Estimation of the probability of freedom from Bovine virus diarrhoea virus in Norway using scenario tree modelling

(1)

ContentslistsavailableatScienceDirect

Preventive Veterinary Medicine

jo u r n al ho me p ag e :w w w . e l s e v i e r . c o m / l o c a t e / p r e v e t m e d

Estimation of the probability of freedom from Bovine virus diarrhoea virus in Norway using scenario tree modelling

Madelaine Norström

^a,∗

, Malin E. Jonsson

^a

, Johan Åkerstedt

^b

, Anne Cathrine Whist

^c

, Anja Bråthen Kristoffersen

^a

, Ståle Sviland

^a

, Petter Hopp

^a

, Helene Wahlström

^d

aNorwegianVeterinaryInstitute,P.O.Box750Sentrum,NO-0106Oslo,Norway

bNorwegianVeterinaryInstitute,P.O.Box295,NO-4303Sandnes,Norway

cNorwegianSchoolofVeterinarySciences,DepartmentofProductionAnimalClinicalSciences,P.O.Box8146Dep.,NO-0033Oslo, Norway

dNationalVeterinaryInstitute,DepartmentforDiseaseControlandEpidemiology,ZoonosisCenter,SE-75189Uppsala,Sweden

a rt i c l e i n f o

Articlehistory:

Received6December2013

Receivedinrevisedform14April2014 Accepted23June2014

Keywords:

Bovinevirusdiarrhoea Scenariotreemodel Surveillance Cattle Norway

Freedomfromdisease

a b s t ra c t

DiseasecausedbyBovinevirusdiarrhoeavirus(BVDV)isnotiﬁableinNorway.Aneradi- cationprogrammestartedin1992.Thenumberofherdswithrestrictionsdecreasedfrom 2950in1994tozeroattheendof2006.From2007,theaimoftheprogrammehasbeen surveillanceinordertodocumentfreedomfromtheinfection.Toestimatetheprobability offreedomfromBVDVinfectionintheNorwegiancattlepopulationbytheendof2011, ascenariotreemodelofthesurveillanceprogramduringtheyears2007–2011wasused.

Threesurveillancesystemcomponents(SSCs)wereincludedinthemodel:dairy,beefsuck- lersampledatfarms(2007–2010)andbeefsucklersampledatslaughterhouses(2011).The designprevalencewassetto0.2%atherdlevelandto30%atwithin-herdlevelforthewhole cattlepopulation.

ThemedianprobabilityoffreedomfromBVDVinNorwayattheendof2011was0.996;

(0.995–0.997,credibilityinterval).Theresultsfromthescenariotreemodelsupportthat theNorwegiancattlepopulationisfreefromBVDV.Thehighestestimateoftheannual sensitivityforthebeefsucklingSSCsoriginatedfromthesurveillanceattheslaughterhouses in2011.Thechangetosamplingattheslaughterhouselevelfurtherincreasedthesensitivity ofthesurveillance.

∗ Correspondingauthor.Tel.:+4723216482;fax:+4723216485.

E-mailaddresses:madelaine.norstrom@hotmail.com (M.Norström),malin.jonsson@vetinst.no(M.E.Jonsson), johan.akerstedt@vetinst.no(J.Åkerstedt),anne.c.whist@nvh.no (A.C.Whist),anja.kristoffersen@vetinst.no(A.B.Kristoffersen), stale.sviland@vetinst.no(S.Sviland),petter.hopp@vetinst.no(P.Hopp), helene.wahlstrom@sva.se(H.Wahlström).

1. Introduction

Bovinevirusdiarrhoea(BVD)iscausedbybovinevirus diarrhoeavirus(BVDV)inthegenuspestivirus.Thevirus isthecauseofmucosaldisease(MD)and haemorrhagic syndrome,buttheeconomicallymostimportantmanifes- tationof the diseaseis related toinfection in pregnant animals,which mayresultinembryonic death,abortion andcongenitaldefects(Radostitisetal.,2000).Ifthedam isinfectedduringday42and125ofthepregnancy,persis- tentlyinfectedcalvesmaybeborn(Radostitisetal.,2000).

http://dx.doi.org/10.1016/j.prevetmed.2014.06.012

licenses/by-nc-nd/3.0/).

(2)

Theseareconsideredtoserveasthemainreservoirofinfec- tiontootheranimals(Baker,1995).InNorway,BVD/MDisa notiﬁabledisease(Anonymeous,1989).From1984to1986, preliminaryinvestigationsindicatedthatnearly30%ofthe dairyherdshadanimalswithantibodiestoBVDV(Løken etal.,1991),whereasforotherproductiontypesthepreva- lencewasunknown.TheannuallossesincattleinNorway duetoBVDVinfectionwasestimatedtobebetween40and 50millionNOK(KrogsrudandLøken,1992).

A surveillance and control programme started in December1992(LøkenandNyberg,2013)ascollaboration betweenGovernmentalinstitutionsandthecattleindus- try.AcattleherdwasconsideredtobeBVDVinfectedifall sequentialtestswerepositiveincludingvirusidentiﬁcation fromatleastoneanimal.

ThelatestBVDVinfectedcattleherdwasidentiﬁedin April2005and therestrictions duetoBVDVwerelifted in November 2006 (Kampen et al., 2007). As a result, theobjectiveofthecontrolandsurveillanceprogramme shiftedfromeradicationofBVDVtosurveillancetodocu- mentfreedomfromdisease.Fromthebeginningof2007,no BVDVinfectedherdshavebeenidentiﬁedinNorwayandno herdhavebeensubjecttorestrictionsforBVD(Åkerstedt etal., 2012).Thecurrent surveillance programincludes dairyaswellasbeefsucklerherds.Theonlybullstation inNorwayisapprovedbytheEuropeanUnion(EU).This requires a testing regime including a severalinfectious diseasesamongstBVDisincluded(EuropeanCommission, 1988).Importedlivecattle,semenandembryosundergo additionaltestingforBVDVinaccordancewiththecattle industry’sown requirementshandledbytheNorwegian LivestockIndustry’sBiosecurityUnit(KOORIMP).

Theaimofthecurrentstudywastoestimatetheprob- abilityoffreedomfromBVDVinfectionintheNorwegian cattlepopulationbytheendof2011.

2. Materialsandmethods

ThisstudywasbasedontheinformationfromtheNor- wegian surveillance program of BVDV in cattle during 2007–2011,aperiodwherenoknownBVDVinfectedherds werereportedinNorway.Theprobabilitythatthecattle populationinNorwaywasfreefromBVDVbytheendof 2011wascalculatedusingscenariotreemodelling(Martin etal.,2007b).

2.1. Datasourcesanddeﬁnitionofcattleherdproduction types

Thefollowingdatasourceswereusedtocalculatethe populationsizeandcategorisetheNorwegiancattlepop- ulationintoproductiontypes:theRegistryofProduction Subsidies(RPS, NorwegianAgricultural Authority,Oslo), Statistics Norway(SSB,Oslo),and theAgriculturalProp- ertyRegister(NorwegianAgriculturalAuthority,Oslo).As of01.01.2011,theNorwegiancattlepopulationconsisted of856,349animalsdistributedin16,401herds.

Theherdswerecategorisedinto

i)dairy herds deﬁned as herds that deliveredmilk to dairies,includingherdswithcombinedproductionof dairyandbeef(66.6%)

ii)beefherdsdividedintobeefsucklerherdsdeﬁnedas herdswithmorethanonebreedingcow(23.8%),with onebreedingcow(1.0%),andbeefﬁnishingherds(8.5%) withnobreedingcows

iii)farmsteaddairy(0.1%),deﬁnedasherdswithon-farm productionofdairyproductsandnodeliveryofmilkto dairies(Table1).

Recordsonmilkdeliveringcattlewereobtainedfrom the dairy industry.Test resultsand sample information wereobtainedfromtheNorwegianVeterinaryInstitute.

2.2. Surveillancesystemcomponents

Thethreesurveillancesystemcomponents(SSCs)ofthe currentofﬁcialNorwegiansurveillanceprogramforBVDV duringthestudyperiodweredairy,beefsucklersampledat farmsandbeefsucklersampledatslaughterhouses(Fig.1).

EachSSCisdescribedbelow.

2.2.1. DairySSC

Annually, 12.5% of all dairy herds were randomly selectedforsamplingwhichensuredbulktankmilk(BTM) samplesfrom atleast 10%of theherds. Thenumber of herdstesteddecreasedfrom1575in2007to1226in2011 (Table2)duetoadecreaseinthenumberofdairyherdsin Norway.In2008,herdsselectedforBVDVtestingtheprevi- ousyearwereexcludedfromthesamplingframe,andfrom

Table1

NumberofNorwegiancattleherdsdistributedonproductiontypesfrom2007to2011.

Year Category No.ofdairy deliveringmilk*^a

No.ofdairyfarm steaddairy

Noofbeef suckler>1cow*

Noofbeef suckler1cow

Noofbeef ﬁnishing

Total

2007 Herd 140,78 26 3926 213 1634 19,877

Animal 732,920 1354 131,372 1731 33,273 900,650

2008 Herd 13,227 25 3716 211 1563 18,742

Animal 725,027 1370 129,247 2257 33,546 891,447

2009 Herd 12,221 23 3834 180 1509 17,767

Animal 701,310 1320 135,904 1740 35,505 875,779

2010 Herd 11,501 21 3883 176 1491 17,072

Animal 686,946 1254 141,368 1780 35,082 866,430

2011 Herd 10,928 20 3903 158 1392 16,401

Animal 672,891 1231 145,707 1491 35,029 856,349

*Includedinthesurveillanceprogram.

aIncludescombinedherds.

(3)

Fig.1. ThescenariotreeoftheNorwegiansurveillanceprogramforBVDVfrom2007to2011withthesurveillancesystemcomponentsdairy,beefsuckler sampledatfarms(2007–2010)andbeefsucklersampledatslaughterhouses(2011).BTMone=bulktankmilksample;BTMpaR=retestofbulktankmilk samples;PoolBloYS(5)=uptoﬁvebloodsamplesinapool;PoolBloYS(10)=uptotenbloodsamplesinapool;IndivBloSameYS=individualbloodsample collectedatthefarm;IndivBloSlh=individualbloodsamplecollectedintheslaughterhouse;Invest=furtherinvestigations.

2009to2011,herdstestedinthetwopreviousyearswere excludedfromthesamplingframe.

2.2.2. BeefsucklerSSCatfarms

In2007–2010,12.5%ofthebeefsucklerherdswereran- domlyselectedforsampling.Thenumberofsampledherds variedbetween370and507(Table2).In2007and2008, uptoﬁvebloodsamplesfromyoungstock(7–15months of age)(PoolBloYS(5))werecollectedper herd.In 2009 and2010,thesampleswerecollectedfromanimalsolder than24months.Allanimalsinthisagegroupweresam- pledin2009andin2010uptotenanimalsperherdwere sampled.Thenumberofbloodsamplescollectedperyear variedbetween4020and5048.Foreachoftheyears,those herdsthathadbeentestedforBVDVtheprevious2years wereexcludedfromthesamplingframe.

2.2.3. BeefsucklerSSCinslaughterhouses

In2011,approximately5000samplesfromadultcattle frombeefsucklerherdswererequestedfromslaughter- houses, i.e. approximately the same sample size as in

previousyearswhenbeefsucklercattleweresampledat farms.The numbers of samplesto becollectedat each slaughterhousewereproportionaltothenumberofslaugh- teredadultbeefsucklercattle(i.e.thecarcasscategories heifers,cowsandbulls)andthesampleswereequallydis- tributedoverworkingdays andmonths,exceptfor July wherenosamplingoccurredduetoreduced numberof adultbeefcattledeliveredtotheslaughterhouses.Samp- lingtookplaceatslaughterhousesthatslaughteredmore than500adultbeefsucklercattlein2010(12intotal)rep- resenting84%ofthetotalslaughterofadultbeefcattlein 2011.Intotal,1278“beef”herdsweresampled,however forthepresentanalysiswecategorisedtheherdsaccord- ingtotheRPSresultingin1094beefsucklerherdswith morethanonecowincludedintheanalysis.

2.2.4. TestingprotocoldairySSC

The ﬁrst step in the surveillance for dairy cattle consistedofBTMtestingforBVDVantibodies(BTMONE).If BTMONEwaspositive,thesamplewasthenretestedtwice (BTMpaR)withthesametestandconcludedaspositiveifat

(4)

Table2

Numberofherdsoranimalsincludedinthesurveillancesystemcompo- nents623forbovineviraldiarrhoeainNorwayfrom2007to2011.

Surveillancesystem component

Year Noofherds No.of primary samples

Dairy^a 2007 1575 1575

2008 1424 1424

2009 1315 1315

2010 1328 1328

2011 1226 1226

Beefsucklersampledat farms

2007^b 370 1485

2008^b 407 1817

2009^c 435 4926

2010^d 507 4018

Beefsucklersampledat slaughterhouses

2011^e 1094^f 4172

aOnebulktankmilksampleperherdwascollected.

b Uptoﬁvebloodsamplesperherdwerecollectedandpooled.

c Allcattleinaherdweresampledandpooledwithuptotensamples.

d Uptotenbloodsamplesperherdwerecollectedandpooled.

eSampleswerecollectedfromindividualcattleatslaughter,onlyherds categorisedbeefsucklerherdswithmorethan1sucklingcowisincluded intheanalyses.

f Onlybeefsucklerherdswithmorethanonesucklingcowwere includedfromthetotalnumberof1278sampledbeefherds.

leastoneofthetestsinBTMpaRwerepositive.IftheBTMpaR

waspositive,uptoﬁvebloodsampleswerecollectedfrom youngstock intheherdandanalysed asapooled sample(PoolBloYS(5)).WhenPoolBloYS(5)waspositive, the bloodsamplesinthepoolwereanalysedindividually(Indi- vBloSameYS).AherdwithpositiveIndivBloSameYSresults wouldbeputunderrestrictionsandfurtherinvestigations (Invest)wereperformedtoclarifyiftheherdwasinfected ornot.Inordertoﬁndpersistentinfectedanimalsallcat- tlewithinsuchaherdweresampledand sampleswith aweakpositiveornegativeserologicalresultswouldbe testedforthepresenceofBVDVusinganantigen-capture ELISA(IDEXXLaboratoriesInc.,Westbrook,Maine,USA).

PositivereactionsforBVDVinnewlyinfectedherdswould beveriﬁedwiththepolymerasechainreaction(PCR)and sequenceanalysis.

2.2.5. TestingprotocolbeefsucklerSSCatfarms

Samplesfromeach herdwere tested forBVDV anti- bodiesinpools,withamaximumoftensamplesineach pool(PoolBloYS(10).Ifthepoolwaspositive,theindividual samples(IndivBloSameYS)formingthepoolwereanalysed separately.Ifanyindividual samplewaspositive, a fur- therinvestigation(Invest)wasperformedasdescribedin Section2.2.4fordairySSC.

2.2.6. TestingprotocolbeefsucklerSSCin slaughterhouses

Atmostﬁvesampleswerecollectedfromasingleherd perday.Thesampleswereanalysedindividuallyorinpools comprisingthesamplescollectedfromthesameherdatthe sameday(IndivBloSlh).Ifapooledsamplewaspositive, theindividualbloodsampleswereanalysedseparately.As mostherdsonlyhadonesampleperday,itwasassumed that no pooling occurred and retesting of pools was

thereforenotincludedinthemodel(Fig.1).Ifthistestwas positive(IndivBloSlh),youngstocksamples(PoolBloYS(5)) werecollectedasdescribedinSection2.2.4fordairySSC andthesamefollow-upprocedurewasused.

2.2.7. Laboratoryanalysesandinterpretationofthetest results

Serum andBTMsamplesweretested forBVDV antibodies,usinganindirectenzyme-linkedimmunosorbent assay(ELISA,SVANOVIR^TMBVDV-Ab;SvanovaBiotechAB, Uppsala,Sweden)(Junttietal.,1987).ResultsofBTMtest- ingweredividedintofourgroupsdependingonthelevels of antibodies: 0.undetectable, 1.low, 2. moderate, and 3.high(Niskanen,1993).Until2009,group0and1were regardedasnegativeandgroup2and3aspositive.From 2010,thegroupswerereclassiﬁedtoincreasethesensi- tivityoftheprogrammesothatgroup0wasregardedas negativewhereasgroup1,2and3wereconsideredaspos- itive(Åkerstedtetal.,2012).Forbloodsamples(individual andpooledsamples),thecut-offvaluegivenbythemanu- facturerforindividualbloodsampleswasused.

2.2.8. Casedeﬁnition

A herd was consideredto be infected if all sequen- tialtestswerepositiveincludingthefurtherinvestigations describedinSections2.2.4–2.2.6(Fig.1).

2.2.9. Designprevalence

TheprobabilityoffreedomfromBVDVwascalculated usingaherddesignprevalenceP_H^∗of0.2%(Table4)i.e.the scenariotreemodelestimatedthesensitivityoftheSSCs astheprobabilityofdetectingatleastoneinfectedherdif 0.2%ormoreofthecattleherdsinthereferencepopulation wereinfected.Thedesignprevalenceof0.2%waschosen asnointernationalacceptedguidelinesfordesignpreva- lencesforBVDVexist,andthisdesignprevalenceisusedfor deﬁningfreedomfromBrucellaabortusandenzooticbovine leukosiswithintheEU(EuropeanCommission,1998).

Inthepresentstudy,awithin-herdprevalence(P^∗_U)of 30%forboth dairyandbeefsucklerherds(Table4)was used.Thisisconsideredasaconservativeestimateasother studieshavereportedhigherwithinherdprevalencesboth fordairy(HoueandMeyling,1991;Braunetal.,1997)and beefcattle(Perezetal.,1994;Paisleyetal.,1996;Brulisauer etal.,2010).

2.3. Scenariotreemodel

Thescenariotreemodel(Fig.1)wasusedtoestimate theprobabilityof gettingapositive outcomegiven that theinfectionwaspresentatthespeciﬁeddesign preva- lences.Thescenariotreemodelincludesthecategorynode Herdtypewithtwobranches,DairyandBeefSucklerand theinfectionnodeHerdstatuswithtwobranches,infected and uninfected.It furtherincludes12differentdetection nodes.Theassumptionsandcalculationsofthetestsen- sitivitiesforthedifferenttestsasexplainedinSection2.2 aredescribedbelow.EachSSCwasassumedtohaveaspeci- ﬁcityof1,becauseallsamplestestingpositivearefollowed upwithfurthertestingandinvestigation.

(5)

2.3.1. Testsensitivityonindividualserumandmilk samples

The manufacturer reported that 99 of 99 individual serum samples positive in thevirus neutralisation test (Svanova)alsowerepositiveintheELISA(personalcom- munication,AfsanehJalali,BoehringerIngelheimSvanova).

ThesensitivityofaserumELISA(individualsample)(Se_Blo) wasdescribedbyaBetadistribution:

Se_Blo=Beta(100,1)

ThesensitivityofthemilkELISAusingtheserumELISA asgoldstandardwasevaluatedonserumandmilksam- plesfrom21individualsthatwerepositiveonserumand 20ofthemwerealsopositivewhenexaminingmilk(per- sonalcommunication,AfsanehJalali,BoehringerIngelheim Svanova).ThetotalsensitivityofthemilkELISA(Se_Mi)was calculatedastheproductofaBetadistribution,Beta(21,2) andthesensitivityoftheserumELISA

Se_Mi=Beta(21,2)×Se_Blo

Thesensitivity oftheindividualfollowuptest (Indi- vBloSameYS)wasassumedtobeequal tothesensitivity oftheindividualbloodtest(Se_Blo)assumingonlyoneposi- tivesamplewouldhavebeenincludedinthepoolsofﬁveor tensamplesalthoughmorethanonepositivesamplecould havebeenincluded.

Thesensitivity ofthetest (IndivBloSlh)wasassumed tobeequaltothesensitivityoftheindividualbloodtest (Se_Blo).Although,pooling ofsamplescollectedthesame dayfromthesameherdoccasionallyoccurred, thiswas nottakenintoconsideration.Thesensitivityofthefurther investigations(Invest)wasassumedtobe1.

Se_In_v=1

2.3.2. Testsensitivityofbulktankmilksamples

In accord with Niskanen (1993) and manufacturer’s information(Svanova,2012)theBTMELISAwasassumed to have the same sensitivity as the individual milk ELISAforpoolswithlessthan50 animals.Thisassump- tion was applicable in Norway as the herd sizes in the Norwegiandairy populationrarely exceed50 dairy cows.

Se_BTMONE=Se_Mi

The sensitivity ofthe retestsof BTM(BTM_par) inter- pretedinparallel(Se_BTMpaR)wascalculatedas(Dohooetal., 2009,page101–102):

Se_BTM_PAR=(Se_BTMONE+Se_BTMONE)−Se_BTMONE×Se_BTMONE

2.3.3. Testsensitivityofthepooledsamples

TheeffectofusingserumELISAonpooledbloodsam- pleshasbeenevaluatedbyCowleyetal.(2012).Intheir study,90poolsconsistingof30serumsampleseachwith between 1 and 30 seropositivesamples per pool were examined. Of these, 35 of the pools included between 10% and 30% individual ELISA positive samples and all thesewerepositiveinthepooledELISA.Poolswith40%

Table3

Theestimatedprobabilities(min,modeandmaxvalues)ofselectingat leastoneinfectedindividualwhenpooling5(a)or10 (b)individual bloodsamplesfromaherdinthesurveillanceprogramofBVDVdur- ing2007–2011.Thevalueswereobtainedbysimulation(5000iterations) includingactualherdsizesoftheﬁnitepopulationwithineachproduction categoryandwithinherddesignprevalence.

Productiontype Year Min Mode Max

Beefsuckler^a 2007 0.819 1.000 1.000

Beefsuckler^a 2008 0.820 1.000 1.000

Beefsuckler^b 2009 0.968 1.000 1.000

Beefsuckler^b 2010 0.967 1.000 1.000

Beefsuckler^a 2011 0.821 1.000 1.000

Dairy^a 2007 0.817 0.846 1.000

Dairy^a 2008 0.813 0.843 1.000

Dairy^a 2009 0.818 0.843 1.000

Dairy^a 2010 0.816 0.842 1.000

Dairy^a 2011 0.816 0.843 1.000

or morepositive sampleswasnot included inthe sen- sitivityestimation astheprevalenceofpositive samples wouldhaveexceededthedesignprevalenceusedinthis study.

Comparedtoindividualbloodsamples(Se_Blo),thesen- sitivityofa pooledsample (PoolBloYS(10)),(SePo10)was describedwithBeta(36,1).

Further, the study from Cowley et al. (2012) indi- cated that anypool with at least 10% positive samples would be positive. Therefore, we assumed that a pool withone positiveoutof ﬁvesamples(20%)wouldhave nolossinsensitivity, i.e.thesensitivity whenanalysing (PoolBloYS(5))was equal to that of individual analysis (Se_Blo).

Theprobabilitiesofselectingatleastonepositiveani- malwhensamplingﬁveortenanimals,respectivelyina herdwith30%infectedanimals,wereestimatedusingthe hypergeometricdistribution,Thiswasbasedontheactual herdsizesinthedairyandbeefsucklerherdpopulationsfor eachyeary,separately.Themode,minimumandmaximum valuesoftheprobabilitieswereobtainedfromsimulations with5000iterations(Table3).Usingtheseestimates,the probabilitythatthepool oftensamples(Se_Samp10,_y)and thepool ofﬁvesamples(Se_Samp5,_y), willinclude atleast oneinfectedanimalwasdescribedasPert(min,mode,max) (Table4).Theoverallsensitivitiesofthesetestswerecal- culatedas

SePoolBloYS(5),y=Se_Samp5,y×Se_Blo

SePoolBloYS(10),y=Se_Pol10×Se_Samp10,y×Se_Blo

2.3.4. Repeatedtestsandinterpretation

For eachof theSSCs thereis a sequenceof repeated tests that will be performed once a test is positive.

Althoughtherepeatedtestsarelikely highlycorrelated, we choose for simplicity to use the same sensitivity values for the primary test as the repeat test as this wouldonlyunderestimatetheoverallsensitivityforeach SSC.

(6)

Table4

InputvaluesusedinthescenariotreemodeltoevaluatethesensitivityoftheNorwegiansurveillanceprogramofBVDVfrom2007to2011.

Parameters Inputs Distribution Notation

Herdleveldesignprevalence 0.002 Fixed P_H^∗

Withinherddesignprevalence 0.3^a Fixed (P^∗_U)

Relativeriskofbeefsucklerherdsversusdairyherdsofbeinginfected 2.59^a Fixed RRHR

Sensitivityonindividualbloodsamples 0.99^a Beta(100,1) SeBlo

Sensitivityonbulktankmilksample 0.91^a Beta(21,2)×SeBlo SeBTM

Sensitivityofpooledbloodsamples(poolsof10)vs.analyzingindividualsamples 0.97^a Beta(36,1) SePo10

Probabilityofselectingatleastoneinfectedindividualwhenpooling5individualblood samplesfromaherd^binyear;y

SeeTable3 Pert(min,mode,max) SeSamp5

Probabilityofselectingatleastoneinfectedindividualwhen10poolingindividualblood samplesfromaherd^binyear;y

SeeTable3 Pert(min,mode,max) SeSamp10

Sensitivityofthefurtherinvestigations 1 Fixed SeInv

Prior(presurveillance)probabilityofinfection 0.5 Fixed PriorPInf

Probabilityofintroduction(peryear) 0.1 – PIntro

aExpectedvalue.

b DairyorbeefsucklerherdasdescribedinTable3.

2.4. Calculationofsurveillancesystemcomponents sensitivities

2.4.1. Adjustedrisks

Theadjustedriskforeachproductiontypecategorywas calculatedasdescribedbyMartinetal.(2007a)as AR_LR=1/(RR_HR×PrRefPop_HR+(1−PrRefPop_HR))

whereARLRwastheadjustedriskforlowriskpopulation, herethedairyherds(Da),RR_HRwastherelativeriskforthe highriskpopulation,beefsucklerherds(Be),whichwascal- culatedbydividingtheproportionsofBVDVinfectedbeef sucklerherds(20.5%)bydairyherds(7.9%)in1993.Theyear 1993waschosenasitwastheearliestyearwithdetailed prevalencedataavailableandtheeradicationprogramme wasconsideredtonothavehadaconsiderableinﬂuence ontheprevalenceestimates.PrRefPop_HRwasthepropor- tionofreferencepopulationfallingintoeachproduction typecategoryforeachoftheyearsincludedinthemodel.

Theadjustedriskforthehighriskpopulation(AR_HR)was calculatedas

AR_HR=RR_HR×AR_LR

Theherddesignprevalenceandtheadjustedrisksfor dairyherdswereusedtocalculatetheeffectiveprobability ofadairyherdbeinginfected(EPInfDa).

EPInf_Da=P^∗_H×AR_LR

whereP_H^∗isthebetweenherddesignprevalenceandAR_LR istheadjustedriskofadairyherdbeinginfected.Theeffec- tiveprobabilityofabeefsucklerherdbeinginfectedwas calculatedinasimilarway.

2.4.2. Annualsensitivityfordairyherdsandbeefsuckler SSC(surveyatfarms)

Thedairyherdsensitivity(Se_HDa),i.e.theoverallsen- sitivityof thetesting including follow up tests (Fig. 1) was estimated by multiplying sensitivities of the ﬁve sequentialtestsinthedairySSC:Se_HDa=Se_BTM×Se_BTM_PAR× SePoolBloYS(5),y×Se_Blo×SeInv

Thebeefsucklersensitivity(SeBe,y),i.e.theoverallsen- sitivityofthetestingincludingfollowuptest(Fig.1),for

theyears2007–2010wasestimatedbymultiplyingsensi- tivitiesofthethreesequentialtestsinthebeefSSCs.

SeHBe,y=

SePoolBloYS(5),y×SeBlo×SeInv fory∈(2007,2008) SePoolBloYS(10),y×SeBlo×SeInv fory∈(2009,2010)

The probability of all dairy herds testing negative (PNegDa,y)ify∈(2007,2011)wascalculatedas:

PNegDa,y=(1−EPInfDa×SeHDa)^nDa,y

whereEPInf_Daistheeffectiveprobabilitythatadairyherd isinfected,SeHDaistheoverallsensitivityofthesequen- tialtestingdoneindairyherds.Furtherisn_Da,ythenumber ofdairyherdstestedinyeary.Theannualsensitivityfor dairyherdsSSC(Se_{SSC Da,y})isthecomplementaryeventofall herdstestingnegative.Theprobabilityofbeefsucklerherds testingnegativeandtheannualsensitivityofthebeefSSC (Se_SSC_Be,y)fory∈(2007,2010)wascalculatedinasimilar way.

2.4.3. AnnualsensitivityforbeefsucklerSSC(surveyin slaughterhouses)

Fortheyear2011,whensamplingofbeefsucklerherds was performed at slaughterhouses, the herd sensitivity was calculated for each herd separately. As herd sizes oftenweresmall,theexpectednumberofinfectedcattle (no.inf_HBe,2011)ineachherdin2011wascalculatedusingthe BinomialdistributionBin(n,p),wherenistheherdsizeand pisthewithinherddesignprevalence(P_U^∗).Theprobability ofaherdtestingnegativewascalculatedusinganapprox- imationofthehypergeometricdistribution(MacDiarmid andHellström,1988).In98.5%oftheanalysesﬁvesam- plesorlesswereincludedinthepoolandthesensitivity wasSePoolBloYS(5)asdescribedinSection2.3.3.Forsimplic- ityweassumedthatthesensitivitywasthesameforthe remaining1.5%oftheanalyses.Theprobabilitythatabeef herdtestednegativein 2011,giventhatit wasinfected (PNegHBloBe,2011)wascalculatedas:

PNegHBloBe,0=2011=(1−(SeBlo×no.tested/herdsize))^no.^inf^HBe,2011 whereSe_Bloisthesensitivityoftheindividualbloodtest, no.testedisthenumberofcattletestedintheherd,herdsize isthenumberofadultanimalsintheherdandno.infHBe,2011

istheexpectednumberofinfected cattlein theherdin

(7)

2011.Asthenumberofsamplescollectedfromeachherd variesbetweenherds,PNegHBloBe,2011willalsovarybetween herds.

Theprobabilitythattheinfectedherd(30%)testedpos- itiveisthecomplementaryevent:

SeHBloBe,2011=1−PNegHBloBe,2011

Theoverallsensitivityforabeefherd(HBe)including thesequentialtesting(Se_HBe,2011)wasobtainedbythefol- lowingcalculation:

Se_HBe,2011=SeHBloBe,2011×SePoolBloYS(5)×Se_Blo×Se_In_v whereSeHBloBe,2011wastheprobabilitythattheherdtested positive in the ﬁrst test; IndivBloSlh and SePoolbloYS(5)

(asdeﬁnedin thedairy SSC)wastheprobabilitythat it testedpositiveinthesecondtestPoolBloYS(5),Se_Blowas the probability that it tested positive in the third test;

IndivBloSameYSandSeInvistheprobabilitythatittested positiveinthefurtherinvestigation;Invest.

The probability that all beef suckler herds tested negative, (PNegBe,2011) taking into account the effective probabilitythatabeefsucklerherdwasinfected,i.e.EPInf_Be wascalculatedas:

PNeg_Be,2011=

∀HBe

(1−EPInf_BE×Se_HBe,2011)

Theprobabilitytodetectatleastoneinfectedbeefsuck- lerherdi.e.thesensitivityforbeefsucklerherdsSSCin2011 (SeBe,2011)wasequalto1−PNegBe,2011.

2.4.4. ThecombinedannualSSCsensitivityforallSSCs Asthesurveillanceindairyandbeefsucklerherdsare mutuallyexclusive,theprobabilitythatbothtestednega- tivefory∈[2007,2011]wascalculatedas:

PNeg_Da_and_Be,y=PNegDa,y×PNegBe,y

The annual sensitivity of the surveillance system in dairyandbeefsucklerherds,i.e.theprobabilitythatatleast onherdtestedpositivewascalculatedas:

Se_Da_and_Be,y=1−PNeg_Da_and_Be,y

2.5. Probabilityoffreedomandtemporaldiscounting TheprobabilityoffreedomfromBVDVwascalculated usingBayestheorem(Martinetal.,2007b).Theposterior probabilityoffreedominyeary(PostPFreey)i.e.afterthe testingthathasbeendoneduringthatyearequalledthe negative predictedvalue.Assuming a perfect speciﬁcity thiswascalculatedasdescribedbyMartinetal.(2007b).

FortheﬁrstyearPriorPInfwaschosenasaneutralprior probabilityof0.5.

The probability of introduction (PIntro) was defined as theannual probabilityof introductionof thedisease in a sufficient number of herds toexceed thespecified designprevalence(P*)(Martinetal.,2007b).Theproba- bilityofintroductionofBVDVintoNorwayforeachofthe studyyearswasconsideredtobelow.However,weused aconservativeapproach,i.e.usingahigherprobabilityof introductionaswethinkisrealisticanditwastherefore

Fig.2.Theprobabilityofinfection,(PriorPInf),theposteriorprobabilityof freedom(PostPfree)andtheposteriorprobabilityofinfection(PostPInf)of thescenariotreemodeloftheNorwegiansurveillanceprogramofBVDV from2007to2011.

setto0.1 for each year inthe model.The mostproba- blesourcesofintroductionwereconsideredtotakeplace duetoimportofanimals,semenorembryo.Theseimports arestronglyregulatedbothbyEUrequirements(European Commission,1964)andvoluntaryadditionalrequirements byKOORIMP.

PriorPInfatthebeginningofthefollowingyearandthe posteriorprobabilityofinfectioninyeary(PostPInfy)were bothcalculatedasdescribedbyMartinetal.(2007b).

2.6. Simulation

ThemodelwasruninR(RCoreTeam,2012).Therewere run5000iterationsforeachoftheSSCs.

2.7. Sensitivityanalysis

ThesensitivityofthePostPFreetovariationinthePIntro wasassessedbyincreasingthePIntroto0.2and0.3andby changingtheRRto1.0and3.0,respectively.

3. Results

Theestimatedmedianvalueofprobabilityoffreedom fromBVDVintheNorwegiancattlepopulationwasabove 0.99from2008andonwardsandattheendof2011,itwas 0.996(0.995–0.997;90%credibilityinterval)(Fig.2).

Theestimatedmedianvalueofannualsensitivityofthe dairySSCvariedanddecreasedfrom0.811in2007to0.709 in2011(Fig.3).Theestimatedmedianvalueofannualsen- sitivityofthesucklingbeefSSCsincreasedfrom0.715in 2007to0.890 in 2011(Fig.3).By changing theannual riskofintroductionfrom0.1to0.2and0.3theprobabil- ityoffreedom decreased(medianvalue(90%credibility interval)from0.996(0.995–0.997);to0.991(0.989–0.992) and0.985(0.981–0.987),respectively.BychangingtheRR themedianvalueofprobabilityoffreedomonlyslightly changedfrom0.969to0.985foraRRof1and3,respectively.

(8)

Fig.3. SensitivitiesforeachSSCwithmedian,loweranduppercredibile interval(indicatedwithhorizontalbars)intheNorwegiansurveillance programofBVDVfrom2007to2011;Da=DairySSC,Be=BeefsucklerSSC andDaandBe=CombinedDairyandbeefsucklerSSCs.

Table5

Theestimatedherdsensitivities(min,modeandmaxvalues)inthe surveillanceprogramofBVDVforeachoftheyears2007–2011obtained bysimulation(5000iterations).

Productiontype Year Min Mode Max

Beefsuckler 2007 0.78 0.96 1.00

Beefsuckler 2008 0.82 0.96 1.00

Beefsuckler 2009 0.75 0.95 1.00

Beefsuckler 2010 0.76 0.95 1.00

Dairy 2007 0.90 0.95 0.96

Dairy 2008 0.89 0.94 0.96

Dairy 2009 0.90 0.94 0.95

Dairy 2010 0.91 0.95 0.96

Dairy 2011 0.94 0.97 0.98

Theherdsensitivities,theprobabilitythataninfected herdwouldhaveapositiveresult,variedfrom0.60to0.93 fordairyherds(Se_HDa)asshowninTable5.Forbeefherds (SeHBe,y)theherdsensitivitiesvariedfrom0.747to0.999 (Table5).In2011,theherdsensitivityforbeefsucklerherds variedfrom0.24to0.95(Table5)withamedianvalueof 0.60asthenumberofsamplescollectedperherdvaried.

4. Discussion

Theresultsfromthescenariotreemodelsupportthat theNorwegiancattlepopulationisfreefromBVDV.Surveil- lanceandcontrolprogrammeimplementedintheNordic countrieshaveresultedineitherfreedomoralmostfree- domfrom BVDV.However,to ourknowledge,no other countrieshavebeenabletoquantifytheprobabilityoffree- domfromBVDVincattlesofar.SeveralotherEuropean countrieshaveimplementedsurveillanceandcontrolpro- grammeforBVDV(StahlandAlenius(2012).

TheannualsensitivityfordairySSCwaslowerthanfor beefsucklerSSCsfrom2009onwards.Mainreasonsforthis arethelowerherdsensitivityfordairyvs.beef,thedeclin- ingnumberofdairysampledandincreasingnumbersof beefsucklerfarmssampledandthehigherinfectionrisk (EPI)forbeefsucklerherds.Thisispartlyduetothefact thatthesensitivity oftheBTMtest waslowerthan the

serologicaltestandthatmoretestswereincludedinthe dairySSCthaninthebeefsucklerSSCs.

Althoughthenumber ofanalysed samplesfrombeef suckler cattle wascomparable withpreviousyears, the sensitivityofthebeefsucklerSSC(SeHBe,y)washigherin 2011.Thisseemsreasonableasthesensitivityisexpected toincreaseifmoreherdsandfewersamplesperherdare collectedsinceinfectiousdiseasesareexpectedtocluster atherdlevel.However,thesensitivityofthebeefsuckler SSCofyear2011wasnotdirectlycomparablewithprevious yearsastheherdsensitivitieswerecalculatedindifferent ways.Whensamplingwasdoneinindividualbeefsuckler herds(2007–2010),thetruenumberofsamplescollected ineachherdnotwastakenintoaccount.Itwasassumed that ﬁveortensampleswerecollectedfromeach herd.

However,when fewersamples werecollected, thiswas due totheherdsizebeingsosmallthatallyoungstock wassampled,whichimpliesthatwehaveunderestimated thesensitivityinthoseyearsasthiswasthecaseinmore than30%oftheincludedherds.Ontheotherhand,forthe samplingatslaughterhouses,thecalculationofthesensi- tivityofthetestingineachherdwasbasedontheherd size,theexpectednumberofinfectedanimalsintheherd (obtainedbysimulation)andthenumberofanalysedsam- plesperherd.However,itwasassumedthatallsamples weretestedindividually despitethefactthat poolingof samplescollectedonthesame dayfromthesameherd occurred.Onlyin1.5%oftheanalysesmorethanfivesam- pleswereincludedinthepool.Astheextentofthispooling wasverysmall,thiswasnotassumedtoaffecttheoutputof themodel.Furthermore,apartfromthecomponenthaving ahighersensitivity,samplingatslaughterhousesismore costefficient.Althoughthenumberofcollectedsamples wasinthesamerange,thenumberofanalysisincreased in2011.However,thebenefitsofnothavingtocollectthe samplesatthefarmsoutweighedtheincreasedcostsdue totheincreasednumberofanalysisin2011.

Before2011,beefsucklerherdswithonlyonebreeding cowwerenotincludedinthesurveillancesystem.Evenif thissubpopulationhadbeensampleditwouldhavebeena seldomphenomenonasthisconstitutesonlyaminorpro- portionofthebeefsucklerpopulation.Also,introduction andpersistenceofBVDVinsuchherdswasconsideredneg- ligibleastheseherdsareverysmall.Incombinedherds,i.e.

herdswithbothdairyandbeefcattle,onlythedairycat- tlewereincludedinthesurveillance.Itwasassumedthat thebeefcattleand thedairy cattlehadclosecontactso iftheinfectionhadbeenintroducedintothebeefsuckler cows,theinfectionwouldhavespreadtodairycattle.Thus, thebeefsucklercowsonthesefarmswereindirectlysur- veyedbytheBTMsampling.Thebeefﬁnishingherdswere not includedin the surveillanceprogramme. No breed- ingoccursintheseherdsandtheanimalswerepurchased fromeitherdairyorbeefsucklerherds.Therefore,itwas consideredunlikelythattheinfectioncouldestablishand persistin theseherds.Farm steaddairy herdshave not beenincludedinthesurveillanceafter2006.Thispopula- tioncoverlessthan0.1%ofthetotalcattlepopulationand theherdsareusuallysmall.Itisconsideredunlikelythat theinfectionwouldestablishintheseherds,buttheinclu- sionoftheseherdsinthesurveillanceactivitiesmightbe

(9)

consideredinthefuturetoincreasecoverageofthesurveil- lanceprogramme.

TheestimateusedfortheriskofintroductionofBVDV, canbeconsideredconservative.Theincreaseoftheproba- bilityofintroductionperformedinthesensitivityanalysis onlyslightlychangedtheresults.AlthoughNorwegianlive- stockisconsideredtobefreefromtheBVDV,import of infected animals and unknown wildlife reservoirs may poseacontinuousthreattothepresentstatus.Fortherapid detection of a potentialreintroduction and consecutive controlofspreading,asurveillancesystemhastomakeefﬁ- cientuseofthecompetenceandawarenessexistingamong farmersandlocalveterinarians.Sincetheoutbreakofblue- tongueandtheeradicationofBVDV,Norwegianfarmers havebecomemoreawareofintroductionofinfectiousdis- eases. Theprobability of introduction of BVDV through illegalimportof animals,semenandembryowerecon- sideredlowbecauseillegalimportmostlikelywouldbe discoveredthrough mandatoryregistrationofherddata forallNorwegiancattleherdsintothenationalindividual cattleregister.Allinformationregardingoriginofanimals mustberecordedbeforetheanimalsmaybeslaughtered.

ImportoflivevaccinescontaminatedwithBVDVmightbea potentialriskofintroductionaswell,buttoourknowledge thereareatpresentonlyonesuchvaccinesonthemarket inNorway(Felleskatalogen,2014).Otherruminantsmay poseariskforintroductionofBVDVandpestivirushasbeen foundtobeendemicintheroedeerpopulationinNorway (Lillehaugetal.,2003).However,theprobabilityofintro- ductionbywildruminantscanberegardednegligible,as shownincamelidsby(Mudryetal.,2010).Thereisasmall populationofcamelids,suchasalpacaandlamainNorway, andapossibleintroductionofinfectiousdiseasesthrough suchpopulationsneedstobeconsidered.Earlierstudies (Løkenetal.,1991)estimatedthemeanprevalenceatherd levelinsheepto18%.Therearenorecentprevalencestud- iesperformedinNorwegiansheep,buttheannualtesting ofcattleherdssincethestartoftheeradicationprogramme in1992havenotdetectedanyreinfectionsduetoinfected sheepevenifBVDVmightbepresentinthesheeppopula- tion.ArecentstudyfromIrelandwhereavoluntaryBVDV controlisrunning(Grahametal.,2013)wasnotableto identifyanyriskofhavingsheepatthefarminrelation toBVDVinfectionincattle.Hencetheriskofintroduction throughotherruminantscanbeconsideredlow.

When calculating the overall sensitivity of all tests theformulasusedrequiresthatthetestareindependent, whichisnotthecasewhenthebulkmilktestisrepeatedon thesamesampleorwhentheseraincludedinthepooled samplewereanalysedindividuallywiththesametest.This wasnotconsideredtocauseanyconcernasthisleadto theestimatedoverallsensitivityoftheteststobeunder- estimated.Furthermore,thechangeincut-offvalueforthe BTMwasnottakenintoaccountresultinginaslightunder- estimationof thesensitivityofthedairySSC after2010.

Whenconsideringbeefsucklerherds,moresamplesthan requestedwascollectedfor1.4%ofallthetestedherdsfor theremainingyearsandin2009alladultanimalsinthe beefsucklerherdsweresampled.Thiswasnottakeninto accountinouranalysisandalsoresultedinunderestima- tionofthesensitivityofthesurveillance.

5. Conclusions

Theresultsfromthescenariotreemodelsupportthat theNorwegiancattlepopulationisfreefromBVDV.Acur- rentchangetosampleattheslaughterhouselevelfurther increasedthesensitivityofthesurveillancesystem.

Acknowledgements

Theauthorswanttothankstaffatthelaboratoriesfor analysingsamplesaswellastheNorwegianFoodSafety Authorityand dairies(TINEBAandQ-meierieneAS)for collectingthesamples.

References

Anonymeous,1989.10.jan.Nr4250.Forskriftomendringifortegnelsen oversjukdommersomomfattesavlovomtiltakmotdyresjukdommer (husdyrloven).NorskLovtidendAvd.I,21–24.

Baker,J.C.,1995.Theclinicalmanifestationsofbovineviraldiarrheainfec- tion.Vet.Clin.NorthAm.FoodAnim.Pract.11,425–445.

Braun,U.,Landolt,G.,Brunner,D.,Giger,T.,1997.EpidemiologischeUnter- suchungenüberdasVorkommenvonBVD/MDbei2892Rindernin95 Milchviehbetrieben.Schweiz.Arch.Tierheilkd139,172–176.

Brulisauer,F.,Lewis,F.I.,Ganser,A.G.,McKendrick,I.J.,Gunn,G.J.,2010.

Theprevalenceofbovineviraldiarrhoeavirusinfectioninbeefsuckler herdsinScotland.Vet.J.186,226–231.

Cowley,D.J.,Clegg,T.A.,Doherty,M.L.,More,S.J.,2012.Bovineviraldiar- rhoeavirusseroprevalenceandvaccinationusageindairyandbeef herdsintheRepublicofIreland.Ir.Vet.J.65,16.

Dohoo,I.,Wayne,M.,Stryhn,H.,2009.VeterinaryEpidemiologicResearch.

VERInc.,Charlottetown,Canada.

EuropeanCommission,1988.CouncilDirective88/407/EECof14June 1988layingdowntheanimalhealthrequirementsapplicabletointra- Communitytradeinandimportsofdeep-frozensemenofdomestic animalsofthebovinespecies.OJL194,10–23.

EuropeanCommission,1998.Councildirective98/46/ECof24June1998 amendingAnnexesA,D(ChapterI)andFtoDirective64/432/EECon healthproblemsaffectingintra-Communitytradeinbovineanimals andswine.OJL198,22–39.

EuropeanCommission,1964.Council Directive64/432/EEConhealth problemsaffectingintra-Communitytradeinbovineanimalsand swine.OJL121,1977–2012.

Felleskatalogen,2014.Felleskatalogen.In:AS,F.(Ed.)(FelleskatalogenAS http://www.felleskatalogen.no/medisin-vet/atc-register/QI02A; last accessed11.04.2014.

Graham,D.A.,Clegg,T.A.,Lynch,M.,More,S.J.,2013.Herd-levelfactors associatedwiththepresenceofbovineviraldiarrhoeavirusinherds participatinginthevoluntaryphaseoftheIrishnationaleradication programme.Prev.Vet.Med112,99–108.

Houe,H.,Meyling,A.,1991.Surveillanceofcattleherdsforbovinevirus diarrhoeavirus(BVDV)-infectionusingdataonreproductionandcalf mortality.Arch.Virol.(Suppl.3),157–164.

Juntti,N.,Larsson,B.,Fossum,C.,1987.Theuseofmonoclonalantibodies inenzymelinkedimmunosorbentassaysfordetectionofantibodiesto bovineviraldiarrhoeavirus.Zentralbl.Veterinarmed.B34,356–363.

Kampen,A.H.,Åkerstedt,J.,Hopp,P.,2007.Thesurveillanceandcon- trolprogrammeforbovinevirusdiarrhoea(BVD)inNorway.Annual report2006.In:Brun,E.,Hellberg,H.,Sviland,S.,Jordsmyr,H.M.(Eds.) NationalVeterinaryInstitute,NorwegianFoodsafetyAuthority,Oslo, 65–71.

Krogsrud,J.,Løken,T.,1992.Pestivirusinfeksjonhosstorfe(BVD/MD).

Husdyrforsøksmøtet:norgeslandbrukshøgskole24.-26.mars1992.

Statensfagtjenesteforlandbruket,Ås,96–99.

Lillehaug,A.,Vikøren,T.,Larsen,I.L.,Åkerstedt,J.,Tharaldsen,J.,Han- deland,K.,2003.Antibodiestoruminantalpha-herpesvirusesand pestivirusesinNorwegiancervids.J.Wildl.Dis.39,779–786.

Løken,T.,Krogsrud,J.,Larsen,I.L.,1991.PestivirusinfectionsinNorway.

Serologicalinvestigationsincattle,sheepandpigs.ActaVet.Scand.

32,27–34.

Løken,T.,Nyberg,O.,2013.EradicationofBVDVincattle:theNorwegian project.Vet.Rec.172,661.

MacDiarmid,S.C.,Hellström,J.S.,1988.Surveillanceforbrucellosisusing askintestoflowsensitivity.ActaVet.Scand.Suppl.84,209–211.

(10)

Martin,P.A.,Cameron,A.R.,Barfod,K.,Sergeant,E.S.,Greiner,M.,2007a.

Demonstratingfreedomfromdiseaseusingmultiplecomplexdata sources2:casestudy—classicalswinefeverinDenmark.Prev.Vet.

Med79,98–115.

Martin,P.A.,Cameron,A.R.,Greiner,M.,2007b.Demonstratingfreedom fromdiseaseusingmultiplecomplexdatasources1:anewmethod- ologybasedonscenariotrees.Prev.Vet.Med.79,71–97.

Mudry,M.,Meylan,M.,Regula,G.,Steiner,A.,Zanoni,R.,Zanolari,P.,2010.

EpidemiologicalstudyofpestivirusesinSouthAmericancamelidsin Switzerland.J.Vet.Intern.Med.24,1218–1223.

Niskanen,R.,1993.Relationshipbetweenthelevelsofantibodiestobovine viraldiarrhoeavirusinbulktankmilkandtheprevalenceofcows exposedtothevirus.Vet.Rec.133,341–344.

Paisley,L.G.,Wells,S.,Schmitt,B.J.,1996. Prevalenceofbovineviral diarrheaantibodiesin256UScow-calfoperations.Asurvey.Therio- genology46,1313–1323.

Perez,M.J.,Wilks,C.R.,Rice,M.,1994.Antibodiestobovineviraldiarrhoea virusinbeefcattle.N.Z.Vet.J.42,73.

Radostitis,O.M.,Gay,C.C.,Blood,D.C.,Hinchcliff,K.W.,2000.Veterinary Medicine,ATextbookoftheDiseasesofCattle,Sheep,Pigs,Goatsand Horses.WBSaunders,HarcourtPublishersLtd.,London.

Stahl,K.,Alenius,S.,2012.BVDVcontrolanderadicationinEurope–an update.Jpn.J.Vet.Res.60(Suppl.1),S31–S39.

Svanova,2012.Svanovir®BVDB-Ab.Setsthestandardforcontrolling bovineviraldiarrhoeavirus(BVDV).Productsheet.

Åkerstedt,J.,Norström,M.,Mørk,T.,2012.Thesurveillanceandcontrol programmeforbovinevirusdiarrhoea(BVD)inNorway2011.In:

Sviland,S.,Hellberg,H.(Eds.),Surveillanceandcontrolprogrammes forterrestrialandaquaticanimalsinNorway.Annualreports2011.

NorwegianVeterinaryInstitute,Oslo,1–7.