Global malaria connectivity through air travel

MISSING IMAGE

Material Information

Title:
Global malaria connectivity through air travel
Physical Description:
Mixed Material
Creator:
Huang, Zhuojie
Tatem, Andrew J.
Publisher:
Bio-Med Central (Malaria Journal)
Publication Date:

Record Information

Rights Management:
All applicable rights reserved by the source institution and holding location.
System ID:
AA00017927:00001

Full Text









-0___

(SS
*jfI




/-I.


^-S ...^





Access Minutes
S0-60
6001 120
1>120


P. falciparum Transmission
rn High :75%
Low : 0.5%
SUnstable Transmission
No Transmission


(DI


1I0


U










*I.



I..

















-~ -~ ~W
-- -~ r
r ~r~-~t~- vcvr ~A ~


a ~.a' ~ S


S --- "




hO* ** ^f ^

oooo o ^ .

I W~
flu


Brazil Community
W South America Community
India and Middle East Community
Southeast Asia Commujnity
China Community "
East Africa and Nieri'- Community
Japan Communnity
Philippine Community
Europe Community
North and Central America Community
Vietnam Community


"- -- 'i:i '* -

_'-C, -, ,,', .., ,n


- . ,' I N
4b
-- OL
-- c



9,
S


0
0
0
*
. 0
0x
0
S*
*
*
S


4 .^ .^f" \



t- .*' i^


'NW



p:














0" **0.0

1 4 4 ..


o_ -,. :"-'-'--.
**" 011 ,'
,, _^


0
2) -1--'-~


.00


| -go,
I .. t O-





*q. ^ ff3 *o 0



fA I ,o T-^ -".
d 0


00'***
o- o*'
O0 O"-


South EastAisa Community
0 India Community
@ South America Community
Africa Community
Nigeria Community
0 Parkisten Community
China Community
P. falciparum Transmission
S High :75%
Low : 0.5%
S notable Transmission
No Transmission


* 0
O0

.


* -I,


0 0


00
0
00


0
0 o -
,-, O 0 -
S .-Oo ,*



0 co
@

o 5 -


0._^%
0'


0,-'_- ,- 0-8 -
.:. ...^ ,- _
^**








' 1- '
0- ,,
0*,- -.,
0

0. ,- .

-_0


A)


0


~0 *~*<~>0 0





~


V w

0***


0 Brazil Community.
0 China and Japan Community
0 Indonesia Community
0 North and Central America Community
* W South America Community
* Philippines Community
* Pan Europe-Africa-Middle-East Community
* South EastAsia
P. vivax Transmission
High 11%
Low : 0.2%


Unstable Transmission
No Transmission


0000080
.- _.
* 0 .... 0 0g





000
,o *ooof i




0 *0
*,- 0 0 @00 00001-:'
0 0.* *. o


o0


00


1


B)


*.-_


J \ L..
,D
0





(I


- 9*


0;

I.


--U(L/


lb.-.


*
v















* S

.5.


*, ,
^. **\

*,

*V* .*'*
lb

ot as'




o o}
0'% o


,v*. .,. ... -' ..









...


A)


q K'


* ,r ~ < .^ "1'




,9 ".- ;. ^

*o' -'
*' \ .. 4 .....

see,
.0^


. L- -.


Betweennese Centrality
0 <1 '-
0 1-5
6-10
* 11 -25
O 26-50


0.


R vivax Transmission
High 11%

Low 0.2%


Unstable Transmission
No Transmission


B)


0 1 -

11 -5
26-50 -
R falciparum Transmission
High :75%

Low 0.5%
Unstable Transmission
No Transmission -


2/


Y.53q;.




;* r-


2/


"\ "4 .
- ." *e ,












-- N v --
H ') S -


S^ *. *
* .. :


** **
.0.6
* *


I&'
r


Incoming P. falciparum Flows
0 5000
5001 20000
20001 50000
50001 1000000
S>100000

P. falciparum Transmission
rn High :75%

Low: 0.5%
Unstable Transmission
No Transmission
:- Malaria Eliminating Countries


, ..- .


E



1%


@*
S-


.3

4
A
69
~ *q~j K

3<-.


t


-% I


-t
K


V


-t -.


V


S. gi
.6.
S L
If_.*.
*


6 -
'* ^ /
4 .


] 6V



n' U

St
*f-4.


Incoming P. vivax Flows
0 5000
5001 -20000
20001 -50000
S50001 100000
S>100000

P. vivax Transmission
rn High 11%

Low: 0.2%
Unstable Transmission
No Transmission
Malaria Eliminating Countries


A.


i~ ~
~SS ~ ft
q
\ ~'
3-


.-^-(-"^---^

(7-
\r
v












** '^ r^ -i

-^


^*<;
^c- ^ .
-1

^ . \


S <^


N\


-. N ,.


-.


>91
S.

I


S
S es


Sv^


<. 9


S
S
S

V U
t
St
-4.-


'St


-4
.SS~
'Vi


Sib

'4~
*'*i;


*1


ll--


liP"




Full Text

PAGE 1

RESEARCHOpenAccessGlobalmalariaconnectivitythroughairtravelZhuojieHuang1,2,3,4*andAndrewJTatem5,6AbstractBackground: Airtravelhasexpandedatanunprecedentedrateandcontinuestodoso.Itseffectshavebeenseen onmalariainratesofimportedcases,localoutbreaksinnon-endemicareasandtheglobalspreadofdrug resistance.Witheliminationandglobaleradicationbackontheagenda,changinglevelsandcompositionsof importedmalariainmalaria-freecountries,andthethreatofartemisininresistancespreadingfromSoutheastAsia, thereisaneedtobetterunderstandhowthemodernflowofairpassengersconnectseach Plasmodium falciparumand Plasmodiumvivax -endemicregiontotherestoftheworld. Methods: Recentlyconstructedglobal P.falciparum and P.vivax malariariskmaps,alongwithdataonflight schedulesandmodelledpassengerflowsacrosstheairnetwork,werecombinedtodescribeandquantifyglobal malariaconnectivitythroughairtravel.Networkanalysisapproacheswerethenutilizedtodescribeandquantifythe patternsthatexistinpassengerflowsweightedbymalariaprevalence.Finally,theconnectivitywithinandtothe SoutheastAsiaregionwherethethreatofimportedartemisininresistancearisingishighest,wasexaminedto highlightriskroutesforitsspread. Results: Theanalysesdemonstratethesubstantialconnectivitythatnowexistsbetweenandfrommalaria-endemic regionsthroughairtravel.Whiletheairnetworkprovidesconnectionstopreviouslyisolatedmalariousregions,itis clearthatgreatvariationsexist,withsignificantregionalcommunitiesofairportsconnectedbyhigherratesofflow standingout.Thestructuresofthesecommunitiesareoftennotgeographicallycoherent,withhistorical,economic andculturaltiesevident,andvariationsbetween P.falciparum and P.vivax clear.Moreover,resultshighlighthow wellconnectedthemalaria-endemicareasofAfricaarenowtoSoutheastAsia,illustratingthemanypossibleroutes thatartemisinin-resistantstrainscouldtake. Discussion: Thecontinuinggrowthinairtravelisplayinganimportantroleintheglobalepidemiologyofmalaria, withtheendemicworldbecomingincreasinglyconnectedtobothmalaria-freeareasandotherendemicregions. Theresearchpresentedhereprovidesaninitialefforttoquantifyandanalysetheconnectivitythatexistsacrossthe malaria-endemicworldthroughairtravel,andprovideabasicassessmentoftherisksitresultsinformovementof infections.BackgroundTheworldwideairtravelnetworkhasexpandedatanexceptionalrateoverthepastcentury.Internationalpassengernumbersareprojectedtorisefrom1.11billionin 2011to1.45billionby2016,withanannualgrowthrate of5.3%[1].Today,thereare35,000directscheduled routesontheairtravelnetwork,with865newroutes establishedin2011[2].Malaria-endemicareasaremore connectedtotherestoftheworldthanatanytimein history,withthediseaseabletotravelatspeedsof600 milesperhourwithininfectedpassengers.Thegrowth oftheairtravelnetworkresultsinsubstantialconcerns andchallengestotheglobalhealthsystem,withaneed toplacemoreemphasisonevidence-drivensurveillance andreportingthatincorporatesspatialandnetworkinformation[3-6]. Risingratesoftravelbetweenmalaria-freeand-endemic countrieshaveledtogeneralpatternsofincreasedratesof importedmalariaoverrecentdecades[7-10].Duetoinfrequentencounters[3,9],importedcasescanchallenge healthsystemsinnon-endemiccountries,withdifficulties indiagnosis[10],misdiagnosisanddelaysintreatment [11,12],aswellassignificanttreatmentexpenses[13]. *Correspondence: seenhzj@gmail.com1CenterforInfectiousDiseaseDynamics,PennsylvaniaStateUniversity, UniversityPark,PA,USA2DepartmentofBiology,PennsylvaniaStateUniversity,PA,USA Fulllistofauthorinformationisavailableattheendofthearticle 2013HuangandTatem;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsofthe CreativeCommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse, distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.HuangandTatem MalariaJournal 2013, 12 :269 http://www.malariajournal.com/content/12/1/269

PAGE 2

Further,flightsmaybringinfectedvectors,resultingin “ airportmalaria ” ,wherepatientswhodonothaveaforeign travelhistorybecomeinfectedthroughbeingbitteninthe vicinityofinternationalairports[14-17].Patternsin importedcasesandairportmalariahavebeenshowntobe relatedtoacombinationofthenumbersoftravellersand themalariariskatthedestinati on[3,16],andtheserelationshipswillcontinuetoevolveasnewroutesbecome established. Theflowofpeopleviaairtravelbetweenendemicareas mayincreasetherisksofre-emergenceorresurgence[18] inpreviouslymalariafreeorlowtransmissionareas[19]. TheautochthonousmalariaoutbreaksinVirginiain2002 [20],Floridain2003[21]andGreecein2011[22],forexample,demonstratethecontinuedrisksoflocaloutbreaks followingreintroductionthroughairtravel,thoughsuch occurrencesarerare[23].Further,theexamplesofmalaria resurgenceinislandnations,suchasSriLanka[24], Mauritius[25]andMadagascar[26],aftercontrolmeasureswererelaxedreinforcetheimportanceofvigilance androbustsurveillanceintermsofhumanmovementin preandpost-eliminationperiods[18].Identifyingtherisks ofmalariamovementthroughtheairtravelnetworkcan provideanevidencebasethroughwhichpublichealth practitionersandstrategicplannerscanbeinformedabout potentialmalariainfluxesandtheirorigins[3,27]. Meanwhile,growingconcernshavebeenraisedabout thepossiblespreadofartemisininresistancefromthe GreaterMekongsubregioninSoutheastAsiatoother endemicareas.Recentresearchhashighlightedincreasing numbersofpatientsshowingslowparasiteclearancerates followingtreatmentwithartemisinin-baseddrugsin theCambodia-ThailandborderandThailand-Myanmar borderregions[28-30].Tremendoushealthandsocioeconomiccostsoccurredwhenchloroquine-resistantparasitesarrivedinsub-SaharanAfricafromSoutheastAsia andspreadacrossthecontinent[31,32].Similarly, sulphadoxineandpyrimethamineresistanceemergedin AsiaandspreadtoAfrica[33,34].TheWHOreportsthat thereisalready “ atleastonestudywithahightreatment failurerate( 10%)reportedfromsixofthe23African countriesthathaveadoptedartesunate-amodiaquinecompound ” [35],andfearremainsoverthespreadofartemisininresistancefromSoutheastAsiatoAfrica,thatcould underminecurrentcontrolandeliminationefforts,with noalternativedrugscomingintheforeseeablefuture. Ratesofimportedmalaria,risksofresurgenceandthe spreadofdrugresistancearealltodayinfluencedbyhow theglobalairtravelnetworkconnectsupthemalariaendemicregionsoftheworld,andthenumbersofpassengersmovingalongit.Here,recentlyconstructedglobal Plasmodiumfalciparum and Plasmodiumvivax malaria prevalencemapsarecombinedwithdataonmodelledpassengerflowsacrosstheairnetwork,todescribeand quantifyglobalmalariaconnectivitythroughairtravelin 2010.Weightednetworkanalysisstatisticsarederivedto examine:(i)whichregionsshowgreatestconnectivityto P.falciparum and P.vivax malaria-endemiczones;(ii)where thelargestestimatedpassengerflowsfromendemicareas occur;(iii)whichregionsform ‘ communities ’ ,wherebymalariainfectionflowswithinthemarelikelytobelargerthan betweencommunities,andfinally,(iv)wherethethreatof importedartemisininresistanceishighestviaairtraffic, andthepossibleriskroutesforthespreadofresistance withinandfromSoutheastAsia.MethodsAirportlocations,flightroutesandpassengerflowmatrixInformationonthelongitude,latitude,citynameandairportcodeforatotalof1,449airportswhichservecities withmorethan100,000people,andamodelled ‘ actual ’ trafficflow(i.e.numberofpassengerstravellingbetween eachlocationandeveryother,irrespectiveofstopovers) connectivitylistwith644,406routesamongsttheseairportswereobtained[6,36,37].Thislistdocumented,for eachoriginandfinaltraveldestination,theestimated numberofpassengerstakingthisroute[36]regardingthe hub-and-spokestructureoftheairtravelnetwork[38].A connectivitymatrixwasthencreatedfromtheconnectivitylist,quantifyingthevolumesandthedirectionalitiesof thepassengerflowsbetweentwoairports.Withinthispassengerflowmatrix,23,785origin – destinationpairswere connectedbydirectflightsbetweentwoairports,291,745 pairswereconnectedbyroutesinvolvingone-stopand 328,876pairsrequiredtwo-stopstoconnect.Thetravel volumesontheroutesweremodelledbasedprimarilyon publiclyavailabledatasetsunderageneralizedlinear modelframework.Fullmodeldetailsareprovidedin Huang etal. [36],butinbrief,toconstructthematrix, topologicalcharacteristicsoftheairtravelnetwork,city population,andlocalareaGDP,amongstothers,wereutilizedascovariates.Actualtravelvolumesfortrainingand validationwereextractedandassembledfromvarious transportationorganizationsintheUSA,Canadaandthe EuropeanUnion.Aloglinearmodelcontrollingforrandomeffectsonorigin,destinationandtheairporthierarchywasthenbuilttopredictpassengerflowsonthe network.Themodeloutperformedexistingairtravelpassengerflowmodelsintermsofpredictionaccuracy[36].MalariadistributionGlobal P.falciparum and P.vivax prevalencemapswere obtainedfromtheMalariaAtlasProject[39]andthe methodsbehindtheirconstructionarepresentedin Gething etal. [40,41].Inbrief,22,212communityprevalencesurveyswereusedincombinationwithmodelbasedgeostatisticalmethodstomaptheprevalenceof P.falciparum globallyin2010withinlimitsoftransmissionHuangandTatem MalariaJournal 2013, 12 :269 Page2of11 http://www.malariajournal.com/content/12/1/269

PAGE 3

definedbyannualparasiteincidenceandsatellitecovariatedata.Similarly,9,970geocoded P.vivax parasiterate ( P.vivax PR)surveyscollectedbetween1985and2010 wereutilizedinaspatiotemporalBayesianmodel-based geostatisticalapproachtomapendemicity,undertherestrictionsofamaskofthestable/unstableendemicity [41]andinformationontheprevalenceoftheDuffy bloodgroup[42].Distributionsof Plasmodiumovale, PlasmodiummalariaeorPlasmodiumknowlesi arenot includedhere,sincesimilardatasetsontheirdistributionsdonotyetexist.Also,theseasonalclimaticconstraintsthataffectthetransmissionof P.falciparum and P.vivax arenotincludedhere,butmodelsofeach[43] willbeincludedinfuturework.WeightednetworkanalysisandcommunitydetectionMalariaprevalencecanvarygreatlyintheregionaround airportsandthecitiestheyserve,andtravellerstaking flightsfromaspecificairportmayresidemany kilometresfromtheairportsinhighertransmissionareas thanfoundinthevicinityoftheairport.Thus,simply assigningthepredictedprevalencefromthemalaria mapsatthelocationofeachairportcouldunderestimate theriskandrateofinfectionexportationattheairport inquestionandunderrepresentitscontributiontoglobal malariaconnectivity.Therefore,followingHuang etal. [6],localaccessibilitytoeachairportwasconsideredby assumingthatpassengerswouldtravellessthan50km withatraveltimelessthantwohourstoaccessan airporttotakeaflight.Underthisassumption,the P.falciparum andthe P.vivax malariaprevalenceassigned toanairportwereobtainedasthemaximumprevalence fromthemalariamapswithinamaskof50kmandtwohourtraveltime(Additionalfile1),inwhichthemask wasgeneratedusingaglobaltraveltimemap[44].This choiceofthemaximumvaluewascreatedfora ‘ worst casescenario ’ assessment,butformostoftheanalyses conductedhere,relativedifferencesbetweenairports wereassessedorpresented,thusmakingthechoiceof, forexample,maximum,minimumormean,irrelevantin mostcases.Likewise,anindicatorthatdefineswhether anairportislocatedinthestable/unstableendemiczone wascreatedaccordingtothesamemask,inwhichthe indicatordefineswhetherthemajorityareaofthemask islocatedinthestable/unstablezone.Additionalfile1 showsthistraveltime/distancemaskwiththeglobal traveltimemap. Theaboveapproachensuredthateachairporthadan assignmentofa P.falciparum and P.vivax prevalence rate(orunstable/malariafree),whichcouldthenbeused asaweightingappliedtothepassengerflowestimatesto deriverelative ‘ malariaflow ’ indicesforeachroute, thatcouldbecomparedtootherroutesacrossthe globalnetworktoanalysemalariaconnectivity.Thus, P.falciparum and P.vivax flowswerecalculatedoneach route(eitherdirect,one-stoportwo-stops)asorigin prevalence*estimatedpassengervolume,toproduce P.falciparum and P.vivax malarianetworks. Agroupofweightedcentralityanalysesandnetwork communitypartitionanalyseswereperformedonthe malarianetworkstoquantifyfeaturesofglobalmalaria connectivity.First,thein-strengthandout-strengthof eachconnectionwascalculatedasthesumofincoming andoutgoingmalariaflowviaallpossibleconnections (directflight,one-stoportwo-stops)asfollows: si XN j 1aijwij 1 inwhich aijistheairportadjacencymatrix(inabinary form)and wijistheweightedmalariaflow.Thismetric estimatesthetotalweightofmalariaflowsthatairports sendandreceive. Followingthis,weighted “ betweenness ” analyseswere performedonthemalariaflowmatrices.Betweennesscentralitymeasuresthenumberofshortestpathsgoing throughaspecificvertex[45].Inaweightednetwork,betweennesscentralityisausefullocalmeasureoftheload placedonthegivennodeinthenetworkaswellasthe node ’ simportancetothenetworkotherthanjustconnectivity[46].Itisoftenusedintransportnetworkanalysisto provideanapproximationofthetraffichandledbythe vertices[47].Thus,hereitprovidesanindicationofthe statusofeachairportasa ‘ malariahub ’ throughitsimportanceintheglobalflowofmalariainfectionsviaair travel-i.e.,ameasureofhowmanyinfectionslikelypass througheachairporteachyear,relativetootherairports, andhowlikelyanairportwouldrouteandspreadmalaria infectionsonward.Thebetweennesscentralityiscalculatedas: CBv Xs v t stv st 2 inwhich stisthetotalnumberofshortestpathsfrom nodestonodetand st( v )isthenumberofthosepaths thatpassthroughv.Notethatontheweighted P.falciparum/P.vivax networks,thedistancebetweenthetwo nodessandtisdefinedbythesumof P.falciparum flows or P.vivax flowsastheedgeweightonthispathundera classicalDijkstrashortest-pathcalculationframework[48]. Anormalizedbetweennesswasusedas CBnormv CBv X CBv = n 3 wherenwasthenumberofnodes(airports)intheair travelnetwork[47].HuangandTatem MalariaJournal 2013, 12 :269 Page3of11 http://www.malariajournal.com/content/12/1/269

PAGE 4

Communitiesinanetworkreflectthepartitionof nodesthataredenselyconnectedandseparatedfrom theothernodesinthenetwork,thusthesenodes “ probablysharecommonpropertiesand/orplaysimilarroles withinthegraph ” [49].Bymappingcommunitiesonthe malarianetworksdefinedhere,groupsofairportsthat showstronglinksintermsoflikelymovementsofinfectionswereidentified.Thispotentiallyhasutilityinterms ofprovidingevidenceuponwhichregionalsurveillance strategiescanbedesigned[50,51].NewmanandGirvan [52]defineamodularityscorewhichmeasuresthequalityofnetworkpartitionsas: Q 1 2 m Xi ; jWijŠ kikj2 m ci; cj 4 inwhich, wijrepresentstheweightoftheedgebetweeni andj(herethesearethe P.falciparum and P.vivax flow matrices), ki=jWijisthesumoftheweightsofthe connectionsattachedtoairporti, ciisthecommunityto whichairportiisassigned; ( ci,cj)is1if ci= cj,otherwise m 1 2Wij. Amultilevelalgorithmforcommunitydetection[53] wasimplemented.Thismethodutilizesaniterativeapproachthatmergescommunitiestomaximizethe modularityscore:Firstly,modularityisoptimizedby allowingonlylocalchangesofcommunities;secondly, theestablishedcommunitiesarecombinedtogetherto constructanewnetwork.Thesetwopassesarerepeated iterativelyuntilnoincreaseofmodularityispossible. Thenumberofcommunitiesreturnedbythisalgorithm yieldsthemaximummodularityscore. Afterward,asimpleWilcoxonrank-sumtest[54]was performedonthedifferencesbetween “ internal ” and “ external ” degreesofacommunityinordertotest whethertheestablishmentofcommunitieswassignificant.Airconnectionsweredefinedwithinacommunity as “ internal ” andtheconnectionsconnectingtheairportsofacommunitywiththerestofthenetworkas “ external ” .Thenullhypothesisofthistestwasthatthere wasnodifferencebetweenthenumberofinternaland externalroutesincidenttoanairportofthecommunity.ResultsTheresultsoftheglobalmalariaconnectivityanalyses arepresentedintwosections:(i)analysesfocussedon theconnectionofendemicmalariaregionstoeachother andtomalaria-freeareas,thathasparticularrelevance toimportedmalariaandmalariaresurgenceandreemergence;and,(ii)analysesexaminingtheconnections betweenSoutheastAsiaandtherestofthemalariaendemicworld,whicharerelevanttothespreadofartemisininresistance.Connectivitywithinendemicareasandtonon-endemic areasFigure1showstheresultsofregionalcommunitystructureanalysesbasedontrafficflowdataoverlaidonthe P.falciparum/P.vivax endemicityandstable/unstable transmissionlimitsmaps.TheWilcoxontestresults showthattheinternaldegreesfortheairportswithinall communitiesaresignificantlydifferentfromtheexternal degrees,withpvaluesof<0.01,thusthecommunitypartitionsshownaresignificant.Themapshighlightthose countriesthatformcommunitieslinkedbyhighlevelsof trafficscaledby P.falciparum/P.vivax prevalenceat theiroriginendemicarea.Additionalfile2describes similaranalysesbasedsolelyonthetravelnetworkdata fromHuang etal. [36].Thecommunitiesdetectedreflectthearchitectureoftheairnetwork,andhowthis relatestomalariaendemicityaroundtheworld.Geographicalcontiguityisclearlyevident,astrafficlevelson shorterdistanceroutesaregenerallyhigherthanonlongerdistanceroutes,butinterestingpatternsrelatingto historicaltiesemerge.Forinstance,for P.falciparum LondonformspartoftheNigeriacommunity,butParis showsstrongertiestotheremainderofsub-Saharan Africa.Theseconnectionsareoftenreflectedinimportedmalariastatistics,withNigeriabeingthemain sourceof P.falciparum casesseenintheUK,butfor France,theFrench-speakingAfricancountriesarethe mainorigin.Similarly,UKairportsalsoformpartofthe India/Bangladeshcommunity,wherehistoricaltiesexist, resultinginsignificanttravelbetweenthetworegions, andconsequent P.falciparum and P.vivax malariaimportationtotheUK.Tiesalsoexistbetweenthewestern USAandEastAsia,whichformasingle P.falciparum community(Figure1A).Additionalfile3showsacommunitydetectionanalysisforairportswithdirectconnections,one-transferconnectionsandtwo-transfer connectionsfrommalaria-endemicareas. Toexaminedirectionalandnetpotentialmovements ofpeopleandparasitesbetweenairportsindifferent countries,theinternationalrouteweightingswere summeduptoidentifypossible “ source ” and “ sink ” airportsofmalariainfections(Additionalfile4).Here,the weightsofallpossibleincomingflowsforairportsinthe non-endemicareas,andtheweightsofallpossibleoutgoingflowsfromairportsinendemicareaswere summeduptodefine “ vertexstrengths ” ofimportation andexportation(notethatonlytheroutesconnecting twodifferentcountriesregardlessofthedomesticroutes wereconsidered).Inthistable,airportsintheFarEast andMiddleAsiasuchasSingapore,HongKong,Dubai, andSharjahdisplaythehighestimportationvalues(note thatSingaporerankedthefirstinbothcategories).Unsurprisingly,majorairhubsinEurope(suchasairports inLondon,ParisandFrankfurt)alsoshowedhighHuangandTatem MalariaJournal 2013, 12 :269 Page4of11 http://www.malariajournal.com/content/12/1/269

PAGE 5

Figure1 (Seelegendonnextpage.) HuangandTatem MalariaJournal 2013, 12 :269 Page5of11 http://www.malariajournal.com/content/12/1/269

PAGE 6

potentialincoming P.falciparum flows.Miamiisthe onlyairportintheUSAontheimportationflowtopten lists,withitsstrongconnectionstoCentralandSouth America.Intermsofexportation,thelargestairportsby trafficcapacityandconnectionstotherestofthe malaria-endemicworldwerehighlighted(Additionalfile4). Mumbaiwasrankedfirstasthelargestexporterof P.falciparum and P.vivax flows,suggestingthatitlikely actsasanimportantportalforspreadingmalariatotherest oftheworld. Thebetweennesscentralitymetricwasutilizedtoinspecttheconnectivityfromendemicareas.Asthebetweennessmetric CB( v )isdefinedasthenumberof shortestpathsconnectinganytwoairportsthatinvolvea transferatairportv,highcentralityairportsinendemic areasprovidehubsforpeopleoriginatingatlessaccessibleairportsinremoteplacestoreachtherestof theworld.Thusthebetweennesscentralitypresentsan initialquantificationforthelikelihoodthatanairport routesorspreadsmalariaflowasa ‘ malariahub ’ .Table1 showsthetoptenhighestbetweennesscentralityairportsfortransferring P.falciparum flowand P.vivax flowelsewhere.Forthe P.falciparum flow,international airportsinAfricaplayimportantrolesashubsforroutinginfections.Someairportsareobservedtohavesmall degrees(lownumbersofconnectingroutes)andlarge centrality(importanceasahub),whichcanbeconsideredasanabnormality[47].Theseairportsconnectless accessibleandconnectedairportsinendemicareasto otherairportsintheworld.For P.vivax flow,Asian internationalhubsplaymoreimportantroles.Ofinterest isPhoenixairport,whichrankedthesixthintermsof P.vivax centrality,suggestingthatitplaysanimportant roleasagatewayinlinking P.vivax -endemicareasto theUSA.Additionalfile5presentsthespatialdistributionofbetweennesscentralityscoresforairports, weightedby P.falciparum or P.vivax flows. Tofurtherinvestigatetheeffectsofflowsfromendemic zones,Additionalfile6AandBshowsthesumsofinternationalincomingriskflowsforalltheairportsinthose36 countriesthathavenationalpoliciesformalariaelimination,andareclosesttoeliminatingthedisease[55].Importationofinfectionsthreatensthesuccessofelimination programmes[19]andwhileairtravelmaynotbethe highestrisksourcefortheseintroductionsformostof thesecountries,itremainsapotentiallyimportantsource ofincominginfections.Fromthesetwomaps,itcanbe seenthatChinaandcountriesinMiddleAsiaaresubjected tothegreatestpressureofincomingflows,relativetoother eliminationcountries,duetotheirlargerincomingtraffic volumesfromendemicregionselsewherearoundthe world.Tosumup,detailedanalysesonairportconnectivity areprovidedinAdditionalfiles2,3,4,5and6.ConnectivitytoSoutheastAsiaFigure2mapsoutthepassengerflowsscaledbyorigin prevalencefor P.falciparum and P.vivax fromthe GreaterMekongsubregion.Significantamountsofflow exchangewithinSoutheastAsiacanbeseeninthe close-upsubsets.Forboth P.falciparum and P.vivax it canbeseenthattheconnectivity,throughnumbersof travellers,toLatinAmericanendemicregionsisweak, butthatmuchstrongerconnectionstosub-Saharan AfricaandtheIndiansubcontinentexist.IncreasingconnectionsthroughtradeandlabourmarketsbetweenAsia andAfricaoverthepastdecadeisexemplifiedherein thestrongconnectionsbetweentheSoutheastAsianregionandallofsub-SaharanAfrica ’ smajorairporthubs. Additionalfile7presentsthetoptenriskroutesspreadingdrugresistanceof P.falciparum and P.vivax from theGreaterMekongsubregiontonon-Asiandestinations,withestimated P.falciparum/P.vivax flowand thenumberofstopsneededtotravelfromtheorigincity tothedestinationcityshown.DiscussionThecontinuinggrowthinairtravelisplayinganimportantroleintheglobalepidemiologyofmalaria.Flight routesnowconnectpreviouslyisolatedmalaria-endemic regionstotherestoftheworld,andtravellersonthese routescancarryinfectionstotheoppositesideofthe worldinlessthan24hours.Whilemanyendemicareas stillremainrelativelyisolated,themalaria-endemic worldisbecomingincreasinglyconnectedtoboth malaria-freeareasandotherendemicregions.Theimpactsofthiscanbeseeninimportedcases,vectorinvasionsandthespreadofdrug-resistantparasitestrains. Hereaspatialnetworkanalysisapproachwerepresented todemonstratetheconnectivitythatexistsacrossthe malariaendemicworldthroughairtravel,andprovide quantitativeindicatorsoftherisksitresultsinformalariamovement. (Seefigureonpreviouspage.) Figure1 Spatialdistributionof Plasmodiumfalciparum/Plasmodiumvivax networkcommunitiesoverlaidon Plasmodiumfalciparum/ Plasmodiumvivax prevalencemaps.A) P.falciparum multilevelmembership. B) P.vivax multilevelmembership.Thesetwomapsshowonly airportsthathavedirectconnectionsfromendemictonon-endemicareas,thoughfullorigin – destinationflowestimateswereusedin calculations.Theinsetmapspresentclose-upviewsoftheUSAandwesternEurope.Airportswiththesamecommunitymembership(indicated bythesamecolour)displaystrongerlinksintermsoflikelymovementsofinfectionsbetweenthemthantoairportsinothercommunities.Note thatinthe P.vivax map,communitieswithlessthantenairportsarenotshown. HuangandTatem MalariaJournal 2013, 12 :269 Page6of11 http://www.malariajournal.com/content/12/1/269

PAGE 7

Resultshighlightthesubstantialconnectivitythatnow existsbetweenandfrommalaria-endemicregions throughairtravel.Whiletheairnetworkprovidesconnectionstopreviouslyisolatedmalariousregions,itis clearthatgreatvariationsexist,withsignificantregional communitiesofairportsconnectedbyhighratesof prevalence-scaledflowstandingout(Figure1).The structuresofthesenetworksareoftennotgeographically coherent,withhistorical,economicandculturaltiesevident.Asnewroutescontinuetobeestablished,these communitieswilllikelychange,withnewpopulartravel routes,suchasthosebetweenChinaandAfrica[3]likely alteringglobalmalariaflowroutes,andnewdestinations thatmightencounterincreasedrisksofimportedmalariawillemerge(Additionalfile4andTable1).These communitymaps(Figure1)andlistsofcitiesbylikely import/exportofinfections(Additionalfile4)andhubs forinfectionflow(Table1)provideaquantitativepicture ofhowmalariainfectionsarelikelymovingglobally throughairtravel,andinformationfromwhichglobal surveillancestrategydesigncandrawupon.Additional file4andTable1highlightthatcertainairportsprovide significanthubsandgatewaysforthemovementofinfectionsandtheirentryintocountries,andthatthese arewidelydistributedacrosstheworld.Theirrolein providingimportantnodesasbothsignificantthroughflowofinfectionsinthenetwork,andentryandexit gatewaysforcasesto/fromregionsmeansthattheypotentiallyrepresentvaluablesentinelsitesforfocussed surveillance.Finally,Figure2providesastarkreminder ofhowwellconnectedthemalaria-endemicareasof AfricaarenowtoSoutheastAsia,illustratingthemany possibleroutesthatartemisinin-resistantstrainscould take.Theseroutescanprovideafirst-stepquantification tosupporttheglobalplanagainstartemisininresistance containment[35]anddesignofsurveillancesystems [56],andshouldberefinedwithinformationonthelocationsofresistancefound.Suchdatacouldalsoinform decisionsonwhereandhowtolimittheriskofspread, forexamplebypre-travelorarrivalscreeningand treatment. Arangeoflimitationsanduncertaintiesexistinthe analysespresentedhere.Intermsofthequantificationof malariatransmission,theuseofstaticmapsofannual Table1Topten Plasmodiumfalciparum/Plasmodiumvivax betweennesscentralityairportswiththeirdegreesina sub-networkthatonlycontains directlinksfromairportsin Plasmodiumfalciparumor Plasmodiumvivax -endemicareasTopten P.falciparum centralityairports AirportCityCountryNormalizedbetweennesscentralityDegree NBONairobiKenya47.3580 MBAMombasaKenya32.4427 JROKilimanjaroTanzania32.3914 BOMMumbaiIndia30.41104 ADDAddisAbabaEthiopia28.2164 DELDelhiIndia23.16111 JIBDjiboutiDjibouti19.7715 ADEAdenYemen18.6315 MGQMogadishuSomalia14.458 HREHarareZimbabwe14.3520 Topten P.vivax centralityairports AirportCityCountryNormalizedbetweennesscentralityDegree BKKBangkokThailand96.43146 ICNSeoulSouthKorea78.12150 DELDelhiIndia59.55133 BOMMumbaiIndia34.17116 KMGKunmingChina30.7990 PHXPhoenixUSA28.6391 DPSDenpasarBaliIndonesia27.9434 SJOSanJoseCostaRica27.7237 DOHDohaQatar25.91100 TASTashkentUzbekistan25.8569Thebetweennesscentralityscoresshowhowmanyshortestpathsgothroughanairport,thustheyhighlightthepotentialthatairportsmightrouteinf ection flows,actingas ‘ malariahubs ’ .Thedegreemeasureshowmanyroutesarelinkedtoanairport.HuangandTatem MalariaJournal 2013, 12 :269 Page7of11 http://www.malariajournal.com/content/12/1/269

PAGE 8

averageprevalence[40,41]neglectstheseasonalityin transmissionthatiscommontomanyareas,andalsothe substantialchangesintransmissionintensityseeninavarietyoflocationsinrecentyears[57].Inmostpartsofthe world,thedensitiesof Anopheles mosquitoeschangeseasonally,thusimpactingthereceptivityoftheseareasto malariaflowsincomingthroughairtravel.Whiledataon changesin Anopheles densitiesgloballyarenotavailable, temperature-drivenmodelsofmalariatransmissionsuitability[43]couldbeintegratedinfutureworktobetter accountforthisandtheseasonallyvaryingtopologiesof theglobalmalariaconnectivitynetworkstudied.The demographicandbehaviouraldifferencesbetweenpassengersarenotaccountedforhere.Thosetakingregularair travelareoftenricher[58],andlesslikelytobeinfected, whilethosethatareactuallyinfectedandshowingsymptomsmaybelesslikelytotravel.Hence,theairtravel passengerdatasetusedhereclearlycontainssomebiases whenaddressingmalariarisks.Further,onlyparasite prevalencewasusedasamalariametric,andwhilethis Figure2 Estimatedrelative Plasmodiumfalciparum/Plasmodiumvivax flowsoriginatingfromtheGreatMekongsubregionoverlaidon Plasmodiumfalciparum/Plasmodiumvivax prevalencemaps.A) P.falciparum flowsoriginatingfromtheGreatMekongsubregion. B) P.vivax flows originatingfromtheGreatMekongsubregion.Theflowsincludeestimatedpassengernumbers,includingdirect,one-transferandtwo-transferflig ht routes.Theinsetmapsshowclose-upviewsforairportsintheGreaterMekongsubregion. HuangandTatem MalariaJournal 2013, 12 :269 Page8of11 http://www.malariajournal.com/content/12/1/269

PAGE 9

maybeanadequatemeasureofpopulationprevalenceatoriginlocations,itisnotsoappropriateforassessingtheriskof infectionacquisitionfornavetravellers,andentomologicalbasedindicesarelikelymoreappropriatehere,asusedin morelocalstudies[27,51,59].Fin ally,theexaminationofrelativeartemisininresistancespreadriskfocusessimplyonall travelfromfourcountries, andthusdoesnotaccountfor anyheterogeneityinresistanceintheregion. Uncertaintiesandlimitationsrelatingtothetraveldata usedalsoexist.Themodelledpassengerflowsrepresent justa2010snapshot,andthusroutesandchangessince thenarenotcaptured,whileinherentuncertaintydueto themodellingprocessalsoexists[36].Moreover,the typesoftravellerandtheiractivitiesduringtraveland theirresidentiallocationareunknown,eachofwhich contributestodifferingmalariainfectionrisks.Finally, overlandandshippingtravelflowsarenotconsidered here,whichalsocontributetolocal,regionalandglobal malariaconnectivityandflows. Thisworkformsthebasisforfutureanalyseson importedmalaria,eliminationfeasibilityandtherisks andpotentialroutesofartemisininresistancespread. Ratesandroutesofimportedmalariahavebeenshown tobesignificantlyrelatedtoacombinationofnumbers oftravellersto/fromendemicdestinationsandthe prevalenceofmalariathere[3].Thepotentialthusexists toconstructamodelbasedonglobalmalariaprevalence [40,41],thelocalspatialinteractionandaccessibilityto anairportwithinaregion[60],transmissionmodelsfor attackrateestimation[27],andtravellerflowdata[36], thatcanbeusedtoforecastimportedmalariarates,validatedwithimportedmalariadatareportedbyhealthfacilities/organizations. Asnationsmakeprogresstowardselimination[55], theimportanceofhumanmovementandimportedcases increases.Thisworkcontributestoanon-goinginitiative,thehumanmobilitymappingproject[61],aimedat bettermodellinghumananddiseasemobility,andwill formoneaspectofcontinuedmultimodalassessments ofmalariamovements[19,27,50,51]andassessmentof malariaeliminationstrategies[23,62].Finally,thepotentiallydisastrousconsequencesoftheriseandspreadof artemisininresistancerequiresthatdetailedandeffective planningbeimplementedinpreparationforcontaining andstemminganyspread[56].Abasicassessmenthere wereprovidedofprevalence-scaledtravelfromthefour SoutheastAsiancountrieswhereresistancehaspreviouslybeenobserved,butsignificantrefinementsofthese estimatesandmodellingmethodsshouldbeundertaken. Thesemayincludeimprovedtrackingandmappingof observedresistanceandhumanmovementpatternsin SoutheastAsia,asisbeingundertakenbytheTRAC project[63],aswellasscenariomodellingoftherisksof resistanceescapetoAfricaorLatinAmerica.Further, theincorporationofaccessibility[64,65]andtraveldata [51,59]withdrugusedata(eg,[66]),prevalenceinformation[40,41]andmodels[67],allundertakenwithina probabilisticmodellingframework(e.g.,[6,60]),could aidinestimationofspreadroutesshouldresistancearise elsewhere.AdditionalfilesAdditionalfile1: Thetraveltime/distancemasktoextractthe estimatedtypicalmaximumprevalenceof Plasmodiumfalciparum/ Plasmodiumvivax attheoriginoftravellers. Insetmap:traveltimeto thenearestmajorsettlement(populationsize>50,000).Theglobalmap ofaccessibilityisobtainedat[44]).Mainmap:eachdotshowsan airportlocationwitha50-kmbufferaroundit,andthecoloursshow theglobal P.falciparum prevalencemap[40]maskedbytheglobal traveltimemapwithathresholdvalueoflessthantwohours.These two-hourand50-kmthresholdswereusedtoassignprevalencevalues toairports(seemaintext). Additionalfile2: Airtravelnetworkcommunitiesweightedby directedestimatesofpassengerflow. Airportswiththesame communitymembership(indicatedbythesamecolour)displaystronger linksintermsoflikelymovementvolumebetweenthemthantoairports inothercommunities.ThemovementvolumeisextractedfromHuang et al ’ s[36]modelledpassengerflowmatrix. Additionalfile3: Communitiesforallpossibleconnections originatingfrom Plasmodiumfalciparum/Plasmodiumvivax -endemic areas.A) P.falciparum multilevelmembership; B) P.vivax multilevel membership.Thesetwomapsshowdirectlyconnected,one-transferand two-transferairportsfromendemicareas.TheinsetmapspresentcloseupviewsoftheUSAandwesternEurope.Airportswiththesame communitymembership(indicatedbythesamecolour)displaystronger linksintermsoflikelymovementsofinfectionsbetweenthemthanto airportsinothercommunities.Notethatinthe P.vivax map,two communitieswithlessthantenairportsarenotshown. Additionalfile4: Toptenairportsbasedonestimatedrelative malariaimportationandexportationrates. P.falciparum / P.vivax flow measuresarecalculatedbasedontheincomingandoutgoingnumbers ofpassengerstravellinginternationally,scaledbythemalariaprevalence attheoriginoftheroutesinthecaseofimportation,andattheairport listedinthecaseofexportation.Theflowsrepresentarelativemeasure ofinfectionmovementandarenotdesignedtorepresentactualnumber ofinfections. Additionalfile5: Spatialdistributionsofairportswith Plasmodium falciparum/Plasmodiumvivax betweennesscentralityscores.A) Airports withnormalizedbetweennessscores>0from P.falciparum -endemicareas, weightedbythe P.falciparum -prevalenceweightedpassengerflows. B) Airportswithnormalizedbetweennessscoresfrom P.vivax -endemicareas, weightedbythe P.vivax prevalence-weightedpassengerflows.Detailson thebetweennessmetricareprovidedinthemainmanuscript. Additionalfile6: Spatialdistributionsofairportnodesin eliminationcountries[55]weightedbyincominginternational Plasmodiumfalciparum/Plasmodiumvivax flows.A) Airportsin countrieswitheliminationobjectivesshownwithdotsizescaledto matchtotalincomingpassengerflowweightedbythe P.falciparum prevalenceatthetravellerorigins. B) Airportsincountrieswith eliminationobjectivesshownwithdotsizescaledtomatchtotal incomingpassengerflowweightedbythe P.vivax prevalenceatthe travellerorigins.Thesetwofigureshighlighttherelativerisksofinfection importationsthroughairtravelforeachcountrywithmalariaelimination objectives[55]. Additionalfile7: Topten Plasmodiumfalciparum routesand Plasmodiumvivax routestonon-AsiandestinationsfromtheGreat Mekongsubregion,asdefinedbymalaria-prevalencescaled passengernumbers. Valueswith(*)arereturnedastheadjustedlargest flowsbetweentwoconnectionflights.HuangandTatem MalariaJournal 2013, 12 :269 Page9of11 http://www.malariajournal.com/content/12/1/269

PAGE 10

Competinginterests Theauthorsdeclarethattheyhavenocompetinginterests. Authors ’ contributions ZHandAJTconceivedtheideaofthisanalysis.ZHandAJTdesignedand performedtheanalysis.ZHandAJTwrotethemanuscript.Bothauthorsread andapprovedthefinalmanuscript. Acknowledgements WethankDrPeterGethingandProfessorSimonHayforsharingthemalaria datasets.AJTissupportedbygrantsfromtheBillandMelindaGates Foundation(#49446,#1032350)andNIH/NIAID(#U19AI089674).AJTalso acknowledgesfundingsupportfromtheRAPIDDprogramoftheScience& TechnologyDirectorate,DepartmentofHomelandSecurity,andtheFogarty InternationalCenter,NationalInstitutesofHealth,USA.Thisworkformspartof theoutputoftheVector-bornediseaseairportimportationriskproject(www. vbd-air.com)andthehumanmobilitymappingproject(www.thummp.org). Authordetails1CenterforInfectiousDiseaseDynamics,PennsylvaniaStateUniversity, UniversityPark,PA,USA.2DepartmentofBiology,PennsylvaniaState University,PA,USA.3DepartmentofGeography,UniversityofFlorida, Gainesville,FL,USA.4EmergingPathogensInstitute,UniversityofFlorida, Gainesville,FL,USA.5DepartmentofGeographyandEnvironment,University ofSouthampton,Highfield,Southampton,UK.6FogartyInternationalCenter, NationalInstitutesofHealth,Bethesda,USA. Received:4February2013Accepted:24July2013 Published:2August2013 References1. IATAPressRelease. http://www.iata.org/pressroom/pr/Pages/2012-12-06-01. aspx 2.IATA: IATA2012AnnualReview. Beijing:IATA;2012. 3.TatemAJ,HuangZ,DasA,QiQ,RothJ,QiuY: Airtravelandvector-borne diseasemovement. Parasitology 2012, 139: 1816 – 1830. 4.WorldHealthOrganization: Theworldhealthreport2007-Asaferfuture: globalpublichealthsecurityinthe21stcentury. Geneva:WHOPress;2007. 5.KhanK,McNabbSJ,MemishZA,EckhardtR,HuW,KossowskyD,SearsJ, ArinoJ,JohanssonA,BarbeschiM,McCloskeyB,HenryB,CetronM, BrownsteinJS: Infectiousdiseasesurveillanceandmodellingacross geographicfrontiersandscientificspecialties. LancetInfectDis 2012, 12: 222 – 230. 6.HuangZ,DasA,QiuY,TatemAJ: Web-basedGIS:thevector-borne diseaseairlineimportationrisk(VBD-AIR)tool. IntJHealthGeogr 2012, 11: 33. 7.OdoliniS,ParolaP,Gkrania-KlotsasE,CaumesE,SchlagenhaufP, Lpez-VlezR,BurchardG-D,Santos-O ’ ConnorF,WeldL,vonSonnenburg F,FieldV,deVriesP,JenseniusM,LoutanL,CastelliF: Travel-related importedinfectionsinEurope,EuroTravNet2009. ClinMicrobiolInfect 2012, 18: 468 – 474. 8.CheckleyAM,SmithA,SmithV,BlazeM,BradleyD,ChiodiniPL,WhittyCJM: RiskfactorsformortalityfromimportedfalciparummalariaintheUnited Kingdomover20years:anobservationalstudy. BMJ 2012, 344: e2116 – e2116. 9.AsklingHH,BruneelF,BuchardG,CastelliF,ChiodiniPL,GrobuschMP, Lopez-VlezR,PaulM,PetersenE,PopescuC,RamharterM,SchlagenhaufP: ManagementofimportedmalariainEurope. MalarJ 2012, 11: 328. 10.AbanyieFA,ArguinPM,GutmanJ: Stateofmalariadiagnostictestingat clinicallaboratoriesintheUnitedStates,2010:anationwidesurvey. MalarJ 2011, 10: 340. 11.KainK,HarringtonM,TennysonS,KeystoneJ: Importedmalaria: prospectiveanalysisofproblemsindiagnosisandmanagement. Clin InfectDis 1998, 27: 142 – 149. 12.NillesEJ,ArguinPM: Importedmalaria:anupdate. AmJEmergMed 2012, 30: 972 – 980. 13.WidmerLL,BlankPR,VanHerckK,HatzC,SchlagenhaufP: Cost-effectivenessanalysisofmalariachemoprophylaxisfortravellersto West-Africa. BMCInfectDis 2010, 10: 279. 14.IsacsonM: Airportmalaria:areview.BullWorldHealthOrgan 1989, 67: 737. 15.ThangH,ElsasR,VeenstraJ: Airportmalaria:reportofacaseandabrief reviewoftheliterature. NethJMed 2002, 60: 441 – 443. 16.TatemAJ,RogersDJ,HaySI: EstimatingthemalariariskofAfrican mosquitomovementbyairtravel. MalarJ 2006, 5: 57. 17.LoupaC,TzanetouK: Autochthonous Plasmodiumvivax malariaina GreekschoolgirloftheAtticaregion. MalarJ 2012, 11: 52. 18.CohenJM,SmithDL,CotterC,WardA,YameyG,SabotOJ,MoonenB: Malariaresurgence:asystematicreviewandassessmentofitscauses. MalarJ 2012, 11: 122. 19.PindoliaDK,GarciaAJ,WesolowskiA,SmithDL,BuckeeCO,NoorAM,Snow RW,TatemAJ: Humanmovementdataformalariacontroland eliminationstrategicplanning. MalarJ 2012, 11: 205. 20.CentersforDiseaseControlandPrevention: Localtransmissionof Plasmodiumvivax malaria– Virginia,2002. MMWRMorbMortalWklyRep 2002, 51: 921 – 923. 21.CentersforDiseaseControlandPrevention: Localtransmissionof Plasmodiumvivax malaria– PalmBeachCounty,Florida,2003. MMWR MorbMortalWklyRep 2003, 52: 908 – 911. 22.DanisK,BakaA,LengletA,VanBortelW,TerzakiI,TseroniM,DetsisM, PapanikolaouE,BalaskaA,GewehrS,DougasG,SideroglouT, EconomopoulouA,VakalisN,TsiodrasS,BonovasS,KremastinouJ: Autochthonous Plasmodiumvivax malariainGreece,2011. EuroSurveill 2011, 16: 1 – 5. 23.ChiyakaC,TatemA,CohenJ,GethingP,JohnstonG,GoslingR, LaxminarayanR,HayS,SmithD: TheStabilityofMalariaElimination. Science 2013, 339: 909 – 910. 24.WijesunderaMDS: MalariaoutbreaksinnewfociinSriLanka. Parasitol Today 1988, 4: 147 – 50. 25.TatarskyA,AboobakarS,CohenJM,GopeeN,BheecarryA,MoonasarD, PhillipsAa,KahnJG,MoonenB,SmithDL,SabotO: Preventingthe reintroductionofmalariainMauritius:aprogrammaticandfinancialassessment. PLOSONE 2011, 6: e23832. 26.LepersJ,DeloronP,FontenilleD,CoulangesP: Reappearnceoffalciparum malariainCentralHighlandPlateauxofMadagascar. Lancet 1988, 331: 586. 27.LeMenachA,TatemAJ,CohenJM,HaySI,RandellH,PatilAP,SmithDL: Travelrisk,malariaimportationandmalariatransmissioninZanzibar. SciRep 2011, 1: 93. 28.FairhurstRM,NayyarGML,BremanJG,HallettR,VennerstromJL,DuongS, RingwaldP,WellemsTE,PloweCV,DondorpAM: Artemisinin-resistant malaria:researchchallenges,opportunities,andpublichealth implications. AmJTropMedHyg 2012, 87: 231 – 241. 29.AndersonTJC,NairS,NkhomaS,WilliamsJT,ImwongM,YiP,Socheat D,DasD,ChotivanichK,DayNPJ,WhiteNJ,DondorpAM: High heritabilityofmalariaparasiteclearancerateindicatesageneticbasis forartemisininresistanceinwesternCambodia. JInfectDis 2010, 201: 1326 – 1330. 30.PhyoAP,NkhomaS,StepniewskaK,AshleyE,NairS,McGreadyR,lerMooC, Al-SaaiS,DondorpAM,LwinKM,SinghasivanonP,DayNPJ,WhiteNJ, AndersonTJC,NostenF: Emergenceofartemisinin-resistantmalariaon thewesternborderofThailand:alongitudinalstudy. Lancet 2012, 379: 1960 – 1966. 31.DondorpAM,YeungS,WhiteL,NguonC,DayNPJ,SocheatD,vonSeidlein L: Artemisininresistance:currentstatusandscenariosforcontainment. NatureRevMicrobiol 2010, 8: 272 – 280. 32.TrapeJF: ThepublichealthimpactofchloroquineresistanceinAfrica. AmJTropMedHyg 2001, 64: 12 – 17. 33.MitaT,VenkatesanM,OhashiJ,CulletonR,TakahashiN,TsukaharaT, NdoungaM,DysoleyL,EndoH,HombhanjeF,FerreiraMU,PloweCV, TanabeK: Limitedgeographicaloriginandglobalspreadofsulfadoxineresistantdhpsallelesin Plasmodiumfalciparum populations. JInfectDis 2011, 204: 1980 – 1988. 34.NaidooI,RoperC: Followingthepathofmostresistance:dhpsK540E dispersalinAfrican Plasmodiumfalciparum TrendsParasitol 2010, 26: 447 – 456. 35.WorldHealthOrganization: Globalreportonantimalarialdrugefficacyand drugresistance:2000 – 2010. Geneva:WHOPress;2010. 36.HuangZ,WuX,GarciaAJ,FikTJ,TatemAJ: Anopen-accessmodeled passengerflowmatrixfortheglobalairnetworkin2010. PLoSOne2013, 8: e64317.HuangandTatem MalariaJournal 2013, 12 :269 Page10of11 http://www.malariajournal.com/content/12/1/269

PAGE 11

37. Anopen-accessmodeledpassengerflowmatrixfortheglobalairnetworkin 2010. http://www.vbd-air.com/data 38.O ’ KellyM: Thelocationofinteractinghubfacilities. TransportSci 1986, 20: 92 – 106. 39. MalariaAtlasProject. http://www.map.ox.ac.uk/ 40.GethingP,PatilA,SmithD: Anewworldmalariamap: Plasmodium falciparum endemicityin2010. MalarJ 2011, 10: 378. 41.GethingPW,ElyazarIRF,MoyesCL,SmithDL,BattleKE,GuerraCA,PatilAP, TatemAJ,HowesRE,MyersMF,GeorgeDB,HorbyP,WertheimHFL,Price RN,MellerI,BairdJK,HaySI: Alongneglectedworldmalariamap: Plasmodiumvivax endemicityin2010. PLoSNegl 2012, 6: e1814. 42.HowesRE,PatilAP,PielFB,NyangiriOa,KabariaCW,GethingPW, ZimmermanPa,BarnadasC,BeallCM,GebremedhinA,MnardD,Williams TN,WeatherallDJ,HaySI: TheglobaldistributionoftheDuffyblood group. NatCommun 2011, 2: 266. 43.GethingPW,VanBoeckelTP,SmithDL,GuerraCA,PatilAP,SnowRW,Hay SI: Modellingtheglobalconstraintsoftemperatureontransmissionof Plasmodiumfalciparum and P.vivax ParasitVectors 2011, 4: 92. 44. Traveltimetomajorcities:AglobalmapofAccessibility .http://bioval.jrc. ec.europa.eu/products/gam/index.htm 45.FreemanL: Asetofmeasuresofcentralitybasedonbetweenness. Sociometry 1977, 40: 35 – 41. 46.BarratA,BarthlemyM,Pastor-SatorrasR,VespignaniA: Thearchitectureof complexweightednetworks. ProcNatlAcadSci 2004, 101: 3747 – 3752. 47.GuimerR,MossaS,TurtschiA,AmaralL: Theworldwideairtransportation network:Anomalouscentrality,communitystructure,andcities ’ global roles. ProcNatlAcadSci 2005, 102: 7794 – 7799. 48.BrandesU: Afasteralgorithmforbetweennesscentrality. JMathSociol 2001, 25: 163 – 177. 49.FortunatoS: Communitydetectioningraphs. PhysRep 2010, 486:75 – 174. 50.TatemAJ,SmithDL: Internationalpopulationmovementsandregional Plasmodiumfalciparum malariaeliminationstrategies. ProcNatlAcadSci 2010, 107: 12222 – 12227. 51.WesolowskiA,EagleN,TatemAJ,SmithDL,NoorAM,SnowRW,Buckee CO: Quantifyingtheimpactofhumanmobilityonmalaria. Science 2012, 338: 267 – 270. 52.NewmanM: Analysisofweightednetworks. PhysRevE 2004, 70: 056131. 53.BlondelV,GuillaumeJ: Fastunfoldingofcommunitiesinlargenetworks. JStatMechTheorExp 2008, 2008: 1 – 12. 54.BauerD: Constructingconfidencesetsusingrankstatistics. JAmerStatist Assoc 1972, 67: 687 – 690. 55.UCSFGlobalHealthGroup: AtlasofMalaria-EliminatingCountries. San Francisco:GlobalHealthSciences;2011. 56.TalisunaA,KaremaC,OgutuB: Mitigatingthethreatofartemisinin resistanceinAfrica:improvementofdrug-resistancesurveillanceand responsesystems. LancetInfectDis 2012, 12: 888 – 896. 57.O ’ MearaWP,MangeniJN,SteketeeR,GreenwoodB: Changesinthe burdenofmalariainsub-SaharanAfrica. LancetInfectDis 2010, 10: 545 – 555. 58.BronsM,PelsE,NijkampP,RietveldP: Priceelasticitiesofdemandfor passengerairtravel:ameta-analysis. JAirTranspManag 2002, 8: 165 – 175. 59.TatemAJ,QiuY,SmithDL,SabotO,AliAS,MoonenB: Theuseofmobile phonedatafortheestimationofthetravelpatternsandimported Plasmodiumfalciparum ratesamongZanzibarresidents. MalarJ 2009, 8: 287. 60.JohanssonMA,Arana-VizcarrondoN,BiggerstaffBJ,GallagherN,MaranoN, StaplesJE: Assessingtheriskofinternationalspreadofyellowfevervirus: amathematicalanalysisofanurbanoutbreakinAsuncion,2008. AmJTropMedHyg 2012, 86: 349 – 358. 61. TheHumanMobilityMappingProject. www.thummp.org 62.HaySI,GuerraCA,TatemAJ,NoorAM,SnowRW: ReviewsTheglobal distributionandpopulationatriskofmalaria:past,present,andfuture.Lancet 2004, 4: 327 – 336. 63. TrackingResistancetoArtemisininCollaboration. http://www.wwarn.org/ partnerships/projects/trac 64.TatemAJ,HemelaarJ,GrayR,SalemiM: Spatialaccessibilityandthe spreadofHIV-1subtypesandrecombinantsinsub-SaharanAfrica. AIDS 2012, 26: 2351 – 2360. 65.LinardC,GilbertM,SnowRW,NoorAM,TatemAJ: Populationdistribution, settlementpatternsandaccessibilityacrossAfricain2010. PLoSOne 2012, 7: e31743. 66.CohenJ,WoolseyA,SabotO: Optimizinginvestmentsinmalaria treatmentanddiagnosis. Science 2012, 338: 612 – 614. 67.FleggJA,GuerinPJ,WhiteNJ,StepniewskaK: Standardizingthe measurementofparasiteclearanceinfalciparummalaria:theparasite clearanceestimator. MalarJ 2011, 10: 339.doi:10.1186/1475-2875-12-269 Citethisarticleas: HuangandTatem: Globalmalariaconnectivity throughairtravel. MalariaJournal 2013 12 :269. Submit your next manuscript to BioMed Central and take full advantage of: € Convenient online submission € Thorough peer review € No space constraints or color “gure charges € Immediate publication on acceptance € Inclusion in PubMed, CAS, Scopus and Google Scholar € Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit HuangandTatem MalariaJournal 2013, 12 :269 Page11of11 http://www.malariajournal.com/content/12/1/269