The primary objectives of this study were to: (a) collect on-farm antimicrobial use (AMU) data in fattener pigs employing two questionnaire-based surveys; (b) assess different quantitative measures for quantifying AMU in fattener pigs; (c) compare AMU in fattener pigs between two different management systems producing finishers: farrow-to-finish (FtF) farms versus finisher farms. Two questionnaires were designed both containing five groups of questions focused on the responder, the farm and AMU (eg, in-feed, in-drinking water and parenteral); both surveys were carried out by means of personal face-to-face interviews. Both surveys started with a sample size of 108 potentially eligible farms per survey; nevertheless, finally 67 finisher farms and 49 FtF farms were recruited. Overall percentages of animals exposed to antimicrobials (AM) were high (90 per cent in finisher farms and 54 per cent FtF farms); colistin (61 per cent and 33 per cent) and doxycycline (62 per cent and 23 per cent) were the most common AMs, followed by amoxicillin (51 per cent and 19 per cent) and lincomycin (49 per cent), respectively. Questionnaire-based surveys using face-to-face interviews are useful for capturing information regarding AMU at the farm level. Farm-level data per administration route can be used for comparative AMU analysis between farms. Nevertheless, for the analysis of the putative relationships between AMU and AM resistance, measures based on exposed animals or exposure events are needed.
- Accepted July 11, 2012.
- British Veterinary Association
Statistics from Altmetric.com
Bacterial antimicrobial resistance (AMR) is one of the main global, health problems of the 21st century, as well as being a multifactorial phenomenon (Torres and others 2010). Antimicrobial use (AMU) has consistently been found as one of the main risk factors for the selection of AMR bacteria in both humans (García Rey and others 2006) and animals (Chauvin and others 2002, Rosengren and others 2007, McEwen and others 2008, Varga and others 2009).
AMU is a complex issue including not only the amount of antimicrobials (AM) employed, but also other key parameters, like specific AMs employed, administration route, dosage, duration and the population exposed. In food animals, these parameters must be measured with an appropriate epidemiological framework, accounting for animals living together in the same facilities, or a batch for slaughtering purposes.
Although different recommendations have been published (Nicholls and others 2001, Grave and others 2006), compilation of animal-related AMU data is complicated owing to different reasons (Carnevale and Shryock 2006). For example, AMU data can be obtained at different levels (eg, national, regional, local, farm) and different sources (eg, animal health pharmaceutical companies, pharmacists, veterinarians or farmers) and methodologies (Chauvin and others 2001, Nicholls and others 2001, Carnevale and Shryock 2006, Grave and others 2006). Other problems are the controversial relationships between the sales, consumption and exposure to AMs. Despite different recognised usage measures, especially weight of active compound (WAC), animal daily dose, or prescribed daily dose, among others (Jensen and others 2004, Chauvin and others 2008), a consensus has not been reached on the interpretation of this data (Grave and others 2004, Carson and others 2008).
At a national level, overall sales can be obtained (Nicholls and others 2001, Grave and others 2006) and easily transformed into WAC. Indeed, some European countries (eg, Scandinavia) have a long experience in the monitoring of AMU and regularly publish the sales figures of AMs at a national level (eg, ‘Danish Programme for surveillance of AM consumption and resistance in bacteria from animals, food and humans’ (DANMAP) 2009; ‘Swedish Veterinary Antimicrobial Resistance Monitoring’ (SVARM) 2010; ‘Usage of Antimicrobial Agents and Occurrence of Antimicrobial Resistance in Norway in animals and humans’ (NORM/NORM-VET 2010 2011); ‘Monitoring of Antimicrobial Resistance and Antibiotic Usage in Animals in the Netherlands’ (MARAN) 2009; or ‘Sales survey of Veterinary Medicinal Products containing Antimicrobials in France’ (ANSES 2011).
In 2009, the EC requested the European Medicines Agency (EMA) (2010) to develop a harmonised approach for the collection and reporting of data based on national sales figures. To carry out this mandate, EMA launched the European Surveillance of Veterinary Antimicrobial Consumption project in 2010. A preliminary report has been published summarising the previously existing national information (Grave and others 2010) as well as a comparative analysis of sales in nine European countries (EMA 2011).
The Spanish Medicines Agency has collected sales data of AMs relating to 2009 from the animal health pharmaceutical companies (Agencia Española de Medicamentos y Productos Sanitarios 2011). Nevertheless, this data source does not allow the partitioning of data according to animal species.
The production of pig meat was 22.0 million tonnes in the European Union (EU)-27 in 2010, considerably higher than the corresponding figures for poultry (11.7 million tonnes, 2009), cattle (7.9 million tonnes) or sheep (0.7 million tonnes) (Eurostat European Commission 2011), with Spain ranking second among EU countries in pork production. In 2010, Spain had almost 100 000 pig farms (Subdirección General de Productos Ganaderos 2011) classified into 11 types (Real Decreto 324/2000, 2000). Among then, finisher farms (ie, those performing only the finishing phase of the pig production cycle, from finishers to slaughterhouse) accounted for approximately half the farms, whereas, the number of farrow-to-finish (FtF) farms (ie, those performing breeding, preweaning, growing and finishing phases at the same site) was approximately 12 000. There are some differences in the distribution and size of pig farms among Autonomic Communities (AC). The highest pig densities, especially in total number of animals, are those in Cataluña, Aragón, Castilla-León, Murcia and Andalucía (Subdirección General de Productos Ganaderos 2011).
The primary objectives of this study were to: (a) collect on-farm AMU data in fatteners employing two questionnaire-based surveys; (b) assess different quantitative measures for quantifying AMU in fatteners; (c) compare AMU in fatteners between two different pig management systems producing finishers: FtF farms versus finisher farms.
Material and methods
The first questionnaire designed was concerning FtF farms. The 14-page questionnaire (Table 1), consisted mainly of open questions for the responder, the farm and the AMU (in-feed, in-drinking water and per injection) covering all phases of the pig production cycle (ie, sows and suckling piglets, growers and finishers), but only the results of the finishing phase (finishers to slaughterhouse) will be presented. Different questions were employed for characterising AMU during the last six months at different levels. The first level was composed of subjective and qualitative questions per administration route (in-feed, in-drinking water and parenteral) to engage the responder's thoughts about AMU frequency (eg, never, occasionally, regularly, frequently and continuous), and AMU reasons (eg, planned, health problems, both). The second-level questions were targeted to identify all the AMs used, including commercial name, dosage, treated animals and indications.
The draft questionnaire was checked by a panel of five experts, and tested in a pilot study with two pig farmers not participating in the final study.
Then, a simplified version of the FtF farms questionnaire was developed for finisher farms. It was a nine-page questionnaire with the same sections as in the previous one, but AMU questions were restricted to only the finishing phase. Both questionnaires (in Spanish) are available from the author on request.
Both surveys were performed by means of personal face-to-face interviews. Data encompassed the six months prior to the interview. Each veterinary medical product containing AMs was counted as an individual record.
Two independent samplings, FtF farms and finisher farms, were conducted. In both cases, their respective sampling frames were obtained from the official Spanish data records (Ministerio de Medio Ambiente, Medio Rural y Marino) updated on January 2010 and June 2010, respectively.
The primary sampling units were the peninsular Spanish AC, and the secondary sampling units were the farms belonging to the first, second and third-sized groups according to the Spanish regulations (ie, Official Spanish classification of pig farms: G1 having until 120 livestock units (1LU = kg live weight of adult cattle); G2: between 121 and 360 LI; G3: between 361 and 864/720 LU; Real Decreto 324/2000, 2000). The sizes of the sampling frames were: 15 AC, 2968 FtF farms (January 2010), and 14215 finisher farms (June 2010).
Sample size calculations (WinEpiscope 2.0; Thrusfield and others 2001) were based on the worst-case scenario (eg, 50 per cent expected relative frequency of farms reporting AMU of any specific AM, with 95 per cent confidence level, an expected error of 10 per cent and an infinite population). These assumptions led to a sample size calculation of 97 farms. An expected participation percentage of 90 per cent was then used to obtain a final sample size of 108 farms for each survey.
In both cases, a multistage sampling methodology was applied, with AC as the primary sampling units and farms as the secondary sampling units. In the first stage, AC and their respective number of farms were selected using freely available software (C-Survey 2.0). This software performed a cluster selection proportional to cluster size (eg, number of eligible farms per AC) and distributed the final sampling selection among them. For the second stage, a linear distribution was chosen for the number of selected farms per AC among the three farm-size groups mentioned above, starting with the G3 group. Finally, the potentially eligible farms were chosen from their respective record lists, arranged from higher to lower size, using a systematic sampling (Levy and Lemeshow 1999). All the interviews were performed by the same interviewer, from April to October 2010 at the FtF farms, and from August to November 2010 at the finisher farms.
Data recording and analysis
Questionnaire-based data were stored in a relational database (Microsoft Office Access). All data were checked for consistency against the original hard copy. Descriptive (eg, mean, SD, median and percentiles) and analytic data analysis (eg, χ2 test and Student's t test) statistics were calculated using commercially available software (IBM SPSS Statistics V.19.0).
Quantifying AMU: three different approaches
The three approaches. These were: number of exposed farms; number of exposed animals; and number of exposure events (ie, animal-days, Rosengren and others 2007). For comparative purposes, all the absolute exposure measures (eg, farms, animals and animal-days) were then transformed into relative exposure measures using their respective units at risk as denominators: farms, animals at risk (ie, number of pigs in the finishing phase during six months), and animal-days at risk (ie, number of animal-day units in the finishing phase), respectively.
Animals at risk were computed for a six-month period from the data obtained through the questionnaire. For the finisher farms, this was calculated from the number of finishing places and the length of the phase in days (eg, departure to the slaughterhouse minus farm arrival). When there were doubts regarding the complete use of the finishing places, then the size and periodicity of entry of batches to the farm were used. In both cases, mortality was estimated to be 4 per cent (MA Higuera, personal communication). For FtF farms, the size and periodicity of the batches coming from the previous phases (eg, suckling and growing), and the length of the finishing phase in days were used, assuming a mortality percentage during the finishing period of 4.5 per cent (MA Higuera, personal communication). Each farm population at-risk figure was multiplied by half its respective phase length for computing total animal-days at risk.
Study participation. Of the 108 potentially eligible FtF farms, 33 (30.6 per cent) were excluded because they did not fulfil the participation criteria (FtF farm in full operation during at least six months before the interview). Thus, 75 farms were considered eligible, and of these, 49 (65 per cent) were surveyed.
Furthermore, of the 108 potentially eligible finisher farms, eight (7.4 per cent) failed to meet the participation criteria (finisher farm in full operation during at least six months before the interview). Of the remaining 100 farms, 67 (67 per cent) were interviewed.
The main reasons for non-participation were: 20 owners (seven in FtF and 13 in finisher farms) could not be reached by telephone despite five or more attempts, and 15 (nine and six, respectively) declined to participate.
The distributions were compared between the potentially eligible and eligible farms, in order to check whether these farms' distributions represented their original population distributions per size-group (Table 2). Major deviations between the potentially eligible figures were on the percentages of the G1 group farms (eg, the smallest farms) that were less represented in both analysed samples. The differences between distributions of potentially eligible and analysed farms were statistically significant only for the FtF farms survey (P < 0.05, χ2 test); however, there were no statistically significant (P > 0.05, χ2 test) differences between eligible and analysed farm distributions in both surveys.
Main descriptive questionnaire-based data. Although a detailed description of all the data obtained in both parts of the study will be reported later, descriptive data of the farms studied are summarised in Table 3. Both types of farms participating in the study were quite similar regarding the pig breeds, farm size (expressed as number of finishers) and days to slaughter; differences were found regarding management structures (eg, integration was very frequent among finisher farms, but it was less common among FtF farms), computer-stored AM data (eg, low percentages on both farm types, specially for FtF farms), available data during the interview (eg, higher in FtF than finisher farms), and people in charge of feeding animals (eg, veterinarians were more involved in feeding in the finisher farms than in the FtF farms).
AMU measuring: three AMU relative measures were calculated as the percentage of farms exposed (Table 4), the percentage of animals exposed (Table 5), and the percentage of exposure units in animal-days (Table 6).
All but two of the FtF farms reported AMU in fatteners for the six-month study period. Thirty-one farms (63 per cent) used both oral and parenteral medications, 4 per cent (2/49) only oral and 29 per cent (14/49) only parenteral. Among the 33 farms registering oral use, 49 per cent (16/33) used in-feed and in-drinking water medications, 15 per cent (5/33) only in-feed medications and 36 per cent (12/33) only in-drinking water medications.
The qualitative AMU data was similar for the finisher farms. All but two farms reported AMU, 90 per cent (60/67) of farms reported oral and injection medication, 3 per cent (2/67) of farms only oral use, and 5 per cent (3/67) of farms only per-injection use. Among the 62 farms reporting oral use, 84 per cent (52/62) used in-feed and in-drinking water medication, 10 per cent (6/62) only in-feed, and 6 per cent (4/62) only in-drinking water medications.
The percentage of farms using oral AMs was higher (P < 0.001, χ2 test) in finisher farms (93 per cent) than FtF farms (67 per cent). AMs were used in-feed in 87 per cent of the finisher farms and 43 per cent of FtF farms; in-drinking water in 84 per cent of the finisher farms and 57 per cent of FtF farms; and per-injection in 94 per cent and 92 per cent farms, respectively (Table 4). All the differences in the oral routes were statistically significant (P < 0.05, χ2 test), indicating that AMU was higher in finisher farms.
Figures by specific AMs highlighted the differences per administration routes mentioned above. For oral medications, doxycycline (76 per cent exposed finisher farms and 45 per cent exposed FtF farms) and colistin (64 per cent and 37 per cent) were the highest AMU compounds, followed by amoxicillin (52 per cent and 18 per cent) and lincomycin (43 per cent and 20 per cent). The most used AMs via the parenteral route were enrofloxacin (69 per cent and 37 per cent), amoxicillin (46 per cent and 39 per cent), and penicillin (46 per cent and 12 per cent). Although there were some differences among farm types related to the ranking of the AMs used, the most remarkable belonged to the quantitative figures demonstrating generally higher amounts in finisher farms than in FtF farms (Table 4).
The overall percentages of animals exposed to AMs (Table 6) were 90 per cent on finisher farms and 54.3 per cent on FtF farms, and parenteral use did not disturb these results because of its lower figures. Furthermore, a single measure summarising all the administration routes could be used as a global AMU measure, indicating that colistin (61.4 per cent in finisher farms and 33.3 per cent in FtF farms) and doxycycline (62.3 per cent and 22.6 per cent) were the highest AMU compounds, followed by amoxicillin (51.4 per cent and 19.5 per cent), lincomycin (48.9 per cent and 14 per cent), neomycin (24.3 per cent and 10.1 per cent), and tylosin (18.7 per cent and 8.4 per cent). Parenteral AMs had lower percentages of exposed animals, including enrofloxacin at 8.9 per cent and 6 per cent, respectively.
For AMs ranked by administration route, there were only slight differences when we compare animal-based with farm-based measures. Doxycycline (62.3 per cent in finisher farms and 22.6 per cent in FtF farms) and colistin (60.2 per cent and 30.8 per cent) were again the highest AMU compounds, followed by amoxicillin (48.9 per cent and 13.5 per cent) and lincomycin (46.2 per cent and 12.2 per cent), whereas, enrofloxacin (8.9 per cent and 6 per cent) and amoxicillin (2.5 per cent and 6 per cent) where the highest AMU parenteral compounds. Nevertheless, there were differences when comparing quantitative figures. Percentages of animals and farms exposed were similar for both oral routes, but they were notably different for the injection route.
Finally, AMU was measured as the percentage of exposure units in animal-days (Table 6). Also, a single measure aggregating all routes of administration could be used, indicating that colistin (14.1 per cent on finisher farms and 8.9 per cent on FtF farms) was the highest AMU compound, followed by doxycycline (8.2 per cent and 3.2 per cent), amoxicillin (8.1 per cent and 3.8 per cent), lincomycin (8.3 per cent and 1.5 per cent), neomycin (5.6 per cent and 5 per cent) and tylosin (2.8 per cent and 5 per cent). Parenteral AMs demonstrated lower percentages of exposure events, including enrofloxacin at 0.5 per cent and 0.2 per cent, respectively. Individual AM data per administration route showed a similar picture than the animal-based AMU estimation. The percentages of general exposure to any AM expressed in animal-days were 22.1 per cent and 16.6 per cent, respectively.
Exploratory analysis. Although the study design was descriptive, associations between some of the factors summarised in Table 3 and farm AMU of specific AMs (yes/no) were explored, especially in the finisher farms survey because of the higher sample size. Based on pig breed, the standard white was associated with higher AMU of most AMs (apart from tetracycline, florfenicol and marbofloxacin/danofloxacin), but only the use of colistin was statistically significant (P < 0.05, χ2 test). Equally, the use of integration as the management organisation was associated with higher AMU (apart from tetracycline and florfenicol), and statistically significant (P < 0.05, χ2 test) for colistin, amoxicillin, doxycycline and enrofloxacin.
Finally, although different variables related to farm size, both qualitative (eg, official farm-size group) and quantitative (eg, sow and finisher numbers) were explored, no statistically significant association with AMU was observed (eg, χ2 test for qualitative variables and Student's t test for quantitative).
The number of farms surveyed was lower than designed, especially for the FtF farms survey (Table 2). This was related to the economic problems in the Spanish pig production sector, which led to many farms closing down, especially the smaller farms. In addition, the response percentages (eg, 65 per cent and 67 per cent) were not at the expected level (90 per cent). Response percentages among similar studies performed in the pig sector included a 25.5 per cent from a mail survey (Stevens and others 2011), 51.2 per cent from an internet-based survey (Jordan and others 2009), and 60 per cent (Timmerman and others 2006) and 98.2 per cent (Casal and others 2007) from face-to-face surveys.
Despite the limitations of this study related to sample size, the medium and larger farms were well represented and, consequently, the potential bias of the AMU estimation for Spanish finishers pig production is believed to be low since these farms produced the majority of the commercial pigs.
The other putative limitation of this study would be the accuracy of the data. According to Table 3, a total of 49 per cent of the responders for the finisher farms survey, and 29 per cent of FtF did not use prerecorded information (eg, computer or paper records) for answering the questions. This probably did not affect the qualitative data on specific AMs used, but may have affected other quantitative data, especially the number of treated animals.
The first objective of this study was to carry out an on-farm AMU data recording employing a questionnaire-based survey. On-farm-based AMU pig data have been published previously in several European countries, for example, France (Chauvin and others 2002), Belgium (Timmerman and others 2006), Great Britain (Stevens and others 2011), and The Netherlands (van der Fels-Klerx and others 2011). Other surveys are also published from Canada (Rajic and others 2006, Rosengren 2007, Rosengren and others 2007), and Australia (Jordan and others 2009). Nevertheless, the differences among them include study design, data sources, and measures for quantifying AMU that can hamper an in-depth comparative analysis. For instance, face-to-face interviews with farmers to collect retrospective data were used (Rajic and others 2006, Timmerman and others 2006), but other authors used different methods, including mail surveys to pig farmers (Stevens and others 2011) or veterinarians (Chauvin and others 2002, Jordan and others 2009), or internet-based surveys (van der Fels-Klerx and others 2011). Similarly, Chauvin and others (2002) focused their study on the last AM prescription, whereas, Jordan and others (2009) analysed previously collected records from a farm data network. In addition, Timmerman and others (2006) obtained data only from group treatments.
In Spain, an on-farm AMU study of finishing pigs conducted in Catalonia in the northeast of the country, has been published previously by Casal and others (2007), and their results agree with those of this study regarding the main AMs used.
We also assessed different quantitative measures for summarising AMU, starting with the percentage of exposed farms (Table 4). This measure is calculated easily and has been previously employed by others for pig production (Rajic and others 2006, Casal and others 2007, Stevens and others 2011, Jordan and others 2009). Nevertheless, this measure is not useful for analysing the relationships between AMU and AMR since it does not capture the selection pressures associated with AMU. This is due to different reasons: farms having a single use (or using AMs in a low number of animals) are weighted equally than farms having multiple uses (or with routine use in all animals); a single measure of use, irrespective of the route of administration, is problematic since parenteral use tends to introduce bias; and the length of the treatment is not considered. It is important to highlight that it is not possible to construct a percentage of exposed farms irrespective of the administration route, since the injection route would overestimate some AMs in the global measure. For instance, enrofloxacin would be the third (finisher farms) or fourth (FtF) ranked AM, if this measure was employed.
The percentage of exposed animals (Table 5) was allowed to circumvent the first obstacle mentioned above, since it was based on the number of animals exposed to each AM and accounted for the number of applications. Nevertheless, the calculation of both absolute figures of animals, exposed and at risk, showed some constraints in our study; first, the number of exposed animals was missing in some records (eg, in 32 of the 90 records in FtF farms and in 37 of the 440 in finisher farms; Table 3); secondly, the number of animals at risk was unknown in one finisher farm; thirdly, the different methods for calculating the number of animals at risk over a six-month period in the finisher farms and in the FtF farms could have been overestimated figures for FtF farms. Despite these pitfalls, the use of percentages as a relative measure, avoided major biases.
Finally, the percentage of animal-days treated (Table 6) integrated the exposure length into the AMU measure. The animal-days at-risk figures were divided by two so as to correct for the fact that not all the animals were in the finishing phase during the entire six-month period. In comparison with the previous data, this measure showed a reduction in all the calculated percentages, and a new reduction in the weight of the injection route in the overall effect, obviously due to its shorter length (Rosengren 2007). Equally, in-drinking water treatments demonstrated a lower effect using this measure.
The third objective of the study was to compare AMU from two different pig management systems producing finishers. Our data concurs with Casal and others (2007), in that higher uses of AMs were for preventive measures for finisher farms due to the higher turnover of animals coming from multiple origins. Equally, van der Fels-Klerx and others (2011) associated these findings with a greater infection risk on finisher farms due to increased animal movements and mixing of animals on the farm.
The most critically important AMs (World Health Organization 2005), especially fluoroquinolones and third-generation cephalosporins, were almost always used per injection, as also the macrolides, erythromycin and tulathromycin (FtF farms). Other β-lactam AMs, including penicillin, were also mainly used per injection (in combination with streptomycin), as well as amoxicillin on the FtF farms, and florfenicol. By contrast, other AMs were mainly (eg, colistin, neomycin, lincomycin and tylosin) or only (eg, doxycycline) used in-feed or in-drinking water administration.
Overall, colistin, doxycycline and amoxicillin were the most used AMs in finishing pigs in Spain. Colistin was also most used for prophylaxis and treatment of enteric diseases, followed by tetracyclines and βlactamics (Casal and others 2007). Timmerman and others (2006) reported doxycycline, amoxicillin, trimethoprim-sulphonamides and polymyxin E (colistin) as the most commonly used oral AMs in group treatments for finishing pigs in Belgium. By contrast, the Canadian data for finishers (Rajic and others 2006, Rosengren 2007) showed a different picture, with tylosin and lincomycin the most commonly used in-feed, and penicillin in-drinking water and per injection. In the Australian study by Jordan and others (2009), the most common AMs for the entire pig cycle were penicillins, tetracyclines, macrolides and sulphonamides. AMU figures in pigs are also available in MARAN, SVARM and DANMAP reports, which demonstrate that tetracyclines, macrolides/lincosamides and pleuromutilins were the most used AMs (MARAN 2009, SVARM 2010, DANMAP 2010). In spite of this limited information, it appears that AMU patterns are similar among European countries but different from those of other continents. This finding demonstrates that local information must be employed for proper risk analysis studies within each country (Nicholls and others 2001).
Although this study was primarily descriptive in nature, an exploratory analysis was done of the associations between AMU and some farm-registered data. This analysis revealed that farm management structure was related with higher uses of some AMs on finisher farms; equally, farms raising Iberian pig had lower AMU for some AMs. These are new findings, and these relationships should be further studied.
Monitoring programmes of AMU in animals supply quantitative data for different purposes (Nicholls and others 2001, Jensen and others 2004, Carnevale and Shryock 2006, Grave and others 2006). Nevertheless, some authors (Grave and others 2004, Carnevale and Shryock 2006, McEwen and Singer 2006, Singer and others 2006) have expressed caution regarding the utility of AMU measures for testing the relationships among AMU and AMR in animals. Measures that take into account the pharmacological activity provide a better measure of the total selection pressure applied to a particular environment (Jensen and others 2004). In addition, data collected at a particular local production site will be the most useful to evaluate responsible AMU (Carnevale and Shryock 2006). Nevertheless, ongoing collection and analysis of valid farm data is laborious and time consuming (Chauvin and others 2008), and may be inefficient and expensive (Grave and others 2006). On the other hand, factors not considered previously, for example, the period from the last AM treatment to slaughter of food-producing animals, could be considered for improving our knowledge of the AM selective pressure due to AMU. Thus, a compromise among scientific, technical and financial aspects is necessary for obtaining a meaningful AMU measure for understanding these relationships with AMR.
Questionnaire-based surveys using face-to-face interviews are useful for capturing information regarding AMU at the farm level. Farm-level data per administration route can be used for comparative AMU analysis between farms. Nevertheless, for the analysis of the putative relationships between AMU and AMR, AMU measures based on exposed animals or exposure events are needed.
This work was funded by Ministerio de Ciencia e Investigación (grant AGL2009-10504). Also, the author appreciates the Ministerio de Medio Ambiente, Medio Rural y Marino for providing the frames for samplings, and for their role as expert on the panel testing questionnaires. Also, thanks to the remaining institutions and veterinarians participating in this panel: Agencia Española de Medicamentos y Productos Sanitarios, Veterindustria, ANAPOC and J Muñoz. The work of E Tejedor performing all the interviews is especially acknowledged. The author is indebted to all the owners, farm operators and veterinarians who kindly participated in the surveys. Finally, the author thanks S. Pyörälä for critical reading, and K Baptiste for editorial comments of the manuscript.
- Accepted July 11, 2012.
Provenance: not commissioned; externally peer reviewed
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.