The prudent use of antimicrobials (AMs) should be widened in pig farming to reduce the risk of AM resistance (AMR) in human and veterinary medicine. It is therefore important to understand pig farmers’ motivators and the barriers to AM usage (AMU) on their farms. The authors investigated pig farmers’ self-estimated levels of AMU, their perceived benefits and risks and the need for AMs in a cross-sectional survey in Belgium, France, Germany and Sweden. The authors also compared these perceptions between the four countries and related them to pig farmers’ actual AMU. The results showed that farmers who used more AMs also estimated their own usage as higher. Farmers perceived many benefits but relatively few risks of AMU in pig farming. Some significant cross-country differences in farmers’ perceptions were found, but they were relatively small. After controlling for country differences and farm differences, only perceived risks had a significant association with AMU. The authors therefore conclude that in order to promote prudent AMU, it seems most promising to focus on the structural differences in pig farming and veterinary medicine (e.g. legislation, role of the veterinarian) among countries. In addition, interventions which aim at reducing AMU should increase farmers’ awareness of the risks of extensive AMU.
- Pig farming
- Risk perception
- Self-reported usage
Statistics from Altmetric.com
The use of antimicrobials (AMs) to prevent or cure diseases in human beings and animals results in the emergence and selection of AM-resistant bacteria (WHO 2014). Consequently, human beings and animals carrying AM-resistant bacteria may be more difficult to treat with AM. It is thus important to use AMs prudently, that is, only when necessary, as targeted treatment and following best clinical practice methods (European Commission 2015). However, the current types and levels of substances used vary greatly between farms and deviate substantially from prudent use principles on many farms (Sjölund and others 2016). To stimulate prudent use of AM, the authors need to understand when, why, by whom and how much AMs are used. One of the areas in which AM usage (AMU) may be substantially reduced is pig farming. Although various farm characteristics have been suggested to be related to when and how much AMs farmers administer to their pigs, such as farm system and farm size (van der Fels-Klerx and others 2011, Vieira and others 2011, Coyne and others 2014), these farm characteristics did not fully explain differences in AMU at different pig farms (van der Fels-Klerx and others 2011), which suggests that there are additional determinants involved.
The study reported here investigated several psychological factors which may relate to pig farmers’ AMU, such as farmers’ beliefs and attitudes about AMU. Moreover, the authors aimed to survey these factors in four European countries and to investigate to what extent these factors are related to AMU in these four countries. An international perspective is needed as AM resistance (AMR) is a global problem (WHO 2014). Moreover, there seem to be noteworthy differences in AMU among European countries (ESVAC 2014), although the European Surveillance of Veterinary Antimicrobial Consumption (ESVAC) data do not allow the conclusion that there are significant country differences in the number of AM treatments per animal. In all European countries, veterinary AM medicines are only available on prescription.
Farmers’ perception of AMs and AMR
Only a few studies have investigated farmers’ perceptions of AMs and AMR. Overall, farmers seemed to perceive high benefits in using AMs in their pigs (Moreno 2014, Visschers and others 2015). For example, AMs were seen as very cost-efficient and highly effective in preventing and curing diseases in pigs.
Farmers appeared to have little awareness of the risks of AMU in pig farming for human and animal health (Moreno 2014), nor were they very concerned about this (Friedman and others 2007, Visschers and others 2015). They rather perceived AMR as a problem caused by other farmers who did not use these drugs prudently (Coyne and others 2014). Nevertheless, in a Swiss study, a higher perceived risk of using AMs was found to be related to fewer AM treatments (Visschers and others 2014).
Additionally, farmers seemed to strongly trust their veterinarians and considered being able to discuss AM prescription decisions with their veterinarians (Friedman and others 2007, Coyne and others 2014, Visschers and others 2015). Pig farmers who were more likely to consult their veterinarians before they administered AMs appeared to use less AMs at the beginning of the fattening period (Visschers and others 2014).
Farmers’ estimated AMU
It is important to know how pig farmers judge their own AMU. Based on their own estimations and awareness of their usage, farmers will be more or less inclined to try to reduce their AMU. Farmers seem to estimate their own AMU as rather low compared with that of other pig farmers in their country (Visschers and others 2015) and to believe that others use AMs less prudently than they do (Coyne and others 2014).
People's estimations or reports of their own behaviour do not always relate to their actual behaviour (Kormos and Gifford 2014). It may be that people report their behaviours incorrectly because they have difficulties recalling behaviours which are habitual and which are therefore performed without much awareness and reflection. Responding in a socially favourable way may also explain the ambiguous results regarding self-reported behaviour measures. Related to this explanation is a sort of optimism bias (Weinstein and Klein 1996); people believe that they perform less risky behaviour than others do, because they are unaware of the behaviour strategies of others.
Aims of this study
The aims of the present study were threefold. First, the authors wanted to know how accurate farmers’ estimations of their own AMU are compared with the actual AMU. Secondly, the authors aimed to investigate farmers’ perceptions of the benefits, need and risks of AMs and of their relationship with their veterinarians and to what extent pig farmers from four different European countries differ in these perceptions. Finally, the authors intended to quantify to what extent psychological factors are related to pig farmers’ AMU in the four investigated countries.
Pig farmers from Belgium, France, Germany and Sweden participated in a study in which their actual AMU and their perception of AMs were assessed. The study was part of the European research project MINAPIG (Evaluation of alternative strategies for raising pigs with minimal antimicrobial usage: Opportunities and constraints, www.minapig.org). The authors aimed to visit, per country, 60 farrow-to-finish farms which had at least 100 sows and 500 finishers per year. The authors estimated that 60 farms would suffice to find significant effects in the analyses they aimed to do. Moreover, this sample size was expected to be manageable within the limited time frame and budgets.
The contact details of the Belgian pig farmers originated from a subscription list from the University of Ghent for a newsletter for pig farmers. In Sweden, the authors used a large pig herd database from the Swedish Animal Health Service as well as herds which had been in contact with researchers from the National Veterinary Institute in the past. German farmers were recruited through consultancy circles for pig farmers and through veterinary practices. In France, the authors restricted the present study to the northwestern part of the country, where 75 per cent of the country's pig production is concentrated and used the national database of the French Institute for the Pig and Pork Industry (IFIP) to randomly choose pig farmers to be invited for the present study. In Belgium, Germany and Sweden, the authors expected that it would be difficult to recruit 60 farmers and therefore contacted all available addresses. This thus resulted in convenient samples.
In the invitation to participate in the present study, it was explained to them that the study would be about AMU, involved a farm visit and the completion of a questionnaire. The questionnaire was sent before the farm visit with the request to complete it before the visit. Data collection took place between November 2012 and December 2013.
After matching the file with the questionnaire data to the file with the actual AMU data using the farmer identification numbers, the sample sizes per country were as follows: 46 farmers in Belgium (response rate: 68 per cent), 56 in France (response rate: 71 per cent), 54 in Germany and 59 in Sweden. The authors could not calculate the response rates of the German and Swedish samples because the various organisations that contacted the farmers did not keep track of the number of initially contacted farmers. The number of participating farmers could thus not be compared with the total number of contacted farmers in these two countries.
The content of the authors’ seven-page questionnaire was based on semi-structured interviews with 14 pig farmers in Switzerlandi and Germany and was developed in close collaboration with the MINAPIG consortium. Topics which were included were, among others, the estimated amount of AMU at the respondent's own farm, the benefits and risks of AMs in pig farming, the need to use AMs in pig farming, the information provided by the farmer's veterinarian regarding health problem solving and three demographic variables (birth year, sex and years of experience in pig farming). Other topics of the questionnaire have already been discussed in other papers (see Visschers and others 2014, 2015). The questionnaire items were initially developed in English and were then translated into the languages of the respective countries.
The estimated relative AMU (see Table 1) was measured with one item on a 5-point response scale, ranging from 1 (‘much less’) over 3 (‘about the same’) to 5 (‘much more’). The other topics were measured with two to five items, which were assessed on 6-point Likert scales, ranging between 1 (‘do not agree at all’) and 6 (‘fully agree’). The variables had acceptable to good internal reliabilities (Cronbach's αs between 0.64 and 0.83, see Table 1), so that the mean scores of all variables were used in the analyses reported below. All items, their means (Ms), SDs, medians (Mdn) and corrected-item total correlations (rpbis) are presented in Table 1.
During the farm visits, interviewers collected information on the AMU of the preceding year in Belgium, Germany and Sweden and over the last batch in Franceii (actual AMU). The recorded AMs product consumption was converted into the active substance used, in milligrams, which was then transformed into the treatment incidence (TI) based on consensus Defined Daily Doses per Animal (DDDAs, Postma and others 2015). The TI is the number of DDDAs used per day and per 1000 pigs at risk (Timmerman and others 2006). The TI was standardised over 200 days (i.e. the average lifespan of a fattening pig, from birth till slaughter), which resulted in the TI200. The TI200 makes it easier to compare the usage across countries, since the period at risk is standardised among the countries. A detailed description of the herd visits, collected data and calculation of the AM TI can be found in Backhans and others (2015), Postma and others (2016) and Sjölund and others (2016).
All analyses were conducted in IBM SPSS Statistics, version 22 (IBM Corporation 2013). The frequency distribution of the TI200 data was positively skewed (M=127, SD=144, Median (Mdn)=69, range: 0–674)—many farms had a low TI200—and were thus not normally distributed. Similarly, the data on the number of sows at a farm were positively skewed (M=298, SD=242, Mdn=224, range: 85–1750). The authors therefore conducted a log transformation on these data, after adding 1 to all data to be able to transform the 0 s (Postma and others 2016). This notably improved the frequency distribution of the TI200 (M=1.76, SD=0.65, Mdn=1.85, range: 0–2.83) and of the number of sows’ data (M=5.48, SD=0.61, Mdn=5.41, range: 4.44–7.47). The authors used the log TI200 and the log number of sows in the further analyses. The relation between the log TI200 and the estimated AMU was calculated using Spearman rank correlation (rs), as the data of the latter variable were also not normally distributed.
To analyse whether farmers in the different countries differed significantly in their perceptions of the risks, benefits and need for AMs and the veterinarian's information provision, the authors conducted a multivariate analysis of covariance (MANCOVA) on these four variables with country as the independent variable. Age, sex and the log number of sows were included as covariates, as the samples from the four countries differed significantly regarding these three variables (see Table 2). Significant main effects of country were further explored by means of post hoc comparisons with Bonferroni correction to prevent an inflation of the probability of a Type I error.
Additionally, a hierarchical linear regression analysis was conducted to predict farmers’ AMU. Log TI200 was the dependent variable, and the authors examined two models. In model 1, the authors included the demographic variables of sex, years of work, log number of sows and country of farmer (the latter as dummies; Belgium was arbitrarily chosen as the reference group). In model 2, the authors added the psychological variables, perceived benefits of, perceived risks of and perceived need for AMs as well as the veterinarian's information provision.
In total, 215 pig farmers completed both the herd visit and the questionnaire, that is, between 46 and 59 farmers per country (see Table 2 for the demographics per country). Data on actual AMU were partly missing for one Belgian farmer, so the TI200 could only be calculated for 214 cases. The average age of the participating farmers was 45 years (SD=9.5), and they reported having on average 23 years of experience in pig farming (SD=10). Overall, 80 per cent were male (n=172) and 18 per cent were female (n=39); four farmers did not report their sex. The participating herds had on average 298 sows (Mdn=114, IQR: 160–330).
Estimated relative AMU and actual AMU
According to their self-assessment, farmers estimated that they used less AMs than did their countrymen (M=2.24, SE=0.06 on a scale ranging from 1 [‘much less’] over 3 [‘about the same’] to 5 [‘much more’]). Although participating farmers from all countries estimated their AMU lower compared with their fellow-country farmers, Swedish farmers’ relative AMU estimates were significantly higher than the relative estimates of the participants of the three other countries (Table 3 and Visschers and others 2015). It should be noted that eight French, 19 German and three Swedish farmers did not complete this question.
The actual AMU (TI200) in the four countries is discussed in detail by Sjölund and others (2016). It suffices here to report that the German farmers had a significantly higher log TI200 than the farmers in the other countries (Table 3). Swedish farmers used significantly less AMs than the farmers from the other three countries. Belgian and French farmers were between their German and Swedish colleagues regarding their TI200.
Additionally, the authors found a significant correlation between log TI200 and the estimated relative AMU among all farmers, rs=0.16, P=0.01. The correlations between log TI200 and estimated relative AMU were even larger in the individual countries, that is, Belgium: rs=0.36, P=0.01; France: rs=0.49, P=0.001; Germany: rs=0.48, P=0.001 and Sweden: rs=0.34, P=0.01 (see Figure 1). In other words, the higher the actual AM TI, the higher farmers estimated their own AMU compared with that of their countrymen.
Cross-country differences in perceived benefits, risks and need for AMs and veterinarian's information provision
Overall, the perceived benefits of AMs (M=4.06, SE=0.06) and the perceived information provision by the veterinarian (M=4.58, SE=0.08) were relatively high in all four countries compared with the midpoint of the 6-point Likert scale (i.e. 3.5). The perceived need for AMs was rather low in all countries (M=2.16, SE=0.08), whereas respondents were neutral regarding the risks from AMs (M=3.15, SE=0.08).
The outcomes of the MANCOVA indicated that farmers from different countries differed significantly in their perceived risks of and need for AMs as well as in the perceived information provision from their veterinarians, but not regarding the perceived benefits of AMs (see Table 4). Belgian farmers perceived more risks from AMs than German farmers. Also, Belgian and Swedish farmers perceived more need to use AMs in pig farming than their French colleagues. Swedish farmers thought that their veterinarians provided more information than German and Belgian farmers believed about their veterinarians, while French farmers perceived a better information provision by their veterinarians than did Belgian farmers. However, the authors should note that these significant differences were relatively small among the countries (see the effect sizes in Table 4).
Predicting actual AMU
Regression of the demographic variables on log TI200 (model 1 in the hierarchical regression) resulted in a significant model which explained 38 per cent of the variance (Table 5, model 1). The number of sows at the farm was a significant positive predictor of the TI with AM. Being a German farmer versus (v) a Belgian increased the log TI200, whereas being a Swedish farmer v a Belgian farmer reduced this.
Adding the psychological variables at model 2 marginally improved the model fit (Table 5, model 2) and resulted in a portion-explained variance of 41 per cent. Being a German v a Belgian farmer, being a Swedish v a Belgian farmer and the number of sows at the farm remained significantly related to the AM TI. In addition, the perceived risks of AMs were significantly related to the log TI200. Higher perceived risks were associated with a lower AM TI, even after controlling for differences in the demographic variables.iii
We found that farmers’ estimated relative AMU was significantly related to their actual AMU: the higher farmers estimated their AMU, the more frequently they treated their pigs with AM. These correlations were not very high, which is quite common to find in behavioural research (Cohen 1988). A higher correlation would be unlikely because of the low variance in farmers’ self-estimated AMU (see Table 3). It was probably not easy to estimate their own AMU compared with other farmers in their country, because they needed to know how many AM treatments they administered on average and how many others applied. Farmers in the investigated countries did not have this knowledge, because their AM TI was not benchmarked against that of other pig farms in their country and communicated to them (as is done in Denmark and the Netherlands).
The moderate correlations between farmers’ estimated AMU and actual AMU as well as their overall tendency to report a lower AMU compared with their countrymen seem to imply that farmers underestimated their AMU. This may have been caused by an optimism bias (Weinstein and Klein 1996). However, the authors recruited a convenience sample in order to find a sufficient number of farmers who were willing to participate in the time-consuming herd visits and to report on the sensitive topic of AMU in pig farming. The possibility can therefore not be excluded that the participating farmers may have been rather interested in their AMU and that they already administered fewer AM treatments than did farmers who did not take part in the present study. Alternatively, socially desirable answering may have caused the moderate correlations between factual and estimated relative AMU in the authors’ sample (Kormos and Gifford 2014). Because the questionnaires were collected during the herd visit, the authors could not guarantee absolute anonymity of the data.
In all four countries, farmers perceived the benefits of AMs to be rather high and the need for AMs as relatively low. This implies that despite the overall favourableness of AM, farmers may believe that pigs can stay healthy without a high AMU. Farmers also rated the risks of AMU and of AMR to be moderate. The latter finding seems to be in line with that of previous studies showing that farmers were neither very aware (Friedman and others 2007, Stevens and others 2007) nor very worried about AMR (Visschers and others 2015). In contrast to the benefits of AM, the risks of AMU are not directly noticeable to farmers as long as they do not experience therapeutic failure in their animals, themselves or the people living and working with them. Farmers thus mostly learn about the risks of AMs indirectly through education and communication.
The farmers seemed to highly appreciate the information provision by their veterinarians. Previous studies revealed that veterinarians are seen as reliable and preferred sources of information (Friedman and others 2007, Garforth and others 2013, Coyne and others 2014). Veterinarians thus seem to have a central role in influencing AMU at pig farms.
However, farmers’ perceived information provision by the veterinarian was unrelated to the TI200 in the present study. It may be that because all farmers were rather positive about their veterinarians, the small variance between farmers with respect to this variable could not significantly correlate with the variance in the number of AM treatments. Alternatively, an important aspect of the relationship between farmers and veterinarians which affects AMU may have been missing in the authors’ questionnaire, for example, the ease with which the veterinarian could be convinced to prescribe AMs (see Coyne and others 2014, Speksnijder and others 2015).
The authors’ findings revealed that country of residence was the most important predictor of farmers’ actual AMU and was thus more important than farmers’ perception of AM. Therefore, it seems necessary to investigate and compare the structural differences between countries regarding pig farming, veterinarian practices and legislation as well as information and communication, which may result in lower AMU and to implement promising structural measures in countries with high AMU. However, some differences between countries, such as the presence and absence of infectious diseases or long-standing veterinary practices, may be difficult to change in the short term.
Farmers’ awareness of the risks of AMU should also be increased for two reasons. First, the perceived risk of AMs was relatively low in the authors’ sample and farmers seemed little aware about the transfer of AMR bacteria between animals and human beings, although this is a high risk among this professional group (e.g. Aubry-Damon and others 2004). The second reason to increase AMU risk awareness is the finding that the perceived risk of AMs was the most important psychological predictor of farmers’ AMU even after controlling for country and farm characteristics. Increasing risk awareness may convince farmers with a relatively high AMU to reduce the number of AM treatments at their farm, even in countries with a low AMU. A potentially effective way may be to benchmark the AMU at all farms against that of other pig farmers in their country (see Andreasen and others 2011, DANMAP 2014). However, because the present study was a cross-sectional study; the authors do not know whether higher perceived risks lead to lower AMU or whether farmers who use more AMs downplay the risks of AM.
Increasing risk awareness will not be sufficient to change farmers’ AMU. Previous research has pointed out several other factors that can persuade people to change their behaviour (van der Pligt 1998). Examples of such additional motivators may be the perceived benefits of alternative behaviours (e.g. alternative preventive measures and treatment options) and their perceived costs. The relations between these additional factors and farmers’ AMU will need to be examined in future research.
The present study had a few limitations which should be addressed. Because the authors conducted it among convenient samples (see above), the variances in risk perception and in AMU may be larger in a bigger and more representative sample (i.e. including various types of pig herds and covering the full range of herd sizes and AMU in a country), so that the actual relation between perceived risks and AMU may even be stronger.
The four countries in this study have different pig farming settings, as well as different policies and regulations regarding the prescription of AM. In all countries, farmers can have a certain amount and type of AMs on stock and administer certain treatments without consulting their veterinarian if they have a written agreement with their veterinarian (Belgian FPS Social Affairs and others 2000, German Federal Ministry for Youth Family and Health 2009, Swedish Board of Agriculture (SJVFS) 2009, German Federal Ministry for Health 2014, Truchet and others 2014). Veterinarians have to visit the farms with a contract regularly to check the animals and whether the prescribed drugs are prudently used. The number of veterinary visits differs between the four countries, as well as the amount of AMs and the period that farmers are allowed to have them on stock. Also, Sweden has had a longer tradition of strict AMU regulation than the other participating countries (see Grave and others 2006).
The number of countries in this study was rather small and their cultures and values are rather homogeneous from a global perspective. This may explain the authors’ finding that farmers from different countries hardly differed in how they perceived AM. Before the authors can generalise their findings to other countries, the relation between psychological variables and AMU should thus also be examined in more representative samples, that is, among various types of pig farms (e.g. pig fattening farms) and various herd sizes, and in countries in which the differences in pig farming settings, AMU regulations and cultural backgrounds are larger.
Despite the structural differences in pig farming and AM prescription regulations among the four countries, it appeared that in all countries, farmers’ estimated AMU was to some extent related to their actual AMU. Country differences as well as perceived risks of AMs served to predict farmers’ actual AMU. The next step is to explore the structural differences in pig farming and veterinary medicine among countries and to test the effectiveness of interventions which aim to reduce AMU among pig farmers in different countries.
This project was part of the European MINAPIG project (Evaluation of alternative strategies for raising pigs with minimal antimicrobial usage: Opportunities and constraints, www.minapig.eu), which was funded by the ERA-NET programme EMIDA (EMIDA19) and by the participating national funding agencies. The authors would like to thank the MINAPIG consortium for helping to develop the questionnaire and for conducting the survey. The MINAPIG consortium consists of the following members in alphabetical order: Annette Backhans, SLU, Sweden; Catherine Belloc, ONIRIS, France; Lucie Collineau, SAFOSO, Switzerland; Jeroen Dewulf, Ghent University, Belgium; Ulf Emanuelson, SLU, Sweden; Elisabeth Grosse Beilage, TiHo Hannover, Germany; Bernd Grosse Liesner, Boehringer Ingelheim, Germany; Christian Alexander Körk, Boehringer Ingelheim, Germany; Ann Lindberg, SVA, Sweden; Svenja Lösken, TiHo, Hannover, Germany; Merel Postma, Ghent University, Belgium; Hugo Seemer, Boehringer Ingelheim, Germany; Michael Siegrist, ETH Zurich, Switzerland; Marie Sjölund, SVA, Sweden; Katharina Stärk, SAFOSO, Switzerland and Vivianne Visschers, ETH Zurich, Switzerland. The authors would also like to thank all farmers who participated in this study.
L. Collineau is also at LUNAM Université, Oniris, INRA, Nantes, France
Provenance: Not commissioned; externally peer reviewed
Correction notice This article has been corrected since it was published Online First. The title “Higher perceived risk of antimicrobials is related to lower antimicrobial usage among pig farmers in four European countries” has been corrected to “Higher perceived risks of antimicrobial use are related to lower usage among pig farmers in four European countries”.
Twitter Follow Merel Postma at @MerelPostma
↵i We also conducted interviews in Switzerland, because Swiss research and consultancy institutes were involved in the MINAPIG project. The same survey as reported in this study was conducted in Switzerland but without herd visits. Therefore, the Swiss farmers’ survey data could not be related to their AMU data and compared with those of the other countries (but see Visschers and others 2014).
↵ii In France, AMU data were only taken from the last batch because we completed the missing data for AMU from the treatment records and an interview with the farmer in this country. If we had chosen a full year, we would not have been able to collect such detailed information as the farmer would not have remembered the treatment details for such a long period.
↵iii To examine whether the relation between the perception variables and actual AMU differed among the four countries, we also conducted a linear regression analysis on AMU (log TI200) which included the nine interaction terms in model 3. Because we wanted to limit the number of interaction terms, we only included those perception variables which significantly differed among countries (see Table 4). The nine interaction terms were computed between the dummy variables of the countries and perceived risks, perceived need and veterinarian's information provision, after we had standardised the data of all continuous predictors (Cohen and others 2003). The inclusion of the interaction terms in the regression model did not significantly improve the model, ΔR2=0.03, Fchange(9, 186)=1.03, P=0.42. Moreover, none of the interaction terms was significantly related to log TI200 (Ps>0.09). In other words, the relation between the TI and perceived risks, perceived need and veterinarians’ information provision did not differ among the four countries. Sweden (v Belgium), the number of sows at the farm and perceived risks of AMs were still significantly related to the TI200 (Ps<0.05).
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.