Identification of potential risk factors for Toxoplasma gondii in fattening pigs in the Netherlands using a Bayesian approach

D.M. Eppink, M. Bouwknegt, D. Oorburg, H.A.P. Urlings, Marcel van Asseldonk, Coen van Wagenberg, Inge Krijger, J.W.P. van der Giessen, M. Swanenburg, H.J. Wisselink

Research output: Contribution to conferenceAbstractAcademic

Abstract

Introduction
Toxoplasma gondii is a relevant foodborne pathogen,it is estimated that up to one third of the worldpopulation has been exposed to the parasite (Tenteret al. 2000). In the Netherlands toxoplasmosis rankssecond on a list of prioritized emerging zoonosis(Havelaar et al. 2010) and also second in disease burden among 14 foodborne diseases (Mangen et al.2017). Data suggest that ingesting improperly cooked meat containing T. gondii is one of the major sources of infection in Europe and North America (Crotta et al. 2017; Guo et al. 2015). The contribution of pork to meatborne T. gondii infections is estimated to be11 % in the Netherlands (Opsteegh 2011) and is seen as an important possible source of human T. gondii infections (Foroutan et al. 2019). The European Food Safety Authority (EFSA) advised to perform serological testing of pigs and on farm audits on risk factors (EFSA 2011). To that end, a serological monitoring program was developed in a slaughterhouse in the Netherlands. In this study, the objective is to determine the association between within herd seroprevalence, corrected for misclassification of samples through Bayesian analyses, and risk factors for T. gondii on fattening pig farms in TheNetherlands.
Materials and MethodsFrom 2015 to 2018, HACCP based audits were performedon 75 fattening pig farms in The Netherlands to identify the presence of potential T. gondii risk factors. All farms were conventional pig farms, with 15 farms being farrow to finish. As overall seroprevalence of T. gondii in pigs in the Netherlands is low, estimated at 5 % (1-12 % 95 % CI) by Foroutanet al. 2019, approached farms were chosen with the knowledge of previous serology data. In this way there would be farms with positive serum samples and farms without them included in the study. The audits were based on an updated version of the questionnaire from Mul et al. (2015) and covered the following topics: outdoor access, farm biosecurity, rodent control, presence of cats, feed and watersupply. In addition, serum samples (n=6272) from fattening pigs were obtained at slaughter throughout the year before the audit on the farm was performed. These samples were used for antibody testing bya PrioCHECK™ Toxoplasma Antibody ELISA. Data were analysed using Bayesian statistics, with the within farm T. gondii prevalence as dependent variable and potential risk factors as independent variables. As always with serology, misclassification due to false-positive or false-negative results can occur. Statistical methods have been developed to account for such misclassification, based on frequentistic as well as Bayesian approaches (Hui & Walter 1980; Josephet al. 1995). First, all independent variables wereanalysed in a univariate logistic model, and variables with a probability ≤0.25 that zero is included in the 95 % interval were analysed in a multivariable model. The multivariate logistic model was fitted using backward elimination until all remaining variables showed a probability ≤0.05 that zero is included in the 95 % interval. Two-way interaction terms were evaluated similarly to the main variables regarding statistical significance.
ResultsDescriptive results showed that 50 out of the 75 farms had 1 or more positive serum samplein the year before the audit was performed. In total 438 samples were positive out of the 6272 samples. Final Bayesian analyses are currently being conducted. However, preliminary results from data analysis using frequentistic logistic multivariate regression identified two significant risk factors: the accessibility of pig feed for cats and theprovision of well water as drinking water for the pigs (Table 1).
Discussion and ConclusionsThe use of serological testing seems to be a valuable guide and monitoring tool for the control of T.gondii in pork production. In a preliminary analysis, a higher within-herd T. gondii seroprevalence on fattening pig farms in the Netherlands was associated with the accessibility of pig feed for cats and the provision of well water as drinking water for the pigs. Improvements in farm management on fattening pig farms will likely contribute to reduction of the human disease burden and is presently studied.
Original languageEnglish
Publication statusPublished - 26 Aug 2019
Event13th SafePork 2019: ONE HEALTH - Tear down interdisciplinary walls - Berlin, Germany
Duration: 26 Aug 201929 Aug 2019

Conference

Conference13th SafePork 2019
CountryGermany
CityBerlin
Period26/08/1929/08/19

Fingerprint

Toxoplasma gondii
finishing
Netherlands
risk factors
farms
swine
audits
blood serum
seroprevalence
burden of disease
logit analysis
cats
sampling
pork
drinking water
food safety
herds
infection
rodent control
Toxoplasma

Cite this

Eppink, D.M. ; Bouwknegt, M. ; Oorburg, D. ; Urlings, H.A.P. ; van Asseldonk, Marcel ; van Wagenberg, Coen ; Krijger, Inge ; van der Giessen, J.W.P. ; Swanenburg, M. ; Wisselink, H.J. / Identification of potential risk factors for Toxoplasma gondii in fattening pigs in the Netherlands using a Bayesian approach. Abstract from 13th SafePork 2019, Berlin, Germany.
@conference{82afe3ff3b7c4523aab49955bd3e0a85,
title = "Identification of potential risk factors for Toxoplasma gondii in fattening pigs in the Netherlands using a Bayesian approach",
abstract = "IntroductionToxoplasma gondii is a relevant foodborne pathogen,it is estimated that up to one third of the worldpopulation has been exposed to the parasite (Tenteret al. 2000). In the Netherlands toxoplasmosis rankssecond on a list of prioritized emerging zoonosis(Havelaar et al. 2010) and also second in disease burden among 14 foodborne diseases (Mangen et al.2017). Data suggest that ingesting improperly cooked meat containing T. gondii is one of the major sources of infection in Europe and North America (Crotta et al. 2017; Guo et al. 2015). The contribution of pork to meatborne T. gondii infections is estimated to be11 {\%} in the Netherlands (Opsteegh 2011) and is seen as an important possible source of human T. gondii infections (Foroutan et al. 2019). The European Food Safety Authority (EFSA) advised to perform serological testing of pigs and on farm audits on risk factors (EFSA 2011). To that end, a serological monitoring program was developed in a slaughterhouse in the Netherlands. In this study, the objective is to determine the association between within herd seroprevalence, corrected for misclassification of samples through Bayesian analyses, and risk factors for T. gondii on fattening pig farms in TheNetherlands.Materials and MethodsFrom 2015 to 2018, HACCP based audits were performedon 75 fattening pig farms in The Netherlands to identify the presence of potential T. gondii risk factors. All farms were conventional pig farms, with 15 farms being farrow to finish. As overall seroprevalence of T. gondii in pigs in the Netherlands is low, estimated at 5 {\%} (1-12 {\%} 95 {\%} CI) by Foroutanet al. 2019, approached farms were chosen with the knowledge of previous serology data. In this way there would be farms with positive serum samples and farms without them included in the study. The audits were based on an updated version of the questionnaire from Mul et al. (2015) and covered the following topics: outdoor access, farm biosecurity, rodent control, presence of cats, feed and watersupply. In addition, serum samples (n=6272) from fattening pigs were obtained at slaughter throughout the year before the audit on the farm was performed. These samples were used for antibody testing bya PrioCHECK™ Toxoplasma Antibody ELISA. Data were analysed using Bayesian statistics, with the within farm T. gondii prevalence as dependent variable and potential risk factors as independent variables. As always with serology, misclassification due to false-positive or false-negative results can occur. Statistical methods have been developed to account for such misclassification, based on frequentistic as well as Bayesian approaches (Hui & Walter 1980; Josephet al. 1995). First, all independent variables wereanalysed in a univariate logistic model, and variables with a probability ≤0.25 that zero is included in the 95 {\%} interval were analysed in a multivariable model. The multivariate logistic model was fitted using backward elimination until all remaining variables showed a probability ≤0.05 that zero is included in the 95 {\%} interval. Two-way interaction terms were evaluated similarly to the main variables regarding statistical significance.ResultsDescriptive results showed that 50 out of the 75 farms had 1 or more positive serum samplein the year before the audit was performed. In total 438 samples were positive out of the 6272 samples. Final Bayesian analyses are currently being conducted. However, preliminary results from data analysis using frequentistic logistic multivariate regression identified two significant risk factors: the accessibility of pig feed for cats and theprovision of well water as drinking water for the pigs (Table 1).Discussion and ConclusionsThe use of serological testing seems to be a valuable guide and monitoring tool for the control of T.gondii in pork production. In a preliminary analysis, a higher within-herd T. gondii seroprevalence on fattening pig farms in the Netherlands was associated with the accessibility of pig feed for cats and the provision of well water as drinking water for the pigs. Improvements in farm management on fattening pig farms will likely contribute to reduction of the human disease burden and is presently studied.",
author = "D.M. Eppink and M. Bouwknegt and D. Oorburg and H.A.P. Urlings and {van Asseldonk}, Marcel and {van Wagenberg}, Coen and Inge Krijger and {van der Giessen}, J.W.P. and M. Swanenburg and H.J. Wisselink",
year = "2019",
month = "8",
day = "26",
language = "English",
note = "13th SafePork 2019 : ONE HEALTH - Tear down interdisciplinary walls ; Conference date: 26-08-2019 Through 29-08-2019",

}

Eppink, DM, Bouwknegt, M, Oorburg, D, Urlings, HAP, van Asseldonk, M, van Wagenberg, C, Krijger, I, van der Giessen, JWP, Swanenburg, M & Wisselink, HJ 2019, 'Identification of potential risk factors for Toxoplasma gondii in fattening pigs in the Netherlands using a Bayesian approach' 13th SafePork 2019, Berlin, Germany, 26/08/19 - 29/08/19, .

Identification of potential risk factors for Toxoplasma gondii in fattening pigs in the Netherlands using a Bayesian approach. / Eppink, D.M.; Bouwknegt, M.; Oorburg, D.; Urlings, H.A.P.; van Asseldonk, Marcel; van Wagenberg, Coen; Krijger, Inge; van der Giessen, J.W.P.; Swanenburg, M.; Wisselink, H.J.

2019. Abstract from 13th SafePork 2019, Berlin, Germany.

Research output: Contribution to conferenceAbstractAcademic

TY - CONF

T1 - Identification of potential risk factors for Toxoplasma gondii in fattening pigs in the Netherlands using a Bayesian approach

AU - Eppink, D.M.

AU - Bouwknegt, M.

AU - Oorburg, D.

AU - Urlings, H.A.P.

AU - van Asseldonk, Marcel

AU - van Wagenberg, Coen

AU - Krijger, Inge

AU - van der Giessen, J.W.P.

AU - Swanenburg, M.

AU - Wisselink, H.J.

PY - 2019/8/26

Y1 - 2019/8/26

N2 - IntroductionToxoplasma gondii is a relevant foodborne pathogen,it is estimated that up to one third of the worldpopulation has been exposed to the parasite (Tenteret al. 2000). In the Netherlands toxoplasmosis rankssecond on a list of prioritized emerging zoonosis(Havelaar et al. 2010) and also second in disease burden among 14 foodborne diseases (Mangen et al.2017). Data suggest that ingesting improperly cooked meat containing T. gondii is one of the major sources of infection in Europe and North America (Crotta et al. 2017; Guo et al. 2015). The contribution of pork to meatborne T. gondii infections is estimated to be11 % in the Netherlands (Opsteegh 2011) and is seen as an important possible source of human T. gondii infections (Foroutan et al. 2019). The European Food Safety Authority (EFSA) advised to perform serological testing of pigs and on farm audits on risk factors (EFSA 2011). To that end, a serological monitoring program was developed in a slaughterhouse in the Netherlands. In this study, the objective is to determine the association between within herd seroprevalence, corrected for misclassification of samples through Bayesian analyses, and risk factors for T. gondii on fattening pig farms in TheNetherlands.Materials and MethodsFrom 2015 to 2018, HACCP based audits were performedon 75 fattening pig farms in The Netherlands to identify the presence of potential T. gondii risk factors. All farms were conventional pig farms, with 15 farms being farrow to finish. As overall seroprevalence of T. gondii in pigs in the Netherlands is low, estimated at 5 % (1-12 % 95 % CI) by Foroutanet al. 2019, approached farms were chosen with the knowledge of previous serology data. In this way there would be farms with positive serum samples and farms without them included in the study. The audits were based on an updated version of the questionnaire from Mul et al. (2015) and covered the following topics: outdoor access, farm biosecurity, rodent control, presence of cats, feed and watersupply. In addition, serum samples (n=6272) from fattening pigs were obtained at slaughter throughout the year before the audit on the farm was performed. These samples were used for antibody testing bya PrioCHECK™ Toxoplasma Antibody ELISA. Data were analysed using Bayesian statistics, with the within farm T. gondii prevalence as dependent variable and potential risk factors as independent variables. As always with serology, misclassification due to false-positive or false-negative results can occur. Statistical methods have been developed to account for such misclassification, based on frequentistic as well as Bayesian approaches (Hui & Walter 1980; Josephet al. 1995). First, all independent variables wereanalysed in a univariate logistic model, and variables with a probability ≤0.25 that zero is included in the 95 % interval were analysed in a multivariable model. The multivariate logistic model was fitted using backward elimination until all remaining variables showed a probability ≤0.05 that zero is included in the 95 % interval. Two-way interaction terms were evaluated similarly to the main variables regarding statistical significance.ResultsDescriptive results showed that 50 out of the 75 farms had 1 or more positive serum samplein the year before the audit was performed. In total 438 samples were positive out of the 6272 samples. Final Bayesian analyses are currently being conducted. However, preliminary results from data analysis using frequentistic logistic multivariate regression identified two significant risk factors: the accessibility of pig feed for cats and theprovision of well water as drinking water for the pigs (Table 1).Discussion and ConclusionsThe use of serological testing seems to be a valuable guide and monitoring tool for the control of T.gondii in pork production. In a preliminary analysis, a higher within-herd T. gondii seroprevalence on fattening pig farms in the Netherlands was associated with the accessibility of pig feed for cats and the provision of well water as drinking water for the pigs. Improvements in farm management on fattening pig farms will likely contribute to reduction of the human disease burden and is presently studied.

AB - IntroductionToxoplasma gondii is a relevant foodborne pathogen,it is estimated that up to one third of the worldpopulation has been exposed to the parasite (Tenteret al. 2000). In the Netherlands toxoplasmosis rankssecond on a list of prioritized emerging zoonosis(Havelaar et al. 2010) and also second in disease burden among 14 foodborne diseases (Mangen et al.2017). Data suggest that ingesting improperly cooked meat containing T. gondii is one of the major sources of infection in Europe and North America (Crotta et al. 2017; Guo et al. 2015). The contribution of pork to meatborne T. gondii infections is estimated to be11 % in the Netherlands (Opsteegh 2011) and is seen as an important possible source of human T. gondii infections (Foroutan et al. 2019). The European Food Safety Authority (EFSA) advised to perform serological testing of pigs and on farm audits on risk factors (EFSA 2011). To that end, a serological monitoring program was developed in a slaughterhouse in the Netherlands. In this study, the objective is to determine the association between within herd seroprevalence, corrected for misclassification of samples through Bayesian analyses, and risk factors for T. gondii on fattening pig farms in TheNetherlands.Materials and MethodsFrom 2015 to 2018, HACCP based audits were performedon 75 fattening pig farms in The Netherlands to identify the presence of potential T. gondii risk factors. All farms were conventional pig farms, with 15 farms being farrow to finish. As overall seroprevalence of T. gondii in pigs in the Netherlands is low, estimated at 5 % (1-12 % 95 % CI) by Foroutanet al. 2019, approached farms were chosen with the knowledge of previous serology data. In this way there would be farms with positive serum samples and farms without them included in the study. The audits were based on an updated version of the questionnaire from Mul et al. (2015) and covered the following topics: outdoor access, farm biosecurity, rodent control, presence of cats, feed and watersupply. In addition, serum samples (n=6272) from fattening pigs were obtained at slaughter throughout the year before the audit on the farm was performed. These samples were used for antibody testing bya PrioCHECK™ Toxoplasma Antibody ELISA. Data were analysed using Bayesian statistics, with the within farm T. gondii prevalence as dependent variable and potential risk factors as independent variables. As always with serology, misclassification due to false-positive or false-negative results can occur. Statistical methods have been developed to account for such misclassification, based on frequentistic as well as Bayesian approaches (Hui & Walter 1980; Josephet al. 1995). First, all independent variables wereanalysed in a univariate logistic model, and variables with a probability ≤0.25 that zero is included in the 95 % interval were analysed in a multivariable model. The multivariate logistic model was fitted using backward elimination until all remaining variables showed a probability ≤0.05 that zero is included in the 95 % interval. Two-way interaction terms were evaluated similarly to the main variables regarding statistical significance.ResultsDescriptive results showed that 50 out of the 75 farms had 1 or more positive serum samplein the year before the audit was performed. In total 438 samples were positive out of the 6272 samples. Final Bayesian analyses are currently being conducted. However, preliminary results from data analysis using frequentistic logistic multivariate regression identified two significant risk factors: the accessibility of pig feed for cats and theprovision of well water as drinking water for the pigs (Table 1).Discussion and ConclusionsThe use of serological testing seems to be a valuable guide and monitoring tool for the control of T.gondii in pork production. In a preliminary analysis, a higher within-herd T. gondii seroprevalence on fattening pig farms in the Netherlands was associated with the accessibility of pig feed for cats and the provision of well water as drinking water for the pigs. Improvements in farm management on fattening pig farms will likely contribute to reduction of the human disease burden and is presently studied.

M3 - Abstract

ER -

Eppink DM, Bouwknegt M, Oorburg D, Urlings HAP, van Asseldonk M, van Wagenberg C et al. Identification of potential risk factors for Toxoplasma gondii in fattening pigs in the Netherlands using a Bayesian approach. 2019. Abstract from 13th SafePork 2019, Berlin, Germany.