Association of Herd Management Factors with Colonization of Dairy Cattle by Shiga Toxin-Positive Escherichia coli 0157

Transcription

1 802 Journal of Food Protection, Vol. 61, No.7, 1998, Pages Copyright, International Association ot Milk, Food and Environmental Sanitarians Association of Herd Management Factors with Colonization of Dairy Cattle by Shiga Toxin-Positive Escherichia coli 0157 DONALD E. HERRIOTT,! DALE D. HANCOCK,2* ERIC D. EBEL,3 LINDA V. CARPENTER,4 DANIEL H. RICE,2 AND THOMAS E. BESSER5!United States Department of Agriculture, Animnl and Plant Health 1nspection Service, 1kterinary Services, 530 Center Street Northeast, Suite 335, Salem, Oregon 97301; 2Field Disease 1nvestigation Unit, Department of1kterinary Clinical Sciences, Washington State University, Pullmnn, Washington ; 3United States Department of Agriculture, Animnl and Plant Health 1nspection Service, 1kterinary Services, 7261 Franklin Road, Boise, 1daho 83709; 4United States Department of Agriculture, Animnl and Plant Health 1nspection Service, Veterinary Services, th Court Southwest, Suite B, Olympia, Washington 98502; and 5Department of1kterinary Microbiology and Pathology, Washington State University, Pullman, Washington , USA MS :Received26August1997/Accepted6 January1998 ABSTRACT Management factors in 36 Pacific Northwest dairy herds were evaluated for their association with the prevalence of Shiga toxin-positive Escherichia coli 0157 (E. coli 0157) in dairy cattle. The within-herd prevalence of E. coli 0157 was estimated by bacteriological culture of fecal pat samples, collected monthly for 6 months (approximately 60 per visit), from heifer cattle. During the first visit to each farm, a management questionnaire was administered that covered a broad range of animal husbandry practices. On each subsequent visit, a brief questionnaire was administered to detect changes in management practices. A significantly higher prevalence of E. coli 0157 was noted in herds that fed corn silage to heifers compared to herds that did not feed corn silage. More tentative associations of E. coli 0157 prevalence were observed for weaning method, protein level of calf starter, feeding of ionophores in heifer rations, feeding of grain screens to heifers, and feeding of animal by-products to cows. Infections involving Shiga toxin-positive Escherichia coli 0157 (E. coli 0157) in humans have been linked to cattle either directly or indirectly (1, 5, 8, 15-18, 21). This has led to interest in the ecology of E. coli 0157 on cattle farms and in the potential factors that might influence this organism's ecology (3, 6, 7, 9-12, 15). Several studies on cattle herds have demonstrated that E. coli 0157 occurs in the majority of cattle herds, including both dairy and beef (10-12). Fecal shedding of E. coli 0157 in cattle is both seasonal, with the highest prevalence in cattle seen in the late summer to early fall, and clustered in periods of high herd prevalence separated by periods of low to undetectable fecal shedding (10, 11, 15). Heifer cattle 3 months to breeding age have been demonstrated to be the most commonly colonized age group of cattle (9-11, 15), and once colonized, cattle typically excrete for no more than 1 month (3). These data indicate that there might be factors on cattle farms that could be modified to reduce the prevalence of E. coli 0157 in cattle and ultimately in the associated disease in humans. During the beef production process, there are practices and procedures in place with the intent of reducing the potential for contamination of finished food products. These practices are both mandated by governing agencies and voluntary on behalf of the producer or processor. However, there has been only limited attention paid to farm-level control of agents that cause foodborne disease. It is possible that farm-level controls aimed at reducing the incidence of human pathogens in cattle, such as E. coli 0157, may provide significant positive results with a minimal level of resource commitment. * Author for correspondence.tel: ;Fax: ; hancock@vetmed.wsu.edu. This report is a follow-up to one that examined the hypothesis that herds that applied manure to grazing land would have a higher prevalence of E. coli 0157 in their cattle (11) than cattle in herds that did not apply manure to grazing land. The current study presents the analysis of a wide variety of other management data that were collected. These data were analyzed in a hypothesis-generating mode to identify herd management practices that might be associated with the prevalence of E. coli 0157 in cattle in these herds. MATERIALS AND METHODS Herd selection and sampling methods. Herd selection and sampling methods utilized in this study are described elsewhere (11). Briefly, the selection of 36 herds was designed to allow maximum statistical power in testing the specific hypothesis that herds that applied manure to grazing land would have a higher prevalence of E. coli 0157; thus, herds were selected on the basis of heifer housing/pasturing practices. In each herd, approximately 60 fresh fecal pat samples were collected monthly for 6 months from female juvenile dairy cattle between 2 and 24 months of age (heifer).previousstudieshavedemonstratedthatthisagegrouphasboth the highestprevalencee. coli 0157 of any age group of cattleand that the prevalencein heifersis correlatedto thatin otheragegroupspresent on farms (10, 15).The associationsexaminedin the presentpaper were not hypothesized a priori, and herds were selected irrespective of managementpracticesbeyondhousingof heifercattle. Bacteriological methods. Fecal detection of E. coli 0157 was based on previously reported methods (19). In this study, E. coli isolatesthatwereunableto fermentsorbitol,thatlackedl3-g1ucuronidase activity,and that possessedboth 0157 antigen and Shiga toxin genes (2) were designatede. coli Possessionof the H flagellar antigen was not determined in this study because a previous study

2 J. Food Prot., Vol. 61, No.7 E. COLI 0157 RISK FACTORS IN CATTLE 803 demonstrated that it is not a consistent marker and that E. coli 0157 isolates meeting the above criteria belong to a single group of clonal organisms capable of producing disease in humans (22). Collection of farm data. During the first visit to each fann, a questionnaire was completed covering a wide range of management variables. On each subsequent visit, an update questionnaire was completed for each herd specifically addressing any changes that may have occurred during the preceding month. Thirty-six management variables (divided into preweaned calf management, heifer management, and cow management) were examined for possible association with E. coli 0157 prevalence in cattle. Management variables affecting groups other than heifers (the age group sampled) were included due to hypothesized herd effects. Data analysis. The Kruskal- Wallis test was used to compute the significance of differences in distributions of E. coli 0157 prevalences among management groups (13). For this analysis, the herd (n = 36) was considered the unit of observation; the dependent variable was herd prevalence of E. coli Log-linear analysis was used to test significance of joint effects of two management variables on herd status (yes orno) for detection of E. coli 0157 (4). RESULTS Prevalence of Shiga toxin-positive E. coli Detailed E. coli 0157 culture results are given elsewhere (11). Briefly, Shiga toxin-positive E. coli 0157 was found in TABLE 1. Association between Shiga toxin-positive E. coli 0157 prevalence and preweaned calf management variables Quartile 1 Median Quartile 3 Calf weaning, feeding, management No. herds No. (%) positive herds prevalence prevalence prevalence Pvalue Weaning age :52 months 0.60 Yes 12 9 (75) No (74) Weaning method 0.07 Abrupt (88) Gradual/amount 6 3 (50) Gradual/dilution 3 0(0) Gradual/interval 9 8 (89) Criteria for weaning 0.22 Set age (68) Grain intake 3 2 (67) Body size 10 8 (80) Convenience 3 3 (100) Grouped with respect to weaning 0.33 Same time (78) Weaned first (73) Grouped first 6 4 (67) > 12 calves in weaning pen 0.34 Yes (78) No (70) Beginning of forage diet relative to 0.63 weaning age Before (74) Simultaneous 10 7 (70) After 6 5 (83) Begin feeding starter <5 days of age 0.28 Yes (79) No (69) Forage: % protein> Yes (67) No (80) Calf starter: % protein> Yes (85) No 15 9 (60) Oats in calf starter 0.55 Yes 11 9 (82) No (71) Barley in calf starter 0.95 Yes (73) No 9 7 (78) Com in calf starter 0.36 Yes (84) No 10 5 (50) Wheat in calf starter 0.87 Yes 8 6 (75) No (74)

3 804 HERRIOTI ET AL. J. Food Prot., Vol. 61, NO.7 TABLE 2. Association between Shiga toxin-positive E. coli 0157 and weaned heifer management variables No. No. (%) Quartile I Median Quartile 3 P Heifer feeding, management variables herds positive herds prevalence prevalence prevalence value Ionophores fed to heifers 0.11 Yes (89) No (59) Monensin fed to heifers 0.10 Yes 6 6 (100) No (70) Lasalocid fed to heifers 0.70 Yes (85) No (70) Decoquinate fed to heifers 0.59 Yes (73) No (76) Decoquinate, monensin, or lasalocid fed to heifers 0.06 Yes (81) No 9 5 (56) >41b (1.82 kg)/headjday 2--6 months of age 0.55 Yes 16 9 (56) No (89) >41b (1.82 kg)/headjday 6-12 months of age 0.94 Yes (67) No (81) >21b (.91 kg)/headjday months of age 0.68 Yes (83) No (71) >4 Ib kg)/headjday 1 month precalving 0.68 Yes (71) No (77) Com silage fed to heifers 2-6 months of age 0.48 Yes 3 3 (100) No (72) Com silage fed to heifers 6-12 months of age 0.20 Yes 9 9 (100) No (67) Com silage fed to heifers months of age 0.03 Yes (92) No (65) Com silage fed to heifers 1 month precalving 0.01 Yes (94) No (58) Grain screens fed to heifers 0.14 Yes 8 7 (88) No (71) Beet pulp fed to heifers 0.50 Yes 13 9 (69) No (78) Whole cottonseed fed to heifers 0.88 Yes 8 6 (75) No (75) Potato by-product fed to heifers 0.94 Yes 2 1 (50) No (76) heifers on 27 of the 36 farms, and the within-herd prevalence method and protein content of calf starter-were associated ranged from 0 to 5.5%. with E. coli 0157 prevalence. Among 17 farms that abruptly Calf management variables. Thirteen preweaned calf weaned calves, the median E. coli 0157 prevalence was management variables were evaluated (Table 1). These 1.67%. This compared to median prevalences of 0.82% focused on weaning and grouping strategies, age of onset of among six herds that gradually weaned calves by reducing calf starter and forage feeding, protein content of forage and the amount fed, 0.0% among three herds that gradually calf starter, and several specific starter ingredients that were weaned by diluting the milk, and 0.83% among nine herds found to vary among study herds. Two variables-weaning that gradually weaned calves by increasing the interval

4 J. Food Prot., Vol. 61, No.7 E. COU 0157 RISK FACTORS IN CATTLE % 5.5% Q) 5.0% g 4.5% Q) ~ 4.0% ~ 3.5% % ~ 25% -. o 2.0% :.g 1.5% u 1.0% W 0.5% 0.0%... 11I11I11I ;,.. III FIGURE 1. Distribution of within-herd prevalence ofe. coli 0157 categorized by the feeding of corn silage and ionophores to heifers. between feedings (P = 0.07). Among the 17 farms that fed calf starter containing more than 16% protein, the median E. coli 0157 prevalence was 1.37%; this compared with 0.28% for farms feeding a starter with a lower protein content (P = 0.13). Heifer management variables. Sixteen heifer management variables related to the period between weaning and first calving were evaluated (Table 2). These focused on feed additives, amount of grain fed, com silage feeding, and use of several by-product feeds. Com silage feeding had the strongest association with E. coli 0157 prevalence. Among the 13 herds in which com silage was fed to 12 to 18-month-old heifers, the median prevalence of E. coli 0157 was 1.73% compared with 0.28% in the remaining herds (P = 0.03). Of the nine herds in which E. coli 0157 was not detected, only one fed com silage to heifers at any age before calving. A more tentative association was observed between E. coli 0157 prevalence and the use of feed additives in heifer rations (Table 2). E. coli 0157 prevalence was higher in herds that used monensin, lasalocid, and/or decoquinate in their heifer rations compared with that of herds not using these additives. This association appeared to be strongest for monensin (median prevalence = 1.73% in herds using it compared with 0.69% in herds not using it, P = 0.10), though monensin was used in only six herds. Among the 27 herds in which at least one of the three feed additives was used, the median prevalence was 1.66% compared with 0.27% in the remaining 9 herds (P = 0.06). The joint effects of the feeding of com silage and ionophores (monensin or lasalocid) to heifers were examined graphically (Fig. 1) and with log-linear analysis. Controlling for ionophore use, com silage feeding was significantly (P < 0.01) associated with E. coli 0157 herd status (positive or negative). Controlling for com silage use, supplementation with ionophores was significantly associated with E. coli 0157 herd status (P = 0.07). No significant interaction was observed (P = 0.48). It seemed noteworthy that 6 of the 9 herds in which E. coli 0157 was not detected were among the 10 herds in which neither com silage nor ionophores were fed (Fig. 1). Feeding of grain screens to heifers between weaning and first calving was found to be associated with E. coli 0157 prevalence in herds (P = 0.14) (Table 2). Though this product was fed in only eight herds, the median prevalence in these herds was 2.10% compared with 0.96% in the remaining herds. Cow management variables. Seven variables related to adult cow management and feeding were examined for possible association with E. coli 0157 prevalence in heifers TABLE 3. Association between Shiga toxin-positive E. coli 0157 and cow management variables No. No. (%) Quartile 1 Median Quartile 3 P Cow feeding and management herds positive herds prevalence prevalence prevalence value Animal by-products (blood, meat, or feather meal) 0.09 fed to cows Yes (100) No (64) Grain screenings fed to cows 0.67 Yes (73) No (76) Beet pulp fed to cows 0.84 Yes (74) No 13 10(77) Whole cottonseed fed to cows 0.86 Yes (76) No 11 8 (73) Bakery waste fed to cows 0.19 Yes 3 3 (100) No (73) Potato by-product fed to cows 0.72 Yes 4 3 (75) No (75) Outside cattle entered herd during study period 0.85 Yes 10 8 (80) No (72)

5 806 HERRIOTI ET AL. J. Food Prot., Vol. 61, No.7 (Table 3). Among six types of by-product feeds fed to cows in the study herds, animal by-product feed was the only type that appeared to be associated with E. coli 0157 prevalence (P = 0.09). E. coli 0157 was detected in all 11 herds that were fed animal by-products (median prevalence = 1.67%) but in only 16 of 25 herds that were not fed animal by-products (median prevalence = 0.55%). Entry of outside cattle into the herds during the study period was not associated with a higher prevalence of E. coli 0157 (P = 0.85). Among 10 herds that admitted new animals, the median prevalence of E. coli was 1.25% compared with a prevalence of 1.09% among herds that did not admit new animals. DISCUSSION On-farm control of organisms that cause foodborne disease in humans has been largely ignored in the ongoing effort to produce safe food. An initial step in the process of identifying potential on-farm control measures includes risk factor studies to identify management factors on a farm that can potentially be modified and that are tentatively associated with the presence of the disease-producing organism. These tentative associations must then be the focus of further study through direct manipulation trials to test the specific hypothesized association. Previous studies have identified associations between management variables and the prevalence of E. coli 0157 in cattle on farms (6, 7, 9). These studies, along with the present one, share several weaknesses inherent in broad, questionnaire-based etiologic studies. These weaknesses include lack of precision in the complexity of correlations among measured and unmeasured variables, suboptimal distribution among herds with respect to many of the management variables such that only a handful of herds are available in some strata, and the heuristic nature of statistical tests conducted on a long list of variables in series where control of confounding and interaction is necessarily incomplete. Efforts were taken to enhance data quality in the current study by clarifying management variables on each of six monthly visits. The dependent variable (within-herd prevalence of E. coli 0157) was stronger in the current study than in any previous study because it was based on several hundred samples collected in each herd over a 6-month period instead of only 30, 60, or 120 samples collected on a single day (as described by Garber et al. (7), Hancock et al. (9), and Dargatz et al. (6), respectively). Even relatively high prevalence herds are commonly negative on single sampling dates (10, 11, 15), and small sample sizes give inherently unstable prevalence estimates for an agent that is rarely present in over 5% of cattle in a herd. Nevertheless, the results presented here must be interpreted with caution, and the associations that were observed are best considered as candidates for more targeted studies. As in previous studies (6, 7, 9), this study found several nutritional variables to be associated with E. coli 0157 prevalence in cattle. Based on these associations, and on the biological plausibility of an effect of diet on intestinal flora, it seems probable that some dietary influences on E. coli 0157 truly exist. Unfortunately, the available published studies are not in agreement as to which dietary variables are important. For example, two previous studies (7, 9) reported a negative relationship between whole cottonseed feeding and E. coli 0157 presence in dairy herds, but no hint of this association could be discerned in the current study. This study found that feeding corn silage, grain screens, and ionophores to heifers and feeding animal by-products to adults is associated with a higher prevalence of E. coli 0157 in cattle. These associations have not been reported in previous studies (6, 7, 9). Given the high correlations among dietary variables that are inevitable in observational research, determining which ones are important in modulating colonization of cattle by E. coli 0157 Will require carefully designed studies in which specific dietary factors can be manipulated. The current study is the first to detect a relationship between feeding corn silage to heifers and E. coli 0157 prevalence. While it is conceivable that this association could have been produced by some unknown confounding factor, a biologically plausible explanation exists for the relationship. Once it is removed from the silo, exposed to air, and mixed with other ingredients, corn silage could provide a moist growth medium for environmental bacteria, including E. coli 0157, thus increasing the exposure probability and the dose to cattle. A recent study confirmed that E. coli can replicate to high concentrations in mixed rations containing corn silage when it is maintained at environmental temperatures for 24 h (14). Cattle rations containing silage are commonly left in the feed bunk for periods of 24 h or more, allowing an opportunity for both contamination with bacteria, such as E. coli 0157, and bacterial replication. The association between E. coli 0157 prevalence and feeding ionophores in the heifer ration merits special attention, even though the association observed in the current study was only tentative. Ionophores have been shown to favor gram-negative bacteria in the gastrointestinal tract (20), and they were introduced in the late 1970s (20), a period that is temporally consistent with the apparent increase in human E. coli 0157:H7-related disease (8). Furthermore, a stratified analysis including corn silage along with ionophores indicated that the association of the latter with E. coli 0157 was not just an artifact of confounding with corn silage. The association between abrupt weaning of calves and higher E. coli 0157 prevalence could be due to intestinal floral disruption, which allows for easier colonization of invader-type E. coli strains, a type to which E. coli 0157 may belong. Important to note, however, is that the association observed was with E. coli 0157 prevalence in groups of heifers, only a small minority of which would be within approximately a month of weaning. If the observed association is real, then it suggests a population effect. Observational studies such as the current study are helpful only to the degree that they provide a means to focus the attention of designed experiments onto a smaller number of variables. The need exists for studies focused on specific farm management factors. Farm-level control of E. coli 0157 may provide significant benefits to human health without great expense.

6 J. Food Prot., Vol. 61, No.7 E. COLl 0157 RISK FACTORS IN CATTLE 807 ACKNOWLEDGMENTS This research project was funded by the United States Department of Agriculture, Animal and Plant Health Inspection Services, Veterinary Services and United States Department of Agriculture NRICGP grant no The authors thank the many field personnel, both state and federal, who provided an invaluable service to this research project. REFERENCES 1. Barrett, T. J., H. Lior, J. H. Green, R Khakhria, J. G. Wells, B. P. Bell, K. D. Green, J. Lewis, and P. M. Griffin Laboratory investigation of a multi-state food-borne outbreak of Escherichia coli 0157:H7 by using pulsed field gel electrophoresis and phage typing. J. Clin. Microbiol. 32: Begum, D., N. A Strockbine, E. G. Sowers, and M. P. Jackson Evaluation of a technique for identification of Shiga-like toxinproducing Escherichia coli by using polymerase chain reaction and digoxigenin-iabeled probes. J. Clin. Microbiol. 31: Besser, T. E., D. D. Hancock, L. C. Pritchett, E. M. McRae, D. H. Rice, and P. I. Tarr Duration of detection of fecal excretion of Escherichia coli 0157:H7 by cattle. J. Infect. Dis. 175: Bishop, Y. M. M., S. E. Fienberg, and P. W. Holland Discrete multivariate analysis: theory and practice, p MIT Press, Cambridge, Mass. 5. Chapman, P. A, C. A Siddons, D. J. Wright, P. Norman, J. Fox, and E. Crick Cattle as a possible source of verocytotoxin-producing Escherichia coli 0157 infections in man. Epidemiol. Infect. 111: Dargatz, D. A, S. J. Wells, L. A Thomas, D. D. Hancock, and L. P. Garber Factors associated with the presence of Escherichia coli 0157 in feces of feedlot cattle. J. Food Prot. 60: Garber, L. P., S. J. Wells, D. D. Hancock, M. P. Doyle, J. Tuttle, J. A Shere, and T. Zhao Risk factors for fecal shedding of Escherichia coli 0157:H7 in dairy calves. J. Am. Vet. Med. Assoc. 207:1: Griffin, P. M., and R V. Tauxe The epidemiology of infection caused by Escherichia coli 0157:H7, other enterohemorrhagic E. coli, and the associated hemolytic uremic syndrome. Epidemiol. Rev. 13:6~ Hancock, D. D., T. E. Besser, M. L. Kinsel, P. I. Tarr, D. H. Rice, and M. G. Paros The prevalence of Escherichia coli 0157:H7 in dairy and beef cattle in Washington State. Epidemiol. Infect. 113: Hancock, D. D., T. E. Besser, D. H. Rice, D. E. Herriott, and P. I. Tarr Longitudinal study of Escherichia coli 0157 in fourteen cattle herds. Epidemiol. Infect. 118: II. Hancock, D. D., D. H. Rice, D. E. Herriott, T. E. Besser, E. D. Ebel, and L. V. Carpenter Effects of farm manure handling practices on Escherichia coli 0157 prevalence in cattle. J. Food Prot. 60: Hancock, D. D., D. H. Rice, L. A. Thomas, D. A Dargatz, and T. E. Besser Epidemiology of Escherichia coli 0157 in feedlot cattle. J. Food Prot. 60: Hollander, M., and D. A Wolfe Nonpararnetric statistical methods, p John Wiley Sons, New York. 14. Lynn, T. v., D. D. Hancock, T. E. Besser, J. H. Harrison, D. H. Rice, N. T. Stewart, and L. L. Rowan. The occurrence and replication of Escherichia coli in cattle feeds. J. Dairy Sci., 81: Mechie, S. C., P. A. Chapman, and C. A Siddons A fifteen month study of Escherichia coli 0157:H7 in a dairy herd. Epidemiol. Infect. 118: Ostroff, S. M., P. A Griffin, R V. Tauxe, L. D. Shipman, K. D. Greene, J. G. Wells, J. H. Lewis, P. A Blake, and J. M. Kobayash A statewide outbreak of Escherichia coli 0157:H7 infections in Washington state. Am. J. Epidemiol. 132: Paros, M. G., P. I. Tarr, H. Kim, T. E. Besser, and D. D. Hancock A comparison of human and bovine Escherichia coli 0157:H7 isolates by toxin genotype, plasmid profile, and bacteriophage A-restriction fragment length polymorphism profile. J. Infect. Dis. 168:130~ Rice, D. H., D. D. Hancock, R. L. Vetter, and T. E. Besser Escherichia coli 0157 infection in a human linked to exposure to infected livestock. Vet. Rec. 138:311. (Letter.) 19. Sanderson, M. W., J. M. Gay, D. D. Hancock, C. C. Gay, L. K. Fox, and T. E. Besser Sensitivity of bacteriologic culture for detection of Escherichia coli 0157:H7 in bovine feces. J. Clin. Microbiol. 33: Schelling, G. T Monensin mode of action in the rumen. J. Anim. Sci. 58: Waters, J. R, J. C. M. Sharp, and V. J. Dev Infection caused by Escherichia coli 0157:H7 in Alberta, Canada, and in Scotland: a five-year review, Clin. Infect. Dis. 19: Whittam, T. S., I. K. Wachsmuth, and R. A Wilson Genetic evidence of clonal descent of Escherichia coli 0157:H7 associated with hemorrhagic colitis and hemolytic uremic syndrome. J. Infect. Dis. 157: