Three-week summer period prevalence of Clostridium difficile in farm animals in a temperate region of the United States (Ohio)

Abstract

The cross-sectional (period) prevalence of Clostridium difficile in 875 farm animals from 29 commercial operations during the summer of 2008 in Ohio, USA was quantified. Compared to an external referent population of intensively managed race horses (12.7%), intensively managed commercially mature food animals (poultry, cattle, swine; < 0.6%) were infrequent shedders of C. difficile (P < 0.00001) during the warmest weeks of 2008.

Clostridium difficile is an enteric pathogen that is present in the intestine of several animal species, causing sporadic intestinal disease due to the production of cell-disrupting exotoxins (A, B, and binary), among other virulence factors (1,2). Of public health concern, C. difficile infections (CDI) in humans have increased over the past decades and appear to be seasonal in northern temperate countries (1,3). It is estimated that at least 1 in every 4 to 5 outpatients admitted to hospitals with symptoms of CDI have no obvious exposure to established sources of C. difficile spores. These persons have no exposure to environments or individuals with direct/indirect physical connection to healthcare settings, where CDIs are common and where indoor environmental contamination with toxigenic strains is high. Because clones of toxigenic C. difficile that affect humans have been repeatedly isolated from foods and animals (also following a seasonal pattern with high prevalence in winter and low in summer) (2,4), there are concerns about exposure of susceptible humans to C. difficile via foods or recreational environments contaminated with feces of colonized animals (2,4,5).

Increasing our understanding of the ecology of this pathogen in environments outside hospitals is of importance, especially because whole genome sequencing recently indicated that most C. difficile strains isolated from patients in hospitals in the United Kingdom did not originate in their own hospital environment (6). Contrasting the traditional assumption of hospital origin for most CDIs based on fingerprint genotyping techniques [e.g., polymerase chain reaction (PCR) ribotyping, pulsed field gel electrophoresis], genome sequencing indicated the need to explore external sources of infection with toxigenic C. difficile (6). Little attention has been paid to the effect of climate on the prevalence of C. difficile in animals (2), especially in regions in which intense farming might facilitate amplification of toxigenic strains responsible for CDIs in humans.

Longitudinal studies on single farm animal species have shown that the prevalence of fecal shedding of C. difficile strains relevant for CDI in humans (e.g., PCR ribotypes 078 and 027) decreases with age, and seemingly with season as temperatures increase in North America (7). Despite the growing body of evidence on age and seasonal factors, there are no comparative studies quantifying the risk of shedding by various animal species in a single region and period. Determining which farm operations are more likely to rear animals that constantly harbor and amplify C. difficile across seasons may provide the basis to implement prevention measures to lower the load of animal-derived C. difficile spores in anthropogenic environments. Given that regional livestock density and farm residency are factors that intersect in agricultural lands, which is relevant for acquisition of enteric bacterial infections by humans, we conducted a cross-sectional study to determine if any particular farm animal species were preferential reservoirs for C. difficile in Ohio, USA in 2008, 2 years after mandatory reporting of CDIs was implemented in the state due to increased CDI frequency and public health concerns (8,9).

To determine if fecal shedding of C. difficile (toxigenic and nontoxigenic strains) in commercially mature farm animals was different across physiologically diverse animal species in a hot climate, we conducted this study i) across poultry, swine, cattle, and horses, ii) during summer when the shedding prevalence was expected to be low) (7), and iii) in a region of the Great Lakes (Ohio, USA). The climatic conditions in this area are similar to those of the region that served to document the effect of age and season on CDI in young dairy calves (southern Ontario, Canada; centroids in a northeast direction 229 km apart across Lake Erie) (7). Intensively managed race horses were the only non-food producing animal population tested. We tested non-toxigenic C. difficile because those strains may protect animals and humans against toxigenic strains (10,11).

The sampled farms were within a 150 km radius of our research center, and were selected based on production and management criteria and volunteer participation. Animals from specialized operations with modern production systems and moderate-to-large market share in Ohio were systematically sampled once. Clinically healthy race horses originating from a farm and residing in a major racetrack were also sampled once. During sampling (and the week before sampling) the average daily maximum air temperature was 26°C ± 3.2°C. As a referent for seasonality, the corresponding average for the coldest month that preceded sampling was 2.5°C ± 6.7°C, 6 months earlier (Wooster, Ohio; http://www.oardc.ohio-state.edu/newweather/).

Fecal samples were cultured to identify C. difficile as described (7,12). In short, samples were enriched for 7 d in C. difficile selective broth supplemented with sodium taurocholate, cycloserine, and cefoxitin, treated with ethanol and plated onto C. difficile fructose agar as described. Suspect isolates were tested for L-proline aminopeptidase activity and confirmed with detection of tpi and 16S rRNA genes, PCR testing of toxin genes (tcdA, tcdB⁺, cdtAB), and PCR ribotyping as described (7,12). Isolates were deemed toxigenic if they had 1 toxin gene and colonies produced the toxin detected by commercial ELISAs, or nontoxigenic if they lacked toxicity for Vero cell monolayers using dialysis tubing and cytotoxicity assay (13).

Binary and continuous data were analyzed using Fisher’s exact or chi-square statistics, and student t-tests, respectively. The sample size to verify a hypothetical prevalence of C. difficile of 5.6 for each animal species assuming simple-random sampling (one-sample, standard deviation = 6%, α = 0.05, power = 0.8) was set at 120 animals per species (13). To compensate for within farm clustering or transmissibility (14), we increased the sample size by 25% for a total of at least 150 animals for each animal species. Binary exact (Clopper-Pearson) confidence intervals are reported (Table 1). All analyses were conducted using STATA intercooled v12 (College Station, Texas, USA).

Table 1.

Three-week summer period prevalence of C. difficile in farm animals in a temperate region, Ohio, USA

	Farms, animals, n = (range, animals per farm)	C. difficile prevalence (exact 95% CI of prevalence)
		Toxigenic	Nontoxigenic
Poultry	11, 340 (min, 30; max, 33)	0/340	1/340
		0% (0, 1.08)a	0.29% (0.007, 1.62)
Swine	5, 150 (min, 30; max, 30)	1/150	0/150
		0.67% (0.017, 3.65)	0% (0, 2.43)a
Cattle	11, 330 (min, 30; max, 30)	2/330	0/330
		0.61% (0.007, 2.17)	0% (0, 1.11)a
Horses*b	2, 55 (min, 5; max 8)	7/55	4/55
		12.72%* (5.27, 24.48)	7.27%* (2.01, 17.58)
Total	29, 875 (min, 5; max, 33)	10/875	5/875
		1.14% (0.55, 2.09)	0.6% (0.185, 1.32)

^*Fisher’s exact P < 0.0001, horses compared to all other species.

^aOne-sided 97.5% confidence interval.

^bHorses sampled at a racetrack with animals originating from at least 8 farms that had resided for at least 1 month prior to sampling, and an independent horse farm. All positive samples corresponded to animals at the racetrack, except 1 toxigenic isolate from a healthy horse at the farm.

In total, we tested 875 adult (or deemed commercially mature and/or ready for harvest) animals raised in 29 farming operations (poultry, swine, cattle, horses) in Ohio, USA, during 3 of the warmest weeks of the year. Following established protocols (7,12), we tested systematically and randomly collected fresh fecal matter or rectal content (5 to 25 g) from 340 commercially modern poultry (Cornish cross broilers, 6 to 8 weeks old, n = 190; commercial turkeys 10 to 12 weeks old, n = 150), 150 finishing pigs (i.e., 16 to 20 weeks of age and 100 to 120 kg body weight), 330 cattle (lactating dairy, n = 150; finishing beef, n = 180), and 55 race horses (about 2 to 4 years old). All animals were reared in high densities. Information regarding diet and management was not collected at the time of sampling as they were deemed comparably balanced to achieve high standard levels of commercial feed conversions. With the exception of samples from horses (which were stabled, and fed hay and concentrated feed), most agar plates had few colonies for testing. When available, up to 4 colonies were tested per plate. The prevalence for C. difficile during the study period was 1.71% [15 of 875 animals, exact 95% confidence interval (CI) = 0.96, 2.81]. Toxigenic isolates were more frequent than nontoxigenic isolates across species (2:1), but the difference was not significant (Fisher’s exact P = 0.3). Concurrent toxigenic and nontoxigenic isolates were identified in only 1 of the positive animals (1/14: 7.1%; 95% CI: 0.18, 33.86), a horse (Table 1).

Toxigenic C. difficile was isolated from 1.14% (10/875) of animals, corresponding to 13.8% of operations (4/29; 95% CI: 3.89, 31.66). Three isolates were derived from food animals: a feedlot beef cow, a dairy cow, and a pig (0.6% of 480 animals). Comparatively, C. difficile was significantly more common in the feces of horses (12.7%, 7/55, Fisher’s exact P < 0.0001; Table 1). Polymerase chain reaction ribotyping indicated toxigenic isolates matched PCR ribotype 078 and 027 from previous publication (1 horse and 1 dairy cow, and 1 horse, respectively) (14) as it has been variably but regularly described in North America (1). Three other PCR ribotypes (A⁺B⁺cdtA⁺/classic-tcdC) did not match strains from our previous studies. Nontoxigenic C. difficile was isolated from 6.9% of operations (2/29; 95% CI: 0.84, 22.76). Further molecular testing of the isolates was not possible after a tornado severely damaged our research center.

Despite various longitudinal studies in animals, this is the first cross-sectional study of this nature in North America comparing multiple species of food animals and horses, sampled over a narrow period of time, in a geographically restricted region with relatively stable climatic conditions. A smaller study with 4.7 times fewer animals (broilers, pigs, and cattle, total n = 187) at abattoirs in Austria showed C. difficile prevalences between 3.3% to 5% in all species (15), which is higher than the upper limits of the 95% confidence intervals reported for our food animals (Table 1). Despite the similarities, the studies are not directly comparable as the Austrian study was conducted during a wider sampling period (March to July, 2008; 16 to 20 wk), with maximum daily air temperatures that differed widely (monthly average range for Vienna, Wien/Schwechat-Flug, from 10.8°C ± 3.71°C to 25.63°C ± 4.09°C; http://freemeteo.com) compared to our present study (26°C ± 3.2°C).

Here we showed that toxigenic and nontoxigenic C. difficile were rare in poultry, swine, and cattle (< 0.6), but common in horses (> 12%), which is in agreement with previous surveys (16). Predisposing factors for shedding in horses have been reported (17). Although horses are not harvested as a primary source of meat in North America, horses are commonly used for transportation and labor during production and harvest of vegetables in the agricultural states in the USA (18). The public health relevance of C. difficile in horses as carriers of C. difficile and the use of animal manure as crop fertilizer remain unclear. Intervention strategies regarding management of animal excrement might be beneficial considering that spores of C. difficile are long-lived within horse manure and soil (19).

Compared to intensively managed horses, collectively, this study indicates that food animals, especially poultry and swine (20,21) were unlikely reservoirs of C. difficile in Ohio, USA during summer in 2008. In contrast, recent studies in Europe continue to highlight that in some regions (22), food animals have variable and higher rates of C. difficile shedding at the time of harvest (up to 10% in cattle). Despite our growing understanding of the epidemiology of this pathogen in animals, it is still uncertain to what extent geography, climate, and season modulate the prevalence in the intestinal tract of animals and the environment, especially, when some clonal strains are found across continents. With seasonal dynamics for C. difficile infections in humans, it is advisable to include geographical and climatic referents (e.g., air temperature) in future studies, and records of local weather stations to further discern in the role of weather and climate on the environmental dynamics of C. difficile.