Modelling the elimination of river blindness using long-term epidemiological and programmatic data from Mali and Senegal

ContentslistsavailableatScienceDirect

Epidemics

jo u rn al h om ep age : w w w . e l s e v i e r . c o m / l o c a t e / e p i d e m i c s

Modelling

the

elimination

of

river

blindness

using

long-term

epidemiological

and

programmatic

data

from

Mali

and

Senegal

Martin

Walker

a,b,∗,1

,

Wilma

A.

Stolk

c,1

,

Matthew

A.

Dixon

c,d,a

,

Christian

Bottomley

c,d

,

Lamine

Diawara

e

,

Mamadou

O.

Traoré

f

,

Sake

J.

de

Vlas

c,2

,

María-Gloria

Basá ˜

nez

c,d,a,2

a_{DepartmentofInfectiousDiseaseEpidemiologyandLondonCentreforNeglectedTropicalDiseaseResearch,ImperialCollegeLondon,NorfolkPlace,W21}

PG,London,UK

b_{DepartmentofPathobiologyandPopulationSciencesandLondonCentreforNeglectedTropicalDiseaseResearch,RoyalVeterinaryCollege,Hawkshead}

Lane,Hatfield,AL97TA,UK

c_{DepartmentofPublicHealth,ErasmusMC,UniversityMedicalCenterRotterdam,Rotterdam,TheNetherlands} d_{MRCTropicalEpidemiologyGroup,LondonSchoolofHygieneandTropicalMedicine,KeppelStreet,London,UK}

e_{Inter-CountrySupportTeamforWestAfrica,WorldHealthOrganization158,Placedel’Indépendance03BP7019,Ouagadougou03,BurkinaFaso} f_{ProgrammeNationaldeLuttecontrel’Onchocercose(PNLO),DirectionNationaledelaSanté(DNS),B.P.233,Bamako,Mali}

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received17December2016

Receivedinrevisedform3February2017

Accepted7February2017

Keywords:

Onchocerciasis

Riverblindness

Elimination

Mathematicalmodelling

Surveillance

a

b

s

t

r

a

c

t

TheonchocerciasistransmissionmodelsEPIONCHOandONCHOSIMhavebeenindependently devel-opedandusedtoexplorethefeasibilityofeliminatingonchocerciasisfromAfricawithmass(annualor biannual)distributionofivermectinwithinthetimeframesproposedbytheWorldHealthOrganization (WHO)andendorsedbythe2012LondonDeclarationonNeglectedTropicalDiseases(i.e.by2020/2025). Basedonthefindingsofourpreviousmodelcomparison,weimplementedtechnicalrefinementsand testedtheprojectionsofEPIONCHOandONCHOSIMagainstlong-termepidemiologicaldatafromtwo WestAfricantransmissionfociinMaliandSenegalwheretheobservedprevalenceofinfectionwas broughttozerocirca2007–2009after15–17yearsofmassivermectintreatment.Wesimulatedthese interventionsusingprogrammaticinformationonthefrequencyandcoverageofmasstreatmentsand trainedthemodelprojectionsusinglongitudinalparasitologicaldatafrom27communities,evaluating theprojectedoutcomeofelimination(localparasiteextinction)orresurgence.WefoundthatEPIONCHO andONCHOSIMcapturedadequatelytheepidemiologicaltrendsduringmasstreatmentbutthat resur-gence,whileneverpredictedbyONCHOSIM,waspredictedbyEPIONCHOinsomecommunitieswiththe highest(inferred)vectorbitingratesandassociatedpre-interventionendemicities.Resurgencecanbe extremelyprotractedsuchthatlow(microfilarial)prevalencebetween1%and5%canbemaintainedfor 3–5yearsbeforemanifestingmoreprominently.Wehighlightthatpost-treatmentandpost-elimination surveillanceprotocolsmustbeimplementedforlongenoughandwithhighenoughsensitivitytodetect possibleresiduallatentinfectionspotentiallyindicativeofresurgence.Wealsodiscussuncertaintyand differencesbetweenEPIONCHOandONCHOSIMprojections,thepotentialimportanceofvectorcontrol inhigh-transmissionsettingsasacomplementaryinterventionstrategy,andtheshortremaining time-lineforAfricancountriestobereadytostoptreatmentsafelyandbeginsurveillanceinordertomeetthe impending2020/2025eliminationtargets.

©2017TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Abbreviations:ABR,annualbitingrate;APOC,AfricanProgrammeforOnchocerciasisControl;CDTI,community-directedtreatmentwithivermectin;MAP,maximuma

posteriori;MDA,massdrugadministration;mf,microfilariae;NTD,neglectedtropicaldisease;OCP,OnchocerciasisControlProgrammeinWestAfrica;PES,post-elimination

surveillance;PTS,post-treatmentsurveillance;SIR,samplingimportanceresampling;WHO,WorldHealthOrganization.

∗_{Correspondingauthor.}

E-mailaddress:[email protected](M.Walker).

1 _{Equalcontributionsasfirstauthors.}

2 _{Equalcontributionsaslastandseniorauthors.}

http://dx.doi.org/10.1016/j.epidem.2017.02.005

1755-4365/©2017TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.

1. Introduction

Human onchocerciasis or river blindness is caused by the filarialnematodeOnchocercavolvulusandisearmarkedfor elim-inationbytheWorldHealthOrganization(WHO) asarticulated bythe2012Roadmap(WHO,2012)andtheLondonDeclaration onNeglectedTropical Diseases(2012).Theprincipalstrategyto achieveeliminationismassdrugadministration(MDA)with iver-mectin.Ivermectinkillstheskin-dwellingmicrofilariae(mf)that aretheprogenyofadultO. volvulusand areinfectioustobiting blackflyspecies vectors. Ivermectin may alsokill and/or steril-ize adultworms(Gardonet al.,2002).Multiplerounds ofmass treatmentareeffectiveinloweringtheprevalenceandintensityof onchocerciasisand–ifgivenforlongenoughathighenough cover-age–canleadtotheinterruptionoftransmissionandelimination oftheinfection. InLatinAmerica, (biannual) MDAhas success-fullyeliminatedonchocerciasisfromColombia(Westetal.,2012), Ecuador(Lovatoetal.,2014), northernVenezuela(Convitetal., 2013)andMexico(Rodríguez-Pérezetal.,2015),withGuatemala awaitingcertification.Goodprogresstowardseliminationhasalso beenmadein Africa(Tekleetal.,2016)which bears99%ofthe onchocerciasisburden,withnotablesuccessesinregionsofMali, Senegal(Diawaraetal.,2009;Traoreetal.,2012),Nigeria(Tekle et al., 2012), Sudan (Higazi et al., 2013) and eastern Uganda (Katabarwaetal.,2014)butalsoconspicuousregionsofongoing transmissiondespiteyears ofinterventioninGhana(Lamberton etal.,2015),Cameroon(Katabarwaetal.,2013a;Wanjietal.,2015; Eisenbarthetal.,2016)andnorthwesternUganda(Katabarwaetal., 2013b),andevidenceofrecrudescenceinBurkinaFaso(Koalaetal., 2017).

EPIONCHO and ONCHOSIM are mathematical transmission modelsofhumanonchocerciasisthathave beenusedto evalu-atetheeffectivenessofdifferentinterventionstrategiesinreaching the2020/2025eliminationtargetsforonchocerciasis(WHO,2012; APOC,2012).Bothmodelshavebeenusedtopredictthetimeto eliminateonchocerciasisunderannualorbiannualMDA(Coffeng et al., 2014a; Turneret al., 2014), and have been used in col-laborationwiththeAfricanProgrammeforOnchocerciaisControl (APOC)toevaluatehowprogresstowardseliminationcanbe accel-eratedbyimplementing alternativetreatmentstrategies (WHO, 2015).ONCHOSIMhasbeenusedtoinformtheOnchocerciasis Con-trolProgrammeinWestAfrica(OCP)andtheAfricanProgramme forOnchocerciasisControlonexpectedtrendsofinfectionduring vectorcontrol(Plaisieretal., 1991), massivermectin treatment (Winnenetal.,2002;Coffengetal.,2014a;Tekleetal.,2016)or theircombination(Plaisieretal.,1997),thehealthimpactofthe interventions(Coffengetal.,2014b),andtheexpectedtimeto elim-ination(Kimetal.,2015).

More recently,thedevelopersof EPIONCHOandONCHOSIM havestartedtocompareandrefinetheirmodels,withthe objec-tiveofreachingconsensusonthefeasibilityofandtimehorizons for elimination, and optimum interventions for achieving the WHO goalsin Africa(Stolk etal., 2015).The first formal com-parison,which attemptedto ‘dock’the two models bymaking parameterassumptionsasequivalentaspossible,revealedsome importantdifferencesbetweenthemodels’structuralassumptions andresultingoutputs.Bothmodelsdemonstratedabenefitof bian-nualversusannualtreatmentbutdifferedintheirpredictedtime toelimination,particularlywhenusingtransmissionbreakpoints as endpoints rather than operational (microfilarial) prevalence thresholdssuchasthoseprovisionallyproposedbyAPOC(2010). Oneoftheconclusionsofthisfirstcomparisonwasthattheuseofa singleprevalencethresholdacrossdifferentendemicity (transmis-sionintensity)settingsmaynotbeappropriateandneedsfurther investigation(Stolketal.,2015).

Theepidemiologyofonchocerciasishaschangedasaresultof theintroductionofivermectinMDA,whichmeansthatEPIONCHO andONCHOSIMarenowbeingusedintransmissioncontextsfar fromthoseinwhichtheywereoriginallydevelopedand parame-terized(Basá ˜nezandBoussinesq,1999;Plaisieretal.,1991;Plaisier etal.,1995).Hence,tomaintainconfidenceinprojections,these modelsmustbetestedandvalidatedagainstdatacollectedbefore, duringandafterMDA,reflectingthespectrumoftransmission con-ditions,fromendemicinfectiontotransmissioninterruptionand elimination(Basá ˜nezetal.,2012a,b).Giventheprolongednatureof onchocerciasisinterventions,capturingandrecordingsuchdatais challengingandresource-intensiveandwouldprobablyhavebeen impossiblewithoutthelogisticalandtechnicalsupportoftheOCP andAPOC.

Here wetest theEPIONCHOand ONCHOSIMmodelsagainst epidemiologicaldatacollectedovertwodecadesfrom27sentinel communities intwo onchocerciasisfociinMaliand Senegal.In thesefoci,MDAwithivermectinalonehassuccessfullybroughtthe prevalenceofinfectivelarvae(inblackflies)andofmf(inhumans) – detectableby skin snips – to zero, indicative of elimination (Diawaraetal.,2009;Traoreetal.,2012).Inparticular,prevalenceof skinmfreachedthepreviouslyproposedoperationalthresholdsset byAPOCof<5%inallsurveyedvillagesand<1%in90%ofthose sur-veyed(APOC,2010).Weuseprogrammaticdataonthefrequency andcoverageofMDAwithivermectintosimulatethese interven-tionsfromendemicbaseline,circa1987,throughtocessationof treatment, circa 2006,andthen beyond, projectingforwards to 2020.Wetrainthemodelprojectionsusingsubsetsof epidemiolog-icaldatacomprisingcommunity-specificestimatesofmicrofilarial prevalencecollectedthroughouttheintervention.Wedetermine whetherthemodelspredictsustainedeliminationorresurgence of infectionin thepost-treatmentand post-elimination periods (WHO, 2016) and explore how these predictions changewhen usingincreasingamountsofepidemiologicaltrainingdata,from pre-interventiondataonlytomultiplelongitudinaldatapointsper community.

2. Modelsandmethods

2.1. EPIONCHO

Fig.1. Endemicmicrofilarialprevalence(beforeintervention)inselectedsentinelcommunitiesin(a)theRiverBakoyefocus,Maliand(b)theRiverGambiafocus,Senegal

circa1987.Endemicitycategoryisdefinedbyamicrofilarialprevalenceinpeopleaged≥5yearsof<40%(hypoendemic),40%–59%(mesoendemic)and≥60%(hyperendemic)

andcolouredsequentiallyfromyellowtored.Errorbarsdenote95%confidenceintervals.Communitynamesareanonymised.

comparedtoannually(Turneretal.,2014);thepotential epidemi-ologicalimpactofMDAwithmoxidectin(Turneretal.,2015a),and thefuturedeploymentofanonchocerciasisvaccine(Turneretal., 2015b).

ThelatestversionofEPIONCHOusedhereincorporatesa num-berofrefinementsfromtheversiondescribedinTurneretal.(2013) andreviewedbyBasá ˜nezetal.(2016).Someoftheseare moti-vatedbytheresultsofourrecent comparisonwithONCHOSIM (Stolketal.,2015),andothersbyourimprovedunderstandingof thepopulationbiologyofO.volvulus.Modificationsinclude:(1)the introductionofa latentperiodforthedevelopmentofO. volvu-luswithintheblackflyvector,implementedbydividingthelarval stateL intoL1, L2and L3 compartments (Basá ˜nez et al., 2007; Basá ˜nezetal.,2009);(2)areparameterizationofthemicrofilarial ‘uptake’curve(describingparasiteestablishmentwithinthe vec-tor)toreflectthatanoverdisperseddistributionofparasitesamong humanhosts affects theoverall severityof density dependence (Churcheretal.,2005);(3)anewfunctionalrelationshipbetween theprevalenceandintensityofmfthatreflectstheunderlying dis-tributionofadultworms,incorporatesthesensitivityofskinsnips (Bottomleyetal.,2016)andbettercapturestheinitial‘bounceback’ ofmicrofilarialprevalenceduringthefirstfewroundsofivermectin MDA(Plaisier etal.,1995);and(4)a morerealisticdistribution ofadultparasitesurvival times(previouslyexponential), imple-mentedbyintroducingmultiplenominal‘age’compartmentsfor adultO.volvulusandparameterisedusingpublishedestimatesof lifeexpectancyaswellasthe‘tail’ofthedistributionofsurvival times(Plaisieretal.,1991).Fulldetailsoftheserefinementsand thecompletedescriptionofthislatestversionofEPIONCHOare givenintheSupplementaryinformation,1.1EPIONCHOandalist ofparameterdefinitions,descriptionsandvaluesinSupplementary information,TableS1.

2.2. ONCHOSIM

ONCHOSIM is a long-established microsimulation model of onchocerciasistransmission(Habbemaetal.,1996;Plaisieretal., 1990).Briefly,themodelconsidersahumanpopulation,typically consistingofapproximately400individuals,withademographic composition that changes stochastically over time. The life-histories ofindividual maleand femaleadult parasitesand the populationsofmfwithinindividualhumanhostsaretrackedand transmissionismediatedbyapopulationofblackflies(the parasite-within-the-vector component is treated deterministically). The survivaltimesofadultwormsfollowaWeibulldistribution.The

bitingrate(numberofbitesperpersonperunittime)isseasonal, variesamonghostsaccordingtoanindividual-specificexposure variable(assignedatbirth),andchangessystematicallywithhost ageandsexinamannersimilartothatofEPIONCHO.Treatmentin ONCHOSIMismodelledinabroadlysimilarmannertoEPIONCHO, albeit(100%)mfarekilledinstantaneouslyinONCHOSIMrather than over1–2 monthsasin EPIONCHOand therates at which femalewormsrecovertheirfertilityaresubtlydifferentbetween thetwomodels(Plaisieretal.,1995;Basá ˜nezetal.,2008;Coffeng etal.,2014a;Basá ˜nezetal.,2016).Humanhosts,excludingchildren aged<5yearsandafractionofwomenofreproductiveage, partic-ipateineachroundofMDAwithaprobabilitythatdepends on ageandsex,andaparametergoverningtheirlifelongadherenceto MDA.ONCHOSIMhasbeenusedtoinformandsupportOCP(Plaisier etal.,1997;Winnenetal.,2002)andAPOC(Coffengetal.,2014a,b) interventionpoliciesandmorerecentlyhasbeenusedtoevaluate theprogressofAPOCprojectstargetingelimination(Tekleetal., 2016).TechnicaldetailsoftheONCHOSIMmodelcanbefoundin Supplementaryinformation,1.2ONCHOSIM,Coffengetal.(2014a) andBasá ˜nezetal.(2016).TheONCHOSIMparameterdefinitions, descriptionsandvaluesaregiveninSupplementaryinformation, TableS2.

2.3. Epidemiologicalandprogrammaticdata

Weuselong-termparasitologicalandprogrammaticdatafrom theRiverBakoyeandRiverGambiafociinMaliandSenegalwhere onchocerciasishasbeendeclaredeliminatedafter15–17yearsof, respectively,annualorbiannualMDA(Diawaraetal.,2009;Traore etal.,2012)(Fig.1).Wedonotusedatafromthecross-borderRiver Falemefocus.Thepredominantblackflyvectorspeciesinbothfoci isSimuliumsirbanum(asavannahmemberoftheS.damnosums.l. complex;Boakyeetal.,1998).Sentinelcommunitieswereinitially selectedbytheOCPcirca1987inareasoftheWesternExtension wherevectorcontrolhadneverbeenimplementedandtherefore providedauniqueopportunitytoevaluatetheimpactofivermectin asthesolemeasureofcontrol.Communitieswereroutinely mon-itoredbytheOCP from1987to2002andfrom2006byateam studyingthefeasibilityof eliminatingonchocerciasiswithMDA alone(Diawaraetal.,2009).

Table1

Summaryofepidemiologicaldataoncommunityprevalenceofmicrofilariae.

Focus,country Numberof

communities

Endemicmicrofilarialprevalence beforeintervention(range)

Sequencelengthofmicrofilarial prevalencetrainingdata(range)

RiverBakoye,Mali 13 31%–70% 3–5

RiverGambia,Senegal 14 36%–82% 4–11

waterforatleast30min,butfor24hifnegativeafterthistime) andstandardizedbyageandsexusingtheOCPreference popu-lation(Moreauetal.,1978).Becauseyoungchildrentend tobe underrepresentedincommunityskin-snipsurveys,weassumed that the data better represented the prevalence in those aged ≥5years rather than≥1year, thereby harmonizingthe sampled demographicgroupwithusualEPIONCHOandONCHOSIMoutputs (Stolketal.,2015).Weexcludedfromouranalysiscommunities whereonlypre-orpost-interventiondatawereavailable(Table1). Theprogrammaticdatacompriseestimates,withinarange,of theoveralltherapeuticcoverage(proportionofthetotal popula-tiontreated)withivermectinwithineachriverbasinfortheperiods 1988–1991,1992–1996and1998–2006(Table2).Ivermectinwas firstdistributedbyOCPmobileteamsbeforeCDTI–which empow-eredcommunitiestoappointlocalcommunity-drugdistributors whotookresponsibilityfordistributingivermectin–was intro-ducedin1997,followingtheinceptionofAPOCin1995(Boatin, 2008).ThereisanotabledropincoverageinthefirstyearofCDTI (Table2)associatedwithlogisticalchallengesinchangingthe dis-tributionsystem(Diawaraetal.,2009).

2.4. Simulatinginterventionsandestimatingcommunity-specific bitingrates

WesimulatedMDAinterventionsusingthe(focus-specific) pro-grammaticdataonfrequencyandcoveragesummarizedinTable2. BothEPIONCHOandONCHOSIMareinitializedusingtheannual bit-ingrate(ABR),definedhereastheaveragenumberofblackflybites receivedperpersonperyear.Hence,toruncommunity-specific simulations,we required an estimateof theABR in each com-munity.WeestimatedABRsbyamaximumlikelihoodapproach (Supplementaryinformation, 1.4Estimatingcommunity-specific bitingrates) using subsets of the microfilarial prevalence time seriesfromeachcommunity.Werefertothesesubsetsastraining datasets,eachcomprisingavariablesequencelengthof longitu-dinalprevalenceestimates fromeach community. Thesmallest training dataset comprised pre-intervention data only and the largest included all data from all communities. We iteratively re-estimatedABRsbyincrementallyincreasingthelengthofthe longitudinaltrainingdatasets,evaluatinghowincreasingamounts oftrainingdataalteredandinfluencedthemodelprojections.Data ontheexacttimeoftreatment(unlikethetimeofepidemiological surveys)werenotavailableandhence,forthepurposesof estimat-ingABRs,weassumedthatsurveyswereundertakenexactly11or 5monthsafterthelasttreatmentfor,respectively,theannualand biannualdistributionsintheRiverBakoyeandRiverGambiafoci (Supplementaryinformation,TableS3).

2.5. Parametricuncertainty

TheestimationofABRsdependsonitsrelationshipwith infec-tion which, in turn, depends on structural assumptions and parametervaluesofthemodel.InONCHOSIM,weestimatedABRs, accountingforseasonalityintransmission,whileallother param-eters were fixed at their default values (Coffeng et al., 2014a; Basá ˜nezetal.,2016).InEPIONCHO,weexplicitlyconsideredthat thereisuncertaintyinsomeessentialparameters:multipleunique parametersets areall concordantwithanindependent dataset

comprisingcoupledpre-interventioncommunity-levelprevalence andABRestimatesfromCameroon(RenzandWenk,1987;Basá ˜nez andBoussinesq,1999)(Fig.2a).Weidentifiedtheseparametersets usingasamplingimportanceresampling(SIR)approach(Gambhir etal.,2015),inwhichwesampledfromtheposteriordistribution ofparametervaluesgivenuniformpriorsdefinedbytheranges of published estimates (Supplementary information, Table S1). DetailsofthisapproacharegiveninSupplementaryinformation, 1.3Parametricuncertainty.

2.6. Selectionofamaximumaposterioriparameterset

TheSIRapproachappliedtoEPIONCHOinvolvesdefiningaprior distributionfordiscreteparametersets,therebysupportingawide rangeofmodelprojectionsforanestimatedABR.Wecalculated theposteriorprobabilityofeachcommunityprojectionasthe like-lihoodofthecombinedtrainingdatasetfromallcommunities(i.e. asinglecombineddatasetforeachfocus)weightedbythe param-eterset(prior)probabilitycalculatedfromtheSIRprocedure.Fora moredirectcomparisonwithONCHOSIM–inwhichweonly con-siderasingledefaultparameterset–weselectedtheparameter setthatmaximisedtheposteriorprobability,akintoamaximum

aposteriori(MAP)approach.DetailsaregiveninSupplementary information,1.5Selectingmaximumaposteriori(MAP)parameter sets.Wehighlightthat,unliketheestimationofABRs,undertaken onacommunitybasis,identificationofMAPparametersetswas doneat theregional level (i.e.includingboth foci); fundamen-talbiologicalandpopulationdynamicsparameterswereassumed constantamong communitiesandbetweenfoci(butABRswere permittedtovary).

2.7. Eliminationcriterion

Weevaluatedtheoutcomeofeachcommunity-based simula-tionassustainedeliminationorresurgenceaccordingtotheMAP parametersetsforEPIONCHOandthedefaultparametersetfor ONCHOSIM.We defined sustainedelimination as local parasite extinction, indicated by the parasitepopulation tending termi-nallytozero.Ifelimination isnotsustained,resurgenceoccurs. InEPIONCHO,sustainedeliminationarisesbycrossingthe trans-missionbreakpoint.Thisarisesbecauseofthenecessityforfemale

stochastic-Table2

Summaryofprogrammaticdataoncommunitycoveragewithivermectin.

Focus,country Timeperiod Treatmentfrequency Treatmentcoverage

RiverBakoye,Mali 1989–1991 Annual 59%–62%

1992–1996 Annual 75%–78%

1997 Annual ∼38%b

1998–2006 Annual 73%–83%

RiverGambia,Senegal 1988–1991 Annual/biannuala _64%–69%

1992–1996 Biannual 76%–77%

1997 Biannual ∼38%b

1998–2006 Biannual 77%–81%

a_{Annual1988–1990;biannualthereafter.}

b _{Firstyearofcommunity-directedtreatmentwithivermectin,coveragewaslow(Diawaraetal.,2009)andhereisassumedtobehalfthatoftheprecedingperiod.}

Fig.2.Modelledrelationshipbetweenannualbitingrate(ABR)ofblackflyvectorsandendemicmicrofilarialprevalenceusing(a)EPIONCHOand(b)ONCHOSIM.Inpanel(a),

thecoupledABR-prevalencedataarefrom9communitiesinnorthernCameroon(Basá ˜nezandBoussinesq,1999)whereeachABRwasmeasuredasanaveragefrommultiple

yearsandlocationswithinandaroundeachcommunity,weightedbytheproportionoftimecommunityresidentsspentattheselocations(RenzandWenk,1987).Eachthin

linecorrespondstoanEPIONCHOparametersetidentifiedbythesamplingimportanceresampling(SIR)proceduredescribedinSupplementaryinformation,1.3Parametric

uncertainty.Thesearecolouredsequentiallyfromyellowtoredinaccordancewithendemicitycategoryasdefinedbyamicrofilarialprevalenceinpeopleaged≥5yearsof

<40%(hypoendemic),40%–59%(mesoendemic)and≥60%(hyperendemic).Thethickblacklinecorrespondstotheparametersetthatachievedthehighestlikelihood.Inset

isthesamegraphwiththex-axistransformedtoalogarithmicscale.Inpanel(b),thethinlinescorrespondtostochasticrealizationsoftheONCHOSIMdefaultparameterset

(Coffengetal.,2014a),colouredsequentiallyaccordingtoendemicitycategory,andthethickblacklineisthemedianof500simulations.ThecoupledABR-prevalencedata

arenotshowninpanel(b)becauseONCHOSIMhasnotbeenre-fittedtothesedata.

ityinthehumanhostpopulation,anon-negligiblephenomenon becauseoftherelativelysmallsimulatedhumanpopulationsize of400–440individuals(seeSupplementaryinformation,TableS2). NeitherEPIONCHOnorONCHOSIMconsiderpossible reintroduc-tionof infection(byblackfliesor humans)fromproximate foci wheretransmissionmaybeongoing.

2.8. Coverageandadherence

We repeated the simulations, re-estimating ABRs (and for EPIONCHO,identifyingMAPparametersets)andevaluating pro-jections of sustained elimination or resurgence using different assumedvaluesoftheestimatedtreatmentcoverage,whichwas givenaswithintherangesshowninTable2.Specifically,weran repeatedsimulationsusingthelower,medianandupperestimates ofcoveragewithineachtimeframe(Table2).Thedegreeof non-adherencetoMDAwasunknownandwassetto5%forbothmodels, inaccordancewithprevioussimulations(Stolketal.,2015).

3. Results

3.1. Infectiondynamicstowardselimination

WepresentinFig.3theobservedmicrofilarialprevalenceand theEPIONCHO-andONCHOSIM-modelleddynamicsintheRiver BakoyefocusinMaliandin theRiverGambiafocus inSenegal. The simulations shown are those with estimated

Fig.3.Observedandmodelleddynamicsofmicrofilarialprevalencein13communitiesfromtheRiverBakoyefocus,Mali(panelsa–d)andin14communitiesfromthe

RiverGambiafocus,Senegal(panelse–h).Panelsontheleft(a,c,eandg)andontheright(b,d,fandh)showEPIONCHOandONCHOSIMprojections,respectively.Thethin

linescorrespondtocommunity-specificsimulationsusingmaximumlikelihoodestimatesofthecommunity-specificannualbitingrates(ABRs)andeitherthemaximuma

posteriori(MAP)parameterset(EPIONCHO)orthedefaultparameterset(ONCHOSIM).TheestimatedABRsandMAPparametersetsarederivedusingthepre-intervention

microfilarialprevalencedataonly(a,b,eandf)orusingthecompletelongitudinalsequenceforeachcommunity(c,d,gandh).Forbrevity,thedynamicspredictedfrom

estimatesusingoneortwointerimtimepointsarenotshown.ForONCHOSIMtherearemanystochasticprojectionsforeachcommunityprojection;forEPIONCHOthereis

asingledeterministicprojectionforeachcommunity,correspondingtotheMAPparameterset.Thethicksolidlinesshowthemediandynamicsbyendemicitycategoryas

categorisedbyamodel-derivedpre-interventionmicrofilarialprevalenceinpeopleaged≥5yearsof<40%(hypoendemic),40%–59%(mesoendemic)and≥60%(hyperendemic)

andcolouredsequentiallyfromyellowtored.InpanelctheestimatedABRs(fromEPIONCHO,usingthecompletelongitudinaldatasequences)indicatedallcommunities

intheRiverBakoyefocuswereeitherhypoendemicormesoendemic.Panelinsetsshowtheperiodbetween2010and2020usingatransformedy-axisforabettervisual

appraisalofthemodelprojectionscomparedtothedataclosetozero.

3.2. Eliminationorresurgence

WepresentinFig.4thenumberandpercentageofcommunities wheretheEPIONCHO-predicted outcomeofMDAwassustained elimination(localparasiteextinction)usingincrementally increas-ing subsets of the longitudinal training datasets (Table 1) and differentassumedlevelsoftreatmentcoveragewithintherange of reported estimates (Table 2). The percentage of (stochastic) ONCHOSIM projections resulting in sustained elimination was greaterthan99%forallcommunities,irrespectiveofthelengthof thetrainingdatasets.Bycontrast,EPIONCHOprojectionswere

Fig.4.Numberandpercentageofcommunity-specificEPIONCHOprojectionsresultingineliminationin(a)theRiverBakoyefocus,Maliand(b)theRiverGambiafocus,

Senegal.Eachgridelementindicatesthenumberandpercentageofcommunitieswithaprojectedoutcomeofsustainedeliminationwhentrainingthemodelprojection

usingdifferentsequencelengthsofthecommunity-specificmicrofilarialprevalenceestimates(x-axis,seeTable1)andassumingdifferentlevelsofcoverage(y-axis,see

Table2).Theoutcomesarecolouredsequentiallyfromred(50%ofcommunitieswithpredictedelimination)togreen(100%ofcommunitieswithpredictedelimination).

Fig.5. ThepercentageofcommunitieswithanEPIONCHO-projectedoutcomeofsustainedeliminationin(a)13communitiesintheRiverBakoyefocus,Maliand(b)

14communitiesintheRiverGambiafocus,Senegal.Endemicityiscategorizedaccordingtothemodel-derivedpre-interventionmicrofilarialprevalenceestimates(using

theestimatedcommunity-specificannualbitingrates)inpeopleaged≥5yearsof<40%(hypoendemic),40%–59%(mesoendemi)and≥60%(hyperendemic)andcoloured

sequentiallyfromyellowtored.Eachbarindicatesthecombinedresultsfromestimatingthecommunityannualbitingrate(andthesubsequentpre-interventionendemicity)

usingdifferentsequencelengthsofthecommunitytrainingdatasets(Table1)anddifferentassumedlevelsoftreatmentcoverage(Table2).Hence,inanysimulationwhere

thepre-interventionendemicitywasestimatedashyperendemic,thechanceofeliminationwaslow,irrespectiveoftheannual(a)orbiannual(b)treatmentstrategy.

Generally,inEPIONCHOtheoutcomeofsustainedelimination issensitivetothemodel-derivedendemicity,asdrivenbythe esti-matedcommunity-specificABRs.Fig.5showsthat–regardlessof thelengthofthetrainingdatasetortheassumedlevelof cover-age–sustainedeliminationwasrarelypredictedinhyperendemic communities(where thepre-intervention endemic microfilarial prevalencewasestimatedtobe≥60%;aslightlyhigherpercentage ofsimulationswithpre-interventionhyperendemicityoccurredin theRiverGambiafocus,SenegalcomparedwiththeRiverBakoye focus, Malibecause of thebiannual treatment strategy).When eliminationispredicted,therateofresurgenceisvariablebut some-timesveryprotracted,dependingoncommunityendemicity(and thecorrespondingABR).WeillustratethisinFig.6byshowingthe EPIONCHO-predictedmicrofilarialprevalencelevelsin2007,2009 and2012–one,threeandfiveyearsafterthelastroundoftreatment in2006–amongcommunitieswhereeithersustainedelimination (microfilarialprevalencetendingtozero)orresurgencewas indi-cated.Manyofthesimulationspredictamicrofilarialprevalence between1%and5%atthesetimesincommunitieswhereinfection is(slowly)resurgent.

4. Discussion

WehavetestedtheEPIONCHOandONCHOSIMonchocerciasis transmissionmodelsagainstlong-termepidemiologicaldatafrom

twofociinMaliandSenegalwherethe(detectable)prevalenceof

O.volvulusmfhasbeenbroughttozeroby,respectively,annual and biannualMDAwithivermectinalone. Bothmodelscapture adequatelycommunityinfectiondynamicsthroughMDAtowards elimination.Eliminationinall27communitiesistheubiquitous outcomepredictedbyONCHOSIM,irrespectiveofinferredvector bitingratesanduncertaincoverage.InEPIONCHO,eliminationis lesscertain,beingdependentontheinferredcommunity-specific blackflybiting rates (ABRs) and corresponding(model-derived) levelsofinfectionendemicity.Inthemorehighlyendemic com-munitiesintheRiverGambiafocus,Senegal–withthehighest vectorbitingrates–EPIONCHOindicatesasubstantiveriskof infec-tionresurgence,despitethebiannualMDAstrategy. Resurgence dynamicscanbeveryprotractedsuchthatthemicrofilarial preva-lenceoftenremainsbetween1%and5%foratleastfiveyearsafter thelasttreatment,beforerisingmoresubstantively.

Fig.6. EPIONCHOpredictionsofmicrofilarialprevalencein2007,2009and2012,one,threeandfiveyearsafterthelastroundoftreatmentin2006,amongcommunitiesin

(a–c)theRiverBakoyefocus,Maliand(d–f)theRiverGambiafocus,Senegal.Endemicityiscategorizedaccordingtomodel-derivedpre-interventionmicrofilarialprevalence

estimates(usingtheestimatedcommunity-specificannualbitingrates)inpeopleaged≥5yearsof<40%(hypoendemic),40%–59%(mesoendemic)and≥60%(hyperendemic)

andcolouredsequentiallyfromyellowtored.Eachcategoryincludesthecombinedresultsfromestimatingthecommunityannualbitingrate(ABR)andthesubsequent

pre-interventionendemicityusingdifferentsequencelengthsofthecommunitytrainingdatasets(Table1)anddifferentassumedlevelsoftreatmentcoverage(Table2).

Thecirculardatapointsineachpanelindicatesimulationswhereresurgenceisthepredictedoutcome(onlyinredhyperendemiccategories,seeFig.5)andthetriangular

datapoints(allcloseto0%)arethosecorrespondingtosustainedelimination.Theviolinplotsillustratethedensityofthemicrofilarialprevalencepredictions.Thehorizontal

dashedlinesrepresentnominal(operational)microfilarialprevalencethresholdsof1%and5%(APOC,2010).

havestillmostlybeenbasedonskinsnippingtodetectmf.APOC setoperationalprevalencethresholdsforstoppingtreatmentand startingsurveillance(<5%microfilarialprevalenceinallsurveyed villagesand<1%in90%ofthosesurveyed)thatshouldbemetfora three-yearPTSperiodtoconfirmelimination(APOC,2010).Indeed, theprogressof54APOCprojectareas(comprisingclustersof5–20 villages)targetingthesethresholdshasrecentlybeenevaluated bycomparingobservedmicrofilarialprevalenceversusONCHOSIM predictions(Tekleetal.,2016),with46/54(85%)judgedontrackor exceedingpredictionintheirprogresstowardsPTS.Theprotracted natureofsomeoftheresurgencedynamicsindicatedbyEPIONCHO, initiallycharacterizedbylatent,difficult-to-detectinfections lev-elsbeforemoremarkedincreases,suggeststhattheproposed3–5 yearsofPTS(APOC,2010;WHO,2016)mustbeextremelyrobust– andincludeoptimizedandefficientsamplingmethodologies–to beeffectiveinidentifyingtheearlysignsofinfectionresurgence. ThisalsomeansthatforAfricancountriestoconfirmeliminationby 2020(WHO,2012)or2025(APOC,2012)theywillneedtobeina positiontostoptreatmentsafelyeitherimminentlyoratthelatest by2022.FormanyAfricancountries,theformerisunlikely. Meet-ingthelatter2022stopping-treatmentdatewillrequirecareful monitoringandevaluationofprogress.

Critically,theresurgencedynamicsprojectedbyEPIONCHOare intimatelylinkedwiththe(poor)sensitivityofskinsnippingatlow infectionlevels(Bottomleyetal.,2016);innear-elimination set-tingsmanyinfectionsarelikelytobemissedbyskinsnippingalone (thedistributionofmfintheskinisaggregatedbycontrasttothe randomPoissondistributionassumedbyONCHOSIM,adifference whichalsohelpstoexplainthemorerapidinitialdeclinein

preva-lenceprojectedbyEPIONCHOcomparedwithONCHOSIM).Thatis, EPIONCHOassumesmoreoftenthanONCHOSIMthatmfcounts arefalse-negative.Thus,applicationofthenewlyrecommended andmoresensitive(serological)diagnostics,mightenableearlier detectionofresurgence.Indeed,anepidemiological survey con-ductedin2014intheRiverGambiafocus,includingsomebutnot allofthecommunitiesincludedinthisanalysis,founda2.5% sero-prevalenceamongchildrenaged5–9years(Wilsonetal.,2016), abovethe0.1%threshold(upper95%confidenceinterval)in chil-drenunder10 yearsrecommendedbytheWHOforverification ofelimination(WHO,2016).Hence,bythiscriterion,elimination cannotbeverifiedintheRiverGambiafocusanditisprobablethat low-leveltransmissioncontinues.Resurgence–basedon parasito-logical(mf)andentomologicaldata–hasalsobeenreportedin BurkinaFasoinafocuspreviouslythoughttohaveeliminatedthe infection(Koalaetal.,2017).

maybetoohigh–andhenceposeariskofresurgenceifapplied tostop treatment– but probably too low in lower meso- and hypoendemicsettings(Stolketal.,2015).Nevertheless,prevalence thresholds (both serological and parasitological) are extremely usefuloperationaltoolsfordecision-supportinnear-elimination settingsiftheycanbeappropriatelyconstructed–andinformed usingmathematicalmodels–toreflectthediversityof transmis-sionsettingsandthevagariesofdifferentdiagnostictools.

The greater circumspection on sustained elimination (and possible resurgence) indicated by EPIONCHO compared with ONCHOSIMreflectsfundamentaltechnicaldifferencesbetweenthe twomodels.Oneimportantdifferenceisthatthehighlynon-linear relationshipbetweenmicrofilarialprevalenceandABRmodelled byEPIONCHO–reflectingadditionaldensity-dependent biologi-calprocessesthatarenotincludedinONCHOSIM(Basá ˜nezetal., 2016)–supportsawiderangeofbitingrates(Fig.2a)andresultant transmissionintensities.Forexample,EPIONCHOpredictsanABR ofapproximately20,000tosimulate80%microfilarialprevalence inthepopulationaged≥5years.InONCHOSIM,the correspond-ingABRisapproximately14,000.Thus,EPIONCHOrequireshigher bitingratestosimulatehyperendemicity,implyingahigherrisk ofresurgenceifsomeresidualinfectionremains after interven-tion. Hence, the epidemiological data from some communities areassociated(byEPIONCHO)withconditionshighlypropitious totransmissionanda parasitepopulationresilienttoevenlong andintensiveinterventions.Resurgenceis a probableprojected outcomeinmanyputativelyhyperendemicvillages,especiallyin theRiverGambiafocusin Senegal,whereinferredABRstended tobehigherthan intheRiverBakoyefocus.In ONCHOSIM,the muchlessnon-linearABR-prevalencerelationship(Fig.2b)ensures morehomogeneityamonginferredcommunitytransmission con-ditions,explaining whysustained eliminationis theubiquitous modelledoutcomeforbothfoci.Furthermore,ONCHOSIMassumes lessefficienttransmissionthanEPIONCHOatlowintensities(due todensity-dependentparasiteestablishment;Basá ˜nezetal.,2002) andadecline inmfproduction withincreasingwormage,both makingresurgencelesslikely.Afuturepriorityistounderstand betterthediscrepancybetweentheABR-prevalencerelationship modelledbyONCHOSIMandthedatacollectedbyRenzandWenk (1987)(where ABRs weremeasuredas averages frommultiple yearsandlocationswithinandaroundcommunities,weightedby theproportionoftimecommunityresidentsspentatthese loca-tions)andtorevisitthedensitydependenciesincludedwithinboth models.

Anotherreasonthat resurgence is more likely inEPIONCHO comparedtoONCHOSIMisthelargepopulationassumption inher-enttodeterministicmodellingapproaches.Chanceeliminationby stochasticdemographiceventslikethedeathof(all)infectedhosts isnotconsidered;eliminationcanonlybeachievedandsustained bycrossingthetransmissionbreakpoint,thetheoretical popula-tiondensitybelowwhichtheparasitepopulationcannotsustain itselfanddeclinesterminallytowardsextinction.Thisthreshold –whichdependsonvectorabundanceandtheresultingABR–is thoughttobeverylowforonchocerciasis(Basá ˜nezetal.,2009), and human helminthiasesmore generally (Anderson and May, 1991)Thisisbecauseoftheassumedpolygamousmatingsystem (abroadly-appliedassumptioninhelminthtransmissionmodels, exceptingschistosomeswhicharethoughttoformmonogamous matingpairs;May,1977)andthedegreeof(adult)worm overdis-persionamonghosts.Alessefficientmatingsystemandamore randomly(lessaggregated) distributionof worms among hosts wouldmake elimination byboth models easiertoreach. Addi-tionally,elimination inONCHOSIMisalsoinfluenced bychance (stochastic)demographicevents.

Typically, ONCHOSIM models a relatively small (400–440) human population such that stochastic events may contribute

appreciablyto theelimination outcome,especially when infec-tionprevalenceislowandbecauseparasitesareaggregatedamong hosts;thedeathofthefewindividualswhoharbourthebulkof theparasitepopulationcanleadtoanextinctionevent.However, forthelongandintensiveinterventions(biannualtreatment fre-quencyfortheRiverGambiafocusinSenegal)consideredhere,even increasingthemodelledpopulationsizeto800individualsdidnot altertheprojectedoutcomeofeliminationacrossall27 communi-ties.However,futureanalysesshouldbeconductedtounderstand bettertheinterplaybetweenhowassumptionsonpopulationsize mightaffecttheminimumrequireddurationorfrequencyof treat-mentwithivermectinandtoupdatecurrentestimatedtimeframes foronchocerciasiseliminationamongdifferentendemicitysettings (Stolketal.,2015).

Although the individual rural communities affected by onchocerciasis tend to be relatively small, whether the large-populationassumptionof EPIONCHOortheexplicitly small(or ‘intermediate’800)populationsizemodelledbyONCHOSIMismost appropriatewill cruciallydependonthedegreetowhich com-munitiesarecoupledbythemovementofblackfliesandpeople. NeitherEPIONCHOnorONCHOSIMconsiderthespatialcouplingof proximatepopulationsandespeciallythepossiblereintroduction ofinfectionsfromfociwithongoingtransmission.Thedegreeto whichindividualcommunitiesareexposedtosuchfociwilldepend on–inadditiontoblackflyandhumanmovement–the geographi-calcoverageofaninterventionoverthewiderareaand,inthecase ofMDA,thetreatmentcoverageinnearbycommunities.These fac-tors,andthemoregeneralroleofspatialtransmissionprocesseson thedynamics,controlandeliminationofNTDsarepoorly under-stood,constitutingamajorgapinourunderstandingofhowbest toreachandsustainelimination.

Animportantlimitationofbothmodels’projectionsisthat ento-mologicaldatafromthesefoci–whichbetween2008and2010 were strongly indicative of transmission interruption (Diawara etal.,2009;Traoreetal.,2012)–werenotusedtotrainthe mod-els.Thisisbecauseentomologicalsurveillancewasconductedaway fromthecommunitiesatafewcapturepoints,inamanner simi-lartoOCPcollectionprotocolsduringvectorcontrol.Inprinciple, thesedatacouldbe–andultimatelyshouldbe–usedtoinform modelprojections,butwithoutdetailedspatially-explicitmodels, this would require assumptions that fliesfroma limited num-berofcollectionpoints(4intheRiverBakoyeand3intheRiver Gambiafocirespectively;Diawaraetal.,2009)were representa-tiveofinfectivityratesamongfliesbitingincommunities.Linking spatially-distinctepidemiologicalandentomologicaldata(where theyexist)willbeanimportantfuturechallengetorefiningour modellingprojections.

isnotpossibletoconcludethatmoreepidemiologicaldataalways giveabetterABRestimate,thekeydeterminantoflocal transmis-sionconditions.Indeed,avarietyofotherfactorssuchasecological orenvironmentalchangesthatmightsystematicallyhavechanged ABRsoverthetwodecadesofinterventionalsoimpedeourability toestimateaccuratelylocaltransmissionconditions.

Theresultsofthis analysistogetherwithourpreviouswork (Stolketal.,2015;Coffengetal.,2014a;Turneretal.,2014) sug-gestthatwhetherthe2020/2025onchocerciasiseliminationgoals willbe metcriticallydepends onwhen anintervention began; howintensivelyandeffectively itwasimplemented(e.g.annual orbiannualMDAwithivermectin;levelsofcoverage)andonthe localtransmissionconditions.Inareaswheretheprevailing eco-logicalconditionsarehighlypropitioustotransmission–including many fociin WestAfricawhere theinitialendemic prevalence wasveryhigh(O’Hanlonetal.,2016)andvectorabundancehas returnedtooriginallevelsfollowingOCPvectorcontrol(Lamberton etal.,2014)–eliminationby2020or2025maybeunfeasiblewith ivermectinalone.Insuchplaces(andelsewhere,suchaswhere sub-optimalresponsestoivermectinarepersistent,Frempongetal., 2016),localizedlow-costvectorcontrolshouldbeconsideredas acomplementtoMDAalongsideotherso-calledalternative treat-mentstrategies(WHO,2015)includinganti-wolbachialtherapies (Walkeretal.,2015)toaccelerateprogresstowardsthe2020/2025 goals.Emergingnewtreatmentoptions(Kuesel,2016),particularly moxidectinwhichissimilarbutmoreefficaciousthanivermectin (Awadzietal.,2014),presentpromisingalternatives(Turneretal., 2015b).Whereeliminationappearstohave beenachieved,it is vitalthatappropriateandrobustPTSprotocolsareimplemented forconfirmation.Duetothepoorsensitivityoftheskinsnip diag-nostic,3-to5-yearPTSperiodsofmicrofilarialprevalencemaynot belongenoughtooffersufficientconfidencethatresurgencehas notoccurred.Modelsmustnowbeusedtoexplorehowresurgence manifestsinseroprevalenceandentomologicaldatacollectedusing thelatestand newlyrecommended diagnostictoolswhich will becomethecornerstoneoffutureonchocerciasissurveillancein Africa.

Authorcontributions

MW,WASandMGBconceivedthestudydesign,undertookthe modelling,thedataanalysisandinterpretationoftheresults.MAD andCBhelpedtorefineEPIONCHOandassistedwiththedata anal-ysis.LDandMOTledthecollectionoftheoriginalepidemiological fieldstudyandassistedwiththeinterpretationofthestudydesign andtheprogrammaticandepidemiologicaldata.MW,WASand MGBdraftedthemanuscriptwhichwasread,revisedforcritical intellectualcontent,modifiedandthenapprovedbyallauthors.

Funding

WeacknowledgefundingoftheNTDModellingConsortiumby theBill and MelindaGates Foundation in partnershipwiththe TaskForceforGlobalHealth.MWandMGBalsoreceivedfinancial supportfrom theUNICEF/UNDP/World Bank WHO Special Pro-grammeforResearchandTraininginTropicalDiseases(TDR)and theWellcomeTrust(grantno.092677/Z/10/Z).MADwouldliketo acknowledgetheMedicalResearchCouncil-DoctoralTraining Pro-grammeatImperial CollegeLondon.Thefundershadnorolein studydesign,datacollectionandanalysis,decisiontopublish,or preparationofthemanuscript.Theviews,opinions,assumptions oranyotherinformationsetoutinthisarticlearesolelythoseof theauthors.

Conflictsofinterest

None.

Acknowledgements

WegratefullyacknowledgeDrHansRemmeforhishelpfuland informativediscussionsduringthepreparationofthemanuscript. WealsothankDrJoaquinPradaandDrLloydChapmanfortheir assistanceinproducingsomeofthefigures,andtoDrRoryPostand fordiscussionofrecentworkonsurveillanceandrecrudescence.

AppendixA. Supplementarydata

SupplementarydataincludingtheEPIONCHOsourcecode asso-ciatedwiththisarticlecanbefound,intheonlineversion,athttp:// dx.doi.org/10.1016/j.epidem.2017.02.005.

References

AfricanProgrammeforOnchocerciasisControl,2010.OperationalFrameworkof

OnchocerciasisEliminationwithIvermectinTreatment.WHO/APOC/MG/10.1.

WorldHealthOrganization,Geneva,Switzerland,Availableat:http://www.

who.int/apoc/oncho eliminationreportenglish.pdf.Accessed22August2016.

AfricanProgrammeforOnchocerciasisControl(APOC),2012.EighteenthSessionof

theJointActionForum,Available:http://www.who.int/apoc/about/structure/

jaf/FinalCommuniqueJAF18Englishfinalwithannexes.pdf.Accessed22

August2016.

Anderson,R.M.,May,R.M.,1991.InfectiousDiseasesofHumans:Dynamicsand

Control.OxfordUniversityPress,Oxford.

Awadzi,K.,Opoku,N.O.,Attah,S.K.,Lazdins-Helds,J.,Kuesel,A.C.,2014.A

randomized,single-ascendingdose,ivermectin-controlleddouble-blindstudy

ofmoxidectininOnchocercavolvulusinfection.PLoSNegl.Trop.Dis.8,e2953,

http://dx.doi.org/10.1371/journal.pntd.0002953).

Basá ˜nez,M.G.,Boussinesq,M.,1999.Populationbiologyofhumanonchocerciasis.

Philos.Trans.R.Soc.Lond.B:Biol.354,809–826.

Basá ˜nez,M.G.,Collins,R.C.,Porter,C.H.,Little,M.P.,Brandling-Bennett,D.,2002.

TransmissionintensityandthepatternsofOnchocercavolvulusinfectionin humancommunities.Am.J.Trop.Med.Hyg.67,669–679.

Basá ˜nez,M.G.,Pion,S.D.S.,Boakes,E.,Filipe,J.A.N.,Churcher,T.S.,Boussinesq,M.,

2008.EffectofsingledoseivermectinonOnchocercavolvulus:asystematic

reviewandmeta-analysis.LancetInfect.Dis.8,310–322,http://dx.doi.org/10.

1016/S1473-3099(08)70099-9.

Basá ˜nez,M.G.,Churcher,T.S.,Grillet,M.E.,2009.Onchocerca-Simuliuminteractions

andthepopulationandevolutionarybiologyofOnchocercavolvulus.Adv.

Parasitol.68,263–313,http://dx.doi.org/10.1016/S0065-308X(08)00611-8.

Basá ˜nez,M.G.,French,M.D.,Walker,M.,Churcher,T.S.,2012a.Paradigmlost:how

parasitecontrolmayalterpatternandprocessinhumanhelminthiases.Trends

Parasitol.28,161–171,http://dx.doi.org/10.1016/j.pt.2012.02.004.

Basá ˜nez,M.G.,McCarthy,J.S.,French,M.D.,Yang,G.J.,Walker,M.,Gambhir,M.,

Prichard,R.K.,Churcher,T.S.,2012b.Aresearchagendaforhelminthdiseases

ofhumans:modellingforcontrolandelimination.PLoSNegl.Trop.Dis.6,

e1548,http://dx.doi.org/10.1371/journal.pntd.0001548.

Basá ˜nez,M.G.,Walker,M.,Turner,H.C.,Coffeng,L.E.,deVlas,S.J.,Stolk,W.A.,2016.

Riverblindness:mathematicalmodelsforcontrolandelimination.Adv.

Parasitol.94,247–341,http://dx.doi.org/10.1016/bs.apar.2016.08.003.

Boakye,D.A.,Back,C.,Fiasorgbor,G.K.,Sib,A.P.P.,Coulibaly,Y.,1998.Sibling

speciesdistributionsoftheSimuliumdamnosumcomplexintheWestAfrican OnchocerciasisControlProgrammeareaduringthedecade1984-93,following intensivelarvicidingsince1974.Med.Vet.Entomol.12,345–358.

Boatin,B.,2008.TheonchocerciasiscontrolprogrammeinWestAfrica(OCP).Ann.

Trop.Med.Parasitol.102,13–17,http://dx.doi.org/10.1179/

136485908x337427.

Bottomley,C.,Isham,V.,Vivas-Martinez,S.,Kuesel,A.C.,Attah,S.K.,Opoku,N.O.,

Lustigman,S.,Walker,M.,Basá ˜nez,M.G.,2016.ModellingNeglectedTropical

Diseasesdiagnostics:thesensitivityofskinsnipsforOnchocercavolvulusin

neareliminationandsurveillancesettings.Parasit.Vectors9,343,http://dx.

doi.org/10.1186/s13071-016-1605-3.

Churcher,T.S.,Ferguson,N.M.,Basá ˜nez,M.G.,2005.Densitydependenceand

overdispersioninthetransmissionofhelminthparasites.Parasitology131, 121–132.

Coffeng,L.E.,Stolk,W.A.,Hoerauf,A.,Habbema,D.,Bakker,R.,Hopkins,A.D.,de

Vlas,S.J.,2014a.EliminationofAfricanonchocerciasis:modellingtheimpactof

increasingthefrequencyofivermectinmasstreatment.PLoSOne9,e115886,

http://dx.doi.org/10.1371/journal.pone.0115886.

Coffeng,L.E.,Stolk,W.A.,Zoure,H.G.,Veerman,J.L.,Agblewonu,K.B.,Murdoch,

M.E.,Noma,M.,Fobi,G.,Richardus,J.H.,Bundy,D.A.P.,etal.,2014b.African

ProgrammeforOnchocerciasisControl1995–2015:updatedhealthimpact

estimatesbasedonnewdisabilityweights.PLoSNegl.Trop.Dis.8,e2759,

Convit,J.,Schuler,H.,Borges,R.,Olivero,V.,Domínguez-Vázquez,A.,Frontado,H.,

Grillet,M.E.,2013.InterruptionofOnchocercavolvulustransmissionin

NorthernVenezuela.ParasitesVectors6,289,

http://dx.doi.org/10.1186/1756-3305-6-289.

Diawara,L.,Traoré,M.,Badji,A.,Bissan,Y.,Doumbia,K.,Goita,S.F.,Konaté,L.,

Mounkoro,K.,Sarr,M.D.,Seck,A.F.,etal.,2009.Feasibilityofonchocerciasis

eliminationwithivermectintreatmentinendemicfociinAfrica:firstevidence

fromstudiesinMaliandSenegal.PLoSNegl.Trop.Dis.3,e497,http://dx.doi.

org/10.1371/journal.pntd.0000497.

Eisenbarth,A.,Achukwi,M.D.,Renz,A.,2016.OngoingtransmissionofOnchocerca

volvulusafter25yearsofannualivermectinmasstreatmentsintheVinadu Nordrivervalley,innorthCameroon.PLoSNegl.Trop.Dis.10,e0004392.

Filipe,J.A.N.,Boussinesq,M.,Renz,A.,Collins,R.C.,Vivas-Martinez,S.,Grillet,M.E.,

Little,M.P.,Basá ˜nez,M.G.,2005.Humaninfectionpatternsandheterogeneous

exposureinriverblindness.Proc.Natl.Acad.Sci.U.S.A.102,15265–15270,

http://dx.doi.org/10.1073/pnas.0502659102.

Frempong,K.K.,Walker,M.,Cheke,R.A.,Tetevi,E.J.,Gyan,E.T.,Owusu,E.O.,Wilson,

M.D.,Boakye,D.A.,Taylor,M.J.,Biritwum,N.K.,etal.,2016.Doesincreasing

treatmentfrequencyaddresssuboptimalresponsestoivermectinforthe

controlandeliminationofriverblindness?Clin.Infect.Dis.62,1338–1347,

http://dx.doi.org/10.1093/cid/ciw144.

Gambhir,M.,Singh,B.K.,Michael,E.,2015.TheAlleeeffectandeliminationof

neglectedtropicaldiseases:amathematicalmodellingstudy.Adv.Parasitol.

87,1–31,http://dx.doi.org/10.1016/bs.apar.2014.12.001.

Gardon,J.,Boussinesq,M.,Kamgno,J.,Gardon-Wendel,N.,Demanga-Ngangue,

Duke,B.O.,2002.Effectsofstandardandhighdosesofivermectinonadult

wormsofOnchocercavolvulus:arandomizedcontrolledtrial.Lancet360,

203–210,http://dx.doi.org/10.1016/S0140-6736(02)09456-4.

Golden,A.,Faulx,D.,Kalnoky,M.,Stevens,E.,Yokobe,L.,Peck,R.,Karabou,P.,

Banla,M.,Rao,R.,Adade,K.,Gantin,R.G.,Komlan,K.,Soboslay,P.T.,deLos

Santos,T.,Domingo,G.J.,2016.Analysisofage-dependenttrendsinOv16IgG4

seroprevalencetoonchocerciasis.Parasit.Vectors9,388,http://dx.doi.org/10.

1186/s13071-016-1623-1.

Habbema,J.D.F.,DeVlas,S.J.,Plaisier,A.P.,VanOortmarssen,G.J.,1996.The

microsimulationapproachtoepidemiologicmodelingofhelminthicinfections, withspecialreferencetoschistosomiasis.Am.J.Trop.Med.Hyg.55(Suppl.5), 165–169.

Higazi,T.B.,Zarroug,I.M.A.,Mohamed,H.A.,ElMubark,W.A.,Deran,T.C.M.,Aziz,N.,

Katabarwa,M.,Hassan,H.K.,Unnasch,T.R.,Mackenzie,C.D.,etal.,2013.

InterruptionofOnchocercavolvulustransmissionintheAbuHamedfocus,

Sudan.Am.J.Trop.Med.Hyg.89,51–57,

http://dx.doi.org/10.4269/ajtmh.13-0112.

Hildebrandt,J.C.,Eisenbarth,A.,Renz,A.,Streit,A.,2012.Singlewormgenotyping

demonstratesthatOnchocercaochengifemalessimultaneouslyproduce

progenysiredbydifferentmales.Parasitol.Res.111,2217–2221,http://dx.doi.

org/10.1007/s00436-012-2983-x.

Katabarwa,M.N.,Eyamba,A.,Nwane,P.,Enyong,P.,Kamgno,J.,Kueté,T.,Yaya,S.,

Aboutou,R.,Mukenge,L.,Kafando,C.,etal.,2013a.Fifteenyearsofannualmass

treatmentofonchocerciasiswithivermectinhavenotinterrupted

transmissionintheWestRegionofCameroon.J.Parasitol.Res.2013,420928,

http://dx.doi.org/10.1155/2013/420928.

Katabarwa,M.N.,Lakwo,T.,Habomugisha,P.,Agunyo,S.,Byamukama,E.,Oguttu,

D.,Tukesiga,E.,Unoba,D.,Dramuke,P.,Onapa,A.,etal.,2013b.Transmissionof

OnchocercavolvuluscontinuesinNyagak-BondofocusofnorthwesternUganda

after18yearsofasingledoseofannualtreatmentwithivermectin.Am.J.Trop.

Med.Hyg.89,293–300,http://dx.doi.org/10.4269/ajtmh.13-0037.

Katabarwa,M.N.,Lakwo,T.,Habomugisha,P.,Agunyo,S.,Byamukama,E.,Oguttu,

D.,Ndyomugyenyi,R.,Tukesiga,R.,Ochieng,G.O.,Abwaimo,F.,etal.,2014.

TransmissionofOnchocercavolvulusbySimuliumneaveiinMountElgonfocus

ofEasternUgandahasbeeninterrupted.Am.J.Trop.Med.Hyg.90,1159–1166,

http://dx.doi.org/10.4269/ajtmh.13-0501.

Kim,Y.E.,Remme,J.H.F.,Steinmann,P.,Stolk,W.A.,Roungou,J.P.,Tediosi,F.,2015.

Control,eliminationanderadicationofriverblindness:scenarios,timelines

andivermectintreatmentneedsinAfrica.PLoSNegl.Trop.Dis.9,e0003664,

http://dx.doi.org/10.1371/journal.pntd.0003664.

Koala,L.,Nikiema,A.,Post,R.J.,Paré,A.B.,Kafando,C.M.,Drabo,F.,Traoré,S.,2017.

RecrudescenceofonchocerciasisintheComévalleyinSouthwestBurkinaFaso.

ActaTrop.166,96–105,http://dx.doi.org/10.1016/j.actatropica.2016.11.003.

Kuesel,A.C.,2016.Researchfornewdrugsforeliminationofonchocerciasisin

Africa.Int.J.DrugsDrugResist.6,272–286,http://dx.doi.org/10.1016/j.ijpddr.

2016.04.002.

Lamberton,P.H.L.,Cheke,R.A.,Walker,M.,Winskill,P.,Osei-Atweneboana,M.Y.,

Tirados,I.,Tetteh-Kumah,A.,Baokye,D.A.,Wilson,M.D.,Post,R.J.,Basá ˜nez,

M.G.,2014.OnchocerciasistransmissioninGhana:bitingandparousratesof

host-seekingsiblingspeciesofSimuliumdamnosumcomplex.Parasit.Vectors

7,511,http://dx.doi.org/10.1186/s13071-014-0511-9.

Lamberton,P.H.L.,Cheke,R.A.,Winskill,P.,Tirado,I.,Walker,M.,

Osei-Atweneboana,M.Y.,Biritwum,N.K.,Tetteh-Kumah,A.,Boakye,D.A.,

Wilson,M.D.,Post,R.J.,Basá ˜nez,M.G.,2015.Onchocerciasistransmissionin

Ghana:persistenceunderdifferentcontrolstrategiesandtheroleofsimuliid

vectors.PLoSNegl.Trop.Dis.9,e0003688,http://dx.doi.org/10.1371/journal.

pntd.0003688.

LondonDeclarationonNeglectedTropicalDiseases,2012.Endingtheneglectand

reaching2020goals,Availableat:http://unitingtocombatntds.org/downloads/

press/ntdeventlondondeclarationonntds.pdf.Accessed22August2016.

Lont,Y.L.,Coffeng,L.E.,deVlas,S.J.,Golden,A.,deLosSantos,T.,Domingo,G.J.,

Stolk,W.A.,2017.Modellinganti-Ov16IgG4antibodyprevalenceasan

indicatorforevaluationanddecisionmakinginonchocerciasiselimination

programmes.PLoSNegl.Trop.Dis.11,e0005314,http://dx.doi.org/10.1371/

journal.pntd.0005314.

Lovato,R.,Guevara,A.,Guderian,R.,Proa ˜no,R.,Unnasch,T.,Criollo,H.,Hassa,H.K.,

Mackenzie,C.D.,2014.Interruptionofinfectiontransmissioninthe

onchocerciasisfocusofEcuadorleadingtothecessationofivermectin

distribution.PLoSNegl.Trop.Dis.8,e2821,http://dx.doi.org/10.1371/journal.

pntd.0002821.

May,R.M.,1977.Togethernessamongschistosomes:itseffectsonthedynamicsof

theinfection.Math.Biosci.35,301–343.

Moreau,J.P.,Prost,A.,Prod’hon,J.,1978.Anattempttonormalizethemethodology

ofclinicalparasitologicalsurveysofonchocerciasisinWest-Africa.Med.Trop. 38,43–51.

O’Hanlon,S.J.,Slater,H.C.,Cheke,R.A.,Boatin,B.A.,Coffeng,L.E.,Pion,S.D.,

Boussinesq,M.,Zouré,H.G.,Stolk,W.A.,Basá ˜nez,M.G.,2016.Model-based

geostatisticalmappingoftheprevalenceofOnchocercavolvulusinWestAfrica.

PLoSNegl.Trop.Dis.15,e0004328,http://dx.doi.org/10.1371/journal.pntd.

0004328.

Plaisier,A.P.,vanOortmarssen,G.J.,Habbema,J.D.,Remme,J.,Alley,E.S.,1990.

ONCHOSIM:amodelandcomputersimulationprogramforthetransmission andcontrolofonchocerciasis.Comput.MethodsProgamsBiomed.31,43–56.

Plaisier,A.P.,vanOortmarssen,G.J.,Habbema,J.D.,Remme,J.,1991.The

reproductivelifespanofOnchocercavolvulusinWestAfricansavanna.Acta Trop.48,271–284.

Plaisier,A.P.,Alley,E.S.,Boatin,B.A.,vanOortmarssen,G.J.,Remme,J.,deVlas,S.J.,

Bonneux,L.,Habbema,J.D.,1995.Irreversibleeffectsofivermectinonadult

parasitesinonchocerciasispatientsintheOnchocerciasisControlProgramme inWestAfrica.J.Infect.Dis.172,204–210.

Plaisier,A.P.,Alley,E.S.,vanOortmarssen,G.J.,Boatin,B.A.,Habbema,J.D.F.,1997.

Requireddurationofcombinedannualivermectintreatmentandvector controlintheOnchocerciasisControlProgrammeinwestAfrica.Bull.World HealthOrgan.75,237–245.

Renz,A.,Wenk,P.,1987.Studiesonthedynamicsoftransmissionofonchocerciasis

inaSudan-savannaareainNorthCameroonV.Whatisatolerablelevelofthe annualtransmissionpotential?Annal.Trop.Med.Paraistol.81,263–274.

Rodríguez-Pérez,M.A.,Fernández-Santos,N.A.,Orozco-Algarra,M.E.,

Rodríguez-Atanacio,J.A.,Domínguez-Vázquez,A.,Rodríguez-Morales,K.B.,

Real-Najarro,O.,Prado-Velasco,F.G.,Cupp,E.W.,RichardsJr.,F.O.,Hassan,H.K.,

González-Roldán,J.F.,Kuri-Morales,P.A.,Unnasch,T.R.,2015.Eliminationof

onchocerciasisfromMexico.PLoSNegl.Trop.Dis.9,e0003922,http://dx.doi.

org/10.1371/journal.pntd.0003922.

Schulz-Key,H.,Karam,M.,1986.PeriodicreproductionofOnchocercavolvulus.

Parasitol.Today2,284–286.

Stolk,W.A.,Walker,M.,Coffeng,L.R.,Basá ˜nez,M.G.,deVlas,S.J.,2015.Required

durationofmassivermectintreatmentforonchocerciasiselimimnationin

Africa:acomparativemodellinganalysis.ParasitesVectors8,552,http://dx.

doi.org/10.1186/s13071-015-1159-9.

Tekle,A.H.,Elhassan,E.,Isiyaku,S.,Amazigo,U.,Bush,S.,Noma,M.,Cousens,S.,

Abiose,A.,Remme,J.H.F.,2012.Impactoflong-termtreatmentof

onchocerciasiswithivermectininKadunaState,Nigeria:firstevidenceofthe

potentialforeliminationintheoperationalareaoftheAfricanProgrammefor

OnchocerciasisControl.ParasitesVectors5,28,http://dx.doi.org/10.1186/

1756-3305-5-28.

Tekle,A.H.,Zouré,H.G.,Noma,M.,Boussinesq,M.,Coffeng,L.E.,Stolk,W.A.,

Remme,J.H.F.,2016.Progresstowardsonchocerciasiseliminationinthe

participatingcountriesoftheAfricanProgrammeforOnchocerciasisControl:

epidemiologicalevaluationresults.ParasitesVectors5,66,http://dx.doi.org/

10.1186/s40249-016-0160-7.

Traore,M.O.,Sarr,M.D.,Badji,A.,Bissan,Y.,Diawara,L.,Doumbia,K.,Goita,S.F.,

Konate,L.,Mounkoro,K.,Seck,A.F.,etal.,2012.Proof-of-principleof

onchocerciasiseliminationwithivermectintreatmentinendemicfociin

Africa:finalresultsofastudyinMaliandSenegal.PLoSNegl.Trop.Dis.6,

e1825,http://dx.doi.org/10.1371/journal.pntd.0001825.

Turner,H.C.,Churcher,T.S.,Walker,M.,Osei-Atweneboana,M.Y.,Prichard,R.K.,

Basá ˜nez,M.G.,2013.Uncertaintysurroundingprojectionsofthelong-term

impactofivermectintreatmentonhumanonchocerciasis.PLoSNegl.Trop.Dis.

7,e2169,http://dx.doi.org/10.1371/journal.pntd.0002169.

Turner,H.C.,Walker,M.,Churcher,T.S.,Osei-Atweneboana,M.Y.,Biritwum,N.K.,

Hopkins,A.,Prichard,R.K.,Basá ˜nez,M.G.,2014.ReachingtheLondon

DeclarationonNeglectedTropicalDiseasesgoalsforonchocerciasis:an

economicevaluationofincreasingthefrequencyofivermectintreatmentin

Africa.Clin.Infect.Dis59,923–932,http://dx.doi.org/10.1093/cid/ciu467.

Turner,H.C.,Walker,M.,Lustigman,S.,Taylor,D.W.,Basá ˜nez,M.G.,2015a.Human

onchocerciasis:modellingthepotentiallong-termconsequencesofa

vaccinationprogramme.PLoSNegl.Trop.Dis.9,e0003938,http://dx.doi.org/

10.1371/journal.pntd.0003938.

Turner,H.C.,Walker,M.,Attah,S.K.,Opoku,N.O.,Awadzi,K.,Kuesel,A.C.,Basá ˜nez,

M.G.,2015b.Thepotentialimpactofmoxidectinononchocerciasiselimination

inAfrica:aneconomicevaluationbasedonthePhaseIIclinicaltrialdata.

Parasit.Vectors8,167,http://dx.doi.org/10.1186/s13071-015-0779-4.

WorldHealthOrganization,2012.AcceleratingWorktoOvercometheGlobal

WorldHealthOrganization,2015.StrategicOptionsandAlternativeTreatment StrategiesforAcceleratingOnchocerciasisEliminationinAfrica

WHO/MG/15/20.WorldHealthOrganization,Geneva,Switzerland.

WorldHealthOrganization,2016.GuidelinesforStoppingMassDrug

AdministrationandVerifyingEliminationofHumanOnchocerciasis

WHO/HTM/NTD/PCT/2016.1.WorldHealthOrganization,Geneva,Switzerland.

Walker,M.,Specht,S.,Churcher,T.S.,Hoerauf,A.,Taylor,M.J.,Basá ˜nez,M.G.,2015.

Therapeuticefficacyandmacrofilaricidalactivityofdoxycyclineforthe

treatmentofriverblindness.Clin.Infect.Dis.60,1199–1207,http://dx.doi.org/

10.1093/cid/ciu1152.

Wanji,S.,Kengne-Ouago,J.A.,Esum,M.E.,Chounna,P.W.N.,Tendongfor,N.,

Adzemye,B.F.,Eyong,J.E.E.,Jato,I.,Datchoua-Poutche,F.R.,Kah,E.,etal.,2015.

Situationanalysisofparasitologicalandentomologicalindicesof

onchocerciasistransmissioninthreedrainagebasinsoftherainforestofSouth

WestCameroonafteradecadeofivermectintreatment.Parasit.Vectors8,

http://dx.doi.org/10.1186/s13071-015-0817-2.

West,S.,Munoz,B.,Sommer,A.,2012.RiverblindnesseliminatedinColombia.

OpthalmicEpidemiol.20,258–259,http://dx.doi.org/10.3109/09286586.2013.

836230.

Wilson,N.O.,BadaraLy,A.,Cama,V.A.,Cantey,P.T.,Cohn,D.,Diawara,L.,Direny,A.,

Fall,M.,Feeser,K.R.,Fox,L.M.,Kabore,A.,Seck,A.F.,Sy,N.,Ndiaye,D.,Dubray,

C.,2016.Evaluationoflymphaticfilariasisandonchocerciasisinthree

Senegalesedistrictstreatedforonchocerciasiswithivermectin.PLoSNegl. Trop.Dis.10,e0005198.

Winnen,M.,Plaisier,A.P.,Alley,E.S.,Nagelkerke,N.J.,vanOortmarssen,G.,Boatin,

B.A.,Habbema,J.D.,2002.Canivermectinmasstreatmentseliminate