Susceptibility to SARS-CoV-2 Infection Among Children and Adolescents Compared With Adults
A Systematic Review and Meta-analysis
Russell M. Viner, PhD; Oliver T. Mytton, PhD; Chris Bonell, PhD; G. J. Melendez-Torres, PhD; Joseph Ward, MBBS; Lee Hudson, PhD;
Claire Waddington, DPhil; James Thomas, PhD; Simon Russell, PhD; Fiona van der Klis, PhD; Archana Koirala, MBChB;
Shamez Ladhani, MD; Jasmina Panovska-Griffiths, DPhil; Nicholas G. Davies, DPhil; Robert Booy, MD; Rosalind M. Eggo, PhD
IMPORTANCEThe degree to which children and adolescents are infected by and transmit severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is unclear. The role of children and adolescents in transmission of SARS-CoV-2 is dependent on susceptibility, symptoms, viral load, social contact patterns, and behavior.
OBJECTIVETo systematically review the susceptibility to and transmission of SARS-CoV-2 among children and adolescents compared with adults.
DATA SOURCESPubMed and medRxiv were searched from database inception to July 28, 2020, and a total of 13 926 studies were identified, with additional studies identified through hand searching of cited references and professional contacts.
STUDY SELECTIONStudies that provided data on the prevalence of SARS-CoV-2 in children and adolescents (younger than 20 years) compared with adults (20 years and older) derived from contact tracing or population screening were included. Single-household studies were excluded.
DATA EXTRACTION AND SYNTHESISPRISMA guidelines for abstracting data were followed, which was performed independently by 2 reviewers. Quality was assessed using a critical appraisal checklist for prevalence studies. Random-effects meta-analysis was undertaken.
MAIN OUTCOMES AND MEASURESSecondary infection rate (contact-tracing studies) or prevalence or seroprevalence (population screening studies) among children and adolescents compared with adults.
RESULTSA total of 32 studies comprising 41 640 children and adolescents and 268 945 adults met inclusion criteria, including 18 contact-tracing studies and 14 population screening studies. The pooled odds ratio of being an infected contact in children compared with adults was 0.56 (95% CI, 0.37-0.85), with substantial heterogeneity (I2= 94.6%).
Three school-based contact-tracing studies found minimal transmission from child or teacher index cases. Findings from population screening studies were heterogenous and were not suitable for meta-analysis. Most studies were consistent with lower seroprevalence in children compared with adults, although seroprevalence in adolescents appeared similar to adults.
CONCLUSIONS AND RELEVANCEIn this meta-analysis, there is preliminary evidence that children and adolescents have lower susceptibility to SARS-CoV-2, with an odds ratio of 0.56 for being an infected contact compared with adults. There is weak evidence that children and adolescents play a lesser role than adults in transmission of SARS-CoV-2 at a population level. This study provides no information on the infectivity of children.
JAMA Pediatr. 2021;175(2):143-156. doi:10.1001/jamapediatrics.2020.4573 Published online September 25, 2020. Corrected on November 2, 2020.
Editorialpage 127 Supplemental content CME Quiz at
jamacmelookup.comandCME Questionspage 215
Author Affiliations: Author affiliations are listed at the end of this article.
Corresponding Author: Russell M.
Viner, PhD, UCL Great Ormond Street Institute of Child Health,
30 Guilford St, London WC1N 1EH, United Kingdom (r.viner@ucl.ac.uk).
JAMA Pediatrics | Original Investigation
T
he degree to which children and adolescents younger than 20 years are infected by and transmit severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is an unanswered question.1-3These data are vital to inform na- tional plans for relaxing social distancing measures, includ- ing reopening schools.Children and adolescents account for 1% to 3% of reported coronavirus disease 2019 (COVID-19) cases across countries4-8and an even smaller proportion of severe cases and deaths.5,9Children appear more likely to have asymp- tomatic infection than adults, and analyses based on symptom-based series underestimate infections in children.
The role that children and adolescents play in transmission of SARS-CoV-2 is dependent on their risk of exposure, their probability of being infected on exposure (susceptibility), the extent to which they develop symptoms on infection, the extent to which they develop a viral load sufficiently high to transmit, and their propensity for making potentially infectious contact with others, dependent on numbers of social contacts across age groups and behavior during those contacts.
Different study types may provide useful information on susceptibility and transmission in children compared with adults, yet each is open to bias. Contact-tracing studies with systematic follow-up of all contacts to estimate second- ary attack rates in children and adults can provide strong evidence on differential susceptibility. Findings from some contact-tracing studies suggest that children have lower SARS-CoV-2 secondary attack rates than adults,10although others have found no difference by age.11One study from South Korea12has suggested adolescents but not children may have higher secondary attack rates, although a separate analysis of child cases from the same population identified minimal transmission from these individuals.13
Population screening studies may identify infection through viral RNA detection or antibodies indicating prior infection. However, the prevalence of SARS-CoV-2 in chil- dren in a population is not a direct indicator of susceptibility or transmission, as the expected prevalence depends on exposure, susceptibility, proportions of children in the popu- lation, mixing rates among children and between adults and children, and timing of social distancing interventions that disrupt mixing.
A number of authors have concluded that children and adolescents may be less susceptible to SARS-CoV-2,2,14al- though there are multiple sources of bias in each study type, which can complicate straightforward analysis. In contact- tracing studies, testing of only symptomatic contacts will in- troduce significant bias, as will seroprevalence studies drawn from clinical contact studies (eg, primary care) or residual laboratory sera. Many studies undertaken quickly during the pandemic are underpowered to identify age differences.
We undertook a systematic review and meta-analysis of published and unpublished literature to assess child and ado- lescent susceptibility to SARS-CoV-2 compared with adults. We limited this review to contact-tracing studies and population- based studies, as these are likely to be most informative and least open to bias.
Methods
Our review question was “What is the susceptibility to SARS- CoV-2 of children and adolescents compared with adults?” We undertook a rapid systematic review and included contact- tracing studies or prevalence studies in published or preprint form as well as data from a national public health website re- porting government statistics and studies. Studies were re- quired to provide data on proven SARS-CoV-2 infection (by polymerase chain reaction or serology) and report either rate of secondary infections in children and adolescents com- pared with adults or infection prevalence or seroprevalence in children and adolescents separate from adults. We ex- cluded reports of single household or single institution out- breaks; studies of hospitalized patients, clinical studies, and cohorts defined by symptoms; studies of unconfirmed cases, ie, cases based on self-report or symptoms, including contact- tracing studies where only symptomatic contacts were traced;
modeling studies or reviews, unless these reported new data;
and prevalence studies with ascertainment based on clinical contact and seroprevalence studies of residual sera, as these are likely to underrepresent children. This study followed the Preferred Reporting Items for Systematic Reviews and Meta- analyses (PRISMA) reporting guideline.
Where studies were drawn from populations that over- lapped, we excluded studies where the time periods overlapped but included studies where time periods did not overlap. We did not include seroprevalence studies only in children in this review, as these did not allow comparison with adults.
We searched 2 electronic databases, PubMed and the medi- cal preprint server medRxiv, on May 16, 2020, and updated this on July 28, 2020. We used the following search terms in PubMed: (“COVID-19”[tw] OR “2019-nCoV”[tw] OR “SARS- CoV-2”[tw]) AND ((child* OR infant*) OR (“transmission”[tw]
OR “transmission” [mh]) OR (“Disease Susceptibility”[tw] OR
“susceptibility”(mh)) OR (“epidemiology”[tw] OR “epidemi- ology” [mh]) OR (“contact tracing”[tw] or “communicable dis- ease contact tracing”[mh])). In medRxiv, we undertook
Key Points
QuestionWhat is the evidence on the susceptibility to and transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) among children and adolescents compared with adults?
FindingsIn this systematic review and meta-analysis including 32 studies, children and adolescents younger than 20 years had 44% lower odds of secondary infection with SARS-CoV-2 compared with adults 20 years and older; this finding was most marked in those younger than 10 to 14 years. Data were insufficient to conclude whether transmission of SARS-CoV-2 by children is lower than by adults.
MeaningPreliminary evidence suggests that children have a lower susceptibility to SARS-CoV-2 infection compared with adults, but the role that children and adolescents play in transmission of this virus remains unclear.
separate searches for “child and covid-19,” “covid-19 and epi- demiology,” “covid-19 and susceptibility,” and “covid-19 trans- mission,” as more complex Boolean searches are not available.
Figure 1 shows the PRISMA flow diagram. One researcher (R. M. V.) screened studies based on titles and abstracts to iden- tify potentially eligible studies for full-text review. Full-text studies were then reviewed by 2 researchers for eligibility (R.
M. V. and O. T. M. or C. W.), and data were extracted indepen- dently by 2 researchers (R. M. V. and either O. T. M. or C. W.).
We hand searched cited references in all potentially eligible studies for additional studies and identified additional stud- ies through authors’ professional networks. Data were ex- tracted on country, study type, study context with regards so- cial distancing measures and school closures at the time of the study, case definition, testing method, sampling method, and infection rates in adults and children.
Methodological quality of included studies was assessed independently by 3 authors (R. V. M., O. T. M., and C. W.) based on a critical appraisal checklist for prevalence studies.15We as- sessed risk of bias using 2 additional criteria: whether symp- tomatic contacts (in contact-tracing studies) or individuals (in population screening studies) were more likely to participate than asymptomatic ones and whether the obtained sample was more than 75% of the intended sample. Studies were catego- rized as high quality if they met all quality criteria and had low risk of bias on both criteria; medium quality if they had low risk of bias on 1 or more criteria and met 5 or more of 7 quality criteria; low quality if they met less than 5 quality criteria; or uncertain quality if multiple domains could not be scored.
Contact-tracing studies and population prevalence stud- ies were considered separately. Random-effects meta- analysis with restricted maximum likelihood estimation was undertaken using the meta commands in Stata version 16 (StataCorp). Odds ratios (ORs) were used as the primary met- ric for contact-tracing studies. Prevalence ratios were used as the primary metric in population-based studies. We planned subgroup analyses using restricted maximum likelihood based on quality of study and age of children and adolescents. P val- ues were calculated using χ2tests. Significance was set at a P value less than .05, and all P values were 2-tailed.
Results
The PubMed search resulted in 3465 studies and the med- Rxiv search resulted in 10 461 studies, of which 113 and 90 stud- ies, respectively, were examined in full, and 16 studies in- cluded (Figure 1). We identified a further 6 studies through reference checking and 10 studies through professional net- works. In total, 32 studies comprising 41 640 children and ado- lescents and 268 945 adults were included (Table),7,10,12,16-45
with quality and bias assessments shown in eTable 1 in the Supplementand weblinks for included studies shown in eTable 2 in theSupplement. A total of 18 studies were contact- tracing studies (CTSs),7,10,12,16-31with 3 based in schools,29-31 and 14 studies were population screening studies.7,32-45 Two were high quality,33,3522 were medium quality,10,12,16,
21,22,24-27,29-32,34,36,38,40-457 were low quality,17-20,23,28,39
and 1 was uncertain quality.7,37
Contact-Tracing Studies
A total of 6 studies were from mainland China,10,16,18,20,22,24
2 from the US,25,26and 1 each from Taiwan,17Japan,19South Korea,12Israel,21the Netherlands,7,28Brunei,27and India,23with 3 CTSs based in schools from Australia,29Ireland,30and Singapore.31Lower secondary attack rates in children and ado- lescents compared with adults were reported by 11 studies:
5 from provinces of China, including Hunan,10,22Hubei,16,18and Beijing,20and 6 from other countries, including Taiwan,17 Japan,19the US,25,26Israel,21and the Netherlands,7,28although confidence intervals were wide in some studies.
No significant differences in secondary attack rates by age were reported in 3 studies from Guangdong province, China,24 Brunei,27and the states of Tamil Nadu and Andhra Pradesh in India,23with 1 study from South Korea12reporting high sec- ondary attack rates in those younger than 19 years. In 3 of these studies,12,23,27secondary attack rates in younger children were low compared with adults, but those among teenagers were as high as or higher than adults.
We undertook a random-effects meta-analysis of secon- dary attack rates in children and adolescents compared with Figure 1. PRISMA Flow Diagram for Search
3465 Studies identified on PubMed 113 Studies screened and deemed potentially eligible
16 Studies included from searches
32 Studies included in analysis
10 461 Studies identified on medRxiv 90 Studies screened and deemed potentially eligible
9 Studies included 7 Studies included
104 Excluded 83 Excluded
6 Studies identified and included from reference checking
10 Studies identified and included from professional contacts
Table.CharacteristicsofIncludedStudies SourceStatusLocationRecruitmentofindexcasesRecruitmentandisolationofcontactsContacttypeNo.ofclusters,index cases,andcontactsCasedefinitionandvirustestingChild/adolescent age;adultage Contact-tracingstudies Zhangetal,10 2020Published andpeer reviewed Hunanprovince, ChinaAllconfirmedcases identifiedbyHunanCenter forDiseaseControlbetween January16andMarch1, 2020 Closecontactswereidentifiedthrough contacttracingofconfirmedcasesand placedundermedicalobservationfor 14d.Aclosecontactwasdefinedasan individualwhohadunprotectedclose contact(within1m)withaconfirmedcase oranasymptomaticinfectionwithin2d beforetheirsymptomonsetorsample collection.
Allcontact types114Clusters representing136index casesand7193contact cases.1Indexcase (0.7%)was<15y.
PositivefindingsonRT-PCR.Allclose contactsweretestedinaccordance withlocalpolicyregardlessof symptoms.Percentageofcontacts testedwasnotstated.
0-14y;≥15y Lietal,16 2020Published andpeer reviewed
Hubeiprovince, China(hospitals inZaoyangCity andChibiCity)
Indexcasesidentifiedfrom2 hospitalsfromJanuary1to February13,2020.Index caseswereexcludedif membersoftheirfamilyhad linkstoWuhanprovince.Not clearifallcasesfrom hospitalweresampledorjust asubset.
Allhouseholdcontactswerequarantined immediatelyfor14dbythelocal governmentandmonitoreddaily.
Household contacts105Indexpatientswith theirhouseholdsand allfamilycontacts (n=392). Theproportionofindex caseswhowerechildren wasnotreported.
PositivefindingsonRT-PCR. Nasopharyngealswabsampleswere collectedatthebeginningandthe middleofquarantine.Allcontacts weretested2to4times.
0-17y;≥18y Chengetal,17 2020Published andpeer reviewed
TaiwanTheinitial100confirmed casesinTaiwanbetween January15andMarch18, 2020.
Closecontactswereidentifiedthrough epidemiologicalinvestigationanddefined asapersonwhodidnotwearappropriate personalprotectionequipmentwhile havingface-to-facecontactwitha confirmedcaseformorethan15min duringtheinvestigationperiod(definedby epidemiologicalinvestigationandtypically upto4dpriortosymptomonsetortest dateforasymptomaticcases).Allclose contactswerequarantinedathomefor 14daftertheirlastexposuretotheindex case.
Allcontact types100Indexcasesand 2761closecontacts. Theyoungestindexcase wasaged11y,although theproportionofindex casesthatwerechildren wasnotreported.
PositivefindingsonRT-PCR.Routine testingforhouseholdandhealthcare workercontacts(30.7%).Other contacts(69.3%)wereonlytestedif symptomatic.
0-19y;≥20y Wangetal,18 2020Published andpeer reviewed
Wuhanprovince, ChinaPatientshospitalizedin UnionHospital(n=85)on February13and14,2020. Notclearifallcasesfrom hospitalweresampledorjust asubset.
Householdcontactsofthehospitalized patients,observedfor14d.Household contacts85Households correspondingtothe85 patientswereenrolled and155household contacts.
PositivefindingsonRT-PCRandthroat swabs.Processfortestinghousehold memberswasnotstated,but33%of householdcontactswerenottestedfor SARS-CoV-2.
Childagenot defined. Mizumoto etal,19 2020
PreprintJapanCasesthatweredomestically acquiredandconfirmedby RT-PCRbyMarch7,2020.
Contactsofindexcases,definition,and methodofascertainmentnotgiven. Nodetailsonisolationofcontacts.
Notstated. Thetotal numberof contacts(8 perindex case) suggests theseare likelyall contacts.
313Casesandtheir 2496closecontacts.PositivefindingsonRT-PCR.Process andeligibilityfortestingofcontacts notdescribed.
0-19y;≥20y (continued)
Table.CharacteristicsofIncludedStudies(continued) SourceStatusLocationRecruitmentofindexcasesRecruitmentandisolationofcontactsContacttypeNo.ofclusters,index cases,andcontactsCasedefinitionandvirustestingChild/adolescent age;adultage Wangetal,20 2020Published andpeer reviewed Beijing,ChinaAlllaboratory-confirmed (ie,RT-PCR–confirmed)cases inBeijinguptoFebruary21, 2020,recruitedthrough BeijingCenterforDisease Control.
FromFebruary28toMarch8,2020,all householdmembersofindexcaseswere observedfor14d.Testingandquarantine ofcontactsnotclearlydefined.
Household contacts124of137Eligible familiesparticipated. Noprimarycaseswere <18y.
Indexandsecondarycasesdefinedas positivefindingsonRT-PCR. ProportionofPCRtestingofsecondary contactswasnotstated.
0-17y;≥18y Parketal,12 2020Published andpeer reviewed
SouthKoreaAlllaboratory-confirmed casesinKorearegistered withtheKoreaCenterfor DiseaseControlfromJanuary 20toMay13,2020.
Allcontactsofindexcasesregisteredwith KoreaCenterforDiseaseControlthrougha comprehensivenationalcontact-tracing systemandobserved forameanof10d.
Household and nonhousehold contacts.Only dataon household contactswere includedin thisreview.
Studied59073contacts (10592werehousehold contacts)of5706index cases.Onlyindexcases whoreported1ormore contactswereincluded; however,only52%of 10962nationalcases reportedintheperiod wereincluded.
Householdandhealthcareworker contactsroutinelytestedbyRT-PCR. Othercontactsonlytestedif symptomatic.
0-19y;≥20y Dattneretal,21 2020PreprintIsraelIdentificationofall householdsinthecityofBnei Brak,Israel,whereall householdmembershad beentestedbyPCRand1or moremembershadpositive findings.Households identifiedthroughtheIsraeli COVID-19databaseuntilMay 2,2020,wereincluded.
Allhouseholdmembersincluded. Atotalof51%ofpopulationwere<20y.Household637Housescomprising 3353people,ofwhom 1510werepositivefor SARS-CoV-2.Alleligible householdswere included.Figures2,3, and4werederivedfrom suppliedestimated probabilitiesofchildren oradultsbeingthe index.
PositivefindingsonRT-PCRtestingof allhouseholdmembers,including indexcasesandcontacts.
0-19y;≥20y Huetal,22 2020PreprintHunanprovince, ChinaAllcaseswithcontactdetails wereidentifiedfromthe notifiableinfectiousdiseases reportingsysteminHunan provincefromJanuary16to April2,2020.
Contactswerequarantinedfor14dand testedwithPCRatleastonceduring quarantine.AfterFebruary7,2020, allcontactsweretested,butonly symptomaticcontactsweretestedbefore (approximately50%ofcontactstested).
Allcontacts1178Casesandtheir 15648contacts.PositivefindingsonRT-PCR.0-19y;≥20y Laxminarayan etal,23 2020
PreprintTamilNaduand AndhraPradesh, India
Indexcasesidentifiedfrom stateregistriesandcontacts tracedbypublichealth agenciesineachstatefrom March5toJune4,2020,in TamilNadu,India,andfrom March5toMay29,2020,in AndhraPradesh,India.
Contactstracedbypublichealthagencies andtestedbetween5and15dof exposure.Insufficientdetailprovided. Notethatthereweretwiceasmanyclose contactsperindexcase<18ycompared with>18y.
Allcontacts4206Confirmedcases and64031contacts. Noteonly4206cases includedof33584total cases(13%),withno detailgivenon nonrecruitment.
PositivefindingsonRT-PCRofall contactsregardlessofsymptoms.0-17y;≥18y Liuetal,24 2020Published andpeer reviewed
Guandong province,ChinaAllcasesidentifiedby intensiveregional surveillancebylocalcenter fordiseasecontrolfrom January15toMarch15, 2020.
ContactstracedandmonitoredwithPCR fromthroatswabstakeneveryfewdaysfor 14d;84%ofcontactswerequarantinedin centralizedstations.
Allcontacts1361Casesreportedand 11868contactstraced andquarantined. Analysisincluded11580 contacts(98%).
PositivefindingsonRT-PCRfrom throatswabs.0-19y;≥20y Rosenberg etal,25 2020
Published andpeer reviewed NewYorkState excludingNew YorkCity Identifiedandstudied229 initiallyconfirmedcasesby PCRfromMarch2to12, 2020.
Activecontacttracingbycountyandstate healthdepartments.Allhousehold contactswereeligibleforPCRtesting. Contactstested0to10dafterindexcase (43%weretestedonday0,ie,initialindex diagnosisday).
Household229Indexcasesand343 householdcontacts.PositivefindingsonRT-PCR.All householdcontactswereeligiblefor PCRtesting;however,notstatedwhat proportionweretested.
0-17y;≥18y (continued)
Table.CharacteristicsofIncludedStudies(continued) SourceStatusLocationRecruitmentofindexcasesRecruitmentandisolationofcontactsContacttypeNo.ofclusters,index cases,andcontactsCasedefinitionandvirustestingChild/adolescent age;adultage Yousafetal,26 2020Published andpeer reviewed Milwaukee, Wisconsin,and SaltLakeCity, Utah AllPCR-positivecasesfrom2 citieswereidentifiedthrough routinepublichealth surveillanceandrecruited betweenMarch22andApril 22,2020.
Activecontacttracingbypublichealth departments.Allcontactswereobserved for14dwith2swabtests(RT-PCR)onday 0andday14,plusifsymptomatic.
Household195of198Contacts participated(98.5%). Numbersofindexcases notstated.
PositivefindingsonRT-PCR.All householdcontactstested.0-17y;≥18y Chawetal,27 2020PreprintBruneiAll71initialcasesinBrunei, whicharosefollowinga religiousevent,withcases detectedafterMarch9, 2020.
DetailedcontacttracingbyMinistryof Health,withRT-PCRtestingofallreported contacts.Allcontactswerequarantinedfor 14dandretestedifsymptomatic.
Allcontacts71Indexcasesand1755 closecontacts.All contactsparticipated.
PositivefindingsonRT-PCR.0-19y;≥20y vanderHoek etal,7,28 2020
Published andpeer reviewed TheNetherlandsNationalsurveillancedata from2Dutchsystems:(1) OSIRIS,aregistryofall laboratory-confirmedcases and(2)HPZone.Dataon contacttracingfrom23of 25Dutchmunicipalities.
DataincludeduptoApril2,2020.Contact tracingwasundertakenforallcases registeredinHPZone.Contactinfection statusidentifiedthroughlinkagetothe mainnationalsurveillancedatabase, suggestingthatonlysymptomatic secondarycaseswereincluded.
Allcontacts231Casesand709close contacts.Proportionof contactstestednot stated.
PositivefindingsonRT-PCR.<19y Schoolcontact-tracingstudies Macartney etal,29 2020
Published andpeer reviewed
NewSouth Wales,AustraliaCOVID-19casesin25 educationalsettings(15 schoolsand10earlylearning centers)forwhichaperson (studentorstaff)with provenCOVID-19(positive findingsonPCR)had attendedwhileinfectious. Identifiedthroughstate NotifiableConditions InformationManagement System.Schoolsremained openbutstudentsdismissed fromMarch23,2020(<5% studentattendance).Note thatschoolattendance remainedhighatthetime thatsecondarycaseswere identifiedinschools,and early-yearssettingsdidnot close.
January25toApril9,2020.Allclose contacts(apersonwhohasbeenin face-to-facecontactforatleast15minor inthesameroomforatleast40minwitha casewhileinfectious)werefollowedup. Allclosecontactsobservedandtestedif symptomaticduringthe14-disolation period.7Settingshadtestingofall contacts5to10dafterlastcontactplus serologyafterday21.
Contactsin educational settingonly
27Primarycases(12 studentand15staff cases)andtheir1448 school-relatedandearly learning–relatedclose contactsfrom25 educationalsettings.12 Highschoolcases(8 students;4staff)from 10schoolshadatotalof 695contacts(598 students;97staff).The 5primaryschoolcases (1student;4staff)from 5schoolshadatotalof 218contacts(179 student;39staff).1448 Contactsidentified;663 (43.5%)weretested (PCRorserologyor both).
PositivefindingsonRT-PCRor serology(specificIgG,IgA,IgM detectionusingindirect immunofluorescence).Swabstaken from542of1448contacts(37.4%). Serologywasperformedin208of 1448contacts(14.3%).
6wk-18y; ≥20y Heaveyetal,30 2020Published andpeer reviewed
IrelandScreenedIrelandnational surveillancetoidentifyall PCR-positivecasesin childrenoradultswho attendedschoolsettings fromMarch1toMarch12, 2020.
Contactstracedandadvisedtoquarantine athomefor14d.Tested(PCR)onlyif symptomatic.
Allcontacts, including school contacts 6Indexcasesidentified (3adult;3<18y).1155 Contactsidentified(924 children;101adults).
PositivefindingsonRT-PCRtesting ifsymptomatic.0-17y;≥18y (continued)
Table.CharacteristicsofIncludedStudies(continued) SourceStatusLocationRecruitmentofindexcasesRecruitmentandisolationofcontactsContacttypeNo.ofclusters,index cases,andcontactsCasedefinitionandvirustestingChild/adolescent age;adultage Yungetal,31 2020Published andpeer reviewed Singapore3PotentialSARS-CoV-2 seedingincidentsin educationalsettingsin Singaporeidentifiedfrom nationalsurveillanceduring FebruaryandMarch2020.
Closeschoolcontacts(eg,classmates) quarantinedfor14d.Contactsin1school and1preschoolweretestedonlyif symptomatic;theseschoolswerenot closed.Contactsin1preschoolweretested byPCRafteranoutbreakcausingschool closure.
School contactsonly3PCR-positivechild indexcaseswere identifiedfrom2 preschoolsand1 secondaryschool.188 Contactswerestudied, ofwhom119(63%) weretested.
PositivefindingsonRT-PCR.1-16y Populationscreeningstudies Gudbjartsson etal,32 2020
Published andpeer reviewed
IcelandFirstinfectiondiagnosedon February28,2020; containmentmeasuresputin place.Primaryschoolsopen butsomesecondaryschools closedandmoderate restrictionsonsocial contactsfromMarch13to April6,2020.
Nationalpopulationscreening.Open invitationfor87%ofparticipantsthrough onlineportalbutwithcollectionofsample from1location(Reyjkavik)andrandom invitationforasubsample(13%).Children <10ymadeup6.4%ofsample. Participationinthestudywasprimarilyby requestofparticipantsratherthanby randomsampling,whichmayhave introducedbiasesinparticipation.
NAOnlypopulation screeningsample reportedhere.
PositivefindingsonRT-PCRon nasopharyngealandoropharyngeal samples.
0-9y;≥10y Lavezzo etal,33 2020
Published andpeer reviewed
Vo,Veneto region,ItalyQuarantinedcommunityin anareaofItalythatwas affectedearlyandseverelyin theepidemic;areawas lockeddownstarting February23,2020,for2 weeks.Studyundertaken closetotheimpositionof verystrictsocialdistancing measuresintheregion.
Allagegroupswerehomogeneously sampled,withage-specificpercentages rangingfrom70.8%to91.6%.2Surveys undertaken;firstsurveyonlyincludedhere (overallresponserate85.9%).Those<21y madeup17%ofsampleandhada participationrateof94%(0-10y)and 95%(11-20y).
NAWepresentdataonly fromthisfirstsurvey, althoughthearticlealso reportsasecondsurvey undertakenduring lockdown.
PositivefindingsonRT-PCRon nasopharyngealsamples.0-20y;≥21y PublicHealth Agencyof Sweden,34 2020
Online reportSwedenFirstdeathreportedin StockholmonMarch11, 2020.Voluntarysocial distancingmeasures recommendedstartingMarch 16,2020,withsecondary schoolsrecommendedto teachvirtually.Primary schoolsandearly-years settingsremainedopen throughout.
2Nationallyrepresentativesurveys undertakenbythePublicHealthAgencyof Sweden.Participantsinvitedbyemail: 2571of4480(57%)participatedinApril and2957of4487(66%)inMay.Children 0-15ymadeup18.9%oftheAprilsample and17.2%oftheMaysample.Participants performedhomeself-samplingusing nasopharyngealswabs.
NANAPositivefindingsonRT-PCRon nasopharyngealsamples.0-15y;≥16y Officefor National Statistics,35 2020
Online reportEnglandStrictnationalsocial distancingmeasuresenacted March20,2020,with gradualeasingoflockdown startingMay25,2020.
Representativesampleof35801 individualsinEngland.Thoseaged2-19y madeup17%ofthepopulation.Cases wereidentifiedbyhomeself-sampling usingnasopharyngealswabs,withcarers swabbingyoungchildren.79%ofInvited participantsprovided1ormoreswabs.
NARepeatedsurveyscarried outeachweek.Data shownherearethe cumulativeprevalenceof thoseeverpositive betweenApril26and June27,2020.
PositivefindingsonRT-PCRon nasopharyngealsamples.2-19y;≥20y (continued)
Table.CharacteristicsofIncludedStudies(continued) SourceStatusLocationRecruitmentofindexcasesRecruitmentandisolationofcontactsContacttypeNo.ofclusters,index cases,andcontactsCasedefinitionandvirustestingChild/adolescent age;adultage Pollánetal,36 2020Online reportSpainStrictsocialdistancingwas imposedonMarch14,2020. Somerestrictionswerelifted onApril27,2020,and furtherrestrictionsliftedon May11,2020.
UndertakenbySpanishMinistryofHealth. Nationalrepresentativesampleobtained fromrandomsamplingofhouseholdsin municipalitiesacrossSpain.Of102803 approached,61075participantsprovided pointofcaresamplesand51958included inbothimmunoassayandpointofcare. Thoseaged0-19y(n=11464)madeup 23%ofthepointofcaresampleand12.6% oftheimmunoassaysample.
NAWeusedthe point-of-caredatahere owingtothesample beingrepresentativeof thechildpopulation, unliketheimmunoassay test.
Point-of-caretest:rapid immunochromatographyIgG(Zhejiang OrientGeneBiotech)andSARS-CoV-2 immunoassay(AbbottLaboratories). ComparisonoftherapidtestIgGwith SARS-CoV-2serologyin16953ofthe studysamplefound97.3%agreement betweentests.
0-19y;≥20y National Institutefor PublicHealth andthe Environment,7,37 2020
Online reportTheNetherlandsSocialdistancingmeasures introducedgraduallyfrom March11,2020.Schools closedfromMarch15,2020.
UndertakenbytheNetherlandsNational InstituteforPublicHealthandthe Environment.Population-basedsampling wasundertakeninarandomsampleofa randomlychosensubsetofmunicipalities acrosstheNetherlands.Totalsampleof 2096.Those<20ymadeup20%ofsample.
NADataprovidedbyauthor F.v.d.K.Positivefindingsonserology(IgG).0-19y;≥20y Hallaletal,38 2020PreprintBrazilFirstcasesreportedFebruary 27,2020,withlocal/state lockdownsduringMarchand April2020.Somestates begantorelaxmeasuresin April2020.
Nationwideseroprevalencesurveyin133 sentinelcitiesin26Brazilianstates. Randomlyselectedhouseholdsvisitedand finger-prickrapidserologytestused.Total samplewas24995withhouseholdresponse rate(55%).Childrenheavily underrepresented—2.2%wereaged0-9y and9.1%wereaged10-19y.
NANARapidlateralflowtestusedinour analysis(WondfoSARS-CoV-2).0-19y Shakibaetal,39 2020PreprintIranPopulation-based seroprevalencestudyin5 countiesinGuilanprovince, Iran,inApril2020—previously veryhighvirusprevalence.
Multistageclusterrandomsampling approachandtelephonerecruitmentofhead ofhousehold.196of632Approached householdsparticipated(31%)with528 participants.
NANAVivaDiagCOVID-19IgM/IgGserology.0-17y;≥18y Biggsetal,40 2020Published andpeer reviewed
GeorgiaStudyundertakenbyUS CentersforDiseaseControl andPreventiontocoincide withendofshelter-in-place orders(April3to30,2020).
Surveyofarandomsampleofhouseholdsin 2metropolitanAtlantacountries.696 Personsfrom394of1675households (23.5%)participated.Children<18ywere 6.9%ofsamplecomparedwith22.4%of population.
NANATotalantibodymeasuredusingVITROS 3600ImmunodiagnosticSystem(Ortho ClinicalDiagnostics).
0-17y;≥18y Stringhini etal,41 2020
Published andpeer reviewed
Genevacanton, SwitzerlandFirstcaseonFebruary26, 2020.Schoolsclosedon March16,2020,andstrict socialdistancingmeasures introducedMarch20,2020. Seroprevalenceinitiatedusing apopulation-basedsamplein canton.
Populationbasedbutnotfullyrandom samplewithincanton(region).1300 Randomlyselectedadultsapproachedeach weekfor5weeksandinvitedtobringall householdmembers≥5yforserology.Only nonsymptomaticindividualsstudied.Atotal of2766of5492(50.4%)agreedto participate,anddatapresentedhereforfirst 1360.Totalof16.4%ofsampleaged0-19y, similartopopulation.
NAIndeterminatecaseswere treatedashaving negativefindingsin calculatingdataforthe meta-analysis.
ELISAtospikeprotein(Euroimmun).5-19y;≥20y Nawaetal,42 2020PreprintUtsunomiya, JapanFirstcasesinJapanstarting January15,2020.Allschools closedFebruary27,2020. Surveyconductedbetween thefirstandsecondspikesof infectioninthecity.
Population-basedseroprevalencesurvey:a randomsampleof1000households approached.Atotalof742of2290persons (32%)participated.13%Were<18y,similar topopulation.
NANAIgG(ShenzhenYHLOBiotech).0-17y;≥18y (continued)
adults, with data included from 14 studies.10,12,16-21,23-28
We combined data on children and adolescents younger than 20 years and compared it with an adult group 20 years and older; thus, ORs and prevalence rates for adults may differ from those reported in studies. The pooled OR estimate for all contact-tracing studies of being a child with secondary infection compared with being an adult was 0.56 (95% CI, 0.37-0.85), with high heterogeneity (I2= 94.6%) (Figure 2).10,12,16-21,23-28
We undertook a meta-analysis of 8 CTSs grouped by age of child (Figure 3)10,12,21,23-25,27,28; the ages differed across studies, and children were defined as those younger than 10 to 14 years, adolescents as those older than 10 to 12 years, and adults as those 20 years and older. The pooled OR for children was 0.52 (95% CI, 0.33-0.82), significantly lower than adults (1 [refer- ence]). For adolescents, this was nonsignificant (OR, 1.23; 95% CI, 0.64-2.36). χ2Test suggested this group difference was significant (χ2= 4.54; P = .03). When only the 8 high-quality and medium-quality studies with low risk of bias were examined,10,12,16,21,24-27this finding was no longer significant (OR, 0.68; 95% CI, 0.41-1.11); however, the difference in estimates be- tween low-quality studies and high-quality and medium-quality studies was not significant (eFigure 1 in theSupplement).
We hypothesized that CTSs including only house- hold contacts might provide a clearer indication of the relative susceptibility to infection of children vs adults be- cause all contacts within households might be assumed to receive a similar exposure to infection from index cases.
A post hoc analysis by type of contacts (eFigure 2 in the Supplement) showed that studies of household contacts had a lower pooled OR (OR, 0.41; 95% CI, 0.22-0.76)) than studies of all contacts (OR, 0.91; 95% CI, 0.69-1.21;
between-group variance: χ21= 5.31; P = .02).
Three studies undertook contac t trac ing in schools.29-31A statewide population-based CTS in edu- cational settings in Australia before and during school closures29found that 27 primary cases (56% staff) across 25 schools or early-years nurseries resulted in 18 second- ary cases in 4 settings, including an outbreak of 13 cases in 1 early-years setting initiated by a staff member, with no evidence of child-to-adult transmission. The second- ary attack rate was 1.2% (18 of 1448) overall, 0.4% (5 of 1411) when excluding the early-years outbreak, and 2.8%
(18 of 633) in those tested. Other national CTSs under- taken in schools in Ireland30and Singapore31before schools closed identified very few secondary cases in schools.
Population Screening Studies
Data from prevalence studies for children and ado- l e s c e nt s c o m p a re d w it h a d u l t s a re s h ow n i n Figure 4.7,32-36,38-45
We did not undertake a meta- analysis of population screening studies given the impor- tant differences in the populations, epidemic time points, and methodologies involved.
Table.CharacteristicsofIncludedStudies(continued) SourceStatusLocationRecruitmentofindexcasesRecruitmentandisolationofcontactsContacttypeNo.ofclusters,index cases,andcontactsCasedefinitionandvirustestingChild/adolescent age;adultage Paganietal,43 2020PreprintLombardy,ItalyThetownofCastiglione d’Adda,4550inhabitantshad highnumbersofinfections fromearlyinthepandemic. Locallockdownoccurredfrom February23,2020.
Entirepopulation(allages)invitedto participate;recruited4174of4550 inhabitants(92%)whohadrapidcapillary testing,ofwhomarandomsampleof562 (stratifiedforageandsex)hadformal serologybyvenipuncture.Those<19y madeup12%oftherapidandformal serologysamples.
NA22%ofPopulation showedoverallpositivity (22.2%onrapidtest, 22.6%onformal serology).Rapidtestused inmeta-analyseshere, asfindingsfromformal serologywerehighly similar.
Rapidcapillarytesting:lateral-flow immunochromatographictest (PrimaLab);serology:CLIA,IgG anti–SARS-CoV-2(Abbott Laboratories).
0-19y;≥20y Weisetal,44 2020PreprintThuringia, GermanySeroprevalencesurveyinthe previouslyquarantined communityNeustadt-am- Rennsteig,from6weeksafter aSARS-CoV-2outbreak (March22,2020).Local lockdowninitiated.
Allcommunityhouseholdsinvited. Atotalof626of883(71%ofcommunity) enrolled.Focusonchildparticipationand bloodcollectiontoberepresentative. Children1-17ywere9.5%ofthesample; 620gavebloodand600participantshadall 6serologicaltestsperformed.
NANASerologyby6quantificationmethods:2 ELISAand4immunoassay:EDINovel CoronavirusSARS-CoV-2IgGELISAkit (EpitopeDiagnostics);SARS-CoV-2IgG ELISAkit(Euroimmun);SARS-CoV-2 S1/S2IgGCLIAkit(DiaSorin);2019- nCoVIgGkit(SnibeCo);SARS-CoV-2 IgGCMIAkit(AbbottLaboratories);and ElecsysAnti-SARS-CoV-2kit(Roche).
1-17y;≥18y Streecketal,45 2020PreprintGangelt, GermanyCarnivalheldonFebruary15, 2020.Strictlocalsocial distancingmeasures introducedonFebruary28, 2020,duetolocaloutbreak anddeaths.
Arandomsampleof600householdswas invitedtoparticipate,and1007individuals from405householdsparticipated.919 Providedserologydata.Atotalof6.0%of samplewasmadeupofthoseaged5-14y.
NAAtotalof62%ofthe88 participantswhocould notbeassessedwere childrennotassessedfor technicalreasons.
Positivefindingsonserology(IgG).5-14y;≥15y Abbreviations:COVID-19,coronavirusdisease2019;ELISA,enzyme-linkedimmunosorbentassay;NA,notapplicable;RT-PCR,reversetranscriptase–polymerasechainreaction;SARS-CoV-2,severeacuterespiratorysyndrome coronavirus2.
Four studies reported virus prevalence.32-35National preva- lence studies from Iceland32and Sweden34undertaken while primary schools were open showed lower prevalence among children and adolescents than adults, as did a municipal study from Italy33undertaken just before lockdown while schools
were open. However, a nationally representative survey from England covering lockdown and the subsequent month iden- tified no significant differences by age.35
A total of 10 studies reported seroprevalence,7,36,38-453 being nationally representative.7,36,38A lower seropreva- Figure 2. Pooled Estimate of Odds of Being an Infected Contact Among Children and Adolescents Compared With Adults
for All Contact-Tracing Studies
Weight,
% Reduced odds
of secondary infection among those <20 y
Increased odds of secondary infection among those <20 y
0.01 0.1 1 10
OR (95% CI) Child
Positive Negative
Adult
Positive Negative Source
OR (95% CI)
Wang et al,18 2020 0.09 (0.02-0.46)
van der Hoek et al,28 2020 0.13 (0.01-2.09)
Li et al,16 2020 0.16 (0.06-0.46)
Wang et al,20 2020 0.25 (0.11-0.56)
Cheng et al,17 2020 0.39 (0.05-2.87)
Rosenberg et al,25 2020 0.39 (0.25-0.62)
Dattner et al,21 2020 0.43 (0.36-0.51)
Mizumoto et al,19 2020 0.43 (0.23-0.83)
Zhang et al,10 2020 0.64 (0.47-0.87)
Yousaf et al,26 2020 0.72 (0.35-1.46)
Chaw et al,27 2020 0.94 (0.49-1.81)
Laxminarayan et al,23 2020 1.08 (0.97-1.20)
Liu et al,24 2020 1.16 (0.92-1.46)
Park et al,12 2020 2.06 (1.54-2.76)
Overall 0.56 (0.37-0.85)
2 0 4 13 1 42 441 10 47 14 12 428 93 50
8 43 96 23 280 114 1297 165 709 55 418 5647 1774 644
130 55 60 64 21 88 432 284 606 33 39 2800 421 2119
49 611 232 28 2265 94 546 2037 5831 93 1278 39 756 9292 56 260
Heterogeneity: τ2 = 0.47; I2 = 94.64%; H2 = 18.64
4.02 1.80 6.00 7.03 2.96 8.59 9.45 7.77 9.12 7.51 7.75 9.53 9.32 9.16
Children and adolescents included those younger than 20 years, and adults included those 20 years and older. OR indicates odds ratio.
Figure 3. Pooled Estimate of Odds of Being an Infected Contact Among Children and Among Adolescents Compared With Adults for Contact-Tracing Studies
Weight,
% Reduced odds
of secondary infection among those <20 y
Increased odds of secondary infection among those <20 y
0.01 0.1 1 10
OR (95% CI) Child
Positive Negative Adult
Positive Negative Source
OR (95% CI) Children
van der Hoek et al,28 2020 (<12 y) 0.17 (0.01-2.90)
Dattner et al,21 2020 (<9 y) 0.25 (0.20-0.32)
Rosenberg et al,25 2020 (<5 y) 0.27 (0.10-0.74)
Chaw et al,27 2020 (<9 y) 0.50 (0.18-1.41)
Laxminarayan et al,23 2020 (<5 y) 0.55 (0.40-0.76)
Park et al,12 2020 (<9 y) 0.57 (0.24-1.39)
Zhang et al,10 2020 (<14 y) 0.64 (0.47-0.87)
Liu et al,24 2020 (<9 y) 1.34 (1.01-1.77)
Adolescents
van der Hoek et al,28 2020 (13-18 y) 0.44 (0.03-7.54)
Dattner et al,21 2020 (10-19 y) 0.66 (0.55-0.80)
Liu et al,24 2020 (10-19 y) 0.93 (0.65-1.33)
Park et al,12 2020 (10-19 y) 2.90 (2.13-3.96)
Chaw et al,27 2020 (10-19 y) 1.69 (0.78-3.68)
Overall 0.72 (0.46-1.10)
Heterogeneity: τ2 = 0.48; I2 = 93.11%; H2 = 14.52
38.83 Heterogeneity: τ2 = 0.31; I2 = 88.05%; H2 = 8.37 0.52 (0.33-0.82)
Heterogeneity: τ2 = 0.41; I2 = 91.59%; H2 = 11.90 1.23 (0.64-2.36) 0
149 5 4 40 5 47 60
0 291 33 8 45
31 742 20 263 1032 232 709 988
12 555 786 155 412
55 432 88 39 2800 2119 606 421
55 432 421 39 2119
611 546 94 1278 39 756 56 260 5831 9292
611 546 9292 1278 56 260
1.92 9.96 6.50 6.43 9.68 7.14 9.72 9.80 61.15
1.89 10.02 9.54 7.67 9.71
Children included those younger than 10 years, adolescents included those aged 10 to 19 years, and adults included those 20 years and older.
OR indicates odds ratio.
lence was identified in children and in some adolescents com- pared with adults in a number of studies, including a nation- ally representative study in Spain (ENE-COVID),36a Dutch nationally representative study (PIENTER Corona study),7,37 and city or regional studies from Iran,39the US,40Switzerland,41 and Japan,42although no difference by age was found in a survey in 133 sentinel cities in 26 Brazilian states.38Two community-based studies following localized outbreaks found lower seroprevalence among children and adolescents than adults in Lombardy, Italy,43and Thuringia, Germany,44with a second German postoutbreak study45finding no overall as- sociation with age. Examination of seroprevalence findings in children separate from adolescents (eFigure 3 in theSupple- ment) suggested that seroprevalence was lower among chil- dren younger than 10 years than adults but not lower among adolescents aged 10 to 19 years than adults, although this was not formally tested.
Discussion
We identified 32 studies from 21 countries that met our eligi- bility criteria and provided information on susceptibility to and transmission of SARS-CoV-2 in children and adolescents com- pared with adults. We excluded studies and study types open to very significant bias, yet studies were predominantly of me- dium and low quality, with only 2 high-quality studies.34,35 Most studies were from middle-income and high-income countries in East Asia and Europe.
We found preliminary evidence from 15 contact-tracing studies that children and adolescents have lower susceptibil- ity to SARS-CoV-2 infection than adults, with a pooled OR of 0.56 (95% CI, 0.37-0.85). This estimate was little changed when
only medium-quality or high-quality studies were examined, although power was reduced and significance was attenu- ated. Only 1 study13found a higher odds of infection in those younger than 20 years than adults, although this finding was confined to those aged 10 to 19 years. When studies were cat- egorized by age, lower susceptibility appeared to be confined to those younger than 10 to 14 years, who had 48% lower odds of infection compared with those 20 years and older. The age bands of the studies were not aligned, making direct compari- sons challenging.
Data from population screening studies were heterog- enous and were not suitable for meta-analysis. Findings con- sistent with lower seroprevalence in those younger than 20 years compared with adults were reported by 2 national studies,7,361 regional study,40and all of the municipal postout- break studies,43-45although confidence intervals were wide in some cases. Two virus prevalence studies similarly reported lower infection rates in those younger than 20 years. In con- trast, other studies reported no age-related differences. No studies reported higher prevalence in children and adoles- cents. Examination of seroprevalence findings in children sepa- rate from adolescents showed that most studies were consis- tent with lower seroprevalence in children compared with adults, although seroprevalence in adolescents appeared simi- lar to adults in all studies.
The findings from the CTSs and prevalence studies are largely consistent in suggesting that those younger than 10 to 14 years are less susceptible to SARS-CoV-2 infection than those 20 years and older, resulting in lower prevalence and sero- prevalence. Data specifically on adolescents are sparse but con- sistent with susceptibility and prevalence rates of adults. Our findings on susceptibility are similar to a modeling analysis by Davies et al,46which estimated that those younger than 20 Figure 4. Ratios of the Prevalence of Severe Acute Respiratory Syndrome Coronavirus 2 Infection in Children and Adolescents
Compared With Adults in Population Screening Studies
0.01 0.1 1 10
Risk ratio (95% CI) Child
Positive Negative Adult
Positive Negative Source
Risk ratio (95% CI) National seroprevalence
National Institute for Public Health and the Environment,7 2020
0.31 (0.13-0.77)
Pollán et al,36 2020 0.62 (0.56-0.69)
Hallal et al,38 2020 0.97 (0.71-1.31)
Regional or municipal seroprevalence
Biggs et al,40 2020 0.34 (0.02-5.54)
Shakiba et al,39 2020 0.77 (0.51-1.15)
Stringhini et al,41 2020 0.90 (0.63-1.29)
Nawa et al,42 2020 0.93 (0.05-17.88)
Postoutbreak seroprevalence
Weis et al,44 2020 0.19 (0.03-1.35)
Pagani et al,43 2020 0.46 (0.36-0.60)
Streeck et al,45 2020 0.59 (0.25-1.38)
Virus prevalence
Gudbjartsson et al,32 2020 0.07 (0.00-1.15)
Lavezzo et al,33 2020 0.22 (0.07-0.68)
Public Health Agency of Sweden,34 2020 0.73 (0.21-2.52)
Office for National Statistics,35 2020 0.94 (0.57-1.55)
5
388 46
0 24 33 0
1 52 5
57 423 50
51 866 132
511 2,802 725
581 69 2445
11 034 3353
48 110 422 98
848 464 482 5959 0
3 3 18
2712 303
19 97 186 3
100 70 17 96
46 941 21 294
629 320 2125 641
12 132 2275 2069 29 728
Lower risk in children
Higher risk in children
years were approximately half as susceptible to SARS-CoV-2 as adults.
We found few data that were informative on the onward transmission of SARS-CoV-2 from children to others. Data from a large Australian school contact-tracing study29sug- gest that, at a population level, children and adolescents might play only a limited role in the transmission of this virus. This is consistent with the data on susceptibility noted above, ie, suggesting that lower rates of secondary infection mean that children and adolescents have less opportunity for onward transmission. There is evidence of transmission from children to others in households and in schools, and there have been reported outbreaks in schools.47,48Other very small studies in Ireland30and Singapore31have found low numbers of secondary cases resulting from infected chil- dren attending school. This is consistent with a national South Korean study,13which found the secondary attack rate from children to household members was extremely low.
The available studies suggest children and adolescents play a lesser role in transmission of SARS-CoV-2, which is in marked contrast to pandemic influenza.49
Limitations
Our study has a number of limitations. We remain early in the COVID-19 pandemic, and data continue to evolve. It is possible that unknown factors related to age, eg, transience of infection or waning of immunity, bias findings in ways we do not yet understand. Some studies were low quality, and nearly all included studies were open to bias. The secondary infection rate in some CTSs was low, and this may represent an underestimate of the unmitigated household attack rate of SARS-CoV-2, as transmission chains were cut short because of strict control measures.50Most of the CTSs were undertaken when strict social distancing measures had been introduced, eg, closures of schools and workplaces and restriction of travel. This would have reduced contacts out- side the home, especially contacts between children, but it may have increased contacts between children and adults by increasing the household contact rate. The number of con- tacts nominated and traced for those younger than 20 years was low compared with adults in some studies,12,23which may have introduced bias. We identified 3 CTSs from Guang- dong province11,51,52that were excluded as they overlapped
with findings from Liu et al2 4; however, findings were unchanged if these studies were included. We included 2 recent large CTSs from India23and South Korea12; however, numbers of children and data quality appeared low, making firm conclusions difficult.
For population screening studies, the numbers of children tested was small in most of the studies and was frequently less than the 15% to 25% of the population that are younger than 18 years in most countries. This likely reflects lower recruitment of children and may be a source of bias, although the direction of this bias is unclear. Age differentials in sensitivity of swab or an- tibody tests may also confound findings. Interpreting the ob- served prevalence and seroprevalence studies requires thorough quantification of social mixing and transmission between age groups and how that changed during lockdowns and social dis- tancing interventions.
Conclusions
There is preliminary evidence that those younger than 10 to 14 years have lower susceptibility to SARS-CoV-2 infection than adults, with adolescents appearing to have similar suscep- tibility to adults. There is some weak evidence that children and adolescents play a limited role in transmission of SARS-CoV- 2; however, this is not directly addressed by our study.
We remain early in our knowledge of SARS-CoV-2, and fur- ther data are urgently needed, particularly from low-income settings. These include further large, high-quality contact- tracing studies with repeated swabbing and high-quality vi- rus detection and seroprevalence studies. Studies that inves- tigate secondary infections from child or adolescent index cases compared with secondary infections from adult index cases are particularly needed to assess transmission. Monitoring of infection rates and contact-tracing studies within child care and school settings will also be important. A range of serologi- cal studies are planned in many countries, and these need to be sufficiently powered to assess differences in seropreva- lence across different age groups and include repeated sam- pling at different time periods as social distancing restric- tions are lifted. We will continue to update this review, including further data as available and updating preliminary data from some included studies.
ARTICLE INFORMATION
Accepted for Publication: August 23, 2020.
Published Online: September 25, 2020.
doi:10.1001/jamapediatrics.2020.4573 Correction: This article was corrected on November 2, 2020, to add the Funding/Support and Role of the Funder/Sponsor sections and fix the degree for Jasmina Panovska-Griffiths.
Author Affiliations: UCL Great Ormond Street Institute of Child Health, London, United Kingdom (Viner, Ward, Hudson, Russell); MRC Epidemiology Unit, University of Cambridge, Cambridge, United Kingdom (Mytton); London School of Hygiene and Tropical Medicine, London, United Kingdom (Bonell, Davies, Eggo); College of Medicine and
Health, University of Exeter, Exeter, United Kingdom (Melendez-Torres); Department of Medicine, University of Cambridge, Cambridge, United Kingdom (Waddington); UCL Institute of Education, London, United Kingdom (Thomas);
National Institute for Public Health and the Environment (RIVM), Bilthoven, the Netherlands (van der Klis); University of Sydney, Sydney, Australia (Koirala, Booy); St George’s University of London, London, United Kingdom (Ladhani);
Department of Applied Health Research, University College London, London, United Kingdom (Panovska-Griffiths).
Author Contributions: Dr Viner had full access to all of the data in the study and takes responsibility
for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Viner, Bonell, Hudson, Eggo.
Acquisition, analysis, or interpretation of data:
Viner, Mytton, Bonell, Melendez-Torres, Ward, Waddington, Thomas, Russell, van der Klis, Koirala, Ladhani, Panovska-Griffiths, Davies, Booy, Eggo.
Drafting of the manuscript: Viner, Mytton, Bonell, Melendez-Torres, Waddington, Thomas, Eggo.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Viner, Melendez-Torres, Waddington, Thomas, Russell, Booy, Eggo.
Administrative, technical, or material support: Viner, Mytton, Waddington, Koirala.
Study supervision: Viner, Bonell, Eggo.
