Antimicrobial-associated diarrhoea in three equine referral practices


email:; Received: 27.07.11; Accepted: 03.05.12


Reasons for performing study: Although antimicrobial-associated diarrhoea (AAD) is the most frequently observed adverse effect of antimicrobial therapy in horses, few multicentred studies on the prevalence of AAD have been performed.

Objectives: To determine the prevalence of AAD in horses that developed diarrhoea after antimicrobial treatment for nondiarrhoeic conditions and identify the antimicrobials used.

Methods: The 2009 database of 3 referral hospitals was searched to identify nonhospitalised horses (weanling age or older) treated with antimicrobials for nongastrointestinal conditions. Horses with these criteria that presented with diarrhoea during 2009 were included in the study. Additional information, including antimicrobial administered and results of faecal pathogen testing, was gathered on each hospitalised case.

Results: Of the 5251 horses treated with antimicrobials for nongastrointestinal signs, 32 were diagnosed with probable AAD, a prevalence of 0.6% (95% confidence interval: 0.43–0.86%). The AAD-diagnosed horses had an 18.8% (6/32) mortality rate. Horses with AAD had been treated for an average of 4.2 days. The most frequently used antimicrobials in horses with AAD were gentamicin in combination with penicillin (n = 7), enrofloxacin (n = 7) and doxycycline (n = 4). Clostridium difficile was identified in faecal samples from 4 horses, 2 of which died and Salmonella from 3 horses.

Conclusions: Results indicated that the prevalence of AAD is low. Any antimicrobial class commonly used in equine practice is a potential cause of equine AAD. Other risk factors, such as opportunistic enteropathogens, may play a part in the development of diarrhoea secondary to antimicrobial usage.

Potential relevance: Although the risk of equine AAD is low, this sequela of antimicrobial treatment is possible especially when opportunistic enteropathogens or other risk factors are present. Because drugs from any antimicrobial class can be potentially involved in AAD, clinicians have additional incentive to ensure the judicious use of antimicrobial agents.


Antimicrobial-associated diarrhoea (AAD) is perhaps the most frequently observed adverse effect of antimicrobial therapy in horses. This is not surprising given the widespread use of antimicrobials to treat or pre-empt equine infectious disease. The clinical relevance of AAD is that it often prolongs hospitalisation, increases cost of treatment and, most importantly, results in a significantly greater mortality risk in affected horses [1]. For example, a 7-year study of 122 horses with acute diarrhoea admitted to a veterinary teaching hospital found that those with a history of antimicrobial treatment that preceded diarrhoea were 4.5 times less likely to survive [2]. In a European study, 8/8 horses died that were positive for β2-toxigenic Clostridium perfringens and treated with antimicrobials (penicillin and gentamicin) prior to onset of diarrhoea [3].

Because diarrhoea is a nonspecific condition with various aetiologies, a diagnosis of AAD is almost always presumptive. Antimicrobial-associated diarrhoea is often diagnosed by a temporal relationship between antimicrobial treatment in nondiarrhoeic horses followed by onset of acute colitis. The traditionally accepted pathogenesis of AAD is that antimicrobials alter the normal gut microflora and enteric environment [4]. In the horse, the enteric microflora is comprised of hundreds of different bacteria, of which the majority are anaerobes. When the microflora population is disrupted, normal metabolism of carbohydrates, volatile fatty acids and bile acids are affected, ultimately resulting in enhanced water secretion, reduced water absorption and diarrhoea. Also the disruption of the normal microflora can lead to proliferation of commensal enteropathogens, such as Clostridium difficile, Clostridium perfringens and Salmonella spp. [1,3,5,6]. In addition, some antimicrobials have a direct prokinetic effect on intestinal motility or result in induction of toxin production by commensal enteropathogens [7,8]. Almost all antimicrobials have been associated with AAD. One of the earliest published reports of AAD was with the administration of oxytetracycline to horses participating in an orthopaedic research study [9]. There are additional reports documenting diarrhoea after the administration of oxytetracycline or its contamination of sweet feed [10,11]. Accidental contamination of feed with lincomycin also has resulted in fatal colitis [12]. Erythromycin has been well documented to cause diarrhoea when administered orally to adults [13,14]. In a study by Baverud et al. 54% of horses receiving β-lactam antimicrobials for a nongastrointestinal disorder developed acute colitis [5]. Potentiated sulphonamides have been associated with diarrhoea but the risk is considered less than with other antimicrobials [15].

There are many studies looking at the incidence of diarrhoea associated with individual antimicrobials [1,5,8,10,13–15]. Based on these studies, the incidence varies from 3–52%. These studies included a small population size in which the frequency of diarrhoea was determined only for an individual antimicrobial. In addition, a majority of these horses were hospitalised, which adds additional stressors that could result in diarrhoea.

This cross-sectional study evaluated the prevalence of probable AAD in horses treated with antimicrobials for nongastrointestinal conditions during a one year period. Additional information was gathered on horses hospitalised with diarrhoea and a history of antimicrobial treatment for a nongastrointestinal condition. Because the number of systematic studies on equine AAD is small, there is a relative lack of objective data about this potential side effect of antimicrobial therapy. Due to the widespread use of antimicrobial agents, a larger study of nonhospitalised patients needs to be performed. Thus, our results and conclusions may contribute to clinicians' understanding of AAD and help guide their approach to antimicrobial treatment of horses.

Materials and methods

Three equine referral hospitals located in Kentucky, Florida and New Jersey participated in the study. All horses of weanling age (nonsuckling) or older, that presented to these practices during 2009 with diarrhoea and a history of antimicrobial administration, were included. A detailed history was obtained for each horse including any antimicrobials used, dosage and treatment dates. Faecal samples were collected at admission and tested in-house using commercially available ELISAs for Clostridium difficile toxins A and B (Premier Toxins A & B)a and Clostridium perfringens enterotoxin (C. perfringens enterotoxin test)b as previously described [16,17]. Faecal samples collected during the first 5 days of hospitalisation were cultured in-house for Salmonella spp. using standard microbiological techniques. If positive for Salmonella spp. isolates were evaluated for antimicrobial sensitivity. Presumptive AAD was diagnosed when antimicrobials were used to treat a nongastrointestinal condition in a horse that subsequently developed diarrhoea during treatment. Duration of diarrhoea was categorised as <1 day, 1–3 days or >3 days. To determine overall antimicrobial usage for each hospital, the hospital's 2009 database was searched and all nonhospitalised horses of weanling age (nonsuckling) or older, treated with antimicrobial agents for nongastrointestinal conditions, were counted. Confidence intervals (CIs) were calculated using Proc Freq of SAS v9.1c.


A total of 5251 nonhospitalised horses were treated with antimicrobials for nongastrointestinal issues during 2009 at the 3 participating referral practices. The prevalence of AAD was 0.6% (32/5251; 95% CI: 0.43–0.86%). The antimicrobial administered, duration of diarrhoea, case outcome and isolation of C. difficile toxins, C. perfringens enterotoxins and Salmonella spp. in the 32 affected horses are shown in Item S1. None of the horses had a prior history of diarrhoea within the preceding 6 months. Their ages ranged from 4 months to 28 years (mean, 3.8 years). The breed most commonly represented was Thoroughbred (24/32). The most common indication for antimicrobial treatment was respiratory infection (15/32). Fourteen (43.8%) horses were treated with more than one antimicrobial. Enrofloxacin was administered as monotherapy in 7 cases of AAD, doxycycline in 4 cases and penicillin-gentamicin combination therapy in 7 cases (Item S1).

Six horses died or were subjected to euthanasia as a result of AAD, an 18.8% mortality rate. Doxycycline was the antimicrobial agent in 3 of the deaths, enrofloxacin in 2 others and potentiated sulphonamides in one. Duration of diarrhoea tended to be relatively brief, lasting ≤1 day in 81.3% (26/32) of horses. Faecal composition tended to be watery to cow pie. All of the horses except 3 were pyrexic (≥38.9°C). Complications during hospitalisation included laminitis in 3 horses and colic in 3 horses. One horse each with colic and laminitis also died.

All classes of antimicrobial agents were represented in the cases that developed AAD. The average duration of antimicrobial treatment for the cases that developed diarrhoea was 4.2 days (range 1–12 days). Table 1 lists the antimicrobials administered over the one year period associated with diarrhoea. Antimicrobial-associated diarrhoea prevalence was higher in horses treated with enrofloxacin 5.4% (7/129) and penicillin/gentamicin combination 3.2% (7/222). Table 2 lists the 5 most commonly administered antimicrobials at each of the 3 referral practices. Collectively, the most frequently used agents were oxytetracycline (n = 1243), potentiated sulphonamides (n = 898), gentamicin (n = 828), penicillin (n = 628) and doxycycline (n = 453).

Table 1. Antimicrobials administered over the one year period associated with diarrhoea
AntimicrobialNo. horses treatedNo. AAD cases95% confidence interval (CI)
Low 95% CI% prevalenceUpper 95% CI
  1. AZI = azithromycin; CEF = ceftiofur; CHL = chloramphenicol; DOX = doxycycline; ENR = enrofloxacin; GEN = gentamicin; MET = metronidazole; PEN = penicillin; RIF = rifampin; POTs = potentiated sulphonamides. **Unable to determine prevalence and confidence interval.

ENR+PEN11 ** ** **
POTs+MET11 ** ** **
Total 32   
Table 2. Antimicrobials used by frequency and practice location
AntimicrobialLocation and No. horses treated
Florida (n = 2854)Kentucky (n = 2042)New Jersey (n = 355)
Potentiated sulphonamides40143463

Clostridium difficile toxins A and B were identified in faecal samples from 4 horses and Salmonella spp. isolated from 3 horses (Item S1). Clostridium perfringens enterotoxin was not detected in any horse. Two of 6 horses that died were diagnosed with C. difficile toxins A and B; all other mortality cases were negative for clostridial or Salmonella infection.


This is the first study to determine prevalence of AAD in a large population of nonhospitalised horses. In previous studies the population was small and the patients were also hospitalised [9,10,13,15]. Additionally, most of these previous studies determined the incidence of AAD in horses treated with one or 2 antimicrobial agents whereas our study looked at a wide range of antimicrobials commonly used by equine practitioners. The overall prevalence of AAD in our study was low. Other studies involving relatively small numbers of horses reported higher rates of AAD. For example, Baverud et al. reported that 25 of 46 horses developed diarrhoea after administration of a β-lactam antimicrobial agent alone or in combination with an aminoglycoside [5]. The occurrence of diarrhoea in hospitalised patients after the administration of potentiated sulphonamides was determined to be 3% [15]. Hospitalisation appears to add additional stressors or nosocomial risks that may precipitate or exacerbate AAD.

Horses affected by AAD in our study were treated with a wide range of antimicrobial agents, supporting that some risk of AAD accompanies the use of any antimicrobial in equine medicine. Alteration of resident intestinal flora results in a disturbance of intestinal metabolism of carbohydrates, volatile fatty acids and bile acids, leading to osmotic diarrhoea. This is because various co-factors, such as carriage of enteropathogens, transportation stress or an antimicrobial's direct prokinetic effect, can initiate or exacerbate AAD and even low-risk antimicrobials can be implicated. It has been proposed that antimicrobials with poor activity against anaerobes, which predominate in the equine intestinal flora, are less likely to cause AAD because they tend to avoid altering the anaerobic bacterial populations in the intestinal tract [18]. This would suggest that antimicrobials with limited activity against anaerobes, such as potentiated sulphonamides, fluoroquinolones and aminoglycosides would be less likely to induce AAD [5,18,19]. In our study, enrofloxacin administration resulted in 7 cases of AAD. The overall prevalence of AAD with enrofloxacin administration was one of the highest for individual antimicrobials (5.4%, 7/129 cases). There are no other reports documenting diarrhoea associated with enrofloxacin administration, although Canadian investigators reported multiple colitis cases associated with the administration of ciprofloxacin [20].

Oxytetracycline, which has had an historic association with AAD, was not associated with diarrhoea in our study, although it was the most commonly administered antimicrobial (1243 cases). In earlier reports of colitis associated with oxytetracycline, the agent was administered at a high dose (40 mg/kg bwt i.v.) [9]. Later reports of colitis following administration of therapeutic doses involved horses that were anaesthetised [10]. Doxycycline was administered as monotherapy to 453 horses, with 4 developing AAD (0.88%). At the suggested therapeutic dose (10 mg/kg bwt orally every 12 h), diarrhoea has not been previously documented with the administration of doxycycline. The only report in the literature of diarrhoea associated with doxycycline administration is when the drug was administered at a higher dose (20 mg/kg bwt orally) [21]. In this study, the dose of doxycycline administered that resulted in diarrhoea was 7.5–10 mg/kg bwt orally every 12 h.

The incidence of potentiated sulphonamides resulting in diarrhoea was low (0.11%, 1/898). This is probably due to their limited activity against anaerobes and being well absorbed from the small intestine. Antimicrobial-associated diarrhoea incidence for potentiated sulphonamides in our study was much lower than others have reported [15,18].

Ceftiofur, which is often anecdotally associated with AAD, was administered to a small number of horses (n = 104), only one of which developed diarrhoea lasting less than one day. The fact that a relatively small number of horses received ceftiofur is probably due to the anecdotal reports of its association with AAD. Foreman reported that horses administered ceftiofur perioperatively developed diarrhoea but the drug was administered at twice the labelled dose frequency (2 mg/kg bwt i.m. every 12 h) [22]. In developmental studies, ceftiofur was given safely i.m. once a day at up to 5 times the therapeutic dose [23]. Interestingly, the ceftiofur-treated horse in our study that developed diarrhoea was given 2.2 mg/kg bwt i.m. twice a day. This is twice the frequency of the labelled dose.

Penicillin was administered to 628 horses as monotherapy with no reported cases of diarrhoea. Gentamicin administered as monotherapy resulted in 2 cases of diarrhoea (0.24%, 2/828). However, horses treated with gentamicin and penicillin had the second highest prevalence of diarrhoea (3.2%, 7/222). Two of the 3 patients with AAD positive for Salmonella spp. had been treated with a combination of penicillin and gentamicin. The second most commonly administered combination of antimicrobials was gentamicin and potentiated sulphonamides. The prevalence of diarrhoea with this combination was 2.6% (2/78). Combining antimicrobial agents may increase the risk of AAD by inducing a broader change in intestinal flora resulting in overgrowth of more resistant Gram-negative organisms.

In human medicine, there is convincing evidence that enteric colonisation with C. difficile is a major contributing factor and the most important enteropathogen in cases of acute colitis during antimicrobial treatment [24,25]. The pathogenesis of C. difficile, including expression of tissue-degrading enzymes and enterotoxins A and B, has been well characterised in cases where colonisation resistance is disrupted by antimicrobial treatment [26]. In one equine study, C. difficile or its cytotoxin were detected in 40% (10/25) of adult horses with acute colitis that were previously treated with antimicrobials for conditions other than diarrhoea [5]. Notably, the nondiarrhoeic horses treated with antimicrobials in the same study were C. difficile-negative. In a later study, the same group found that 42.9% (18/42) of horses that developed acute colitis during antimicrobial treatment were also positive for C. difficile[27]. In our study, 2 of 6 horses that died were positive for C. difficile toxins, suggesting that this pathogen had an exacerbating effect in cases of AAD. In a prospective study, Weese et al. found that increased mortality in adult horses with diarrhoea was associated with faecal isolation of C. difficile[17].

Clostridium perfringens enterotoxin was not identified in horses with AAD in our study. This is unusual because previous reports found 12–19% of adult horses with diarrhoea were positive for C. perfringens enterotoxin [28,29]. Reasons for possible negative samples include lack of production of toxins in vivo, degradation of toxin during sample handling, dilution of the toxin in a liquid sample or administration of metronidazole prior to sampling. To overcome this, consecutive samples could have been obtained. It is known that C. perfringens produces other toxins so it is possible that other exotoxins or enterotoxins could have produced clinical disease. Clostridial organisms are more commonly associated with diarrhoea secondary to antimicrobial administration but Salmonella spp. have also been reported [1,13,14]. In our study, Salmonella spp. were isolated from 3 horses (9.4%, 3/32), all of which survived. Owen et al. identified antimicrobial administration as a risk factor for faecal shedding of Salmonella spp. and development of clinical salmonellosis [30].

The prevalence of AAD did differ among the 3 referral hospitals, ranging from 2.8% (10/355) in New Jersey to 0.7% (14/2042) in Kentucky and 0.3% (8/2854) in Florida. No previous studies have identified geographical or regional differences in the susceptibility of horses to AAD. Geographical differences reflect differences in the management of horses such as housing practices (pasture vs. stalled) and feeding regimes. A difference in the enteric flora probably exists due to feeding regimes, soil composition and administration of concurrent medications

Pathogenicity of the enteropathogens involved may vary depending on their antimicrobial resistance and virulence factors. The development of antimicrobial resistance is best illustrated by numerous nosocomial outbreaks involving multidrug resistant Salmonella spp. [6,31,32]. Variations in the usage of antimicrobials may also account for the difference in the geographical incidence of AAD. Veterinarians often chose antimicrobials based on ease of administration and anecdotal experience with a particular agent or drug class. The most common antimicrobial administered in each region differed (Table 2). The differences in antimicrobial administration may reflect the type of horse industry in each state. Florida and Kentucky have a higher resident horse population, specifically breeding farms, than New Jersey. Overall, New Jersey had the highest prevalence of AAD, but the number of horses receiving antimicrobials was lower than the other 2 states.

A weakness of this study is that the frequency of antimicrobial usage was determined based on the hospital's field/ambulatory veterinarians antimicrobial usage. Several of the horses referred for diarrhoea post antimicrobial administration were from veterinarians outside the practice. Obtaining details of antimicrobial usage from the outside veterinarians would have strengthened the study and made the prevalence of AAD even smaller. There are 2 antimicrobial combinations (enrofloxacin+penicillin and potentiated sulphonamides+metronidazole) recorded in Table 1 that had been administered once and resulted in diarrhoea. The antimicrobial usage from referring veterinarians would have strengthened the CIs for some of the antimicrobials (Table 1). Overall, the purpose of this paper was to determine the prevalence of AAD in a large group of nonhospitalised patients. This study brings to light the fact that the occurrence of AAD is low and sets the ground for future research into the association of antimicrobials and diarrhoea.

The prevalence of AAD in our study was low and does not minimise the importance of appropriate use of antimicrobials. The American College of Veterinary Internal Medicine has released a consensus statement with guidelines for the prudent use of antimicrobial agents [33]. Antimicrobials should be administered only when there is a known or suspected infection. Prophylactic administration is appropriate in certain instances such as in patients with compromised immune function or contamination of a surgical site. Antimicrobial selection should be based on culture and sensitivity results. Consideration should be given to the ability of the antimicrobial to reach the infected site and its efficacy against the identified or suspected bacterium. Accepted antimicrobial dosages and treatment regimes should be used. Inappropriate use of antimicrobials increases the chances of resistance.

Results of our study indicate that agents from any antimicrobial class can potentially result in AAD in horses. Combining antimicrobial agents may increase the likelihood of AAD. The aetiology of AAD is multi-factorial and likely involves several possible synergistic mechanisms. Prudent antimicrobial selection, dose and treatment duration are important to ensure treatment efficacy and to reduce the risk of AAD.

Authors' declaration of interests

No conflicts of interest have been declared.

Source of funding

The project was funded by Pfizer Animal Health.


The authors acknowledge the contribution of Mark Dana of Scientific Communications Services in the writing and editing of the manuscript.

Manufacturers' addresses

a Meridian Bioscience Inc, Cincinnati, Ohio, USA.

b TechLab Inc Blacksburg, Virginia, USA.

c SAS Institute Inc, Cary, North Carolina, USA.