2019 A Case Study Evaluating the Risk of Infection from Middle Eastern Respiratory Syndrome Coronavirus (MERS_CoV) in a

2019·1.02 MB·English

RiskAnalysis,Vol.0,No.0,2019 DOI:10.1111/risa.13389
A Case Study Evaluating the Risk of Infection from Middle
Eastern Respiratory Syndrome Coronavirus (MERS-CoV)
in a Hospital Setting Through Bioaerosols
UmeshAdhikari,1 AlexandreChabrelie ,1 MarkWeir ,2 Kevin Boehnke ,3
EricaMcKenzie ,4 LuisaIkner,5 MengWang,6 QingWang ,7 KyanaYoung,8
CharlesN.Haas ,9 JoanRose,8 andJadeMitchell 1,∗
MiddleEasternrespiratorysyndrome,anemergingviralinfectionwithaglobalcasefatality
rateof35.5%,causedmajoroutbreaksfirstin2012and2015,thoughnewcasesarecontinu-
ouslyreportedaroundtheworld.Transmissionisbelievedtomainlyoccurinhealthcareset-
tingsthroughaerosolizedparticles.ThisstudyusesQuantitativeMicrobialRiskAssessment
todevelopageneralizablemodelthatcanassistwithinterpretingreportedoutbreakdataor
predictriskofinfectionwithorwithouttherecommendedstrategies.Theexposurescenario
includes a single index patient emitting virus-containing aerosols into the air by coughing,
leading to short- and long-range airborne exposures for other patients in the same room,
nurses,healthcareworkers,andfamilyvisitors.Aerosoltransportmodelingwascoupledwith
Monte Carlo simulation to evaluate the risk of MERS illness for the exposed population.
Results from a typical scenario show the daily mean risk of infection to be the highest for
thenursesandhealthcareworkers(8.49×10−4 and7.91×10−4,respectively),andthelow-
estforfamilyvisitorsandpatientsstayinginthesameroom(3.12×10−4 and1.29×10−4,
respectively).Sensitivityanalysisindicatesthatmorethan90%oftheuncertaintyintherisk
characterizationisduetotheviralconcentrationinsaliva.Assessmentofriskinterventions
showedthatrespiratorymaskswerefoundtohaveagreatereffectinreducingtherisksforall
thegroupsevaluated(>90%riskreduction),whileincreasingtheairexchangewaseffective
fortheotherpatientsinthesameroomonly(upto58%riskreduction).
KEYWORDS: Hospital;MERS-CoV;mitigation;QMRA;riskcharacterization
1. INTRODUCTION
1DepartmentofBiosystemsandAgriculturalEngineering,Michi-
1.1. HistoricalBackground
ganStateUniversity,EastLansing,MI,USA.
2Division of Environmental Health Sciences, College of Public
Coronaviruses (CoVs) are a common cause
Health,TheOhioStateUniversity,Columbus,OH,USA.
3DepartmentofAnesthesiology&theChronicPainandFatigue of upper respiratory infections in humans. Strains
ResearchCenter,UniversityofMichigan,AnnArbor,MI,USA.
4Department of Civil and Environmental Engineering, Temple 8DepartmentofFisheriesandWildlife,MichiganStateUniversity,
University,Philadelphia,PA,USA. EastLansing,MI,USA.
5DepartmentofSoil,WaterandEnvironmentalScience,Univer- 9DepartmentofCivil,ArchitecturalandEnvironmentalEngineer-
sityofArizona,Tucson,AZ,USA. ing,DrexelUniversity,Philadelphia,PA,USA.
6DepartmentofCivil&Environmental Engineering, University ∗AddresscorrespondencetoJadeMitchell,DepartmentofBiosys-
ofSouthFlorida,Tampa,FL,USA. tems and Agricultural Engineering, Michigan State University,
7Department of Animal and Food Sciences, University of 524S.ShawLane,EastLansing,MI48824,USA;tel:+1(517)
Delaware,Newark,DE,USA. 353-4544;fax:[phone];[email protected].
1 0272-4332/19/0100-0001$22.00/1(cid:3)C2019SocietyforRiskAnalysis
2 Adhikarietal.
endemic to the human population include 229E, individuals to prevent contact with others is rec-
HKU, NL63, and OC43 (Chan et al., 2015), which ommended (CDC, 2017a, 2017b). The published
circulate continuously among children and adults literature on MERS has consistently estimated a
worldwide with infection trends typified by season- reproductive number (R ; the average number of
0
ality and mild symptoms in healthy individuals. The secondarycasesgeneratedbyaprimarycase)of<1,
new millennium, however, has marked the emer- suggestingthatMERS-CoVdoesnotyetposeapan-
genceviazoonosisoftwohighlyvirulentCoVstrains demicrisk(Brebanetal.,2013;Nishiura,Miyamatsu,
noveltothehumanpopulation.In2003,anovelCoV Chowell,&Saitoh,2015;Poletto,Pelat,Le´vy-Bruhl,
emerged in the Guangdong Province of China that Boelle,&Colizza,2016;WorldHealthOrganization,
caused a new and deadly outbreak of respiratory 2018). In Jeddah, Saudi Arabia, 82 of 168 clinical
disease in humans termed as severe acute respira- samples stemmed from a single hospital, and phy-
torysyndrome(SARS-CoV).Withinmonths,SARS- logenetic analyses of seven confirmed MERS-CoV
CoV spread rapidly to 25 countries (in part due to isolates from those cases were found to cluster in a
the modern, highly globalized nature of air travel), singlemonophyleticclade(Drostenetal.,2015).
withthousands sickened and closeto 800 fatalcases MERS-CoV is primarily transmitted through
(Hilgenfeld & Peiris, 2013). Investigation of the ori- infectious aerosolized particles. Under hospital
ginofSARS-CoVledfirsttotheexoticanimalmar- settings, the attack rate has been reported to be
ketsofChinaandtheinitialimplicationofpalmcivet 1.1–10% (Al-Abdallar, 2014; Al-Tawfiq & Perl,
catsandraccoondogs(whichwerefoundtobeinter- 2015),while3.6–5%attackrateshavebeenreported
mediate hosts), with further study indicating bats as for the persons in close contact with infected pa-
the true natural reservoir of SARS-like CoVs (Han tients (Al-Tawfiq & Perl, 2015; Memish, Assiri, &
etal.,2015).InApril2012,anoutbreakofsevereres- Al-Tawfiq, 2014). Mean incubation period for the
piratory viral illnesses localized in several intensive virus has been reported to range from 2 to 15 days,
careunitsoccurredintheMiddleEasterncountryof withamedianvalueoffivedays(Banik,Khandaker,
Jordan; both patients and healthcare workers were & Rashid, 2015). MERS-CoV infection results in
infected. Within several months, cases had also sur- fever,coughsorethroat,headache,andoccasionally
facedinseveralnearbyMiddleEasterncountriesin- results in nausea, vomiting, and diarrhea. In more
cluding Saudi Arabia, Qatar, and the United Arab severe cases, patients may experience shortness of
Emirates, with rapid movement into 20 additional breath, pneumonia, and death (Banik et al., 2015).
countries in North Africa and Europe (Al-Tawfiq, In the South Korean outbreak, the morbidity rate
2013; Breban, Riou, & Fontanet, 2013). Early indi- was estimated to be 1.08% (Ki, 2015). The patient
cations pointed toahighlyvirulent infectious agent, mortality rate has been reported to vary greatly
as a high percentage of patients were dying, partic- depending on the age and underlying conditions,
ularly those with comorbidities. A novel strain of such as diabetes, heart disease, and chronic lung
CoVwassoonisolatedandnamedaftertheregionof disease. In the South Korean outbreak, the overall
origininconjunctionwiththeprimarymanifestation mortality rate was reported to be 19.4%. MERS
ofsymptoms—MiddleEasternrespiratorysyndrome infected persons who were already hospitalized for
CoV(MERS-CoV)(deGrootetal.,2013). other medical conditions had a higher mortality
rate(33.8%)thanthepersonswithoutpriormedical
conditions (9.2%). Similarly, patients over 60 years
1.2. MERS-RelatedHealthIssues
of age had a higher mortality rate (38.1%) than
MERS-CoV affects the lungs and respiratory youngerpatients(6.4%)(Ki,2015).
systemwithanestimated35.5%mortalityinpatients
globally (World Health Organization, 2018). There
1.3. TheLargeOutbreakinSouthKorea,2015
arecurrentlynohumanvaccinesavailabletocounter
infection with MERS-CoV, while veterinary vac- A cluster of MERS-CoV cases arose in South
cines for camels are currently under developments Korea during May 2015. The visitation of a single
(Widagdo, Okba, Stalin Raj, & Haagmans, 2017). indexpatienttofivedifferenthospitalsisbelievedto
Therefore,todate,containmentofinfectiousviruses haveresultedin185downstreamnosocomialcasesof
via personal hygiene, use of personal protective MERS-CoV (Cowling et al., 2015; Park et al., 2015;
equipment (PPE), isolation of MERS symptomatic World Health Organization, 2015), although con-
persons, and quarantine of potentially exposed firmatory phylogenetic analyses have not yet been
ACaseStudyEvaluatingtheRiskofInfectionfromMERS-CoVinHospital 3
performed. Unlike the previously documented case sharingthesameroom;familyvisitorscomingtovisit
clusters, the South Korean outbreak was well doc- theindexpatient;andtheotherpatientssharingthe
umented with regard to incubation time, transmis- sameroom(Choetal.,2016).Riskestimationiscon-
sionchains(i.e.,28first-generationcases,125second- ductedbyusingtheMonteCarlosimulationmethod
generationcases,and32third-generationcases),and toincorporateuncertaintyandvariabilityintherisk
contact tracing of infected patients (Ki, 2015). The characterization.Sensitivityofthemodelparameters
majority of infections were hospital-acquired; only is assessed to determine where additional data or
one of the 186 patients in the South Korean clus- knowledge could potentially reduce uncertainty and
terwasbelievedtobeinfectedoutsideofahospital, increaseourunderstandingoftheserisks.Finally,the
andtwootherindividualswereinfectedbymodesof effectiveness of mask and increased ventilation risk
transmissionthatarecurrentlyunknown(Ki,2015). management measures is evaluated. Rather than a
Despite the fact that MERS has been reported retrospective case analysis, the study is intended to
to survive a maximum of 24–48 hours on surfaces contribute a framework for analyzing current and
(Van Doremalen & Munster, 2015), it has been futureMERSriskinsimilarsettings.
proposed that based on the South Korean MERS
outbreak, the virus would not survive long enough
to be capable of involving spread through indirect
2. MATERIALSANDMETHODS
fomite route (Cho et al., 2016). On contrast, studies
suggestedthatthemaintransmissionrouteofMERS
2.1. ExposureScenarioandAssessment
was via the airborne route, especially over close
contact airborne exposure (Xia et al., 2014). Hence, The basis of the exposure scenario involves a
isolation of index patient in a negative-pressure symptomatic patient infected with MERS-CoV who
roomandquarantineofpotentiallyexposedpersons has been admitted to a hospital without implemen-
are considered key risk management measures for tation of isolation or quarantine procedures. It was
literaturethatinvestigatedtheSouthKoreanMERS assumedthatallexposedpeopleweresusceptibleto
outbreak (Cowling et al., 2015; Kim et al., 2015; infection and all infections led to illness (or death).
Park et al., 2017; Park et al., 2016; World Health Atypicalsizeof230m3hospitalroomwassetforthe
Organization, 2015). In consequence, isolation and model, which is four times the single patient room
quarantinewouldbemeasuresthatwoulddrastically sizenotedinYin,Gupta,Zhang,Liu,&Chen(2011)
lower the risk of MERS infection once patients andisbasedonthefactthatover50%ofthehospital
are identified. From previous outbreaks, the index rooms in South Korea have four or more beds. The
patient stays unidentified as a MERS carrier for up symptomaticpatientwasconsideredtheonlysource
totwodays(Choetal.,2016).Additionally,thetime ofinfectionwithintheroom(seeFig.1).
for identifying MERS from a diagnostic laboratory MERS-CoVisthoughttobetransmittedprimar-
in a patient takes up to three days (Cowling et al., ily via aerosols in a manner similar to endemic hu-
2015),soprobableexposuredurationsaroundtwoto manrespiratoryCoVstrainssuchas229EandOC43.
threedaysarerelevantscenariostomodel. Forthepresentassessmentscenario,onlytheriskof
infectionfromaerosolizedparticlesanddropletsex-
pelledbycoughingwasconsidered.Theinfluenceof
1.4. StudyObjectives
nebulizer treatments that can be done on the index
TheobjectiveofthisstudywastousetheQuan- patientwasconsiderednegligibleandnotincludedin
titative Microbial Risk Assessment (QMRA) ap- the model. Although the contribution of this treat-
proachtodevelopageneralizablemodelforquanti- ment was suggested by Park et al. (2016), studies
fyingtheriskofinfectionassociatedwithin-hospital havealsodemonstratedthatnebulizersdonotspecif-
exposures to MERS through infectious aerosols. ically impact transmission (Seto, 2015; Thompson
The parameter values were selected from multiple et al., 2013). Fomites may also serve as a poten-
sources including the latest reported large outbreak tial reservoir for MERS-CoV due to the settling of
that occurred in South Korea, and data from other aerosols after release from infected persons. How-
sources.Riskofinfectionisestimatedforfourtypes ever,somestudiesstatedthatfomite-basedexposure
of at-risk populations: nurses and healthcare work- pathways were not significant compared to airborne
ers visiting the index patient (before the patient routes, and so it was not considered in this study
wasidentifiedascarryingMERS)andotherpatients (Xiao,Li,Sung,Wei,&Yang,2018).
4 Adhikarietal.
Fig.1. ExposurescenarioandQMRAoutlinesteps.QMRA=QuantitativeMicrobialRiskAssessment;MERS=MiddleEasternrespi-
ratorysyndromevirus;HCW=healthcareworker.
ACaseStudyEvaluatingtheRiskofInfectionfromMERS-CoVinHospital 5
Two forms of modeling were included in this with aerosols that were likely to be inspirable and
MERS assessment: (1) modeling aerosol concen- respirable. Aerosol production values were taken
trations to identify at-risk populations in hospital fromStilianakisandDrossinos(2010)andtherefer-
settings;and (2)estimatingexposure doseand char- ences therein. Particles with a diameter of <10 µm
acterizing risk. The risk of infection for several ex- were considered as respirable aerosols. Respirable
posure populations was considered as follows: (1) aerosols are expected to be easily transported, due
otherpatientsinthesameroomofindexpatient;(2) to their small diameter, and thus represent a po-
nurses; (3) other healthcare workers (e.g., doctors) tentialexposurepathwayforpeoplethatarefarther
visitingtheindexpatientandothersintheroom;and away from the source (e.g., more than 1–2 m from
(4)familymemberscomingtovisittheindexpatient. the source). Thus, respirable particles were the only
Viruses released via coughing and transport in evaluated exposure pathway for patients sharing a
the hospital room were modeled using a mass bal- roomwithaninfectedsymptomaticpatient.Aerosols
anceapproachtoapproximateasteady-stateconcen- with a diameter of 10–100 µm were considered as
tration of viruses contained in aerosol droplets. The inspirable aerosols as these large particles are not
droplets are being removed from the system either expected to be transported long distances and are
due to settling to the floor or ventilation-based air only relevant for persons in close contact. Nurses,
exchange. The risk of infection for each of the four healthcareworkers,andvisitorswereassumedtobe
populations wasassessedbasedonexposuresoccur- exposed to both respirable and inspirable aerosols.
ring over 1, 8, 20, and 41 days. These time periods Viral release into the room was calculated using
werebasedonreporteddurationsfromthesymptom Equation(2):
onset to discharge from the hospital during the Ko-
πd2
rean outbreak—a median of 20 days, minimum of V = i ×10−12, (2)
i
6
8 days, and maximum of 41 days (Ki, 2015)—
and from estimated durations for other patient where Vi (mL)isthevolumeforeachdropletsizedi
exposure—upto44hours(Choetal.,2016). that are released into the room as inspirable or res-
pirable droplets during each coughing event. Each
cough produced N number of droplets of size d,
i i
2.2. AerosolTransportModeling where each droplet is assumed to be spherical, and
thedropletvolumeiscalculatedas 1 ×π ×d3,where
Aerosol transport modeling was undertaken to 6 i
d is the diameter (µm). The droplets were assumed
assess virus inputs from coughing and removal via i
tobeproducedfromapatientlyingsupine,suchthat
settling onto surfaces and the air exchange pro-
thedropletcloudwasproducedata1mheight.
cesses(i.e.,heating,ventilation,andairconditioning
Following Stilianakis and Drossinos (2010),
[HVAC] systems). The model room system was as-
pathogen generation (e.g., coughing) and removal
sumed to have reached steady state, meaning that
(e.g., settling, ventilation) were assumed to be a
thereisnoaccumulationorlossfromthesystemover
continuous process. Exhalation by the infected pa-
time, and that the input flow rates must equal the
tientwasnotconsideredasourceofvirus-containing
removalflowrates.Thisinput–outputrelationshipis
droplets.
showninEquation(1):
After the particles were produced during a
N(in,coughing) = N(out,setting) +N(out,ventilation) coughing event, droplet evaporation, droplet set-
tling, and droplet removal via the ventilation were
+N , (1)
(inhalation) considered. Postevaporation particle transport was
whereNisthenumberofdropletscontainingviruses. evaluated, accounting for two removal mechanisms:
InEquation(1), N orthenumberofviruses droplet settling and ventilation-based droplet re-
(inhalation)
removed through inhalation by infected or unin- moval. Stoke’s law was used to calculate droplet
fected persons (patients in the same room, health terminal settling velocity vi(terminal) (m/hr) (Equa-
care workers, and visitors) was assumed to be non- tion (3)), which was assumed to be impacted only
significant as compared to the other two terms, byparticlediameterdi (Nicas,Nazaroff,&Hubbard,
N(out,settling) and N(out,ventilation), and thus was ne- 2005). (cid:2) (cid:3)
glected as a sink. Expiratory events (i.e., cough- 0.166
ing) produces a broad distribution of aerosol par- vi(terminal) = 0.108 × di2× 1+ d . (3)
i
ticles, however, this analysis was only concerned
6 Adhikarietal.
Terminalsettlingvelocities werecalculated foreach andwasconsideredentirelyanelasticcollision.Air-
oftherepresentativeparticlesizes,d.Acriticalset- borne particles were assumed to be homogeneously
i
tling velocity, v (m/hr), was calculated as the distributed within the volume of the room. Hence,
(critical)
required settling velocity to fall from the height of the number of droplets containing viruses removed
the patient bed h (m) during the air residence throughsettlingforeachdropletiis:
(cough)
time τ (hour) (Equation (4)). Air residence time, τ, (cid:2) v (cid:3)
is the average amount of time that a “parcel” of air Ni(out,settling) = Ni(in,cough) × 1− v (critical) .(8)
is in the room, which depends on the volume of the i(terminal)
roomv andtheventilationrateq . Air exchanges via ventilation was also consid-
(room) (ventilation)
v ered a removal mechanism, in which air, including
τ = (room) , (4) the homogeneously mixed virus-containing aerosol
q
(ventilation)
droplets, was removed from the hospital room and
replaced with new air. It was assumed that the re-
h
v = (cough) . (5) placement air contained no viruses. During each
(critical) τ
air replacement, all the remaining droplets were as-
Ventilation flow rate q(ventilation) was quantified by sumed to be removed by the ventilation, which im-
thenumberofairexchangesperhour(ACH)ofthe pliestherelationshipinEquation(9).
room volume, which was defined as shown in Equa-
tion(6): Ni(out,ventilation) = Ni(in,cough) −Ni(out,settling). (9)
q = V × ACH, (6) It was assumed that Ni(out,ventilation), number of
(ventilation) (room)
droplets remaining after settling, are suspended in
whereV isthevolumeofthehospitalroom(m3) the room until they are removed by ventilation.
(room)
andq istheairflowrate(m3/hr)determined Hence, the concentration of saliva in the air pro-
(exchange)
by the number of ACH v . As stated previ- ducedbyasinglecoughperunitvolumeofroomair
(exchange)
ously,heightofthepatientbed,h ,was1m.For iscalculatedas
(cough) (cid:4)
particlesthathadaterminalsettlingvelocitygreater n NV
thanthecriticalvelocity(vi(terminal) >v(critical)),itwas C(salivainair) = Vi=1 i i , (10)
assumed that settling was a viable removal mecha- (room)
nism. It was further assumed that droplets that hit where C is the concentration of saliva in
(salivainair)
the floor were permanently removed from the sys- the room air produced by a single cough per hour
temwithnoresuspension.Thisacknowledgesthatall (mL/m3),Vi isthevolumeofeachdropletcalculated
thesettleableaerosoldropletssettledtothefloorina usingEquation(2),andV istheroomvolume.
(room)
timeintervallessthanτ.However,duetothecontin- Wefurtherassumedastandardairexchangerate
uousgeneration,thereweresomefractionsoftheset- ofsixtimesperhour(Zumla&Hui,2014).Thehalf-
tleabledropletsthatwerenotyetsettled.Atagiven life of CoVs in the air is 67.33 hours (Ijaz, Brunner,
time, for the droplets with terminal velocity greater Sattar,Nair,&Johnson-Lussenburg,1985),butsince
thanthecriticalvelocity(v >v ),itwas we assumed that the air in the room was exchanged
i(terminal) (critical)
assumedthattheaerosolconcentrationofsettleable sixtimesperhour,decaywasnotconsidered.
droplets was proportional tothe ratioof settlingve-
locities, as shown in Equation (7). For the droplets
2.3. AerosolConcentrationsintheAir
thathadterminalsettlingvelocitieslessthanthecrit-
ical settling velocity (v ≤v ), it was as- To model the amount of virus released into the
i(terminal) (critical)
sumedthattherewasnodropletremovalviasettling. air, several studies were compared that specified
the number and size of droplets expelled during
(cid:2) (cid:3)
v coughing (Duguid, 1946; Loudon & Brown, 1967;
Ni(room,settleable) = Ni(in,cough) × v (critical) . (7) Nicas et al., 2005; Papineni & Rosenthal, 1997) for
i(terminal) selection of the data set that best fits the condition
Fortheselaterparticles,itwasassumedthataircur- of patients exposed to MERS-CoV. The number
rentsintheroomdictatedtheirtransport.However, of cough events per hour was modeled based on
this transport and homogeneous mixing did not in- Loudon and Brown (1967), using the estimates for
clude settling onto another surface resulting in re- the number of cough events in nonsmokers with
moval (i.e., striking a piece of furniture, or a wall) pneumonia. Based on Nicas and Jones (2009), we
ACaseStudyEvaluatingtheRiskofInfectionfromMERS-CoVinHospital 7
assumed that 0.044 mL of saliva was emitted per PFU)reportedbyHoungetal.(2004)andbasedon
cough, which represents the most conservative es- a SARS-CoV qPCR assay was employed to calcu-
timate compared to other published volumes in the late infectious PFU values for the MERS-CoV ex-
literature (Duguid, 1946; Loudon & Brown, 1967; posuremodeling.RecoveredMERS-CoVconcentra-
Papineni & Rosenthal, 1997). Saliva volume was tiondatawerefittedtoalognormaldistribution.
assumedtohaveauniformdistributionwitha±10%
ofthereportedvalue.Oftheexpiredfluid,0.00015%
2.5. ExposedPopulationBehavior
wasconsideredrespirableand0.54%wasconsidered
inspirable. In other words, about 99.45% of the Exposurescenariosforthenursesandhealthcare
volumeexpiredduringeachcoughwasconsideredto workers were modeled based on the frequency and
be nonrespirable and noninspirable, and therefore duration of their patient visits. For healthcare per-
wasnotincludedinthisanalysis.Respirabledroplets sonnel, due to the wide range of reported durations
were modeled as aerosols with mean postevapora- per visit by Cohen, Hyman, Rosenberg, and Larson
tiondiametersof4µmand8µm(forsmallandlarge (2012), a triangular distribution was specified with a
respirabledroplets),whichStilianakisandDrossinos medianoftwominutesandarangeof1–72minutes)
(2010) estimated were produced at a rate of 160 (Table I). Similarly, a triangular distribution with a
and 7.5 droplets per coughing event, respectively. median value of two minutes per visit and a range
Similarly,basedonStilianakisandDrossinos(2010), of 1–120 minutes was assumed for the nurses as in-
representative inspirable droplets corresponded to putsintheexposuremodel(Cohenetal.,2012).For
aerosols with mean postevaporation diameters of boththehealthcareworkersandnurses,thenumber
7.3µmand74µmdiameterdroplets(corresponding ofpatientvisitsandnumberofdifferentpatientsvis-
toinhalableaerosols),whichwereproducedat41.47 ited were also taken from Cohen et al. (2012) and
and 138.48 droplets per cough, respectively. Other are tabulated in Table I with all model inputs and
than this initial evaporation, it is assumed that the distributions. Nurses and healthcare workers were
aerosoldropletsdidnotchangeinsize,includingthat assumed to be exposed to inspirable and respirable
neither further evaporation nor particle aggregation particleswhilevisitingtheindexcase,andtotheres-
occurred. Uncertainty in the droplet production pirable particles while visiting other patients in the
numbers was investigated by holding the number sameroom.Otherpatientsintheroomwereassumed
of particles constant, and using bootstrap iterations to be exposed to respirable particles only 24 hours
to compare the uncertainty in the relative number a day (Ki, 2015). For the family visitors, a median
of particles for each of the four respective repre- visit duration of 14 minutes was used (Cohen et al.,
sentative particle sizes. The results of the bootstrap 2012). Furthermore, based on Cohen et al. (2012),
uncertainty analysis were used to model particle frequencyofvisitorswasassumedtorangefrom0to
productionasastochasticinput. 6.4 visits per hour with a median value of 1.3. Daily
exposure doses for nurses, healthcare workers, the
otherpatients,andfamilyvisitorswerecalculatedby
2.4. MERS-CoVConcentrationinSaliva
aggregating the exposure doses over the entire day
MultiplepapershavequantifiedlevelsofMERS- consistingofmultiplevisits.
CoV in sputum, nasopharyngeal secretions, and
saliva samples using the quantitative polymerase
2.6. EstimatedExposureDose
chainreaction(qPCR)methodology(Cormanetal.,
2015; Min et al., 2016; Muth et al., 2015). MERS- The daily exposure dose for the nurses and
CoV titer data specified in these studies are in to- healthcare workers was calculated by considering
tal viral units (noninfectious + infectious) of RNA that once entering the room, they would expose
genomic copies per milliliter (GC/mL) as the val- themselves both through respirable and inhalable
uesweregeneratedusingreal-timeqPCR.Sincethe aerosols during their visit to the MERS index pa-
dose–response model unit was in plaque-forming tient, and through only respirable aerosols when
unit (PFU), according to the used best-fit dose– visiting the other patients in the room. Hence, daily
response for SARS-CoV taken from the QMRA exposure dose for nurse and healthcare worker
Wiki website (Huang, 2013), a conversion factor of consisted of the sum of each of these two exposure
1,239:1 (1,239 GC equivalent units to one infectious routes:
8 Adhikarietal.
TableI. ParametersUsedintheModel
Parameters Unit Description InputValues(a;b)* Distribution Sources
V(saliva/cough) mL Volumeofsaliva 0.044(0.0396;0.0484) Uniform NicasandJones
expelled/cough(±10%) (2009)
R – Genomiccopies-to-PFU 1,239:1 Pointvalue Houng(2004)
(GC:PFU)
conversionfactor
C PFU/mL Virusconc.saliva=Conc. 41,734(7;201,945) Lognormal Corman(2015),
(MERSinsaliva)
[#GC/mL]×R Min(2016),
(GC:PFU)
Muth(2015)
N(cough/day) day−1 Numberofcoughs/day= 6.25(0.125;39.25) Triangular Loudon(1967)
N(cough/hr)× 24
di µm Dropletdiameter(4µm 4;8;7.3;74 Pointvalue Stilianakisand
and8µmfor<10µm Drossinos(2010)
respirabledroplets,
7.3µmand74µmfor10–
100µminspirable)
Ni # Numberof 160;7.5;41.47;138.48 Pointvalue Stilianakisand
droplets/diameterdi Drossinos(2010)
emitted/cough
Vi mL Volumeofeach Calculated Pointvalue Stilianakisand
droplet/diameterdi = Drossinos(2010)
(πdi2)/6×10−12
V m3 Hospitalroomsize 230 Pointvalue Yin(2011)
(room)
v m/hr Requireddropletsettling Calculated Pointvalue Nicas(2005)
(critical)
velocitytofallonground
= 0.108 ×d2×
i
(1+0.166/di)
C #/m3 Conc.dropletsinthe Calculated Normal Stilianakisand
(salivainair)
air/cough Drossinos(2010)
(cid:4)n
= ( NiVi)/V(room)
i=1
N(roomentries/hr) hr−1 Visitfrequencyofnurse 2.5(0;12.6) Triangular Cohen(2012)
N(roomentries/hr) hr−1 Visitfrequencyof 1.6(0;8.12) Triangular Cohen(2012)
healthcareworkers
N(roomentries/hr) hr−1 Visitfrequencyofafamily 1.3(0;6.4) Triangular Cohen(2012)
member
N(patientsvisited/entry) # Numberofdifferent 4.5(0.5;18) Triangular Cohen(2012)
patientsvisitedbya
nurse
N(patientsvisited/entry) # Numberofpatientsvisited 2.8(0.5;7) Triangular Cohen(2012)
byahealthcareworker
t(spent/entry) min Timespend/visitofanurse 2(1;120) Triangular Cohen(2012)
t(spent/entry) min Timespend/visitofa 3(1;72) Triangular Cohen(2012)
healthcareworker
t(spent/visit) min Timespend/visitofafamily 14(1;124) Triangular Cohen(2012)
member
texposed/d hr/d Contacttimeofother 24 Pointvalue Assumed
patientinthesame
room/d
V(inhaled/d) m3/hr Respirationrateofan 0.5 Pointvalue EPA(2011)
exposedperson
k PFU−1 Parameterofthe 0.00246(0.00135;0.00459) Normal Huang(2013)
exponential
dose–response
ACH hr−1 Airexchangerate(forthe 6 Basecase ZumlaandHui(2014)
basescenario)
F % %dropletsoutmask(from 0.032(0.010;0.100) Uniform Borkow(2010),Wen
(dropletsoutmask)
logreduction) (2013)
*a=Minvaluefortriangularandlognormaldistributionand5thpercentilevaluefornormaldistribution,respectively;b=maxvaluefor
triangularandlognormaldistributionand95thpercentilevaluefornormaldistribution,respectively.
ACaseStudyEvaluatingtheRiskofInfectionfromMERS-CoVinHospital 9
D(expo/d,n−hcw) =C(MERSinsaliva)×C(salivainair) tionwasconductedusingtheCrystalBall(cid:3)R program
(Version 11.1.4512.0, Oracle, Redwood Shores, CA,
1
×N(cough/hr)× ×V(inhaled/d) USA) to incorporate variability and uncertainty in
q
(ventilation) theinputparametersandtopropagateittotheout-
× N(Npa(trioeonmtsveinstitreieds//ehnrt)ry) ×t(spent/entry)×t(work/d), ptiuotn,parirsakmseotferisnf(ei.cet.i,oenx)p.oRsiusrkesdoofsiensfepcetriosnubfpoorpeualcah-
scenario were calculated using a published dose–
(11)
response model as described in Section 3. A differ-
entialsensitivityanalysisofmodelvariancewasper-
where D is the daily MERS virus inhaled by
(expo)
formed to determine which input variables have the
exposed personnel while being one time near index
greatest effect on the risk estimates. To reduce the
patient and another time near patients sharing the
riskofMERSinfection,twotypesofriskmitigation
room(PFU/day),C istheconcentration
(MERSinsaliva)
strategieswereevaluatedusingthefinalriskmodels:
of MERS in saliva (PFU/mL), C is the
(salivainair)
increasingairexchangerateandusingamaskasPPE.
concentration of droplets in the air after one cough,
N(coughs/hr) is the number of coughs per hour (#/hr), 2.7. Dose–ResponseModel
q is the ventilation air flow rate of the
(ventilation)
room derived from the ACH (#/hr), V(inhaled/d) is A primary knowledge gap in the study is the
the air intake rate of the exposed person (m3/hr), absence of a dose–response model for MERS-CoV.
N(roomentries/hr) is the number of entries nurse or Therefore,theSARSdose–responsemodel(Huang,
healthcare worker makes per hour to visit either 2013) was employed as a surrogate. MERS has sev-
the index patient or the other patients (#/hr), eral similarities to SARS: both have an animal ori-
N(patientsvisited/entry) is the number of patients visited gin and appeared around 2002 in approximately
bynursesorhealthcareworkersperroomentryvisit the same regions—Asia and Middle East (Sutton
(for the index patient or other patients) (#/visit), & Subbarao, 2015), both are respiratory CoVs with
t(spent/entry) is the amount of time spent during each the same transmission route, both have a compara-
visit (hr/visit), and t(work/d) is the number of daily ble protein structure for binding to host cells (Lu,
working hours for nurses and healthcare personnel Wang, & Gao, 2015), and both have reported simi-
(assumed8hr/day). lar tropism within cells (Zhou, Chu, Chan, & Yuen,
For the other patients in the same room, daily 2015).Hence,despiteprobable differencesinattack
exposuredosewascalculatedasfollows: rates and mortality rates between the two viruses
(Chanetal.,2015),itwasassumedinthisstudythat
Dexpo,op =C(MERSinsaliva) ×C(salivainair)× N(cough/hr) the SARS dose–response model is the best avail-
1 able model for MERS. Several dose–response stud-
× ×V(inhaled/d)×texposed/d, ies for SARS were evaluated to determine a rec-
q
(ventilation)
ommended dose–response model (De Albuquerque
(12) et al., 2006; DeDiego et al., 2008; Mitchell & Weir,
n.d.; Watanabe, Bartrand, Weir, Omura, & Haas,
where the daily exposure duration texposed/d was as-
2010). Recommended SARS dose–response model
sumedtobecontinuous(i.e.,24hr/d).
follows the exponential dose–response relationship
For the family visitors, daily exposure dose was
(Equation(14))forexposuredoseexpressedinPFU
calculatedbasedontheirnumberofvisitsperdayof
and the probability of a response based on an end
theindexpatient Nfamilyvisits/d:
point of death in mice (De Albuquerque et al.,
Dexpo,fm =C(MERSinsaliva)×C(salivainair)×N(cough/hr) 2006; DeDiego et al., 2008). For translating this an-
imal dose–response relationship to a human dose–
1
× ×V(inhaled/d)×N(familyvisits/d) response relationship, a generally accepted assump-
q
(ventilation) tionthatadeathendpointforananimalmodelmay
×t(spent/visit). (13) be used for examining the human risk of infection
wasapplied(Haas,Rose,&Gerba,2014).
Thegeneralequationfortheexponentialmodel
Asystematicliteraturereviewwasconductedto is:
determine the best estimates for each input param-
eter in the exposure model. A Monte Carlo simula- P = 1−e(−kd), (14)
Inf
10 Adhikarietal.
where P is the risk (probability) of infection, k 3. RESULTS
Inf
is the optimized dose–response function parameter
(PFU−1), and d is the dose (PFU). In the Monte 3.1. RiskofInfection
Carlo analysis, the k value in the dose–response
Based on the results obtained from the Monte
modelwasmodeledwithanormaldistributionbased
Carlo simulation, the risk of infection to each
on the 5th, 50th (median), and 95th percentile val-
exposed group was characterized. Fig. 2 shows the
ues reported by Huang (2013) and reported in
boxplotofdailyriskofinfectiontoeachgroupforthe
TableI.
base scenario, meaning without any preventive in-
Thecumulativeriskofthemorbidityacrossmul-
terventions,thestandardrateforhospitalsof6ACH
tiple exposure days was modeled by Equation (14)
was considered (Zumla & Hui, 2014). The median
(Haas,Rose,&Gerba,2014):
(mean) daily risk for the nurses coming to visit the
index patient and other patients in the same rooms;
P = 1−(1− P )n, (15) the healthcare workers (e.g., doctors); the family
M Inf
members coming to visit the index patient; and the
other patients sharing the room were found to be
where P istheprobabilityofmorbidityandnisthe 1.33×10−8 (8.49×10−4),1.18×10−8 (7.91×10−4),
M
number of days of exposures with P , probability 6.36 × 10−9 (3.12 × 10−4), and 2.73 × 10−9 (1.29 ×
Inf
of infection from a daily exposure. The risk associ- 10−4) respectively. The estimated highest daily risk
,
atedwitheachpopulationwasassessedfor8,20,and of infection for the healthcare workers and nurses
41daysofexposure,whichrepresentstheminimum, suggested the frequency of airborne close-range ex-
median,andmaximumhospitalizationperiodsforan posurerouteplaysabiggerroleinthetransmissionof
MERSinfectedpatients(Ki,2015). MERS, compared to the long-range airborne route
towhichotherpatientsareexposed,confirmingwhat
wassuggestedbyXiaoetal.’s(2018)work.Statistical
t-tests showed that the daily risk of infection for
2.8. RiskManagementEvaluation—AirChange healthcare workers was significantly higher than
PerHourandWearingofMask the one for the other patients or the family visitors
(p-value = 0.0014 and 0.0240, respectively, at α =
Toreducetheamountofairbornerespirablepar-
0.05). When comparing nurses and other healthcare
ticles, Zumla and Hui (2014) recommend increasing
worker, the result is not significant (p-value =
theairchangesperhour(ACH)from6to12inhos-
0.8475), so they have similar risks. Other patients in
pital facilities or rooms with high risk of airborne
the same room had a statistically significant lower
disease. Thus, in addition to the worst-case scenario
risk of infection compared to nurses (p-value =
considering 0 ACH and the Korean outbreak sce-
0.0017),buthadnonsignificantstatisticaldifferences
nario using 3 ACH (Cho et al., 2016), standard 6
inriskwithfamilyvisitors(p-value=0.0547).
ACH (Zumla & Hui, 2014), along with increased 9
Fig. 3 shows the aggregated risk of infection
and12ACHwereevaluatedfortheirefficacyinmin-
for the exposed populations during multiple daily
imizingtheinfectionrisk.
exposures to the MERS-infected patient for typical
Weevaluatedtheuseofrespiratorymasks(N95)
hospital durations. As expected, the results show
as a means of personal protection. Laboratory stud-
iesshowedalargedecrease(upto>4logreduction) increasedriskofinfectiontoalltheexposedpopula-
tionsovertime.Therateofincreasewashighestfor
in virus exposure when wearing masks (Borkow,
the healthcare workers and nurses, in comparison
Zhou, Page, & Gabbay, 2010). However, the de-
to the family visitors, which itself had higher rate of
crease did not take into account imperfect mask fit
increase compared to the other patients sharing the
or lack of compliance in wearing the masks. Due
room. Similar to the daily risk, aggregated risk of
tothesefactors,MERS-CoVreductionduetowear-
infection was the highest for the healthcare workers
ing N95 respirators was assumed to have a uniform
and nurses, followed by family visitors and other
distribution spanning 1–2 log reductions in MERS-
patients. By day 41, the average risk of infection to
CoVconcentration(Bałazy,Toivola,Adhikari,etal.,
the nurses was 1.01, 1.2, and 2.4 times the risk for
2006; Bałazy, Toivola, Reponen, et al., 2006; Gupta,
the healthcare workers, family members, and other
2011; Rengasamy, Zhuang, & Berryann, 2004; Wen
patients,respectively.
etal.,2013).