Hostname: page-component-7c8c6479df-24hb2 Total loading time: 0 Render date: 2024-03-28T21:08:05.789Z Has data issue: false hasContentIssue false

An analysis of cattle farmers' perceptions of drivers and barriers to on-farm control of Escherichia coli O157

Published online by Cambridge University Press:  27 November 2014

L. TOMA*
Affiliation:
Land Economy, Environment and Society, SRUC, Edinburgh, UK
J. C. LOW
Affiliation:
Royal (Dick) School of Veterinary Studies, University of Edinburgh, Edinburgh, UK
B. VOSOUGH AHMADI
Affiliation:
Land Economy, Environment and Society, SRUC, Edinburgh, UK
L. MATTHEWS
Affiliation:
College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, UK
A. W. STOTT
Affiliation:
Future Farming Systems, SRUC, Edinburgh, UK
*
*Author for correspondence: Dr L. Toma, SRUC Edinburgh Campus, King's Buildings, West Mains Road, Edinburgh EH9 3JG, UK. (Email: luiza.toma@sruc.ac.uk)
Rights & Permissions [Opens in a new window]

Summary

Structural equation modelling and survey data were used to test determinants' influence on farmers' intentions towards Escherichia coli O157 on-farm control. Results suggest that farmers more likely to show willingness to spend money/time or vaccinate to control Escherichia coli O157 are those: who think farmers are most responsible for control; whose income depends more on opening farms to the public; with stronger disease control attitudes; affected by outbreaks; with better knowledge and more informed; with stronger perceptions of biosecurity measures’ practicality; using a health plan; who think farmers are the main beneficiaries of control; and whose farms are dairy rather than beef. The findings might suggest that farmers may implement on-farm controls for E. coli O157 if they identify a clear hazard and if there is greater knowledge of the safety and efficacy of the proposed controls.

Type
Original Papers
Copyright
Copyright © Cambridge University Press 2014 

INTRODUCTION

There is increasing evidence that the farm environment is an important hazard resulting in a considerable number of sporadic Escherichia coli O157 infections [Reference Parry1Reference Matthews6]. The presence of E. coli O157 in animal manure can lead to contamination of soil and grass, farm buildings, fences, machinery and water-courses, and the organism may survive for months in animal faeces and soil. In relation to measures for the control of E. coli O157 on-farm, EU food regulations recognize that ‘The application of the Hazard Analysis and Critical Control Point (HACCP) principles to primary production is not yet generally feasible. However, guides to good practice should encourage the use of appropriate hygiene practices at farm level’ [7, p. 5]. Despite significant effort in the past 10 years to understand the carriage of E. coli O157 by cattle both on and between farms, current knowledge is still incomplete thus limiting the understanding of what can be assumed as good practice for on-farm control. An additional potential constraint to on-farm control is the fact that no production losses are associated with cattle infection and therefore controls are necessary only to prevent human infection.

Applying E. coli O157 control measures on-farm is assumed to decrease the risk of transmission of E. coli O157 disease from livestock to humans and, implicitly, reduce the risks posed by E. coli O157 to human health. Understanding which determinants influence farmers’ behavioural intentions and, potentially, behaviour towards livestock disease control has been the focus of a number of research studies over time and increasingly so during the past couple of decades.

The study analyses the impact of a priori determinants on adoption of E. coli O157 on farm control measures by cattle farmers in the UK. We used a dataset collected through a stratified telephone survey and analysed it using a structural equation model (SEM) based on behavioural economics theory. This is, to the best of our knowledge, the first paper using SEM applied to representative survey data to analyse farmers’ attitudes and intentions to control E. coli O157 on farm.

METHODS

Research hypotheses

Based on a review of literature and expert opinion, we built and tested five research hypotheses:

Hypothesis 1:

Farm characteristics (e.g. farm type; use of a livestock health plan; income from opening the farm to public) influence farmers’ willingness to control E. coli O157 on farm.

Farm characteristics influence the type of disease control measures required and the level of investment (financial or labour) needed [Reference Gunn8, Reference Toma9]. As well as the farm's physical constraints, the financial situation of the enterprise will impact on what measures the enterprise can afford to implement [Reference Chilonda and Van10, Reference Stott11]. Some authors [Reference Garforth, Bailey and Tranter12] found that pig and sheep farmers did not see health plans as a useful disease risk measure and mostly members of farm assurance schemes were more likely to have one. Other research [Reference Toma9] found that membership in cattle/sheep health schemes influenced biosecurity behaviour indirectly through other factors such as access to information and advice.

Hypothesis 2:

Farmers’ access to information and knowledge about E. coli O157 influences their willingness to control E. coli O157 on farm.

An important factor influencing farmer behaviour is the access to information on disease control measures and animal health issues. Some authors [Reference Garforth, Bailey and Tranter12] found that improving farmers’ access to information, targeting it through training events, the farming press, veterinarians, farmer groups, and tailoring it to different categories of livestock farmers could increase uptake of disease risk measures. Several studies analysed the importance of knowledge (awareness) of E. coli O157 of farmers, among other stakeholders, in influencing behaviours and dealing with E. coli O157 risk and prevention [Reference Ellis-Iversen13Reference Strachan15].

Hypothesis 3:

Incidence of outbreaks on farm influences farmers’ willingness to control E. coli O157 on farm.

In an outbreak situation the perceived and potential risks are elevated and the likelihood of farmers’ implementing measures to control the disease increases significantly [Reference Toma9, Reference Coleman, Hemsworth and Hay16Reference Lindberg19]. Garforth et al. [Reference Garforth, Bailey and Tranter12] found that farmers associated risk with the local disease status. If they were aware of neighbours’ livestock having a transmittable disease, they were likely to take additional precautions. Additionally, their study [Reference Garforth, Bailey and Tranter12] found that several farmers who stated they stopped vaccinating against some diseases when the risk was low said that they would consider vaccinating again if the disease risk increased in the area.

Hypothesis 4:

Farmers’ perceived practicality of E. coli O157 on-farm control measures (e.g. biosecurity) influences their willingness to control E. coli O157 on farm.

The literature has established that farmers are more likely to take up disease control measures if they find them practical/suitable to their farms. Braun et al. [Reference Braun20] found that demonstrations of successful implementation of biosecurity measures and their benefits increase the level of uptake. One study [Reference Cross, Rigby and Edwards-Jones21] used best-worst scaling to elicit experts’ assessment of the relative practicality and effectiveness of measures to reduce human exposure to E. coli O157, while another study [Reference Garforth, Bailey and Tranter12] found that farmers perceived the impracticality of some measures as constraints to the ability to implement them.

Hypothesis 5:

Farmers’ attitudes and perceptions regarding benefits of and responsibility towards E. coli O157 on-farm control influence their willingness to control E. coli O157 on farm.

Farmers’ attitudes towards and perceptions of disease control measures have an important role in farm decision-making processes and, more specifically, in disease control behaviour [Reference Toma9, Reference Chilonda and Van10, Reference Garforth, Bailey and Tranter12, Reference Ellis-Iversen13].

SEM

We used SEM with observed and latent variables to test the hypotheses and assess the strength of these relationships, i.e. how much these factors influence one another and primarily the behavioural willingness to control E. coli O157 on farm. As each variable will influence behavioural willingness both directly or indirectly (through their effect on other variables in the model, which in turn will directly influence behavioural willingness), the variance explained by the model is higher than when other techniques, such as regression analysis, are used [Reference Jöreskog and Sörbom22].

The model consists of two parts, namely the measurement model (which specifies the relationships between the latent variables and their constituent indicators), and the structural model (which designates the causal relationships between the latent variables). The model is defined by the following system of three equations in matrix terms [Reference Jöreskog and Sörbom22]:

the structural equation model: η =  + Γξ + ζ,

the measurement model for y: y = Λ y η + ε,

the measurement model for x: x = Λ x ξ + δ,

where η is an m*1 random vector of endogenous latent variables; ξ is an n*1 random vector of exogenous latent variables; B is an m*m matrix of coefficients of the η variables in the structural model; Γ is an m*n matrix of coefficients of the ξ variables in the structural model; ζ is an m*1 vector of equation errors (random disturbances) in the structural model; y is a p*1 vector of endogenous variables; x is a q*1 vector of predictors or exogenous variables; Λ y is a p*m matrix of coefficients of the regression of y on η; Λ x is a q*n matrix of coefficients of the regression of x on ξ; ε is a p*1 vector of measurement errors in y; δ is a q*1 vector of measurement errors in x.

We performed model estimation with the diagonally weighted least squares (DWLS) method using the statistical package Lisrel 8·80 [Reference Jöreskog and Sörbom22]. We combined Prelis (to analyse the raw data and compute the asymptotic covariance matrix) and Lisrel (to obtain estimates and test statistics, e.g. t test values, which estimate the statistical significance of causal relationships). SEM estimation is performed by minimizing the discrepancy between the covariance matrix of observed variables and the theoretical covariance matrix predicted by the model structure, which is a function of the unknown parameters. For the case of discrete indicators, Muthén [Reference Muthén23] and others developed procedures based on the application of polychoric correlations (rather than the Pearson correlations used for continuous indicators) to estimate the covariance matrix of the latent continuous indicators from the discrete indicators. Consistent estimates of the parameters can then be obtained by minimizing the discrepancy between the estimated covariance matrix and the theoretical covariance matrix [Reference Bollen24]. DWLS estimation method is consistent with the types of variables included in the model (ordinal and categorical) and the deviation from normality in some of these variables [Reference Finney, DiStefano, Hancock and Mueller25].

The model is validated using absolute, incremental and parsimonious goodness-of-fit (GoF) indicators [Reference Hair26]. The absolute fit indicators include: root mean square error of approximation and GoF index. Incremental fit indicators include: adjusted GoF index, non-normed fit index, normed fit index, relative fit index, comparative fit index and incremental fit index. Parsimonious fit indicators include normed χ 2.

An acceptable level of overall GoF does not guarantee that all constructs meet the requirements for the measurement and structural models. The validity of the SEM is assessed in a two-step procedure, the measurement model and the structural model. In the measurement model the reliability of single-indicator latent variables is tested using the ‘theory-testing extremes’ of reliability within the range of 0·7–1 [Reference Ping27] to determine if any structural coefficients become non-significant at these extremes.

Model selection is performed using a nested model approach, in which the number of constructs and indicators remains constant, but the number of estimated relationships changes.

The structure of the model was based on the survey questionnaire detailed in the following ‘Questionnaire and survey’ section.

Questionnaire and survey

The questionnaire was built based on a review of the literature, and expert opinion was used to develop it into its final version. The questionnaire was circulated in several rounds to experts from academia, the private sector and the policy environment, e.g. from the Scottish Agricultural College (currently SRUC), University of Glasgow, Bioniche Life Sciences, Food Standards Agency, Department for Environment, Food and Rural Affairs (Defra), who commented on the design of the questionnaire.

The questionnaire was consistent with the aim of testing the research hypotheses and the use of SEM. It included closed-ended questions on the following: socio-demographical information about the farmer (gender, age, education); farm economic information (status with respect to the farm holding, total farm land area, number of livestock, full-time and part-time labour, share of income from livestock production, organic certification, open farm characteristics, proportion of farm income dependent on opening to the public); access to information sources; knowledge about E. coli O157; attitudes regarding the use of control measures for E. coli O157; perceived benefits of controlling E. coli O157; perceived responsibility in controlling E. coli O157; influence on business of factors such as regulations and E. coli O157 outbreaks; perceived practicality of bio-security measures; intentions to change farm size; intentions to change public access to the farm; intentions to change E. coli O157 control measures on-farm; willingness to use E. coli O157 control measures.

Table 1 presents a description of the latent variables and their corresponding indicators included in the SEM model.

Table 1. Description of latent variables and their corresponding indicators

The data used in this study was collected through a representative survey of UK cattle farmers. The sampling frame was derived from the June 2010 Survey of Agriculture and Horticulture for England, Wales, Scotland and from the Public Health Information System data for Northern Ireland and included all holdings with cattle. The criteria for inclusion in the study were as follows: main farm type (classification derived from the June survey information as the standard measure of farm activity and type; to be classed in a particular area, a holding must have at least two-thirds of its activity in one particular area, otherwise it is deemed to be of mixed type); farm size [using only holdings which have a standard labour requirement (SLR) >0·25 FTE (full time equivalent) to avoid inclusion of hobby farmers]; stocking density or less favoured area (LFA) marker (used in place of stocking density when data is not available); livestock groups (holdings can either have dairy and/or beef – any one activity or all); region (England, Wales, Scotland, Northern Ireland).

A stratified sample was drawn from this population in which the sample had the same proportionate split of holdings according to farm type. Farmers were removed from the sample if they met any of the following criteria: they were no longer active on the register (ceased farming); they were listed as a ‘stop’ (people to whom no correspondence was sent, e.g. recent bereavements).

During the 3 weeks prior to the survey (April 2011), 1420 opt-out letters were sent to farmers in England, Scotland, Wales and Northern Ireland. The opt-out letter stated the aim of the survey, approximate duration of the interview, underlined that the survey was voluntary and that it ensured respondent anonymity. The letters sent to the Welsh farmers were in both English and Welsh. Farmers who did not wish to participate were asked to return an enclosed form in a reply paid envelope provided, within 1 week. We allowed 2–3 weeks for opt-out letters to be returned by farmers before the survey started, and 81% of the farmers contacted by postal mail (opt-out letter stage) did not return their opt-out letters and implicitly agreed to participate in the telephone interview.

A pilot survey of 10 farmers from England, Scotland, Wales and Northern Ireland was conducted to identify any changes needed to the questionnaire before administration.

The telephone survey took place during May–June 2011. Overall, 405 farmers were contacted by telephone for the interview. The average duration of the interview was 17 minutes. The interviews were not audio-recorded but notes were taken by the interviewer and answers compiled in an SPSS database. Farmers were reassured that all information provided would be completely anonymous in any subsequent reports or publications and that they and their farms would never be individually identifiable. Any farmers wishing to opt out after the data was collected were able to do so.

RESULTS

Descriptive statistics for the sample

A total of 405 completed questionnaires were obtained forming a representative sample at the UK level (147 England, 123 Wales, 101 Scotland and 34 Northern Ireland; 309 beef and 96 dairy cattle farms). The total sample size of 405 farmers is consistent with methodological requirements (estimation method and number of measured parameters).

Regarding socio-economic characteristics, the sample consisted of 85% male farmers and 15% female farmers. Age distribution showed 38% of farmers aged <50 years, 40% between 51–65 years and 22% >65 years. For educational level, 45% of farmers finished school, 42% finished college and 12% finished university. With respect to farm holding, most farmers (61%) owned their farms, 25% were partly tenants/partly owners and 12% were tenants. About two thirds of farmers (63%) used an animal health plan to manage the health of their livestock. Regarding income, 75% of farmers had half or more of their income coming from livestock production and only about 5% of farmers had more than 5% of their income dependent on opening their farm to the public.

Regarding knowledge about the impact of E. coli O157 on human health, the majority of farmers were aware that E. coli O157 causes disease in people (82%), that people touching calves/cows may become infected with E. coli O157 (73%), that livestock were an important source from which E. coli O157 spreads (62%) and that E. coli O157 could be present on raw meat (76%), in raw milk (49%) and could contaminate produce such as lettuce, apples, spinach (51%) or rural drinking water (54%)Footnote . The survey also identified a number of farmers that demonstrated a lack of awareness of the different potential means of E. coli O157 transmission, which might imply that not all farmers implemented the necessary controls to prevent cross-contamination.

With regard to sources of useful information on E. coli O157, 79% of farmers stated media, followed by veterinary surgeons (44%), government (33%), other farmers (23%), industry organizations (19%) and the internet (6%)Footnote .

As regards perceived beneficiaries of on-farm controls to reduce E. coli O157 in cattle, 75% of farmers answered that all (farmers, processors, retailers, public, government) would benefit. Regarding perceived responsibility for controlling E. coli O157 on farms, 66% of farmers stated that responsibility remained with them; however, 21% of farmers stated that all should share responsibility and 12% of farmers considered that the government should be responsible for the control of E. coli O157 on-farms. Only around 19% of farmers agreed that E. coli O157 might be present in cattle on their farm, including 3% who strongly agreed with this statement.

With respect to E. coli O157 on farm control, the majority of farmers found as practical/very practical the following biosecurity measures: separating animals into different age groups for the majority of the time (74%), keeping bedding dry and replacing contaminated/wet bedding on a daily basis (65%), quarantine and testing of livestock brought to the farm (57%) and cleaning feed troughs daily (54%). Reducing current livestock numbers on the farm and disinfecting the animal sheds/pens weekly were found to be not at all practical by 44% and 40% of farmers, respectively.

Regarding willingness to control E. coli O157 on farm, a low majority of farmers (59%) stated they would be willing to use a treatment such as two doses of vaccine that would cost £5 to buy for each animal every year and given to cattle aged 3–18 months. For the majority (91%) of the farmers not willing to use vaccination, one of the reasons was lack of information, for 69% of farmers the cost was too expensive, while 49% said that it would take too much time to administer. However, 61% of farmers said that they would be encouraged to use vaccination if it was part of a national programme to benefit the reputation of the industry, while 44% of farmers stated they would be encouraged to use vaccination if it was used by other farmers. Additional reasons given by farmers against vaccination were the need for clear evidence of disease, regulation related (‘if it was obligatory’) and practical difficulties relating to the implementation of the vaccine.

While almost half of the farmers (47%) indicated that they would be willing to pay £1–5 and a seventh of farmers (14%) more than £5 per animal per year to ensure that E. coli O157 was not present on their own farm, almost a sixth (17%) of farmers answered that they would not be willing to spend any money. A ninth (11%) of farmers would be willing to spend time on a daily basis (30 minutes per day) to ensure that E. coli O157 was not present on their own farm; however, about a fifth (18%) would not spend more than 1 day per year and about an eighth (12%) of farmers would not be willing to spend any time at all.

Table 2 presents some descriptive statistics for the variables included in the model.

Table 2. Descriptive statistics

s.d., Standard deviation.

Results of the SEM

The path diagram for the estimated SEM is presented in Figure 1.

The model has a good fit according to the measures of absolute, incremental and parsimonious fit. The main GoF indicators (estimated and recommended values) for the estimated models are presented in Table 3.

Table 3. Goodness-of-fit indicators

The table presents absolute, incremental and parsimonious goodness-of-fit indicators. The absolute fit indicators include: root mean square error of approximation and goodness of fit index. Incremental fit indicators include: adjusted goodness of fit index, non-normed fit index, normed fit index, relative fit index, comparative fit index and incremental fit index. Parsimonious fit indicators include normed χ 2.

Additional testing of the appropriateness of the model was achieved by comparing the estimated model with two other models that acted as alternative explanations to the proposed model in a competing models strategy using a nested model approach. The results across all types of GoF measures favoured the estimated model in most cases. Therefore, we confirmed the accuracy of the proposed model and discarded the competing ones.

In the measurement model we tested the reliability of the single-indicator latent variables using the ‘theory-testing extremes’ of reliability within the range of 0·7–1 and determined that none of the structural coefficients became non-significant at these extremes.

After assessing the overall model and aspects of the measurement model, the standardized structural coefficients for both practical and theoretical implications were examined. Table 4 presents the standardized total effects between the latent variables in the model.

Table 4. Standardized total (direct and indirect) effects (t values in parentheses)

The latent variable scores and observational residuals depend on the unit of measurement in the observed variables. As some of these units are the result of subjective scaling of the observed variables the observational residuals were standardized (rescaled such that they have zero means and unit standard deviations in the sample) [Reference Jöreskog and Sörbom22]. Total effects represent how much a 1-unit change in an independent variable will change the expected value of a dependent variable.

Behavioural willingness (i.e. willingness to spend time to control E. coli O157 on farm; willingness to pay money to control E. coli O157 on farm; willingness to use vaccination to control E. coli O157 on farm) is significantly influenced by perceptions of farmers being most responsible for E. coli O157 on-farm control; attitudes towards E. coli O157 on-farm control; proportion of farm income dependent on opening to the public; and perceived effects of E. coli O157 on business. Additionally, willingness to pay money or vaccinate are significantly influenced by frequency of access to information and knowledge about E. coli O157, while willingness to spend time to control the disease is influenced by perceived practicality of biosecurity measures for E. coli O157 on-farm control; use of a health plan; perceptions of farmers as main beneficiaries of E. coli O157 on-farm control; and farm type.

The model has a good level of prediction as it explains more than half (52% and 76%) of the variance in willingness to pay money and, respectively, willingness to use vaccination to ensure that E. coli O157 is not present on their own farm. A lower but still significant level of prediction (42%) is shown for willingness to spend time to control E. coli O157 on farm.

Of the factors influencing all behavioural willingness variables, the highest effect is shown by the perceptions of farmers being most responsible for E. coli O157 on-farm control (50%, 36% and 50% ceteris paribus on willingness to spend time, money and vaccinate, respectively). This suggests that farmers who feel responsible towards controlling E. coli O157 on-farm control are more likely to be willing to use control measures. The variable has a direct impact on willingness to vaccinate and, through it, an indirect impact on willingness to spend money to control the disease (Fig. 1). Its impact on willingness to spend time to control the disease is both direct and indirect through use of a health plan (Fig. 1). The latter implies that responsibility towards disease control influences a farm's adoption of an animal health plan.

Farmers’ attitudes towards E. coli O157 on-farm control is another main factor influencing all behavioural willingness variables (14%, 31% and 42% ceteris paribus on willingness to spend time, money and vaccinate, respectively). The variable has a direct impact on willingness to vaccinate and, through it, an indirect impact on willingness to spend money to control the disease (Fig. 1). Its impact on willingness to spend time to control the disease is indirect through perceptions of biosecurity practicality (Fig. 1). This confirms one of the main facts of behavioural theories, namely that attitudes precede intentions and behaviour. Farmers with stronger attitudes towards E. coli O157 control are more likely to have stronger perceptions of the practicality of control measures and be willing to control the disease.

Another main influence on behavioural willingness is the proportion of farm income dependent on opening to the public (34%, 26% and 37% ceteris paribus on willingness to spend time, money and vaccinate, respectively). The variable has a direct impact on willingness to spend time to control the disease and on willingness to vaccinate and, through the latter, an indirect impact on willingness to spend money to control the disease (Fig. 1). This suggests that farmers whose income depends more on opening their farms to public are more likely to be willing to vaccinate or spend more money/time to control E. coli O157.

Perceived effect of reports/experience/incidents of E. coli O157 outbreaks or incidents on the way of managing business during the past 5 years has a significant influence on behavioural willingness (4%, 24% and 33% ceteris paribus on willingness to spend time, money and vaccinate, respectively). This suggests that farmers whose livestock was affected by disease in the past or who know other farmers affected by it are more likely to be willing to control the disease. There is a large difference between the impact on willingness to spend time and the impact on willingness to vaccinate/spend money to control disease, which may suggest that farmers affected by E. coli O157 in the past consider vaccination as more effective than other less-expensive but more time-consuming measures. The variable has a direct impact on willingness to vaccinate and, through it, an indirect one on willingness to spend money to control the disease (Fig. 1). Its impact on willingness to spend time to control the disease is indirect through perceptions of farmers as main beneficiaries of E. coli O157 on-farm control (Fig. 1). The latter might suggest that farmers who experienced the impact of disease are more likely to be aware of the benefits of controlling it.

Besides the four factors above influencing all behavioural willingness variables, willingness to vaccinate or spend money to control the disease are also influenced by information and knowledge.

There is a very strong relationship between intention to vaccinate and the more general intention to spend money to control E. coli O157 on farm (72% ceteris paribus). This suggests that, as vaccination would be an expensive exercise, farmers willing to use it are more likely to be willing to spend money on this and/or other measures of E. coli O157 control.

Access to information has a significant influence – directly and indirectly through knowledge (Fig. 1) – on willingness to use vaccination (18% ceteris paribus) and, through it, an indirect one on willingness to spend money to control E. coli O157 on farm (13% ceteris paribus). This suggests that farmers with more frequent access to information are more likely to have knowledge about E. coli O157 and be willing to control E. coli O157 on farm.

Knowledge about E. coli O157 has a significant influence on willingness to use vaccination (16% ceteris paribus) and, through it, an indirect one on willingness to spend money to control E. coli O157 on farm (12% ceteris paribus) (Fig. 1). This supports the scientific evidence on the established linkage between knowledge and behavioural intentions and shows knowledge of E. coli O157 as a necessary antecedent of intention to vaccinate or spend money to control E. coli O157 on farm.

Willingness to spend time to control E. coli O157 on farm is also influenced by perceived practicality of biosecurity, use of a health plan, perceived benefits and farm type. The model included a significant relationship between access to information and use of a health plan (25% ceteris paribus); however, this influence did not mediate a significant impact of information on willingness to spend time to control disease (Fig. 1).

Farmers’ perceptions of biosecurity measures to be practical/suited to the needs of their farms significantly influence willingness to spend time to control E. coli O157 on farm (27% ceteris paribus). As the biosecurity measures included in the model are time consuming, farmers’ perceptions of their practicality and suitability for the control of E. coli O157 on farm would influence their willingness to spend more time in controlling the disease.

Use of a health plan has a lower but significant indirect – through perceptions of biosecurity practicality (Fig. 1) – influence on willingness to spend time to control E. coli O157 on farm (9% ceteris paribus). This might suggest that farmers using a health plan, which is likely to include biosecurity measures, are more likely to find these measures as suitable to control the disease and, implicitly, be willing to spend more time on disease control.

Perceptions of farmers as the main beneficiaries of E. coli O157 on-farm control have a lower but significant indirect – through perceptions of biosecurity practicality (Fig. 1) – influence on willingness to spend time to control E. coli O157 on farm (9% ceteris paribus). This implies that farmers who think they benefit from disease control are more likely to perceive the practicality of biosecurity measures and, implicitly, be willing to spend more time on disease control.

Farm type has a low but significant influence (5% ceteris paribus) on willingness to spend time to control E. coli O157 on farm. The effect is indirect through use of a health plan (Fig. 1). This implies that dairy farmers rather than beef farmers are more likely to use a health plan and be willing to spend more time in controlling E. coli O157 on farm.

DISCUSSION

The study analysed the impact of a priori determinants of adoption of E. coli O157 control measures by cattle farmers in the UK. We used a dataset collected through a stratified telephone survey of 405 cattle farmers in the UK and SEM with observed and latent variables to test the influence of a priori identified determinants on behavioural intentions towards E. coli O157 control.

The results confirm findings from the literature and expert opinion. The model has a good level of prediction as it explains a high percentage of the variance in willingness to control E. coli O157 on farm. However, the level of prediction could improve if other factors were added to the model. The literature on farmers’ attitudes and behaviour towards control measures of E. coli O157 is currently limited. More research is needed, especially of the exploratory type (e.g. using qualitative data from in-depth interviews or focus groups) to identify other factors influencing farmers’ behaviour regarding disease control.

The results of this study will contribute to the existing evidence and will potentially assist policy makers in finding means of behavioural change.

The model suggests that farmers more likely to show a higher willingness to control E. coli O157 on farm are those: with stronger perceptions of responsibility towards E. coli O157 on-farm control [Reference Ellis-Iversen13]; with stronger attitudes towards E. coli O157 on-farm control [Reference Toma9, Reference Garforth, Bailey and Tranter12, Reference Ellis-Iversen13]; with higher proportion of the farm income dependent on opening to the public [Reference Chilonda and Van10, Reference Stott11]; who were affected by E. coli O157 incidents in the past [Reference Toma9, Reference Garforth, Bailey and Tranter12, Reference Coleman, Hemsworth and Hay16Reference Lindberg19]; who are more informed and have better knowledge about E. coli O157 [Reference Garforth, Bailey and Tranter12Reference Strachan15]; with stronger perceptions of practicality of biosecurity measures for E. coli O157 on-farm control [Reference Garforth, Bailey and Tranter12, Reference Braun20]; who use a health plan [Reference Toma9]; with stronger perceptions of benefits of E. coli O157 on-farm control [Reference Garforth, Bailey and Tranter12]; and whose farms are dairy rather than beef.

This might imply that increasing access to information to all farmers and targeting more specifically dairy farmers, farmers who open their farms to public and farmers affected by past outbreaks might lead to better knowledge, stronger perceptions and attitudes and, consequently, higher willingness to control E. coli O157 on farm.

The fact that responsibility perceptions were found to have the strongest effect on behavioural willingness to control the disease might suggest the need not only to increase access to information, but to provide information on sources and modes of E. coli O157 transmission.

Similarly, the fact that perceived practicality of biosecurity measures was found to have a strong effect on willingness to control the disease might suggest the need to provide information on control measures to suit the specific circumstances of farms.

Farmers’ intentions to control E. coli O157 on-farm are influenced by their attitudes with regard to potential rewards, such as increase in the price of their products or enhanced reputation with customers if they used control measures. This might suggest that if major retailers and buyers of milk and beef would provide incentives, farmers would be more willing to apply proven E. coli O157 on-farm control.

The findings might suggest that farmers may implement on-farm controls for E. coli O157 if they identify a clear hazard and if there is greater knowledge of the safety and efficacy of the proposed controls. Despite farmers recognizing a responsibility for the potentially negative consequences that maintaining cattle and spreading this pathogen poses to the public, for the majority of farmers there is a lack of validated on-farm control options, and the lack of a clear link between human cases of infection and their own livestock. This might suggest the need to provide information on safety and efficiency of control options in addition to modes of disease transmission.

Fig. 1. Conceptual path diagram for the estimated model showing the drivers of farmers’ behavioural intentions towards Escherichia coli O157 on-farm control. The arrows indicate direction of influence of each latent variable on another; thick bold arrows represent direct influences on behavioural intentions; thin bold arrows represent indirect influences on behavioural intentions. WTP, Willingness to pay.

ACKNOWLEDGEMENTS

We thank the Food Standards Agency in Scotland and Defra who funded this research. We also thank the respondents to our survey. SRUC receives grant-in-aid from the Scottish Government.

DECLARATION OF INTEREST

None.

Footnotes

The percentages relate only to those farmers who had heard of E. coli O157 prior to the survey (73% of the total sample).

References

REFERENCES

1. Parry, SM, et al. Risk factors for and prevention of sporadic infections with verocytotoxin (shiga toxin) producing Escherichia coli O157. Lancet 1998; 351: 1019–22.CrossRefGoogle ScholarPubMed
2. O'Brien, SJ, Adak, GK, Gilham, C. Contact with the farming environment as a major risk factor for sporadic cases of Shiga toxin-producing Escherichia coli O157 infection in humans. Emerging Infectious Diseases 2001; 7: 10491051.CrossRefGoogle Scholar
3. Locking, ME, et al. Risk factors for sporadic cases of Escherichia coli O157 infection: the importance of contact with animal excreta. Epidemiology and Infection 2001; 127: 215–20.Google Scholar
4. Rangel, JM, et al. Epidemiology of Escherichia coli O157:H7 outbreaks, United States, 1982–2002. Emerging Infectious Diseases 2005; 11: 603609.Google Scholar
5. Griffin, G. Review of the major outbreak of E. coli O157 in Surrey, 2009. Final Report of the Independent Investigation Committee, 2010.Google Scholar
6. Matthews, L, et al. Predicting the public health benefit of vaccinating cattle against Escherichia coli O157. Proceedings of the National Academy of Sciences USA 2013; 110: 16265–16270.Google Scholar
7. European Parliament. European Parliament legislative resolution on the Council common position with a view to the adoption of a European Parliament and Council regulation on the hygiene of foodstuffs (10543/2/2002-C5-0008/2004-2000/0178(COD)). P5_TA(2004)0216. European Parliament, 2004, Strasbourg.Google Scholar
8. Gunn, GJ, et al. Measuring and comparing constraints to improved biosecurity amongst GB farmers, veterinarians and the auxiliary industries. Preventive Veterinary Medicine 2008; 84: 310334.CrossRefGoogle ScholarPubMed
9. Toma, L, et al. Determinants of biosecurity behaviour of British cattle and sheep farmers – a behavioural economics analysis. Preventive Veterinary Medicine 2013; 108: 321333.Google Scholar
10. Chilonda, P, Van, Huylenbroeck G. A conceptual framework for the economic analysis of factors influencing decision-making of small-scale farmers in animal health management. Revue Scientifique et Technique (International Office of Epizootics) 2001; 20: 687700.Google Scholar
11. Stott, AW, et al. A linear programming approach to estimate the economic impact of bovine viral diarrhoea (BVD) at the whole-farm level in Scotland. Preventive Veterinary Medicine 2003; 59: 5166.Google Scholar
12. Garforth, CJ, Bailey, AP, Tranter, RB. Farmers’ attitudes to disease risk management in England: a comparative analysis of sheep and pig farmers. Preventive Veterinary Medicine 2013; 110: 456466.Google Scholar
13. Ellis-Iversen, J, et al. Perceptions, circumstances and motivators that influence implementation of zoonotic control programs on cattle farms. Preventive Veterinary Medicine 2010; 93: 276285.CrossRefGoogle ScholarPubMed
14. Jones, CDR, et al. Escherichia coli O157: comparing awareness of rural residents and visitors in livestock farming areas. Epidemiology and Infection 2011; 139: 15221530.Google Scholar
15. Strachan, NJC, et al. The relationship between lay and technical views of Escherichia coli O157 risk. Philosophical Transactions of the Royal Society of London, Series B 2011; 366: 19992009.Google Scholar
16. Coleman, GJ, Hemsworth, PH, Hay, M. Predicting stockperson behaviour towards pigs from attitudinal and job-related variables and empathy. Applied Animal Behaviour Science 1998; 58: 6375.Google Scholar
17. Ekboir, JM. The role of the public sector in the development and implementation of animal health policies. Preventive Veterinary Medicine 1999; 40: 101115.Google Scholar
18. Delabbio, J. An assessment of bio-security utilization in the recirculation sector of finfish aquaculture in the United States and Canada. Aquaculture 2004; 242: 165179.CrossRefGoogle Scholar
19. Lindberg, A, et al. The control of bovine viral diarrhoea virus in Europe: today and in the future. Revue Scientifique et Technique (International Office of Epizootics) 2006; 25: 961979.Google Scholar
20. Braun, AR, et al. A global survey and review of farmer field school experiences. International Livestock Research Institute Report, ILRI, 2006.Google Scholar
21. Cross, P, Rigby, D, Edwards-Jones, G. Eliciting expert opinion on the effectiveness and practicality of interventions in the farm and rural environment to reduce human exposure to Escherichia coli O157. Epidemiology and Infection 2012; 140: 643654.Google Scholar
22. Jöreskog, KG, Sörbom, D. LISREL8·80: Structural Equation Modeling with the SIMPLIS Command Language. Chicago, USA: IL Scientific Software International, 2007.Google Scholar
23. Muthén, B. A general structural equation model with dichotomous, ordered categorical, and continuous latent variable indicators. Psychometrika 1984; 49: 115132.Google Scholar
24. Bollen, KA. Structural Equations with Latent Variables. New York: John Wiley and Sons, 1989.Google Scholar
25. Finney, SJ, DiStefano, C. Non-normal and Categorical data in structural equation modeling. In: Hancock, GR, Mueller, RO, eds. Structural Equation Modeling: a Second Course. Greenwich, Connecticut: Information Age Publishing, 2006, pp. 269314.Google Scholar
26. Hair, JF, et al. Multivariate Data Analysis, 6th edn, Upper Saddle River, NJ: Pearson Prentice Hall, 2006.Google Scholar
27. Ping, RA. How does one specify and estimate latent variables with only 1 or 2 indicators? (http://home.att.net/~rpingjr/Under_Det.doc), 2008.Google Scholar
Figure 0

Table 1. Description of latent variables and their corresponding indicators

Figure 1

Table 2. Descriptive statistics

Figure 2

Table 3. Goodness-of-fit indicators

Figure 3

Table 4. Standardized total (direct and indirect) effects (t values in parentheses)

Figure 4

Fig. 1. Conceptual path diagram for the estimated model showing the drivers of farmers’ behavioural intentions towards Escherichia coli O157 on-farm control. The arrows indicate direction of influence of each latent variable on another; thick bold arrows represent direct influences on behavioural intentions; thin bold arrows represent indirect influences on behavioural intentions. WTP, Willingness to pay.