Cattle enrolled in this study were managed at four western Canadian feedlots in south central Alberta, under production conditions that are typical of those used at large commercial cattle feedlots throughout western Canada and the United States. Feedlots had one-time capacities between 15,000 and 20,000 animals, with pens capable of housing 50 to 350 animals. Animals are housed in open-air, dirt-floor pens arranged side by side with central feed alleys and 20% porosity wood-fence windbreaks. All feedlots have modern cattle handling facilities. Each animal handling facility has a hydraulic chute equipped with an individual animal scale, a chute-side computer with software for individual animal data collection and management (iFHMS, Feedlot Health Management Services Ltd., Okotoks, Alberta) and separation alleys to facilitate the return of animals to designated pens. All animal handling and sampling procedures were approved prior to the initiation of the study by the University of Guelph Animal Care Committee, the Animal Care Committee of Feedlot Health Management Services (FHMS) and the Institutional Animal Care and Use Committee of Colorado State University.
Candidate animals utilized in the study were procured through the auction market system across western Canada. Various cattle types were fed at these feedlots including cattle of various entry weights, age classes (calves and yearlings), frame sizes, sources (e.g., ranch-direct cattle and back-grounded cattle), and genders (bulls, steers and heifers). A 2-stage random sampling plan was used to determine which pens and animals within those pens were selected for enrolment. Animals were allocated to the study from January 17, 2009 to September 11, 2009. During the enrolment period, 30% of all new pens of cattle were randomly selected for inclusion in the study using a pen randomization table as the cattle arrived at the feedlot. Within each selected pen, 10% of all animals in that pen were then randomly enrolled in the study at initial processing using an individual animal randomization table. Cattle enrolled in the study weighed a mean of 375 kg (152 to 513 kg).
All animals enrolled in the study were subject to standardized animal health management and feedlot production procedures as per the protocols developed by the feedlot animal health/production consultants (FHMS). In brief, each animal received a unique individual animal identification ear tag, a trial-specific ear tag to help identify individuals for future sample collection, a subcutaneous hormonal growth implant in the middle third of the ear, vaccine(s) to immunize against selected bacteria and viruses that cause disease in feedlot cattle, and application of topical avermectin for internal and external parasite control. In animals at higher risk of developing disease, a parental antimicrobial was administered as part the prevention and control strategies for bovine respiratory disease. Water and standard mixed complete feedlot diets, formulated to meet or exceed the National Research Council nutritional requirements for beef cattle feedlot cattle, were offered ad libitum throughout the feeding period.
Individual animals enrolled in the trial were sampled twice over the course of the study: at the time of arrival and initial processing, and then again at various times in the middle of the feeding period when cattle were processed again to perform standard feedlot management procedures. Feces were collected from individual animals per rectum using a new palpation sleeve (#33, Almedic, Montreal, Canada) for collection and transfer (minimum 4 g) into a new sterile plastic fecal cup (# 109117, Globe Scientific, Paramus, New Jersey). Fecal samples were labelled, refrigerated (4°C) and transported to FHMS within 7 days of collection. At FHMS, samples were placed into ZipLoc bags and shipped (once a week in a chilled cooler, by air courier (Purolator Corporation) from Calgary, Alberta to the microbiology laboratory (University of Guelph, Guelph, Ontario) for further processing.
Each fecal sample collected over the course of the trial was assigned a unique identification number to ensure blinding of the laboratory staff and uniform labelling of samples. All treatments of cattle housed in the pens of cattle that were enrolled and sampled for the Individual antimicrobial use data were recorded at each feedlot over the course of the study using a chute-side computer system (iFHMS, FHMS). These data were available for each animal and included the product, the dose, the route and the number of days administered. Data on both individual animal and in-feed antimicrobial exposure were collected, with the in-feed data compiled from the pen-based feeding records. All study data were subsequently compiled, collated in a computer spreadsheet program (Microsoft Office Excel 2003), and verified.
Clostridium difficile culture
Approximately 2 g of feces were inoculated into 9 mL of C. difficile moxalactam norfloxacin (CDMN) enrichment broth (Oxoid Ltd; Nepean, ON Canada) containing 0.1% sodium taurocholate and incubated anaerobically at 37°C for 7 days. Two millilitres of broth were then added to 2 mL of anhydrous alcohol and incubated at room temperature for 60 min. After centrifugation (3,980 rcf for 10 min), the pellet was inoculated onto CDMN (Oxoid Ltd; Nepean, ON Canada) agar and incubated in an anaerobic chamber at 37°C for 48 h and, if negative, re-checked 3 days later. Isolation and identification of C. difficile was based on the characteristic morphology and odour of the colonies, Gram stain and the presence of the L-proline aminopeptidase activity (Remel Inc, Lenexa, KS, USA). One single colony for each isolate was subcultured and stored at -80°C and re-cultured prior to molecular analysis.
Clostridium difficile was grown on blood agar for 24 h and approximately 10 colonies were suspended in 1 mL of distilled water and centrifuged at 12,100 rcf for 1 min. The supernatant was discarded and 200 μL of a commercial DNA extraction kit (InstaGene Matrix; Bio-Rad, Richmond, CA, USA) were added and incubated at 56°C for 30 min and at 100°C for 8 min. The mixture was centrifuged and 200 μL of the supernatant were frozen at -20°C until processing.
Ribotyping was performed as described by Bidet et al. . Ribotype patterns were evaluated visually and compared to an internal library of ribotypes. The international numerical designation (e.g., ribotype 078) was used for bacterial strains recognized as a known international ribotype based on comparison with reference strains. A multiplex PCR was used for detection of genes encoding toxin A (tcdA) and toxin B (tcdB) as described by Lemee et al. . A second PCR was performed for detection of toxin A gene constitutive difference between A-/B+ strains and A+/B+ strains, and thus, identification of toxin A negative strains was performed according to Kato et al. . Detection of cdtA, the gene encoding for the enzymatic component of CDT, was performed according to Stubbs et al. . Sequence analysis of the tcdC gene was performed  and the result was classified according to Curry et al. .
Pulsed-field gel electrophoresis was performed following the protocol used by Miller et al.  with modifications. Briefly, C. difficile was cultured for 48 h and inoculated into 3 mL of pre-reduced brain and heart infusion (BHI) solution and grown anaerobically for 6 h. The solution was then adjusted to have an optical density (OD)600 between 0.3 and 0.7. Four hundred microlitres of the solution were centrifuged at 12,000 rcf for 1 min and the pellet suspended in 150 μL of cell lysis buffer and added 150 μL of melted 1% SeaKem Gold PFGE agarose (Cambrex BioScience Rockland Inc., ME, USA) plus 1% SDS for pipetting into plug molds. After solidified, plugs were transferred into 500 μL of lysis buffer added by 25 μL of lysozyme (final concentration 20 mg/mL) and 25 μL of mutanolysin (final concentration 12.5 U/mL) and incubated at 37°C overnight. The solution was then replaced by 500 μL of proteinase K (PK) buffer added by 25 μL of PK (final concentration 20 mg/mL) and incubated overnight at 56°C for 4 h. Plugs were rinsed three times with 1 × TE buffer and placed on a shaker for 5 min. The last step was repeated once and three more washes were performed with intervals of 10, 15 and 20 min. The solution was replaced by 150 μL of buffer A (New England Biolabs, ON, Canada) and after 10 min replaced by buffer A plus 60 U of restriction enzyme SmaI and incubated at 25°C overnight. Half of the plug was cut off and transferred to wells of a 1.3% pulsed-field certified agarose gel (BioRad, CA, USA). DNA separation was performed in a CHEF-DR II chamber (BioRad, CA, USA) added of 2.2 L of 0.5× TBE buffer plus 500 μL of 0.2 M thiourea, set to run for 22 h at 6 V/cm with initial switch time of 1 sec and final switch time of 40 sec. The gel was stained in ethidium bromide and images were obtained using a computerized system (SynGene, Synoptics, MD, USA).
Antimicrobial use data
Individual animal exposure data regarding antimicrobial drugs were recorded at each feedlot over the course of the study using a chute-side computer system (iFHMS, Okotoks, Alberta). These data included the product used, the dose, and the date and route of administration. All study data were subsequently compiled, collated in a computer spreadsheet and verified. Ionophores, and coccidiostats were not included in this analysis.
Dosage information for exposures to antimicrobial drugs was converted into an Animal Defined Daily Dose (ADD). The ADD metric represents the number of days of treatment for an animal based on an assumed average maintenance dosage needed for clinical therapy. Dosage conversion to ADD was based on the expected length of drug effect as indicated by approved dosages.
The least square means estimates and 95% confidence intervals for the prevalence of C. difficile at arrival and the second (post-arrival) time point were modelled using logistic regression. Regression analysis using generalized estimating equation (GEE) methods, was used to correct prevalence estimates for lack of independence related to sampling of multiple individuals from the same pens using a compound symmetry (exchangeable) correlation structure. Logistic regression was performed using commercially available software (SAS version 9.2, SAS Institute Inc, Cary, NC).
Date of sampling during the feeding period (< 62 days on feed (DOF), 62-71 DOF, or > 71DOF) and exposure of cattle to antimicrobial drugs. Antimicrobial exposure data were summarized as ADDs administered parenterally or in feed between arrival and second sampling, by class of antimicrobial drug, and by route of administration (beta lactams, macrolides, phenicols, quinolones, tetracyclines, and sulfonamides for parenteral exposures and tetracyclines for in-feed exposures).
The outcome for the logistic models was the presence or absence of C. difficile. Variables were screened in univariable models to determine those to be included in multivariable model building using a critical alpha for inclusion of 0.25. Multivariable models were not assessed since none of the antimicrobial exposure variables met the inclusion criteria. Odds ratios, 95% confidence intervals (95%CI), and the associated P-values were reported from logistic regression models.