Quantifying spatiotemporal heterogeneity of MERS-CoV transmission in the Middle East region: a combined modelling approach

HAL Id: hal-01246336

https://hal.sorbonne-universite.fr/hal-01246336v2

Submitted on 4 Jan 2016

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Distributed under a Creative Commons Attribution - NonCommercial - NoDerivatives| 4.0

International License

Quantifying spatiotemporal heterogeneity of MERS-CoV

transmission in the Middle East region: a combined

modelling approach

Chiara Poletto, Vittoria Colizza, Pierre-Yves Boëlle

To cite this version:

Chiara Poletto, Vittoria Colizza, Pierre-Yves Boëlle. Quantifying spatiotemporal heterogeneity of

MERS-CoV transmission in the Middle East region: a combined modelling approach. Epidemics,

Elsevier, 2016, 15, pp.1-9. �10.1016/j.epidem.2015.12.001�. �hal-01246336v2�

ContentslistsavailableatScienceDirect

Epidemics

jo u r n al h om ep ag e :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

Quantifying

spatiotemporal

heterogeneity

of

MERS-CoV

transmission

in

the

Middle

East

region:

A

combined

modelling

approach

Chiara

Poletto

_,

_Vittoria

_Colizza

_,

_Pierre-Yves

_Boëlle

a_{SorbonneUniversités,UPMCUnivParis06,INSERM,InstitutPierreLouisd’épidémiologieetdeSantéPublique(IPLESPUMRS1136),F75012,}

27rueChaligny,Paris75012,France

b_{InstituteforScientificInterchangeFoundation,viaAlassio11/c,Torino10126,Italy}

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received9July2015

Receivedinrevisedform28October2015 Accepted9December2015

Availableonline17December2015 Keywords:

MiddleEastrespiratorysyndrome Spatiotemporalheterogeneity Outbreakanalysis

Multi-modetransmission

a

b

s

t

r

a

c

t

MERScoronaviruscasesnotifiedintheMiddleEastregionsincetheidentificationofthevirusin2012 havedisplayedvariationsintimeandacrossgeography.Throughacombinedmodellingapproach,we estimatetheratesofgenerationofcasesalongthezoonoticandhuman-to-humantransmissionroutes andassesstheirspatiotemporalheterogeneity.WeconsiderallcasesnotifiedtoWHOfromMarch2012to mid-September2014.WeuseastochasticmodellingofthetimeseriesofcaseincidenceintheMiddleEast regiontoestimatetime-andspace-dependentzoonoticandhuman-to-humantransmissionparameters. Themodelalsoaccountsforpossiblelackofidentificationofsecondarytransmissionsamongnotified cases.Thisapproachiscombinedwiththeanalysisofimportedcasesoutoftheregiontoassessthe rateofunderreportingofcases.Outofatotalof32possiblemodels,basedondifferentparameterisation andscenarioconsidered,thebest-fitmodelischaracterisedbyalargeheterogeneityintimeandacross spaceforbothzoonoticandhuman-to-humantransmission.Thevariationintimethatoccurredduring Spring2014ledtoa17-foldand3-foldincreaseinthetwotransmissions,respectively,bringingthe reproductiveratetovaluesabove1duringthatperiodforallregionsunderstudy.Themodelsuggests that75%ofMERS-CoVcasesaresecondarycases(human-to-humantransmission),whichissubstantially higherthanthe34%ofreportedcaseswithanepidemiologicallinktoanothercase.Overall,estimated reportingrateis0.26.Ourfindingsshowahigherlevelofspatialheterogeneityinzoonotictransmission comparedtohuman-to-human,highlightingthestrongenvironmentalcomponentoftheepidemic.Since sporadicintroductionsarepredictedtobeasmallproportionofnotifiedcasesandareresponsiblefor triggeringsecondarytransmissions,amorecomprehensiveunderstandingofzoonoticsourceandpath oftransmissioncouldbecriticaltolimittheepidemicspread.

©2015TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

TheMiddleEastrespiratorysyndromecoronavirus(MERS-CoV)

wasdetectedforthefirsttimein2012intheArabianPeninsulaand

sincethenithasbeenthesourceofglobalhealthconcern(WHO).

Thediseasecanbesevere,itischaracterisedbysporadicemergence

ofcasesandassociatedclustersinarestrictedgeographicalarea,the

MiddleEast,butfewcasesalsotravelledinternationallyandwere

diagnosedworldwide(ECDC).Morerecentlyitalsoledtoalarge

localoutbreakinSouthKorea,followingacaseimportationfrom

theMiddleEastregion(WHOandMERS-Cov).

∗ Correspondingauthor.Tel.:+33144738436. E-mailaddress:chiara.poletto@inserm.fr(C.Poletto).

Infection control outbreak investigations and observational

studieshavehighlightedimportantaspectsoftheepidemiology

ofthedisease.Bothzoonoticintroductionsandhuman-to-human

transmissions are responsible for new infections. Dromedary

camelshavebeenfoundtobeintermediaryhostsforthevirusand

mayrepresentthesourceforzoonoticinfection(Mulleretal.,2015;

Alagailietal.,2014).Human-to-humantransmissioncanoccur

fol-lowing closeand prolongedcontactsand it hasbeenidentified

asthemain mechanismresponsiblefor thedocumented

hospi-taloutbreaksofAl-Hasa(Assirietal.,2013)andJeddah(Drosten

etal.,2015).Modellingstudieshighlightedthesubcriticalnature

oftheepidemic(Chowelletal.,2014;Polettoetal.,2014;Breban

etal.,2013;Cauchemezetal.,2014;Majumderetal.,2014)and

fewofthemalsocharacterisedtherateofzoonotictransmission

(Polettoetal.,2014;Cauchemezetal.,2014).Apopulation-based

serologicalinvestigationrecentlyprovidedevidenceforsubstantial

http://dx.doi.org/10.1016/j.epidem.2015.12.001

1755-4365/©2015TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/ 4.0/).

2 C.Polettoetal./Epidemics15(2016)1–9

under-detectionofcasespossiblyduetosub-clinicalformsofthe

disease(Mulleretal.,2015).

Despitetheseadvances,therearestillmanygapsinknowledge

regardingMERSepidemiologyandecology.Geographicalvariation

intheepidemichasrarelybeenconsideredinthesestudies,even

thoughepidemicdatashowa clearspatialheterogeneityacross

countriesintheMiddleEastregionandacrossprovincesofSaudi

Arabia.Occasionalsparksinactivityhavebeenobservedaswell.

Aparticularlylargeincreaseinthenumberofcaseswasreported

duringSpring2014,withamaximumof133weeklycasesatthe

endofApril(epidemiologicalweek2014-17),confinedmainlyin

theprovincesofRiyadhandMakkahinSaudiArabia(WHO).

Theobservedstrongspatialandtemporalheterogeneities

cha-racterising a persistent yet subcritical epidemic pose a set of

challengesforthecomprehensiveunderstandingofMERS-CoV

cir-culationintheMiddleEastregion.Whetherthisbehaviourresults

fromacombinationofdifferenttransmissionmodesorseasonal

effectsorothermechanismsisstillunclear.Hereweaimtoquantify

the spreading potential associated to zoonotic and

human-to-humantransmissionroutesandcharacterisetheirtemporal and

geographicalvariation,touncoverpossiblevariationsinthe

epi-demiologyofthedisease.Weproposeacombinedapproach,akin

totheoneintroducedin(Polettoetal.,2014),basedonthejoint

analysisofimportedcasesoutoftheMiddleEastregionand

inci-denceofcaseswithintheregion.Bymaximisingtheuseofavailable

informationthemodelisabletoestimatethetransmission

param-etersforeachrouteandtheirspatiotemporalvariation,accountfor

unidentifiedhuman-to-humantransmissionamongdetectedcase,

andassesstherateofcasedetection.

2. Methods

2.1. Data

DataconsistedofallcasesnotifiedtoWHOwithonsetbetween

thebeginningoftheepidemicinMarch2012(specificallyweek

2012-12)andmid-September2014(i.e.week2014-38),retrieved

asofMarch2015from(Rambaut,2013).Casesarisinginthe

Mid-dleEastregion(SaudiArabia,Qatar,Oman,Kuwait,Jordan,United

ArabEmirates,andYemen)andcasesimportedtoWestern

Euro-peancountriesandNorthAmericawereanalysedseparately.The

datasetusedfortheanalysisdidnotinclude113casesoccurring

betweenMay2013andMay2014identifiedthroughretrospective

reviewandreportedbyWHOonJune26(MERS-Cov)forwhich

information of neitherdate of onset or date of hospitalisation

wasavailable.Completenessofthedatasetusedinboth

tempo-ralandgeographicalinformationisaddressedinSection1ofthe

SupportingInformation.

IntheMiddleEastregion,casesweregroupedaccordingto

loca-tionattheprovincelevelinSaudiArabia(11overthe13provinces

experiencecasesandwereconsideredintheanalysis)andatthe

countrylevelotherwise,finallyyielding17regions(11provinces

inSaudiArabia,plustheremaining6countriesoftheMiddleEast

region,i.e.Qatar,Oman,Kuwait,Jordan,UnitedArabEmirates,and

Yemen).Foreachregionwecomputedtheepidemiccurveusing

datesofonset.Asthisdatewasmissinginapproximatelyhalfthe

cases,imputationwasusedbasedonhospitalisationornotification

datesasexplainedinthenextsubsection.

Aproportionof34%ofcasesreportedtoWHOweredescribedas

“secondary”whentheywereepidemiologicallylinkedtoanother

case.Accordingly,wereferinthefollowingtocasesarisingfrom

human-to-humantransmissionas“secondarycases”andtoall

oth-ersas“primarycases”,i.e.ofzoonoticorigin.Lackofinformation

ontheidentificationofsecondarytransmissionisalsoconsidered

inthemodel.

WefinallyconsideredtheimportedcasesinWesternEurope

andNorthAmerica(n=9),astheseregionshavesetuphigh

sen-sitivitysurveillancesystemstodetectimportations.Weincluded

Austria, Belgium, Denmark, Finland, France, Germany, Iceland,

Ireland,Italy,Liechtenstein,Luxemburg,theNetherlands,Norway,

Portugal,Spain,Sweden,Switzerland,theUnitedKingdom,Greece,

CanadaandtheUnitedStatesaspossibledestinationsoutofthe

MiddleEast.Inthisarea,1casewithtravelhistorytoMiddleEast

wasnotifiedinFrance,Germany,Greece,ItalyandtheUnited

King-domand2casesintheNetherlandsandtheUnitedStatesoverthe

periodunderstudy.

2.2. Reconstructionofincidencetime-series

Incidenceofonset time-seriesinthe17 MiddleEastregions

werereconstructedasfollows.Ifhospitalisationdatethwas

avail-able, the onset date was imputed at th−th where th was

sampledfromthedistributionoftimefromonsetto

hospitalisa-tion,determinedfromothercaseshospitalizedinthesameperiod

(i.e.[th−30days,th+30days]);ifonlythenotificationdatetrwas

known,theonset date wasimputedat tr−tr where tr was

sampledfromthedistributionoftimefromonsettonotification,

determinedfromothercasesnotifiedinthesameperiod,asbefore.

Cumulativedistributions fromonsettohospitalisationandfrom

onsettonotificationarereportedintheSupplementary

Informa-tion.Thisprocedurewasrepeatedtoyield 20epidemiccurves:

therewasanaverage707cases(range704–712casesdependingon

imputation)withonsetintheperiodconsideredforanalysis.The

overallepidemiccurveprofilewaslittleaffectedbyimputation(see

SupplementaryInformation).Overall,thefiveregionsreportingthe

mostcaseswereMakkah(245cases),Riyadh(223cases),Eastern

Province(68),UnitedArabiaEmirates(65)andAlMedinah(42).

2.3. Model

Weusedatwo-stepapproachtoestimatethereportingrate,

therateofsporadicgenerationofcasesfromnon-humansource

pr

sp(t)andthereproductionratioRr(t)asafunctionoftime tin

eachregionr.Thefirststepincludedmodellingtheincidencetime

seriesintheMiddleEastregion,thesecondstepusedimportation

ofcasesandastochasticdata-drivenmodelforspatialdiffusionof

theepidemicworldwide.Aschematicrepresentationoftheanalysis

isreportedinFig.1.Allparametersandvariables arelistedand

describedinTable1.

Table1

Variablesandcorrespondingdescriptions.

Variable Description

nr_{(t) Totalnumberofcasesinregionrattimet} Dr_{(t) Numberofnotifiedcasesinregionrattimet}  Reportingrate,Dr_(t)=nr_(t)

Nr Populationofregionr dr

s(t) Numberofsecondarytransmissionsamongallnotifiedcases Dr_{(t)inregionrattimet}

sr_{(t) Numberofnotifiedsecondarycasesinregionrattimet} pr

sp(t) Rateofgenerationofsporadiccasesinregionrattimet qr

sp(t) Rateofgenerationofsporadiccasesinregionrattimet obtainedfromnotifiedcases,qr

sp(t)=prsp(t)

Rr_{(t) Reproductionratioforhuman-to-humantransmissionin} regionrattimet

˛r Geographicalvariationoftherateofgenerationofsporadic casesinregionr,pr

sp(t)=˛rpsp(t)

ˇr Geographicalvariationofthereproductionratiofor human-to-humantransmissioninregionr,Rr_(t)=ˇ

rR(t) qsp,1,qsp,2 Baselinelevelandpeaklevelofthesplinefunctionassumedto

modelqr sp(t)

Fig.1. ModelA:Schemeofthe2-stepapproach.Step1isbasedonthefitof20imputedepidemiccurvesforeachofthe17regionsintheMiddleEast(onlyonecurveisdisplayed forthesakeofvisualisationcorrespondingtotheregionexperiencingthelargestnumberofcases).ItallowsmodelselectionandestimationofRr_(t)andqr

sp(t)=prsp(t).Step 2isbasedonthefitofimportedcasesinWesternEuropeandNorthAmericaandallowsestimating.B:SchemeofthefunctionalformsassumedforR(t)andqsp(t)when temporalheterogeneity(eitherforoneoftheparametersorboth)isconsideredinthemodel.Parametersqsp,1,qsp,2,R1,R2,R3areestimated.C:Combinationofparameters yieldingthe32modelsforexploration.

Firststep:Casesweregroupedbyweek,correspondingtothe mean generationtime (Assiriet al.,2013; Poletto etal., 2014).

In each regionr, incidence was modelled as thesuperposition

of sporadiccases at a region-specific time-varyingweekly rate

pr

sp(t)andofcasescausedbyhuman-to-humantransmissionwith

region-specifictime-varyingreproduction ratioRr_(t).More

pre-cisely,weassumedpr

sp(t)=˛rpsp(t)andRr(t)=ˇrR(t),where˛rand

ˇrareregionspecificrescalingfactorsofsporadicgenerationof

casesandreproductiveratio,respectively.Wethenexpressedthe

expectedvalueoftotalnumberofcasesnr_{(t)inregionrattime}

t asE(nr_(t))=˛

rNrpsp(t)+ˇrR(t−1)nr(t−1)where Nris the

pop-ulationoftheregion.Asasensitivityanalysis,wealsoexplored

a model where thenumber ofcases dependedonincidence in

thethreeweeksbefore(i.e.generationtimedistributionspanning

3weeks).

Weassumedaconstantanduniformreportingrate,sothat

the number of notified cases Dr_{(t) in region r at time t was}

Dr_(t)=nr_{(t). We further assumed that D}r_{(t) was Poisson}

dis-tributed.Theobservedrateofgenerationofsporadiccasesbased

onnotificationsisthendefinedasqr

sp(t)=prsp(t).

AsindicatedintheDatasection,anumbersr_{(t)ofcasesamong}

theDr_{(t),wereidentifiedassecondary.We definedtwo}

scenar-iosforinterpretingthisinformation.Inthecompleteinformationon

transmissionsscenario,sr_{(t)wasassumedtobeequaltotheactual}

number of secondary transmissions among notifiedcases dr

s(t),

sr_(t)=dr

s(t);inthepartialinformationontransmissionsscenario,

sr_{(t)wastakenasalowerboundofthisnumber:wethusassumed}

that somecasesnot described as secondary couldnevertheless

haveresultedfromunidentifiedhuman-to-humantransmission,

i.e.ds(t)≥sr(t).

Inthecompleteinformationontransmissionsscenario,we

com-putedineachregiontheprobabilityofhavingDr_{(t)notifiedcases,}

among which sr_{(t) human-to-human transmissions: P(d}r

sp(t)=

Dr_(t)−sr_(t)&d

s(t)=sr(t))wheredsp(t)arethenotifiedcasesarising

fromsporadicgeneration.ConditionalonDr_{(t),assumingthesame}

reportingrateforsecondaryandprimarycases,thedistribution

ofdr

s(t)wasbinomialwithprobabilityˇrR(t−1)Dr(t−1)/E(Dr(t)),

sothat P(drsp(t)=Dr(t)−sr(t)&ds(t)=sr(t)) =e−E(Dr(t))(ˇrR(t−1)D r_(t−1))sr_(t) sr_(t)! (Nr_˛ rqsp(t))D r_(t)−sr_(t) (Dr_(t)−sr_(t))! .

Forthepartialinformationontransmissionsscenario,we

com-puted P(drs(t)=D(t)−sr(t)&dsp(t)≥sr(t)) =e−E(Dr(t))(ˇrR(t−1)D r_(t−1))sr_(t) sr_(t)! ×





withtheprobabilitythatacasegeneratedbyhuman-to-human

transmissionwascorrectlydescribedassecondary.Inbothcases,

thefinallog-likelihoodwasobtainedasasumoverallweeksand

regions.

Therescalingfactors˛randˇr,assumedtomodelthepresence

ofspatialheterogeneityinthezoonoticgenerationofcasesand

secondary transmissions, respectively, are described asrandom

parameterslog-normallydistributed.Theabsenceofgeographical

variationwasalsoconsidered,wheretheseparametersweresetto

1.

For temporal changes, we modelledqsp(t)as a linear spline

andR(t)asastepwisefunction(seeFig.1B)withnodes/jumpsat

weeks2012-12,2014-13,2014-17,2014-21,2014-38tocapture

thelargestepidemicwaveoccurringduringSpring2014.qsp(t)was

thereforedescribedby2parameters(namelyqsp,1andqsp,2)and

R(t)by3parameters(R1,R2,R3).Intheabsenceoftemporal

vari-ation,thesequantitieswereconstantintime,i.e.qsp(t)=qsp,1and

4 C.Polettoetal./Epidemics15(2016)1–9

Consideringsystematicallyallparameterisations(possible

pres-enceoftemporaland/orgeographicalvariationinpsp(t)andR(t))

andscenarios(completeandpartialinformationontransmissions)

yieldedatotalof32modelformulations(Fig.1C),andeachmodel

wasfittedtothe20imputedepidemiccurves.We exploredthe

posterior distribution of all parameters of interest using Gibbs

sampling(100,000iterations,2chains).Priordistributionsforall

parametersarereportedintheSupplementaryInformation.We

usedDeviance InformationCriterion (DIC) toidentify the

best-fitting model, averaging DIC differences over the 20 epidemic

curves,and reportedposteriormeansand credibleintervals for

parameters.

Forsensitivity analysis,we shiftedthe peakof psp(t)by one

week,allowedfora10-week-longchangeinR(t)andpsp(t),fora

singlestepincreaseinR(t)ofvaryingduration,andfordispersion

inthegenerationtimeduration(seeSupplementaryInformation

fordetails).

Secondstep:Similarlyto(Polettoetal.,2014),weusedaglobal

epidemicandmobilitymodel(GLEAM)(Balcanetal.,2009a,b)tofit

thenumberofimportedcasesincountriesoutoftheMiddleEast

(Fig.1b).

GLEAM is based on a spatially structured meta-population

approach comprising 3362 subpopulations in 220 countries in

theworld coupledthrough mobility connections. The model is

informedwithhigh-resolutiondemographicdatafor6billion

indi-vidualsandmulti-scalemobilitydataincludingthefullairtraffic

databasefromtheInternationalAirTransportAssociation(IATA)

(IATA)and short-range groundmobility obtainedfromnational

commutingdata(Balcanetal.,2009b).WemodelledMERS-CoV

infectiondynamics withineach subpopulationas the

combina-tionofgenerationofcasesfromsporadicinfections,assumedtobe

aPoissonprocess,andaSEIRcompartmentalization(susceptible,

exposed,infectious,recoveredindividuals)withgenerationtimeof

7.6daysandlatencyperiodof5.2days(Assirietal.,2013;Poletto

etal.,2014).Transmissibilityisparameterisedbysettingqr

sp(t)and

Rr_{(t),totheestimatedvaluesfromthebest-fitmodelselectedin}

step1.

The detection rate  was thefree parameter of the model.

Foreachvalueof,GLEAMallowedthegenerationofstochastic

numericalrealizationsoftheMERS-CoVoutbreaksimulatingthe

localepidemicinthesourceregionandinternational

dissemina-tioneventsthroughmobilityprocesses.Wethusgeneratedwith

aMonteCarloproceduretheprobabilitydistributionPi(ni)ofthe

numberniofimportedMERS-CoVcasesincountryioutoftheseed

region.Beingallindependentimportationevents,wecandefinea

likelihoodfunctionLimp













empiricallyobservednumberofimportedcasespercountry.

3. Results

3.1. Modelselection

Altogether,modelsinthepartialinformationontransmissions

scenarioprovidedamuchbetterfitthantheonesofthecomplete

information on transmissions case, suggesting that more

trans-missions occurred than actually reported (see Supplementary

InformationforallDICvalues).

Table2summarisestheDICdifferencesforthe16models

cor-respondingtothepartialinformationontransmissionsscenario.The

bestmodelincludedheterogeneityintimeandspaceforboththe

sporadicgenerationofcasesandthereproductiveratio.TheDIC

differencesshowthatalltheothermodelsfitsubstantiallyworse.

Asexpected,themodelswithnotemporalheterogeneityineither

pr

sp(t)orRr(t)hadthelargestDICshowingthatthehypothesisof

Table2

DICdifferencesforthe16modelstestedinthescenariopartialinformationon transmissions.Differenceswereaveragedover20imputedepidemics.

Temporalvariation Geographicalvariation

None R psp BothpspandR

None 329 295 136 121

psp 308 279 54 45

R 249 219 51 41

BothpspandR 232 206 14 0

constanttransmissionparametersfailedtoproperlyexplainthe observedincidenceprofile.Assumingtemporalvariationineither transmissionmodealreadyimprovedthefit,withatimevarying reproductionratio providing thelargestimprovement. For spa-tialheterogeneity,allowingthezoonotictransmissiontochange betweenregionswasalwaysfavoured.Themodelprovidingthe closestDICtothebest-fitoneincludedtemporalchangeintherate ofsporadiccasesandinRr_{(t),withspatialheterogeneityonlyfor}

thezoonoticcomponent.

Thisconclusionwasrobusttochangesinthemodels(see sensi-tivityanalysisintheSupplementaryInformation).

3.2. Spatialandtemporalheterogeneityinthetransmission

modes

Table3showstheestimatedparametersofthefunctionsR(t)

andqsp(t),alongwiththereportingrate,estimatedforthe

best-fitmodel.Thebivariateposteriordistributionsoftheparameters

qsp,1,qsp,2R1,R2,R3areshownintheSupplementaryInformation.

Wepredictedequalto0.26[0.16–0.52]correspondingtoan

aver-ageunderreportingratioof3.8.Foreachregiontherateofsporadic

generationofcasesandthereproductiveratiocanbecomputedby

multiplyingthetemporalrescalingfunctionsbythefactors−1˛r

andˇr,respectively(estimatesfor˛randˇrareprovidedinthe

SupplementaryInformation).ThebarchartofFig.2presentsan

overviewoftheresultsbyfocusingontheminimumandthe

max-imumvaluesoftheparametersintime foreachregion,namely

pr

sp,1=−1˛rqsp,1 and prsp,2=−1˛rqsp,2 fortherateofzoonotic

transmission,and Rr

1=ˇrR1 and R3r=ˇrR3 forthereproductive

ratio.DuringtheepidemicpeakofSpring2014thegenerationof

sporadiccaseswasestimatedtoincreaseofa factor17.4. More

indetail,fortheperiodoflowepidemicactivitywepredicteda

weeklynumberofsporadiccasesonthewholeregion(including

undetectedcases)equalto6.22[0.56–43.1]thatincreasesupto

amaximumof108.0[8.0–734.0]duringSpring2014.Themean

reproductiveratioestimated for theperiod oflow activity was

belowone inall regions,rangingfrom0.31 [0.06–0.60] to0.70

[0.50–0.92].Duringthefourweeksbetween2014-13and

2014-17,itwaspredictedtoincreasebyafactor3.3,raisingaboveone

forallregions(Fig.2).Itwasfoundtodecreasetovaluesbelowthe

epidemicthresholdandslightlyhigherthanthebasevaluesduring

thesecondpartoftheSpringwave(from2014-17to2014-21).

Zoonotictransmissionischaracterisedbylargergeographical

variationswithrespecttohuman-to-humantransmission,as

con-firmed by the coefficient of variations associated to˛r and ˇr

Table3

Parameterestimatesobtainedfromthebestfit.

Parameterestimate qsp,1 0.026[0.011–0.056]×10−6 qsp,2 0.45[0.15–1.07]×10−6 R1 0.40[0.22–0.57] R2 1.32[0.72–2.0] R3 0.60[0.32–0.85]  0.26[0.16–0.52]

p sp [x10 ]-6 Eastern Province Tabuk Al-Jawf Al-Qassim Nejran 'Aseer UAE Yemen Oman Jordan Al Bahah Riyadh Makkah Al Madinah Northern Borders Region Kuwait Qatar R p sp,1 Rr1 r p sp,2 Rr3 r

Fig.2.RateofzoonoticcaseintroductionandreproductiveratioinMiddleEast regions.Baselinevalue(minimum,lightcolour)andSpring2014value(maximum, darkcolour)areshowninbluefortherateofzoonotictransmissionineachregion, pr

sp,1=−1˛rqsp,1andprsp,2=−1˛rqsp,2,andinredforthereproductiveratioRr1= ˇrR1andRr3=ˇrR3.ThedashedredlineindicatesthethresholdvalueR=1.

equalto0.99and0.20respectively.Thelargestreproductiveratios wereobtained for theEastern Region (2.3[1.4–3.7] duringthe maximumepidemicactivity),Makkah(1.8[1.3–2.4]),AlMedinah (1.7[1.0–2.6])andRyiadh(1.6[1.1–2.2]),whilethesmallest val-uesof theparameterwere obtainedforAl Jawf(1.0 [0.2–2.1]), Oman (1.0 [0.2, 2.1]). The probability of zoonotic transmission was thehighest in the provincesof Riyadh(pr

sp,2 equal to 9.5

[3.5–18.3]×10–6_{),EasternRegion(4.7[2.0–8.4]×10}–6_)andQatar

(4.1[1.3–8.5]×10–6_{),and thesmallestfor AlBahah(0.34[0.06,}

1.15]×10–6_{)andYemen(0.05[0.01,0.14]×10}–6_).

Resultsof the sensitivityanalysis showedthat theposterior distributionsof theparameters werelittleaffectedby arbitrary and simplifying modelling choices (date of the peak, width of the increase, span of the generation time distribution – see

SupplementaryInformation).

3.3. Assessmentoftheepidemicsituation

Fig.3showsthepredictednumberofcasesaggregatedoverthe

wholeperiodbrokendownbytypeoftransmission,obtainedforthe

best-fitmodel.Resultsshowalargerfractionofsecondarycasesout

ofthetotalnumberthanreported.InWHOreports,secondarycases

amountto34%ofallreportedcases,withregionallevelsranging

between0%and57%.Accordingtomodelpredictions,this

propor-tionwas75%,withvaluesinthemostaffectedregionsequalto89%

inMakkah(vs.thereported28%),76%inEasternRegion(vs.50%),

75%inAlMedinah(vs.36%),74%inUAE(vs.54%)and69%inRiyadh

(vs.26%).Boththetotalnumberofcasesandtheratioofsecondary

toprimarycasesweregeographicallyheterogeneous.

Fig.4 shows predictedtime series ofincidence and

propor-tionofhuman-to-humantransmissioncasesfortheMiddle East

regionasawholeandforthefiveregionswiththemostintense

epidemicactivity.Periodswiththehighestlevelsofincidencewere

characterisedbyanincreaseintheproportionofsecondarycases

above50%,indicatinganincreaseinhuman-to-human

transmis-sionabovetheepidemicthreshold.InparticularSpring2014saw

anoverallincreaseintheproportionofhuman-to-human

trans-missioncasesfrom52%to74%.Attheregionalleveltheincidence

patternswerediverse.Peaksinproportionofsecondarycaseswith

valuesabove95%werereachedinMakkah(duringtheSpring2014

wave)AlMedinah(inSeptember2013andMay2014)andEastern

Province.TheprovinceofRiyadh,despitealargeincidence,never

sawaproportionoftransmissionabove80%.

4. Discussion

Wepresentedhereanassessmentoftherelativeroleof

human-to-humanvs.zoonotictransmissiononthedynamicsofMERS-CoV

in theMiddle Eastpeninsulaaccounting fortemporal and

geo-graphicalvariability.Wedesignedacombinedmodellingapproach

in order tomake optimal useof theavailable information and

provideacomprehensivedescriptionoftheepidemic.The

com-parison among a wide set ofmodels highlighted thefollowing

characteristicsoftheMERS-CoVepidemiology.

First,manyhuman-to-humantransmissioneventslikelywent

unidentified, as shown bythe betterfit of the setof scenarios

considering partial information ontransmissions compared with

theonesassumingcompleteinformationontransmissions.Indeed,

human-to-humantransmissionwasreportedin34%ofthecases,

butthebest-fitmodelpredictedthatthisproportionwasupto75%.

Notaccountingfortheimperfectinformationonsecondarycases

ledtounder-estimatinghuman-to-humantransmission.

Interest-ingly, this larger proportion of human-to-human transmission

predictedbythebest-fitmodelisconsistentwiththefactthatonly

few contactsbetweenknown sourcesof zoonotic transmission,

suchasdromedarycamels,andhumanshavebeenreportedamong

MERS-CoVpatients(Gossneretal.,2014)andoftennosuchcontacts

werefoundforinfectionsreportedasprimary.Ourresultsshow

thatalargepartofhuman-to-humantransmissionmayhavewent

unidentifiedinthedata,aspreviouslysuggested(Drostenetal.,

2015).Theproportionofunidentifiedhuman-to-human

transmis-sionwasthelargestwhenepidemicactivityincreased,whichcould

beexplainedbythemoredifficultassessmentoftheoriginofcases

inpresenceofalargenumberofcases.Indeed,whileinthereported

datatheproportionofsecondarycasesdidnotchangesubstantially

duringtheSpring2014periodcomparedtotherestofthetime(34%

vs.33%),ourmodelshowedthatthispercentageincreasedfrom

52%to74%inSpring2014.Thissharpincreasehaspreviouslybeen

describedinanobservationalstudyandattributedmainlyto

noso-comialtransmissions(Drostenetal.,2015).Thepredictedvaluein

thelowactivityperiod(52%)issimilartothevaluereportedfor

theperiodbetweenSeptember2012andOctober2013(59%)(The

WHOMERS-CoVResearchGroup,2013).

Our analysis provided further insight in the Spring 2014

increase.Wefoundthatthisepidemicwaverequireda

substan-tial increase in both the rate of introduction of the virus and

human-to-humantransmission. We obtainedindeeda riseof a

factor17intherateofsporadicintroductions,andathreetimes

increaseinthereproductiveratiogoingfromvaluesbelowthe

epi-demicthreshold(inaccordancewithpreviousstudies(Polettoetal.,

2014;Brebanetal.,2013;Cauchemezetal.,2014))tovaluesabove

suchthresholdforalltheregions.Morefrequentsporadic

intro-ductionscouldberelatedtoanincrease intheviruscirculation

inthezoonoticsource.Arecentstudy(Werneryetal.,2015)on

MERS-CoVspreadamongdromedarycamelsshowsthatthevirus

producesacuteepidemicsincalves,oftenborninSpring.Such

out-breaksmaycauseanincreaseinthenumberofprimarycasesand

6 C.Polettoetal./Epidemics15(2016)1–9

Fig.3.Casesofzoonoticoriginandfromhuman-to-humantransmissionineachregionduringthewholeperiodunderstudy.Foreachregion,theareasoftheredand yellowportionsofthecirclesareproportionaltothenumberofhuman-to-humantransmissionsandzoonoticcasesrespectively.Theactualcountsareshownforthefive regionsexperiencingthemostcases.

thenumberofadmissionsofMERS-CoVcasestohospitalswiththe

possibilityoffurthertriggeringhospitaloutbreaksaspreviously

reported(Drostenetal.,2015;Saadetal.,2014;Obohoetal.,2015).

Despitea suggestionthat thehighlyseasonalbreedingcycle of

camelsmaydriveMERS-CoVinfectionsinhumans(Werneryetal.,

2015),wedidnotfindevidencetoexplainaseasonalpatternon

human-to-humantransmission.We testedseasonalpatternsfor

qsp(t)andR(t)toexamineiftherehadbeenariseinthenumberof

casesduringSpring2013.Overall,themodelfitswerenotimproved

(seeSupplementaryInformation).However,thisdoesnotruleout

thatsuchaseasonalpatternmaybecomemoreapparentasthe

numberofcasesincreases.Sensitivityanalysisonfivealternative

modelsshowedthatposteriordistributionofparameterswerelittle

affectedbythemodeldetails.

Modelselectionshowsthat addinggeographical

heterogene-ityin R(t) doesnotimprove substantially thefit, meaning that

geographical variation in transmission settings and

human-to-human contact behaviour does not play a critical role in the

MERS-CoVspreadingdynamicswithintheArabianPeninsula.On

theotherhand,zoonotictransmissionpresentedahighest level

ofheterogeneity.Thisresultpointsoutthestrongenvironmental

component of the infection. Such variations indeed should be

relatedtoaspatiallyheterogeneousforceofinfectioninducedby

thezoonoticsource.Recentstudiesprovidedinformationonspatial

densityofcamelsovertheArabianPeninsula(Gossneretal.,2014)

and the prevalence of MERS-CoV infection among the animals

(Alagailietal.,2014).Besidesthesetwoingredients,however,the

forceof infectionfromcamelstohumanis alsodeterminedby

human-to-camelcontactbehaviour.Informationonfarming

prac-ticesandtheirgeographicalvariationwouldbecriticaltoexplain

theobservedpatternandtoinformriskassessmentanalysis.

Thefiveregionswiththehighestepidemicactivitypresented

differentbehaviours.ThetwolargestoutbreaksoccurredinMakkah

andRiyadh.Makkahshowedthemostintensehuman-to-human

transmission,whiletheepidemicinRiyadhwascharacterisedby

alargenumberofvirusintroductions.Thesefindingsare

consis-tentwiththephylogeneticanalysisconductedby(Drostenetal.,

2015).Furthermoreinaccordancewiththestudyin (Majumder

etal.,2014)werecoveredahigherreproductiveratioinMakkah

thaninRiyadh(Rr3equalto1.9against1.6).Theepidemicactivityin

bothAlMedinahandUAEshowshightransmissionandlow

intro-ductions.EventuallytheincidenceprofileintheEasternProvince

ismorescatteredalloverthestudyperiod(Assirietal.,2013).

Anotherimportantfindingis theconfirmation that

underre-portingintheregioniscommon,aspreviouslyreportedinother

modellingstudies(Polettoetal.,2014;Cauchemezetal.,2014),

consistently with surveillance works (Saad et al., 2014; Oboho

Fig.4. PredictedtimeseriesofincidenceandproportionofsecondarycasesforthewholeMiddleEastandthefiveregionswiththemostcases.Foreachregionthe averagenumberofcasesisdisplayedwiththegrey-scalecolour-map,whilevaluesfortheproportionofsecondarycasesarecolour-coded(white,yelloworangeandred) accordingtothelegend.Averagenumberofcumulativecasespredictedbythemodelisindicatedonthetopofeachcolormap.Dashedlinesindicatethetimeperiodsfor increasedactivityassumedbythemodel.

andarecentnationwideseroprevalenceinvestigationinSaudi

Ara-bia(Mulleretal.,2015).Here,wefoundthatthereportedcases

amountedtobetween16%and52%ofallcases,sothatthetotal

epi-demicsizeintheMiddleEastregionasofnowcouldrangebetween

∼2500and∼8000cases.Casescouldgoundetectedbecausethey

aresubclinical(Mulleretal.,2015;Saadetal.,2014;Obohoetal.,

2015;Memishetal.,2013;Al-TawfiqandMemish,2014).Assuming

mildcaseslesstransmissiblethensymptomaticoneswouldhave

ledtoalargerestimateforthenumberofcases,anaspectthatwas

notconsideredinourmodelbecauseoflackofinformation.Current

medicalknowledgeofthediseasedoesnotallowamoredetailed

modellingoftheinfectionnaturalhistoryandfurther

epidemiolog-icalinvestigationisneededtoassesssuchlevelofheterogeneityin

transmission.

Our modelling approach introduces a seamless transition

between analysing cluster of cases and epidemic curves. Most

publishedanalyses of theMERS-CoVepidemics focusedon the

characteristicsof case cluster sizestoinformhuman-to-human

transmission(Polettoetal.,2014;Brebanetal.,2013;Cauchemez

et al., 2014), and would not be adequate toanalyse sustained

human-to-humantransmission.Amorerecentanalysisused

tra-jectorymatchingbasedonadeterministicSEIRmodel(Chowell

etal.,2014)thatmayresultinlargerinaccuraciesforsmallcounts.

To overcomethesetwo limitations,we formulateda stochastic

birth and death process combining the occurrence of sporadic

casesandtheiroffsprings,extendingmethodsfortheestimation

ofreproductionnumbersbasedonsecondarycasesonlytoinclude

8 C.Polettoetal./Epidemics15(2016)1–9

analysis,weconsideredcasesoverintervalsofthesameduration

asthemeangenerationtime,butextensiontoarbitrarygeneration

timedistributionisstraightforward.WealsoassumedPoisson

vari-abilityinincidence,butallowedforover-dispersionbyintroducing

regionspecificparameters.

Ourstudyisaffectedbyasetoflimitations.First,thedetection

ratiowasassumedtobegeographicallyuniform.Thismaybe

unrealistic,especiallyacrossdifferentcountriesduetothedifferent

surveillancesystems.We believe,however,thattheassumption

ofuniformdetectionrateacrosstheprovincesofSaudiArabiais

justifiedbythenationalsurveillancesystemandconsequentlythe

geographicalsignaturepredictedbythemodelwithintheregionis

likelytobegenuine.Weassumedtobealsoconstantintime.Such

assumptionwasmadeinotherstudiesreferringtotheperiodbefore

Spring2014(Chowelletal.,2014);thiswasjustifiedbythefactthat

pro-activesurveillancewasputinplaceduringthatperiod(Memish

etal.,2014).ThesharpincreaseincasesduringSpring2014may

havecausedanoverloadinsurveillancesystemsandadecrease

inthedetectionprobability.Ifthisisthecase,theincreaseinthe

zoonotictransmissionduringtheperiodcouldbehigherthanthe

resultsprovidedhere.Itisimportanttonote,however,thatgiven

themodelassumptionssuchbiasdoesnotaffecttheestimatedratio

betweenprimaryandsecondarycasesnorthereproductiveratio.

Second,ourmodelassumestransmissionprobabilitytobethe

sameacrossdifferentsettings(e.g.households,hospitals).Wethus

provideunstratifiedestimatesofthereproductiveratio.Current

datadonotallowamoredetailedmodellingoftheepidemicspread

indifferentsettings.

Also,wedonotaccountfortheoccurrenceofsuper-spreading

events.Heterogeneityintransmissionwasaddressedbyanalysing

clusterdistributioninMiddleEast(KucharskiandAlthaus,2015)

andlocaltransmissionincountriesoutoftheMiddleEastregion

followingcaseimportation(Nishiuraetal.,2015).Resultsindicate

substantial potential for super-spreading that may have

con-tributedtothecaseinsurgenceduringtheSpring2014wave.

Eventually,thestudyrestrictstotheMiddleEastregionand

excludescontiguouscountrieslikeLebanonandIran,where

MERS-CoVcasesweredetected.Thiswasmotivatedbythefactthatthe

ArabianPeninsulaexperiencedthegreatmajorityofcasesandwas

alsothesourceofimportationofcasesatthegloballevel(ECDC).

Thustheterritoryrepresentsthefocusofmajorglobalhealth

con-cern.

Inconclusion,weintroducedanewmodellingapproach

appli-cabletoazoonoticdiseaseinasubcritical/criticalspreadingregime

thatisabletodisentangletherelativeroleofthedifferent

trans-mission routes. Applied to MERS-CoV, the model showed that

human-to-humantransmission is morefrequent than expected

andhighgeographicalheterogeneityandtemporalvariation

char-acterisethe zoonoticinsurgenceof caseswithbursts that have

thepotentialtotriggeroutbreakswithintensehuman-to-human

transmission.Assuch,priorityshouldbegiventothecontrolofthe

zoonotictransmissionwithabetterassessmentofthemechanisms

attheoriginoftheobservedvariationinthegenerationofcases.

Acknowledgements

ThisworkhasbeenpartiallyfundedbyReacting(INSERM);the

EC-HealthContractNo.278433(PREDEMICS)toV.C.;theANR

Con-tractNo.ANR-12-MONU-0018(HARMSFLU)toV.C.andC.P.

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in

theonlineversion,atdoi:10.1016/j.epidem.2015.12.001.

References

Alagaili,A.N.,Briese,T.,Mishra,N.,Kapoor,V.,Sameroff,S.C.,deWit,E.,etal.,2014. MiddleEastrespiratorysyndromecoronavirusinfectionindromedarycamelsin SaudiArabia.mBio5(2),http://dx.doi.org/10.1128/mBio.00884-14,e00884-14. Al-Tawfiq,J.A.,Memish,Z.A.,2014.MiddleEastrespiratorysyndromecoronavirus: transmissionandphylogeneticevolution.TrendsMicrobiol.22(10),573–579,

http://dx.doi.org/10.1016/j.tim.2014.08.001.

Assiri,A.,McGeer,A.,Perl,T.M.,Price,C.S.,AlRabeaah,A.A.,Cummings,D.A.,etal., 2013.HospitaloutbreakofMiddleEastrespiratorysyndromecoronavirus.N. Engl.J.Med.369(5),407–416,http://dx.doi.org/10.1056/NEJMoa1306742. Balcan,D.,Hu,H.,Gonc¸alves,B.,Bajardi,P.,Poletto,C.,Ramasco,J.J.,etal.,2009a.

Sea-sonaltransmissionpotentialandactivitypeaksofthenewinfluenzaA(H1N1): aMonteCarlolikelihoodanalysisbasedonhumanmobility.BMCMed.7,45,

http://dx.doi.org/10.1186/1741-7015-7-45.

Balcan,D.,Colizza,V.,Gonc¸alves,B.,Hu,H.,Ramasco,J.J.,Vespignani,A.,2009b. Multiscalemobilitynetworksandthespatialspreadingofinfectiousdiseases. Proc.Natl.Acad.Sci.U.S.A.106(51),21484–21489,http://dx.doi.org/10.1073/ pnas.0906910106.

Breban,R.,Riou,J.,Fontanet,A.,2013.InterhumantransmissibilityofMiddleEast respiratorysyndromecoronavirus:estimationofpandemicrisk.Lancet382 (9893),694–699,http://dx.doi.org/10.1016/S0140-6736(13)61492-0. Cauchemez,S.,Fraser,C.,VanKerkhove,M.D.,Donnelly,C.A.,Riley,S.,Rambaut,A.,

etal.,2014.MiddleEastrespiratorysyndromecoronavirus:quantificationofthe extentoftheepidemic,surveillancebiases,andtransmissibility.LancetInfect. Dis.14,50–56,http://dx.doi.org/10.1016/S1473-3099(13)70304-9.

Chowell,G.,Blumberg,S.,Simonsen,L.,Miller,M.A.,Viboud,C.,2014.Synthesizing dataandmodelsforthespreadofMERS-CoV,2013:Keyroleofindexcasesand hospitaltransmission.Epidemics9,40–51,http://dx.doi.org/10.1016/j.epidem. 2014.09.011.

Drosten,C.,Muth,D.,Corman,V.,Hussain,R.,AlMasri,M.,HajOmar,W.,etal.,2015. Anobservational,laboratory-basedstudyofoutbreaksofMERS-Coronavirus inJeddahandRiyadh,KingdomofSaudiArabia,2014.Clin.Infect.Dis.60(3), 369–377,http://dx.doi.org/10.1093/cid/ciu812.

EuropeanCentreforDiseasePreventionandControl(ECDC).RapidRisk Assess-ment:Severe respiratory disease associated with Middle Eastrespiratory syndromecoronavirus(MERS-CoV).Stockholm:ECDC;9March2015. Avail-able from: http://ecdc.europa.eu/en/publications/layouts/forms/Publication DispForm.aspx?List=4f55ad51-4aed-4d32-b960-af70113dbb90&ID=1274. Gossner,C.,Danielson,N.,Gervelmeyer,A.,Berthe,F.,Faye,B.,KaasikAaslav,K.,2014.

Human–DromedaryCamelinteractionsandtheriskofacquiringzoonotic Mid-dleEastrespiratorysyndromecoronavirusinfection.ZoonosesPublicHealth,

http://dx.doi.org/10.1111/zph.12171.

InternationalAirTransportAssociation(IATA)(accessedon25Mar2014).Available from:www.iata.org.

Kucharski,A.J.,Althaus,C.L.,2015.TheroleofsuperspreadinginMiddleEast Respira-torySindromeCoronavirus(MERS-CoV)transmission.Eurosurveillance20,25,

http://dx.doi.org/10.2807/1560-7917.ES2015.20.25.21167.

Majumder, M.S., Rivers, C., Lofgren, E., Fisman, D., 2014. Estimation of MERS-coronavirus reproductive number and case fatality rate for the Spring2014Saudi Arabiaoutbreak:insights frompublicly availabledata. PLoSCurr.Outbreaks(Dec18.),http://dx.doi.org/10.1371/currents.outbreaks. 98d2f8f3382d84f390736cd5f5fe133c.

Memish,Z.A.,Zumla,A.I.,Assiri,A.,2013.MiddleEastrespiratorysyndrome corona-virusinfectionsinhealthcareworkers.N.Engl.J.Med.369,884–886,http://dx. doi.org/10.1056/NEJMc1308698.

Memish,Z.A.,Al-Tawfiq,J.A.,Makhdoom,H.Q.,Al-Rabeeah,A.A.,Assiri,A.,Alhakeem, R.F.,etal.,2014.ScreeningforMiddleEastRespiratorysyndromecoronavirus infectioninhospitalpatientsandtheirhealthcareworkerandfamilycontacts: aprospectivedescriptivestudy.Clin.Microbiol.Infect.20,469–474,http://dx. doi.org/10.1111/1469-0691.12562.

MiddleEastrespiratorysyndromecoronavirus(MERS-CoV)–update,26June2014.

http://www.who.int/csr/don/20140626/en/.

Müller,M.A.,Meyer,B.,Corman,V.M.,Al-Masri,M.,Turkestani,A.,Ritz,D.,etal., 2015.PresenceofMiddleEastrespiratorysyndromecoronavirusantibodiesin SaudiArabia:anationwide,cross-sectional,serologicalstudy.LancetInfect.Dis. 15,559–564,http://dx.doi.org/10.1016/S1473-3099(15)70090-3.

Nishiura,H.,Miyamatsu,Y.,Chowell,G.,Saitoh,M.,2015.Assessingtheriskof observingmultiplegenerationsofMiddleEastrespiratorysyndrome(MERS) casesgivenanimportedcase.Eurosurveillance20(27),http://dx.doi.org/10. 2807/1560-7917.ES2015.20.27.2118.

Oboho1,I.K.,Tomczyk,S.M.,Al-Asmari,A.M.,Banjar,A.A.,Al-Mugti,H.,etal.,2015. 2014MERS-CoVoutbreakinJeddah—alinktohealthcarefacilities.N.Engl.J. Med.372(9),846–854,http://dx.doi.org/10.1056/NEJMoa1408636.

Poletto,C.,Pelat,C.,Levy-Bruhl,D.,Yazdanpanah,Y.,Boëlle,P.-Y.,Colizza,V.,2014. AssessmentoftheMiddleEastrespiratorysyndromecoronavirus(MERS-CoV) epidemicintheMiddleEastandriskofinternationalspreadusinganovel max-imumlikelihoodanalysisapproach.Eurosurveillance19(23),20824,http://dx. doi.org/10.2807/1560-7917.ES2014.19.23.20824PMID:24957746.

Rambaut,A.,2013.MERS-covSpatialTemporalandEpidemiologicalInformation, Availablefromhttp://epidemic.bio.ed.ac.uk/coronavirusbackground(accessed on31.3.15).

Saad,M.,Omrani,A.S.,Baig,K.,Bahloul,A.,Elzein,F.,Matin,M.A.,2014.Clinical aspectsandoutcomesof70patientswithMiddleEastrespiratorysyndrome coronavirusinfection:asingle-centerexperienceinSaudiArabia.Int.J.Infect. Dis.29,301–306,http://dx.doi.org/10.1016/j.ijid.2014.09.003.

TheWHOMERS-CoVResearchGroup,2013,November.Stateofknowledgeand datagapsofMiddleEastrespiratorysyndromecoronavirus(MERS-CoV)in humans.PLoSCurr.Outbreaks,http://dx.doi.org/10.1371/currents.outbreaks. 0bf719e352e7478f8ad85fa30127ddb8,Edition1.

Wallinga,J.,Teunis,P.,2004.Differentepidemiccurvesforsevereacuterespiratory syndromerevealsimilarimpactsofcontrolmeasures.Am.J.Epidemiol.160(6), 509–516,http://dx.doi.org/10./aje/kwh255.1093.

Wernery,U.,Corman,V.M.,Wong,E.Y.M.,Tsang,A.K.L.,Muth,D.,Lau,S.K.P.,etal., 2015.AcuteMiddleEastrespiratorysyndromecoronavirusinfectioninlivestock dromedaries,Dubai,2014.Emerg.Infect.Dis.21(6),http://dx.doi.org/10.3201/ eid2106.150038.

WorldHealthOrganization(WHO),WorldHealthOrganization(WHO),Global Alertand Response (GAR), Middle Eastrespiratory syndromecoronavirus (MERS-CoV) Available from: http://www.who.int/csr/disease/coronavirus infections/en/.

WorldHealthOrganization(WHO),MiddleEastrespiratorysyndrome corona-virus (MERS-CoV). Summary of MERS statistics inthe Republic of Korea (translated from the www.mers.go.kr website)asof 24June 2015, 9:00. Available from: http://www.wpro.who.int/outbreaksemergencies/summary. of.MERS.stats/en/(accessed24.06.15).