An investigation of classical swine fever virus seroprevalence and risk factors in pigs in East Nusa Tenggara, eastern Indonesia.

(1)

An investigation of classical swine fever

virus seroprevalence and risk factors in

pigs in East Nusa Tenggara, eastern

Indonesia

ARTICLE

in

PREVENTIVE VETERINARY MEDICINE · FEBRUARY 2015

Impact Factor: 2.17 · DOI: 10.1016/j.prevetmed.2015.02.002

CITATIONS

2

READS

120

9 AUTHORS

, INCLUDING:

Gusti Ngurah Mahardika

Udayana University

42

PUBLICATIONS

110

CITATIONS

SEE PROFILE

Edwina Leslie

University of Sydney

5

PUBLICATIONS

17

CITATIONS

SEE PROFILE

Ian D Robertson

Murdoch University

191

PUBLICATIONS

3,465

CITATIONS

SEE PROFILE

Jenny-Ann L M L Toribio

University of Sydney

67

PUBLICATIONS

392

CITATIONS

SEE PROFILE

(2)

Please cite this article in press as: Sawford, K., et al., An investigation of classical swine fever virus 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

An

investigation

of

classical

swine

fever

virus

seroprevalence

and

risk

factors

in

pigs

in

East

Nusa

Tenggara,

eastern

Indonesia

Kate

Sawford

a,b

,

Maria

Geong

c

,

Petrus

M.

Bulu

d

,

Emily

Drayton

a

,

Gusti

N.K.

Mahardika

e

,

Edwina

E.C.

Leslie

a

,

Ian

Robertson

d

,

Anak

Agung

Gde

Putra

f

,

Jenny-Ann

L.M.L.

Toribio

a,∗

a_Farm_Animal_&_Veterinary_Public_Health,_Faculty_of_Veterinary_Science,_University_of_Sydney,₄₂₅_Werombi_Road,_Camden,

NewSouthWales2570,Australia

b_Greater_Sydney_Local_Land_Services,_Level_4,_2–6_Station_Street,_Penrith,_New_South_Wales_2750,_Australia

c_Animal_Health_&_Veterinary_Services,_Provincial_Department_of_Livestock,_East_Nusa_Tenggara_Province,_Kupang,_West_Timor,_Indonesia d_College_of_Medicine,_Murdoch_University,_South_Street_Murdoch,_Western_Australia_6150,_Australia

e_Animal_Biomedical_and_Molecular_Biology_Laboratory,_Faculty_of_Veterinary_Medicine,_Udayana_University,_Bali,_Indonesia f_Disease_{Investigation}_Centre_Denpasar,_Ministry_of_Agriculture,_Bali,_Indonesia

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received7April2014

Receivedinrevisedform29January2015 Accepted2February2015

Keywords:

Classicalswinefevervirus EastNusaTenggara Indonesia Epidemiology Riskfactors Seroprevalence

a

b

s

t

r

a

c

t

Classicalswinefevervirus(CSFV)isahighlyinfectiousdiseaseofpigs.Ithashadsignificant impactsonEastNusaTenggara,easternIndonesiasinceitsintroductionin1997.Inspiteof itsimportancetothisregion,littleisknownaboutitsseroprevalenceanddistribution,and pig-levelandfarmer-levelfactorsthatmayhaveanimpactontheserologicalstatusofan individualpig.Toaddressthisknowledgedeficit,across-sectionalseroprevalencesurvey wasconductedin2010involving2160pigsand805farmersfromfourislandsintheregion. Farmerquestionnairesandpigrecordformswereusedtocollectdataaboutthefarmers andpigssurveyed.BloodwascollectedfromeachpigtodetermineitsCSFVserological status.Apparentandtrueprevalencewerecalculatedforeachisland,district,subdistrict, andvillagesurveyed.CSFVserologicalstatuswasusedasanoutcomevariableinmixed effectslogisticregressionanalyses.

OveralltrueCSFVseroprevalencewasestimatedat17.5%(lowerCI16.0%;upperCI19.5%). Seroprevalenceestimatesvariedwidelyacrosstheislands,districts,subdistricts,and vil-lages.ManggaraiBarat,adistrictonthewesternendofFloresIsland,containedpigsthat werepositiveforantibodytoCSFV.Thisresultwasunexpected,asnoclinicalcaseshad beenreportedinthisarea.OlderpigsandpigsthathadbeenvaccinatedforCSFVwere morelikelytotestpositiveforantibodytoCSFV.Thefinalmultivariablemodelaccounted foralargeamountofvariationinthedata,howevermuchofthisvariationwasexplained bytherandomeffectswithlessthan2%ofthevariationexplainedbypigageandpigCSFV vaccinationstatus.

∗ _{Corresponding}_author._Tel.:₊₆₁₂₉₃₅₁_1609;_fax:₊₆₁₂₉₃₅₁_1618.

E-mailaddresses:kate.sawford@gmail.com(K.Sawford),geongmaria@yahoo.com(M.Geong),pietmb@gmail.com(P.M.Bulu),

edra4815@uni.sydney.edu.au(E.Drayton),gnmahardika@gmail.com(G.N.K.Mahardika),eles2597@uni.sydney.edu.au(E.E.C.Leslie),

I.Robertson@murdoch.edu.au(I.Robertson),aagputra@gmail.com(A.A.GdePutra),jenny-ann.toribio@sydney.edu.au(J.-A.L.M.L.Toribio).

http://dx.doi.org/10.1016/j.prevetmed.2015.02.002

(3)

Please cite this article in press as: Sawford, K., et al., An investigation of classical swine fever virus InthisstudywedocumentedtheseroprevalenceofCSFVacrossfourislandsinEastNusa Tenggara,easternIndonesia.Wealsoidentifiedriskfactorsforthepresenceofantibody toCSFV.FurtherinvestigationisneededtounderstandwhyclinicalCSFVhasnotbeen reportedonthewesternendofFloresIsland,andtoidentifyadditionalriskfactorsthat explainCSFVserologicalstatustoinformdiseasecontrolstrategies.

1. Introduction

Classicalswinefever virus(CSFV), orhog cholera, is a pestivirusassociated withhighmorbidity and mortal-ityratesin pigs.CSFVhasbeeneradicatedfromseveral countriesincludingtheUnitedStates,Australia,andNew Zealand,andanumberofcountriesinCentraland West-ernEurope(Artoisetal.,2002;Edwardsetal.,2000;World Organisationfor Animal Health,2014).However,recent outbreaksincountriespreviouslyfreeofCSFVindomestic pigshavehadsignificanteconomicandanimalhealth con-sequences(Elbersetal.,1999;Moennigetal.,2003;World OrganisationforAnimalHealth,2014),andCSFVremains endemicinpartsofAsia,CentralAmericaandSouth Amer-ica.

EastNusaTenggara(NTT)istheprovincewiththe high-estlevelofporkconsumptioninIndonesia,andthelargest pigpopulationwithanestimatedsizeof1.8million ani-mals(BPSStatistics,2013).Smallholderpigfarmers(total herdsizesof ≤20 pigs) are thepredominant producers inthisregion,with85%ofhouseholdsraisingpigs(Johns etal.,2009;Santhiaetal.,2006)andagricultureisthe pri-maryincomesourceforthemajorityofhouseholds(Wang, 2007).InNTT,pigsprovideafoodsourceandfinancial secu-rity,andarehighlyvaluedsociallyandculturally(Santhia etal.,2006;Leslieetal.,2014).Therefore,morbidityand mortalityeventsinthepigpopulationimpactalarge pro-portionofthehumanpopulation.

CSFVisahighlycontagioustransboundarydisease.Pigs aregenerallyinfectedoronasally,andspreadisbothdirect viahorizontalandverticaltransmission,andindirectvia contaminatedfomitesandporkproducts.Clinicaldisease causedbyCSFVisclassifiedasacute,subacute,orchronic, andisdeterminedbyCSFVstrain,aswellashostfactors, includingpigage,breed,stageofpregnancy,previousCSFV exposurestatus,and CSFVvaccination status.Thereare nopathognomonicsignsfor CSFV,andtherefore labora-torydiagnosticsarerequiredtomakeadiagnosis(Moennig etal.,2003).

CSFVwasconfirmedin NTTin1998.It subsequently spreadacrosstheprovince largelythroughuncontrolled livepigmovement,causing substantiallosses.It contin-uestolimitpigproductionintheregion(TriSatyaetal., 1999;Christie,2007).InNTT,districtsareclassifiedbyCSFV infectionstatus,whichisbasedonclinicalcasereportsto theNTTLivestockOfficeandlimitedgovernment-led sero-logicalsurveys.In2010,alldistricts onWestTimorand SumbaIslandandonedistrictontheeasternendof Flo-resIslandwereclassifiedasinfected;onedistrictineast FloresIslandwasclassifiedassuspect;andtherestof Flo-resIsland plusLembata Islandclassifiedasnot infected (Fig.1).Vaccinationcampaignsareconductedindistricts

withthehighestpigdensitiesandannualreportsofcases inanattempttocontroldisease.However,fluctuationsin thesizeofthepigpopulationhavecontinued,andtheNTT LivestockOfficehasdocumentedanincreaseinthe num-berofannualreportedcases(DinasPeternakanPropinsi, 2011).

Liveattenuated‘Chinese’strain(C-strain)vaccineCSFV vaccineisusedinNTTtocontroldisease.Itseffectiveness hasbeendemonstratedbyanumberofstudies,and pro-tection lastsatleast6–18monthsand maybelifelong (vanOirschot,2003).Neutralizingantibodyusuallyappears withintwoweeksandincreasesuntilatleast4–12weeks postvaccination(vanOirschot,2003).Antibodycan per-sistmanyyears afterinoculationwithasingledose,but also disappearsin someindividuals and may disappear atahigherrateunder‘real’fieldconditionscomparedto field trialconditions(vanOirschot,2003).It isgenerally acceptedthatthepresenceofneutralizingantibody con-fersCSFVprotection(Suradhatetal.,2007).Similarly,pigs thatrecoverfromacuteCSFVinfectiondevelop neutraliz-ingantibodyasearlyastwoweekspostinfection(Moennig, 2000).Theseanimalsareprotectedagainstfutureinfection forseveralyearsandimmunitymaybelifelong(Moennig, 2000).

InspiteoftheimportanceofCSFVtoNTT,littleisknown abouttheseroprevalenceanddistribution.Noserological surveyshavebeenconductedinmanypartsofNTT, includ-ingthewesternhalfofFloresIsland.Inconsistencieshave beennotedbetweenthenumberofCSFVcasesreported bytheNTTLivestockOfficeandthefewpublishedstudies (Santhiaetal.,2003;DinasPeternakanPropinsi,2011).It hasbeenrecognizedthatasaresultofgovernment decen-tralization,communicationbetweenandwithindifferent governmentsectorsislacking,whichmaybethecauseof datainconsistencies(BrandenburgandSukobagyo,2002). However,previousstudieshavealsonotedthat farmers acrossNTTarereluctanttoreportCSFVcases(Robertson et al., 2010; Deveridge, 2008).Moreover, Santhia et al. (2003)statedthatfarmersandanimalhealthworkerson AlorIslandinNTTwerenotreportingallCSFVcases.

(4)

Please cite this article in press as: Sawford, K., et al., An investigation of classical swine fever virus

Fig.1.ThedistributionandspreadofCSFVacrossNusaTenggaraTimur,easternIndonesiaupuntiltheendof2010.Thenumberscorrespondtothe followingevents:(1)thefirstsuspectedCSFVcasesreportedinSumbaTimurandFloresTimurdistrictsinmid1997;(2)thefirstdiagnosticlaboratory confirmedcasesofCSFVinNTTinMarch1998;(3)additionalcasesidentifiedin1999;(4)thefirstsuspectedCSFVcasesfromclinicalreportsfrominSikka districtin2000;and(5)thefirstcaseofCSFVdetectedonAlorIslandinJuly2002.AdaptedfrominformationobtainedfromTriSatyaetal.(1999),Christie (2007),Santhiaetal.(2003),andGeong,M(perscomm.,2011).TheclassificationofdistrictsaccordingtoCSFVinfectionstatusisillustratedusingboxes –redboxescontaindistrictsclassifiedasinfected,yellowboxescontaindistrictsclassifiedassuspect,andgreenboxescontaindistrictsclassifiedasnot infected.(Forinterpretationofthereferencestocolorinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)

2. Materialandmethods

ThesurveywasconductedasdescribedfromAprilto September2010followingapprovaltheUniversityof Syd-ney’sHumanResearchEthicsCommittee(08-2009/11866).

2.1. Questionnairedesign

Aquestionnairewasdevelopedtorecordinformation onfarmerdemographics,farmstructureandperformance, pighusbandry,reproductivemanagement,pigmovements, pighealthhistoryandresponse,andfarmerCSFV knowl-edgeandawareness,andvaccinationpractices.Apigrecord sheetwasdevelopedtorecordinformationregardingthe sex,age,healthinthelastthreemonths,source,body con-ditionscore(BCS),andCSFVvaccinationstatusofthepigs fromwhichabloodsamplewascollected.Bothconsisted ofopenand closedquestions.Closedquestionsincluded multiple choice, checklist, or short answer type ques-tions.Throughoutthefarmerquestionnaireandpigrecord forms,CSFVwasreferredtoashogcholeraasthistermis usedcommonlyinIndonesia.Ittookapproximately30min tocompletethefarmerquestionnairewitheach partici-pant.

ThedocumentsweredevelopedinitiallyinEnglishand thentranslatedintoBahasaIndonesiabyDr.MariaGeong, DirectorofLivestockServicesNTTandanativespeakerof localorigin.Veterinariansfromeachislandattendedajoint trainingeventduringwhichthefarmerquestionnaireand

pigrecordformwerepilottestedwith12pigownersin Kupang,whichallowedquestionrefinement.

2.2. Samplingstrategy

Amulti-stageapproachtosamplingwasusedtoselect thedistricts,subdistricts,villagesandfarmers.

2.2.1. Selectionofdistricts

Purposivesamplingwasusedtoselectdistrictswithin each island. District inclusion was based on reported clinicalcasesofCSFV(DinasPeternakanPropinsi,2011), geographicaldiversity,andperceivedhighimportanceof pigproductionandtradewithinthedistrictaccordingto LivestockServicesNTTveterinarians.

On West Timor Island, the districts Belu and Kota Kupangwere included.Bothhave a history of reported CSFVand vaccinationcampaignsfor CSFV.Belu borders TimorLeste andpigtradingacrosstheborder isknown tooccur.KotaKupangisthemainpig-producingdistrict onWestTimorIslandandincludestheprovincialcapital Kupang.

OnSumbaIsland,thedistrictsSumbaBaratDayaand SumbaTimurwereincluded.Bothhaveahistoryofclinical casesofCSFVandCSFVvaccinationcampaigns.

(5)

Please cite this article in press as: Sawford, K., et al., An investigation of classical swine fever virus district,andthereforeCSFVseroprevalencewasexpected

tobeverylow,orzero.SikkaisaCSFVsuspectdistrictas therehavebeenveryfewreportedcaseswithonlyonecase reportedfrom2002to2009,andthereforeCSFV seropreva-lencewasalsoexpectedtobelow.

ForLembataIsland,adistrictinitself,clinicalCSFVhad notbeenreportedpriortothestudyandtherehadbeenno CSFVvaccinationcampaigns.

2.2.2. Selectionofsubdistrictsandvillages

Simplerandomsamplingwasusedtoselecttwo sub-districtsperdistrictinFlores,WestTimor,andLembata, andonesubdistrictperdistrictinSumba.Subdistricts con-sidered remote, unsafe,or unlikely toco-operate were excludedfromthesamplingframe.Thenumber of sub-districtssampledwasbasedonlogisticalissues relating totimeandfundsavailable.Foreachselectedsubdistrict, simplerandomsamplingwasusedtoselectthreevillages persubdistrictinFloresandLembataandtwovillagesper subdistrictinSumbaandWestTimor.

2.2.3. Selectionoffarmers

InWestTimorandSumba,30farmersfromeach vil-lagecompletedthefarmerquestionnaire,whileinFlores andLembata20farmersfromeachvillagecompletedthe farmerquestionnaire.Foreachselectedvillage,asampling framewas constructed by obtaininga listof pig farm-ersfromtheVillageHead.LivestockServicesveterinarians requestedthis informationduring apreliminaryvisit to eachselectedvillage.Atthesametimepermissionto con-ductthesurveyinthevillagewasobtained.

Simplerandomsamplingwasusedtoselecttwentyto 50%morefarmersthanrequiredfromeachvillage.Extra farmers wereselected toensurea sufficient number of farmersweresurveyed.Farmershadtobepresentinthe villageonthedayoftheinterviewteamvisitto partici-pate.In FloresandLembata,farmersalsohad toownat leastthreepigsovertheageofthreemonths.InSumba, pigsownedbymultipleindividualswereoftengroupedin pensandunderthecareofasinglefarmer.Thereforein Sumba,farmershadtohaveatleastthreepigsovertheage ofthreemonthsundertheircaretoparticipate.InWest Timor,farmers hadtoown a minimumofone pigover threemonthsofagetoparticipate.Ifaselectedfarmerdid notmeettheselectioncriteriaorwasunwillingto partici-patethenextfarmerselectedduringtherandomsampling processwhometthecriteriareplacedthem.Farmerswere informedoftheirselectiononthedaypriortothevillage visit,andthereforeahighrateoffarmerattendanceinthe villagewasexpected.Farmerswereprovidedwithafree healthcheckoftheirpigsandadministrationof medica-tionsasrequiredasanincentive.

2.2.4. Selectionofpigs

InSumba,threepigsthreemonthsofageorolderwere selectedforbloodsamplecollectionfromeachinterviewed farmerusingconveniencesamplingforatotalof90pigs sampledpervillage.InWestTimor,1–4pigsgreaterthan threemonthsofagewereselectedfromeachinterviewed farmerusingconveniencesampling.IneachvillageinWest Timor,attheendoftheinterviewprocesstherewerefewer

than90pigssampledandthereforethedecisionwasmade tosamplepigsgreaterthanthreemonthsofagefrom farm-erswhodidnottakepartinthefarmerinterview.InFlores andLembata,threepigsgreaterthanthreemonthsofage wereselectedfromeachinterviewedfarmerusing conve-niencesamplingforatotalof60pigssampledpervillage. Previousstudieshaveshownthatmaternallyderived anti-bodylevelsreachaminimumlevelbyabout10weeksof age,andthereforeitwasassumedthatmaternalderived antibodywouldnotbepresentinpigsgreaterthanthree monthsofage(Klinkenbergetal.,2002a).

Table1listsbyislandthenamesofthestudydistricts, subdistricts,andvillages,andthenumberoffarmersand pigssampled.

2.2.5. Estimationofrequiredsamplesize

InSumbaandWestTimor,thenumberofpigsrequired toestimateCSFVantibodyprevalencewascalculatedwith Epitools(Sergeant,2010)using:(1)anexpected seropreva-lenceof20–30%basedontheexpertopinionofDr.Maria Geong;(2)avillagepigpopulationof2000pigsbasedon theexpertopinionofDr.MariaGeong;(3)alevelof pre-cisionof10%;(4)alevelofconfidenceof95%;and(5)an imperfecttestwith95%sensitivityand95%specificity.The samplesizerequiredrangedfrom81to99pigsdepending ontheexpectedprevalence,andthereforethemidrange valuewaschosen.

In Flores and Lembata, thenumber of pigs required to detect CSFV antibody was calculated with Epitools (Sergeant,2010)using:(1) aminimum expected preva-lence in the selected villages of 5%; (2) a village pig populationrangingfrom500to5000pigs;(3)alevelof confidence of 95%;and (4) an imperfect test with 95% sensitivityand100%specificity.Theseinputsresultedin a samplesizerangingfrom59 to62 pigsdependingon thevillagepigpopulationsize,andthereforethemidrange valuewaschosen.

2.3. Dataandsamplecollection

Thefarmerquestionnairewascompletedduringeach participantinterview.Thisprocesswasconductedpriorto bloodcollectionfromthesampledpigs.Thepigrecordform wascompletedaftersamplecollection.Eachfarmerandpig wasassignedauniqueidentificationcode.

Pigsweremanuallyrestrainedwithanosesnareand 3mlofbloodwascollectedfromthejugularveinusinga serumvacutainerand20-gaugeneedle.Whenbloodcould notbecollectedfromthejugularvein,a23-gaugeneedle and3mlsyringewereusedtocollectbloodfromthelateral earvein.Thissamplewasthenimmediatelytransferredto aserumvacutainer.

Vacutainerswerelabeledwiththecorrespondingpig identificationcodefromthepigquestionnaireandstored on ice during the sample collection period and trans-portationtotheLivestockServicesDepartmentLaboratory. Serumseparationwasperformedwithin12hofsample col-lection–sampleswerecentrifugedfor5minat80,000rpm and the serum transferred to a serum vacutainer and storedat−5◦_C._Samples_were_later_transported_on_ice_to

(6)

Table1

Totalnumberoffarmersandpigssurveyedacrosstheislands,districts,subdistricts,andvillagesinEastNusaTenggara,easternIndonesia,fromAprilto September2010.

Island District Subdistrict Village #offarmers(#ofpigs)

Flores

Sikka

Alok

KotaUneng 20(60)

Madawat 20(60)

Nangalimang 20(60)

Nita

Bloro 20(60)

Tilang 20(60)

Tebuk 20(60)

Manggarai Barat

Komodo

WaeKelambu 20(60)

BatuCermin 20(60)

GoloBilas 20(60)

Lembor

Tangge 20(60)

Amba 20(60)

PocoRutang 20(60)

Islandtotal 240(720)

Lembata Lembata

Nubatukan

Selandoro 20(60)

Lewoleba 20(60)

Bakalerek 20(60)

Lebatukan

Lamatuka 20(60)

Merdeka 20(60)

Waienga 20(60)

Sumba

Timur Haharu

Rambangaru 30(90)

Praibakul 30(90)

Sumba

BaratDaya Loura

Waitabula 30(90)

RadaMata 30(90)

West Timora

Belu

Tasifeto Barat

Naitimu 37(90)

Naekasa 41(90)

Atambua Selatan

Fatukbot 50(90)

Lidak 42(90)

Kota Kupang

Maulafa Sikumana 39(90)

Oepura 32(90)

Oebobo

Oebobo 39(90)

Oebufu 45(90)

Total 805(2160)

a_Thirty_farmers_per_village_in_West_Timor_completed_the_farmer_{questionnaire.}_The_number_of_farmers_in_the_table_indicates_the_number_of_farmers_that ownedthesampledpigs.

UniversityofUdayana,Denpasar,andstoredat−20◦_C_until

serologicalanalysis.

2.4. Serologicalanalysis

Serum samples were analyzed using a commercial CSFV enzyme-linked immunosorbent assay (ELISA) kit (PrioCHECK®_CSFV_Ab,_Lleydstat,_{Netherlands).}_ELISAs_were

performedaccordingtothemanufacturer’s instructions. ELISAplateswerereadusinga450nmfilteronanELISA readertodetermineopticaldensityandthesevalueswere usedtocalculatepercentinhibition(PI)(Colijnetal.,1997). AsamplewasconsideredpositiveforCSFVantibodywhen PIwas≥50%,inconclusivewhenPIwas31–50%and neg-ativewhenPIwas<30%(Colijnetal.,1997).Samplesthat hadhaemolysedorappearedcontaminatedwereincluded intheanalysis.

Serumsamplesthatwereclassifiedasinconclusiveor positivebythePrioCHECKCSFVAb,orforwhichtheresult wasmissing,werereanalyzedusingthePrioCHECK®_CSFV

Ab2.0.ThissecondgenerationELISAismorespecificfor CSFVantibodycomparedtootherpestivirusantibody,and thereforewasusedtoreducethelikelihoodthatpositive

resultswereduetocrossreactionwithantibodytoanother pestivirus.AsamplewasconsideredpositiveforCSFV anti-bodywhenPI≥40%,negativewhenPIwas<40%according tothemanufacturer’sinstructions.

AsamplethatwaspositiveforCSFVantibodyonthe Pri-oCHECKCSFVAbbutforwhichtherewasinsufficientserum toconductthePrioCHECKCSFVAb2.0wasconsidered posi-tiveforCSFVantibody.Samplesthathadnoresultavailable forthePrioCHECKCSFVAbbuttestedpositivewith Pri-oCHECKCSFVAb2.0weredeemedpositive,whilesamples thathad noresultavailablefor thePrioCHECKCSFVAb buttestednegativeonthePrioCHECKCSFVAb2.0were deemednegative.Samplesthathad aninconclusivetest resultwithPrioCHECKCSFVAbbuthadnoresultonthe PrioCHECKCSFVAb2.0weredeemedtohavenoresult.

2.5. Datamanagement

Datawereentered intotwodatabasescreatedinEpi InfoTM_Software_(version_3.5.1,_CDC,_{www.cdc.gov/epiinfo}_,

(7)

Please cite this article in press as: Sawford, K., et al., An investigation of classical swine fever virus identificationcode. Thedata were cleaned in Microsoft

ExcelandexportedforanalysisinR(version3.0.2.©_2013,

TheRFoundationforStatisticalComputing).

2.6. Calculationofapparentandtrueseroprevalence

Apparent seroprevalence and confidence intervals, usingthenormalapproximationinterval,werecalculated foreach island, district,subdistrict, andvillagein theR statistical package (prevalence, v 0.2.0). Estimated true seroprevalenceandconfidenceintervalswerecalculatedin theRstatisticalpackage(prevalence,v.0.2.0)(Roganand Gladen,1978).Testsensitivitywassetat89%and speci-ficityat100%. Thesetest performance parameters were calculatedinEpitools(Sergeant,2014)foruseinseriesof thePrioCHECKCSFVAb(sensitivity98%andspecificity99% determinedbyColijnetal.,1997;Moseretal.,1996)and thePrioCHECKCSFVAb2.0(sensitivity91%andspecificity 100%determinedbySchroederetal.,2012).

2.7. Riskfactoranalysis

2.7.1. Outcomeandexplanatoryvariables

Theunitofinterestwastheindividualpig.Theoutcome variablewasCSFVserologicalstatus.Pigswereclassifiedas eitherCSFVantibodypositiveorCSFVantibodynegativeas previouslydescribed.

Twenty-six explanatoryvariables werederived from thequestionnaires:20farmer-levelvariablesandsix pig-levelvariables.Numberofpigsonthefarmwastheonly continuousvariable.Theremainingexplanatoryvariables werecategorical,20of whichwerebinaryvariables. All explanatoryvariablesarepresentedinTable2.

2.7.2. Descriptiveanalyses

Contingencytableswerecreatedtoexplorethe relation-shipbetweeneachofthecategoricalexplanatoryvariables andCSFVserologicalstatus.Inaddition,summarystatistics werecalculatedforthenumberofpigsonfarm(theonly continuousexplanatoryvariable),bothaloneandaccording toCSFVserologicalstatus.

2.7.3. Univariableanalysis

Theassociationofeachexplanatoryvariablewiththe binaryoutcomevariablewasassessedusing univariable mixedeffectslogisticregressionanalysesintheRstatistical package(lme4,v.1.0-5).Tocontrolfortheeffectof clus-tering,farmer,village,subdistrict,district,andislandwere fittedseparatelyasrandomeffects.Basedonthe associa-tionbetweeneachexplanatoryvariableandtheoutcome variable,allexplanatoryvariableswithap-valueof≥0.20 wereexcludedfromthemultivariableanalyses.In addi-tion,variableswithmorethan10%ofmissingvalueswere excludedfrommultivariableanalyses(Dohooetal.,2009, pp.369).

Accordingto expert opinion fromLivestock Services NTT,farmerswithherdsizesof1–3pigsgenerallykeeppigs forhomeconsumptionoruseintraditionalceremonies, whilefarmerswithlargerherdswereresponsibleforthe majorityofpigmovementsintoandoutofavillage. There-fore,thedecisionwasmadetoexcludethedatafromWest

Timorfromtheunivariableandmultivariableanalysis.The datafromLembatawerealsoexcludedasallsamplestested negativeforCSFVantibody.

2.7.4. Multivariableanalyses

Amultivariablemixedeffectslogisticregressionmodel was constructed using the R statistical package (lme4, v.1.0-5) witha manual backward stepwiseapproach to evaluatetheassociationofexplanatoryvariableswiththe outcomevariableafteradjustingforeachother.Variables that were statistically significant (p-value<0.05) were retainedinthefinalmodel.Thecorrelationbetween covari-ates wasevaluatedusing achi-square testand deemed significantatap-valueoflessthan0.05.A2-wayinteraction betweenageandvaccinatedforCSFVwastestedwithinthe multivariablemodel.

Tocontrolfortheeffectofclustering,farmer,village, subdistrict,district,and island werefittedseparately as randomeffects.Goodness-of-fitofthefinallogistic regres-sionmodelwasassessedbycalculatingconditionalR2_for

thefinalmodel(R2

GLMM(c)).Theamountofvariationinthe

dataexplainedbythefixedeffectswasassessedby calculat-ingmarginalR2_for_the_fixed_effects₍_R2

GLMM(m))(Nakagawa

andSchielzeth,2013).

2.7.5. Intra-classcorrelationcoefficient

Intra-classcorrelation(ICC)wascalculatedforeach ran-domeffectusingthelatentvariableapproachtoquantify theamountofclusteringbetweenunitsateachofthe dif-ferentlevelsofclustering(Browneetal.,2005).Clustering wasdeemedhighforrandomeffectsthathadanICCgreater than0.3(Dohooetal.,2009,pp.537,583).

3. Results

Sevenhundredandtwentyfarmersand2160pigsfrom 805farmersweresurveyedacrossthefourislands.Herd sizerangedfrom1to48pigs,withanaverageof4.6pigs. Onethousandfourhundredfifty-two(67.2%)ofthe2160 pigsincludedinthesurveywereborninthefarmer’sherd. Approximately42%(898/2160)ofthepigssurveyedwere 3–5monthsofage,30%(652/2160)were6–11monthsof age,andtheremaining28%(610/2160)wereequaltoor greaterthan12monthsofage.Fivepercent(113/2160)of thepigshadbeensickinthethreemonthspriortothetime ofthesurvey.

3.1. Seroprevalence

(8)

Table2

ExplanatoryvariablesanalyzedforassociationswithCSFVserologicalstatusamongst1080pigsfrom360farmerssurveyedinEastNusaTenggara,eastern Indonesia,fromApriltoSeptember2010.

Variablegroup Variables

Pig-levelvariables Sexb_;_Agea_;_Vaccinated_for_CSFV;_Sick_in_the_last_three_months;_Source_of_pigb_;_Body_condition_scorea

Farmerlevelvariables Cattleonfarm;Goatsonfarm;Buffaloonfarm;Numberofpigsc_;_Pigs_free_to_roam;_Pigs_fed_swill;_Pigs_fed agriculturalwaste;Cookswill;Litterswithdeadpigletsbeforeweaningb_;_Pigs_introduced_in_the_last₁₂_months; Pigsexitedinthelast12months;Useownboarforbreeding;Pigsslaughteredathome;Suddendeathofpigsin thelastthreemonths;Sickpigsthatdiedinthelastthreemonths;Livestockservicescontactedintheeventofa sickpig;Pigshavecontactwithotherpigsoutsidetheherdb_;_Heard_of_CSFV;_Pigs_vaccinated_for_CSFV

Allvariablesarebinary(1–yes,2–no)unlessotherwiseindicated. a_Ordinal_variable.

b_Categorical_variable. c_Continuous_variable.

whichtherewasnoresultonthePrioCHECKCSFVAb,two testednegativeandtwotestedpositiveonthePrioCHECK CSFVAb2.0,while27hadnoresultavailable.Therefore serologicalfindingswereavailablefor2131ofthe2160 sample collected. Overall apparentCSFVseroprevalence acrossthefourislandswas15.8%(95%CI14.3,17.4),while overalltrueCSFVseroprevalencewasestimatedat17.5% (95%CI16.0, 19.5).Apparent prevalenceand true preva-lence estimatesacrosstheislands,districts, subdistricts, andvillagesarepresentedinTable3.

3.2. Univariablemixedeffectslogisticregressionanalyses

Elevenvariables wereassociatedwithCSFV serologi-calstatusattheunivariablecut-off p-valueof<0.20,six variablesatthepiglevelandfivevariablesatthefarmer level(Tables4–7).Thevariables‘Useownboarfor breed-ing’and‘Bodyconditionscore’wereexcludedduetotoo manymissingresponses.Atotalofelevenvariableswere consideredinmultivariableanalyses.

3.3. Multivariablemixedeffectslogisticregression analyses

ThefinalmodelforCSFVserologicalstatusispresented inTable8.Thetwovariablesinthefinalmodelwereboth pig-level characteristics.Pigs witha previoushistory of vaccinationforCSFVwere3timesmorelikelytotest pos-itive forantibodytoCSFV.Pigsequal toor greaterthan 12monthsofagewere2.5timesaslikelytotestpositive forantibodytoCSFVcomparedtopigs3–5monthsofage, whilepigs6–11monthsofagewereequallyaslikelytotest positiveforantibodytoCSFVcomparedtopigs3–5months ofage(Table8).

Thevariables‘Age’and‘VaccinatedforCSFV’were sig-nificantly correlated. However,pigs 3–5months of age weremorelikelytobevaccinatedforCSFVcomparedtothe othertwoagecategories,andtheproportionofpigs6–11 monthsofagevaccinatedforCSFVwassimilartothatof pigs≥12monthsofage.Thereforebothvariableswereleft inthemultivariablemodel.Theinteractiontermforageand vaccinatedforCSFVwasnotsignificant.Noneoftheother variablesinthefinalmodelweresignificantlycorrelated.

The conditional R2 _value _for _the _overall _model _was

0.638,thoughthemarginalR2 _value_for_the_fixed_effects

was0.0181,indicatingthatthefixedeffectsaccountedfor 1.8%ofthevariationinthedata.

3.4. Intraclasscorrelationcoefficient

ThevariancesandICCsforthefiverandomeffectterms areshowninTable9.Thedatawerehighlyclusteredatthe subdistrict,village,andfarmerlevels.

4. Discussion

CSFVisahighlyinfectiousdiseaseofpigswithmajor animal health and economic consequences. In regions whereCSFVis endemic,thefirststeptocontrollingthe diseaseistounderstanditsseroprevalenceand distribu-tion.ThereforeweundertookaCSFVseroprevalenceand riskfactorstudyintheNTTprovinceofeasternIndonesia. InthesamplesizecalculationsforCSFVantibody preva-lence,theinputsfor testsensitivityandspecificitywere estimated to be lower than the values reportedin the literature(Colijnetal.,1997;Moseretal.,1996).The deci-siontousemoreconservativeestimateswasbasedonthe factthat theperformance of first-generationCSFV anti-bodyELISAshasnotbeenevaluatedinIndonesia,andthe desiretoensurethatasufficientnumberofpigswere sam-pledduringeachvillagevisit.Thedecisiontousethe first-andsecond-generationCSFVantibodyELISAsinserieswas madeaftertheunexpectedresultfromFloresIsland.The numberofpigsrequiredtoestimateCSFVantibody preva-lencewassufficientata levelof precisionof10% anda levelofconfidenceof95%giventhattestsensitivitywas 89%andtest specificitywas100%forthetestsin series (Sergeant,2014).In2010,Epitoolsassumedatest speci-ficityof100%whencalculatingthesamplesizerequired fordiseasedetection,whichisthereasonthattest speci-ficitywassetat100%whendeterminingthenumberofpigs tobesampledonFloresandLembata.Asamplesizeof60 wassufficientfordetectingCSFVantibodyataminimum expectedprevalenceof6%,assumingavillagepig popula-tionof5000,alevelofconfidenceof95%,andanimperfect testwith89%sensitivity and100% specificity(Sergeant, 2014).

(9)

Table3

ApparentprevalenceandtrueprevalenceofCSFVantibodyacrosstheislands,districts,subdistricts,andvillagessurveyedinEastNusaTenggara,eastern Indonesia,fromApriltoSeptember2010.

Island District Subdistrict Village Numberof

sampleswitha testresult

Number positive

Apparent prevalence

True prevalence

Flores 706 55 7.8(5.9,10.1) 8.9(6.8,11.2)

Sikka 357 15 4.2(2.5,7.0) 5.0(2.9,7.7)

Alok 179 6 3.4(1.4,7.5) 4.3(1.8,8.0)

KotaUneng 60 1 1.7(0.087,10.1) 3.6(0.46,10.0)

Madawat 60 0 0(0,7.5) 1.8(0,5.4)

Nangalimang 59 5 8.5(3.2,19.4) 11.1(4.3,20.7)

Nita 178 9 5.1(2.5,9.7) 6.3(3.0,10.6)

Bloro 59 4 6.8(2.2,17.3) 9.2(3.1,18.0)

Tilang 59 0 0(0,7.6) 1.8(0,5.5)

Tebuk 60 5 8.3(3.1,19.1) 10.9(4.1,20.3)

Manggarai Barat

349 40 11.5(8.4,15.4) 13.1(9.6,17.2)

Komodo 177 0 0(0,2.6) 0.63(0,1.9)

WaeKelambu 59 0 0(0,7.6) 1.8(0,5.4)

BatuCermin 59 0 0(0,7.6) 1.8(0,5.4)

GoloBilas 59 0 0(0,7.6) 1.8(0,5.4)

Lembor 172 40 23.3(17.3,30.4) 26.4(19.7,33.7)

Tangge 56 12 21.4(12.0,34.8) 25.3(14.4,38.4)

Amba 59 22 37.3(25.3,50.9) 42.3(29.3,56.2)

PocoRutang 57 6 10.5(4.4,22.2) 13.3(5.6,23.8)

Lembata 360 0 0(0,1.3) 0.31(0,0.94)

Nubatukan 180 0 0(0,2.6) 0.6(0,1.8)

Selandoro 60 0 0(0,7.5) 1.8(0,5.4)

Lewoleba 60 0 0(0,7.5) 1.8(0,5.4)

Bakalerek 60 0 0(0,7.5) 1.8(0,5.4)

Lebatukan 180 0 0(0,2.6) 0.6(0,1.8)

Lamatuka 60 0 0(0,7.5) 1.8(0,5.4)

Merdeka 60 0 0(0,7.5) 1.8(0,5.4)

Waienga 60 0 0(0,7.5) 1.8(0,5.4)

Sumba 358 139 38.8(33.8,44.1) 43.7(38.2,49.5)

SumbaTimur 179 60 33.5(26.8,41.0) 37.8(30.2,45.9)

Haharu 179 60 33.5(26.8,41.0) 37.8(30.2,45.9)

Rambangaru 89 35 39.3(29.3,50.3) 44.5(33.7,56.0)

Praibakul 90 25 27.8(19.1,38.4) 31.7(21.9,42.7)

SumbaBarat Daya

179 79 44.1(36.8,51.7) 49.7(41.7,58.0)

Loura 179 79 44.1(36.8,51.7) 49.7(41.7,58.0)

Waitabula 90 36 40.0(30.0,50.9) 45.2(34.4,56.8)

RadaMata 89 43 48.3(37.7,59.1) 54.3(42.7,66.0)

WestTimor 707 142 20.1(17.2,23.3) 22.7(19.5,26.1)

Belu 352 83 23.6(19.3,28.4) 26.6(21.8,31.7)

Tasifeto Barat

178 49 27.5(21.2,34.8) 31.2(24.2,38.7)

Naitimu 90 18 20.0(12.6,30.0) 23.2(14.6,32.9)

Naekasa 88 31 35.2(25.5,46.2) 39.9(29.1,51.5)

Atambua Selatan

174 34 19.5(14.1,26.4) 22.4(16.2,29.4)

Fatukbot 87 22 25.3(16.8,35.9) 29.1(19.6,39.5)

Lidak 87 12 13.8(7.6,23.2) 16.4(9.1,25.4)

KotaKupang 355 59 16.6(13.0,21.0) 18.9(14.7,23.5)

Maulafa 178 9 5.1(2.5,9.7) 6.2(3.0,10.5)

Sikumana 89 3 3.4(0.87,10.2) 4.9(1.4,10.6)

Oepura 89 6 6.7(2.8,14.6) 8.6(3.6,15.6)

Oebobo 177 50 28.2(21.9,35.6) 32.0(24.8,39.7)

Oebobo 88 17 19.3(12.0,29.4) 22.5(13.9,32.7)

Oebufu 89 33 37.1(27.3,48.0) 42.0(31.3,53.4)

towardlargerherdsinFlores,Lembata,andSumbacould haveimpactedtheseroprevalencecalculations.Giventhat largerherdsareresponsibleforthemajorityofpig move-mentsintoandoutofavillage,seroprevalenceestimates fortheislandsofFlores,LembataandSumbamaybehigher

thanthosethatwouldhavebeencalculatedhadfarmers withfewerthanthreepigsoverthreemonthsofagebeen eligibletoparticipate.

(10)

Table4

Descriptiveresultsforpig-levelexplanatoryvariablessignificantlyassociated(p<0.20)a_with_CSFV_serological_status_amongst₁₀₈₀_pigs_from₃₆₀_farmers surveyedinEastNusaTenggara,easternIndonesia,fromApriltoSeptember2010.

Variables Categories CSFVserologicalstatus Total

Negative(row%) Positive(row%)

Age

3–5months 242(80%) 59(20%) 301

6–11months 310(88%) 44(12%) 354

≥12months 318(78%) 91(22%) 409

Sex

Male 224(84%) 42(15%) 226

Female 516(80%) 128(20%) 664

Malecastrated 123(85%) 22(15%) 145

Sickinthelastthreemonths

Yes 37(70%) 16(30%) 53

No 829(82%) 177(18%) 1006

Pigsource

Borninyourherd 507(82%) 109(18%) 616

Other 362(81%) 83(19%) 445

Bodyconditionscore

1 28(65%) 15(35%) 43

2 95(75%) 32(25%) 127

3 228(90%) 25(10%) 253

4 392(91%) 41(9%) 433

5 27(93%) 2(7%) 29

VaccinatedforCSFV

Yes 82(54%) 70(46%) 152

No 787(87%) 120(13%) 907

CSFV–classicalswinefevervirus.

a_All_pig-level_variables_had_p_-values_<0.20.

eveninthedistrictofKotaKupang,anareawithahistory ofclinicalreportsofCSFVandCSFVvaccinationcampaigns, village-leveltrueseroprevalenceestimatesrangedfrom5% inSikumanato42%inOebufu.Thisfindingissignificant becauseit showsthat CSFVseroprevalenceisdissimilar withinandbetweenvariouslevelsofspatialaggregation.

Further,it demonstrates that in areas where CSFV vac-cinationcampaignshave beenundertaken,thelevelsof herdimmunityrequiredtocontroldiseasearenotbeing achieved(Klinkenbergetal.,2002b).

Ofthe1080pigsincludedintheunivariableand multi-variableanalysis,152werereportedvaccinatedforCSFV.

Table5

Descriptiveresultsforfarmer-levelexplanatoryvariablessignificantlyassociated(p<0.20)a_with_CSFV_serological_status_amongst₁₀₈₀_pigs_from₃₆₀ farmerssurveyedinEastNusaTenggara,easternIndonesia,fromApriltoSeptember2010.

Variables Categories Totalherds CSFVserologicalstatus Totalpigs

Negative(row%) Positive(row%)

Cookswill

Yes 114 268(79%) 72(21%) 340

No 241 558(83%) 121(17%) 709

Pigsslaughteredathome

Yes 125 274(73%) 100(27%) 374

No 231 588(87%) 90(13%) 678

Sickpigsthatdiedinthelastthreemonths

Yes 17 33(65%) 18(35%) 51

No 338 825(83%) 173(17%) 998

Livestockservicescontactedintheeventofasickpig

Yes 193 512(90%) 57(10%) 569

No 164 350(72%) 136(28%) 486

PigsvaccinatedforCSFV

Yes 59 98(56%) 77(44%) 175

No 299 767(87%) 116(13%) 883

CSFV–classicalswinefevervirus.

a_Variables_with_p_>_0.20_not_included_in_this_table:_Cattle_on_farm;_Buffalo_on_farm;_Goats_on_farm;_Number_of_pigs;_Pigs_free_to_roam;_Pigs_fed_swill;_Pigs

(11)

Table6

Univariablemixedeffectslogisticregressionresultsforpig-levelvariablesassociatedwithCSFVserologicalstatus(p<0.20)a_amongst₁₀₈₀_pigs_from₃₆₀ farmerssurveyedinEastNusaTenggara,easternIndonesia,fromApriltoSeptember2010.

Variablesandcategories ˇ SE(ˇ) Oddsratio LCL(OR) UCL(OR) p-Valueb

VaccinatedforCSFV <0.0001

No – – 1.0 – –

Yes 1.26 0.32 3.51 1.87 6.58

Age <0.001

3–5months – – 1.0 – –

6–11months 0.0160 0.304 1.02 0.560 1.843

≥12months 1.023 0.298 2.78 1.551 4.989

Sickinthelastthreemonths 0.005

No – – 1.0 – –

Yes −0.38 0.42 0.68 0.298 1.555

Bodyconditionscore <0.0001

1 – – 1.0 – –

2 −0.252 0.446 0.78 0.32 1.86

3 0.889 0.610 2.43 0.735 8.05

4 1.144 0.727 3.14 0.756 13.05

5 0.521 1.117 1.68 0.189 15.03

Sourceofpig 0.023

Other – – 1.0 – –

Borninyourherd −0.00146 0.280 0.999 0.577 1.73

Sex 0.0116

Female – – 1.0 – –

Male −0.124 0.249 0.884 0.542 1.44

Castratedmale −0.766 0.314 0.465 0.251 0.861

a_All_pig-level_variables_had_p_-values_<0.20.

b _p_-Values_based_on_likelihood_ratio_X2_-test_of_{significance.}

Table7

UnivariablemixedeffectslogisticregressionresultsforfarmerlevelvariablesassociatedwithCSFVserologicalstatus(p<0.20)a_amongst₁₀₈₀_pigs_from 360farmerssurveyedinEastNusaTenggara,easternIndonesia,fromApriltoSeptember2010.

Variables ˇ SE(ˇ) Oddsratio LCL(OR) UCL(OR) p-Valueb

PigsvaccinatedforCSFV 0.000186

No – – 1.0 – –

Yes 0.988 0.306 2.686 1.474 4.893

Livestockservicescontactedintheeventofasickpig 0.0032

No – – 1.0 – –

Yes −0.284 0.316 0.752 0.405 1.39

Sickpigsthatdiedinthelastthreemonths 0.00892

No – – 1.0 – –

Yes 0.0498 0.446 1.051 0.438 2.52

Pigsslaughteredathome 0.00040

No – – 1.0 – –

Yes 0.380 0.277 1.25 0.850 2.52

Cookswill 0.0020

No – – 1.0 – –

Yes −0.029 0.424 0.971 0.423 2.23

a_Variables_with_p_>_0.20_not_included_in_this_table. b _p_-Values_based_on_likelihood_ratio_X2_-test_of_{significance.}

Table8

Finalmultivariablemixedeffectslogisticregressionmodel(p<0.05)forCSFVserologicalstatusamongst1080pigsfrom360farmerssurveyedinEastNusa Tenggara,easternIndonesia,fromApriltoSeptember2010.

Variables ˇ SE(ˇ) Oddsratio LCL(OR) UCL(OR) p-Valuea

Fixedeffects

Intercept −3.31 1.02 – – – 0.0012

Age <0.001

3–5months – – 1.0 – –

6–11months −0.044 0.31 0.96 0.524 1.75

≥12months 0.925 0.30 2.52 1.40 4.54

VaccinatedforCSFV <0.001

No – – 1.0 – –

Yes 1.15 0.32 3.17 1.68 5.98

N=1059,Log-likelihood=−374.41;d.f.=9;p<0.001;Goodness-of-fitR2-teststatisticR2

(12)

Table9

Variancesandintraclasscorrelation(ICC)coefficientsforeachrandomeffectterminthefinalmultivariablemixedeffectslogisticregressionmodelfor CSFVserologicalstatusamongst1080pigsfrom360farmerssurveyedinEastNusaTenggara,easternIndonesia,fromApriltoSeptember2010.

Randomeffectterm Numberinlevel Variance ICC

Island 2 0.591 0.0645

District 4 0.00056 0.0646

Subdistrict 6 3.504 0.447

Village 16 0.687 0.522

Farmer 359 1.090 0.641

Ofthepigsreportedvaccinated,only46%(70/152)tested positiveforCSFVantibody(Table4).Thereareanumber of factors that could be contributing to this low sero-conversionrate.Maternallyderivedantibodyisthemost commoncauseofCSFVvaccinationfailure,particularlyin highlyendemicareas (Suradhatetal.,2007),and there-forepigletsthathavecirculatingmaternalantibodymay notseroconvertwhenvaccinated. Thisinterferencemay beparticularlyimportantinNTTwherefarmersmaynot activelyweanpiglets,prolongingthetimeduringwhich pigletsnursefromthesow.Alternatively,vaccinestorage anddeliverymaynotbeadequateforachievingthehigh levelsofefficacyreportedfortheC-strainvaccineinthe literature(vanOirschot,2003).Further,itispossiblethat farmersintheregionarenotaccuratelyreportingtheCSFV vaccinationstatusoftheirpigs.Finally,co-infectionwith otherpathogens(e.g.,pseudorabies,porcinereproductive andrespiratorysyndromevirus),aswellasmycotoxinsand chemicalsareknowntointerferewithCSFVvaccination (Suradhatetal.,2007),andtheirnegativeimpactinthis regioncannotberuledout.

None of the farmer-level variables were significant determinantsofCSFVserologicalstatus,andonlytwo pig-levelfactorswereincludedinthefinalmultivariablemodel (Table8).Pigsequaltoorgreaterthan12monthsofage were more likely to test positive for antibody toCSFV. Thisresult is expected,as olderanimalsare both more likelytohavebeenexposedtoCSFVandvaccinated dur-ing a campaign. Pigsthat were reported vaccinated for CSFVwerethreetimesaslikelytotestpositiveforCSFV antibodycomparedtothosethatwerenotreported vac-cinated.Thisresultisexpected,andimportantlysuggests thatthevaccinationcampaignsundertakenbyNTT Live-stockServicesarecontributingtoCSFVseroprevalencein theregion.

Incontrasttotheexpectationthatolderpigsaremore likely to bereported vaccinated, in this studypigs 3–5 monthsweremorelikelytobereportedvaccinated com-paredtopigsintheothertwoagecategories.Thisfinding could be the result of reporting bias whereby farmers are more likely to recall the vaccination of youngpigs comparedtoolderpigs,inparticularbecausevaccination of younger pigs would have had to occurin the more recentpastandthereforemightbemorememorable. Alter-natively, vaccinated pigs maybehealthiercompared to unvaccinatedpigsandthereforemaybemorelikelytobe slaughteredoncetheyreachthe6–11monthor≥12month agecategory.Finally,farmerscouldassumethatany injec-tionisaCSFVvaccination,andtherefore‘CSFVvaccination status’mayinfactrepresent‘TreatmentbyNTTLivestock Services’.

The final multivariable model accounted for a large amountof thevariation in thedata, however thefixed effects,pigageandCSFVvaccinationstatus,accountedfor only1.8%ofthevariation.HighICCsatthefarmer,village, andsubdistrictindicatethedatawerehighlyclustered.It maybethatunmeasuredfactorsatanyoneofthesethree levelsofspatialaggregationbetterexplainCSFV serolog-icalstatuscompared tothepigandfarmer-level factors exploredinthisstudy.ThisfindingisofrelevancetoCSFV controleffortsintheregionbecauseitsuggeststhat inter-ventionsmaybebestimplementedatthesubdistrictlevel toaccount for thehighlevel of clustering atthis level, aswellasthelevelofvillageandfarmer.Arecentsocial network analysisfoundthatthemajorityof formaland informalpigmovementsoccurbetweensubdistricts, lend-ingfurthersupportforinterventionatthesubdistrictlevel (Leslie,2012).

The C-strain vaccine has a number of advantages, includingearlyonsetofCSFVimmunityandfull protec-tionagainstverticaltransmission(Suradhatet al.,2007; Schroeder et al., 2012). However, one of its disadvan-tagesisthattheantibodyresponseitinducescannotbe differentiatedfromthatcausedbyCSFVinfection.While marker vaccines have been developed in the hope of enablingdifferentiationofinfected fromvaccinated ani-mals,thesevaccinesarelessprotectiveandtheimmune response is delayed when compared to the C-strain vaccine (Suradhat et al., 2007; Schroeder et al., 2012). Additionally,theantibodyELISAsdevelopedas accompa-nying markertestshave beenshown tolacksensitivity (Schroeder et al., 2012). These characteristics of CSFV vaccines and ELISAs impact significantly our ability to understand the epidemiology of the virus in the field. Future studies should consider using additional detec-tiontechniques,includingpolymerasechainreaction,virus genotyping,andsentinelpigs,tobetterunderstandCSFV herdimmune status,pathogenesis,and epidemiologyin NTT.

(13)

Please cite this article in press as: Sawford, K., et al., An investigation of classical swine fever virus mayshowfewclinicalsignsand recover.CSFV

vaccina-tionisnotpermittedinthisarea(TriSatyaetal.,1999), howeveritsoccurrencecannotberuledout.Finally,while movementofpigsfromCSFVinfectedtosuspector unin-fectedareasisnotpermittedinIndonesia(TriSatyaetal., 1999),illegalmovementsfromcentraldistrictstowestern districtsonFloresareknowntooccur(Leslie,2012).While suchmovementscouldresultintheintroductionofCSFV intoManggaraiBarat,itcouldalsoresultinthepresenceof vaccinatedpigsintheregion.

5. Conclusions

In2010,LembataIslandwasconfirmedfreeofantibody toCSFV.However,thedistrictofManggaraiBaratonthe westendofFloresIslandcontainedpigsthatwere seropos-itiveforCSFV,inspiteofnoreportsofclinicalCSFVandno government-ledvaccinationprogramsinthis region.Pig ageandCSFVvaccinationstatuswereassociatedwithCSFV serologicalstatus,witholderpigsandpigsvaccinatedfor CSFVmorelikelytotestpositiveforantibodytoCSFV.Our resultsindicatethatfurtherresearchtoidentifythestrains ofcirculatingCSFVanddeterminetheeffectivenessof dis-easecontrolstrategiesis required.Such activitieswould contributetoabetterunderstandingofCSFVepidemiology inNTT.

Acknowledgements

The Australian Centre for International Agricultural Research(ACIAR)fundedthisstudyundertheAH/2006/156 project.TheauthorsgratefullyacknowledgetheNTT Live-stockOfficeinKupang,thevillageleadersandparticipating farmers for their collaboration. The authors are also indebtedtotheveterinariansandfieldstaffwhoassisted withthesamplecollectionandsurveyadministration.The assistanceofNavneetDhandandPeterThompson(Faculty ofVeterinaryScience,UniversityofSydney)wascriticalto thedataanalysisandmuchappreciated.

References

Artois,M.,Depner,K.R.,Guberti,V.,Hars,J.,Rossi,S.,Rutili,D.,2002. Clas-sicalswinefever(hogcholera)inwildboarinEurope.Rev.Sci.Tech. 21,287–303.

BPSStatistics,2013.NusaTenggaraTimurDalamAngka2013,Integration ProcessingandStatisticalDisseminationDivision.Publication Num-ber53000.06.02.BadanPusatStatistikOfficeofNusaTenggaraTimur Province,Indonesia.

Brandenburg,B.,Sukobagyo,P.,2002.AnUpdateonLivestockSector PerformanceinResponsetotheEconomicCrisis,Government Decen-tralizationandLocalAutonomy.IndonesiaLivestockSectorStudy, Document37565. DirectorGeneralof LivestockServices,Jakarta, Indonesia.

Browne,W.J.,Subramanian,S.V.,Jones,K.,Goldstein,H.,2005.Variance partitioninginmultilevellogisticmodelsthatexhibitoverdispersion. J.R.Stat.Soc.A:Stat.Soc.168,599–613.

Christie,B.(Ed.),2007.ACIARTechnicalReports65:AReviewof Ani-malHealthResearchOpportunitiesin NusaTenggaraTimurand NusaTenggaraBaratProvince,EasternIndonesia.ACIAR,Canberra, Australia.

Colijn,E.O.,Bloemraad,M.,Wensvoort,G.,1997.AnimprovedELISAforthe detectionofserumantibodiesdirectedagainstclassicalswinefever virus.Vet.Microbiol.59,15–25.

Deveridge,A.,(Honoursthesis)2008.ChallengesintheDevelopmentof aLivestockDiseaseSurveillanceProgramonAlor,EasternIndonesia. SchoolofNaturalSciences,UniversityofWesternSydney,Sydney.

DinasPeternakanPropinsi,2011.DinasPeternakan(ProvincialLivestock Service)AnnualReport2011.DinasPeternakanPropinsi,KotaKupang, Indonesia.

Dohoo,I.,Martin,W.,Stryhn,H.,2009.VeterinaryEpidemiologicResearch. VERInc.,Charlottetown,Canada.

Edwards,S.,Fukusho,A.,Lefevre,P.C.,Lipowski,A.,Pejsak,Z.,Roehe,P., Westergaard,J.,2000.Classicalswinefever:theglobalsituation.Vet. Microbiol.73,103–119.

Elbers,A.R.W.,Stegeman,A.,Moser,H.,Ekker,H.M.,Smak,J.A.,Pluimers, F.H., 1999. The classical swine fever epidemic 1997–1998 in the Netherlands: descriptive epidemiology. Prev. Vet. Med. 42, 157–184.

Johns,C.,Cargill,C.,Patrick,I.,Geong,M.,Ly,J.,Shearer,D.,2009. Small-holderCommercialPigProductioninNTT–OpportunitiesforBetter MarketIntegration.SADIFinalReport.AustralianCentrefor Interna-tionalAgriculturalResearch,Canberra.

Klinkenberg,D.,Moormann, R.J.M.,de Smit,A.J.,Bouma,A.,de Jong, M.C.M.,2002a.Influenceofmaternalantibodiesonefficacyofa sub-unitvaccine:transmissionofclassicalswinefevervirusbetweenpigs vaccinatedat2weeksofage.Vaccine20,3005–3013.

Klinkenberg,D.,deBree,J.,Laevens,H.,deJong,M.C.,2002b. Within-andbetween-pentransmissionofclassicalswinefevervirus:anew methodtoestimatethebasicreproductionratiofromtransmission experiments.Epidemiol.Infect.128,293–299.

Leslie,E.C.C.,(DoctorofPhilosophyPh.D.)2012.PigMovementsAcross Eastern Indonesiaand Associated Risk of Classical Swine Fever Transmission. UniversityofSydney,Sydney http://hdl.handle.net/ 2123/9316

Leslie,E.E.C.,Geong,M.,Abdurrahman,M.,Ward,M.P.,Toribio,J.-A.L.M.L., 2015.Adescriptionofsmallholderpigproductionsystemsineastern Indonesia.Prev.Vet.Med.118,319–327.

Moennig,V.,2000.Introductiontoclassicalswinefever:virus,diseaseand controlpolicy.Vet.Microbiol.73,93–102.

Moennig,V.,Floegel-Niesmann,G.,Greiser-Wilke,I.,2003.Clinicalsigns andepidemiologyofclassicalswinefever:areviewofnewknowledge. Vet.J.165,11–20.

Moser,C.,Ruggli,N.,Tratschin,J.D.,Hofmann,M.A.,1996.Detectionof antibodiesagainstclassicalswinefevervirusinswineserabyindirect ELISAusingrecombinantenvelopeglycoproteinE2.Vet.Microbiol.51, 41–53.

Nakagawa,S.,Schielzeth, H.,2013.Ageneralandsimplemethodfor obtainingR2fromgeneralizedlinearmixed-effectsmodels.Methods Ecol.Evol.4,133–142.

RDevelopmentCoreTeam,2008.R:ALanguageandEnvironmentfor Sta-tisticalComputing.RFoundationforStatisticalComputing,Vienna, Austria.

Robertson,I.,Holyoake,C.,Ramsay,G.,Mesiti,L.,Geong,M., Pullingo-mang,D.,2010.TheDevelopmentofaNationalSurveillanceSystemfor ClassicalSwineFever,AvianInfluenza,andFootandMouthDisease. AnnualReport.ProjectAH/2004/020.AustralianCentrefor Interna-tionalAgriculturalResearch(ACIAR),Canberra.

Rogan,W.J.,Gladen,B.,1978.Estimatingprevalencefromtheresultsofa screeningtest.Am.J.Epidemiol.107,71–76.

Santhia,K.A.P.,Purnatha,N.I.,Sunarno,C.,Wulan,S.S.N.,Pullingomang,D., 2003.EpidemiologicalStudyofHogCholerainAlorDistrict,EastNusa TenggaraProvince.BalaiPenyidikandanPengujianVeterinerRegional VIDenpasar,DiseaseInvestigationCentre,Denpasar,Bali,Indonesia.

Santhia,K.A.P.,Dibia,I.N.,Purnatha,N.I.,Sutami,N.N.,2006.Surveillance forHogCholeraeradicationinAlorDistrict,NusaTenggaraTimur. BalaiPenyidikandanPengujianVeterinerRegionalVIDenpasar, Dis-easeInvestigationCentre,Denpasar,Bali,Indonesia.

Schroeder,S.,vonRosen,T.,Blome,S.,Loeffen,W.,Haegeman,A., Koe-nen,F.,Uttenthal,A.,2012.Evaluationofclassicalswinefevervirus antibodydetectionassayswithanemphasisonthedifferentiationof infectedfromvaccinatedanimals.Rev.Sci.Tech.31,997–1010.

Sergeant, E.S.G., 2010. Epitools epidemiological calculators. AusVet Animal Health Services and Australian Biosecurity Cooperative ResearchCentre forEmergingInfectiousDisease,Availablefrom:

http://epitools.ausvet.com.au

Sergeant, E.S.G., 2014. Epitools epidemiological calculators. AusVet Animal Health Services and Australian Biosecurity Cooperative ResearchCentre forEmergingInfectiousDisease,Availablefrom:

http://epitools.ausvet.com.au

(14)

Tri Satya, P., Hutabarat, N., Santhia,K., 1999. The distribution and controlstrategiesofclassicalswinefeverinIndonesia.In: Black-sell, S. (Ed.), Classical Swine Fever and Emerging Diseases in Southeast Asia. ACIAR Proceedings No. 94, Vientiane, Loa. , pp.111–115.

vanOirschot,J.T.,2003.Vaccinologyofclassicalswinefever:fromlabto field.Vet.Microbiol.96,367–384.

Wang,Y.,2007.RuralPovertyinIndonesia–EnablingtheRuralPoorto OvercomePovertyinIndonesia.InternationalFundforAgricultural Development(IFAD),Rome,Italy.