Home

Publisher version

1. Source Selection Broilers Change scenario target prevalence gt Layers Use Subtype Target Prevalence V _ ENTERITIDIS a E AGONA m ANATUM F Groups of Subtypes a BOVISMORBIFICANS be Turkeys TYPHIMURIUM P 0 0240 A 0 0100 E BRAENDERUP ys 0 20 a j I cz Turkeys ENTERITIDIS P 0 0400 A 0 0100 E BRANDENBURG cz Turkeys TYPHIMURIUM P 0 0140 A 0 0100 BREDENEY DE Turkeys TYPHIMURIUM P 0 0180 A 0 0100 E DERBY Wes Turkeys TYPHIMURIUM P 0 0160 A 0 0100 FR Turkeys ENTERITIDIS P 0 0120 A 0 0100 HADAR FR Turkeys TYPHIMURIUM P 0 0150 A 0 0100 E HEIDELBERG I Hu Turkeys ENTERITIDIS P 0 0130 A 0 0100 m INFANTIS IT Turkeys TYPHIMURIUM P 0 0390 A 0 0100 PL Turkeys TYPHIMURIUM P 0 0300 A 0 0100 kentucky ik Tem TYPHIMURIUM P10 0320 A 0 0100 E aras i as vee m LIvinGsTONE LONDON MBANDAKA MONTEVIDEO E NEWPORT l RISSEN i SAINTPAUL TYPHIMURIUM Here you select the food source and whether you want to change the target prevalence for individual subtypes or for groups of subtypes 6 1 Individual subtypes As shown in the screenshot above you will see the subtypes present in the analysis in the right side table Click the check box and set a target prevalence value for that specific subtype Multiple subtypes can be
2. serotypelj i lt sum truecasescji 1 ReportingCountryCount j i serotypescenario j i lt sum truecasesscenariocji 1 ReportingCountryCount j i for c in 1 ReportingCountryCount for i in 1 SeroVarCount 1 truecasescil c i lt sum truecasescjil c 1 FoodSourceCount i truecasesscenarioci c i lt sum truecasesscenariocji c FoodSourceCount i for j in 1 FoodSourceCount source c j lt sum truecasescji c j SeroVarCount 1 sourcescenario c j lt sum truecasesscenariocji c j 1 SeroVarCount 1 for k in 1 SourceCountryCount for j in 1 FoodSourceCount source2 k j lt sum truecasesorigin 1 ReportingCountryCount k j source2scenario k j lt sum truecasesoriginscenario 1 ReportingCountryCount k j for c in 1 ReportingCountryCount travel c lt sum yt c unknown 1 c lt spdo c SeroVarCount sum uk c 1 SeroVarCount 1 unknown c lt unknown1 c Supporting publications 2012 EN 318 69 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food S
3. source CY BROILERS source CY LAYERS source Cy PIGS source CY TURKEYS source CZ BROILERS source CZ LAYERS source CZ PIGS source CZ TURKEYS source DE BROILERS source DE LAYERS source DE PIGS source DE TURKEYS source DK BROILERS source DK LAYERS source DK PIGS The data from the variables selected before running the model is added to individual grids and tables The node names are reformatted to contain the dimension values for easy identification of the data points As seen in the screen shot above source 1 1 the node name actually outputted from WinBUGS is renamed to source AT Broilers which gives the number of estimated human cases attributed to broilers in Austria All data including imported data analysis settings resulting data and model code will be saved in the database for later retrieval This ensures that it is possible to see exactly how the results were produced years after it was done Also it is possible to export all this as a zip file Each individual data file is also available The resulting data values can also be exported as individual files for further analysis in an easy to use format To export an analysis as a zip file select Open Analysis and press Export to file Then select a destination folder for your file and press Save a Analyses Analysis Name Created gt test45 21 05 2012 test34 21 05 2012 Ne
4. awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella institu source attribution model ory Food D3 Variable definitions for TT SAM without wo underreporting RequiredFields FoodImportExport Country_From Country_To Food_Source Amount Prevalence Country Genus Species Food_Source Units_Tested Units_Positive HumanCase Country Genus Species Incidence Travel Outbreak DimensionMinimums FoodSources 3 ReportingCountries 5 SourceCountries 5 Subtypes 5 Distributions a ReportingCountryCount FoodSourceCount q SeroVarCount Data Incidence n ReportingCountryCount SeroVarCount Travel yt ReportingCountryCount SeroVarCount Outbreak outb ReportingCountryCount SeroV arCount OutbreakSum outbreak ReportingCountryCount ImportExport m ReportingCountryCount SourceCountryCount FoodSourceCount Prevalence p SourceCountryCount FoodSourceCount SeroV arCount Distribution_a Grouping FoodSourceCount Default_Range 0 100 Visible True FloatingPoint False UseMaxRangeArray True Transpose True Distribution
5. caused by e g values of zero in distributions or errors in the WinBUGS code written by the creator of the model If the rare occasion arises press the Recalculate Initial Values button Supporting publications 2012 EN 318 25 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella Ins U source attribution model DTU Food j 8 Monitoring of convergence and goodness of fit If no problems occurred during the execution you can now press Kernel Density and a window displaying the Kernel density plots for the food source dependent factors a and subtype dependent factors q will be shown All the plot names are listed to the left and by clicking the name the plots for that particular variable name is displayed in the right side There will be one curve for each
6. BE Belgium I HU Hungary Iv SE Sweden BG Bulgaria IE Ireland x SI Slovenia l l CH Switzerland IT Italy iv SK Slovakia iv CY Cyprus LT Lithuania Iv UK United Kingdom CZ Czech Republic LU Luxembourg I DE Germany iv LV Latvia l DK Denmark MT Malta EE Estonia NL Netherlands ES Spain Cs FI Finland PL Poland FR France PT Portugal Check All Uncheck All Depending on the data input or model choice it might be the case that it can only handle one country at a time Hence only one country can be selected When this is the case the window title as well as the title in the sidebar will reflect this so instead of Select Reporting Countries the title will be Select Reporting Country The number of countries presented will also depend on the data imported If only 10 countries are present in the data files then only 10 countries will be listed By default EFSA_SAM automatically selects all countries present in the dataset but it is possible to deselect countries if needed 4 2 Selecting the food sources A quite self explaining window Supporting publications 2012 EN 318 16 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and t
7. For the Single member state model Travel is replaced by Travel_Yes Travel_No and Travel_Unknown Also Country is replaced by Year 4 digits Supporting publications 2012 EN 318 11 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model Required columns for food production and import export data A Country_From Either two character short code or the full country name A Country_To Either two character short code or the full country name A Food_Source Text Broilers Pigs Turkeys Layers more food sources can be defined by the user A Tonnes Integer number no decimals amount in tonnes of food source When Country_From and Country_To is the same the required value is the amount produced and available for consum
8. Graphical Statistics 7 434 455 Hald T Vose D Wegener HC and Koupeev T 2004 A Bayesian approach to quantify the contribution of animal food sources to human salmonellosis Risk Analysis 24 255 269 Hald T Pires SM and de Knegt L 2012 Development of a Salmonella source attribution model for evaluating targets in the turkey meat production Report to contract CT EFSA BIOHAZ 2011 02 Supporting publications 2012 259 36 pp Havelaar AH Ivarsson S L fdahl M and Nauta MJ in press Estimating the true incidence of campylobacteriosis and salmonellosis in the EU 2009 Epidemiology and Infection published online 13 April 2012 DOI http dx doi org 10 1017 S09502688 12000568 Little CL Pires SM Gillespie IA Grant K and Nichols GL 2010 Source attribution of Listeria monocytogenes in England and Wales Adaptation of the Hald Salmonella Source Attribution Model Foodborne Pathogens and Disease 7 7 749 756 Miillner P Jones G Noble A Spencer SEF Hathaway S French NP 2009 Source attribution of food borne zoonoses in New Zealand a modified Hald model Risk Analysis 29 7 970 984 Pires SM and Hald T 2010 Assessing the differences in public health impact of Salmonella subtypes using a Bayesian microbial subtyping approach for source attribution Foodborne Pathogens and Disease 7 143 151 Pires SM Evers EG van Pelt W Ayers T Scallan E Angulo FJ Havelaar A and Hald T 2009 Attributing the human disease burden of f
9. The EFSA_SAM model is set up in a Bayesian framework and estimates the number of human sporadic and domestic cases attributed to each source per country A assuming that the observed number of sporadic cases per subtype per country 0 is Poisson distributed Poisson 04i gt Aci and 1 Nekji Pui Mey acj q where cgiis the expected number of human cases per subtype i and source j reported in country c and caused by food produced in country k px is the prevalence of subtype i in source j in country k Megis the amount of source j available for consumption in country c produced in country k a j is the source dependent factor for source j in country c and q is the subtype dependent factor for subtype i When c is equal to k the food originates from the country in which the case is reported Specification of the priors for a and q is described in more details in section 2 5 For the single MS model time period most likely year is replacing MS as a third dimension giving the following equations Poisson 0 gt A and 2 Mii Pij a M 5 ay q where t represents the time period Supporting publications 2012 EN 318 10 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure T
10. The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food s i l Development of a user friendly interface version of the Salmonella National Food Institute source attribution model User friendly Interface for EFSA Source Attribution Modeling EFSA_ SAM Manual version 1 0 Software Version 1 0 20120623 Supporting publications 2012 EN 318 1 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella N I source attribution model TABLE OF CONTENTS Table OF contents sisirin ra aE a e E E a E EE EiS 2 Mntr od CHOM scier snai E E E EE EE E E Eis 3 K stalan snr eee e aea EE E Ee A EE E a EEE EE E 4 1
11. ea aea GEEA E E AEE E SATENE EEES 4 Background as provided by ERSA c lt csceccscesscsstesessncpoceeseslocsegtaaevesessupessiensdepsndfuntusungebecsbbeavescasedseess 5 Terms of reference as provided by EFSA 0 eee ees essecsseceseceseceseceseeeseeeeeeseeeecaeeeaeecaeecaaecaaecaecnaeenaeenaeenes 6 ODIJCCUHVES sccvecendaedsn casseuelusceessceesceustca pelsdchcswussan Gotisbecessvia td ey susectwevisen coy lasotuaeel e a e i a 7 Matetials and Method sisien e E ea gedaan caebece devant nee T E eenits 7 1 Principle of the Bayesian subtyping approach for source attribution modelling eee ences 7 2 Development of a user friendly interface for the Source Attribution Model EFSA_SAM 9 2 1 Model Cimensions lt ices cates saucchtunadiar2vsacd execs sgccduesupestas encod leans R ENEE EEEREN EEEE 9 2 2 The mathematics of the EFSA_SAM mocdel ccc cccccccccccccececececececececececececececeeueeeseueuss 10 2 3 Model software for the interface eee esseesseceseceseceseceeeeseeeseeeeneeeseecsaeeaaecsaecaeenaeesaeenes 11 2 4 Data input to the EFSA_SAM model 00 000 ceeceeceeeeeeeeeeeeeeesceeaeecaeceaeceaecnaeesseeeeeeeeeeeeneeeaes 11 2 4 1 Reported human cases of Salmonella or other pathogens escesesseceeeceeeeeeeeeeeeeenee 11 2 4 1 1 Genus species and subtype specification 0 eee eseessecseceeceseceeeeseceseeeseeeeeeeeaee 12 2A 2 Underreporting factors cerien sees s Ladeivstasdewesssbcabouiss heey i a eee i e
12. ear Genus Species SeroType PhageType GenoType Food_Source Units_Tested Units_Positive HumanCase Year Genus Species SeroType PhageT ype GenoType Incidence Travel_Yes Travel_No Travel_Unknown Outb reak DimensionMinimums FoodSources 3 Subtypes 5 Years 2 Distributions a YearCount FoodSourceCount q SeroVarCount ptrav YearCount SeroVarCount Data Incidence n YearCount SeroVarCount Travel_Yes yt YearCount SeroVarCount Travel_No nt YearCount SeroVarCount Travel_Unknown pt YearCount SeroVarCount Outbreak outb YearCount SeroVarCount ImportExport m YearCount FoodSourceCount Prevalence p YearCount FoodSourceCount SeroVarCount Distribution_a Grouping FoodSourceCount Default_Range 0 0 01 Visible True FloatingPoint True UseMaxRangeArray True Transpose False Distribution_q Grouping Default_Range 0 0 4 NA_Elements 1 SeroVarCount Visible True FloatingPoint True UseMaxRangeArray True Transpose False Distribution_ptrav Grouping YearCount Default_Range 0 0 2 Visible True FloatingPoint True UseMaxRangeArray False Transpose False KernelDensityPlot a Coda q Coda Winbugs_Defaults BurnInIterations 10000 RunIterations 30000 Chains 3 Output_Variables Supporting publications 2012 EN 318 71 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European
13. either individual subtypes or for groups of subtypes You can make any number of scenarios Scenarios Se li i N lj i Add Delete Q P Coxe In order to create a new scenario press add and a dialog box will ask you to give the scenario a name Add Scenario Scenario Name Enteritidis 1 pct Typhimurium 1 pct ox cac It is important to enter a meaningful name here as this name will be used for identifying the scenarios afterwards Press Ok Supporting publications 2012 EN 318 20 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model A new Scenario Settings window appears K Scenario Settings
14. ins the number of iterations and the number of Markov chains the model is ready for execution in WinBUGS EFSA_SAM automatically sends the code generated on the basis of the users selections to the WinBUGS environment Here a Markov Chain Monte Carlo MCMC simulation is applied to arrive at the posterior distributions for a qi and u and consequently the overall results of the model Supporting publications 2012 EN 318 15 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella Institu source attribution model DTU Food j 2 6 Model check Monitoring of convergence and goodness of fit A first step to check how the model is performing is to examine the Kernel density plot of the posterior distribution of acj qi and uc These should depict distributions that at the left
15. or revise the underreporting factors for human cases if relevant for the model selected and add What If scenarios 5 1 Distribution settings The number of distributions presents depends of the model selected In the Distribution Settings window for the priors the ranges for each distribution grouped or ungrouped can be set independently In the example below you can see that distribution a is a two dimensional array over Reporting Countries and Food Sources You can also see that the dimension is grouped by Food Source Since four food sources have been chosen earlier four lines appear in the window one for each food source F Distribution Settings _ D Distribution Names _ Distribution Settings a ReportingCountryCount FoodSou q SeroVarCount Number of elements 96 24 u ReportingCountryCount z oe Generated True E Grouped By FoodSourceCount Rangel Broilers 0 100 Range2 Layers 0 100 a Range3 Pigs 0 100 Range4 Turkeys 0 100 i i Initial Values i 86 4 81 4 11 40 15 31 84 35 35 61 93 85 60 21 2 4 95 53 84 25 30 34 68 94 1 32 53 40 24 51 30 8 S5 29 32 0 4 7 12 3 32 DBD B M BH 3B 4 9 84 21 53 89 93 6 78 22 94 10 78 33 39 23 57 3 Supporting publications 2012 EN 318 18 The present document has been produced and adopted by the bodies identified above as author
16. published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food B DATA TEMPLATES FOR THE EFSA_SAM MODEL Development of a user friendly interface version of the Salmonella source attribution model Table B1 Example of data template for reported cases of human salmonellosis Year Country Genus Species Serovar Total_cases Travel_cases Domestic_cases Unknown_travel Outbreak_cases 2010 Ms1 Salmonella S enterica AGONA 37 9 21 7 0 2010 MS1 Salmonella S enterica ANATUM 13 4 7 2 0 2010 MS1 Salmonella S enterica BOVISMORBIFICANS 14 0 10 4 0 2010 MS1 Salmonella S enterica BRAENDERUP 35 17 13 5 0 2010 MS1 Salmonella S enterica BRANDENBURG 9 0 7 2 0 2010 MS1 Salmonella S enterica BREDENEY 8 0 6 2 0 2010 MS1 Salmonella S enterica DERBY 10 1 7 2 0 2010 MS1 Salmonella S enterica ENTERITIDIS 6102 673 4072 1357 217 2010 MS1 Salmonella S enterica HADAR 62 22 30 10 0 2010 MS1 Salmonella S enterica HEIDELBERG 16 2 10 4 0 2010 MS1 Salmonella S enterica INFANTIS 107 5 76 26 0 2010 MS1 Salmonella S enterica KENTUCKY 22 11 8 3 0 2010 MS1 Salmonella S enterica KOTTBUS 17 5 9 3 0 2010 MS1 Salmonella S enterica LIVINGSTONE 10 0 7 3 0 etc etc etc etc etc et
17. reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors rl CaAad lmetitiite ta PF UUU INSTITUTE DTU Food National f Development of a user friendly interface version of the Salmonella source attribution model D4 WinBUGS code for TT SAM without wo underreporting Model SALMONELLA including others Estimating the number of sporadic and domestically acquired cases excluding travel and outbreak cases for c in 1 ReportingCountryCount for i in 1 SeroVarCount obl c i lt outb c i 1 ob2 c i lt max ob1 c i 0 o c i lt n c i ob2 c i spdo c i lt o c i yt c i Excl others Attributing cases to sources for c in 1 ReportingCountryCount for i in 1 SeroVarCount 1 o c i dpois lambdaexp c i lambdaexp c i lt lambda c i yt c i lambda c i lt sum lambdacji c 1 FoodSourceCount i uk c i lt o c i lambdaexp c i for j in 1 FoodSourceCount for k in 1 SourceCountryCount lambdackji c k j i lt p k j i m c k j a c j q i truecasesckji c k j i lt lambdackji c k j 1 lambdacji c j i lt sum lambdackji c 1 SourceCountryCount j i truecasescji c j i lt sum truecasesckji c 1 SourceCountryCount j i for c in 1 ReportingCountryCount for i in 1 SeroVarCount 1 for j in 1 FoodSourceCoun
18. set as needed each with a different target prevalence It is however only possible to set target prevalence values for one food source at the time in each scenario In the window to the left a list of affected subtypes will appear telling the actual prevalence P and the adjusted prevalence A Supporting publications 2012 EN 318 21 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food i Development of a user friendly interface version of the Salmonella National Food Institute source attribution model 6 2 Groups of subtypes In the table to the right you will see the subtypes present in the analysis An editbox above the table enables you to enter the target prevalence for a group of subtype the subtypes selected using the check boxes Oee O M i UM Change scenario target prevalence 1 00 use Subtype V EN
19. source attribution model Tine Hald and Jan Lund National Food Institute Technical University of Denmark M rkh j Denmark gt NextPhase IT Jersie Denmark ABSTRACT The objective of the work described in this report was to develop a flexible and user friendly interface for attributing human cases of food borne pathogens to the responsible food animal reservoirs and or food sources The interface is based on two existing Salmonella source attribution models developed in previous EFSA contracts The developed interface is called the EFSA Source Attribution Model EFSA_SAM and the programming language used is Embarcaderos Delphi XE2 Enterprise Based on the user s imported data and model selections the interface generates a WinBUGS code that is executed in WinBUGS and the resulting data are then imported from WinBUGS to the interface software for tabulation and graphical display This approach ensures consistency in both model and data setup eliminating the need for user knowledge of WinBUGS syntax EFSA_SAM requires data by country on reported number of human cases by subtypes food source prevalences by subtypes and food production and trade Users can specify which countries food sources and subtypes e g Salmonella serovars to include in the model The EFSA_SAM also includes the possibility to run different scenario analyses where the user can explore the effect on human cases by changing the prevalence of specific subtypes i
20. staff Winy Messens Luis Vivas Alegre BIOHAZ Unit and Jos Cortinas Abrahantes SAS Unit for the support provided to this external scientific report submitted to EFSA Any enquiries related to this output should be addressed to biohaz efsa europa eu The EFSA_SAM model is available upon request to biohaz efsa europa eu Suggested citation Tine Hald and Jan Lund Development of a user friendly interface version of the Salmonella source attribution model Supporting Publications 2012 EN 318 77 pp Available online www efsa europa eu publications European Food Safety Authority 2012 Development of a user friendly interface version of the Salmonella source attribution model DTU Food SUMMARY EFSA has been working on a series of Scientific Opinions originating from a mandate received by the European Commission EC in July 2008 on the review of Salmonella targets in poultry primary production For evaluating targets in the broiler and turkey production specific Salmonella source attribution models have been developed by external contractors Both models were based on the Hald model and use a Bayesian approach employing microbial subtyping data in both cases Salmonella serovar data These types of source attribution models allow for the identification of the most important animal reservoirs of the zoonotic agent assisting risk managers to prioritize interventions and focus control strategies at the animal production level T
21. that these models require specialist knowledge both in statistics and programming However with the development of EFSA_SAM a flexible user friendly interface for attributing human cases of food borne pathogens to the responsible food animal reservoirs and or food sources is now available which will hopefully make the approach accessible and useful for a broader audience It is emphasised though that the interpretation and validation of the model results require some statistical knowledge but with this technical report and the user manual at hand the users should be able to apply and run the model as well as making an appropriate evaluation the results It is also stressed that like for all models the results are never more valid than the data that is applied Data validation is therefore an important part of any source attribution exercise Inclusion of all major animal reservoir sources for human infections and the availability of representative data on prevalence and subtype distribution from the same reservoirs are therefore a requirement Likewise incidence data and subtype distribution from human cases is needed Finally the usefulness of the model has also so far only truly been evaluated for Salmonella Application of the model for other pathogens therefore needs further research and evaluation before the results can be used for risk management decision support Supporting publications 2012 EN 318 19 The present document
22. the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model truecasesscenariocji c j i lt sum truecasesscenariockji c SourceCountryCount j i summing over k by c for j in 1 FoodSourceCount for k in 1 SourceCountryCount for c in 1 ReportingCountryCount truecasesorigin c k j lt sum truecasesckji c k j 1 SeroVarCount 1 truecasesoriginscenario c k j lt sum truecasesscenariockji c k j 1 SeroVarCount 1 OUTPUT serovar per source for i in 1 SeroVarCount 1 for j in 1 FoodSourceCount serotypel j i lt sum truecasescji 1 ReportingCountryCount j i serotypescenario j i lt sum truecasesscenariocji 1 ReportingCountryCount j i for c in 1 ReportingCountryCount for i in 1 SeroVarCount 1 truecasescil c i lt sum truecasescji c 1 FoodSourceCount i truecasesscenarioci c i lt sum truecasesscenario
23. types Next thing to set up is the model settings such as revising the underreporting factors if these are applied see select model below and then setting up different scenarios to analyse along with the baseline analysis see scenarios below The final step is to generate the WinBUGS code and execute the analysis in WinBUGS Afterwards the WinBUGS results are imported into the user friendly interface The number of items present in the sidebar will change depending on the model selected It is fully configurable when setting up the models in the advanced part of EFSA_SAM You can read more about this later in the manual Supporting publications 2012 EN 318 6 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella d lI E q i j gt 1 1 National Food I
24. used for FoodImportExport e Year e Country_From e Country_To e Food_Source e Amount Following field names can be used for Prevalence e Year Country Genus Species Food_Source Units_Tested Units_Positive Following field names can be used for HumanCase e Year e Country Supporting publications 2012 EN 318 31 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella I source attribution model l DTU Food Genus Species Incidence Travel Travel_Yes Travel_No Travel_Unknown Outbreak For the single member state model the fields year travel_yes travel_no and travel_unknown is used For multiple member state model the fields country country_from country_to and Travel are used Dimension Ranges In order for the WinBUGS code to determine the rang
25. 05907 1001 200 Jambdacji 1 74 0 01282 0 0128 2 504E 4 2 676E 4 0 008882 0 04717 1001 2000 Jambdacji 1 1 8 0 002049 0 002048 4 186E 5 4 162E 5 0 001388 0 007529 1001 200 Jambdacji 1 1 9 0 03493 0 03492 6 923E 4 7 021E 4 0 02381 0 1292 1001 2000 Jambdacji 1 1 10 8 672E 4 8 714E 4 1 705E 5 1 836E 5 6 001E 4 0 003221 1001 200 Jambdacji 1 1 11 0 8097 0 8057 0 01593 0 01661 0 5533 2 961 1001 2000 Jambdacji 1 1 12 1 087 1 082 0 02152 0 02204 0 7472 4 061 1001 2000 Jambdacji 1 1 13 4 163E 4 4 143E 4 8 261E 6 8 277E 6 2 844E 4 0 001543 1001 200 Jambdacji 1 1 14 0 009101 0 009072 1 811E 4 1 852E 4 0 006248 0 03403 1001 2000 Jambdacji 1 1 15 2 038E 6 2 036E 6 4 134E 8 4 243E 8 1 381 6 7 438E 6 1001 200 Jambdacji 1 1 16 0 01072 0 01066 2 102E 4 2 185 4 0 007401 0 03973 1001 2000 Jambdacji 1 1 17 0 2104 0 2098 0 004195 0 004451 0 1419 0 7635 1001 2000 Jambdacji 1 1 18 0 006695 0 006679 1 333 4 1 36E 4 0 004589 0 0246 1001 2000 Jambdacji El 719 0 0 0 0 2 236E 12 0 0 0 0 0 0 1001 2000 lambdac i 1 1 20 9 477E 4 9 501E 4 1 932E 5 1 935E 5 6 45 4 0 00352 1001 2000 Jambdacji i 0 5572 0 5562 _ 0 01138 0 01136 0 3818 2 05 1001 2000 lambdac i 1 0 04565 0 04557 8 935E 4 9 384E 4 0 03125 0 1664 1001 2000 Jambdacji 1 587 9 36 86 2 283 517 0 587 8 660 2 1001 2000 Jambdacji i 1 713 0 1191 0 005262 1 484 1 712 1 953 1001 2000 Tambdacti 1 1 713 0 1396 0 004006 1 46 1 708 2 016 1001 2000 Jambdacji 1 1 101 0 07893 0 002059 0 9519 1 099 1 262 1001 2000 Jam
26. 1 WinBUGS instalati Ohsisssseisisussessessisitsnsssasosssi satioun asena Enni Sar EE ESEN SEEEN EPESOES EKES 4 1 2 EFSA_SAM installation esesesesesesesesesererererererererererererererererererererererererererererererererererererereeere 5 Qe TheuserinterfacE isss i a a a a SE a a a NSE 5 2 1 The workflow SIGEDAL sisser E nacenevenacccave vagedapnoace casts nacevasevacedasedagecesasaeedevo ons 6 2 2 The center information WindoW ccccccccccccccccecececececececececececececececeescececesecucecseuseeseueseueueueass 7 3 Creating a new analysis ss ceccedsiesthies auvetedeensseseaetaeseledheatboeduvetsdebvaccacedustalesnstedaneubeculonactensetn EEEE 7 3 1 Data TM ports s scdicecsiusasdaesunesatsssoeciees a ERAEN EEE ONERE ERE EEE ENEO IERE EEEE KEE EEOSE 9 3 1 1 Data format and validation eeeeeeeeeeeessseseseseseseseseserererererererererererrrerererererererererererererererere 10 31 2 Importing datan siiiiceieiciecedi dives taitvesteiegend tedetisd i a a a 10 3 1 3 Column name substitution eessesssesesesesesesesesesererererererererererererererirererererererererererererererrrere 13 3 1 4 Dimension data value substitution essssseessessssesesssesesereserererererererererirererrrererererrrerererere 13 3 1 5 6 Pilling inthe planks s ereroaesosieronss ieiet irere onie aiioe eiea Ei eiei e Ti 4 Selection the model dimensions 4 1 Selecting reporting and source Countries oe eee eee eeseceeceeceseceeceeeceseceeeeeseeeeneee
27. 273 0 1674 1001 2000 lambdac i 1 3 6 0 0 0 0 2 236E 12 0 0 0 0 0 0 1001 2000 Jambdacji TEHA 20 18 2 512 0 06531 15 69 20 14 25 44 1001 2000 Tambdacji 1 3 8 1 329 0 1706 0 006183 1 02 1 321 1 684 1001 2000 4 n As you can see in the window above the format of the output that WinBUGS makes is not very easy to read or interpret Therefore it will be re arranged into a user friendly tabular format Supporting publications 2012 EN 318 28 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model source Baseline sourcescenario Scenario Mean Std dev 2 5 Median 97 5 source AT LAYERS source AT PIGS source AT TURKEYS source BE BROILERS source BE LAYERS source BE PIGS source BE TURKEYS
28. 5 Spain 5 1 0 8 219 1 214 2 Sweden 1 0 8 0 5 0 5 The Netherlands 3 0 8 26 8 26 2 United Kingdom 1 7 0 8 7 5 73 2 4 3 Prevalence data on Salmonella or other pathogens For each specified food source included in the EKFSA_SAM model the number of units tested i e sample size and the number of positive units per subtype and MS or year for single MS model are required as input see data templates in Appendix B2 and C2 The EFSA_SAM will based on these values calculate a subtype specific prevalences in each food source 2 4 4 Production and trade data Ideally the EFSA_SAM model should employ consumption data of the specified food sources However national consumption data do not generally include information of the origin of the food i e the country in which the food was produced which is considered to be an essential part of the model because of the extensive trade of foods between MSs Therefore an approximation is recommended where the amount available for consumption produced in a MS is estimated as Amount available for consumption production export The amount of food imported to one MS from another MS should be estimated in order to consider trade between MSs In the EFSA_SAM model the prevalences of the specific subtypes are weighted by production and import figures as explained in section 2 2 Supporting publications 2012 EN 318 13 The present document has been produced and adopted by the bodies i
29. AM model is in theory able to address other food borne pathogens than Salmonella by specifying other pathogen subtypes in the Genus and Species tabs see user manual in Appendix A It is emphasized however that there are some biological requirements e g clonal dissemination and data needs e g sample representativeness of the included sources that have to be fulfilled in order for the model to produce meaningful results for other pathogens Pires et al 2009 Other pathogen models based on the pathogen subtyping principles initially described for Salmonella by Hald et al 2004 include Campylobacter source attribution in New Zealand using MLST typing M llner et al 2009 and Listeria monocytogenes source attribution in UK using serotyping and AFLP typing Little et al 2010 For other pathogens the approach still needs to be validated through further research So even though the EFSA_SAM model will not require in depth knowledge of the syntax and technical specifications of the model code it does require good understanding of the model principles and model limitations in order to interpret the results correctly Users of the interface are therefore recommended to read this report before starting using the interface to become familiar with the model principles and the mathematics behind which is required in order to interpret the model results and assess the validity of the model 2 2 The mathematics of the EFSA_SAM model
30. AT data extracted 3 January 2010 http faostat fao org site 569 DesktopDefault aspx PageID 569 ancor http epp eurostat ec europa eu newxtweb Supporting publications 2012 EN 318 14 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model l DTU Food Remember that the a s are based on two dimensions meaning that the number of priors specified for a is the number of food sources multiplied with the number of countries or years for single MS model included in the model For the subtype dependent factor the g value for first subtype specified e g Enteritidis is by default set to 1 meaning that all the other subtypes are estimated relative to this Since the priors are simply multiplication factors it makes sense to select a common subtype as an a
31. Downloaded from orbit dtu dk on Dec 18 2015 Technical University of Denmark DIU Te Development of a user friendly interface version of the Salmonella source attribution model Hald Tine Lund Jan Publication date 2012 Document Version Publisher final version usually the publisher pdf Link to publication Citation APA Hald T amp Lund J 2012 Development of a user friendly interface version of the Salmonella source attribution model S borg The National Food Institute Technical University of Denmark General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights e Users may download and print one copy of any publication from the public portal for the purpose of private study or research e You may not further distribute the material or use it for any profit making activity or commercial gain e You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details and we will remove access to the work immediately and investigate your claim DTU Food Supporting Publications 2012 EN 318 EXTERNAL SCIENTIFIC REPORT Development of a user friendly interface version of the Salmonella
32. EU MS q Subtype related factor one per subtype The q value for the first subtype in the demo data S Enteritidis is by default set to 1 and the others q values are estimated relative to this one Baseline scenario EU MS u Underreporting factor one per MS Supporting publications 2012 EN 318 18 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model DTU Food Table 3 Description of the variables that the user is able to select as model outputs in the single MS model Variable name Description source Number of estimated human cases per source per year travel Number of estimated travel related human cases per year outbreak Number of estimated outbreak related human cases per year unknown Number of estimated human cases with unknown source
33. Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model source YearCount FoodSourceCount Number of estimated cases per source per year MS MS Baseline travel YearCount Number of estimated travel related cases per year MS Baseline unknown YearCount Number of estimated cases with unknown source per year MS Baseline total YearCount Number of estimated total cases per year MS Baseline percsource YearCount FoodSourceCount Percentage of estimated cases per source per year MS Baseline lambdatji YearCount FoodSourceCount SeroVarCount Number of estimated sporadic and domestic cases per serovar source and year MS Baseline lambdaexp YearCount SeroVarCount Number of estimated cases per serovar and year MS Baseline a yearcount Food source related factor one per source MS Baseline q SerovarCount Serovar related factor one per serovar The q value for the first serovar S Enteritidis is by default set to 1 and the others q valu
34. Jan_Novi2 Description Settings Variable Definitions Code Data Initial Values Script Items Available When Setting Up An Analysis Data Import Analysis Dimensions Model Settings Run Analysis J Reported Cases M Single Prevalence Year I Dimension Distributions J Generate Winbugs Code Prevalence In Food Sources 7 Multiple Prevalence Years M Under Reporting Factor IJI Run Analysis E Food Import Export E Single Human Case Year M Scenarios F Multiple Human Case Years E Single Reporting Country Multiple Reporting Countries E Single Source Country F Multiple Source Countries Z Food Sources Bacteria Types 10 2 Variable definitions The variable definitions tab is where the interface between EFSA_SAM and WinBUGS is defined These settings define Distributions their grouping NA items and their default range values This also includes which variables that are going to have Kernel density plots default values for the Baseline and the Scenarios scripts iterations and chains etc Supporting publications 2012 EN 318 30 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency p
35. R THE TT SAM AND SINGLE MEMBER STATE MODEL D1 Variable definitions for TT SAM with underreporting RequiredFields FoodImportExport Country_From Country_To Food_Source Amount Prevalence Country Genus Species Food_Source Units_Tested Units_Positive HumanCase Country Genus Species Incidence Travel Outbreak DimensionMinimums FoodSources 3 ReportingCountries 5 SourceCountries 5 Subtypes 5 Distributions a ReportingCountryCount FoodSourceCount q SeroVarCount u ReportingCountryCount Data Incidence n ReportingCountryCount SeroVarCount Travel yt ReportingCountryCount SeroVarCount Outbreak outb ReportingCountryCount SeroV arCount OutbreakSum outbreak ReportingCountryCount ImportExport m ReportingCountryCount SourceCountryCount FoodSourceCount Prevalence p SourceCountryCount FoodSourceCount SeroVarCount UnderReporting u ReportingCountryCount Distribution_a Grouping FoodSourceCount Default_Range 0 100 Visible True FloatingPoint False UseMaxRangeArray True Transpose True Distribution_q Grouping Default_Range 0 100 NA_Elements 1 SeroVarCount Visible True FloatingPoint False UseMaxRangeArray True Transpose False Distribution_u Grouping Default_Range 2 5 Visible True FloatingPoint True UseMaxRangeArray False Transpose False KernelDensityPlot a Coda q Coda WinBUGS_Defaults BurnInIterations 10000 Supporting publications 2012 EN 318 60 The present document has been produced and adopt
36. S BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 41 SSSSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 66 SSSSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 8 SSSSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium NT SSSSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium U312 SSSSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 193 BEEF 2267 0 2009 Salmonella Enterica Agona BEEF 2267 0 2009 Salmonella Enterica Anatum BEEF 2267 0 2009 Salmonella Enterica Brandenburg BEEF 2267 0 2009 Salmonella Enterica Derby BEEF 2267 0 2009 Salmonella Enterica Dublin BEEF 2267 2 etc etc etc etc etc etc etc etc etc Supporting publications 2012 EN 318 58 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food National Food Development of a user friendly interface version of the Salmonella source attribution m
37. SAM with underreporting Model SALMONELLA including others Estimating the number of sporadic and domestically acquired cases excluding travel and outbreak cases for c in 1 ReportingCountryCount for i in 1 SeroVarCount ob1 c i lt outb c i 1 ob2 c i lt max ob1 c i 0 o c i lt n c i ob2 c i spdo c i lt o c i yt c i Excl others Attributing cases to sources for c in 1 ReportingCountryCount for i in 1 SeroVarCount 1 o c i dpois lambdaexp c i lambdaexp c i lt lambda c i yt c i lambda c i lt sum lambdacji c 1 FoodSourceCount 1 uk c i lt o c i lambdaexp c i for j in 1 FoodSourceCount for k in 1 SourceCountryCount lambdackji c k j i lt p k j i m c k j a c j q i truecasesckji c k j i lt lambdackji c k j i uf c lambdacji c j i lt sum lambdackji c 1 SourceCountryCount j i truecasescji c j i lt sum truecasesckji c 1 SourceCountryCount j 1 for c in 1 ReportingCountryCount for i in 1 SeroVarCount 1 i for j in 1 FoodSourceCount for k in 1 SourceCountryCount i casescenariockji c k j i lt p_scen k j i m c k j a c j q i truecasesscenariockji c k j i lt casescenariockji c k j i uf c Supporting publications 2012 EN 318 62 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in
38. SS 6 1 0 5 1 2009 Salmonella Enterica Typhimurium DT 193 RSRRSSSSS 41 5 17 20 4 2009 Salmonella Enterica Typhimurium NT RSRSSSSSS 1 0 0 1 0 2009 Salmonella Enterica Typhimurium U302 RRRRSSSSS 2 0 1 0 1 2009 Salmonella Enterica Typhimurium DT 12 SSSSSSSSS 10 0 2 8 0 2009 Salmonella Enterica Typhimurium DT 120 SSSSSSSSS 3 0 0 2 1 2009 Salmonella Enterica Typhimurium DT 17 SSSSSSSSS 5 0 0 4 1 2009 Salmonella Enterica Typhimurium DT 193 SSSSSSSSS 1 0 0 1 0 2009 Salmonella Enterica Typhimurium DT 41 SSSSSSSSS 0 0 0 0 0 2009 Salmonella Enterica Typhimurium DT 66 SSSSSSSSS 0 0 0 0 0 2009 Salmonella Enterica Typhimurium DT8 SSSSSSSSS 4 0 0 2 2 2009 Salmonella Enterica Typhimurium NT SSSSSSSSS 21 0 2 14 5 2009 Salmonella Enterica Typhimurium U312 SSSSSSSSS 3 0 0 1 2 2009 Salmonella Enterica Typhimurium DT 193 1 1 1 0 0 2009 Salmonella Enterica Agona 13 0 9 2 2 2009 Salmonella Enterica Anatum 3 0 0 1 2 2009 Salmonella Enterica Brandenburg 3 0 1 1 1 etc etc etc etc etc etc etc etc etc etc a Ifthe data do not distinguish between no to travel travel_no and unknown travel history travel_uk these cases should be reported as travel_no Supporting publications 2012 EN 318 57 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender
39. Salmonella National Food Institute source attribution model If UseMaxRangeArray True then an additional array will be created with the values of Maximum upper range setting of the distribution It may look like this Distribution Ranges for a a_MaxRange c 100 100 100 100 And in the code it may be used like this Food source dependent factor a two dimensional and uniform priors for c in 1 ReportingCountryCount for j in 1 FoodSourceCount a c j dunif 0 a_MaxRange j Kernel Density Plot The KernelDensityPlot section simply tells which distributions density plots are made for KernelDensityPlot a Coda q Coda WinBUGS_Defaults The WinBUGS_Defaults section defines the default values for the number of chains and for the number of Burn In and Run iterations WinBUGS_ Defaults BurnInIterations 10000 Runlterations 30000 Chains 3 Supporting publications 2012 EN 318 34 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards
40. Statistics and Computing 10 325 337 Supporting publications 2012 EN 318 4 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella Institute source attribution model DTU Food Select everything and save it to a text file Then follow the instructions After a successful upgrade you should see Version 1 4 3 in the About WinBUGS window 38 About WinBUGS G WinBUGS l with Y Y DoodleBUGS io Yo Ys Version 1 4 3 A August 6th 2007 l Ys Y BUGS 1990 1996 Medical Research Council MRC UK B U G S WinBUGS 1996 2007 Imperial College and MRC UK C License key download and installation Navigate to http www mrc bsu cam ac uk bugs winbugs WinBUGS14_immortality_key txt Select all text and copy it
41. TERITIDIS AGONA ANATUM BOVISMORBIFICANS Change prevalence of __ BRAENDERUP Individual Subtypes T BRANDENBURG Groups Of Subtype BREDI _ ees 3 DERBY Turkeys TYPHIMURIUM P 0 0240 A 0 0100 E HADAR Turkeys ENTERITIDIS P 0 0400 A 0 0074 j Turkeys TYPHIMURIUM P 0 0140 A 0 0026 l PERRIS Turkeys TYPHIMURIUM P 0 0180 A 0 0100 INFANTIS Turkeys TYPHIMURIUM P 0 0160 A 0 0100 KENTUCKY Turkeys ENTERITIDIS P 0 0120 A 0 0044 i Turkeys TYPHIMURIUM P 0 0150 A 0 0056 __ ees Turkeys ENTERITIDIS P 0 0130 A 0 0100 LIVINGSTONE Turkeys TYPHIMURIUM P 0 0390 A 0 0100 konna Turkeys TYPHIMURIUM P 0 0300 A 0 0100 j Turkeys ENTERITIDIS P 0 0150 A 0 0100 MBANDAKA Turkeys TYPHIMURIUM P 0 0320 A 0 0100 E MONTEVIDEO NEWPORT RISSEN Turkeys In the window to the left a list of affected subtypes will appear telling the actual prevalence P and the adjusted prevalence A Due to minor rounding errors the calculated new grouped prevalence may not for some countries equal exactly the selected target value 0 0001 7 Running the analysis When all settings and choices have been made it is time to run the analysis for the baseline In the tree view to the left you can see the baseline and if you have chosen to create scenarios these will also be present in the view In order to run the baseline you must select the baseline likewise if you want to run a scenario select th
42. The Grouping defines whether groups of initial values are needed In this example distribution a is defined by the dimensions ReportingCountryCount and FoodSourceCount If there are 24 reporting countries and 4 food sources there will be 96 elements If no grouping is selected it will be a single dimension array containing 96 elements If like in the example above a grouping is applied this case Grouping FoodSourceCount a will be defined as a two dimensional array of 4 rows by 24 columns If Transpose True then the array is transposed and will now be 24 rows by 4 columns If Floating point is false the data elements will be integer numbers otherwise they will be floating points Supporting publications 2012 EN 318 33 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the
43. UM V DERBY VIRCHOW V HADAR V OTHER V HEIDELBERG V INFANTIS i V KENTUCKY V KOTTBUS i V LIVINGSTONE i Check all Uncheck al Only subtypes present in the imported data either in Food source prevalences or Reported human cases will be shown By default EFSA_SAM automatically selects all subtypes present in the dataset but it is possible to deselect subtypes if needed Supporting publications 2012 EN 318 17 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model It should be noted that models require a minimum number of subtypes to be able to run and provide reliable results around 10 subtypes 5 Model settings These are the settings for the model where the user can check
44. URIUM Broilers 380 4 2009 MS1 Salmonella S enterica ENTERITIDIS Broilers 380 4 2009 MS1 Salmonella S enterica MONTEVIDEO Broilers 380 6 2009 MS1 Salmonella S enterica INFANTIS Broilers 380 8 2010 MS1 Salmonella S enterica TYPHIMURIUM Broilers 450 13 etc etc etc etc etc etc etc etc Supporting publications 2012 EN 318 55 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food National Food Institute Development of a user friendly interface version of the Salmonella source attribution model Table B3 Example of data template for animal food production and trade data Year Country_From Country_To Food Source Tons 2010 AT AT Broilers 239 681 2010 AT BE Broilers 60 2010 AT CY Broilers 0 2010 AT CZ Broilers 1 336 2010 AT DE Broilers 42 965 2010 AT DK Broilers 6 2010 AT EE Broilers 54 2010 AT ES Broilers 23 2010 AT Fl Broil
45. V file format The columns needed in the files depend on the model chosen and must be documented for each model created The List Separator in the files must be Semicolon the Decimal Separator must be Period and the Thousands Separator must be Comma C The first row of the data file must contain the column names also separated by semicolons The sequence of the columns does not matter but all columns must be present and preferably correctly spelled Datasets must be clean and verified before importing them into EFSA_SAM although a number of checks will be performed to validate the data entry and headings i Column Naming 11 Text values for the dimensions Country Subtype Food Source iii Data range checks for e g Year iv Negative values v sums differences of some columns and vi duplicate data on the combination of dimensions In order to ensure proper results from the analysis we need to ensure that the terms used for countries subtype names etc are the same for prevalences human cases and food import export For instance if the term S Enteritidis is used in the prevalences file and the human cases file uses the term Enteritidis the software will never find a match between Prevalence and Human Cases for this particular subtype Fortunately EFSA_SAM makes it easier to ensure consistency as all country names and the most frequently used subtypes Salmonella serovars and phage types and food source names
46. _q Grouping Default_Range 0 100 NA_Elements 1 SeroVarCount Visible True FloatingPoint False UseMaxRangeArray True Transpose False KernelDensityPlot a Coda q Coda WinBUGS_Defaults BurnInIterations 10000 RunIterations 30000 Chains 3 Output_Variables serotype FoodSourceCount SeroVarCount Number of estimated cases in EU per source and serotype EU Baseline serotypescenario FoodSourceCount SeroVarCount Number of estimated cases in EU per source and serotype EU Scenario source ReportingCountryCount FoodSourceCount Number of estimated cases per source and reporting MS MS Baseline sourcescenario ReportingCountryCount FoodSourceCount Number of estimated cases per source and reporting MS MS Scenario source2 SourceCountryCount FoodSourceCount Number of estimated cases per source and MS of origin MS Baseline source2scenario SourceCountryCount FoodSourceCount Number of estimated cases per source and MS of origin MS Scenario truecasescji ReportingCountryCount FoodSourceCount SeroVarCount Number of cases pr MS food source and serotype MS Baseline Supporting publications 2012 EN 318 66 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparen
47. ables etc It is emphasized that even if the software seems simple to use it is required that the users know the data requirements and have some statistical background for running the analyses and know how to identify the pitfalls and evaluate the analysis results It is therefore recommended that users also read the technical report describing the modeling principles and mathematics in more detail before starting to use the software Supporting publications 2012 EN 318 3 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model 1 INSTALLATION The user needs to have both WinBUGS and EFSA_SAM installed on the PC as EFSA_SAM lt call WinBUGS for execution of the model code developed in EFSA_SAM It does not matter whe
48. afety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella NSUtUTE source attribution model DTU Food R TC j s Anag f YG UNII IG WU total c lt travel c unknown c sum source c 1 FoodSourceCount outbreak c totalscenario c lt travel c unknown c sum sourcescenario c 1 FoodSourceCount outbreak c GoodFit c lt o c 1 SeroVarCount 1 lambdaexp c 1 SeroVarCount 1 for evaluating goodness of fit excluding other serovars totalEU lt sum total 1 ReportingCountryCount unknownEU lt sum unknown 1 ReportingCountryCount travelEU lt sum travel 1 ReportingCountryCount unktravEU lt unknownEU travelEU totalEUscenario lt sum totalscenario 1 ReportingCountryCount difftotalEU lt totalEU totalEUscenario for j in 1 FoodSourceCount sourceC j lt sum source 1 ReportingCountryCount j percsourceC j lt sourceC j 100 totalEU sourceCscenario j lt sum sourcescenario 1 ReportingCountryCount j percsourceCscenario j lt sourceCscenario j 100 totalEUscenario diffsourceC j lt sourceC j sourceCscenario j diffpercsourceC j lt percsourceC j percsourceCscenario j percdiffsourceC j lt diffsourceC j 100 sourceC j DEFINITION OF PRIORS Fo
49. and or genotypic subtyping methods The principle is to compare the distribution of subtypes in potential sources typically food animals with the subtype distribution in humans and the approach is enabled by the identification of strong associations between some of the dominant subtypes and a specific food animal reservoir providing a heterogeneous distribution of subtypes among the sources Subtypes exclusively or almost exclusively isolated from one source are regarded as indicators for the human health impact of that particular source assuming that all human infections with these subtypes originate only from that source Human infections caused by subtypes found in several reservoirs are then distributed relative to the prevalence of the indicator types The Bayesian model first described by Hald et al 2004 attributes domestically acquired laboratory confirmed human infections caused by different Salmonella subtypes e g serovars phage types antimicrobial resistance profiles as a function of the prevalence of these subtypes in animal and food sources and the amount of each food source consumed However the number of people being infected by a particular subtype in a particular food source supposedly depends on additional factors related to the subtype and food source in question Therefore a multi parameter prior which accounts for the presumed but undefined differences between subtypes and food sources with respect to cause human infection
50. ar country Same thing happens if a serovar country is present in the reported human cases data but not in the food source prevalence data Then zero values will be added for that serovar country for all food sources If Import Export information is missing from to one or more countries these are also replaced with amounts of zero Supporting publications 2012 EN 318 15 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model 4 Selection the model dimensions 4 1 Selecting reporting and source countries The windows for selecting reporting and source countries are the same and are pretty much self explaining Select reporting countries v AT Austria l GR Greece i RO Romania
51. are stored in a database The data from all imported files will be checked against this database and as you will see EFSA_SAM can replace differently spelled items throughout the whole file in a few mouse clicks Also new items e g additional subtypes or food sources can be added to the database for future use Even though this is a very valuable function it is highly recommended to use the same naming conventions throughout the datasets Also if the data file contains columns named differently from the columns required a window will ask the user to substitute the column names If disagreeing data entries are neither substituted nor added as new to the database the data file cannot be imported Note EFSA_SAM can handle both full country names and the country short codes e g Denmark and DK will be handled the same way and does not need substitution 3 1 2 Importing data In order to provide a concrete example we will use the Salmonella Turkey Target model TT SAM as a reference Hald et al 2012 The data columns needed for this model is as follows 7 Hald T Pires SM and de Knegt L 2012 Development of a Salmonella source attribution model for evaluating targets in the turkey meat production Report to contract CT EFSA BIOHAZ 201 1 02 Supporting publications 2012 259 36 pp Supporting publications 2012 EN 318 10 The present document has been produced and adopted by the bodies identified above as author s Th
52. as aR E ERE 12 2 4 3 Prevalence data on Salmonella or other pathogens eeeeseeseseserssressssrrerrsresrrsresserrsss 13 2 4 4 Production and trade data s cccccssscseveetcotesccssscsestsesneusneuscrnesicsnonasccuseoeseceseaseaasovtoosssvenae 13 2D Specification of prior distributions initial values and model execution eee errr 14 2 6 Model check Monitoring of convergence and goodness Of fit 16 2 7 Baseline and scenario analyses cescessceseceseceseeeseeeeeeeeaceeacecsaecsaecsaeceaecnaeeeeeeseeseneseneeeaee 17 2 8 Model output and presentation of the results eeceeeeeseeeeeeeeeeeeseeeaeecaeecsaecaeceaeenseenes 18 2 9 Saving and exporting an analysis ee eee eeseeseeseecssecssecesecsseessecsseeseeesseeeseeseaeeeaeeeaaesaaeenaees 19 Conclusions and RecommMendations cssccssecsseceseceseceseeeseeeseeecaeeeaeeeaeecsaecsaecsaeceaeeeseeeeeseeeeeeeeeneeaee 19 References airesin eo siete gestae one eda h dered es out EE ta dade saeed ot E E E E E anasto 20 Appendice Surenen diiran aan oie eea EEEa EA AROA EESE OaE A ASO e AREE EEE EA A DS 21 A EFSA_SAM manual version 1 0 noseesessssssssssssesssesssesesrsrseseseseserererererererererererrrerererererererererererere 21 B Data templates for the EFSA_SAM model oe eeceeceeseeeeeeeeeeseeeseecaeceaeceseceaeeseesseeeeeeseneeeaee 54 C Data templates for the Single Member State model sseeesseeeeesseeesrsessrsrrssrrsrsstesresreersressesreses 57 D Variable defin
53. at scenario and set the variables In order to avoid using and maintaining too many different models WinBUGS source code the EFSA_SAM is built in a way that it will run both a baseline and a scenario for each run All scenarios use the same WinBUGS code Supporting publications 2012 EN 318 22 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food National Food Institute Development of a user friendly interface version of the Salmonella source attribution model When selecting the baseline the baseline data will be used and the results will reflect the results of the baseline When selecting a scenario both the baseline data and the scenario data with the adjusted prevalences will be used and results from both the baseline and the selected scenario will be outputted CS STE Models 4 Baseline Output Var
54. bdacji 1 1 453 0 1233 0 003576 1 22 1 449 1 709 1001 2000 lambdac i 1 4 724 0 3177 0 01005 4 106 4 717 5 366 1001 2000 Jambdacji He 0 1668 0 01419 8 008E 4 0 1397 0 1664 0 1958 1001 2000 i Jambdacji 1 12 24 0 7395 0 02011 10 85 12 26 13 72 1001 2000 Jambdacji i 0 1297 0 00992 6 624E 4 0 1108 0 1295 0 149 1001 2000 lambdacji i 0 01741 0 001967 1 115E 4 0 01386 0 01733 0 0217 1001 2000 Tambdac i iL 19 89 1 281 0 07055 17 45 19 88 22 47 1001 2000 Jambdacji 1 0 0 0 0 2 236E 12 0 0 0 0 0 0 1001 2000 lambdacji 1 0 002692 2 444E 4 1 181E 5 0 002239 0 00268 0 003203 1001 2000 lambdac i i 1 451 0 1086 0 004033 1 24 1 45 1 678 1001 2000 Jambdacji 1 2 779 0 2012 0 005705 2 397 2 775 3 19 1001 2000 Jambdacji 1 0 06623 0 005397 2 941E 4 0 05626 0 0661 0 07702 1001 2000 Tambdac i fi 1 114 0 079 0 003263 0 9654 1 114 1 27 1001 2000 Jambdacji s c 1 211 0 08035 0 004055 1 052 1 209 1 372 1001 2000 Jambdacji 1 1 045 0 0829 0 001935 0 8852 1 042 1 21 1001 2000 lambdac i 1 0 04673 0 003822 2 375E 4 0 03955 0 04663 0 05449 1001 2000 lambdac i 1 518 6 29 54 0 819 461 9 518 8 577 9 1001 2000 lambdac i 11 22 1 222 0 085 0 004902 1 058 1 222 1 393 1001 2000 Jambdacji 1 3 1 82 84 10 53 0 4124 64 04 82 29 105 0 1001 2000 Jambdacji 1 3 2 6 256 0 8045 0 02661 4 832 6 23 7 915 1001 2000 Jambdacji 1 3 3 0 06851 0 009426 3 Zae 0 0519 0 06799 0 08826 1001 2000 lambdac i 1 3 4 0 0 0 0 2 236E 12 0 0 0 0 1001 2000 Jambdacji 1 3 5 0 1284 0 01817 6 43E 4 0 09465 o 1
55. butions defined and or the other calculations done in the model Typically if the model during an iteration results in a value of zero for some calculated parameter this causes problems for the following estimations and then the model stops running In the EFSA_SAM initial values are generated automatically but users may experience that the WinBUGS log file returns an error massage which will require a that a new set of initial values are generated as explained in the user manual Appendix A When the prior distributions and the associated initial values have been specified the user is requested to specify the number of burn ins iterations and Markov chains that the model should run Burn ins is the number of iterations used for the model to arrive at some steady state after which the model is producing stable results The number of iterations is the set of iterations on which the final results will be calculated Finally the number of chains specifies how many independent runs of the model should be done It is important to note that for each chain a new set of initial values are to be generated and it is emphasised that the initial values should be widely dispersed in the prior distribution model The results of the different chains are used to monitor convergence as explained below Typically for this kind of model a burn in of 5 10 000 iteration 20 30 000 model iterations and 3 to 5 chains are appropriate After having specified the burn
56. c etc etc etc etc Supporting publications 2012 EN 318 54 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors beer ood Ins Development of a user friendly interface version of the Salmonella source attribution model Table B2 Example of data template for reported animal food data Year Country Genus Species Serovar Food Source Units_Tested Units positive 2008 MS1 Salmonella S enterica INFANTIS Broilers 408 1 2008 MS1 Salmonella S enterica KENTUCKY Broilers 408 1 2008 MS1 Salmonella S enterica TYPHIMURIUM Broilers 408 1 2008 MS1 Salmonella S enterica ENTERITIDIS Broilers 408 2 2008 MS1 Salmonella S enterica MONTEVIDEO Broilers 408 4 2009 MS1 Salmonella S enterica HADAR Broilers 380 1 2009 MS1 Salmonella S enterica INFANTIS Broilers 380 1 2009 MS1 Salmonella S enterica KENTUCKY Broilers 380 2 2009 MS1 Salmonella S enterica TYPHIM
57. cally all you have to do is to select the Baseline select the output variables set the number of chains and iterations and then press the Run WinBUGS button Once these values have been set the initial values for the distributions will be created and added to the file that WinBUGS uses for initial values Once the initial values have been created they are saved in the analysis file and will not be changed unless the dimensions or the distribution range are changed Then EFSA_SAM will generate all the files that WinBUGS needs in order to run the analysis and start WinBUGS execution File Tools Edit Attributes Info Model Inference Options Doodle Map Text Window Help __ at 4 7 427E 4 7 426E 4 7 716E 4 1001 2 712E 5 3 403E S 1 985E 4 2 076E 4 al2 3 5 261E 4 5 26E 4 6 066E 4 1001 7 1 798E 5 253 5 1 691E 6 2 681E 6 3 503E 5 When WinBUGS has finished running the analysis it will automatically close and return to EFSA_SAM The run time depends on the complexity of the model the number of iterations and chains and the power of the PC running the analysis but it can easily take several hours using the recommended values and settings above If an error occurs during execution EFSA_SAM will detect this and ask if the user wants to view the WinBUGS log which can provide more information about the problem For more serious errors WinBUGS will stop execution with a so called TRAP window This will happen rarely and can be
58. ces can differ between the subtypes The EFSA_SAM will automatically change the original prevalence to the set target prevalence but only if the original prevalence is greater than the target prevalence In case the original prevalence is lower than the target prevalence the original prevalence is kept 2 Setting a combined target prevalence for a group of subtypes Here the users can select any number of subtypes for which a combined prevalence should be equal or less to a set target prevalence The EFSA_SAM generates a new set of subtype specific prevalences that are proportionally scaled down from the original prevalences in order to result in an overall prevalence equal to or less than the target prevalence In both types of scenarios the original and adjusted prevalences will be presented for the user for comparison Supporting publications 2012 EN 318 17 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached i
59. cji c FoodSourceCount i for j in 1 FoodSourceCount source c j lt sum truecasescji c j 1 SeroVarCount 1 sourcescenario c j lt sum truecasesscenariocji c j 1 SeroVarCount 1 for k in 1 SourceCountryCount for j in 1 FoodSourceCount source2 k j lt sum truecasesorigin 1 ReportingCountryCount k j source2scenario k j lt sum truecasesoriginscenario 1 ReportingCountryCount k j for c in 1 ReportingCountryCount Supporting publications 2012 EN 318 63 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model travel c lt sum yt c uf c unknown 1 c lt spdo c SeroVarCount sum uk c 1 SeroVarCount 1 unknown c lt unknown1 c uf c total c lt trave
60. cy principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model truecasesscenariocji ReportingCountryCount FoodSourceCount SeroVarCount number of cases pr MS food source and serotype MS Scenario travel ReportingCountryCount Number of estimated travel related cases per MS MS Baseline Scenario unknown ReportingCountryCount Number of estimated cases with unknown source per MS MS Baseline Scenario total ReportingCountryCount Number of estimated total cases per MS MS Baseline totalscenario ReportingCountryCount Number of estimated total cases per MS MS Scenario GoodFit ReportingCountryCount Goodness of fit ratio reported number of cases per MS divided by the estimated number of cases per MS per serovar MS Baseline Scenario totalEU Number of estimated total cases in EU EU Baseline unknownEU Number of estimated cases with unknown source in EU EU Baseline Scenario travelEU Number of estimated travel related cases in EU EU Baseline Scenario unktravEU Sum of unknownEU and travel EU EU Baseline Scenario totalEUscenario Number of estimated total cases in EU EU Scenario diff
61. dentified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model DTU Food The data specifications requires that the users of the EFSA_SAM model applies EUROSTAT data or another relevant data source on production export and import to estimate the amount available for consumption in each MS by the MS of origin which are used as input data for the model see data template in Appendix B3 The single MS model also requires data on the food produced see template in Appendix C3 If only domestic food sources are included in the model national statistic on production or consumption may be used Technically the EFSA_SAM model could make the estimations of foods available for consumption in a MS based on raw data input on production import and export However during the development of the BT SAM model it was realised that t
62. e The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella l source attribution model OBJECTIVES The overall objectives of the tasks covered by this report were e To develop a flexible and user friendly interface for attributing human cases of Salmonella to responsible food animal reservoirs and or food sources The interface is based on a Salmonella source attribution model developed for setting target for Salmonella in the turkey production the Turkey Target Source Attribution Model TT SAM which has been described in a previous report Hald et al 2012 The TT SAM model was based on two existing Salmonella source attribution models developed in the WinBUGS software as part of previous EFSA service contracts CFT EFSA BIOHAZ 2010 02 and NP EFSA ZOONOSES 2010 01 e To prepare a user manual explaining the use of the interface software MATERIALS AND METHODS 1 Principle of the Bayesian subtyping approach for source attribution modelling The microbial subtyping approach involves characterisation of isolates of the pathogen by phenotypic
63. e issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model Dl Import Human Cases ncidence Travel Outbea a ee looks Country Genus Species Serotype Incdence Travel Outbreak a Loading and validating C EFSA_SAM dataimport_dem AT Salmonela Enterica AGONA 37 3 o Chisa vaak Data validation ok AT Salmonela Enterica ANATUM 13 0 AT Salmonella Enterica BOVISMORBIFICANS 14 0 0 AT Salmonela Enterica BRAENDERUP 35 17 0 AT Salmonela Enterica BRANDENBURG 9 0 0 AT Salmonela Enterica BREDENEY 8 0 0 AT Salmonela Enterica DERBY 10 1 0 AT Salmonela Enterica ENTERITIDIS 6102 673 217 AT Salmonela Enterica HADAR 62 22 0 AT Salmonela Enterica HEIDELBERG 16 2 0 AT Salmonela Enterica INFANTIS 107 5 0 AT Salmonela Enterica KENTUCKY 22 11 0 AT Salmonela Enterica KOTTBUS 17 5 0 AT Salmonela Enterica LIVINGSTONE 10 0 0 AT Salmonella Enterica LONDON 8 0 0 AT Salmonela Enterica MBANDAKA 12 4 0 AT Salmonela Enterica MONTEVIDEO 21 4 0 AT Salmonela Enterica NEWPORT 38 11 0 AT Salmonela Enterica OTHER 853 126 0 AT Salmonella Enterica RISSEN 12 4 0 AT Salmonela Enterica SAINTPAUL 65 7 10 AT Salmonela Enterica TYPHIMURIUM 958 65 194 AT Salmonela Enterica VIRCHOW 58 18 0 BE Salmonela Enterica AGONA 31 0 0 BE Salmonela Enterica ANATUM 18 0 0 BE Sa
64. e a good overview of the settings selected and how far the user is in the process 3 Creating a new analysis Click New Analysis from the General menu In the window that appears enter a name and description for the analysis select Genus and Species of the bacteria under study from the drop downs It should be noted that for most pathogens other than Salmonella the approach still needs to be validated through further research see also the technical report Then click in the Select Model edit box or press the ellipsis button to the right Supporting publications 2012 EN 318 7 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model New Analysis Analysis Name Test 1 Genus Sp
65. e is Embarcaderos Delphi XE2 Enterprise The interface generates a WinBUGS code based on the user s imported data and model selections The interface exports this code with corresponding data to WinBUGS where the code is executed automatically The model results are then imported from WinBUGS to the interface software for tabulation and graphical display and possible exportation to other softwares for further analysis e g MS Excel This approach ensures consistency in both model and data setup eliminating the need for user knowledge of the WinBUGS syntax Users can import data into the EFSA_SAM from semicolon separated files Required data are i the reported number of human cases per country and subtype including data on the number of travel and outbreak related cases also per country and subtype ii food animal prevalence data per country and subtype including the number of units tested and the number of positive units and iii data on the production and trade of the different food animal sources in the EU Member States The EFSA_SAM also allows for the inclusion of underreporting factors recognizing that the reported number of human cases only reflects a part of the disease burden and the degree of underreporting varies hugely between countries In the interface users can specify which countries food sources and subtypes e g Salmonella serovars to include in the model It is also possible to run an analysis for a single country only but wh
66. e scenario MS travel Number of estimated travel related human cases per MS Baseline scenario MS unknown Number of estimated human cases with unknown source per MS Baseline MS total Number of estimated total human cases per MS Scenario MS totalscen1 Number of estimated total human cases per MS Baseline scenario MS GoodFit Goodness of fit ratio reported number of human cases per MS divided by the estimated number of cases per MS Baseline EU totalEU Number of estimated total human cases in EU Baseline scenario EU unknownEU Number of estimated human cases with unknown source in EU Baseline scenario EU travelEU Number of estimated travel related human cases in EU Baseline scenario EU unktravEU Sum of unknownEU and travel EU Scenario EU totalEUscen1 Number of estimated total human cases in EU Scenario EU difftotalEU totalEU totalEUscen1 Baseline EU sourceC Number of estimated human cases per source in EU Baseline EU percsourceC Percentage of estimated human cases per source in EU Scenario EU sourceCscen1 Number of estimated human cases per source in EU Scenario EU percsourceCscenl Percentage of estimated human cases per source in EU Scenario EU diffsourceC sourceC sourceCscen1 Scenario EU diffpercsourceC percsourceC percsourceCscen1 Scenario EU percdiffsourceC Percentage difference between baseline and scenario diffsourceC 100 source Baseline scenario EU MS a Food source related factor per source and MS Baseline scenario
67. ecies Salmonella w Enterica Select Model TT 5AM wi Underreporting A window with a list of models appears Double click on the model you want to run Model List ono del Nam Model Description Single Memberstate analysis TT SAM w Underreporting Turkey Target Source Attribution Model with under reporting TT SAM wo Underreporting Turkey Target Source Attribution Model without under reporting ten oe There exist three demo models in the software e TT SAM Single MS In the single member state model the user can specify a model including data from a single country only but for more years e TT SAM w underreporting Here the user can specify a model for several countries accounting for underreporting of human cases e TT SAM wo underreporting Here the user can specify a model for several countries but choose not to account for underreporting of human cases After clicking the appropriate model a new window appears Here you need to select which properties of your bacteria subtypes you want to use The properties you select here must be present as data columns in the Reported human cases and Prevalence in food sources data files that you will be importing later in the process Supporting publications 2012 EN 318 8 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the au
68. ed by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model DTU Food RunIterations 30000 Chains 3 Output_Variables serotype FoodSourceCount SeroVarCount Number of estimated cases in EU per source and serotype EU Baseline serotypescenario FoodSourceCount SeroVarCount Number of estimated cases in EU per source and serotype EU Scenario source ReportingCountryCount FoodSourceCount Number of estimated cases per source and reporting MS MS Baseline sourcescenario ReportingCountryCount FoodSourceCount Number of estimated cases per source and reporting MS MS Scenario source2 SourceCountryCount FoodSourceCount Number of estimated cases per source and MS of origin MS Baseline source2scenario SourceCountryCount FoodSourceCount Number of estimated cases per source and MS of origin MS Scenario truecasescj
69. endly interface is Embarcaderos Delphi XE2 Enterprise Delphi is a native Windows development tool and therefore has the advantage of very fast code execution and a very small footprint with no dependencies to external frameworks This is in contrast to for instance languages like C and VB NET where the resource consuming NET framework is needed on the clients computers or the Java language where the Java Runtime Environment also must be installed on the clients computers The independency to external frameworks means that the hardware requirements for applications developed in Delphi are smaller and the risk of framework or runtime version inconsistencies is eliminated Software produced in Delphi will work on both 32 bit and 64 bit Microsoft Windows versions Delphi has a very long history of backwards code compatibility ensuring excellent maintainability of existing code Delphi was originally manufactured by Borland and the first version appeared in 1995 and the latest 15 version known as XE2 was released in September 2011 The developed interface generates a WinBUGS code based on the users selections and export this code with corresponding data to WinBUGS There the model code is executed automatically inside WinBUGS The model results are then imported from WinBUGS to the interface software for tabulation and graphical display This approach ensures consistency in both model and data setup eliminating the need for user knowledge of Wi
70. equired Fields Year Country Genus Species Incidence Travel Outbreak Serotype Phagetype Genotype Analysis a Food Sources 0 Generate WinBugs Code PERRE Sia 0 Distributions 2 2 Not Generated t q Not Generated If an appropriate substitution is not available it is possible to add the new entry into the database This value is now available for all future data imports It is important to note that you must not add names that you can find a substitution for This will obscure the database and you will not be able to trust the data mapping in the future for this particular value which in turn will produce unreliable results So even though EFSA_SAM does its best to ensure that data is mapped correctly it is highly recommended to use the same naming convention for the dimension values countries food sources and serovars throughout all the datasets Supporting publications 2012 EN 318 14 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards th
71. ere several periods typically years of data are included This can provide an indication of the trend over time Required data for this type of model are i the reported number of human cases by subtype including data on the number of travel domestic unknown travel history and outbreak related cases also per subtype ii food animal prevalence data per subtype including the number of units tested and the number of positive Supporting publications 2012 EN 318 2 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model l DTU Food units and iii data on the amount of the included animal foods available for consumption in the country The data imported into EFSA_SAM will be used for a baseline analysis providing estimates on the number of human cases at
72. ers 0 2010 AT FR Broilers 103 2010 AT GR Broilers 102 etc etc etc etc etc Supporting publications 2012 EN 318 56 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella source attribution model C DATA TEMPLATES FOR THE SINGLE MEMBER STATE MODEL Table C1 Example of data template for reported cases of human salmonellosis Phage Resistance Year Genus Species Serotype type profile Incidence Outbreak Travel_yes Travel_no Travel_uk 2009 Salmonella Enterica Enteritidis PT 21 7 0 4 1 2 2009 Salmonella Enterica Enteritidis PT 4 6 0 1 4 1 2009 Salmonella Enterica Enteritidis PT 8 8 0 0 6 2 2009 Salmonella Enterica Typhimurium DT 104 RRRRSSSSS 9 0 0 7 2 2009 Salmonella Enterica Typhimurium DT 120 RSRRSSSSS 20 11 2 16 2 2009 Salmonella Enterica Typhimurium DT 120 RSRSSSS
73. erves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model If you need to review the data the code or script the EFSA_SAM has created click on the WinBUGS code tab As shown in the following window it is possible for advanced users to review and edit the model code The code will not be shown unless the user chooses to show it If the model code is changed these changes will be saved in the final analysis files but it will not be saved into the original model settings in the database If the user wants to save the changes for future use it will be possible to add a changed model to the model repository WinguGs code enteritidis 1 pet Typhimunum O Models 4 Baseline Output Variables WinBUGS Code Enteritidis 1 pct Typhimuric 7 Enteritidis 1 pct Typhimuriu Code data and script passed to WinBUGS Model Code Initial Values Batch Script list a 2 Dimension Ranges SourceCountryCount 24 ReportingCountryCount 24 SeroVarCount 23 FoodSourceCount 4 Distribution Ranges for a a_MaxRange c 100 100 100 100 Distribution Ranges for q q MaxRange c 100 Incidence data Data for n ReportingCountr
74. es are estimated relative to this one MS Baseline Supporting publications 2012 EN 318 72 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella NSttUte source attribution model l DTU Food Nationa Eqad D6 WinBUGS code for Single Member State Model Model for t in 1 YearCount Incl others for i in 1 SeroVarCount o t i lt n t i outb t i ft t i lt nt t i 1 at t i lt yt t i 1 ptrav t i dbeta at t i ft t i xtrav t i lt ptrav t i pt t 1 dt t i lt xtrav t 1 yt t i spdo t i lt o t i dt t i Excl others for i in 1 SeroVarCount 1 n t i dpois lambdaexp t 1 lambdaexp t i lt lambda t i dt t i outb t i lambda t i lt sum lambdatji t 1 FoodSourceCount i uk t i lt o
75. es of the different dimensions a set of Dimension Count variables are available ReportingCountryCount The number of selected reporting countries SourceCountryCount The number of selected source countries FoodSourceCount The number of selected food sources SeroVarCount The number of selected serovars YearCount The number of selected years So if e g three food sources have been selected FoodSourceCount will be set to 3 and inserted into the WinBUGS data file These ranges can be used throughout the Variable Definitions configuration and in the WinBUGS code itself Dimension Minimums In order to have a proper execution of the model it is possible to set a minimum number of values that must be selected before running the model DimensionMinimums FoodSources 3 ReportingCountries 5 SourceCountries 5 Years 2 Subtypes 10 Data The data section defines the names of the variables and their dimensions EFSA_SAM makes output data to WinBUGSS based on these settings Data Incidence n ReportingCountryCount SeroV arCount Travel yt ReportingCountryCount SeroV arCount Outbreak outb ReportingCountryCount SeroVarCount OutbreakSum outbreak ReportingCountryCount ImportExport m ReportingCountryCount SourceCountryCount FoodSourceCount Prevalence p SourceCountryCount FoodSourceCount SeroV arCount Supporting publications 2012 EN 318 32 The present document has been produced and adopted by the bodies identified ab
76. estimating cases value insignificant for i in 2 SeroVarCount 1 qli dunif 0 q_MaxRange 1 Underreporting factor defined as a lognormal distribution replace mu and tau for mean and sdev for c in 1 ReportingCountryCount uf c lt exp u c u c dnorm m u c pr u c pr u c lt pow s u c 2 Supporting publications 2012 EN 318 64 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model End of model Supporting publications 2012 EN 318 65 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s
77. eys was received by EFSA in June 2010 EFSA Q 2010 00899 and this Opinion was published in April 2012 EFSA was asked by the EC to indicate and rank the Salmonella serotypes with public health significance according to Annex III of Regulation EC No 2160 2003 to assess the impact of a reduction of the prevalence of Salmonella in breeding flocks of turkeys on the prevalence of Salmonella in flocks of fattening turkeys and to assess the relative public health impact if a new target for reduction of Salmonella is set in fattening turkeys being 1 or less of flocks remaining positive for all Salmonella serovars with public health significance compared to e the theoretical prevalence at the end of the transitional period 1 or less of flocks remaining positive for Salmonella Enteritidis or Salmonella Typhimurium and e the real prevalence in 2010 to be reported by the Member States MSs The three Opinions addressing Gallus gallus have employed a different approach to address the quantitative aspects of the questions received The Opinion related to Salmonella in broilers was supported by the work of a contractor CT EFSA BIOHAZ 2010 02 who provided quantitative estimates using a broiler target Salmonella source attribution model BT SAM This model was based on the Hald model and uses a Bayesian approach employing microbial subtyping data Hald et al 2004 The Hald model is a Markov Chain Monte Carlo model written in the WinBUGS environment
78. g year and country which can also increase the robustness of the model and consequently improve the parameter estimation for instance by assuming that the q values remain unchanged over at least shorter time periods Pires and Hald 2010 and are independent on country The model calculates the expected number of human cases per subtype i according to the above equation From this a back calculation is made by adding the number of travel and outbreak related cases with known subtype in order to get the expected number of reported cases The observed data i e the reported number of human cases per subtype is then linked with the prior distribution by assuming that the number of cases per subtype is Poisson distributed the likelihood function with a parameter value equal to the expected number of cases This results in posterior estimates for the unknown parameters q and a and consequently for the number of cases per subtype and source Aj which can then be summarised over subtypes to get to the number of cases per source Aj The microbial subtyping approach requires a collection of temporally and spatially related isolates from various sources and humans and is consequently facilitated by an integrated food borne disease surveillance programme focused on the collection of pathogen isolates from the major food animal reservoirs and from humans Pires et al 2009 The data quality and availability are considered the biggest limitati
79. gCountryCount FoodSourceCount Food source related factor per source and MS EU MS Baseline Scenario q SeroVarCount Serovar related factor One per serovar EU MS Baseline Scenario uf ReportingCountryCount Underreporting factor One per MS EU MS Baseline Scenario 10 3 Model code The model code is standard WinBUGS code In order for the code to work along with the settings that EFSA_SAM produces some variables can be used for controlling the loops The variables will be the count of Reporting Countries Food Sources Serovars Source Countries ensuring that the loops will run within the limits of the data arrays also produced by EFSA_SAM based on the selections the user makes during analysis setup Supporting publications 2012 EN 318 35 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface
80. has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella N In source attribution model The results of EFSA_SAM are expected to be useful for the delineation of risk management strategies particularly if the model is applied on a regular basis to evaluate the impact of targeted interventions and dynamic changes in the sources of human salmonellosis It is recommended that all user of EFSA_SAM read both this technical report and the user manual before starting to use the software REFERENCES AVEC Association of Poultry Processors and Poultry Trade in the EU Countries 2011 2011 Annual Report Available from http www avec poultry eu Default aspx ID 4731 52 pp Brooks SP and Gelman A 1998 Alternative methods for monitoring convergence of iterative simulations Journal of Computational and
81. he author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food i Development of a user friendly interface version of the Salmonella National Food Institute source attribution model Select Food Sources Min 3 mn fen se Broilers Layers l Pigs If new food sources have been added during data import these food sources will be listed here By default EFSA_SAM automatically selects all food sources present in the dataset but it is possible to deselect food sources if needed It should be noted that models require a minimum number of food sources to be able to run and provide reliable results For the TT SAM and Single MS model the minimum number of food sources is three 4 3 Selecting the bacteria subtypes Here the bacteria subtypes are selected Available Subtypes ENTERITIDIS V LONDON V AGONA V MBANDAKA V ANATUM V MONTEVIDEO V BOVISMORBIFICANS V NEWPORT V BRAENDERUP V RISSEN V BRANDENBURG V SAINTPAUL V BREDENEY V TYPHIMURI
82. he bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella l source attribution model Table 1 Estimated means and standard deviations for the underreporting factors uc applied in the EFSA_SAM model Mean Ln Sdev Ln Mean dist Mean data Austria 2 1 0 8 11 2 11 Belgium 0 9 0 9 3 6 3 5 Bulgaria 6 3 0 8 734 8 718 4 Cyprus 4 9 0 8 177 2 173 3 Czech Republic 3 1 0 8 29 6 28 9 Denmark 1 2 0 8 4 5 4 4 Estonia 2 5 0 8 17 4 16 9 Finland 1 3 0 8 0 4 0 4 France 3 0 8 27 5 26 9 Germany 2 0 8 10 9 8 Greece 6 8 0 8 1257 1229 Hungary 3 9 0 8 68 3 66 8 Ireland 1 1 1 1 5 6 5 4 Italy 4 0 8 73 4 71 8 Latvia 33 0 8 45 4 44 3 Lithuania 3 8 0 8 60 5 59 1 Luxembourg 1 1 4 6 4 4 Malta 5 1 0 8 227 8 222 6 Poland 4 5 0 8 116 6 114 Portugal 7 4 0 8 2131 2 2083 8 Romania 5 5 0 8 358 4 350 2 Slovakia 3 7 0 8 54 3 53 1 Slovenia 3 4 0 9 41 7 40
83. he model can provide estimates for the effect on the number of human cases originating from a particular reservoir if the observed prevalence in that reservoir is changed or for specific subtypes e g specific serovars of Salmonella occurring in that reservoir The source attribution approach has been considered by EFSA Working Groups and Panel Experts as valid when addressing these types of questions where the use of a classical quantitative risk assessment model i e transmission models would be impaired due to a lack of data and time limitations As these models require specialist knowledge it was requested by EFSA to develop a flexible user friendly source attribution model for use for example in future mandates dealing with similar questions The objective of the work described in this report was therefore to develop a flexible and user friendly interface for attributing human cases of food borne pathogens to the responsible food animal reservoirs and or food sources The interface is based on a Salmonella source attribution model developed for setting target for Salmonella in the turkey production the Turkey Target Source Attribution Model TT SAM Results from this model were used by the BIOHAZ panel in their related Scientific Opinion The developed interface described in this report is called the EFSA Source Attribution Model EFSA_SAM The programming language development environment used for developing the user friendly interfac
84. he present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella l source attribution model It should be noted that food sources in the single MS model can include imported food products if data for these exist 2 3 Model software for the interface The BT SAM the EU SSA and the TT SAM models are all compatible with WinBUGS 1 4 3 which is a shareware that can be downloaded free of charge from http www mrc bsu cam ac uk bugs WinBUGS 1 4 3 is by experience of the authors of this report more stable than the OpenBUGS version for this type of complex models and is therefore used for executing the model in this project The WinBUGS programming language is however quite complicated particular for a 3 dimensional model as the EFSA_SAM model The objective of this project was therefore to develop a flexible and user friendly interface that can be used by food and or public health scientists or food safety risk managers with no or only little knowledge of the WinBUGS syntax The programming language development environment used for developing the user fri
85. he user can verify by scrutinizing the data that this seems to be the case the model results can still be considered valid conditioning that that convergence otherwise appears to have occurred l Density Plot for q Typhimurium DT 193 Untype BGR 1 00021 Figure 2 Kernel density plot of the posterior distribution for the subtype dependent factor q for Typhimurium DT193 The posterior distribution is similar to the defined prior distribution which may indicate that this specific subtype is only found in a single food animal source A crucial part of every Bayesian model is to monitor convergence i e to check that the model is producing robust and stable results The use of more Markov chains and monitoring that these chains after a specified number of iterations are ending up at the same results is an appropriate way of doing Supporting publications 2012 EN 318 16 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and
86. hed complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model l DTU Food e number of reported cases per subtype and MS or year for single MS model in the study period e number of travel related cases per subtype and MS in the study period For the single MS model travel data should be specified as yes to travel no to travel or unknown travel history Appendix C1 e number of outbreak related per subtype and MS or year for single MS model in the study period Based on these data EFSA_SAM will estimate the number of domestic and sporadic cases which will then through the model be allocated to specific food sources or to a group of unknown for instance if the subtype seen in humans is not found in any of the food sources included 2 4 1 1 Genus species and subtype specification There is the possibility to include three levels of subtypes which by default for Salmonella is called Serovar Phage type and Genotype In the EFSA_SAM a list of Salmonella serovars and phage types are already included in the database and imported data will be compared to this lis
87. here is often disagreement between import and export data i e the amount reported as exported in one MS to another MS do not correspond to the amount reported as imported in the receiving MS Therefore some subjective judgement is needed when analysing the trade data and it is considered best that the decision of the final data to be included is taken by the model users The EFSA_SAM includes a demo dataset for production and trade data This dataset is equivalent to the data used in the TT SAM model Hald et al 2012 In this dataset data on production of the animal food sources were extracted by EFSA from the EUROSTAT and provided as Excel files Production data for broilers and turkeys were taken from the 2010 AVEC report AVEC 2011 as the EUROSTAT data does not provide information for the separate poultry species For pigs the weight of slaughtered carcasses per MSs in 2010 was used as a measure of domestic production Finally for eggs data on the production of shell eggs were extracted from FAOSTAT since these data were missing from many MSs in the EUROSTAT data All information on trade between MSs was extracted from EUROSTAT database dataset name DS 016890 EU27 Trade Since 1988 By CN8 Export data as reported by the MSs were used for both estimating import and export This was done in order to use only one table realising that there was a high degree of disagreement of the data reported in the export and import tables for each fo
88. i ReportingCountryCount FoodSourceCount SeroVarCount Number of cases pr MS food source and serotype MS Baseline truecasesscenariocji ReportingCountryCount FoodSourceCount Sero VarCount number of cases pr MS food source and serotype MS Scenario travel ReportingCountryCount Number of estimated travel related cases per MS MS Baseline Scenario unknown ReportingCountryCount Number of estimated cases with unknown source per MS MS Baseline Scenario total ReportingCountryCount Number of estimated total cases per MS MS Baseline totalscenario ReportingCountryCount Number of estimated total cases per MS MS Scenario GoodFit ReportingCountryCount Goodness of fit ratio reported number of cases per MS divided by the estimated number of cases per MS per serovar MS Baseline Scenario totalEU Number of estimated total cases in EU EU Baseline unknownEU Number of estimated cases with unknown source in EU EU Baseline Scenario travelEU Number of estimated travel related cases in EU EU Baseline Scenario unktravEU Sum of unknownEU and travel EU EU Baseline Scenario totalEUscenario Number of estimated total cases in EU EU Scenario difftotalEU totalEU totalEUscen1 EU Scenario sourceC FoodSourceCount Number of estimated cases per source in EU EU Baseline percsourceC FoodSourceCount Percentage of estimated cases per source in EU EU Baseline sourceCscenario FoodSourceCount Number of estimated cases per source in EU EU Scenario percsourceCscenario FoodSourceC
89. iables WinBUGS Code Enteritidis 1 pct Typhimuriu Enteritidis 1 pct Typhimuriu Output variables from the model Drag a column header here to group by that column Use Type Scope Variable Description V Baseline EU serotype Number of estimated cases in EU per source and serotype V Scenario EU serotypescenaric Number of estimated cases in EU per source and serotype V Baseline MS source Number of estimated cases per source and reporting MS E Scenario MS sourcescenario Number of estimated cases per source and reporting MS E Baseline MS source2 Number of estimated cases per source and MS of origin a Scenario MS source2scenario Number of estimated cases Per source and MS of origin E Baseline MS truecasescji Number of cases pr MS food source and serotype E Scenario MS truecasesscenar number of cases pr MS food source and serotype E Baseline Sa EU Ms travel Number of estimated travel related cases per MS m Baseline Sa EU MS unknown Number of estimated cases with unknown source per MS C Baseline MS __ total NumberofestimatedtotalcasesperMS I Scenario ms totalscenario _ Number of estimated total cases per MS_ Baseline EU totalEU Number of estimated total cases in EU Baseline 50 MS GoodFit Goodness of fit ratio reported number of cases per MS divided by the estimated number of cases per MS per serovar m E Baseine soEu unknonnEu Nu
90. iendly interface version of the Salmonella institu source attribution model l DTU Food ber diagnostic based on the results of a and q for the second half of the iterations resulting in a single value for R for each posterior distribution Convergence may be assumed for practical purposes if R lt 1 05 WinBUGS user manual Finally the predictive ability or goodness of fit of the model can be assessed by estimating the ratio between the observed human cases sporadic human cases reported in each country and the number of cases predicted by the model A ratio around 1 indicates a good fit A plot of the GoodFit values from each country can indicate if the model for some countries is predicting poorly as compared to the observed data Poor fit is often related to data quality and the user may therefore want to exclude specific countries in order to evaluate the influence of these countries on the overall model results For the single MS model the goodness of fit can be assessed by comparing the variables lambdaexp expected number of cases per subtype and year with n the reported number of cases per subtype and year When the user is satisfied with the performance of the model it is time to run the Scenarios This time select a scenario in the tree view set the output variables and WinBUGS run settings and press Run WinBUGS just like you did when you ran the Baseline Supporting publications 2012 EN 318 27 The prese
91. is task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model l DTU Food Required columns for food source prevalences A Country Either two character short code or the full country name A Food_Source Text Broilers Pigs Turkeys Layers more food sources can be added by the user Genus Text as defined for the bacteria in the software Genus Salmonella for the TT SAM Species Text as defined for the bacteria in the software Species Enterica for the TT SAM SubTypeFields like Serotype Phagetype Genotype as chosen when you created the analysis For genotypes e g MLST or MLVA types the user specifies the format solely In fact the Genotype option doesn t need to be a genotype but could also be a third phenotypic subtyping level such as antimicrobial resistant patterns For the TT SAM serotypes are used Other bacteria types tha
92. itions and model codes for the TT SAM and Single Member State model 60 Supporting publications 2012 EN 318 4 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella l source attribution model BACKGROUND AS PROVIDED BY EFSA EFSA has been working on a series of Scientific Opinions originated by a mandate received by the European Commission EC in July 2008 on the review of Salmonella targets in poultry primary production The Opinions have been adopted by the BIOHAZ Panel and published on the EFSA website They have provided a quantitative estimate of the public health impact of setting new targets for the reduction of Salmonella in breeding flocks laying hens and broilers of the species Gallus gallus A similar mandate for flocks of breeding and fattening turk
93. l c unknown c sum source c 1 FoodSourceCount outbreak c totalscenario c lt travel c unknown c sum sourcescenario c 1 FoodSourceCount outbreak c GoodFit c lt o c 1 SeroVarCount 1 lambdaexp c 1 SeroVarCount 1 for evaluating goodness of fit excluding other serovars totalEU lt sum total 1 ReportingCountryCount unknownEU lt sum unknown 1 ReportingCountryCount travelEU lt sum travel 1 ReportingCountryCount unktravEU lt unknownEU travelEU totalEUscenario lt sum totalscenario 1 ReportingCountryCount difftotalEU lt totalEU totalEUscenario for j in 1 FoodSourceCount sourceC j lt sum source 1 ReportingCountryCount j percsourceC j lt sourceC j 100 totalEU sourceCscenario j lt sum sourcescenario 1 ReportingCountryCount j percsourceCscenario j lt sourceCscenario j 100 totalEUscenario diffsourceC j lt sourceC j sourceCscenario j diffpercsourceC j lt percsourceC j percsourceCscenario j percdiffsourceC j lt diffsourceC j 100 sourceC j DEFINITION OF PRIORS Food source dependent factor a two dimensional and uniform priors for c in 1 ReportingCountryCount i for j in 1 FoodSourceCount a c j dunif 0 a_MaxRange j Subtype dependent factor q uni dimensional and uniform priors q 1 lt 1 q for SE set to 1 all others will be estimated relative to this serovar q SeroVarCount lt 1 not used for
94. l purposes if R lt 1 05 WinBUGS user manual Finally the predictive ability or goodness of fit of the model can be assessed by estimating the ratio between the observed human cases and the number of cases predicted by the model A ratio around 1 indicates a good fit A plot of the ratios from each country or year and subtype in single MS model can indicate if the model for some countries or years is predicting poorly as compared to the observed data Poor fit is often related to data quality and the user may therefore want to exclude specific countries or years in order to evaluate their influence on the overall model results 2 7 Baseline and scenario analyses The data imported into EFSA_SAM will be used for a baseline analysis providing attribution estimates based on the data applied in the model Baseline results therefore give an indication of the most important sources in the current situation The results of the baseline analysis can then be compared with the results from different scenario analyses specified by the user The interface allows for two types of scenarios 1 Setting target prevalences for individual subtypes In this type of scenario the user can change the original prevalence of specific subtypes to see the effect on the number of human cases if the prevalence for instance is reduced due to targeted control The prevalence can be change for several subtypes in the same scenario and the new target prevalen
95. lmonela Enterica BOVISMORBIFICANS 49 0 0 BE Salmonela Enterica BRAENDERUP 29 0 0 BE Salmonella Enterica BRANDENBURG 72 0 0 BE Salmonela Enterica BREDENEY 18 0 0 BE Salmonela Enterica DERBY 151 0 0 BE Salmonela Enterica ENTERITIDIS 2395 0 52 BE Salmonela Enterica HADAR 72 0 0 BE Salmonela Enterica HEIDELBERG 16 0 0 BE Salmonela Enterica INFANTIS 108 0 0 BE Salmonela Enterica KENTUCKY 90 0 0 BE Salmonela Enterica KOTTBUS 19 0 0 ue Required Fields Country Genus Species Incidence Travel Outbreak Serotype cancel After column names and dimension values have been validated the file is displayed and the Import Data button is now active The data will not be imported before this button is pressed When the import button is pressed the data will finally be loaded into the analysis and the data import windows will disappear if the import succeeds If the import fails the window will not disappear and the status window to the right will make a note of what went wrong Adding values and data substitution does not change the data file on the disk 3 1 5 Filling in the blanks During the import procedure EFSA_SAM will add zero values to non existent data E g if a prevalence for a serotype foodsource country is found in the food source prevalence import file and that same serovar country is not found in the reported human cases import file then zero values for human cases will be added for this serov
96. mber of estimated cases with unknown source in EU E Baseline So EU_ travelEU Number of estimated travel related cases in EU o Baseline S EU lunktravEU Sum of unknownEU and travel EU I Scenario _EU_ totalUscenario Number of estimated total cases in EU F Scenario EU difftotalEu totalEU totalEUscen1 r F Baseline EU sourcec Number of estimated cases per source in EU E Baseline EU percsourceC Percentage of estimated cases per source in EU Total Chains 5 E Scenario EU sourceCscenario Number of estimated cases per source in EU E Iterations al Scenario EU percsourceCscer Percentage of estimated cases per source in EU Burn In 10000 A Scenario EU MS diffsourceC sourceC sourceCscen1 Run 30000 ai p eee ms nar SE menmam on Edit Distributions Density Plot Show Results Close EE Now you can select the output variable that you want to obtain results from It is always a good idea to select the GoodFit variable as this will give an indication of how well the model is performing i e able to fit the data see also the section on model convergence and goodness of fit below Here you also enter the values for the number of Markov chains the number of Burn In iterations and the number of Run iterations that WinBUGS will use during execution Note When running WinBUGS in script mode as we do here the total number of iterations r
97. mber of human cases per country is calculated A poor fit of the model for some countries is often linked to poor data quality The EFSA_SAM interface is delivered with a user manual which is also part of this report Users of the interface are recommended to read this report before starting using the interface to become familiar with the model principles and the mathematics behind which is required in order to interpret the model results and assess the validity of the model Supporting publications 2012 EN 318 3 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella I source attribution model l DTU Food TABLE OF CONTENTS a ONN ge la a E E cvedasadesvideaseyblassauyeasseeveugead 1 DS UTI ALY seeen e E N E E E E E E EAEE E 2 Table of Contents sureres oiee nae e Ea ea i De
98. mension the model was able to produce more robust results and it was assessed that even with only serotyping data available the model could still produce meaningful results This was considered to be useful for countries that use only serotyping in their national surveillance of Salmonella Through the EFSA service contracts CT EFSA BIOHAZ 2010 02 and CT EFSA ZOONOSES 2010 02 the Hald model was adapted to the EU level by including MS as a third dimension The model produces attribution estimates at the overall EU level as well as MS specific estimates and allows for exploring the Salmonella contribution from food traded between MSs by accounting for export and import figures for the included food sources 2 Development of a user friendly interface for the Source Attribution Model EFSA_SAM As described above two mathematical models for Salmonella source attribution at the EU level have been developed through two independent EFSA service contracts the BT SAM model from the CT EFSA BIOHAZ 2010 02 and the EU Salmonella Source Attribution EU SSA model from the CT EFSA ZOONOSES 2010 02 However the two models are in principle structured in the same way and employing the same type of data The user friendly model will in the following be referred to as the EFSA Source Attribution Model EFSA_SAM A user manual for the EFSA_SAM interface is provided in Appendix A 2 1 Model dimensions Both the BT SAM and the EU SSA are 3 dimensional model
99. n Salmonella can be defined and the subtype names of these can be defined by the user and are not included in the EFSA_SAM database see more about this in the advanced section of the manual It is also emphasized again that the modeling approach still needs to be validated for most other pathogens A Units_Tested integer number number of samples tested will be the denominator for calculating the prevalence per food source and country A Units_Positive integer number number of positive samples will be the nominator for calculating the prevalence per subtype food source and country The food source prevalence in percent is calculated by EFSA_SAM and is outputted as a floating point numbers with 4 decimals For the Single MS model Country is replaced by Year four digits Required columns for reported human cases Incidence travel outbreak Country Either two character short code or the full country name gt Genus Text as defined for the bacteria in the software gt Species Text as defined for the bacteria in the software gt Subtype Fields like Serotype Phagetype Genotype same as for Prevalence data described above gt Incidence integer number reported number of cases per subtype and country Travel integer number reported number of cases related to travel abroad per subtype and country A Outbreak integer number reported number of cases related to outbreaks per subtype and country
100. n the included food sources A critical part of all Bayesian models is to check for model convergence and goodness of fit The EFSA_SAM describes different ways for checking convergence and exploring model fit providing the user with some tools to assess the validity of the model The EFSA_SAM interface is delivered with a user manual which is also part of this report Users of the interface are recommended to read this report before starting using the interface to become familiar with the model principles and the mathematics behind National Food Institute Technical University of Denmark KEY WORDS Salmonella source attribution flexible user friendly interface version scenario analysis DISCLAIMER The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Question No EFSA Q 2012 00672 gt Acknowledgment The authors would like to thank EFSA
101. n the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model DTU Food 2 8 Before running the model the user is requested to select for which outcome variables statistics should be reported Table 2 and Table 3 Statistics reported per outcome variables are mean standard deviation median and 95 credibility interval Results are reported in tables that can be exported as comma separated files for further analysis or graphical display in other softwares e g MS Excel Model output and presentation of the results Table 2 Description of the variables that the user is able to select as model outputs in the EU model Type of analysis Scope Variable name Description Baseline EU serotype Number of estimated human cases in EU per source and subtype Scenario EU serotypescen Number of estimated human cases in EU per source and subtype Baseline MS source Number of estimated human cases per source and reporting MS Scenario MS sourcescen1 Number of estimated human cases per source and reporting MS Baseline MS source2 Number of estimated human cases per source and MS of origin Scenario MS source2scen 1 Number of estimated human cases per source and MS of origin Baseline MS truecasescji Number of human cases pr MS food source and subtype Scenario MS truecasescen cji Number of human cases pr MS food source and subtype Baselin
102. nBUGS syntax which can be error prone All model settings i e selection of subtypes MSs food sources etc the generated WinBUGS code and the data are stored in a scalable SQL database This ensures track keeping of all models executed through the user friendly interface making it possible to retrieve the exact setup at a later point in time for documentation Also easy to use backup and restore routines are implemented The interface software are delivered as a program file that can be installed from any kind of mass storage media e g CD USB keys or be downloaded from an internet link to the users personal computers 2 4 Data input to the EFSA_SAM model Below is a general description of the types of data that it is possible to employ in the EFSA_SAM model for source attribution analyses 2 4 1 Reported human cases of Salmonella or other pathogens Data on reported human cases of Salmonella or other pathogens with the following levels of details are required as input for the EFSA_SAM model see data template in Appendix B1 for EU model and Appendix C1 for single MS model Supporting publications 2012 EN 318 11 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is publis
103. nario unknown ReportingCountryCount Number of estimated cases with unknown source per MS MS Baseline Scenario total ReportingCountryCount Number of estimated total cases per MS MS Baseline totalscenario ReportingCountryCount Number of estimated total cases per MS MS Scenario GoodFit ReportingCountryCount Goodness of fit ratio reported number of cases per MS divided by the estimated number of cases per MS per serovar MS Baseline Scenario totalEU Number of estimated total cases in EU EU Baseline unknownEU Number of estimated cases with unknown source in EU EU Baseline Scenario travelEU Number of estimated travel related cases in EU EU Baseline Scenario unktravEU Sum of unknownEU and travel EU EU Baseline Scenario totalEUscenario Number of estimated total cases in EU EU Scenario difftotalEU totalEU totalEUscen1 EU Scenario sourceC FoodSourceCount Number of estimated cases per source in EU EU Baseline percsourceC FoodSourceCount Percentage of estimated cases per source in EU EU Baseline sourceCscenario FoodSourceCount Number of estimated cases per source in EU EU Scenario percsourceCscenario FoodSourceCount Percentage of estimated cases per source in EU EU Scenario diffsourceC FoodSourceCount sourceC sourceCscen1 EU Scenario diffpercsourceC FoodSourceCount percsourceC percsourceCscen1 EU Scenario percdiffsourceC FoodSourceCount Percentage difference between baseline and scenario diffsourceC 100 sourceC EU Scenario a Reportin
104. nchor point Often the datasets of the reported human cases and the food source prevalence data will also be constructed so that they include a group of other subtypes This will by default be specified as the last subtype in the dataset and data referred to this group is not used for estimating the attribution values meaning that human cases belonging to this group will by default be allocated to the group of unknown Because the q value for the other subtypes is not included in the actual calculations the number of priors for q ends up to equal the number of subtypes included minus two i e minus a prior for the first and last subtype which are by default set to 1 Finally the underreporting factors are included as distributions in the model one for each country As explained in section 2 4 2 these are included as lognormal distributions and prior distributions ranging from 2 to 5 are included as default in EFSA_SAM For each parameter included as a distribution with specified priors we also need to specify proper initial values Initial values tell the model where in the specified distributions to start drawing random numbers So the initial values should lie within the ranges for the prior distributions Specifying appropriate initial values can be a bit tricky and often when a WinBUGS model stops running and returns an error message the reason is that the initial values chosen resulted in implausible results given the distri
105. nderscore The column names are case insensitive 3 1 4 Dimension data value substitution Just like column names can be substituted dimension data values can also be substituted In the example shown below an entry called S Enterica has been found in the import file and the software tells that a value Enterica exists in the database and a substitution is possible Supporting publications 2012 EN 318 13 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model a gt General Settings Loading F NextPhase DelphiSource RiskAnalysis Datal Validating columns Column validation ok New Items Found New Items found for column Species S enterica enterica R
106. neeeaeesaaeenaees 16 4 2 Selecting the f00d SOULCES 35 esscidsseisscuessucdecavsesssseysbenedopueba cece ve snares sa aee a aie E 16 4 3 Selecting the bacteria sUbtyp Sissesureiorsoeeiicapcinines c ni aiae a Ee eai 17 Dt Model setina S seirer onneani ee E a e E e E Deea eea GEE E TEE AE E ETERS 18 5 1 Distribution setna S reer e REE EE EELEE EE S A EEES EE EEEE 18 5 2 Underreporting factors for human cases eeceseceseeeeeeeeeeesceeseecaecsaeceaeeeseeeseeeseeeeeeeeneeeaee 19 E o E E1 E E A E EE E E T T E E T EEE 6 1 Individual subtypes 6 2 Groups of subtypes 7 Running the amal ysis osonencs anai a a e a o iaaii iE 7 1 Da Kora GIR EXC eat To a AEP EEE E A A AA 8 Monitoring of convergence and goodness Of fit 0 00 eee eeceeeeeeeeeeeeeeeeeaeesaeecaaecsaecsaecnaeenaeenseeees 26 9 Vhevoutputiess ntedietitiss iendsiiteadcleaiiieitash tite deen dltewn lelenslidinas E A R 28 10 The advanced part of EFSA_SAM seisein eiia e EE e EE E EEE 30 10 1 Modelsettin S iiersssnoorermeineinn oaii ea a e A EEA EE 30 10 2 Variable d finiti nS esciniorsssenioninens eiee iare oer ai E EEEN AESi eSa ET 30 10 3 Modelc denirreiiprecnre etii neis a E e Re a aA AEE TE EEE EE 35 Supporting publications 2012 EN 318 2 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safet
107. nstitute source attribution model fe EFSA Source Attribution Modelling General _ Settings New Analysis E port To File Open Analysis Import From File Quit Analysis Analysis Settings Name Testing TT SAM Model TT SAM w Underreporting Turkey Target Source Attribution Model with under reporting factors n Bacteria S Genus Salmonella Species Enterica Reporting Countries Subtype Properties Serotype Source Countries Reported Cases Incidence Travel Outbreak Food Sources Loader Bacteria Types Prevalence Data Loaded Model Settings R Food Import Export Data Loaded Unde ting Facto i Herping Fanar Reporting Countries 25 Scenarios AT BE CY CZ DE DK EE ES F FR GR HU IE IT LT LU LV NL PL PT RO SE SI SK UK Source Countries 25 Analysis nz AT BE CY CZ DE DK EE ES FI FR GR HU IE IT LT LU LV NL PL PT RO SE SI SK UK Food Sources 4 Generate Winugs Code BROILERS LAYERS PIGS TURKEYS Bacteria Subtypes 24 ENTERITIDIS AGONA ANATUM BOVISMORBIFICANS BRAENDERUP BRANDENBURG BREDENEY DERBY HADAR HEIDELBERG INFANTIS KENTUCKY KOTTBUS LIVINGSTONE LONDON MBANDAKA MONTEVIDEO NEWPORT RISSEN SAINTPAUL SENFTENBERG TYPHIMURIUM VIRCHOW OTHER Scenarios 1 BaseLine A d 2 2 The center information window The center information window is where the user can follow the progress of the model setup The window information is updated every time the user makes a change in the setup and will therefore giv
108. nt document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food National Food Institute Development of a user friendly interface version of the Salmonella source attribution model 9 The output After the analysis runs the data produced by WinBUGS will be imported to EFSA_SAM z F a e 2 7 telai m SLE EE ne _ E E File Edit Format View Help stats lambdacji Node statistics node mean sd MC error 2 5 median 97 5 start sample Jambdacji 1 1 1 1 491 1 489 0 03008 0 03082 1 021 5 461 1001 2000 Jambdacji 1 1 2 0 01173 0 01172 2 354E 4 2 436E 4 0 00801 0 04353 1001 2000 Jambdacji 1 1 3 0 005317 0 005321 1 re 1 077E 4 0 00361 0 01954 1001 2000 Jambdacji 1 1 4 0 0 0 0 2 236E 12 0 0 0 0 0 0 1001 2000 Jambdacji 1 1 5 0 0 0 0 2 236E 12 0 0 0 0 0 0 1001 2000 Jambdacji 1 1 6 0 001608 0 001603 3 298E 5 3 193E 5 0 001106 0 0
109. nterface for source attribution modelling of food borne pathogens based on the codes of the existing Salmonella source attribution model This interface can be explored to evaluate the public health impact of different prevalence levels and target serovars of Salmonella spp in fattening turkey flocks in EU MSs Appropriate training to future users of the model should also be provided According to the Technical Specifications of the Negotiated Procurement Procedure NP EFSA BIOHAZ 201 1 04 the tasks to be carried out in particular are to develop a flexible user friendly generic interface for a source attribution model e The broiler target Salmonella source attribution model BT SAM should be employed to develop a windows platform user friendly generic and flexible model interface e The model which is now 3 dimensional should allow for flexibility It should be possible to change the input parameters of the model i e the number of animal food sources the number of MSs and the type and number of subtypes e It should be possible to evaluate various scenarios using the user friendly interface It should allow to change the prevalence of single subtype in one or more animal food sources and the combination of different subtypes in one or more sources e g for evaluating targets e The user friendly interface should allow for generating outputs for different scenarios the expected rates of human cases in the EU MSs including
110. od source The unit used for expressing the amount of food produced and exported was tonnes The demo model does not include data on food imported from outside of EU due to lack of data on both imported amounts and Salmonella prevalences The amount available for consumption produced in a MS was as mentioned above estimated as Production export In some instances this resulted in negative production values i e the amount exported were larger than the amount produced within the country In order to ensure that MSs would still have nationally produced food available in their own country it was assumed that imported products could also be re exported Data availability and data used are described in Hald et al 2012 2 5 Specification of prior distributions initial values and model execution Every variable or parameter included as a probability distribution in a Bayesian framework needs to have a prior distribution specified in order for the model to estimate a posterior distribution As explained in section 1 the mathematical model behind the EFSA_SAM interface relies on the estimation of a posterior distribution for the food source a and subtype q dependent factors as these along with the inputted data is used for estimating the number of human cases per food source subtype and country The prior distribution for a and q in the EFSA SAM model are as default included as Uniform distribution with values ranging from 0 100 4 FAOST
111. od source dependent factor a two dimensional and uniform priors for c in 1 ReportingCountryCount i for j in 1 FoodSourceCount a c j dunif 0 a_MaxRange j Subtype dependent factor q uni dimensional and uniform priors q 1 lt 1 q for SE set to 1 all others will be estimated relative to this serovar q SeroVarCount lt 1 not used for estimating cases value insignificant for i in 2 SeroVarCount 1 qli dunif 0 q_MaxRange 1 End of model Supporting publications 2012 EN 318 70 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model D5 Variable definitions for Single Member State Model RequiredFields FoodImportExport Y ear Food_Source Amount Prevalence Y
112. odel Table C3 Example of data template for animal food production data Year Food_Source Amount 2009 PORK 90928171 5 2009 BEEF 46099552 2009 LAYERS 31100000 2009 BROILERS 26580904 5 2009 DUCK 403698 2009 IMP_PORK 43651806 2009 IMP_BEEF 51305993 5 2009 IMP_CHICK 21333407 5 2009 IMP_DUCK 4370019 2009 IMP_TURKEY 2437732 5 2010 PORK 90930660 5 2010 BEEF 46101118 2010 LAYERS 31101230 2010 BROILERS 26582393 5 2010 DUCK 403818 2010 IMP_PORK 43657206 2010 IMP_BEEF 51311262 5 2010 IMP_CHICK 21336663 5 2010 IMP_DUCK 4370143 2010 IMP_TURKEY 2437901 5 2011 PORK 90930330 5 etc etc etc Supporting publications 2012 EN 318 59 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella ISN source attribution model ory Food D VARIABLE DEFINITIONS AND MODEL CODES FO
113. of more Markov chains and monitoring that these chains after a specified number of iterations are ending up at the same results is another appropriate way of monitoring for convergence A very simple way of monitoring convergence is by visually looking at the Kernel density plot for the food source dependent factors a and subtype dependent factors q for each chain If these appear to be overlying the model seems to have converged well A more formal approach is to calculate the bgr diagnostics Here convergence is considered to have occurred when the variance across all chains B is no larger than the variance within each individual chain W and when the chains had reached a stable level EFSA_SAM makes a calculation of the Supporting publications 2012 EN 318 26 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user fr
114. of the chains run in the model r I Kernel Density _ Sex AT BROILERS a IAT LAYERS Density Plot for a BE BROILERS ae ia 2E a8 5 Ba m CZ BROILERS BGR 1 00074 CZ TURKEYS Save This Image Save All Images FI BROILERS FI LAYERS FL PIGS FI TURKEYS FR BROILERS FR LAYERS SS FR PIGS x Ok t d If needed the plots can be saved to bitmap files bmp You can save either the plot shown or choose to save all plots When saving plots you are asked for a location folder for the plots Each file will automatically be given the name of the variable plotted A first step to check how the model is performing is to examine these Kernel density plots They should depict distributions that at the right side end in a smooth curve approximating zero If the density distribution is cut off to the right the ranges of the prior distribution appear inappropriate and need to be extended see the technical report for more information on this EFSA_SAM will display a small warning three stars if the density peak is off centered A warning will be displayed if the peak is in the right 15 percent of the range Even though warnings are displayed it is a good idea to verify that all plots are ok If changes are made to the distribution ranges new initial values for the ranges changed will be made and the baseline analysis must be run again The use
115. on a PC The Contractor s model uses the latest version called OpenBUGS see http www openbugs info w and has added the EU MSs to the model as a third dimension This type of model allows for the identification of the most important reservoirs of the zoonotic agent assisting risk managers to prioritize interventions and focus control strategies at the animal production level The model can provide estimates for the effect on the number of human cases originating from a particular reservoir if the observed prevalence in that reservoir is changed or for specific subtypes e g specific serovars of Salmonella in that reservoir This source attribution approach has been considered by WG and Panel Experts as valid when addressing this type of questions where the use of a classical quantitative risk assessment model i e transmission model would be impaired due to a lack of data and time limitations Therefore for the Opinion related to Salmonella in turkeys a similar model named the turkey target Salmonella source attribution model TT SAM was used to support the BIOHAZ Panel see previous contractor s report Hald et al 2012 The BT SAM model developed by the former Contractor is written in an OpenBUGS code and the TT SAM model is written in WinBUGS 1 4 3 As these models require specialist knowledge it was requested by EFSA to develop a flexible user friendly source attribution model for use for example in other mandates dealing with
116. on of this approach A strong advantage of the microbial subtyping approach is that it allows for the identification of the most important pathogen reservoirs assisting risk managers to prioritize interventions and focus control strategies at the animal production level Particularly if repeated on a regular basis the approach is regarded as a powerful tool to monitor the progress of control and follow the trends in the sources of human infections Hald et al 2004 Pires et al 2009 The results of this type of model can also provide estimates for the effect on the number of human cases originating from a particular reservoir e g turkeys if the observed prevalence in that reservoir is changed for instance following the implementation of a control program Given the nature of the model it will also be able to provide estimates on the expected change in human cases for specific subtypes e g specific serovars of Salmonella It should be stressed that the model attributes human cases to the reservoir level meaning that the model is not able to differentiate between transmission routes within the same reservoir For instance human cases linked to the pig reservoir will include cases both infected through consumption of pork and cases infected through direct contact with pigs Therefore and in contrast to a traditional farm to consumption risk assessment model the model does not give detailed insight into transmission routes and cannot p
117. onela Enterica VIRCHOW 58 18 0 BE Salmonela Enterica AGONA 31 0 0 BE Salmonella Enterica ANATUM 18 0 0 BE Salmonela Enterica BOVISMORBIFICANS 49 0 0 BE Salmonela Enterica BRAENDERUP 29 0 0 BE Salmonela Enterica BRANDENBURG 72 0 i BE Salmonella Enterica BREDENEY 18 0 0 BE Salmonela Enterica DERBY 151 0 0 BE Salmonela Enterica ENTERITIDIS 2395 0 52 BE Salmonela Enterica HADAR 72 0 0 BE Salmonela Enterica HEIDELBERG 16 0 0 BE Salmonella Enterica INFANTIS 108 0 0 BE Salmonela Enterica KENTUCKY 90 0 0 BE Salmonela Enterica KOTTBUS 19 0 0 Required Fields Country Genus Species Incidence Travel Outbreak Serotype In the right side of the data import window a status of the import progress can be seen A list of the fields required can be seen at the bottom of the window The required fields will change accordingly to the model chosen and the type of data that you are going to import Supporting publications 2012 EN 318 12 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights
118. oodborne infections to specific sources Foodborne Pathogens and Diseases 6 417 424 Supporting publications 2012 EN 318 20 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model APPENDICES A EFSA_SAM MANUAL VERSION 1 0 Supporting publications 2012 EN 318 21 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority
119. otype Reported Cases Incidence Travel Outbreak Not Loaded Prevalence Data Not Loaded a Food Import Export Data Not Loaded Reporting Countries 0 Source Countries 0 Food Sources 0 Generate WinBugs Code Bacteria Subtypes 0 Scenarios 1 BaseLine 3 1 Data imports The import windows for Reported human cases Prevalence in food sources and Food Import Export are more or less the same so there is no need to cover all three of them in detail However all the different features of the imports will be covered in detail Supporting publications 2012 EN 318 9 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model l DTU Food 3 1 1 Data format and validation EFSA_SAM can import data in CS
120. ount Percentage of estimated cases per source in EU EU Scenario diffsourceC FoodSourceCount sourceC sourceCscen1 EU Scenario diffpercsourceC FoodSourceCount percsourceC percsourceCscen1 EU Scenario percdiffsourceC FoodSourceCount Percentage difference between baseline and scenario diffsourceC 100 sourceC EU Scenario a ReportingCountryCount FoodSourceCount Food source related factor per source and MS EU MS Baseline Scenario q SeroVarCount Serovar related factor One per serovar EU MS Baseline Scenario uf ReportingCountryCount Underreporting factor One per MS EU MS Baseline Scenario Supporting publications 2012 EN 318 61 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model D2 WinBUGS code for TT
121. ove as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella I source attribution model l DTU Food Underreporting u ReportingCountryCount consisting of MeanUnderReporting m u ReportingCountryCount StandardDeviationUnderReporting s u ReportingCountryCount Distributions The Distributions section defines the distributions and their dimensions Distributions a ReportingCountryCount FoodSourceCount q SeroVarCount u ReportingCountryCount Distribution_a Grouping FoodSourceCount Default_Range 0 100 Visible True FloatingPoint False UseMaxRangeArray True Transpose True Distribution_q Grouping Default_Range 0 100 NA_Elements 1 SeroVarCount Visible True FloatingPoint False UseMaxRangeArray True Transpose False Distribution_u Grouping Default_Range 2 5 Visible True FloatingPoint True UseMaxRangeArray False Transpose False
122. per year total Number of estimated total human cases per year percsource Percentage of estimated human cases per source per year lambdatji Number of estimated sporadic and domestic human cases per subtype source and year lambdaexp Number of estimated human cases per subtype and year a Food source related factor one per source q Subtype related factor one per subtype The q value for the first subtype in the demo data S Enteritidis is by default set to 1 and the others q values are estimated relative to this one 2 9 Saving and exporting an analysis After having finished an analysis the whole analysis is automatically saved The analysis including set up and results can also be exported as individual files in an easy to use format for further analysis or graphical display in other softwares e g Excel The whole analysis including results can also be saved as a zip file for future documentation purposes CONCLUSIONS AND RECOMMENDATIONS The source attribution approach based on subtyping has been assessed to be a valuable tool for pointing out the most important source of human salmonellosis in the EU as well as to predict changes in human incidence if prevalences in different animals reservoirs and for specific serovars are changed for instance as a result of targeted control The principle has also been applied in individual countries for prioritizing risk management strategies A drawback of the models used for source attribution is
123. procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella source attribution model Table C2 Example of data template for reported animal food data Resistance Year Genus Species Serotype Phage type profile Food_Source Units_Tested Units_positive 2009 Salmonella Enterica Enteritidis PT 21 BEEF 2267 0 2009 Salmonella Enterica Enteritidis PT 4 BEEF 2267 0 2009 Salmonella Enterica Enteritidis PT 8 BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 104 RRRRSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 120 RSRRSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 120 RSRSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 193 RSRRSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium NT RSRSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium U302 RRRRSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 12 SSSSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 120 SSSSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 17 SSSSSSSSS BEEF 2267 0 2009 Salmonella Enterica Typhimurium DT 193 SSSSSSSS
124. prus Czech Republic SRF 183 18 FS 1 ea Ve 18 3 18 MRR 2 2 S R 8 Havelaar AH Ivarsson S L fdahl M and Nauta MJ in press Estimating the true incidence of campylobacteriosis and salmonellosis in the EU 2009 Epidemiology and Infection published online 13 April 2012 DOI http dx doi org 10 1017 S09502688 12000568 Supporting publications 2012 EN 318 19 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model 6 Scenarios Scenarios can be added to EU multi member state models by simply copying the baseline data and settings and allowing the user to change target prevalence data and then execute different scenarios in WinBUGS It is possible to set target prevalences for
125. ption in the country i e domestic production typically estimated as total production minus export For the Single MS model Country is replaced by Year 4 digits When choosing one of the data imports from the sidebar a file dialog will appear When the appropriate file has been chosen the data import window will appear ry Import Human Cases Incidence Travel Outbreak LL ee E Se a pos LI Country Genus Speces Serotype Incidence Travel Outbreak a Loading and validating C EFSA_SAM dataimport_dem AT Salmonela Enterica AGONA 37 9 o cee Seaton Data validation ok AT Salmonela Enterica ANATUM 13 4 0 AT Salmonela Enterica BOVISMORBIFICANS 14 0 0 AT Salmonela Enterica BRAENDERUP 35 17 0 AT Salmonela Enterica BRANDENBURG 9 0 0 AT Salmonela Enterica BREDENEY 8 0 0 AT Salmonela Enterica DERBY 10 1 0 AT Salmonela Enterica ENTERITIDIS 6102 673 217 AT Salmonella Enterica HADAR 62 22 0 AT Salmonela Enterica HEIDELBERG 16 2 0 AT Salmonela Enterica INFANTIS 107 5 0 AT Salmonella Enterica KENTUCKY 22 11 0 AT Salmonella Enterica KOTTBUS 17 5 0 AT Salmonela Enterica LIVINGSTONE 10 0 0 AT Salmonela Enterica LONDON 8 0 0 AT Salmonella Enterica MBANDAKA 12 4 0 AT Salmonela Enterica MONTEVIDEO 21 4 0 AT Salmonela Enterica NEWPORT 38 11 0 gt AT Salmonella Enterica OTHER 853 126 0 AT Salmonela Enterica RISSEN 12 4 0 AT Salmonella Enterica SAINTPAUL 65 7 10 AT Salmonella Enterica TYPHIMURIUM 958 65 194 AT Salm
126. rinciple to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food i Development of a user friendly interface version of the Salmonella National Food Institute source attribution model Model Settings Model Name Description Settings Variable Definitions Code Data Initial values Script Distributions a SourceCountryCount FoodSourceCount q SeroVarCount Distribution_a Grouping FoodSourceCount Default_Range 0 100 Distribution_q Grouping SeroVarCount Default_Range 0 100 NA_Elements 1 SeroVarCount KernelDensityPlot a Coda sam_output_a1 txt sam_output_alndex txt q Coda sam_output_q1 txt sam_output_gIndex txt BaseLine_Script BurnInIterations 1000 Runiterations 2000 Chains 5 Scenario_Script BurnInIterations 1000 Runiterations 2000 Chains 1 The required fields section defines which column names must be present in the import data files These are the fields that the importer will validate is present RequiredFields FoodImportExport Country_From Country_To Food_Source Amount Prevalence Country Genus Species Food_Source Units_Tested Units_Positive HumanCase Country Genus Species Incidence Travel Outbreak Following field names can be
127. rovide estimates for the expected changes in human infections by the introduction of specific intervention strategies The model can however still employ data from carcass or food Supporting publications 2012 EN 318 8 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella l source attribution model sampling as long as the subtype distribution obtained from this sampling is assessed to reflect the subtypes distribution in the original reservoir e g cross contamination from other sources is minimal The Hald model described above was initially designed with two dimensions Salmonella subtype and food source In 2010 the model was extended by Pires and Hald 2010 to include a temporal dimension year for trend analyses within a single country By including a temporal di
128. s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model The Dimension distributions window will be available when the baseline analysis has been run in order to adjust the ranges if the Density plots show that the posterior distributions are out of range how to check this is explained later 5 2 Underreporting factors for human cases If the model selected supports underreporting factors this window will be available Here it is possible to verify and adjust the factors distribution means and standard deviations used The underreporting factors available as default in the EFSA_SAM interface are based on the methodologies published by Havelaar et al 2012 Verify Under Reporting Country Abbr Country Name Under ReportingMean Under Reporting SDev Austria Belgium Bulgaria Sudtzertand Cy
129. s was introduced The bacteria dependent factor qi can be interpreted as combining survivability virulence and pathogenicity of the pathogen to estimate the ability of that subtype to cause disease whereas the food source dependent factor aj estimates differences between food sources in characteristics that affect their ability to act as vehicles for foodborne infections e g general differences in bacterial load food characteristics influencing growth behaviour or preparation procedures It is however emphasised that the estimated values of the bacteria and food source dependent factors are simply multiplication factors comparable to regression coefficients in regression analyses that helps us to arrive at the most likely solution given the observed data Their relative size can provide an idea about the differences between subtypes and food types with respect to causing human infections but estimates based on the results of a single model should be interpreted with care However by applying the model on a regular basis as new data become available it may be possible to monitor the main sources and dynamics in the occurrence of human salmonellosis and to improve the estimation of the model parameters including the bacteria and food source dependent factors Supporting publications 2012 EN 318 7 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusivel
130. s with the following dimensions e EU Member State MS e Animal food source e Salmonella serotype However the two models include different numbers of MSs and Salmonella serotypes i e the length of the arrays varies For future models it is expected that different array lengths would be needed for instance by the inclusion of more or fewer MSs or animal food sources The EFSA_SAM model is therefore made flexible so that a user specified number of MSs food sources and Salmonella serotypes can be included There is a specific model script addressing the situation for a model with only a single MS included as it is expected that some MSs would like to apply the model to national data only In this single MS model it is possible to include time e g year as a new third dimension replacing the MS dimension from the EFSA_SAM model The single MS model may be useful for MSs that want to apply the model on a regular basis to evaluate the effect of current Salmonella control The number of food sources that can be included is in theory unlimited but a minimum of three food sources should be included in order to produce sensible results The included food sources should represent the most important sources of Salmonella in the MSs included in the model and will for the Supporting publications 2012 EN 318 9 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusivel
131. side end in a smooth curve approximating zero Figure la If the density distribution is cut off to the right the ranges of the prior distribution appear inappropriate and need to be extended Figure 1b This is easily done in the EFSA_SAM interface as explained in the user manual Appendix A l Density Plot for a AT TURKEYS Density Plot for q Typhimurium DT 193 55555 BGR 1 00098 a BGR 1 00005 b Figure 1 Kernel density plot of the posterior distribution for a the food source dependent factor a for turkeys in Austria where the ranges of the prior distribution appears to be properly specified plot from EFSA_SAM model and b the subtype dependent factor q for Typhimurium DT 193 fully susceptible where the ranges of the prior distribution needs to be expanded to obtain a well defined posterior distribution for this q values plot from single MS model Sometimes and particularly in the single MS model the density plot of a posterior distribution can appear similar to the defined prior distribution Figure 2 This means that the model cannot find a proper value for the specific parameter e g g Most often this is caused by the fact that the subtype in question is found in only one single food animal source and because the a and q values are interlinked in a multiparameter prior the uncertainty around q in such a situation is included in the estimate for a i e the model doesn t care what value q takes If t
132. similar questions 3 Regulation EC No 2160 2003 of the European Parliament and of the Council of 17 November 2003 on the control of salmonella and other food borne zoonotic agents OJ L 325 12 12 2003 p 1 15 Supporting publications 2012 EN 318 5 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model l DTU Food TERMS OF REFERENCE AS PROVIDED BY EFSA The purpose of the contract was to provide the EFSA BIOHAZ panel with a source attribution model to evaluate targets specifically answering the Terms of Reference of the Salmonella in turkey flocks mandate EFSA Q 2010 00899 which has been covered in a previous report Hald et al 2012 In addition the objective is to provide EFSA with a generic flexible but at the same time user friendly i
133. t i for k in 1 SourceCountryCount i casescenariockji c k j i lt p_scen k j i m c k j a c j q i truecasesscenariockji c k j i lt casescenariockji c k j i ee et lt sum truecasesscenariockji c 1 SourceCountryCount j i Supporting publications 2012 EN 318 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model summing over k by c for j in 1 FoodSourceCount for k in 1 SourceCountryCount for c in 1 ReportingCountryCount truecasesorigin c k j lt sum truecasesckji c k j 1 SeroVarCount 1 truecasesoriginscenario c k j lt sum truecasesscenariockji c k j 1 SeroVarCount 1 OUTPUT serovar per source for i in 1 SeroVarCount 1 for j in 1 FoodSourceCount
134. t as explained in the user manual For genotypes e g MLST or MLVA types the user specifies the format solely In fact the Genotype option doesn t need to be a genotype but could also be a third phenotypic subtyping level such as antimicrobial resistant patterns 2 4 2 Underreporting factors To take account for differences in human case underreporting between MSs the EFSA_SAM model includes the possibility to employ underreporting factors for the MSs The underreporting factors included in the EFSA_SAM are those used for the model answering the turkey target mandate the TT SAM model as described in Hald et al 2012 The methodologies used for estimating the underreporting factors are described in Havelaar et al in press The underreporting factors are based on data from 2009 and included as probability distributions in the EFSA_SAM model in order to account for uncertainty around the data A lognormal distribution was found to provide a good fit of the data and the estimated means and standard deviations were used as model input Table 1 It is possible to revise these if when better or more recent data become available in the future Underreporting factors u may differ depending on the pathogen in question Alternative underreporting factors should therefore be considered by users if the model is applied for other pathogens Supporting publications 2012 EN 318 12 The present document has been produced and adopted by t
135. t i lambdaexp t i for j in 1 FoodSourceCount lambdatji t j i lt p t j i m t j a t j qi for j in 1 FoodSourceCount source t j lt sum lambdatji t j 1 SeroVarCount 1 percsource t j lt source t j 100 total t travel t lt sum dt t unknown t lt spdo t SeroVarCount sum uk t 1 SeroVarCount 1 outbreak t lt sum outb t 1 SeroVarCount total t lt sum source t FoodSourceCount travel t unknown t outbreak t End for t in 1 YearCount for t in 1 YearCount for j in 1 FoodSourceCount a t j dunif 0 a_MaxRange j q 1 lt 1 q SeroVarCount lt 1 for i in 2 SeroVarCount 1 qli dunif 0 q_MaxRange 1 End Model Supporting publications 2012 EN 318 73 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors
136. tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella Institu source attribution model DTU Food 5 A Export To File Exports the current analysis to a zip file This includes the model data and results A Quit Closes EFSA_SAM EFSA Source Attribution Modelling F General Settings New Analysis Open Analysis Quit Import From File Analysis The Settings menu is where the more advanced features are located That is creating and modifying models maintaining the more general data such as countries food sources bacteria types and subtypes etc Also database backup and restore functionality is located here 2 1 The workflow sidebar The workflow sidebar is where the user controls the work flow in an easy to understand sequence of tasks to perform First data imports such as number of reported human cases incidence food source prevalences and food import export values Then if needed changing the analysis dimensions such as reporting and source countries food sources and bacteria
137. the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella I source attribution model l DTU Food Output_Variables The Output_Variables section defines which variables from the code are available to EFSA_SAM and that subsequently be produced by WinBUGS when the model is run Output_Variables serotype FoodSourceCount SeroVarCount Number of estimated cases in EU per source and serotype EU Baseline serotypescenario FoodSourceCount SeroVarCount Number of estimated cases in EU per source and serotype EU Scenario source ReportingCountryCount FoodSourceCount Number of estimated cases per source and reporting MS MS Baseline sourcescenario ReportingCountryCount FoodSourceCount Number of estimated cases per source and reporting MS MS Scenario source2 SourceCountryCount FoodSourceCount Number of estimated cases per source and MS of origin MS Baseline source2scenario SourceCountryCount FoodSourceCount Number of estimated cases per source and MS of origin MS Scenario truecasescji ReportingCountryCount FoodSourceCount Sero VarCount Number of cases pr MS food source and serotype MS Baseline truecasesscenariocji ReportingCountryCount FoodSourceCount Sero VarCount number of cases pr MS food source and serotype MS Scenario travel ReportingCountryCount Number of estimated travel related cases per MS MS Baseline Sce
138. the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model l DTU Food this For some modeling problems more solutions may be available which is why it is important to check that this is not the case or if it is try to explain the possible reasons A very simple way of monitoring convergence is by visually looking at the Kernel density plot for the acj and qi values for each chain If these appear to be overlying the model seems to have converged well In Figure 1 all three chains indicated by different colours overlap in an appropriate way A more formal approach is to calculate the bgr diagnostics as described by Brooks and Gelman 1998 Here convergence is considered to have occurred when the variance across all chains B is no larger than the variance within each individual chain W and when the chains had reached a stable level In WinBUGS the ratio R B W is plotted over time iterations and will tend to 1 as convergence is approached However these plots cannot be exported from WinBUGS Consequently EFSA_SAM makes its own calculation of the bgr diagnostic based on the results of a and q for the second half of the iterations resulting in a single value for R for each posterior distribution The bgr diagnostic value is presented in each density plot Figure 1 Convergence may be assumed for practica
139. ther WinBUGS or EFSA_SAM is installed first 1 1 WinBUGS installation Prerequisites WinBUGS 1 4 3 must be installed in order to use EFSA_SAM WinBUGS is a freeware and a lot of resources can be found here http www mrc bsu cam ac uk bugs winbugs contents shtml WinBUGS Download and installation Download http www mrc bsu cam ac uk bugs winbugs winbugs 14 zip This version will work on both Windows XP Windows Vista and Windows 7 32 and 64 bit versions and you will be able to control exactly where WinBUGS will be located Create a folder C EFSA_SAM Winbugs 14 and unpack the contents of the zip file to this folder Start your copy of WinBUGS 14 by double clicking C EFSA_SAM Winbugs14 Winbugs 14 exe Verifying the WinBUGS version When WinBUGS is running press Help and then About WinBUGS The version number must be 1 4 3 If you just see 1 4 then you need to upgrade WinBUGS to 1 4 3 28 About WinBUGS gt D WinBUGS zoe with ha Yel DoodleBUGS Yj Ys Version 1 4 I 268 Lys Ly BUGS 1990 1996 Medical Research Council MRC UK B UG S WinBUGS 1996 2003 Imperial College amp MRC UK How to upgrade Navigate to http www mrc bsu cam ac uk bugs winbugs WinBUGS14_cumulative_patch_No3_06_08_07_RELEASE txt Lunn DJ Thomas A Best N and Spiegelhalter D 2000 WinBUGS a Bayesian modelling framework concepts structure and extensibility
140. thor s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model Select Sub Type Properties to ivi l Phagetype l Genotype After pressing Ok the window will disappear and the sidebar and center information window will appear At this point information about the bacteria types and the chosen model will be present Also a list of settings you are going to set up will be present oO e EFSA Source Attribution Modelling General Settings New Analysis Export To File Open Analysis ous Import From File Analysis Data Imports a H H Analysis Settings Reported Human C AE EE Name TT SAM Analysis with underreporting Prevalence In Food Sources Model TT SAM w Underreporting Food epartcoort Tiskey Target Source Aliribulion Model with under reporting kactors Pe Bacteria ki Genus Salmonella Species Enterica Subtype Properties Ser
141. to your clipboard Click File New Paste the text into the blank window From the Tools menu pick the Decode option A dialog box will appear Click on the Decode All button to install the key Quit and restart WinBUGS to start using the full version Failing to install the license key properly will result in an error message and termination of WinBUGS when trying to run a model regardless whether it is done directly from WinBUGS or via EFSA_SAM 1 2 EFSA_SAM installation Insert the CD ROM or USB Key Run EFSA_SAM_ Setup exe The rest of the installation will run automatically A shortcut to the program will be placed on the user s desktop The software will by default be installed in the C EFSA_SAM folder 2 The user interface When starting EFSA_SAM a simple main window will appear It contains a top menu containing two main items General and Settings The General menu has three sub menu items New Analysis Creates a new analysis Open Analysis Opens a previously created analysis Import From File Will copy an already existing analysis making it possible to change settings and rerun the analysis Supporting publications 2012 EN 318 5 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a
142. totalEU totalEU totalEUscen1 EU Scenario sourceC FoodSourceCount Number of estimated cases per source in EU EU Baseline percsourceC FoodSourceCount Percentage of estimated cases per source in EU EU Baseline sourceCscenario FoodSourceCount Number of estimated cases per source in EU EU Scenario percsourceCscenario FoodSourceCount Percentage of estimated cases per source in EU EU Scenario diffsourceC FoodSourceCount sourceC sourceCscen1 EU Scenario diffpercsourceC FoodSourceCount percsourceC percsourceCscen1 EU Scenario percdiffsourceC FoodSourceCount Percentage difference between baseline and scenario diffsourceC 100 sourceC EU Scenario a ReportingCountryCount FoodSourceCount Food source related factor per source and MS EU MS Baseline Scenario q SeroVarCount Serovar related factor One per serovar EU MS Baseline Scenario uf ReportingCountryCount Underreporting factor One per MS EU MS Baseline Scenario Supporting publications 2012 EN 318 67 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority
143. tributable to the different food animal sources in the actual situation The results of the baseline analysis can be compared with the results from one or more scenario analyses specified by the user The interface allows for two types of scenarios i the setting of target prevalences for individual subtypes and ii the setting of a combined target prevalence for a group of subtypes In the first type EFSA_SAM will automatically change the original prevalence to the set target prevalence but only if the original prevalence is greater than the target prevalence In the latter type the users can select any number of subtypes for which a combined prevalence should be equal or less to a set target prevalence The EFSA_SAM generates a new set of subtype specific prevalences that are proportionally scaled down from the original prevalences in order to result in an overall prevalence equal to or less than the target prevalence A comparison of the baseline and scenario results can be used to assess the effect on the predicted number of human cases if targeted control measures are implemented for specific subtypes or groups of subtypes A critical part of all Bayesian models is to check for model convergence and goodness of fit The EFSA_SAM describes different ways for checking convergence and include the calculation of the bgr diagnostics that is also a part of the WinBUGS software For exploring goodness of fit a ratio between the observed and predicted nu
144. un is the Burn In iterations PLUS the Run Iterations When running in GUI mode using the WinBUGS user interface the total number of iterations is the run iterations where the user can selecct how many of these should be counted as Burn In As a start chains and iteration values will be set to a default value chosen by the creator of the model the first time you enter the baseline and scenarios The number of iterations needed depends on the complexity of the model but a Burn In of 10 000 and a number of iteration of 30 000 run in 3 5 chains appears appropriate for the EU Salmonella model the demo data Since you might want to run the scenarios under different conditions and wanting different data outputs each scenario has its own set of variables and WinBUGS run settings Chains and Iterations Above the table with the variables you can see two tabs The Output Variables and the WinBUGS code tab Supporting publications 2012 EN 318 23 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority res
145. underreporting and the percentages in terms of the EU expected rate mean statistics 2 5 and 97 5 statistics by food type subtype as done in the previous contractor s report The interface should include some graphical tools as well e g density plots of the uncertainty distributions of the total reduction of human cases of foodborne related salmonellosis and these related to a targeted animal food source only for the scenarios A menu should be provided to include only a selection of the data for example for incorporating only one MS and excluding including trade The interface should also have a reporting menu in which results to be used are output and presented in a sort of report format This contract grant was awarded by EFSA to The National Food Institute Technical University of Denmark Denmark Contract title Development of a flexible user friendly interface version of the Salmonella source attribution model developed under CFT EFSA BIOHAZ 2010 02 for evaluating targets in turkey meat production EFSA Q 2010 00899 and use in future source attribution assessments Contract number CT EFSA BIOHAZ 201 1 02 Supporting publications 2012 EN 318 6 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedur
146. version of the Salmonella National Food Institute source attribution model Model Settings Model Name SA_Jan_Nov12 Estimating the number of sporadic and domestically acquired cases excluding travelling a cin 1 ReportingCountryCount for jin 1 SeroVarCount spdo c i lt ofc i yt c i Exd serovar others We don t indude these in the attribution step Attributing cases to sources for cin 1 ReportingCountryCount for jin 1 SeroVarCount 1 ofc i dpois lambdaexp c i lambdaexp c i lt lambdaexp1 c i u c lambdaexp 1 c i lt lambda c i yt c i lambda c i lt sum lambdacji c 1 FoodSourceCount i uk c i lt ofc i lambdaexp c i uk2 c i lt o c i ambdaexp c i It is highly recommended to describe document all variables used in the code This can be done either as comments in the code or in the description section of the model Also meaningful names for the variables are highly recommended This will make it easier for other people to understand what is going on in the inner workings of the model Supporting publications 2012 EN 318 36 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is
147. view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model 3 1 3 Column name substitution After selecting a file EFSA_SAM will check if the columns required are present in the file If one or more columns are missing a misspelled column will appear as missing the user will be given the choice to substitute one or more column names in the file with the names that the software expects This will not change anything in the file on the disk but just serve as a way to explain the software what to look for instead The EFSA_SAM will display the column names it expects throughout the rest of the analysis setup i e if a column in the file was called Serovar and a column name Serotype was expected the term Serotype will be used Substitute Column Names Please substitute missing column names Current Items Units_Positive YEAR p COUNTRY GENUS SPECIES SEROVAR e FOOD_SOURCE UNITS_TESTED UNITS POSITIVE PHAGETYPE GENOTYPE In the example shown above you can see to the left that two columns in the file were missing Units_Positive and Serotype and the columns that were found in the file Here we will just substitute SEROVAR with Serotype and UNITS POSITIVE with Units_Positive note the missing u
148. w Datra No Under REporting 20 05 2012 Test New Data 20 05 2012 vr Supporting publications 2012 EN 318 29 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella National Food Institute source attribution model 10 The advanced part of EFSA_SAM This part of the manual is aimed for advanced users needing to adapt EFSA_SAM to WinBUGS models Adding new models into EFSA_SAM is done in the Model Settings windows found in the Settings Menu 10 1 Model settings On the settings tab it is possible to select which items are to be shown in the sidebar Items not needed in order to run the model should be unchecked making the user interface simpler and easier to use when actually running the analyses based on the model Model Settings Model Name A_
149. y by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors DTU Food Development of a user friendly interface version of the Salmonella l source attribution model majority of countries in EU include pigs pork meat broilers chicken meat and layers eggs The EFSA_SAM model is by default specifying pigs pork meat broilers chicken meat layers eggs and turkeys turkey meat but with the possibility to add additional sources if relevant data are available The number of Salmonella subtypes that can be included is also unlimited It will to a large extend be the data including data availability and quality that in the end will define the number included for a specific model The EFSA_SAM model is developed to address Salmonella source attribution including the specification of Salmonella serovars and phage types for S Enteritidis and S Typhimurium In addition it is possible to include further subtyping levels such as antimicrobial resistance patterns and genotypes e g MLV A The EFSA_S
150. y by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model l DTU Food The basic equation used to estimate the number of human cases per source and subtype is defined as follows dij Pij Mj aj Gi Equation 1 where Aj is the expected number of cases per subtype i and source j p is the prevalence of subtype i in source j M is the amount of source available for consumption in the country a is the food source dependent factor for source j and q is the bacteria dependent factor for subtype i To avoid problems related to identifiability i e overparameterisation of the model described in Eq 1 the number of estimated parameters needs to be reduced The pooling of some subtypes or food sources into groups with similar characteristics is one way of addressing this problem Depending on the available data the model can be extended to include other dimensions such as time period e
151. y Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model DTU Food INTRODUCTION The flexible user friendly interface software for the EFSA Source Attribution Model hereafter called EFSA_SAM is split into two parts A general user part and an advanced user part The reason for this is that higher flexibility often leads to higher complexity and higher complexity often leads to lower user friendliness Also very often software in general is going to be used by two groups of people The occasional users that just needs to perform a simple task and the advanced users that need to be able to adjust a lot of details Transferring this to EFSA_SAM the software is split into two main areas One general area is used for running analyses based on predefined models importing data setting up analysis dimensions and running the model The other advanced area is used for setting up the models e g model code WinBUGS Scripting WinBUGS EFSA_SAM interface vari
152. yCount SeroVarCount ENTERITIDIS AGONA ANATUM BOVISMORBIFICANS BRAENDERUP BRANDENBURG BREDENEY DERBY HADAR HEIDELBERG INFANT n structure Data c AT 102 37 13 14 35 3 8 10 62 16 107 BE 2395 31 18 43 29 72 18 151 72 16 108 CY 213 4 4 0 8 i is 4 a 0 4 CZ a H E 35815 51 14 20 7 4 a 35 76 a 230 DE E Chains 74622 206 151 463 171 367 45 570 267 34 1778 Total Chains 0 Ea E Iterations Burn In Run 0 0 Run winbugs Edit Distributions Density Plot Show Results Supporting publications 2012 EN 318 24 The present document has been produced and adopted by the bodies identified above as author s This task has been carried out exclusively by the author s in the context of a contract between the European Food Safety Authority and the author s awarded following a tender procedure The present document is published complying with the transparency principle to which the Authority is subject It may not be considered as an output adopted by the Authority The European food Safety Authority reserves its rights view and position as regards the issues addressed and the conclusions reached in the present document without prejudice to the rights of the authors Development of a user friendly interface version of the Salmonella source attribution model 7 1 Model execution Basi

Publisher version

Contents

Download Pdf Manuals

Related Search

Related Contents