Probability ofclassical swine fever virus introduction in the European Union: analysis and modelling C.J.deVos,H.W.Saatkamp andR.B.M.Huime FarmManagementGroup,DepartmentofSocialSciences,WageningenUniversity,Hollandseweg1, 6706KNWageningen,TheNetherlands.E-mail:[email protected] 1. Introduction Recent classical swinefever (CSF)epidemicsin theEuropean Union(EU)incurred high economiclosses.In 1993/94Germany and Belgium wereseverelyhitbyaCSFepidemic, 217farmsbeingaffected inGermany (Kramer etal., 1995;Pittleretal., 1995)and 55in Belgium (Laevensetal., 1998;Koenen etal., 1996;Vanthemsche, 1996).Even more disastrouswasthe 1997/98 CSFepidemic intheNetherlands,whichaffected 429 farms, whilemorethan 10million pigsweredestroyed preventively and for welfare reasons (Anonymous, 1998).Thecostsofthisepidemic (i.e.direct costsand consequential lossesto farms andrelated industries)wereestimated at US$2.3billion (Meuwissen etal., 1999). Import ofinfected piglets from theNetherlands alsolead toamajor epidemicinSpain,99 farms being infected in 1997/98(Edwardsetal.,2000;Greiser-Wilkeetal.,2000). Theintroduction ofCSFisacontinuing threat tothepigproduction sectoroftheEU.The diseaseisstillpresent in somecountriesofcentral andeastern Europe(Edwardsetal.,2000). Besides,CSFoccursinan endemic form inwild boarpopulations insomeareasof Germany, Franceand Italy(Laddomada,2000),representing apermanent CSFvirus(CSFV) reservoir. Thedisease can thuseasily bereintroduced intofreeregions ifnopropermeasuresaretakel Prevention ofCSFVintroduction should therefore getthehighestprioritypossible. Most outbreaks inthemajor CSFepidemicsmentioned aboveoccurred intheso-called denselypopulated livestock areas(DPLAs)thathaveanaveragepigdensityofmorethan3ÓÖ pigs/km2(Michel &DeVos,2000).Theseareascameintobeingduetoeconomic factors, such astheavailability ofcheapfeedstuff andreasonably pricedland andthevicinityof urban markets(Dijkhuizen &Davies, 1995;Huirneetal., 1995).Theconcentration ofpig production in theseareas issupposed tobecorrelated withtheriskofintroduction andspread ofepidemicdiseases(Dijkhuizen &Davies, 1995).Pigandpigfarm density are,however, nottheonlydeterminantsintheriskofvirusintroduction.Insight intoallfactors contributing totheriskofvirusintroduction isaprerequisite for takingpreventiveactionsthatareboth epidemiologically effective and economically sensible,andistherefore ofutmostimportance insupporting policy-making. This study focuses ontheprobability ofCSFVintroduction for severalregions intheEU. The main objective wastogain moreinsight intothemajor factors contributing tothisprobability. Two approaches wereused:(i)aqualitative assessment thatprovidesatoolforaquick analysisthat canbeperformed withrelatively littleinformation and (ii)aquantitative approach using acomputermodel,thatprovidesmoredetailed insightintothefactors that contribute totheprobability ofCSFVintroduction,but forwhich alotofquantitativedataare needed. Thepaper startswithanoverviewofsomerelevant definitions. Then apathwaydiagram giving anoverview ofallpathwayspossiblycontributing totheprobability ofCSFV introduction ispresented. This was used as the basis for both the qualitative'assessment SfirT the computer model. Results of the qualitative assessment are given for five densely and five sparsely populated livestock areas (SPLAs) in the EU. Model results are presented for five EU member states.The paper concludes with discussion of the approaches used and the results obtained. 2. Definitions As not all definitions in the field of animal health risk analysis are standardised yet, a brief overview of key terms used throughout this paper is given first. Virus introduction is defined as entrance of virus into the livestock production sector of a region free of the disease, causing a primary outbreak1. The definition of a primary outbreak was derived from EU Council Directive 82/894/EEC: 'an outbreak not epizootiologically linked with a previous outbreak in the same region of a member state, or the first outbreak in a different region of the same member state' (CEC, 1982).The regions used for this definition are areas with a surface of at least 2000 km 2 , that must be controlled by competent authorities and at least comprise one of a certain, member state-dependent, administrative area, e.g., provinces in Belgium, Italy and Spain, counties in the United Kingdom and Ireland and departments in France (Council Directive 64/432/EEC, Article 2 (CEC, 1964)). Using the EU definition of aprimary outbreak, more primary outbreaks can occur within one epidemic Thefocus inthisarticleisonCSFVintroduction intothedomesticpigpopulationofaregion.Virus introduction intothewildboarpopulation hasnotbeenconsidered,becauseprevention andcontrolstrategies usedaredifferent (CEC, 1980),aswellastheeconomicconsequencesofsuchanintroduction. ifvirusisspreadfromoneregion toanother.Thiswasthecase,for example,duringthe 1997/98CSFepidemic intheNetherlands,inwhich fourprimary outbreakswererecorded. Introduction ofvirusisinducedby so-called pathways.Thesearethecarriersand mechanisms that cantransmitvirusfrom aninfected toasusceptibleanimal.Pathwayscan eitherbeexogenous orendogenous.Exogenouspathwaysarelinked withvirussources outside theregion wheretheymight causeaprimaryoutbreak,whereasendogenouspathways residewithin theregion affected. Theregionswhereexogenouspathwaysmaycomefrom are called regionsoforigin,whereastheregion affected iscalled thetarget area. 3. Pathway diagram Toobtain moreinsight intotheprobability ofCSFVintroduction, aso-called pathway diagram wasconstructed toshow allpossiblepathways for CSFVintroduction, including theirmain eventsandinterrelations (Fig. 1). Apathwaydiagram isatree-like approach that provides insight intoallpossiblecausesof anadverseevent(DeVosetal.,2001).Inorder for theadverseeventtooccur,allcontributing eventsofacertainpathwayhavetobetrue. Assigningprobabilitiesto alleventsinthediagrammakesitpossible toestimatethe probability ofoccurrenceoftheadverseevent. Insert FigureJ Thepathway diagram consistsoffour levels.Startingatthetopofthediagram,thefirstlevel comprises thepathways forvirusintroduction intoaregion,includingboth exogenousand endogenouspathways.Apathway canonly contributetotheprobability ofvirus introduction ifitispresent.Theextentofpresence isexpressed inpathway-units whicharetheunitsin which apathwayismeasured, e.g., ananimal,ametrictonofanimalproductsorareturning livestock truck. Atthesecond level,itisdetermined whether anypathway-units arepresent that are infected orcontaminated withvirus.Obviously,exogenous pathwaysonly constitute arisk,ifthey originatefroman areawherethediseaseisprevalent. Endogenouspathwaysonlyposea risk, iftheycontain avirusreservoir. Someexogenouspathwayscanonlycontributetothe probability ofvirus introduction, iftheyoriginate from aneighbouring area,e.g.,air currents andbirds,pets,arthropods androdents.Thesepathwaysplay aminorroleinvirus introduction,because theytransport thevirusoveronly shortdistances. Atthethird level,itisevaluated whether preventiveactionssucceed indetectingand/or inactivating thevirus.Only aselection ofpreventivemeasuresisgiveninthediagram.Most additional preventive measuresthat canbetakenbyaregion,e.g.,testingorquarantine,can beadded tothepathway diagram atthislevel. Ifthevirusisstillpresent after passing thethird level,virustransfer tosusceptibledomestic animalscanoccurbytwomainroutes:swill feeding toordirectorindirect contact with susceptible animals.Whichrouteisrelevant dependsonthepathway for virusintroduction. , Virustransfer willonly result in anoutbreak ifthevirusconveyed constitutes an infective dose.There is,however, oneexception tothisgeneral pattern:import ofaninfected live animal,legallyorillegally, will alwayslead toanoutbreak iftheanimal survivesand infection isnotdetected intime.Then swill feeding orcontactwith susceptible animals isnot required tocauseaprimary outbreak, sincetheimported animalbecomespart ofthelivestojjg population. Foreachpathway,themain eventsleading toaprimary CSFoutbreak areshown inthe pathway diagram.Theoretically, each event inthepathwaydiagram canbeassigneda probability thatitwill occur.Theseareall conditional probabilities,i.e.,theprobabilityof occurrenceofacertain event given that allpreviouseventshaveoccurred. Togivean example,virusintroduction bythepathway 'returning livestock trucks' will onlyoccurifa livestock truckreturnstotheregion after visiting aninfected region,ifthistruck is contaminated withvirus,ifitisnot disinfected properly, ifitcomesintocontact with susceptible animalsand ifthevirusdoseconveyed isatleast theminimum infective dose. 4. Qualitative assessment of the probability ofCSFV introduction into densely and sparsely populated livestock areas Thepathwaydiagram ofFigure 1 wasusedtoqualitatively assesstheprobability ofCSFV introduction forfivedensely andfivesparselypopulated livestock areas intheEU. Information aboutthepresenceofpathways(level 1 ofthepathwaydiagram)andthe possibility ofpathwayspassing virustosusceptibledomesticpigs(level4)wasusedto classify theregionsaccording theirprobability ofCSFV introduction. Basic information available for eachregion ispresented inTable 1-A. Thisinformation was used tocalculatepig and farm densityandtoderivean estimateofthenumberofpigs exported orimported bytheregion (Table 1-B).Theestimatesderived aretheminimum numbersofpigs imported orexported.Obviously, inmostcases thesewillbean underestimate ofthegrossimports andexports. Insert Tables1-Aand1-B Table 1-B indicatesthatregionswith ahighpigdensity havelargernet importsorexportsof pigsthanregionswithalowpigdensity. Südoldenburg (GE),Mantova(IT),and WestFlanders(BE)aremajor net importersofpiglets,whereas South (NL)and Côtesd'Armor (FR)aremajor net exporters.Südoldenburg (GE)hasanet importoffattening pigs,whereas South (NL),Mantova (IT),West-Flanders (BE),and Côtesd'Armor (FR)haveanetexportof fattening pigsduetoashortageofslaughtercapacity. Regionswithahighpigdensity also haveahigh pig farm density,with theexception ofMantova(IT)wherepigsareconcentrated inlargeintensive farms.Thenumberofairportswithregular flights ishighest forCôtes d'Armor (FR)whichhasfour ofthem.Thenumberoflaboratoriesworking with CSFVis highest for Hannover(GE)which hasthree,including theEUReference Laboratory forCSF. Thenumberofwildboarishighest for Ome(FR)with anestimated population of5,500 animals.NoCSFinfections inwild boarwere,however,detected inrecentyearsinthe regionslisted inTable 1-A (Laddomada,2000).Swill feeding isforbidden inallbutthe Germanregions.InGermany swill feeding isstill allowed,butonly after adequateheat treatment. Hence,ifallregulations wereobserved,noneoftheregionsofTable 1-A would suffer from primary CSFoutbreaks causedbyswill feeding. Toclassify theregions ashaving alow,moderateorhighprobability ofCSFV introduction, eightcriteriawereused (seeleft column ofTable2).Thesecriteria werebased on(i)the (extent of)presenceofpathways and(ii)pigand farm densityand swill feeding, which are indicators fortheprobability that CSFVwillreach susceptibledomestic pigsoncevirushas been introduced.Ifaregionmeetsnoneoronlyoneofthecriteria,itisdenoted ashavinga lowprobability ofCSFVintroduction. Ifaregionmeetstwoorthreeofthesecriteria, itis denoted ashavingamoderateprobability ofCSFVintroduction,and ifitmeets four ormore, itisindicated ashaving ahigh probability ofCSFVintroduction.Onthebasisofthese criteria,theregions South(NL),Südoldenburg (GE),Hannover (GE),West-Flanders(BE), and Côtesd'Armor(FR) wereclassified ashaving ahighprobability of CSFVintroduction. South-West (NL),Mantova (IT),and Ome(FR)wereconsidered tohaveamoderate probability ofCSFV introduction, whereastheremaining regions- the SPLAsofItalyand Belgium- wereconsidered tohavealowprobability ofCSFV introduction (Table2). Insert Table2 5. Computer model to assess theprobability ofCSFV introduction quantitatively Acomputermodelwasdeveloped tocalculatetheprobabilitiesof CSFVintroduction for individual regionsin theEUquantitatively andinmoredetail.Mainaimofthismodel isto analysewhichpathwayscontribute totheprobabilityof CSFVintroduction foraparticular region,theso-called target area,and from where,i.e.,whichregionsoforigin,thesepathways originate.Furthermore,themodel providesinsight intodifferences betweentarget areaswith regard totherankingofcausativepathways andregions,andtheirinteractions,andhence the need fordiversification ofpreventivemeasures. 5.1.Modellingapproach Themodel calculates theprobability ofCSFVintroduction intothedomesticpigpopulation ofatarget areabydifferent pathways(seeFig. 1).Theprobability ofvirus introductionby endogenouspathwaysdependsonlyonthesituation inthetarget area,whereastheprobability ofvirusintroduction byexogenouspathwaysdependsonconditionsintheirregionoforigin aswell asinthetarget area.Mostprobabilities calculated inthemodel arevery low,asthey arecalculated perepidemicineachregion oforiginand separately for each exogenous pathway.Therefore, thescenariopathway approachwasused asmodelling technique,andnot simulation,because itrequiresrelatively littlecomputing timeandcan easilycalculate extremely lowprobabilities (Vose, 1997).Themodelwasconstructed inMicrosoft Excel97 with theadd-inprogramme @Risk 4.0.5 (Palisade corporation). Foreachpathway inthemodel,ascenario treewasconstructed inwhichthesequenceof eventsthatultimately leadstointroduction ofCSFVintothedomesticpigpopulation ofthe targetarea isdetermined (Milleret al., 1993).Only thoseeventsthat aredecisiveinwhether aninfected pathway-unit will transmitvirustosusceptible animalswereincluded inthe scenario trees.Eachevent inthescenariotreeswasassigned aprobability that itwill occur. Thesewereallconditional probabilities.TogettheprobabilityofCSFVintroductionbya certain pathway,allprobabilities alongitsscenariotreeweremultiplied.Thescenariotrees fortheexogenouspathwayswerecalculated separately for eachregionoforigin.Combining theoutcomeofall scenariotreecalculations gaveinsightintotherelativecontributionof regions andpathwaystotheprobability ofCSFVintroduction for thetargetarea. Calculations wereperformed forfiveEUmemberstates,i.e., Germany, France,Italy,the Netherlands and Belgium.Thesearetheso-called targetareas.All EUmemberstateswere included in themodel aspossibleregionsoforigin.Nothird countries wereincluded. 5.2. Input 5.2.1.OccurrenceofCSF CSFisnotendemicinthedomesticpigpopulation oftheEU,exceptfor theisland of Sardinia,Italy(Anonymous, 1997).Thediseasesituation withregard toCSFintheEU member stateswasobtainedfromtheAnimal DiseaseNotification System (ADNS).Thisisa computerised databaseoftheEUinwhich allprimary andsecondary outbreaksofhighly contagious animaldiseasesarerecorded.ADNSdoes,however,notprovide information on ' linksbetweenoutbreaks andhencethenumberofCSFepidemicsthatoccurred inthe member statescould notdirectly bederivedfromthisdatabase.Therefore, thenumberof regionsexperiencing atleastoneprimaryoutbreak wasused toestimate theaveragenumber ofepidemicsperyear.Dataoftheperiod 1990-1999wereused,becausefrom1990onwards massvaccination against CSFwasnolongerapplied intheEU(Terpstra &DeSmit,2000). Forthis 10-yearperiod,theminimum, average andmaximum numberofregionswith primary outbreaksperyearisgiven inTable 3.Theaveragenumber wasused astheexpected numberofepidemicsperyear inthemodel calculations.Forthosemember statesthatdidnot suffer from CSFatallduring thisperiod,theexpected number ofepidemicsperyearwasset at0.095usingtheexponential distribution (Vose, 1997): Epidemicsyear= 1 - exp(-x/ß) (1) withx=timeintervalused for thecalculations (i.e.1year)and ß=minimum intervalbetween twoepidemics(i.e. 10year).This isthehighestpossiblevalue,sinceitassumesthat (i)the event ispossible and(ii)itdidoccurbefore thefirstmoment ofobservation andwill occur again immediately after the lastmomentofobservation, i.e.,in theeleventh year (whichis 2000). Insert Table3 Furthermore,thediseasesituation inthewildboarpopulationsofthetarget areaswas considered.CSFisendemicinpartofthewildboarpopulations ofGermany,FranceandItaly (Laddomada,2000).Theprevalenceofinfection variespermetapopulation ofwildboarand ismonitored by samplingprogrammes (e.g.Eibers&Dekkers,2000;Crucièreet al.,1998; Kern, 1998).Theregionsinwhich infected wildboarroam aretheBundesländerofLower Saxony,Mecklenburg-Vorpommernand Brandenburg inGermany,thedepartmentsof MoselleandBas-Rhin inFrance,andtheprovincesofNuoro,Sassari andOristano(all onthe island of Sardinia)andVareseinItaly (Laddomada,2000). 5.2.2. Pathways inthemodel Inthepathway diagram ofFigure 1 allpathwaysthatpossibly contributetotheprobabilityof CSFVintroduction wereincluded. Aselectionofthesepathwayswasmadefor inclusionin thecomputer model forvirus introduction onthebasisoftwocriteria:(i)expected importance for CSFVintroduction on thebasisofhistorical data and scientific literatureand (ii)availability ofknowledge and data toquantify theunderlying probabilities. Themajor routes forCSFVintroduction in theEUsince theprohibition ofmassvaccination werefeeding ofimproperlyheated swill,directorindirect contact with wild boarand animal movements (DeVosetal.,2000;Fritzemeier etal.,2000).Information from ADNSgave similarresults,indicating thatpurchaseofanimals,waste food feeding, and spread by fomites2 caused themajority ofprimary CSFoutbreaks for which theorigin ofdiseasewas given inADNS(DeVoset al.,2001). Therefore, theexogenouspathway importofbatch ofdomesticanimalsand the endogenous pathway wildlife wereincluded inthemodel forvirusintroduction. Furthermore,thepathway importofanimalproducts forhuman consumption wasincluded,asitisoneofthe major routesthatmight contributeto(illegal)swillfeeding. Illegal importsofanimalsandanimal productsmight evenbemoredangerous with regard totheprobability ofCSFV introduction, becausetherearenochecksonwheretheseimportscomefrom and theirdiseasestatus.These pathways were,however,not included inthemodel,asnodatawereavailableontheir numbers.Thepathwayreturning livestock truckswasincluded intothemodel,becausedata couldbederivedfromthetotal numberofanimalsexported and expertsconsidered ittobean importantriskfactor for theriskfor CSFVintroduction intotheNetherlands (Horst etal., 1998).Although CSFVcanbetransmitted potentially overlongdistancesbydistribution of virus-contaminated semen (DeSmit,2000),thepathway geneticmaterial wasnot included in themodelduetolackofdata.All otherpathwaysdisplayed inthepathway diagram werenot included intothemodel astheywereconsidered tocontributeonlymarginally tothe probability ofCSFVintroduction andhardly anydatawereavailable for quantification. 2 Spreadby fomites isall virus spreadcaused byobjects that arecontaminated with thedisease agentandhence covers those indirect contactsbetween animals thatare notincluded inothertransmission routesdistinguished in ADNS (Laddomada,personal communication). Italso includes spreadby wild boar(Pittleret al., 1995). 12 5.2.3. Occurrenceofpathways InTable4,dataontheoccurrenceofpathways included inthemodel aregivenfor the five target areas.Imports from and exportstoallEUmemberstatesareshown.Inthemodel,these numbersweresubdivided according theirregionsoforigin.Thenumbersofpigsimported and exported weredivided bytheaveragebatch sizetoget estimatesofthenumberofbatches ofdomesticanimalsimported andreturning livestock trucks,respectively. Insert Table4 Germany isamajor importer ofpigs,whereastheNetherlands isamajor exporter.Italy hardly exportsanypigs.Withregard toporkproducts,itcanbeseenthatmosttradeisin fresh andchilled products.The wildboarpopulationsofGermanyandFrancearemuch biggerthanthoseoftheNetherlands and Belgium. ForItalynodataonthenumberofwild boarin thecountrywereavailable.In Germany, wildboararesprea"dalloverthecountry, whereasintheothercountries theyaremoreregionallydistributed (Anonymous, 1999). Thepathways weredivided into subgroups toperform themodel calculations.Theimports and exportsofpigsweresubdivided intothreecategories,i.e.,piglets,breedingpigsand fattening pigs.Itwasassumed that therisksofCSFVintroduction mightdiffer between these categories.Fattening pigsarebrought toslaughterhouses, whereaspigletsandbreeding pigs will becomepartofthepigpopulation in thetargetarea.Furthermore,itwasassumed that morepreventivemeasureswillbeinusefor theimportofbreeding pigsthanpiglets.The import ofporkproductswassubdivided intofour categories,i.e.,fresh orchilled, frozen, non-heat treated,andheat treated pork products,because theprobability that CSFVwill survivematuringandprocessing differs betweenthosecategories.CSFVcansurviveupto 4,5yearsinfrozen pork products (FarezandMorley, 1997)anduptoseveralmonthsin fW, orchilled porkproducts(Terpstra, 1991; 1986).Innon-heat treatedporkproducts,i.e.,saltad. inbrine,smoked ordried, CSFV cansurviveuptothreemonths(Terpstra, 1986).Inheattreatedproducts,CSFVwill not surviveifthecombination oftemperatureanddurationof heattreatment was sufficient. 5.3. Calculations Themodel calculationsweresubdivided intothreemaingroups,i.e.,regionsoforigin, exogenouspathways,and endogenouspathways. 5.3.1.Regionsoforigin Thisgroupcontainsall calculations ofparameters that arerelated totheCSFsituationinthe; regionsoforigin,e.g., theannual numberofCSFepidemics,thelength oftheseepidemics, thelength oftheirhighriskperiod (HRP),i.e.,theperiod from first infection withthevirusjp first detection (Horstet al., 1998),thecumulative incidenceofdiseaseintheregion,etc. The valuesoftheseparameters willvary from yeartoyearandepidemictoepidemicdueto inherentrandomness.Averagevalueswereused inthecurrentmodel calculations. 5.3.2. Exogenouspathways Thisgroup containsallcalculations for theprobability ofCSFVintroductionbyexogenous pathways.Thecalculation ofthisprobability isperformed infour steps.Inthefirststep,the probability that asinglepathway-unit will causeCSFV introduction intothetarget areais calculated,given that CSFVispresent intheregionoforigin.Thescenariotreesdescribedin gection5.1. areusedinthisstep.Pointofdeparture in thescenariotreesisthat CSFVis presentintheregionoforigin.Thethreemaineventsthat canbedistinguished inallscenario treesare: isthepathway-unit infected orcontaminated withCSFV? istheinfection orcontamination detectedoreliminated bypreventive measures? doesthe infected orcontaminated pathway-unit comeintocontact with susceptiblepigsin thetargetarea andtransmit an infective viral dose? Inthesecond step,theprobability that CSFVisintroduced byaparticularpathwayis calculated,combining theprobability ofstep 1 withthetotal numberofpathway-unitsgoing from theregionoforigin tothetarget areaduring theperiod thatCSFV ispresent inthe regionoforigin inabinomial distribution (Vose,2000).Steps 1 and2arerepeated for all exogenouspathways inthemodel.Inthethird step,theprobabilities ofstep2aresummedto obtain theoverall probability ofCSFVintroduction intothetargetareaduringanepidemicin theregionoforigin.Inthefourth step,theprobability ofstep3ismultiplied withthe expected numberofCSFepidemicsintheregion oforigin toobtaintheannualprobability of CSFVintroduction intothetarget areafrom thisparticularregion oforigin.Thiswhole procedureisrepeated for allregionsoforigin inthemodel.Aschematicrepresentation ofthe stepscanbefound inFigure2. InsertFigure2 Thecalculations for thepathway 'import ofbatch ofpiglets'aredescribed inmoredetail to illustratethemodelcalculations inthefirst twosteps.Thescenariotreeused instep 1 isgiven inFigure3.Thefirstevent inthisscenariotreeiswhetherthebatchofanimals transported from theregion oforigin tothetarget areacontainsinfected piglets.Thisprobability wasset 15 equal tothecumulative incidence (CI)ofdiseaseintheregion oforigin.Thesecondevent is whether theinfected animalswill bedetected bypreventive measures taken intheregionof • origin, such asclinical inspection ofpigsbefore issuing ahealth certificate.Thisprobability wasset equal tothesensitivity ofpreventivemeasures(Sereg).Thethird event iswhetherthe infected animalswillbedetected by preventivemeasurestakeninthetarget area.This probability wasalsoset equal tothesensitivity ofpreventivemeasures (Sela).These sensitivities depend on thekind ofpreventivemeasurestaken and thequalityofveterinary , services.Theirvaluescanbechanged inordertocalculate theeffect ofmoreor different preventive measures.Inthis scenario tree,noeventwasincluded fortheprobabilityofr. contact with susceptiblepigsand transmission ofaninfective viral dose,because importofan infected pigletwillalwayslead to aprimary outbreakiftheinfection isnotdetectedtimely and thepiglet isbrought ontoafarm inthetarget area.Theprobability that abatch ofpiglets causes CSFVintroduction ifcoming from aregion whereCSFV ispresent ishencecalculated by: Pbatchofpigie,s= CI * ( 1 - S e r e g ) * ( 1 - S e u ) '(2) Insert Figure3 Inthesecond steptheprobability that thepathway importofbatch ofpigletscausesCSFV introduction iscalculated by: "pathway batchofpiglets ~ » ~ ( ' ~""batchofpiglets ) with n=number ofbatchestransported during theperiod that CSFV ispresent. *• ' Theprobabilities used in the scenariotreemightbedifferent fortheHR? and theremainder oftheepidemic(PostHRP).Therefore Pbachofp.giœwillhavedifferent values for theHRPand PostHRPand thecalculationshavetobeperformed separately for eachperiod andthen i added. 5.3.3. Endogenouspathways Thisgroupcontains all calculations for theprobability ofCSFVintroduction by endogenous pathways.Onlythepathwaysdirect and indirectcontact with wild boarwereincluded inthe model(asexplained before). Thethreemain events inthescenario treesfor thesepathways are: - isanindividual wildboar infected with CSFV? - doesthiswildboar comeintocontact withsusceptible domestic pigs? - isaninfective viral dosetransmitted tothosepigs? Toillustratethemodelcalculations for endogenous pathways,thescenariotree for the pathwaydirect contact with wildboar isgivenin Figure4.Theprobability ofinfection (INF) wasestimated using resultsfromserological surveys (e.g.Eibers and Dekkers,2000;Kern, 1998).Theprobability ofdirect contact with susceptiblepigs(CSAdlrcct)wasestimated using information onthegeographical distribution ofwild boarand domesticpigproduction, whetherwildboararerestricted intheirmovementsby,for example, fences,andthe proportion ofpig farms withoutdoor facilities.Theprobability oftransmitting an infective doseifdirect contact occurs(IDdino)wasassumed toberatherhigh and therefore setat0.9. Theannual probability of CSFV introduction bydirect contact with asinglewildboaris hencecalculated by: 17 Pboar=INF *CSAd,rœ,*IDdlrecI (4) Insert Figure4 Toget theannual probability ofCSFV introduction by thispathway, thisprobability isthen combined with thetotal numberofwild boar inthetarget areausing abinomial distribution (Vose,2000): ^pathway boar — I ~ ( I — rboar / withn=total numberofwildboar inthetarget area. 5.4. Results Theannua!probabilities ofCSFV introduction for thetarget areas aregiven inTable5. Results aregiven for the exogenouspathways and separately for theendogenous pathways .. direct and indirect contact with wildboar.Theannual probability ofCSFVintroduction by . exogenouspathways ishighest for Germany with0.2.Germany can thus expectCSFV introduction byexogenous pathwaysonaverageonceperfiveyears.For theothertarget areastheannualprobability ofCSFV introduction byexogenous pathways isabout0.05. They can thusexpect CSFV introduction byexogenous pathwayson averageoncepertwenty years.Theannual probability of CSFV introduction duetodirect orindirect contact withwild boariszero for theNetherlands and Belgium,asnoinfected wildboarpopulationsare present.Theprobability that Germany and Italy will experience CSFV introduction dueto \P) eitherdirect orindirect contact with wildboar isveryhigh.This indicatesthatitmight happen almost everyyearoreven several timesperyear. Insert Table 5 5.4.1.Moredetailedresultsfor exogenouspathways Theannual probability ofCSFV introduction byexogenouspathways isdetermined bythe regionsoforigin (all EUmemberstates except thetarget area)and theexogenouspathways included inthemodel.Figure5gives insight intothemain regionsoforigin contributingto thisprobability for thetarget areas.Intheleft column agraph isdisplayed for each target area inwhichtheannual probability ofCSFVintroduction isshownperregion oforigin. Inthe right column alsoagraph isdisplayed for each target area,but inthesegraphstheprobability ofCSFVintroduction during aCSFepidemic intheregion oforigin isshown.Togivean example,theannual probability that CSFVisintroduced from theNetherlands into Germany isalmost 0.1 (left uppergraph),whereastheprobability thatCSFVisintroduced into Germany duringanepidemic intheNetherlands is0.16(right uppergraph).Theprobability perepidemic ishigherthantheprobability peryear,astheexpected numberofepidemicsin theNetherlands islessthan 1 (Table3). InsertFigureS Ingeneral,theNetherlands,Belgium and Germanycontributemosttotheannual probability ofCSFVintroduction into thetarget areas(left column ofgraphs).Theprobability that CSFV isintroduced intooneof thetarget areasduring asingle epidemic inGermany is,however, low ifcompared with epidemicsin theNetherlands and Belgium (right column ofgraphs). 19 Therelatively high annual probability that CSFVis introducedfromGermany isthusmainly due tothehigh averagenumberofepidemics peryearinthisregion oforigin. Furthermore,it canbeconcluded that aCSFepidemic inLuxembourg constitutes aconsiderableriskof CSFVintroduction for Belgium and France, whereas anepidemic in Denmark constitutesa considerableriskofCSFVintroduction for Germany,FranceandItaly. Figure6gives anoverview oftherelativecontribution ofexogenouspathwaystotheannual probability ofCSFV introduction by exogenouspathwaysinto thetargetareas.Ingeneral,the pathwayscontributing most are importofpiglets,returning livestock trucks andimportof fresh orchilled pork products.Theirrelativecontribution differs pertargetareaand depends bothon thenumberofpathway-units present (seeTable4)and theprobability ofCSFV introduction perpathway-unit. Thelatter is,for example,quitehigh for pork products imported intoGermany incomparison totheother target areas,asthisistheonlytarget area whereswill feeding isstill alluwed(seeTable 1-A).Thisexplainstherelativelyhigh contribution ofthispathway totheannual probability ofCSFV introduction intoGermany. On thebasisof thenumberofpathway-units present only,onewould expect amuchbigger contribution oftheimportofpigletsand fattening pigs.ForItaly,therelativecontributionof imported pork productstotheprobability ofCSFV introduction isalsohigh,which ismainly duetothe largeamountof fresh and chilled porl products imported.FortheNetherlands, returning livestock truckscontributemost totheannualprobability ofCSFV introduction. This isduetothehugenumber ofpigsexported.Belgium alsoexportsquitealotofpigs. Returning livestock trucksdo,however, notcontributemuch tothe annua!probabilityof CSFVintroduction, asin Belgium almost 60%ofthefarmers disinfects incoming vehicles reducing theprobability ofCSFV introduction perpathway-unit -, whereas inallother member states,thispercentagevaried between 0and 12.5% (data obtained froma 20 questionnaire conducted during theEUResearch Project FAIR5-PL97-3S66,refer to Italian paper).ForBelgium theimport ofpigletsisthemost importantpathway which isexplained by thelargenumberofpiglets imported. ForFrance,bothimportofpigletsandpork products contributed alot totheannual probability of CSFV introduction. Insert Figure6 5.5.Sensitivityanalysis Sensitivity analysiscan indicate theimpact ofinputparameters onmodel outcome.Forthis purpose,thevaluesofsomeimportant inputparameters werechangedbyincreasing and reducing them with afactor 2.Theparameters changed were(1)theaveragelength ofthe HRPper epidemic intheregionsoforigin,(2)the numberof lifepigsimported,(3)the numberoflifepigsexported andhencethenumber ofreturning livestock trucks,and (4)the probability that swill feeding ispractised, which affects boththeprobability ofCSFV introduction byimportofporkproducts andby indirect contactswith wild boar. Furthermore, theprobability ofCSFV introduction for thehypothetical situation thatnoswill isfed atall wascalculated. Resultsofthe sensitivity analysis for theprobability ofCSFVintroductionby exogenouspathways arepresented inTable6and for theendogenous pathway indirect contactwithwildboarinTable 7.The sensitivity analysisperformed did not changethe probability ofCSFV introduction by direct contact withwildboar. Insert Table 6 Table6givestherelativechangeoftheprobability ofCSFVintroduction byexogenous pathways.Theeffects ofchanging the abovementioned parameters differ largelybetween l target areas.Theimpact ofchanging the averagelength oftheHRPintheregionsoforigin wasbiggest for theNetherlands and Belgium.Theimpact of changing thenumberof imported life pigs wasbiggest for Belgium, whereas theimpactofchanging thenumberof exported life pigswasbiggest for theNetherlands.Thisisinaccordance withtheresults presented in Figure6,showing that importofpigscontributes mosttotheprobabilityof CSFV introduction intoBelgium and returning livestock truckstotheprobability ofCSFV. introduction intotheNetherlands.Theimpactofchanging theprobability ofswill feeding washigh for Germany, Franceand Italy. Ingeneral,theprobability ofCSFV introduction byexogenous pathways increased whenthe valuesoftheinputparameters weremultiplied by2,and diminished when thesevalueswere multiplied by0.5.Theonlyexceptions arethe changesinprobability ofCSFV introduction for Germany and Italywhen reducing theaveragelength oftheHRP intheregionsoforigin. Thisisdue tothefact that thetotal length oftheepidemic wasnot changed inthiscalculation and that therefore theprobability ofCSFV introduction by import ofporkproductsduringthe PostHRP wasincreased asthePostHRP lasted longer. Import ofpork productswasthemajor pathway contributing totheprobability ofCSFV introduction for thesetarget areas(Fig.6). Insert Table 7 Table 7givestherelative changeoftheprobability ofCSFV introduction byindirectcontact with wildboar whenchanging theprobability that swill feeding ispractised.This increasesor reduces theprobability of indirect contact per individual wildboar.Theimpactofchanging 22 thisinput parameter waslargest for FranceandItaly.ForGermany italsohad impact,butdue tothe largewild boarpopulation present inthiscountry,theannual probability ofCSFV introduction duetoindirect contact with wildboarwaschanged toalesser extent. Indirect contactwith wildboardidnot constitute ariskofCSFVintroduction for thedomesticpig population intheNetherlands and Belgium andhencechanging theprobability ofswill feeding had noimpact onmodel outcome for thesetargetareas. 6. Discussion and conclusion 6.1. Pathwaydiagram Insight intoall factors contributing totheriskofCSFV introduction isneeded todecideon preventiveactionsthat arecost-effective, i.e.,achieveconsiderableriskreductionat reasonable costs.Inthisstudyboth aqualitative andquantitative approach wereusedto estimatetheprobability of CSFV introduction for several regions inthe EUandtoevaluate which factors contribute most totheprobability ofCSFVintroduction. Both approaches were based onapathway diagram especially constructed for CSFVintroduction intoregionsofthe EUunderanon-vaccination policy. Constructing such apathway diagram providesmoreinsight intoallpossiblepathwaysand events contributing totheoccurrence ofanadverseevent.Therefore itisrecommended as partofhazard identification, which isthefirststepinriskanalysis(Wooldridge et al., 1996). Themorepathways involved,themorecomplex thediagrambecomes.Therefore onlythe main events leading totheoccurrenceofaprimary CSFoutbreak were involved inthe 23 pathwaydiagram ofFigure 1.Scenario treeswereusedtodescribe eachpathway inmore detail. 6.2.Qualitative assessment oftheprobabilityo/CSFV introduction intodenselyandsparsely populatedlivestockareas Thequalitative assessment oftheprobability ofCSFVintroduction into DPLAs and SPLAs demonstrated that DPLAsgenerally had ahigherprobability ofCSFVintroduction than SPLAs,although thiscould notbeattributed topigdensity only.Theresults shouldbe interpreted with care,because not for allpathways in thediagram information ontheir presence wasavailable.Furthermore, thecontribution ofpathwaystotheoverall probability of CSFV introduction will differ, but itwasnotknown towhat extent..Somepathwaysplaya moreimportant roleinintroducing CSFVtoaregionthan others.ThecriteriaofTable2 were,however,given equal weight inthequalitativeassessment.Therelativecontributionof thepathwayswill alsodiffer perregion,becauseitdependson thenumberofpathway-units present,theregion oforigin ofthepathway-units and theiruse.Horst etal.(1998)obtained expert estimatesoftherelative importance ofpathways for CSFVintroduction.Thesewere, however, specifically for CSFV introduction into'heNetherlands and could therefore notbe used for different European regions.Toestimatetheprobability ofCSFV introduction for regionsinthe EUmore adequately,quantitative information on thepresence ofpathwaysand theirmain events isthus necessary. 24 6.3. Computer model to assess theprobability of CSFV introduction quantitatively The computer model used to estimate the probability of CSFV introduction quantitatively gave much more detailed insight into the factors determining this probability for a target area. It shows which regions of origin and which exogenous pathways contribute most to the annual probability of CSFV introduction by exogenous pathways and gives estimates of the probabilities of CSFV introduction by wild boar separately. This information is very useful for policy-makers as it helps to set priorities for preventive measures. The impact of preventive measures can be estimated by the computer model as well by changing relevant input parameters. 6.3.1. Modelling approach The computer model developed has a clear structure. Calculations were subdivided into three main groups, which makes it easy to add or change input parameters. For each target area results can be obtained as aggregates (probability of CSFV introduction per year), but also separately for each combination of exogenous pathways and regions of origin. Not all pathways that might contribute to theprobability of CSFV introduction could be included in the model. The model structure is,however, such that additional pathways can easily be included ifnew information and data would become available. Some of the pathways that were not included are considered to contribute only marginally to the probability of CSFV introduction, e.g., import of manure, birds, pets, arthropods and rodents, air currents and laboratories. Others might contribute considerably to the probability of CSFV introduction, but no quantification was possible. This was the case for, e.g., import of genetic material, illegal imports, tourists and professional people. To get a more comprehensive 25 overview ofthemajor causingpathways oftheprobability ofCSFV introduction intothe target areas,theircontribution totheprobability of CSFV introduction maybeestimated qualitatively. Nothird countries wereincluded in themodel asregionsoforigin.Therefore their contribution totheprobability ofCSFVintroduction for thetarget areascould notbe calculated. Inordertoincludethird countries intothemodel without anexponential increase in thenumberofcalculations,third countries could best beclustered taking intoaccount their geographical situation and disease status withregard toCSF(e.g.endemic, free with vaccination, free without vaccination). Formost regionsoforigin,i.e.,EUmember states,theexpected number ofCSFepidemics peryear islessthan one.For Germany and Italy,however, theaveragenumberofepidemics peryearoverthelast ten years exceedsone(Table3).Simply multiplying theprobability per epidemic with theexpected number ofepidemics instep4ofthecalculations for exogenous pathways might inthiscaselead toanoverestimateoftheannualprobability ifthetotal epidemic period (which alsodependson theaveragelengthoftheepidemics) exceedsayear. Therefore the annual probabilities ofCSFViitroduction intothetarget areas from these regionsoforiginsmighthavebeen overestimated bythemodel.Thecontribution ofItalyto the annual probability ofCSFV introduction for thetarget areas wasratnersmall,but the contribution ofGermany wasquitelarge,mainly duetothehigh averagenumberof epidemicsperyear(see Fig.5). The target areas for which model calculations wereperformed were EUmember states,and not densely and sparsely populated livestock regions.Calculating theprobability ofCSFV 26 introduction for regions in theEUappeared tobeimpossibledueto lackofdata.Itwastried toderivedata on importsand exportsoflife animalsperregionfromtheanimal movement system (ANIVIO),but thesedata werenot readily available.Furthermore,dataontheanimal movements within the memberstates areneeded toperform model calculations atregional level.For thispurpose,national identification and recording systems(I&R)shouldbeused, but data from such systemswerenot available inall EUmember states.Estimatingthe probabilities for the scenario trees isalsomuchmoredifficult atregional level. 6.3.2. Results Theannual probabilities ofCSFV introduction by exogenouspathwayswerequitelowif compared withexpert estimates (Horst etal., 1998).Alsocomparing themodelresultswith theaveragenumberofregionswith primary outbreaks inTable3,suggeststhat themodel underestimates theprobability ofCSFV introduction by exogenouspathways,although for Germany, France,and Italy wild boarmight alsocontribute considerably totheprobability of CSFV introduction.Theprobability ofCSFVintroduction bydirect and indirect contactwith wild boar wereindeed quitehigh,especially for Germany and Italy.This isinaccordance withreality,asmanyprimary outbreaks inthesecountriescanbeattributed todirector indirect contactswith wildboar.Fritzemeier etal.(2000)concluded that 59%ofallprimary outbreaks inGermany in theperiod 1993-1998,which wereabout 90,wereduetodirector indirect contactswith infected wild boar.And manyoftheCSFoutbreaks inItalyreported to ADNSwereintheinfected regionsofSicilia. Mostprobably themodel underestimates theprobability ofCSFV introduction byexogenous pathways.Itshould bekept inmind thatnot allpathwayscontributing totheprobabilityof CSFV introduction were included in themodel and neitherthird countries.Absolutevaluesof 27 model outcomeshould therefore notbeconsidered astruevaluesfortheprobability ofCSFV introduction.Themodel can,though,beused torank thetargetareas for theirprobabilityof CSFV introduction.Theultimate aimofthemodel wasnot togiveexact estimatesofthe probability ofCSFV introduction,buttogainmore insight intothepathways andregionsof origin contributing most tothisprobability.Theprobability ofCSFV introductionby exogenouspathways was,for example,much higher for Germany than for theothertarget areas.And although Germany isimporting millionsof life pigs,thispathwaycontributed less totheprobability ofCSFVintroduction than theimport ofpork products (Fig.6).Analysing themodel output explainsthat this isduetothefact that swill feeding isstill allowed in Germany. Thecontribution ofimport ofpork productstotheannualprobability ofCSFVintroduction for thetarget areas wasratherhigh,especially for Germany and Italy(Fig.6).This isnotin accordance withhistorical dataonthemost important sourcesof CSFV introduction (seee.g. DeVosetal.,2000;Fritzemeier et al.,2000).Itmightbethat someofthe underlying probabilitiesused tocalculatetheprobability ofCSFV introduction bypork productswere estimatedtoohigh. Import ofinfected porkproductscan,however, easily leadtoCSFV introduction if fed asswill.Thiswas,for example, seen intheCSFepidemics intheUnited Kingdom in 1986and2000(Sharpe et al.,2000;Williams&Matthews, 1988). 6.3.3. Sensitivity-analysis Sensitivity analysiswasperformed toget insight intothe impactofinputparameterson model results.Only someofthemost important inputparameters werescreened. Further sensitivity analysis canbeperformed toestimatethe impact ofpreventive measuresonthe probability ofCSFV introduction for thetarget areas.Figures 5and 6alreadygiveagood 28 indication of theregions oforigins and exogenouspathwaysatwhichpreventive measures should bedirected.Themodel providesalldata tomakemoredetailedpicturesofthe exogenouspathwayscontributing most to theprobability ofCSFV introduction perregionof origin.Prevention ofCSFVintroduction can eitherbereached bychanging (i)theCSF situation intheregionsoforigin,(e.g.the averagenumberofepidemicsperyear, thelength of theHRP,thenumberofinfected premises),(ii)thenumberofpathway-units coming from certain regionsoforigin,or(iii) thepreventive measurestakentoreducetheprobability of CSFVintroduction perpathway-unit. Themodel can provide information onwhich typeof measures willbemost effective. Inordertocalculate thecosts and benefits ofsuchmeasures, alsotheir impacton thespread of CSFVonceintroduced intoatarget area should betaken intoaccount. Forthispurpose simulation modelscanbeused (e.g.Jalvingh etal., 1999; Saatkampetal., 1996). References Anonymous(1997).- Report on AnnualMeeting ofNational Swine FeverLaboratories. Vienna, Austria, 16-17June 1997.European Commission, doc.VI/7888/97. Anonymous (1998).- Theoutbreak ofclassical swinefever intheNetherlands.Final evaluation (Deuitbraak vanklassiekevarkenspest. Eindevaluatie). ReportfromtheMinistry ofAgriculture,NatureManagement and Fisheries,95pp.(in Dutch) Anonymous (1999).- Classical swinefever inwildboar. Scientific CommitteeonAnimal Health and Animal Welfare. Adopted 10August 1999.European Commission, Brussels. Report XXIV/B3/R09/1999,46pp. Anonymous(2000).- Development ofpreventionandcontrol strategiestoaddressanimal health andrelated problems indenselypopulated livestock aieasintheCommunity.Periodic Progress Report II.April 1999- April 2000.EUResearch Project FAIR5-PL97-3566. CEC(Commission oftheEuropean Communities) (1964).- Council Directive 64/432/EEC on Community measures for intracommunitytrr leincattleandpigs.Off.J. Eur. Communities, 121(July29, 1964),1977-2003. CEC(Commission oftheEuropean Communities) (1980).- Council Directive80/217/EEC (asamendedby Council Directive91/685/EEC)on Community measures for thecontrolof classical swine fever. Off.J.Eur.Communities, L47(February 21,1980), 11-23. 30 CEC(Commission ofthe European Communities) (1982).- CouncilDirective 82/894/EEC onnotification of animal diseases in theCommunity. Off.J.Eur.Communities, L378 (December 31, 1982),58-62. Crucière,C, Burger, C.&Gonzague,M.(1998).- Laboratory investigationsofthe 'Massif Vosgien' CSFwildboaroutbreak. In:Proc.Meeting 'Measures tocontrol classical swine fever inthe European wild boar', Perugia, Italy6-7 April 1998.European Commission,doc. VI/7196/98,93-97. DeSmit,A.J.(2000).- Laboratory diagnosis,epizootiology, and efficacy ofmarker vaccines inclassical swinefever: areview. Vet. 0., 22,182-188. DeVos,C.J., Horst,H.S.&Dijkhuizen, A.A.(2000).- Riskofanimalmovements for the introduction ofcontagious animal diseasesintodenselypopulated livestock areasofthe European Union.InProc.of theSociety for Veterinary Epidemiology and Preventive Medicine(M.V.Thrusfield &E.A.Goodall,eds).Edinburgh, 29-31March 2000. 124-136. DeVos,C.J., Saatkamp,H.W.,Huirne,R.B.M.&Dijkhuizen, A.A.(2001).- Riskof classical swinefever virus introduction atregional level intheEuropeanUnion:aconceptual framework. Rev.sei.tech.Off.Int. Epiz.,submitted. Dijkhuizen, A.A.&Davies,G.(eds) (1995).- Animal health andrelatedproblemsindensely populated livestock areasoftheCommunity. Proceedingsofaworkshopheld inBrussels,2223November 1994.Report EUR 16609EN.Office for Official Publications oftheEC, Luxembourg, 216pp. Edwards, S., Fukusho, A., Lefèvre, P.C., Lipowski, A., Pejsak, Z., Roehe, P. & Westergaard J. (2000). - Classical swine fever: the global situation. Vet. microbial., 73, 103-119. Eibers, A.R.W. & Dekkers, L.J.M. (2000). - Sero-surveillance of diseases in wild boar of interest to domestic livestock in the Netherlands 1996-2000. In: Proc. Annual meeting of the Dutch Society for Veterinary Epidemiology and Economics (A. Van Nes, ed). Utrecht, the Netherlands, 13 December 2000, 49-54. Farez, S. & Morley, R.S. (1997). - Potential animal health hazards of pork and pork products. Revue Scientific et Technique, Office Internationa! des Epizooties, 16: 65-78. Fritzemeier, J., Teuffert, J., Greiser-Wilke, I., Staubach, Ch., Schlüter, H. & Moennig, V. (2000). - Epidemiology of classical swine fever in Germany in the 1990s. Vet. microbiol., 77, 29-41. Greiser-Wilke, I., Fritzemeier, J., Koenen, F., Vanderhallen, H., Rutili, D., De Mia, G.-M., Romero, L., Rosell, R., Sanchez-Vizcaino, J.M & San Gabriel, A. (2000).- Molecular epidemiology of a large classical swine fever epidemic in the European Union in 1997-1998. Vet. microbiol., 77, 17-27. Horst, H.S., Dijkhuizen, A.A., Huime, R.B.M. & De Leeuw, P.W. (1998). - Introduction of contagious animal diseases into The Netherlands: elicitation of expert opinions. Livest. prod, sei., 53, 253-264. Huirne,R.B.M.,Hopman,J.K.K.,Stelwagen, J.&Dijkhuizen, A.A.(1995).- Spatial distributionsofdensely populated livestock areas:historical development and reflections from economic theory.In Proc.Workshoponanimal health and related problems indensely populated livestock areasoftheCommunity (A.A.Dijkhuizen &G.Davies,eds).Brussels, 22-23November 1994.Office for Official Publications of theEC,Luxembourg, 19-31. Jalvingh,A.W.,Nielen,M.,Maurice,H.,Stegeman,A.J.,Eibers,A.R.W. &Dijkhuizen, A.A. (1999).- Spatial andstochastic simulation toevaluatetheimpactofeventsandcontrol measuresonthe 1997-1998classical swine fever epidemicinTheNetherlands.I.Description ofsimulation model.Prev.vet. Med.,42,271-295. Kern,B.(1998).- Experiences with CSF-vaccination ofwildboarin Brandenburg.In:Proc. Meeting 'Measures tocontrol classical swinefever intheEuropean wildboar', Perugia,Italy 6-7 April 1998.European Commission,doc.VI/7196/98, 122-127. Koenen, F.,Van Caenegem,G., Vermeersch, J.P.,Vandenheede,J. &Deluyker, H.(1996). Epidemiological characteristics of an outbreak of classical swine fever in an area ofhigh pig density. Vet. Rec, 139,367-371. Kramer,M.,Ahl,R.,Teuffert, J., Kroschewski, K., Schlüter, H.&Otte,J.(1995).- Classical SwineFeverinGermany-someepidemiological aspects.InProc.oftheSociety for Veterinary Epidemiology and Preventive Medicine(E.A.Goodall,ed).Reading,29-31 March 1995.110-118. 33 Laddomada, A.(2000).- Incidence and control ofCSFinwildboar inEurope. Vet. microbiol, 73,121-130. Laevens,H.,Deluyker,H., Koenen,F.,Van Caenegem,G.,Vermeersch,J.P.&DeKruif,A. (1998).- Anexperimental infection with aclassical swine fever virusinweanerpigs.II.The useof serological data toestimatethedayofvirus introduction innatural outbreaks. Vet. Q., 20,46-49. Meuwissen,M.P.M.,Horst, H.S.,Huime,R.B.M.&Dijkhuizen, A.A.(1999).- Amodelto estimatethefinancialconsequences of classical swine fever outbreaks:principlesand outcomes.Prev.vet. Med.,42,249-270. Michel,I.&DeVos,C.J.(2000).- Denselypopulated livestock areas inthe European Union:definition and location.InProc.X"1 International CongressonAnimal Hygiene (M.J.M.Tielen&M.Th.Voets, eds).Maastricht,2-6July2000.332-337. Miller, L.,McElvaine,M.D.,McDowell,R.M.&Ahl,A.S.(1993).- Developinga quantitativeriskassessment process.InRisk analysis,animalhealth and trade(R.S.Morley, ed).Rev. sei.tech.Off.int. Epiz., 12(4),1153-1164. Pittler,H., Fiedler,J.,Jentsch, D.,Hasselbach, P.&Kramer, M.(1995).- Theproblemsand consequences oftheswine fever epidemic during 1993/94inGermany (Der SchweinepestSeuchenzug 1993/94,Probleme und Konsequenzen). Tierarzt!. Umschau, 50,522-530.(in German) 34 Saatkamp, H.W., Huirne,R.B.M.,Geers,R., Dijkhuizen, A.A.,Noordhuizen,J.P.T M & Goedseels,V.(1996).- State-transition modelling ofclassical swinefever toevaluate national identification andrecording systems- general aspects andmodel description.Agric. Syst.. 51,215-236. Sharpe,K.,Gibbens,J., Morris,H.&Drew,T.(2001).- Epidemiology ofthe2000CSF outbreak in EastAnglia:preliminaryfindings.Vet. Rec, 148, 91. Terpstra, C.(1986).- Hetvoeren vankeukenafvallen alsoorzaak vanvarkenspest. Tijdschr. Diergeneeskd., Ill, 254.(in Dutch) Terpstra, C.(1991).- Hogcholera:anupdateofpresentknowledge.Br. Vet. J., 147,397406. Terpstra,C.&DeSmit,A.J.(2000).- The 1997/98epizooticofswinefever inthe Netherlands:control strategiesunderanon-vaccination policy. Vet. microbioi, 77,3-15. ' Vanthemsche,P.(1996).- Classical SwineFever 1993-1994Belgium.PigJournal, 37, 4353. Vose,D.J.(1997).- Risk analysisinrelationtotheimportation and exportation ofanimal products.Rev.sei.tech.Off. Int. Epiz., 16,17-29. Vose,D.(2000).- Risk analysis.Aquantitativeguide.Second edition.John Wiley &Sons, Ltd., Chichester,418pp. Williams, D.R. &Matthews,D.(1988).- Outbreaks ofclassical swine fever in Great Britain in 1986. Vet. Rec. 122, 479-483. Wooldridge,M, Clifton-Hadley, R.&Richards,M.(1996).- "I don't want tobetold whatto doby amathematical formula"- Overcoming adverseperceptions ofrisk analysis.InProc.of theSociety for Veterinary Epidemiology andPreventive Medicine(M.V.Thrusfield &E.A. Goodall,eds).Glasgow,27-29 March 1996.36-47. 36 < a. p-- — — rr, t< « **• y y ~ 8 |> v a if - s. -= -5 4* r- ï E S S 8- £ ^ 't Z P "8 e % 1 'i — »n <£ o © — \=: .2 — s *= u \ s s s < J c Table3 The minimum, average and maximum number of regions peryear in eachcountry that experienced primary CSF outbreaks for the period 1990-19«0 (source: ADNS) Countries Minimum Average Maximum Germany 2 10.2 25 France 0 0.5 2 Italy 0 2.8 6 The Netherlands 0 0.6 4 Belgium 0 0.4 ! Luxembourg 0 0 0 United Kingdom 0 0 0 Ireland 0 0 0 Denmark 0 0 0 Greece 0 0 0 Spain 0 0.7 5 Portugal 0 0 0 Austria 0 0.2 I Finland 0 0 0 Sweden 0 0 0 Table4 Quantification orpathwaysincluded inthemodelfor virusintroduction The Germany France Italy Netherlands Belgium 1,416,941 283,049 306,213 31,735 702,767 1 Import ofpigs(numbers} Piglets Breedingpigs 132,633 14,805 11,753 16,319 61,418 Fatteningpigs 1,261,402 164,555 715,212 390,452 68,494 684,326 177,581 1,773 1,633,211 84,531 Breedingpigs 5,484 64,456 18,545 107,770 11,923 Fatteningpigs 371,897 72,106 7,749 1,427,590 1,012,768 Fresh/chilled 725,832 296,606 633,062 68,997 83,836 Frozen 114,365 97,759 73,167 40,759 19,038 Non-heattreated 32,380 60,150 8,778 10,058 25,435 Heattreated 42,109 25,548 12,721 31,032 33,662 W l d boar (head)" 600,000 450,000 n.a.' 3,600 10,000 Exportofpigs(numbers)* Piglets Importofpork products(metric tons)* 1 Imports from andexportstoall EUmemberstatesin 1999.Source:EL'ROSTATCOMEXTdatabase b Sources: Eibersand Dekkers(2000),Anonymous(1999),andAnonymous(1997) c Nodataavailable * ï vri — Fig. 1 Pathway diagram containing allpathways that contribute totheprobability of CSFV introduction in the European Union Fig. 2 Schematic representation of model calculations for exogenous pathways Steps 1 & 2 for region of origin A H « f trwtad pork Non*««truud product! y- PloMti pork producM - BrMflirtg pigi !""**l* Fig.3 Scenario tree for the exogenous pathway 'import ofbatch ofpiglets' i> e u u <?U O •o <**o e t: o > C o ,u to r\ ^ .S eC ra O P CJ . r •s s M y £H 5 c D - r\ Fig. 4 Scenario tree for the endogenous pathway 'direct contact with wild boar' o •a V > y S \ O •a '5. Ü u • — .0 Eo rt '5. C J2 o Q A Fig. 5 Contribution of regions oforigin tothe probability ofCSFV introduction into the target areas Probability of CSFV introduction by exogenous pathways per EPIDEMICineach region oforigin Probability of CSFV introduction by exogenous pathways per YEARfor each region of origin Germany I Germany j 012 . ' 0 « ... | 0.04 \.. /yyyysyssy/sy 1 g .:::.:.:.•-: iJlJLsJL syy/ysyssy/ss region of origin region of origin France Franc« dotsT F3l JtlLULi /yy/ysysjy/sy / yy/ysys'y/yy 'S region of origtn Kary Kary 003 , • 008, n 0(0 -. ;F! ,0 °<* j a»4 o=J k f' :; °"i !» B _--B ys/y/yvysss region of origin region of origin The Netherlandi T h t Netherlands .—jn. /v'yyysyssysyy Belgium i I Belgium I 00« . . .. on: . •tl I' iys y/ss/ss/sss region of o n g « ;/ / yyy/ysyyss/ V Fig. 6 Relative contribution ofeach exogenous pathway to the annual probability of CSFV introduction into thetarget areas byexogenous pathways 1 01 o D in TJ u 3 •o g H u 3 TJ O n. j t o- o -£ o Q. Tl £co CD «i 0) .c o r o u a 3 -o • ï ni O o CD a a <11 i t Tl o "ra o. .o CU c <D c N !c V) o g CD c •c o O o Q. TD f o a t" tQ. •e •e o o Q. a. t- a • a .*o V> u Ol Ol o c c w 'c TJ OS CD CU > ü h (U CD .O O) fc (X D V) ? aai ai a. o t: O •c o o Q. a t- b dl Q- O T* O a F • • a a !7 i/ ;'' !.. 'l • • • .«i*?v„ Hf '/te: 1 TO a> ai ai VA1R5-PL97-3566 'Development of prevention and control strategies'to address animal health and related problems in densely populated livestock areas of the Community" Final Report Edited by R. Huirne 0 and H.-W Windhorst2' l) Wageningen University, Department of Social Sciences, Farm Management Group IIollandseweg 1, NL-6706 KN Wageningen, The Netherlands 2) University of Vechta, Institute for Spatial Analysis and Planning in Areas of Intensive Agriculture (ISPA), Vechta, Germany FAIR5-PL97-3566 'Developmentofprevention andcontrolstrategies toaddressanimalhealthand related problems indensely populated livestockareasoftheCommunity" FinalReport Typeofcontract: Shared-costresearch project Totalcost: 1,285,200EURO ECcontribution: 1,285,200EURO (100%) Commencementdate: 01-04-1998 Duration: 3yearsand6months Completiondate: 01-10-2001 ECcontact: DGVIF.n.3 Mrs.Isabel Minguez-Tudela Coordinator: Prof.dr.RuudHuirne WageningenUniversity DepartmentofSocial Sciences FarmManagement Group Hollandseweg1 6706KNWageningen TheNetherlands Telephone+31317484065 Telefax:+31 317482745 E-mail:[email protected]
© Copyright 2026 Paperzz