates susceptible to imipenem over time, despite imipenem not being used commonly on horse farms or in our hospital (<5 foals/year in our hospital). Imipenem is used more often in human medicine but is reserved for treatment of special cases in which no other drugs appear to be effective [18]. One hypothesis to explain the above finding would be nosocomial transmission of imipenem-resistant Enterococcus spp. from man to horses in the farm or equine hospital environment. Proof of this hypothesis would require further research. Besides increased resistance to imipenem, Enterococcus spp. also showed a significant increase in MIC values in more recent years for ceftiofur and ticarcillin/clavulanic acid, potentially indicating a trend towards the development of resistance. Resistant Enterococcus spp. isolates not only pose a risk to equine health but are also of serious concern in human health care [22]. Sepsis caused by Enterococcus spp. in man is often nosocomial in origin and has become more prevalent in recent years [29, 30].

Despite the frequent use of TMS in equine practice, no significant changes in susceptibility were found in the present study.

In contrast to the observed increases in resistance to antimicrobial drugs, reduced resistance (increased susceptibility) was also noted for some antimicrobial drugs, particularly those that are no longer used commonly in foals in California. Enterobacteriaceae (and subgroups E. coli and Klebsiella spp.) showed an increased percentage of isolates susceptible to tetracycline, to the extent that almost 70% of Enterobacteriaceae were susceptible to tetracycline in 1998–2010. Whereas 70% susceptibility is not sufficient to make tetracycline a suitable alternative first-choice antimicrobial for treating sepsis in foals, this finding is consistent with the results of one previous study [12], and suggests that reduced use of this drug may have led to less selection pressure for the emergence of resistance and, in turn, favoured the re-emergence of susceptible strains. Staphylococcus spp. (and subgroup coagulase-positive Staphylococcus spp.) showed an increase in percent of isolates susceptible to chloramphenicol. The use of chloramphenicol is prohibited in large parts of the world owing to legislation and, in our hospital, its use to treat infections in foals has been infrequent during the past 20 years. The sparse use of this drug may have reduced or reversed the selection pressure for the emergence of resistance, resulting in an increase in percent of staphylococci susceptible to chloramphenicol.

Coagulase-negative Staphylococcus spp. showed an increase in percent susceptible isolates to ceftiofur and penicillin over time, even though both drugs are used commonly in our hospital. The reason for this seemingly paradoxical finding is not clear.

Implications for antimicrobial use in foals

Guidelines and consensus statements have been published in recent years to educate veterinarians about the importance of judicious and rational use of antimicrobials, to assist them in the process of rational antimicrobial selection, and to reduce the likelihood of development of antimicrobial resistance [16, 18]. Although culture and susceptibility testing of bacteria isolated from the particular case prior to initiation of treatment provides the best evidence on which to base antimicrobial selection, these publications acknowledge that antimicrobial drugs will always need to be used empirically in some patients [16, 18]. Empirical choice of antimicrobial drugs should be based on previous experience, knowledge of the agents most likely to be recovered from a particular species with disease in a particular organ system, and knowledge of local resistance patterns [16, 18]. Empirical use of antimicrobials is clearly justified in initial treatment protocols for rapidly progressive, life-threatening conditions such as systemic sepsis, while awaiting the results of culture and susceptibility tests on samples collected at admission. Historical information regarding the predominant bacterial isolates from septic foals and their susceptibility to antimicrobials in a particular geographic location, as presented in this study, is therefore important to guide the empirical selection of antimicrobials to initiate treatment of foal sepsis.

Based on the in vitro results of this study, the current first choice combination of amikacin with ampicillin remains an appropriate initial treatment for foal sepsis. Amikacin continues to have a high level of in vitro activity against Gram-negative isolates and Staphylococcus spp. Amikacin is strongly preferred over gentamicin, owing to the high level of in vitro resistance of several species of bacteria to gentamicin. Ampicillin remains highly active in vitro against β-haemolytic Streptococcus spp. and is one of the drugs with the highest in vitro activity against Enterococcus spp. (1998–2010: 70.7% susceptible).

The decrease in in vitro activity of ceftiofur against Enterobacteriaceae over time is of concern and suggests that ceftiofur will potentially be less effective for treating foals with sepsis caused by Gram-negative enteric organisms than was the case in years past.

Trimethoprim/sulfamethoxazole is not sufficiently active in vitro against many of the tested isolates to recommend its use to initiate treatment of sepsis in foals.

A high percentage of bacteria were found to be susceptible to imipenem in the current study; therefore, this drug might seem to be an attractive alternative for treatment of bacterial infections in foals. Use of imipenem should, however, be strictly limited to those cases in which infecting bacteria have been shown to be resistant to all feasible alternatives, or other antimicrobial drugs have proven to be ineffective.

It should be recognised that the findings of the current study may be unique to the hospital population at UC Davis. As noted above, climatic conditions and management factors, including the selection pressure imposed by antimicrobial use, will influence resistance patterns of bacteria in the local environment. Together with the knowledge that the susceptibility patterns of Enterobacteriaceae and Enterococcus spp., 2 of the most important groups of pathogens causing neonatal sepsis, are inherently unpredictable, the importance of determining the susceptibility of each individual isolate retrieved from a foal with sepsis cannot be overstated. Additionally, the observed changes in susceptibility patterns of groups of bacteria to antimicrobial drugs over time underscores the need for continuous monitoring, as well as judicious antimicrobial use.

Authors' declaration of interests

No competing interests have been declared.

Ethical animal research

Ethical review not required by this journal, it is a retrospective study based on clinical records.

Source of funding

This project was supported by the Center for Equine Health with funds provided by the State of California Pari-Mutuel Fund and contributions by private donors.

Acknowledgements

The authors would like to thank Spencer Jang, Eileen Samitz and Dr Barbara Byrne from the UC Davis Microbiology Laboratory for their contributions to this study and Drs Astrid Watzin, Kirsten Wroolie and Maria DeCarlo for the work they have put into this study over the years. The authors would also like to thank the clinicians, residents, students and nursing technicians at the William R. Pritchard VMTH at the UC Davis, USA, for their hard work and the excellent level of care they have provided to save hundreds of foals over the years.

Authorship

All authors made significant contributions to the completion of this study and have played an important role in drafting and/or revising the final article. All authors have approved the final version submitted for publication.

Manufacturers' addresses

  1. aBecton Dickinson and Co., Sparks, Maryland, USA.

  2. bWampole, Cranbook, New Jersey, USA.

  3. cPro-Lab Diagnostics, Richmond Hill, Ontario, Canada.

  4. dTrek Diagnostic Systems Inc., Cleveland, Ohio, USA.

  5. eCYTEL Software Corporation, Cambridge, Massachusetts, USA.

Ancillary