|Home | About | Journals | Submit | Contact Us | Français|
Food safety risks due to Escherichia coli O157:H7 may be affected by variability in prevalence in or on live cattle at slaughter. Our objectives were to assess the prevalence and risk factors associated with E. coli O157:H7 in feedlot pens immediately prior to slaughter, and assess relationships among methods of monitoring the E. coli O157:H7 status of pre-harvest pens. We studied 84 pens containing a total of nearly 27 000 head of cattle in commercial feedlots in Alberta during 2003 and 2004. Sampling devices (ROPES) prepared from manila ropes were used to detect high prevalence pens. Forty of 84 pens (48%) were classified ROPES-positive. Within pens, fecal prevalence ranged between 0% to 80% (median = 20%) and the hide prevalence ranged between 0% and 30% (median = 0%). Pens that were ROPES-positive had a higher median prevalence for feces (40%) and for hides (3.8%) than those that were ROPES-negative (13.3% and 0%, respectively). The prevalence of E. coli O157:H7 in pens immediately prior to slaughter was found to be quite high or very low even within feedlots and seasons. Factors such as sampling month, temperature, precipitation, pen floor conditions, and water tank cleanliness were associated with E. coli O157:H7 outcome measures, although associated factors were not completely consistent among years and outcome measures. Fecal and hide prevalence are considered primary pre-harvest indicators of potential carcass contamination, but other methods such as ROPES that are associated with these outcomes may provide logistic advantages to efficiently classify pens of cattle as high or low risk to food safety.
En salubrité alimentaire, les risques associés à Escherichia coli O157:H7 peuvent être influencés par la variabilité de la prévalence de cet agent chez ou sur les animaux vivants au moment de l’abattage. Nos objectifs étaient d’évaluer la prévalence et les facteurs de risque associés à E. coli O157:H7 dans des enclos de parcs d’engraissement avant l’abattage, et évaluer les associations parmi les méthodes de surveillance du statut des enclos en regard de E. coli O157:H7. L’étude menée en 2003 et 2004 a porté sur 84 enclos contenant un total de près de 27 000 têtes de bétail dans des parcs d’engraissement en Alberta. L’outil d’échantillonnage (ROPES), préparé à partir de cordage en manille, a été utilisé pour identifier les parcs avec une prévalence élevée. Quarante des 84 parcs (48 %) ont été classés ROPES-positifs. À l’intérieur d’un parc la prévalence fécale variait entre 0 et 80 % (médiane = 20 %) et la prévalence sur la peau variait entre 0 et 30 % (médiane = 0 %). Les parcs qui étaient ROPES-positifs présentaient une prévalence médiane plus élevée pour les fèces (40 %) et pour la peau (3,8 %) que ceux qui étaient ROPES-négatifs (13,3 % et 0 %, respectivement). La prévalence de E. coli O157:H7 dans les parcs immédiatement avant l’abattage était relativement élevée ou très basse et ce à l’intérieur d’un même parc d’engraissement ou des saisons. Des facteurs tels que le mois d’échantillonnage, la température, les précipitations, l’état du plancher des enclos, et la propreté du réservoir à eau étaient associés avec les critères d’évaluation bien que les facteurs associés n’étaient pas complètement constants entre les années et les critères d’évaluation. La prévalence dans les fèces et sur la peau sont considérées comme des indicateurs pré-récolte primaires de la contamination potentielle des carcasses, mais d’autres méthodes telles que ROPES qui sont associées avec ces critères peuvent fournir des avantages logistiques pour classifier de manière efficace les enclos à bovins comme présentant un risque élevé ou un risque faible pour la salubrité alimentaire.
(Traduit par Docteur Serge Messier)
Escherichia coli serotype O157:H7 is a major public health and food safety concern in North America and other parts of the world. Cattle are considered a primary source of these bacteria and the consumption of food and water contaminated with cattle feces, as well as direct or indirect contact with live cattle, are considered major routes of human infection. The North American beef industry has made significant changes in the processing environment in order to reduce beef contamination and improve beef safety, and recent evidence suggests that in-plant processing procedures and interventions are effective at reducing levels of bacteria that are found pre-harvest (1–3). The feces and hides of cattle are considered major sources of contamination during beef processing, and there is a correlation between fecal and hide prevalence of E. coli O157:H7 pre-harvest and post-harvest carcass contamination (1). Furthermore, simulation data indicate that decreasing fecal prevalence and altering the slaughter order based on the E. coli O157:H7 status of cattle may be useful in decreasing carcass contamination (4).
If cattle pens with a high prevalence of E. coli O157:H7 could be identified accurately and efficiently prior to slaughter, it has been suggested that pre- and/or post-harvest interventions could be targeted toward cattle that may pose a higher risk to food safety and public health (5,6). In addition, such a pen-testing strategy could be used to evaluate, or monitor, the effectiveness of E. coli O157:H7 control procedures in commercial cattle feeding systems. The pre-harvest fecal prevalence of E. coli O157:H7 within a pen or group of feedlot cattle can vary widely (0% to > 80%) among pens and over time (6–8). Cattle that shed at high levels and for increased periods of time may be critical in the spread and maintenance of E. coli O157:H7, and may result in an increased prevalence of fecal shedding and hide contamination as well as an increased risk for carcass contamination (9,10). Several strategies have been proposed for identifying pens of feedlot cattle that may pose a higher risk to food safety, including using individual fecal, oral or hide samples, composite fecal samples, and manila ropes placed within pens (1,5,11). Potentially, groups of cattle with a high prevalence of E. coli O157:H7 could be slaughtered at the end of the shift before cleaning, targeted for potential interventions to decrease shedding and hide contamination, or otherwise managed to mitigate potential risk. Our objectives were to assess the prevalence and risk factors associated with E. coli O157:H7 in feedlot pens immediately prior to slaughter, and to assess the relationship among methods of monitoring the E. coli O157:H7 status of pre-harvest pens in Alberta feedlots.
Cross-sectional studies of cattle pens in commercial feedlots in Alberta were used to assess the E. coli O157:H7 status immediately prior to slaughter. Feedlot sampling was conducted in 2 study periods during 2 consecutive years. Period 1 was conducted from July to December 2003, and Period 2 was conducted from April to November 2004. Period 1 assessed the E. coli O157:H7 status of pre-slaughter feedlot pens in 4 different commercial feedlots using rope-devices placed within pens for cattle to lick, rub, or chew (ROPES); and freshly voided fecal samples collected from the pen floor (5,12). Period 2 assessed E. coli O157:H7 status of pre-slaughter feedlot pens in one commercial feedlot using ROPES and hide samples. Feedlots were selected based on the owners’ willingness to participate and the availability of pens for sampling immediately prior to slaughter.
Data on cattle, pen, and environmental characteristics were collected for all pens when ROPES were collected. The number of days on feed, number of cattle in the pen, and predominant cattle type within the pen (steer or heifer; and British, continental, or mixed/other breed) were collected from the feedlots. The conditions of the pen floor, water tank, and cattle hides (tag) were observed and recorded as scores following predetermined protocols and under the direction of one technician. Pen floor scores were based on the conditions observed over most of the pen, where: 5 = very wet and muddy, 4 = mostly muddy, 3 = damp/moist, 2 = neither damp or dry, and 1 = dry and dusty. Water tank conditions (degree of cleanliness) were scored as: 4 = dirty (bottom covered with sediment, water murky), 3 = moderate (feed debris and sediment, water cloudy), 2 = mild (some debris, water mostly clear), and 1 = clean (very little debris, water clear). Tag scores (amount of mud/manure accumulating on the hide of most of the cattle) were: 4 = heavy (at or above the hocks and on sides), 3 = moderate (below the hocks and on the belly), 2 = mild (below the hocks on some), 1 = none (very little tag on any cattle). Data on the daily and weekly mean high temperature, low temperature, and total precipitation were retrieved from Environment Canada for the nearest local weather station (http://www.weatheroffice.ec.gc.ca/canada_e.html).
All samples were stored in a cooler with frozen ice packs, shipped by overnight courier to the Veterinary Microbiology Laboratory of the Food Safety Division of Alberta Agriculture and Food, and processed within 24 to 48 h after collection. Samples were cultured to identify E. coli O157:H7 and colonies were confirmed as E. coli O157:H7 by serology and PCR.
Using a ROPES protocol previously described, all pens were sampled for E. coli O157:H7 using manila ropes as sampling devices (5,12). In brief, 7 sampling devices were hung in the study pens at least 1 h prior to sundown on the evening before the scheduled day of slaughter. Devices were attached to pen fences and along feed bunks in locations where cattle within the pen (but not in neighboring pens) could readily access the devices orally and by hide contact. The following morning the devices were collected aseptically and placed in individual, sterile Whir-pack bags to which we added 50 mL of Brilliant Green Broth (BGB) 2% (40 g/L; Difco Laboratories, Becton Dickinson Microbiology Systems, Sparks, Maryland, USA). Pens were defined as ROPES-positive when one or more of the rope samples were culture-positive for E. coli O157:H7 (5,12).
During Study Period 1, samples of freshly voided feces were collected from individual fecal pats on the pen floor on the day study pens were scheduled to be shipped for slaughter and ROPES were collected. Plastic spoons were used to place approximately 50 g of feces from each sampled fecal pat into 60 mL sterile specimen containers. Clean spoons, latex gloves, and aseptic techniques were used to minimize the potential for cross contamination. Due to the variability in pen sizes, 30 fecal samples were collected from pens with approximately 300 cattle or more, 20 samples from pens with 150 to 300 cattle, and 15 samples from pens with < 150 cattle. Pens were considered E. coli O157:H7 positive when at least one fecal sample cultured positive.
During Period 2, hide swabs were collected from 40 cattle per pen on the day that cattle were being shipped for slaughter and ROPES were collected. Every 3rd to 5th animal, depending on pen size, was restrained and sampled. Two 3 × 3″ sterile gauze pads moistened with sterile water (15 mL) were used to vigorously swab an area of approximately 450 cm2 at the dorsal midline between the shoulders (withers). Each hide swab was placed in a 60 mL sterile container. Again, aseptic techniques were used to minimize potential cross contamination. Pens were considered hide-positive when at least one hide sample cultured positive for O157:H7.
The ROPES samples were cultured using methods similar to those previously described (5,12). Briefly, an additional 250 mL of BGB 2% (prewarmed to 37°C) were added to each Whirlpak, which were then manually mixed for approximately 1 min. Following incubation for 6 h at 37°C, immunomagnetic separation (IMS) was performed using a programmable KingFisher mL (KF) automated magnetic particle processor designed for automated transfer and processing of magnetic particles in a 1-mL tube scale (Thermo Electron Corporation, Vanii, Finland). The KF was programmed according to times and procedures outlined for the manual IMS method. One milliliter of sample broth and 20 μL of anti-E. coli O157 immunomagnetic beads (Dynal Biotech, Oslo, Norway) were mixed for 30 min and the beads-bacteria complex was washed 3 times with 1 mL of phosphate buffered saline containing 0.05% Tween 20 (PBS/Tween). Beads were resuspended in 100 μL of PBS/Tween following the final wash. Fifty microliters of final suspension were plated onto a Sorbitol-MacConkey Agar with Cefixime (0.05 mg/L) and Tellurite (2.5 mg/L) (CT-SMAC), streaked for isolation and incubated at 35°C for 18 to 24 h. Individual nonsorbitol fermenting colonies (3–5) were subcultured to a blood agar (BAP), MacConkey agar (MAC), and CT-SMAC plates. After incubation (35°C for 18 to 24 h), one colony from BAP was further identified by biochemical testing with the IMViC test and by E. coli O157 agglutination testing by following manufacturer’s instructions (both Difco). Absence of β-glucuronidase activity was confirmed by culture on Fluorocult agar (Merck, KgaA, Darmstadt, Germany). Colonies presumptively classified as E. coli O157:H7 based on morphology, biochemical, and serological characterization, were confirmed as E. coli O157:H7 by PCR using corresponding colonies from MAC plates and the procedures described in the following text. Up to 3 colonies per sample were stored at −70°C for further characterization.
Fecal samples were mixed thoroughly using a sterile tongue depressor. A 10% suspension was prepared by mixing 10 g of fecal material in 90 mL of pre-enrichment E. coli broth (Difco) containing 1.0% novobiocin (Sigma-Aldrich, St. Louis, Missouri, USA) (mEC). The suspension was incubated at 41.5°C for 6 h in a shaker incubator then refrigerated overnight. Subsequently, 1 mL of mEC broth was used in the IMS protocol and then isolation and identification of E. coli O157:H7 was carried out as described in the ROPES procedure.
Hide samples were enriched using methods described by others (13). Twenty milliliters of prewarmed (37°C), concentrated BGB (60g/L) were added to sample containers, which were then mixed thoroughly, incubated at 37°C for 6 h, and refrigerated overnight. The IMS protocol and all other isolation and identification procedures for E. coli O157:H7 were performed as described for ROPES.
Six to 8 colonies were removed from MAC plates and dispersed in 200 uL of Ultra-PrepMan® Sample Preparation Reagent (Applied Biosystems, Foster City, California, USA) using an inoculating needle. After vortexing for 5 s and centrifuging briefly, samples were placed in a boiling waterbath for 10 min and then cooled at room temperature for 15 min. Samples were vortexed for 10 s and centrifuged at 13 000 rpm for 3 min at 4°C. A 1:10 dilution was made by adding 50 μL of supernatant to 450 μL of 12 mM Tris buffer, pH 8.0. Diluted deoxyribonucleic acid (DNA) was stored at 2 to 8°C until polymerase chain reaction (PCR) was performed. A multiplex PCR method with modifications was used to confirm isolates as E. coli O157:H7 by detecting virulence genes for Shiga toxins (stx1 and stx2), intimin (eaeA), and H7 (fliC) as previously described (14). Primers (DNA Lab, Biochemistry, University of Alberta, Edmonton) were as follows: VT3-5′CCATGACAACGGACAGCAGTT 3′ and VT4-5′CCTGTCAACTGAGCACTTTG 3′ (stx2); VT5-5′CATTGTCTGGT GACAGTAGCT 3′ and VT6-5′CCCGTAATTTGCGCACTGAG 3′ (stx1); EC1-5′GCGCTGTCGAGTTCTATCGAGC 3′ and EC2-5′CAACGGTGA CTTTATCG CCATTCC 3′ (fliC); EC3-5′CAGGTCGTCGTGTCTG CTAAA 3′ and EC4-5′TCAGCGTGG TTGGATCAACCT 3′ (eaeO157). Escherichia coli O157:H7 strain AP0087-10 (Agri-Food Laboratories Branch, Alberta Agriculture and Food, Edmonton) was used as a positive control and water as a negative control in all assays. Ten μL of diluted DNA was added to 40 μL of master mix consisting of 1X PCR buffer, 2.0 mM magnesium chloride (MgCl2), 0.2 mM each deoxyribonucleotide triphosphate (dNTP), 0.2 μM each primer and 0.02 U/μL Platinum Taq polymerase. All reagents were from Invitrogen, Mississauga. Amplification parameters consisted of preheating at 94°C for 3 min followed by 35 cycles of 94°C for 1 min; 65°C for 1 min; 72°C for 2 min and a final extension of 72°C for 5 min. Fifteen microlitre aliquots of PCR product were run on a 1.2% agarose gel stained with ethidium bromide. A 100 base pair (bp) marker (Invitrogen) was the size standard on each gel and digital images were captured under UV illumination using a Kodak DC290 camera. Isolates with product of correct size for stx1 (732 bp), stx2 (779 bp), eaeO157 (1087 bp) and fliC (625 bp) were confirmed to be E. coli O157:H7.
Prevalence of E. coli O157:H7 was estimated from samples collected within pens, and the probability for E. coli O157:H7 being detected was determined at the sample and pen levels. Exact binomial 95% confidence intervals (CI) were calculated for prevalence proportions (given in brackets within the text). All analyses were performed using SAS (version 9.1; SAS Institute, Cary, North Carolina, USA); P ≤ 0.05 was used for all hypothesis testing. Nonparametric methods were used to compare E. coli O157:H7 pen classifications based on ROPES versus those based on fecal and hide samples. Wilcoxon 2-sample tests were used to compare the median prevalence for feces and hides to ROPES classifications, and Spearman rank tests to investigate correlations between ROPES results and within-pen prevalence estimates for fecal and hide samples. Logistic regression was used to model the probability of pens testing ROPES positive based on within-pen prevalence estimates of E. coli O157:H7 for fecal and hide samples. Within-pen prevalence estimates then were categorized based on natural data breaks and least-square means tests were used to separate probability estimates.
Associations among pen and climate characteristics and the E. coli O157:H7 statuses of pens were determined using generalized mixed linear models (Proc Glimmix). For each analysis, statistical models were developed using a series of steps that were similar to procedures used in previous investigations of E. coli O157:H7 prevalence (12,15,16). Four separate pen-level analyses were performed to investigate factors associated with within-pen fecal prevalence (Period 1), within-pen hide prevalence (Period 2), and the presence or absence of E. coli O157:H7 for each pen based on ROPES (2 separate analyses; one for each of the 2 study periods). Pens were considered ROPES-positive when at least 1 rope sample cultured positive (5,12). For descriptive purposes (Table I), categorical variables with sparse data (for example, month of sampling) were analyzed for unconditional associations using Fisher’s exact tests. For all Period 1 regression analyses, a random effect for feedlot was included to account for the non-independence of pens within feedlots. Unconditional associations among the E. coli O157:H7 status of pens and each potential explanatory variable were identified initially using a screening step. Each variable associated with E. coli O157:H7 on screening (P < 0.2) was entered into multivariable models and then removed using a backward selection approach until all variables remaining were significant at P ≤ 0.05; all significance testing was two-sided. Specific interaction terms were investigated after main effects were determined. To check for collinearity, associations among all factors that were associated with E. coli O157:H7 based on screening (P < 0.2) were examined. Fisher’s exact, Pearson correlation coefficient, and Spearman’s correlation tests were performed for binary and nominal variables, continuous variables, and ordinal variables, respectively; Kruskal-Wallis tests were used to compare ordinal and continuous outcomes with categorical outcomes.
Over the 2 study periods during 2003 and 2004, we studied 84 pre-slaughter pens containing nearly 27 000 head of cattle in 4 commercial feedlots in Alberta. For Period 1 (2003), the cattle capacity of the 4 participating feedlots ranged from < 10 000 to > 50 000. The largest of these feedlots also participated during Period 2 (2004). The number of cattle within a pen ranged from 70 to 535 cattle (mean = 318, median = 277). The mean number of days on feed for study pens was 147.5 (median = 150) and ranged between 90 and 229 days. Of the 44 pens studied in Period 1, 33 held steers and 11 held heifers; and 5 pens of cattle were British breeds, 16 were continental breeds, and 23 were mixed/other breed pens. In Period 2, 40 study pens were comprised of 35 steer and 5 heifer pens; and 13 British breed, 15 continental breed, and 12 mixed/other breed pens. All cattle were fed grain-based, high concentrate finishing rations typical of feedlots in western Canada.
During Period 1, we recovered E. coli O157:H7 from 39 of 44 pens (88.6%; CI 75.4% to 96.2%) using fecal sampling and from 22 of 44 pens (50%; CI 34.6% to 65.4%) using ROPES (Figure 1). Therefore, the probability of pens to test ROPES-positive given that E. coli O157:H7 was detected in feces (that is, the pen-level diagnostic sensitivity of ROPES for detecting fecal positive pens) was 0.56 (22/39) or 56.4% (CI 39.6% to 72.2%). Escherichia coli O157:H7 was recovered from feces in all pens in which E. coli O157:H7 were recovered from ROPES; thus, the pen-level diagnostic specificity of ROPES was 100%. Within pens, prevalence ranged between 0% and 80% (median = 20%) for fecal samples with an overall prevalence of 29.6% (95% CI: 26.8% to 32.5%). Escherichia coli O157:H7 were recovered from all 20 pens studied in the late summer (July 20 to September 14, 2003). During that time, within-pen fecal prevalence ranged from 7% to 77% (median = 40%) and E. coli O157:H7 were recovered from 37% of the cattle feces (CI 33.1% to 41.0%). One or more ROPES were positive in 15 of these 20 pens, which equates to 75% (CI 50.9% to 91.3%) sensitivity for ROPES to detect pens with at least 1 positive fecal sample. In pens studied in the fall and early winter (October 19 to December 7, 2003), E. coli O157:H7 was recovered from feces from 19 of 24 (79.2%) pens and the within-pen sample prevalence ranged from 0% to 80% (median = 13.3%). During that time, E. coli O157:H7 were recovered from 18.2% of the fecal samples (CI 14.5% to 22.4%) and from ROPES in 7 of 24 pens (29.2%). Thus, the estimated sensitivity of ROPES for detecting pens with at least 1 positive fecal sample was 36.8% (CI 16.3% to 61.6%) for the fall/winter period. The median within-pen prevalence of E. coli O157:H7 in fecal samples, and the proportion of pens with ≥1 culture-positive rope samples were both significantly higher during the later summer than the fall and earlier winter.
Overall sample prevalence of E. coli O157:H7 in feces from pens where one or more ROPES were positive was 39.8% (CI 35.8% to 44.0%); whereas the fecal sample prevalence in ROPES negative pens was 15.7% (CI 12.4% to 19.6%). The median sample prevalence for feces from ROPES positive pens (40.0%) was significantly higher than the median prevalence (13.3%) for ROPES negative pens. Prevalence in fecal samples within a pen was correlated with the number of ROPES that cultured positive for the pen (r = 0.63, P < 0.01) and also was correlated with whether at least one ROPES was culture positive for the pen (r = 0.59, P < 0.01). Within-pen fecal sample prevalence was significantly associated with the probability of pens testing ROPES positive. The probability of testing ROPES positive increased as fecal sample prevalence increased (Figure 2). The probability for pens with < 10% fecal prevalence (n = 12) differed from those with 25% to 50% (n = 10) and > 50% prevalence (n = 8), but not from those with 10% to 25% fecal prevalence (P = 0.28). The probability of testing ROPES positive for pens with 10% to 25% fecal sample prevalence (n = 10) also differed from those with 25% to 50% and > 50% prevalence, but the probability did not differ between the latter 2 categories (P = 0.67).
During Period 2, E. coli O157:H7 was recovered from either cattle hide samples or ROPES in 24 of 40 pens (60.0%; CI 43.3% to 75.1%) (Figure 3). Escherichia coli O157:H7 were recovered from ≥ 1 ROPES in 6 pens where all hide samples were culture negative. Escherichia coli O157:H7 were also recovered from hides in 6 pens where all ROPES were culture negative, yet only one hide sample (2.5%) was positive in each of these pens. Therefore, estimates of the pen-level diagnostic sensitivity and specificity of ROPES for detecting pens with at least one positive hide sample were 66.7% (12/18; CI 41.0% to 86.7%) and 72.7% (16/22; CI 49.8% to 89.3%), respectively. There was an overall prevalence on individual cattle hides of 3.13% (CI 2.33% to 4.10%), and individual hide prevalence within pens ranged from 0% to 30% (median = 0%). In the 18 pens where ≥ 1 ROPES were positive, the prevalence of E. coli O157:H7 on individual cattle hides was 6.11% (CI 4.48% to 8.12%); whereas the overall hide prevalence for the 22 ROPES negative pens was 0.68% (CI 0.25% to 1.48%). The median within-pen hide prevalence for ROPES positive pens (3.75%; range: 0% to 30%) was significantly higher than the median (0%; range: 0% to 2.5%) for ROPES negative pens. Prevalence of E. coli O157:H7 on hides within a pen was correlated with the number of ROPES that cultured positive within a pen (Spearman’s r = 0.49, P < 0.01) and also was correlated with whether or not at least one ROPES was culture positive for the pen (r = 0.50, P < 0.01). The probability of testing ROPES positive increased as hide sample prevalence increased (Figure 4). The probability of testing ROPES positive for pens with ≥ 5% hide sample prevalence (n = 9) differed from that for pens with < 5% but > 0% prevalence (n = 9) and those with 0% prevalence (n = 22), but did not differ between the latter 2 prevalence categories (P = 0.77).
All potential explanatory pen and climate factors that were investigated were associated with at least one of the E. coli O157:H7 outcome measures based on initial screening (P < 0.2) during the years of the study (Table I). However, all factors associated with E. coli O157:H7 based on screening also were associated with at least one other independent variable for that year (data not shown). For example, the number of cattle within pens, number of days on feed, conditions of the pen floor, hide tag scores, water tank scores, and all temperature and precipitation measures varied by season. Although some similarities existed, the 4 final multivariable models had different combinations of independent variables (Tables II and III).
For pens in Period 1 (2003), the probability of detecting E. coli O157:H7 in feces was higher when the pen floor conditions were drier (lower scores); given the month of sampling and the mean daily low temperature for the week of sampling (Table II). The respective probability of detecting E. coli O157:H7 decreased from July to December, and, after accounting for the month and pen floor conditions, an increase in the mean low temperature for the week of sampling also resulted in a decrease in the detection probability (Table II). An interaction between month and pen floor conditions approached statistical significance (P = 0.07). The probability of pens testing ROPES positive during Period 1 increased with fewer days on feed and an increase in the mean low temperature for the week of sampling (Table II).
For Period 2 (2004) pens, the probability of detecting E. coli O157:H7 on hides was higher when there were more cattle in the pen, there was less tag on the hides, the water tanks were dirtier, and the mean precipitation for the sampling week was higher (Table III). The probability of pens being ROPES positive increased as the precipitation on the day of sampling and the mean low temperature on the week of sampling increased (Table III). An interaction between these variables approached statistical significance (P = 0.07).
We studied feedlot cattle immediately prior to slaughter and found the prevalence of E. coli O157:H7 from ROPES, fecal, and hide samples varied considerably between feedlot pens. We have shown that ROPES results are associated with fecal and hide prevalence, and thus we concur with previous work suggesting ROPES may be a valuable tool for identifying pens of cattle that are potentially of high risk to food safety (5,11,12). The seasonal, climatic, and pen condition factors that were associated with E. coli O157:H7 outcome measures are factors that have been investigated previously, yet the associations in this study were not entirely consistent with previous literature, and varied to some extent among years and outcome measures studied. Although the feedlots and cattle studied were fairly typical for Alberta, the study pens were from a limited number of feedlots that were not chosen randomly, and thus may not be completely representative of cattle in commercial feedlots in Alberta.
Prevalence estimates of E. coli O157:H7 varied tremendously, but were similar to previously reported estimates. The within-pen fecal sample prevalence of E. coli O157:H7 herein ranged from 0% to 80%, which is consistent with previous descriptions of 0% to 100% (7), 0% to 93.3% (17), and 0.7% to 79.8% (5) in US feedlots, and 0% to 90% in Canadian feedlots (6). Our hide prevalence ranged from 0% to 30% within-pens, which is similar to previous estimates of hide prevalence in feedlot cattle that range from 0% to 25% at the feedlot and 11% to 60% at slaughter (1,3,18–20). The large variability in pre-harvest prevalence for both feces and hides illustrates that there is a real opportunity to identify pens with higher levels of E. coli O157:H7 prior to slaughter, and that associated food safety risks should not be assumed equal for all pens of cattle. Approximately half of the pens that we studied were positive for E. coli O157:H7 based on ROPES, which have been used to detect high prevalence pens (5,11,12). A previous study evaluating 1160 pens in Alberta feedlots found 6.8% of pens ROPES-positive, but only 1 rope device was used per pen compared to 7 in the current study (11). Animal contact with ROPES is greater when more ROPES are used per pen (21). In a study of feedlots in the Midwestern United States, 7 ROPES per pen were used and E. coli O157:H7 were recovered in 52% of pens (5). As discussed in the following text, there are several factors associated with E. coli O157:H7 prevalence pre-harvest, yet much of the variability over time and among feedlot pens is unexplained (12,16,22,23).
Previous evidence indicates that culturing rope devices may serve as an indicator of pens with high prevalence of E. coli O157:H7 in feces (5,11). The current study provides further support that ROPES may identify pens of cattle with a high prevalence of E. coli O157:H7 in feces, and also demonstrates an association between ROPES and hide prevalence. Smith et al (5) found a slightly stronger correlation (r = 0.72) between the number of positive ROPES and fecal prevalence than we did (r = 0.63), but they tested feces from all cattle within a pen rather than just a subset. They further established pen-level estimates of sensitivity (82%) and specificity (92%) for classifying a pen as positive for high prevalence (using a cut-point of 16% within pen fecal prevalence) based on one or more ROPES culturing positive. Although in our study the confidence intervals for fecal prevalence estimates for many pens overlapped their established 16% cut-point (Figure 1), pens with higher point estimates for fecal prevalence were more likely to be ROPES positive and only a limited number of pens appeared to be potentially misclassified with ROPES. In addition, this study has shown that the probability of a pen testing positive with ROPES increases when the within-pen prevalence of E. coli O157:H7 fecal and hide samples increases (Figures 2 and and4).4). A previous study of Alberta feedlots showed that if only one sample per pen was collected, then E. coli O157:H7 would be detected more frequently using a rope than a sample of either feces, oral fluids, feed or water sources (11). Our study is the first to demonstrate a significant association between ROPES and the hide prevalence within pens. Although the correlation between the number of positive ROPES and hide prevalence within a pen was not as strong as for fecal prevalence, this may be due in part to the lower overall prevalence for hides. This study has shown that ROPES were positive for all pens in which the point estimate for hide prevalence was 5% or higher. Unfortunately we were unable to sample cattle hides or carcasses at the abattoir, which would have provided more information on the utility of ROPES for assessing potential food safety risk.
Even if ROPES or other strategies result in occasional misclassification of E. coli O157:H7 levels of pre-harvest pens, using simple pen-level monitoring strategies may allow for more effective applied research or surveillance. Epidemiologic studies to examine relationships between pre-harvest factors and E. coli O157:H7 typically have required large numbers of fecal and/or hide samples to be collected to accurately categorize the status of individuals and groups of cattle. Associated logistic and budgetary impacts can result in fewer pens being studied and a limited ability to efficiently evaluate risk factors or interventions. In feedlot operations, risk factors or interventions would be identified or implemented at the pen or whole feedlot level. Thus, identification of pen status without the expense of individual animal samples may be valuable, and a practical pen-sampling strategy would not have to be perfect to have utility. Others have provided further discussion of pen-level sampling and have shown the utility of using ROPES as compared to other strategies such as composite fecal or water sampling (5,11).
All of the factors corresponding to season, climate, and pen conditions that we studied were associated with E. coli O157:H7 outcome measures either on screening or multivariable statistics (Table I). The ability to evaluate the effects of potential risk factors was affected by the limited number of pens sampled within time and place, and by associations among risk factors. Others have provided further discussion of how limitations in effective sample size and the inter-relatedness among risk factors, especially within hierarchical units and seasons, can be constraints to evaluating risk factors for E. coli O157:H7 (19,22,23).
A higher prevalence in summer months and with warmer temperatures is well described. Previous studies have reported similar patterns of prevalence in feedlots, dairies, and abattoirs (3,12,22,23). In a previous study of Alberta feedlots, the likelihood of fecal samples and ROPES testing positive for E. coli O157:H7 was more than 2 times greater if collected from April to September than the other 6 months of the year (11). Smith et al (12) also found the probability for pens of cattle to test ROPES-positive was higher in the summer than in the winter. The multiple temperature parameters investigated herein were all associated with fecal, hide, and ROPES prevalence (Table I); however, only mean daily low temperature for the week of sampling was significant in multivariable models. Temperature parameters were extremely correlated, but perhaps mean low temperature is more reflective of the seasonal aspects of E. coli O157:H7. A higher mean low temperature increased the probability of pens testing ROPES positive for E. coli O157:H7 in both years of our study (Tables II and III). However, this factor was associated with a decreased probability of fecal samples testing positive after considering the month of sampling and pen floor conditions (Table II). Our ROPES findings were consistent with previous observations that higher 7-day mean air temperatures were associated with a greater probability for detecting ROPES-positive pens (12). Additionally, increasing heat index, a combination of air temperature and humidity, has been shown to increase the likelihood of feed samples testing positive for E. coli O157:H7 (15). Although not significant, we found a trend towards and interaction between temperature and precipitation factors in our model of 2004 ROPES prevalence. Season and temperature seem consistently associated with E. coli O157:H7 in feedlot cattle, yet prevalence even within season varies. Season and temperature are not determinants that can be modified directly for feedlot cattle, yet they may represent other, potentially modifiable factors.
Many risk factors investigated varied with season; muddy pen conditions, for example, occurred more frequently in late fall or winter harvested pens (data not shown). A higher fecal prevalence of E. coli O157:H7 was found with drier pen conditions, even when accounting for the month of sampling (Table II). The same association was seen in preliminary analysis of ROPES and hide prevalence (Table I). Although not significant, the trend toward an interaction between month and pen conditions in the fecal prevalence model suggests that the effect of pen condition depends on the month of sampling. Previous data from US feedlots indicate a higher within-pen fecal prevalence in muddy (median 22.4%) and dusty (17.9%) pens compared to normal (6.5%) (8). The probability of a ROPES positive pen also has been reported to be higher in both dry or wet pens compared to ideal pen conditions (12). Another US study found a higher prevalence of E. coli O157:H7 in feces when pens were dry versus wet (24). Although similar types of scoring systems were used in these studies, it is difficult to compare results across studies when pen conditions are assessed based on ordinal scores assigned by different researchers in different situations.
Hide tag consisting of mud, manure, or bedding that is adhered to live animals arriving at an abattoir could have implications in contamination of facilities and beef products. However, a model of pre-harvest control of E. coli O157:H7 indicated that reducing tag scores would not substantially reduce carcass contamination (4). In addition, a study of Alberta cattle found no consistent relationship between tag scores and carcass contamination as indicated by counts of aerobic organisms, coliform bacteria, and generic E. coli (25). A higher fecal prevalence of E. coli O157:H7 in US feedlots has been associated with cattle having mud/manure below fetlock compared to dry cattle, but the prevalence was lowest for cattle with tag above the fetlock (16). The cleanliness of hides was not associated with ROPES positive status in previous studies of summer or winter-fed cattle in US feedlots (12). In our study, preliminary analysis revealed that hide tag was inversely related to fecal prevalence, hide prevalence and Period 1 ROPES prevalence, but final models showed a significant inverse relationship only between hide tag and hide prevalence (Table III). In both 2004 models, an increasing amount of precipitation on the week or day of sampling increased the probability of E. coli O157:H7 on hides or ROPES respectively. Perhaps clean, yet damp cattle were more conducive to recovery, growth and/or maintenance of E. coli O157:H7 in the pens that we studied.
The cleanliness of cattle water tanks also has been associated with E. coli O157:H7 prevalence in our study and others. The impact of water tank cleaning frequency or degree of cleanliness at sampling has been reported previously with mixed results. Dewell et al (24) observed a negative relationship (OR = 0.98) on univariate analysis between the number of days since cleaning and E. coli O157:H7 positive fecal samples. Another US study found no correlation between water tank cleanliness and E. coli O157:H7 presence in fecal samples (8). Other US studies have indicated that cleanliness of water tanks was not related to presence of E. coli O157:H7 in water or feces; however, presence of the E. coli O157:H7 organism in the water did increase the likelihood of fecal and ROPES samples testing positive (12,15,16). Our initial analysis revealed that water tank condition was associated with fecal, hide, and ROPES prevalence of E. coli O157:H7 (Table I). However, tank cleanliness was only significant in the final model for hide prevalence in which pens with clean tanks had the lowest probability of hide samples testing positive (Table III). Presence of E. coli O157:H7 in water tanks was not directly evaluated in this study.
Results indicated that a greater hide prevalence of E. coli O157:H7 was associated with a larger number of cattle per pen (Table III), and that pens with a higher number of days on feed were more likely to be ROPES positive in Period 1 (Table II). Larger pens are not always associated with higher E. coli O157:H7 prevalence. Sargeant et al (16) found that the number of cattle per pen was inversely related to fecal prevalence. Another study failed to observe any relationship between cattle per pen and fecal shedding of E. coli O157:H7 (8). Our initial analysis indicated an association between the number of days on feed for a pen and the probability of a pen testing positive with ROPES and hide prevalence within pens (Table I). The final model for ROPES prevalence in Period 1 indicated that an increasing number of days on feed was associated with a reduced probability of pens testing positive. A recent Alberta study showed that cattle on feed for less than 30 days were more likely to be fecal positive for E. coli O157:H7 than cattle on feed for more than 30 days (11). However, cattle that had recently arrived at the feedlot were not studied herein, only those from pens that were pre-slaughter. Other researchers have not observed any association between days on feed and shedding of E. coli O157:H7 (8,24).
It was demonstrated that the prevalence of E. coli O157:H7 in Alberta feedlot pens immediately prior to slaughter can be quite high or very low even within feedlots and seasons. As others have suggested (5,6), pens with a high prevalence of E. coli O157:H7 could be targeted for mitigation strategies if they could be effectively identified prior to slaughter. Fecal and hide prevalence are considered major pre-harvest indicators of potential carcass contamination, but other methods such as ROPES that are associated with these outcomes can provide logistic advantages (5,11,12). Associations between carcass contamination or other more direct measures of food safety risks and ROPES or other pre-harvest measures should be further investigated. Although several factors associated with pen-level outcome measures of E. coli O157:H7 were identified, the complexity of the epidemiology and interrelatedness of risk factors may limit the ability to predict or control E. coli O157:H7 based on characteristics of the pen (19,22). Further efforts should be directed towards determining practical and valid methods for identifying cattle that pose the greatest risk to food safety so that appropriate mitigation and monitoring strategies can be developed and adopted.
This work was supported by the Food Safety Division of Alberta Agriculture and Food. The authors thank Ludovic Silasi, Annette Visser, Arlene Otto, Carol Goertz, Suzanne Gibson, Cathy Sheppard, Kyla Kennedy, Wayne Lazaroff, and Pat Layton for their excellent technical assistance. This is contribution number 07-168J from the Kansas Agricultural Experiment Station.