1 1 of 23 6/21/2005 8:41 AM A Mail Survey of Ontario Beef Producers' Attitudes about Antibiotics Wendy J. M. Powell Department of Population Medicine 1 Project in Epidemiology POPM*6250 Financial Support for this program was provided by: Ontario Agricultural Adaptation Council OMAFRA animals (beef) program Ontario Cattlemen s Association 1 University of Guelph, Guelph, Ontario, N1G 2W1 Prepared in the style of The Canadian Veterinary Journal.
2 2 of 23 6/21/2005 8:41 AM Abstract -- Surveys were distributed to the beef producers in Ontario, in the fall of 1999, to establish use patterns and perceptions about antibiotics. Most of the 587 respondents thought that antibiotic use needed to be reduced, but as many as half of the respondents may not have known the difference between antibiotics and other injectable products. The results indicate that any guidelines should explain what an antibiotic is, when they should be used, and ways to reduce unnecessary use. Antibiotic use below the median annual percentage of animals treated (10%), was positively associated with: ionophore use in feed; calling a veterinarian for a common animal health problem; and having a larger number of animals on the farm. Additional research is needed to better explain the large variation observed in the percentage of animals treated on the farm and to enhance the quality of prudent use guidelines. Introduction The agricultural contribution to the development of antimicrobial resistance in bacteria is being increasingly scrutinized (1-4). The potential for agricultural uses of antimicrobials to have a negative impact on human health is influencing policy and regulations (1, 2, 5). Denmark banned the use of avoparcin in 1997 and virginiamycin in 1998 (6), the E.U. banned bacitracin, spiramycin, tylosin and virginiamycin for use in animal feed in 1999 (7) and the U.S.A. is currently considering revoking the approval for fluoroquinolone use in poultry (8). Antimicrobials, an essential tool in the humane and cost effective rearing of cattle, are used for growth promotion and to treat, control and prevent disease, under most modern farming conditions. Any antimicrobial use, including use in beef production, has the potential to increase the selection of antimicrobial resistant bacteria, contribute to the development of resistance in bacteria and lead to the spread of the genes that encode resistance factors (9-13). In response to the concerns regarding the efficacy of antimicrobials for both human and veterinary use, medical doctors, veterinary groups and agricultural groups are developing guidelines to maximize the benefits of antimicrobial use while minimizing the risk of antimicrobial resistance. Judicious or prudent use guidelines have been produced by several major veterinary organizations including the Canadian Veterinary Medical Association (CVMA), The World Veterinary Association (WVA), the International Federation of Agricultural Producers (IFAP), the World Federation of Animal Health Industry (COMISA) and the World Health Organization (WHO) (1, 3, 14). These documents explain, in very general terms, the ideal use of antimicrobials. These principles need to be applied to very specific circumstances so that they have practical applications. The purpose of this survey was to obtain information on how the beef producers in Ontario are currently using antimicrobials and to acquire information on their perceptions about the risk of antimicrobial resistance.[d1] Materials and Methods Distribution of the survey An accurate sampling frame for the beef producers in Ontario was not available. The best available alternative was the mailings list for the Ontario Cattlemen s Association's (OCA) magazine. The OCA endorsed the questionnaire that was developed for the study, and entitled it: "The OCA Producer Survey" (Appendix 1). The survey was distributed with the October 1999 issue of their periodical, Ontario Beef. The OCA mailing list for Ontario Beef included everyone who has marketed an animal for beef, including cull dairy cows, unless they had specifically asked to be taken off of the mailing list. It also included anyone that was interested in the industry and requested a copy, including veterinarians, feed manufacturers, researchers and pharmaceutical company personnel. The issue of the magazine that included the questionnaire had a column explaining the purpose of the survey (Appendix 2). It was not possible to contact individual farmers to remind them to fill out the survey because individual names and addresses were not made available. A reminder was printed in the August 2000 issue of Ontario Beef.
3 3 of 23 6/21/2005 8:41 AM Structure of the survey There were three different research perspectives in the survey: food safety issues, husbandry practices and antimicrobial resistance. The food safety part of the survey consisted of nine questions that were a mixture of open-ended questions, yes/no answers and Likert scales. The Likert scales were used to measure agreement with various statements and ranged from one, strongly agree, to six, strongly disagree. The results from the food safety component of the survey will be described in another report. The second part of the survey consisted of twenty-two questions that dealt with farm type, husbandry practices and antimicrobial use. These questions were a mixture of open-ended questions, selections from a list and blanks to fill in. The third part of the survey comprised fifteen questions regarding attitudes about antimicrobial use. Most of these questions used a Likert scale, however there were also three open-ended questions. The respondents were also asked for their age, where they routinely received information on new developments in science and agriculture and for any additional comments. Pretesting The food safety questions had been tested and used in previous surveys. The antimicrobial resistance questions were distributed to the members of the Ontario Agribusiness Association, at a Feed and Grain Association meeting in Toronto, Ontario, in February of Eighty-seven people answered the survey. Questions that were frequently left unanswered or caused confusion, as assessed by written comments, were rewritten or removed. The entire survey, for this study, was pretested by eight of the people that work in the offices of the OCA, resulting in minor changes to the questions. Data verification All data were checked for validity of entry against the original questionnaires and any errors were corrected. The responses for the number of 'cows'were checked to make sure that all of the cows, that were included in the analysis, were beef cows and not dairy cows. The surveys that were completed by people not involved in beef production were removed from the analysis. The number of respondents varied between questions, so the actual number of respondents is reported for each question. The numbers of answers to each question were compared to ensure that there was not a pattern to the response distribution. Data manipulation Several variables were derived from the answers to the questions in the survey to facilitate analysis of the data. The variable CVSICK was created by dividing the sum, of the number of calves, on a cow/calf farm, during the year prior to the survey, that the respondent reported treating for each illness listed (Table 1), by the total number of calves born on the cow/calf farm during the year prior to the survey. The variable TREAT was created by dividing the sum, of the number of beef animals, on the feedlot, during the year prior to the survey, that the respondent reported treating for each illness listed (Table 1), by the total number of beef animals on the feedlot, during the year prior to the survey. The variables CVSICK and TREAT therefore represent the fraction of animals that were treated on cow/calf farms and feedlots respectively. The median value of both CVSICK and TREAT was ten per cent. To facilitate data analysis, the variable THERAP was created and coded 1 if CVSICK or TREAT was ten per cent or higher and coded 0 if they were not. A variable ANIMALS was created by summing all of the beef animals on the farm. This included beef cows, calves born on the farm and calves, steers and heifers that were purchased. The variable AGE was created by recoding the answers to question F17, What is your age please? When the respondent answered 'old'or 'over 50', AGE was given the median value of all of the values that were given for F17, which was 54 and satisfies both descriptions, 'old'and 'over 50'.
4 4 of 23 6/21/2005 8:41 AM The variable NOTANTI was created by examining the responses to open-ended questions that asked specifically for types of antimicrobials. If the respondent included any products that were not antimicrobials in their responses they were coded 1 if not, they were coded 0. A variable FARMTYPE was created by recoding the answers to a question that asked what type of farm the respondent owned, or when necessary, checking to see what types of beef animals the respondent reported having. The coding was 1 for a feedlot, 2 for a cow/calf operation, 3 for a combination of feedlot and cow/calf, 4 for backgrounders and veal producers, 5 for a combination of feedlot and backgrounder and 7 for a combination of cow/calf and backgrounder. A farm that reported having a combination of cow/calf, backgrounder and feedlot was originally coded 6. These farms were recoded 3 because a cow/calf operator that also has a feedlot would have some component of backgrounding as well. Statistical Analysis The data were initially entered into a Microsoft Excel '98 spreadsheet (Microsoft Corporation, Redmond, Washington, USA) and converted into STATISTIX (Analytical Software, St. Paul, Minnesota, USA) and SAS (SAS Institute, Cary, North Carolina, USA) files for statistical analysis. Lists of the answers to the open-ended questions were compiled and the percentage of respondents giving each answer was calculated. The means, minimum and maximum values, standard deviations and confidence intervals were determined for the continuous variables. The frequency of each response in the Likert scales and categorical variables were determined. Two-way comparisons between variables were performed using t-tests, Wilcoxon comparisons, chi square goodness of fit tests, Spearman correlations, Pearson's correlations, odds ratios and scatter plots. Associations were considered significant when p < To determine if respondents were representative of the Ontario beef industry, comparisons of survey results were made with Ontario census data from The expected number of beef animals in various counties of Ontario was based on data obtained from the Ontario Ministry of Agriculture, Food and Rural Affairs (15). The actual total numbers of beef cows, steers and heifers, in each county of Ontario, as reported on the survey, were compared to the expected numbers from OMAFRA census data, using a one-way, standard chi-square, goodness-of-fit test. Statistical associations between the dependent variable THERAP (1,0) and potential independent variables were determined. Two different approaches were used to build multivariable models. The first approach was to examine two-way comparisons, using either a chi-square or a t-test as appropriate, for each independent variable that could be associated with the outcome THERAP (Appendix 3 and 4). Only variables that had a minimum of 300 responses were included in the analysis so that the model would represent as many of the respondents as possible. Any two variables that demonstrated a Pearson s correlation of 0.75 or greater were not used in models together to avoid collinearity problems (16). Variables significant at a p-value < 0.15 were made available to a logistic model and then removed using backward elimination until all remaining variables had a p-value < The variables ANIMALS, FARMTYPE and AGE were forced into the model because they were considered potential confounders. The second approach was to make all variables, with 300 or more responses, available to the model, followed by backward elimination removing the least significant variables (p > 0.20). Non-significant variables were only left in the model, if removing them resulted in at least a ten per cent change in the coefficient of at least one other variable. Some interaction terms, of the variables that remained in the model, were added to the model but none were found to be significant so they were removed. Results
5 5 of 23 6/21/2005 8:41 AM Response Rate Responses were obtained to 632 (4%) of the 17,000 questionnaires that were sent with Ontario Beef magazine. Of these, 23 were completed and returned by farmers involved in dairy, but not beef production and 22 were completed and returned by people not actively involved in either dairy or beef production. The 587 remaining surveys that were completed and returned by beef producers represented 8.0% of the marketed steers, 7.5% of the marketed heifers and 4% of the beef cows in Ontario. Types of Beef Producers There were 341 respondents that were involved in cow/calf production and feedlot producers accounted for 106 respondents. Eighty-nine respondents identified themselves as having a combination of the above. In addition, 7 farmers classified themselves as either backgrounders or veal producers and fourty-four were a combination of feedlot and backgrounders. The feedlot producers were divided into those that bought calves, heifers or steers. The numbers of calves bought ranged from , with a median of 55 calves. 11,050 calves of the total 18,961 calves in the survey (58.3%) being raised on the seven largest farms (Figure 1). A total of 11,534 heifers were raised on 40 study farms, with a median of 80 heifers and a range of The largest four farms raised 7,200 heifers, or 62.4% of the total reported in the survey. On 65 farms, 23,518 steers were reported demonstrating a range from 1-6,000 animals and a median of 125 steers. The largest seven farms accounted for 14,200 steers or about 60.4% of the total reported on the survey. In 1999 the beef herd in Ontario comprised 434,000 steers and heifers (15). The mean size of the cow/calf farms was 40.1 cows with a median of 20 cows. The range in size was from cows on 419 study farms. The beef herd in Ontario is comprised of 483,000 beef cows and heifers for breeding (15). A total of 17,077 beef cows were represented by the survey. The majority of the farms (70.6%) reported having one hundred beef animals or less on their farms. Conversely, half of the animals (49.5 %) were raised on the 26 largest farms (5% of the total number of farms). The proportion of each type of animal, for each county, as reported in the surveys, was compared to the expected proportion in each county, as determined by OMAFRA census data (15). There was a very significant relationship, (c 2 = 30,475.26, df=14 and p-value < for steers, c 2 = 27,515.09, df=14 and p-value < for heifers and c 2 = 15,458.76, df=14 and p-value < for beef cows) demonstrating that the actual number of animals in each county was very different from the expected number of animals. Illness and treatment As described in Table 1, the most common reported illness of a calf on a cow/calf farm was diarrhea, followed by respiratory illness. In finisher operations the most common reported illness was respiratory illness (shipping fever) followed by foot rot and pinkeye. Approximately one third of the study farms reported treating 5% or less of their animals. There was no significant association between mortality and the number of animals on the farm. Morbidity was significantly associated with the number of animals of the farm (p=0.039) when examined within a multivariable model. Attitudes about antimicrobial use Most of the respondents (94.2%) agreed or strongly agreed that bacterial infections cause serious economic losses. One hundred and eighty-four respondents (40.6 %) felt that the cost of antimicrobial use was moderately important in determining whether or not to use the antimicrobial, while 13.4% stated that it was extremely important.
6 6 of 23 6/21/2005 8:41 AM Over 53% said that they had heard of hygiene practices designed to decrease the need for antimicrobials. In an open-ended question, over 80 distinct practices were suggested as ways to reduce the use of antimicrobials. Some of the more frequent responses included: keep environment clean and dry (134 responses), calve in clean area or on grass (47 responses), provide a clean water source (28 responses), use disinfectants (27 responses), provide good ventilation (27 responses) or fresh air (15 responses) and keep the herd closed (10 responses). Most of the respondents (94.4%) agreed that it is important to reduce the use of antimicrobials. Only 2 people (0.4%) disagreed. A slightly smaller number (92.1 %) agreed that it was necessary to reduce the use of antimicrobials in order to slow the development of resistance in bacteria. Only 5 people (0.9%) strongly disagreed with this statement. Thirteen per cent of the respondents felt that reducing antimicrobial resistance was not the main reason to reduce antimicrobials. About half (45.4%) of the respondents were not confident that pharmaceutical companies would produce new antimicrobials by the time that they were needed to combat antimicrobial resistant bacteria. A majority (78.0%) agreed, to some extent, that a controversy in the media would change their current antimicrobial use. The respondents were split in their beliefs with 48.1% agreeing and 51.9% disagreeing that the cost of antimicrobials in feed was low. If an increase in profits could result from a decrease in the use of antimicrobials 96.0% agreed (87.4%-strongly) that they would decrease their use. Almost all of the respondents (95.4%) felt that the use of antimicrobials in feed could lead to the development of antimicrobial resistance in bacteria. Antimicrobial use in feed and water As outlined in the Table 2, 269 (61.0%) of the 441 respondents reported using a feed additive. Ionophores (44.2%) were by far the most common additive with less than ten per cent of the respondents listing one of antimicrobials, vitamins and minerals or anthelmintics. In a separate question that asked specifically about ionophores, 44.6% reported using them. Of these, 17.4% reported using them in % of their feeds. The responses to the two questions asking about the use of ionophores in feed were negatively correlated and not significantly different (Spearman rank= , p=0.239). Alternatively, the responses to the question asking how often the respondent used antimicrobials in the feed, were significantly different from the responses to the question asking how often ionophores only, were used in the feed (Spearman rank = , p <0.01) suggesting that respondents did not consider ionophores to be antimicrobials. In an open ended question, a total of nine feed antimicrobials were reported (the survey did not distinguish therapeutic use from other uses) by 38 (8.6%) of respondents (n=441) (Table 2). The answer premix was given as a feed additive. The Compendium of Veterinary Products (17) lists several products that could be called a premix. A premix is a generic term that describes a product containing a range of ingredients that may include vitamins, minerals, salts, antimicrobials and protein. It was not assumed that a premix contained an antimicrobial in the analysis, which may have resulted in an underestimate of how many feeds contained antimicrobials. Also, some custom manufactured feeds already contain antimicrobials and they are not represented in these data. When asked, in a separate question, if they used feed medicated with an antimicrobial, 61 (12.6%) respondents (n=484) said that they used medicated feed at least some of the time. Fourty-three of these or 70.5% of those who medicated feed, said that they did it less than 5% of the time, which suggests that they are not using it continuously for growth promotion. A majority (81.5%) of all study respondents either only slightly agreed or did not agree that there was a significant increase in average daily gain resulting from the use of antimicrobials in feed. Prescriptions from a veterinarian were used for medicated feed by 41 (10.0%) of the respondents (n=411). Some of the reasons for adding antimicrobials to the feed included: recommended by a veterinarian, unfavorable weather, routinely given on arrival and that they were given as needed. Fourty-two (15.5%) respondents (n=270) reported adding antimicrobials to the water. This was distributed between sulfonamides (4.0%) and tetracyclines (12.2%).
7 7 of 23 6/21/2005 8:41 AM Antimicrobials given individually The injectable antimicrobials that the farmers reported using are listed in Table 3. Half reported using oxytetracycline (50.0%) or penicillin (48.5%); fewer reported using tilmicosin (27.2%), trimethoprim sulfadoxine (23.0%) or florfenicol (14.6%). Less than five per cent of the farmers reported using the other ten injectable antimicrobials. Lincomycin and enrofloxacin are not approved for use in cattle in Canada (17) but were reportedly used on some farms. In the open response questions that asked what antimicrobials were being used, 50.4% of the respondents included an inappropriate injectable product, like a vaccine and in the topical or oral question that specifically asked for types of antimicrobials, 24.4% included an inappropriate product, like an anthelmintic. Injectable antimicrobials were used prophylactically by 28.5% of respondents (n=228). The answer to this question was not significantly associated with whether or not the respondent reported having a regular veterinarian. Only 3.2% of the farmers (n=587) treated the entire herd when only some animals were sick. Fifty per cent of the respondents (n=495) were classified as having a higher than median level of therapeutic use (³10% of animals treated per annum). In two-way comparisons, being classified as having low therapeutic use was not significantly related to having a regular veterinarian (p = 0.056). High therapeutic use tended to be associated positively with the use of injectable antimicrobials for metaphylactic treatment (p = 0.015), with disagreeing that it is important to decrease the use of antimicrobials (p = 0.022) and with disagreeing that it is possible for bacteria to become increasingly resistant to antimicrobials (p = 0.032). Logistic Regression Modeling of Risk Factors for Higher than Median Levels of Therapeutic Treatment Multivariate models were built to adjust for confounding and assess the interrelationships among multiple independent variables and the dependent variable THERAP. The two logistic models that were constructed are presented in Tables 4-7. Some interaction terms, of the variables that remained in the model, were added to the model but none were found to be significant so they were removed. Smaller farm size (in terms of the number of animals on the farm), the percentage of feed that was medicated with ionophores and agreeing that there was a significant average daily gain from feeding antimicrobials were all positively associated with THERAP in both models. Calling a veterinarian when there was a problem, such as a difficult calving, was negatively associated with THERAP in model 2. Discussion This survey points to several areas of interest with regards to prudent use guidelines. The inclusion, by survey respondents, of products that were not actually antimicrobials, on the questions asking specifically for antimicrobials, can be interpreted in at least two ways. Either approximately half of the farmers, in this survey, did not know the difference between antimicrobials and other injectable products, or the farmers were trying to be inclusive by naming all injectable products. A conservative approach assumes the former, and prudent use guidelines need to start with an explanation of what products are and are not antimicrobials. It would be counter-productive for farmers that were concerned about antimicrobial use, to forgo vaccinations because they simply misunderstood the difference. This might actually increase the amount of illness on their farms, resulting in the need for more antimicrobial use. One of the more controversial uses of antimicrobials is to enhance feed efficiency (2, 4, 6, 7). In-feed antimicrobials can result in an increase in growth rate of up to 15% and a reduction in feed intake of up to 9% in beef (18, 19). This is less than the potential 30% increase in growth rate and the potential 12% decrease in feed intake, that can be seen in young pigs (20, 21), or the 26% increase in growth rate and up to a 5% decrease in feed intake, that can be seen in broilers (20, 22). Most of the respondents did not agree that there was a significant increase in average daily gain resulting from the use of antimicrobials in feed. The increase in feed efficiency that is seen in beef may not
8 8 of 23 6/21/2005 8:41 AM consistently off set the cost of the antimicrobials in the feed, as much as it does for some other species. The respondents did not consider ionophores to be antimicrobials. Approximately half of the respondents treated less than ten per cent of their animals. This is lower than the average morbidity of 21-29% reported in 1980 for the Ontario beef-calf feedlots (23). Five per cent or less of the animals were treated on roughly a third of the farms. Research examining some of the reasons for the large variation in the number of animals, that farmers reported treating, would be valuable. Hygienic practices, like those recommended by some of the respondents, have been shown to reduce the number of animals that need to be treated (24). Reducing handling as much as possible, especially immediately after arrival, providing adequate bunk space and pen maintenance have been shown to reduce morbidity (24). Farmers that reported treatment rates below the median of 10% per annum, were associated with larger farms, called a veterinarian for calving difficulties and used ionophores in the feed, but did not use other in-feed antimicrobials for growth promotion. Ionophores are associated with health benefits, in addition to the enhanced feed conversion, which makes them a good management practice (25-27). The willingness to consult a veterinarian, for calving difficulty, is consistent with a willingness to seek expert advice on other veterinary issues, such as prudent use of antimicrobials. Alternatively, only cow/calf producers would encounter calving difficulties and this question may simply target this group. Since half of the animals were raised on the largest 5% of the study farms, most of the animals, represented in this survey, were on farms that reported a treatment level of less than 10% per annum and they were also less likely to use antimicrobials (not including ionophores) for growth promotion. The most commonly reported ailments were respiratory disease, diarrhea, foot rot and pink eye. Respiratory disease is the most common health problem for the beef industry (28, 29) and foot rot is the most common cause of lameness (30-32). Diarrhea is also a common condition and it can be caused by numerous etiologies (33). The severity, cause and treatment vary with age (34, 35). Guidelines could specifically address how to minimize the occurrence of these conditions and specifically deal with the best way to treat them or reduce their likelihood, thereby reducing antimicrobial use. For example, vaccinations (23) or feeding of colostrum from vaccinated cows (35), maintaining clean, dry pens (24), and decreasing stress (23, 36) are conservative ways to reduce some of these illnesses. All of the farms in the survey performed routine husbandry practices such as deworming and vaccinating, which provides a framework to add some of the other procedures if the farm would benefit from them. Less than one per cent of the respondents reported using enrofloxacin, a fluoroquinolone. In Canada, this constitutes extra label drug use (17). Extra label drug use is only legal in Canada under the guidance of a veterinarian in an established veterinary-client relationship. Fluoroquinolones are considered to be essential as a drug of last resort for some infections in human medicine (37-40). Prudent use guidelines could include a statement encouraging farmers to only use these drugs, under a veterinary prescription, if they are essential. Over 90% of the respondents agreed that there was a need to reduce antimicrobial use and that bacterial infections can cause serious economic losses. About half of the farmers did not agree that pharmaceutical companies would solve the problem for them by developing new antimicrobials. A majority of the respondents were also worried about negative media coverage and admitted that this would cause them to re-evaluate their current use of antimicrobials. Almost all of the respondents felt that in-feed antimicrobials led to antimicrobial resistance. With this questionnaire design it was not possible to meet all of the requirements of the Total Design Method (TDM) suggested by Dillman (41), such as sending up to three follow-up letters to the individuals that did not reply to the survey and this resulted in a lower return rate. Erdos said that if no mailing list is available, it is impossible to conduct a reliable mail survey that is representative of the population being surveyed (42). A better response rate has been documented with an actual stamp on the return envelope (41, 43). This was not done, in the present study, because the cost was prohibitive. The survey did have the endorsement of the OCA and this may have increased or decreased the response rate (41). Distributing this survey with the Ontario Beef magazine preserved most of the benefits of a traditional mail survey such as allowing a wide distribution, giving each farmer an equal opportunity to reply and sufficient time to answer the questions and allowing
9 9 of 23 6/21/2005 8:41 AM centralized control (41, 42, 44). There was also no interviewer bias, as in personal or telephone interviews. The anonymous nature of the mailed survey may have increased the chances of unbiased replies especially if farmers were embarrassed about how many animals they treated or their use of in-feed antimicrobials (41). It also allowed farmers to check for information such as the names of the medications or how many animals were treated (42, 43). A deadline was given to add a note of urgency. This may have reduced the response rate however, because people may not bother to fill it out after the deadline (42). The largest drawback of this type of distribution was the bias introduced by self-selection. The reason that people volunteer to answer a survey is often related to the topic of the survey (45). This means that their responses may not be typical of the responses of all of the beef producers in Ontario. The four per cent response rate makes it highly improbable that the results of this survey have much external validity. Robinson concluded that the reasons for non-response, in his survey, were unrelated to the topic of the survey (44). Over 17,000 copies of the survey went out but it was difficult to determine the actual denominator for the survey. It was not possible to know which surveys were received by people that were actively involved in beef production or how many people received multiple copies. A desirable first mailing response rate is 40-57% (42, 43). A mail survey can be considered reliable if it has a minimum of 50 per cent response, unless it demonstrates with some form of verification that the non-respondents are similar to the respondents (42). Neither of these conditions were met with this survey. One attempt to compare the survey respondents to all of the beef producers in Ontario was to compare the distribution of the beef animals per county, represented in the survey, to the distribution of beef animals per county in the census data. There was no relationship. A survey could have been based on the clients of a teaching hospital. The response rate could have been improved. Unfortunately, the prior association with the university would also have a selection bias. The individuals that had a prior association with the university could have opinions about antimicrobials that may differ from the remainder of the beef producers in Ontario in unpredictable ways. The external validity of such a survey may not have been any better than with the current study. The acceptability of the response to the survey has to be judged by the nature of the goals, which the researcher sets for a particular piece of research (42). The purpose of this questionnaire was to acquire a knowledge base that would facilitate the development of prudent use guidelines for the use of antimicrobials in beef. The survey respondents were presumably those individuals that were interested in this topic but represent a restricted population that may not be representative of all of the beef producers in Ontario. They are, however, the initial target population for the prudent use guidelines and the associations that were derived from this survey are internally valid (45). The answers to this survey can arguably be directly applied to the initial farmers that will be called upon to evaluate the first set of guidelines. To that end, the biases inherent in the self-selection are acceptable as long as no attempt is made to generalize these results to all beef producers in Ontario. The variation in treatment rates between the farms suggests that there are many areas that need to be investigated to account for these differences. Some differences may have occurred only in the year of study. Other differences, the ones that are of more interest to the investigator, are the ones that are consistently associated with lower morbidity. The survey illustrated when and how antimicrobials are currently being used and which practices the farmers considered less important. Prudent use guidelines need to concentrate on explaining which products are and are not antimicrobials, when they should be used and the husbandry practices that are most likely to decrease their unnecessary use. Acknowledgments Financial assistance for this research was provided by Ontario Agricultural Adaptation Council, OMAFRA animals (beef) program and the Ontario Cattlemen s Association. We thank the members of the Ontario Cattlemen s Association for their participation in this survey and William Sears for statistical assistance.
10 10 of 23 6/21/2005 8:41 AM References 1. World Health Organization. Global strategy for the containment of antimicrobial resistance, Anonymous. A proposed framework for evaluating and assuring the human safety of the microbial effects of antimicrobial new animal drugs intended for use in food-producing animals. U.S. Food and Drug Administration s Center for Veterinary Medicine, Canadian Veterinary Medicine Association. Guidelines on The Prudent Use of Antimicrobial Drugs in Animals, Anonymous. Evaluation of the Human Health Impact of the Microbial Effects of Antimicrobial New Animal Drugs Intended for Use in Food- Producing Animals. Food and Drug Administration. U.S. Federal Register, 1998; 63: European Commission. Opinion of the Scientific Steering Committee on Antimicrobial Resistance. DGXXIV, Consumer Policy and Consumer Health Protection, out50_en.html 6. Aarestrup FM, Bager F, Jensen NE, Madsen M, Meyling A, Wegener HC. Resistance to antimicrobial agents used for animal therapy in pathogenic-, zoonotic- and indicator bacteria isolated from different food animals in Denmark: a baseline study for the Danish integrated antimicrobial resistance monitoring programme (DANMAP). APMIS 1998, 106, European Commision. Council Regulation (EC) No 2821/98 of 17 December 1998 amending, as regards withdrawal of the authorisation of certain antibiotics, Directive 70/524/EEC concerning additives in feedingstuffs, _ html 8. Anonymous. Human Health Impact of Fluoroquinolone Resistant Camplylobacter Attributed to the consumption of Chicken. U.S. Federal Register, 2000; 65: Roberts MC, Chung WO, Roe D et al. Erythromycin-resistant Neisseria gonorrhoeae and oral commensal Neisseria spp. carry known rrna methylase genes. Antimicrob Agents Chemother 1999; 43: Levy SB. Microbial rsistance to antibiotics. An evolving and persistent problem. Lancet 1982; 10: Chaslus-Dancla E, Pohl P, Meurisse M, Marin M, Lafont JP. High genetic homology between plasmids of human and animal origins conferrring resistance to the animoglycosides gentamicin and apramycin. Antimicrob Agents Chemother 1991; 35: Johnson AP, Burns L, Woodford N et al. Gentamincin resistance in clinical isolates of Escherichia coli encoded by genes of veterinary origin. J Med Microbiol 1994; 40: Karam GH, Hefner JE. Emerging issues in antibiotic resistance in blood-borne infections. Am J Respir Crit Care Med, 2000; 162: Anonymous. Prudent use of antibiotics in animals; world farmers, veterinarians and animal health industry commit to new global principles. The World Veterinary Association, Ontario Ministry of Agriculture and Food and Rural Affairs. Website: html 16. McNeil D. Epidemiological research methods. New York: John Wiley and Sons, Canadian Animal Health Institute. Compendium of Veterinary Products. Hensall, Ontario: North American Compendiums Ltd, Aarestrup FM. Occurrence, selection and spread of resistance ot antimicrobial agents used for growth promotion for food animals in Denmark. APMIS 2000; 108: supp. 101.
11 11 of 23 6/21/2005 8:41 AM 19. Rogers JA, Branine ME, Miller CR et al. Effects of dietary virginiamycin on performance and liver abscess incidence in feedlot cattle. J Anim Sci 1995; 73: DeCraene A, Vianese J. Economic effects of technology in agricultrue. Do performance enhancers for animals benefit consumers? Ghent: University of Ghent, Stahly TS, Cromwell GL, Monegue HJ. Effects of the dietary inclusion of copper and (or) antibiotics on the performance of weanling pigs. J Anim Sci 1980; 51: Proudfoot FG, Hulan HW, Jackson ED, Salisbury CDC. Effect of lincomycin as a growth promoter for broiler chicks. Br Poultry Sci 1990; 31: Martin SW, Meek AH, Davis DG et al. Factors associated with mortality in feelot cattle: The Bruce county beef cattle project. Can J Comp Med 1980; 44: Morisse JP, Cotte JP. Evaluation of some risk factors in bovine salmonellosis. Vet Res 1994; 25: Duffield TF, Bagg RN. Use of ionophores in lactating dairy cattle: A review. Can Vet J 2000; 41: Lowe LB, Ball GJ, Carruthers VR et al. Monensin controlled-release intraruminal capsule for control of bloat in pastured dairy cows. Aust Vet J 1991; 68: Stock RA, Laudert SB, Stroup WW et al. Effect of monesin and tylosin combination on feed intake variation of feedlot steers. J Anim Sci 1995; 73: National Animal Health Monitoring System. Feedlot Management Practices. Animal and Plant Health Inspection Service, U.S. Department of Agriculture, US Department of Agriculture. Cattle and calves death loss. National Agricultural Statistical Service, Greenough PR, MacCallum FJ, Weaver AD. Lameness in Cattle. Edinburgh: Oliver & Boyd, Merrill JK, Morck DW, Olson ME, Dick CP. Evaluation of the dosage of tilmicosin for the treatment of acute bovine footrot. Bovine Practitioner 1998; 33: Dawson JC. Treatment of digital dermatitis. Cattle Practice 1999; 7: Jones JR. Peri-weaning calf diarrhoea syndrome: findings to date and further investigation. Cattle Practice 2000; 8: Blowey RW. Chronic peri-weaning diarrhoea -The clinical syndrome. Cattle Practice 2000; 8: Le Rousic S, Klein N, Houghton S, Charleston B. Use of colostrum from rotavirus-immunised cows as a single feed to prevent rotavirus-indusced diarrhoea in calves. Vet Rec 2000; 147: Lofgreen GP, El Tayeb AE, Kiesling HE. Millet and alfalfa hays alone and in combination with high eneregy diet for receiving stressed calves. J Anim Sci 1981: 52: Conte JE. Manual of antibiotics and infectious diseases. Baltimore: Williams and Wilkins, Anonymous. FDA panel urges restricted use of fluoroquinolones in animals. ASM News 1994; 60: Beam Jr TR. Fluoroquinolones in animal feeds. ASM News 1994; 60: Wilcox MH, Spence RC. Quinolones and Salmonella gastroenteritis. J. Antimicrob Chemother 1992; 30: Dillman DA. Mail and Telephone Surveys: The Total Design Method. New York: Wiley, Erdos PL. Professional Mail Surveys. Malabar: Robert E. Krieger Publishing Company Inc., Kelsey JL, Whittemore AS, Evans AS, Thompson WD. Methods in Observational Epidemiology. New York: Oxford University Press, Robinson RA, Agisim P. Making mail surveys more reliable. J Marketing 1951; 15: Torrence ME. Understanding epidemiology. Toronto: Mosby, 1997.
12 12 of 23 6/21/2005 8:41 AM Table 1. The median number and range of each type of animal that required treatment, for each illness, on each type of farm, as reported by producers Diarrhea Septic Respiratory Calf Pink Eye Wound Urinary Tract Foot rot Arthritis Illness Septicemia Infections Infections median median median median median median median median range range range range range range range range Cow/calf * * * * Feedlot * 0 * * * *not applicable Table 2. Feed additives reportedly used by Ontario beef producers Feed Percentage of respondents Antibiotics Percentage of respondents additives reporting use reporting use ionophores 44.2 tylosin phosphate 2.7 lasalocid 9.3 chlortetracycline 1.8 decoquinate 6.6 salinomycin sodium 1.4 melengestrol acetate 5.9 tetracycline 1.1 minerals and salt 5.4 spectinomycin <1.0 vitamins 3.2 penicillin <1.0 Other additives sulfonamides <1.0 neomycin sulfate <1.0
13 13 of 23 6/21/2005 8:41 AM sodium diacetate 2.5 oxytetracycline <1.0 prepared premixes 1.8 levamisole <1.0 No feed additives 39 ethylenediamine dihydoiodide <1.0 molasses <1.0 selenium <1.0 sodium bicarbonate <1.0 <1.0
14 14 of 23 6/21/2005 8:41 AM Table 3. Injectable antibiotics reportedly used by Ontario beef producers Injectable antibiotics Percentage of respondents reported reporting use oxytetracycline 50.0 penicillin 48.5 tilmicosin 27.2 trimethoprim sulfadoxine 23.0 florfenicol 14.6 tetracycline 5.0 lincomycin spectinomycin 3.1 ceftiofur hydrochloride 4.2 ampicillin, sulbactam benzathine 3.8 spectinomycin 2.7 penicillin/streptomycin 1.2 No injectable antibiotics used 7.5 tylosin <1.0% gentamicin <1.0% erythromycin <1.0% enrofloxacin <1.0%
15 15 of 23 6/21/2005 8:41 AM Table 4. Definitions of variables included in the multivariable regression model with the outcome variable THERAP Variable Definition ANIMALS Total number of beef animals on the farm FARMTYPE Category of beef farm (coded 1 if feedlot, 2 if cow/calf, 3 if a combination of cow/calf and feedlot, 4 if backgrounder or veal and 5 if feedlot and backgrounder AGE The age of the respondent BJ Percentage of feed that was medicated with ionophores (coded on a Likert scale from 1 if none was medicated to 6 if % was medicated) F12 There is a significant gain from in-feed antibiotics (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) CALVING A veterinarian is consulted to assist with calving (coded 1 if yes and 0 if no) Table 5. Two-way comparisons between the dependent variable THERAP and the independent variables Variables p-value ANIMALS BJ C D D E WORMERS Variables p-value CALVING F F QF
16 16 of 23 6/21/2005 8:41 AM C FEED REDUCE
17 17 of 23 6/21/2005 8:41 AM Table 6. Multivariable Model 1. Regression coefficients, odds ratios and significance levels for variables included in the model of THERAP derived from variables that initially had unconditional associations Variable Coefficient Odds Ratios SE(b) p-value ANIMALS FARMTYPE AGE BJ F Table 7. Multivariable Model 2. Regression coefficients, odds ratios and significance levels for variables included in the model of THERAP selected by making all variables available to the model Variable Coefficient Odds Ratios SE(b) p-value ANIMALS BJ F CALVING Appendix 2. The column that accompanied the survey in the October 1999 issue of Ontario Beef Magazine. OCA Producer Survey Food safety is a top priority for Ontario cattlemen. Through the Quality Starts Here program and other aggressive, on-farm initiatives, producers are constantly striving to deliver a high-quality, safe product to the slaughterhouses and, ultimately, to kitchen tables, at home and abroad. OCA has recently undertaken three separate, but related, projects with researchers at the University of Guelph, led by Doug Powell and Scott McEwen, to further enhance the safety of Ontario beef in general, and to specifically help manage the responsible use of antimicrobials on farms in Ontario. As part of the initial work, a survey is being mailed with this issue of Ontario Beef. Yes, Ontario producers have been excessively surveyed, often with a lack of clear goals. But you are being encouraged by the OCA executive, which have reviewed and revised the survey to best meet the needs of Ontario producers, to complete and return the
18 18 of 23 6/21/2005 8:41 AM surveys at your earliest convenience -- make that Nov. 1, The results will form the basis for the development of prudent-use guidelines to better use antimicrobials on farms, and, perhaps more importantly, send a clear message to consumers that Ontario cattlemen are serious about food safety. Appendix 3. Definitions of variables used in two-way tests of association with the outcome THERAP Variable Definition THERAP High therapeutic use was defined as a farm that treated ten per cent of the animals on the farm or more with antibiotics (coded 1 if high use and 0 if not) ANIMALS Total number of beef animals on the farm FARMTYPE Category of beef farm (coded 1 if feedlot, 2 if cow/calf, 3 if a combination of cow/calf and feedlot, 4 if backgrounder or veal, 5 if feedlot and backgrounder and 7 if cow/calf and backgrounder) BJ Percentage of feed that was medicated with ionophores (coded on a Likert scale from 1 if none was medicated to 6 if % was medicated) CCCC New hygiene practice will reduce the need forantibiotics (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) QC2 The number of calves born that developed an illness that required medication C3 The percentage of calves born that were never sold due to illness or death D2 The percentage of cattle that developed an illness that required medication D5 Injectable antibiotics are used prophylactically (coded 1 if yes and 0 if no) QE2 Individual I.D. tags are used (coded 1 if yes and 0 if no) E3 The entire herd is treated when some animals are sick (coded 1 if yes and 0 if no) E5 Regular veterinarian visits farm (coded 1 if yes and 0 if no)
19 19 of 23 6/21/2005 8:41 AM VISIT Frequency of veterinary visits (coded 1 if weekly, 2 if monthly, 3 if 5-6 times per year, 4 if 3-4 times per year, 5 if twice per year, 6 if once per year and 7 if less than once per year) VACCINAT Vaccination is routinely performed (coded 1 if yes and 0 if no) CASTRATI Castration is routinely performed (coded 1 if yes and 0 if no) DEHORNIN Dehorning is routinely performed (coded 1 if yes and 0 if no) HORMONES Hormones are routinely used (coded 1 if yes and 0 if no) WORMERS Wormers are routinely used (coded 1 if yes and 0 if no) CALVING A veterinarian is consulted to assist with calving (coded 1 if yes and 0 if no) HERDHEAL A veterinarian is consulted for herd health (coded 1 if yes and 0 if no) SICK A veterinarian is consulted for a sick animal (coded 1 if yes and 0 if no) OUTBREAK A veterinarian is consulted for an outbreak (coded 1 if yes and 0 if no) VISITS A veterinarian visits regularly (coded 1 if yes and 0 if no) PURCHASE A veterinarian is consulted to purchase antibiotics (coded 1 if yes and 0 if no) AGE The age of the respondent QF3 It is possible for bacteria to become more resistant to antibiotics (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F4 Bacterial infections are expensive (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F5 The cost of antibiotics is important (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F6 Aware of hygiene practices designed to decrease the need for antibiotics (coded 1 if yes and 0 if no) F7 It is necessary to decrease antibiotic use to slow antibiotic resistance (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F8 It is necessary to decrease antibiotic use (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree)
20 20 of 23 6/21/2005 8:41 AM F9 Clear label instructions are important (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F10 Companies will develop new antibiotics before antibiotic resistance becomes a problem (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F11 Controversy over antibiotic use would cause a change in current antibiotic practices (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F12 There is a significant gain from in-feed antibiotics (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F13 The cost of antibiotics is low (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F14 The respondent would decrease antibiotic use if it increased profits (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F15 The use of in-feed antibiotics can lead to antibiotic resistance (coded on a Likert scale from 1 if strongly agree to 6 strongly disagree) F17 Age of the respondent QF2 Prescriptions have been used to add antibiotics to feed (coded 1 if yes and 0 if no) C The percentage of feed that is medicated with an antibiotic (coded on a Likert scale from 1 if none was medicated to 6 if % was medicated) FEED Antibiotics are used in feed (coded 1 if yes and 0 if no) A7 The farmer has implemented programs to improve the safety of beef (coded 1 if yes, 2 if no and 3 if unsure) A2 The food in the supermarket is safe (coded on a Likert scale from 1 if completely confident to 5 not sure) RESISTAN 1 if yes and 0 if no) It is important to reduce the use of antibiotics in order to slow the development of antibiotic resistance in bacteria (coded REDUCE It is important to reduce the use of antibiotics (coded 1 if yes and 0 if no)