Clostridioides (Clostridium) difficile in neonatal foals and mares at a referral hospital

Abstract Background Understanding the epidemiology of Clostridium difficile is important for the development and assessment of infection prevention and control practices, as well as surveillance methods and interpretation of diagnostic testing results. Objective Our objective was to longitudinally evaluate C. difficile shedding in neonatal foals and mares admitted to a referral hospital neonatal intensive care unit. Animals Foals admitted to a neonatal intensive care unit, along with their dams. Methods Rectal swabs were collected from mares and foals at admission, and then approximately every 3 days, when possible. Selective culture for C. difficile was performed and isolates were characterized by toxin gene PCR and ribotyping. Results Clostridium difficile was isolated from 103/409 (25%) samples; 65/208 (31%) from foals and 38/201 (19%) from mares. Cumulatively, C. difficile was isolated from at least 1 sample from 50/113 (44%) foals and 30/97 (31%) mares. No association was found between hospitalization day and isolation of C. difficile (P = .13). Twenty‐three different ribotypes were identified, with ribotype 078 predominating. Fifteen foals had 2 positive samples during hospitalization. In only 6/15 (40%) foals was the same strain identified both times (5 ribotype 078 and 1 ribotype 012). Conclusions and Clinical Importance Clostridium difficile is an important pathogen in adult horses and foals, and our findings highlight the complexity surrounding the epidemiology of this opportunistic pathogen. It can be found commonly, transiently, and cluster within a facility in the absence of identifiable disease occurrences or clusters.


| INTRODUCTION
Clostridioides (Clostridium) difficile is a spore-forming bacterium that is an important cause of colitis in adult horses and foals. [1][2][3] In contrast to human medicine, where C. difficile infection (CDI) is diagnosed predominantly in older individuals in healthcare facilities, C. difficile and CDI are commonly found in horses outside of veterinary facilities. [4][5][6][7][8][9] However, the transmission dynamics of C. difficile in horses are poorly understood, both in veterinary facilities and on farms. Hospitalassociated (HA) CDI has been described in horses, 10

| Data analysis
Isolates were classified as community-associated (CA) if detected from samples collected at the time of admission and HA if first detected ≥48 hours after hospitalization, after an initial negative sample. 18 Source was not ascribed for positive samples collected after 48 hours of admission when an admission sample was not collected. Because of variation in sampling dates caused by logistical issues, the day of sampling was categorized as follows: admission, hospitalization 1 (H1; days 2-4), hospitalization 2 (H2; days 5-7), and hospitalization 3 (H3; days 8-10).
Chi-square or Fisher's exact tests were used for categorical comparisons. A P < .05 was considered significant.

| Study population
Ninety-seven mare and foal pairs, and an additional 16 unaccompanied foals were enrolled, corresponding to 210 horses (113 foals and 97 mares). All were thoroughbreds, apart from 2 mare-foal standardbred pairs. Horses were from 76 farms, with a median of 2 horses (typically a single mare-foal pair) admitted per farm (range, 1-8 Mares ranged from 2 to 21 years of age (median, 9 years; IQR, 7 years). The ages of 2 mares were unknown.
There were 209 admission samples, 175 from H1, 17 from H2, and 8 from H3. Two foals were sampled twice during H2 (on both days 3 and 4). In both cases, results from the days were the same (negative). The day 4 samples were removed from statistical analyses so that only 1 sample was present for that sampling timepoint, but both were retained for descriptive analyses. Similarly, 1 mare was sampled on both days 2 and 3. Both were positive, and the day 2 sample was removed from statistical analysis. Results by sampling period are presented in Figure 1. No association was found between sampling day (admission, H1, H2, H3) and isolation of C. difficile overall (mares and foals combined, P = .13), or specifically for mares (P = .26) or foals (P = .26). When individual sampling points were compared to admission, no differences were found for mares (P = .3-1.0).
A significant difference from baseline was found in foals at H2 Longitudinal results for mares and foals that ≥2 samples collected are presented in Table 2. Twenty mares were negative at admission alongside positive foals and had subsequent rectal swabs collected.
Four (20%) of those mares subsequently were positive, but only 1 of F I G U R E 2 C. difficile shedding status of 97 mare-foal pairs at admission to an equine referral hospital F I G U R E 3 Timing of isolation of C. difficile ribotypes that were detected at least 3 times in mares and foals at an equine hospital them shed the same strain (078) as the foal was shedding at admission.
Conversely, 7 foals that were negative at admission accompanied positive mares. Two (29%) of these foals subsequently were positive, but in both instances, the foal shed a different strain than the mare harbored at admission. In 1 pair, both the mare and foal harbored the same strain on day 3, but it was different from the strain the mare shed at admission.
Three mare and foal pairs both were shedding C. difficile at admission ( Figure 2). In only 1 of those was the same strain (ribotype 078) identified in both the mare and foal.
Some temporal clustering of ribotypes was apparent ( Figure 3).

| DISCUSSION
The rate of C. difficile shedding at the time of admission was high in our study; but the study population must be considered. Although limited clinical data are available, all foals were critically ill with suspected noninfectious disease, based on their admission to the intensive care unit (rather than the general ward or isolation unit).
Many could have received antimicrobials before referral and most would have been treated with antimicrobials during hospitalization.
Based on close contact between mares and foals, and the normal coprophagic behavior of foals, it is logical to assume there would be a strong influence on C. difficile transmission risk (especially mare to foal). One small earlier study using a less discriminatory typing method reported the same C. difficile strain in 36% of mare/foal pairs. 24 This finding is in contrast to our study where only 3% of pairs were shedding at admission and in only 1 pair was the same strain shed by both mare and foal. Apparent transmission from mare to foal, or vice versa, occurred during hospitalization but was uncommon. Although numerically higher, no significant difference was found in C. difficile acquisition of foals from infected mares vs mares from infected foals. Thus, the exposure risk of mares and foals does not seem to be strongly linked, and other endogenous or exogenous (eg, environmental contamination, human cross-contamination) factors may play important roles in the epidemiology of C. difficile in this rather unique subset of the population. Other potential factors such as severity of illness (and corresponding chance of coprophagia of the mare's feces or extent of contact with veterinary personnel) could not be analyzed.
Shedding was characterized as HA or CA using common definitions from human medicine. 18 Although we can be confident that positive results at admission from horses that were not recently hospitalized indicate community acquisition, we have less confidence in subsequent samples because of the potential for false-negative baseline cultures and lack of information about the incubation period or, more specifically, the time from oral exposure to shedding in adult horses and in foals. In an experimental study of foals, the meantime to first isolation of C. difficile was 48 hours, 25 and thus false-negative cultures are likely in a population of neonates that were exposed shortly after birth, before hospitalization. Therefore, although a standard approach to classification of HA and CA was used, this approach may overestimate HA rates and is a reason we have defined the term HA as "hospital-associated" rather than "hospital-acquired". Clustering of some strains strongly suggested transmission within the hospital.
However, because ribotype 078 dominated, clustering only could be detected with strains other than 078. More discriminatory approaches such as whole-genome sequencing would be needed to better assess acquisition within the hospital setting. The identified clusters all were relatively small and self-limiting (or controlled by routine infection control practices).
Changes in C. difficile shedding during hospitalization occurred.
Although the cumulative prevalence increased throughout hospitalization, there was limited apparent difference in point prevalence over time and acquisition or loss of C. difficile shedding did not appear to be associated with duration of hospitalization, apart from a significantly higher prevalence in foals at H2 vs admission. Whether this finding represents a true increased risk during the day 4-6 time period compared to other time periods requires further study. The sample sizes for post-admission could have impacted our ability to detect changes at other time points.
The term "shedding" has been used in our study vs "colonization." Although the term colonization is used commonly, it represents a state in which the bacterium is present and actively growing in the host. However, that conclusion is difficult to make, versus transient passage of the bacterium or metabolically inert bacterial spores.
Whereas detection of C. difficile in feces likely reflects at least transient colonization, shedding has been used as a more accurate description of what can be assessed.
Limited longitudinal study of horses has been performed to characterize shedding dynamics. However, a study of 25 healthy horses that involved monthly fecal sample collection reported a cumulative horse-level prevalence of 40% over the course of the year. 26 However, the sample level prevalence was only 5.5%, and only once was the same strain found in consecutive monthly samples in the same horse. Another study reported significantly higher recovery of Furthermore, the study population was mares and foals from breeding farms, and C. difficile transmission may differ on other types of farms.
Our study provides useful information about the dynamics of C. difficile, but whether results would be similar in different populations requires further investigation.
Clostridium difficile is an important pathogen in adult horses and foals, and our study emphasizes the complexity surrounding the epidemiology of this opportunistic pathogen. It can be commonly found, transiently present and cluster within a facility in the absence of identifiable disease occurrences or clusters. These features highlight the importance of routine infection control and antimicrobial stewardship to try and limit transmission of this pathogen. Our results contribute to a better understanding of the epidemiology of C. difficile in hospitals and interpretation of routine or outbreak surveillance.

ACKNOWLEDGMENT
Funding provided by Equine Guelph.