1

RapidScopingReviewofEvidenceofOutdoorTransmissionofCOVID-19

.MikeWeed(1)&AbbyFoad(2)(1)ProfessorofAppliedPolicySciences&ProVice-Chancellor(Research&Enterprise)(2)ResearchDirector,CentreforSport,PhysicalEducation&ActivityResearch(spear)CanterburyChristChurchUniversity,NorthHolmesCampus,Canterbury,CT11QUCorrespondence:mike.weed@canterbury.ac.ukv.1(4.7.2020)–notpeerreviewed|ProtocolpreviouslypublishedonmedRxivhttps://doi.org/10.1101/2020.08.07.20170373

Abstract

The COVID-19 pandemic is both a globalhealthcrisis,andacivicemergencyfornationalgovernments, including the UK. As countriesacross the world loosen their lockdownrestrictions, the assumption is generally madethat the riskofCOVID-19 transmission is loweroutdoors, and this assumption has shapeddecisions about what activities can re-commence, the circumstances in which theyshouldre-commence,andtheconditionsunderwhich they should re-commence. This isimportant for events and activities thatgenerate outdoor gatherings of people,includingbothparticipatoryandspectatorsportevents, protests, concerts, carnivals, festivals,andothercelebrations.

The review, which was designed to beundertaken rapidly in 15 days, returned 14sourcesofevidenceofoutdoor transmissionofCOVID-19, and a further 21 sources that wereused to set the context and understand thecaveats that should be considered ininterpretingthereviewfindings.

The review found very few examples ofoutdoor transmission of COVID-19 in everydaylife among c. 25,000 cases considered,suggesting a very low risk. However risk ofoutdoor transmission increases when thenatural social distancing of everyday life isbreached,andgatheringdensity,circulationandsize increases, particularly for an extendedduration.Therewasalsoevidencethatweatherhad a behavioural effect on transmission, withtemperatures that encourage outdoor activityassociated with lower COVID-19 transmission.Dueto lackofsurveillanceandtracingsystems,

and confounding factors and variables, therewas no evidence that robustly testedtransmission at outdoormass gatherings (circa10,000+people),whichareaslikelytogeneratetransmission from the activities they prompt(e.g.communaltravelandcongregationinbars)as from outdoor transmission at the gatheringitself.

The goal of hosts and organisers of eventsandactivities that generateoutdoorgatheringsof people is to prevent the escalation of riskfrom sporadic transmission to the risk oftransmission through a cluster outbreak.Considerations for such hosts and organisersinclude: (1) does the gathering prompt otherbehaviours that might increase transmissionrisk?; (2) for eachpart of theeventor activity,how dense is the gathering, how much dopeoplecirculate,howlargeisthegathering,andhowlongarepeoplethere?;(3)israpidcontacttracing possible in the event of an outbreak?Theseconsiderationsshouldtakeplacerelevanttothesizeoftheunderlyingrisk,whichincludesthe rateof infection in thecommunityand thelikely attendance of vulnerable groups. Riskmustbebalancedandmitigatedacrosstheriskfactorsofdensity,circulation,sizeandduration.Nooneriskfactorpresentsaninherently largerrisk thananyother, butneither is anyone riskfactoramagicbullettoeliminaterisk.Keywords:COVID-19, outdoor transmission, outdoorgatherings, events, running, sport, fans,spectators, physical activity, protests, concerts,carnivals,festivals

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.

2

IntroductionFollowing 213 global deaths and 9,800infections, on 30th January 2020 the WorldHealthOrganisation categorised COVID-19 as aPublic Health Emergency of InternationalConcern (PHEIC), and five weeks later, on 11thMarch, as a pandemic, atwhich point 118,000cases and 4,291 deaths in 114 countries hadbeen reported (Ghebreyesus, 2020). The firstcluster of COVID-19 cases were recorded inWuhan,Chinaon21stDecember2019,and thefirstdeathon11thJanuary2020.IntheUK,thefirstdomesticallycontractedcasewasrecordedon 28th February, and the first death on 5thMarch. A global spread across Europe andNorthandSouthAmericameant that countriesineverycontinentaroundtheworld,facedwitha virus forwhich therewas no vaccine and notreatment,implementedlockdownmeasurestodealwithaglobalpandemicthathadresultedinover850,000recordeddeathsworldwideby1stSeptember2020(Worldometer,2020).

Following reductions in the spread of thevirus as a result of these periods of lockdown,many countries are now developing andimplementingplanstocautiouslyopenuptheirsocieties and economies. Most of Europe hasloosened lockdown measures and is nowputtinginplacemitigationmeasurestolivewiththevirus,includingdecidingwhichactivitiescanre-commence, and which cannot. In the UK,after six weeks of lockdown, the governmentpublishedaCOVID-19recoverystrategyon11thMay, with an update for further expansionpublishedon24thJuly(HMGovernment,2020a).

One of the assumptions in the UK recoveryplan,andthoseofmanyothercountries,isthatthe risk of transmission of the virus is loweroutdoors (HM Government, 2020a). Thisassumption is central in informing decisionsaboutwhichpartsof theeconomyandsociety,andwhatactivities,canre-commence,includingdecisions about the mitigation measuresrequired.Oneareainwhichthisisofparticularimportance is physical activity and sports, andparticularly mass participation events such asrunning. Following government guidelinesissuedon10th July (HMGovernment, 2020b) ithasbeenpossibleformassparticipationrunning

events to take place, and some have done sounder principles drawing on governmentguidelines issuedbyUKAthletics (UKAthletics,2020). These principles require additionalmitigation measures to ensure some socialdistancing, including but not limited to,staggered wave starts, and strategies to limitpre-eventgathering (suchasnothavingon-sitebriefings,registrationandannouncements),andto accelerate post-event dispersal. However,for many event hosts, the issue of the risk oftransmission of COVID-19 outdoors remainsparamount, and this paper sets out a protocolfor a rapid scoping review to explore evidenceof such outdoor transmission. The findings ofthe review will also be relevant to events inother sectors that generate mass gatherings,suchasconcerts,carnivalsandfestivals.

ReviewScopeThereviewwasdesignedtoseek,evaluateandanalyse evidence of incidents of outdoortransmission of COVID-19, the settings andenvironmentsofsuchtransmission,and,whereavailable, all relevant circumstances, including,but not limited to, temperature, windconditions, social crowding or distancing, andthe existence or otherwise of any COVID-19mitigation measures. The review was alsodesignedtoseek,evaluateandanalyseevidenceof the prevalence of outdoor transmissioncomparedtoindoortransmission,andevidenceof the impact of high profile mass gatherings,both immediately before (e.g. ChampionsLeague soccer matches) and during (e.g. BlackLivesMatterprotests)lockdowns.

While not designed to seek, evaluate oranalyse evidence relating to the science ofoutdoor transmissionofCOVID-19, key insightsfrom the extant science and literature, mostlyidentified as part of the review protocol, havebeen included to set the context andunderstand the caveats that should beconsideredininterpretingthereviewfindings

The review was designed to be undertakenrapidly in 15 days. Therefore, a key additionalpurposeof thereviewwastoassesswhetherafurther extended more detailed and

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

3

comprehensivereviewwouldcapturewiderandmoreextensiveevidence.

ReviewProtocolThe protocol for the review comprised threesearch elements: electronic searches usingGoogle Scholar; pursuit of chains of sourcesreferredto inpapersreturned intheelectronicsearches to the original source or sources ofevidence;handsearchesofpapersandevidencesources considered by the UK government’sScientific Advisory Group for Emergencies(SAGE), and its feeder groups, and of theresearch of known authors researching COVID-19transmission.

The electronic searches were undertaken inGoogleScholar,usingthesearchstring<“COVID19” “outdoor” transmission >. Experimentalsearches suggested that “COVID 19” with aspace within quotation marks would be themost effective search term, and that addingalternatives for COVID 19 (e.g. SARS-CoV 2,Coronavirus) added little to the efficacy of thesearch. Similarly, adding alternatives for“outdoor”, such as “outside” or “open air”,detracted from rather than improved searchefficacy.

Search results were initially reviewed usingthe article title and preview text containingsearch terms returned by Google Scholar toevaluatewhether a returnedpaper referred tooutdoortransmission.Ifitappearedthatitdid,the full textof thepaperwas searched for theword “outdoor” and the relevant passages oftext were reviewed to establish whether anyevidence, or references to other sources ofevidence, of outdoor transmission wereincluded. Chains of sources referred to inpapers were pursued to the original source orsources of evidence of outdoor transmission,which were also included for evaluation andanalysis. Sources that provided opinion,summation or only onward reference toevidence of outdoor transmission were notincluded. Given the rapidlyemergingevidencerelatingtoCOVID-19,neitherpeer-reviewstatusnor the outlet in which the source waspublished, was used as inclusion/exclusioncriteria.

It was expected that evidence of outdoortransmission would be limited, and while suchevidence was referred to in many sources,sources containing actual evidence werelimited. For this reason, the initial search,orderedbyGoogleScholar’srelevancefunction,was limited to the first 100 sources returnedbecause itwas assumed that the searchwouldbesaturatedatthatpoint,andnonewsourcesof evidence would be added. The pre-agreedmarker of saturation was that the last 20returns (81-100)wouldnotaddfurthersourcesofevidence,andthismarkerwasmet.Hadthismarker not been met, then the search wouldhavebeenextendedtothenext20sourcesuntilthepointwas reached that the last 20 sourcesdidnotaddanyadditionalsourcesofevidence.

The search usingGoogle Scholar’s relevancefunction was undertaken on 10/8/20 andresultedin55ofthe100returnedsourcesbeingevaluated for inclusion. Of these55sources,9were included for joint review by the authorteam,aswellasafurther12fromthepursuitofthe chain of sources referred to in the 55papers.

The search was then repeated, ordered byGoogleScholar’sdateaddedfunction.Thiswasto ensure that new sources of evidence werenotoverlooked.Thesameprotocolasdescribedabove was followed in relation to establishingthe relevanceof the returns, referencemining,and establishing the saturation point for thesearch,whichwasreachedat100returns.

ThesearchusingGoogleScholar’sdateaddedfunction was undertaken on 11/8/20, andresultedin28ofthe100returnedsourcesbeingevaluated for inclusion. Of these28sources,2were included for joint review by the authorteam,butnonewereaddedfromthepursuitofthe chain of sources referred to in the 28papers.

InadditiontotheGoogleScholarsearch,thepapersandevidencesourcesconsideredbytheUK government’s Scientific Advisory Group forEmergencies (SAGE), and its feeder groups,including the Scientific Pandemic InfluenzaGroup onModelling (SPI-M) and the New andEmerging Respiratory Virus Threats AdvisoryGroup (NERVTAG), were hand searched forevidence and discussion of outdoortransmission on 13/8/20. Of these papers and

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

4

evidence sources, 8 were included for jointreviewbytheauthorteam.

Finally, specific manual searches of theresearch of known authors researching COVID-19 transmission were undertaken during 10-16/8/20, of which 6 were included for jointreviewbytheauthorteam.

Overall, across all search elements, 37sources were included for joint review by theauthor team. MW initially reviewed andevaluated sources for inclusion on the basis ofrelevanceandwasresponsibleforthedecisionsto include the sources for joint review by theauthor team as set out above. AFindependently reviewed the inclusiondecisionson the basis of relevance to the review. Theauthor teamthendiscussed the inclusions,andthis resulted in the removal of two sources onthebasisofrelevance.

Theauthorteamalsoconcludedthatsomeofthe included sources (14) provided directevidence of incidents of outdoor transmission.However, others (21) did not provide evidenceof incidents of outdoor transmission, but didprovide insights, both from the science oftransmission and the extant literature, thatwould be important in understanding thecontext and caveats that should be consideredin interpreting the review findings.Consequently, the author team decided toinclude only the 14 sources providing directevidenceofoutdoor transmission in the formalreview, but to also develop a separate contextandcaveatsdiscussionthatwouldconsider,butnot be limited to, the remaining 21 sources.Figure 1 provides a summary decision flow ofthesearchprotocolandtheinclusiondecisions.

As evidence of incidents of outdoortransmission of COVID-19 was expected to belimited, search inclusioncriteria relatedonly torelevance – there were no inclusion/exclusioncriteria relatedtoqualityorproxies forquality.However, qualitywas evaluated in the analysisand synthesis of the included sources. Theproduct of this analysis is a critical narrativesynthesis (Pope & Mays, 2006) which, whilstdescribing and synthesising evidence insubstantive terms, also highlights potentialweaknesses in returned evidence throughoutthenarrative.

MW wrote the first draft of the narrativesynthesis, includinganembeddedevaluationofevidencequality.AFreviewedtheevaluationofevidencequalityembedded in the firstdraftofthenarrative synthesis. The author team thendiscussedandagreedtheembeddedevaluationofevidencequality.Evidence of Outdoor TransmissionofCOVID-19:ReviewFindingsWhat is the evidence of incidents of outdoortransmission?The majority of the sources considered forinclusion inthereviewstatedthattransmissionof COVID-19 outdoors is a lower risk thanindoors. However,theevidencebaseforthesestatementscanbetracedbacktojustthreerootsources:astudyof110casesinJapan(Nishiuraet al, 2020), a review of case reports of 7,324casesinChina(Qianetal,2020),andadatabasecompiled at the London School of Hygeine &TropicalMedicine, totalling20,471casesacross616 clusters in the most recent publication(Lakha, Rudge&Holt, 2020), fromwhich therehad also been two earlier publications (Knight,Leclerc& Kucharski, 2020; Leclerc et al, 2020).Ofthese,onlyLeclercetal(2020)hasbeenpeerreviewed.

Both Nishiura et al (2020) and Qian et al(2020) considered cases that largely occurredbeforeanylockdownrestrictionswereimposed.The 110 cases in Japan were before 28/1/20,and Japan’s ‘state of emergency’ was imposedon 9/4/20. In China, the 7,324 cases werebetween 4/1/20 and 11/2/20, but all wereoutsideHubei province,where local lockdownswere variously introduced from 2/2/20. Assuch, the vast majority of cases occurred in‘normal’ life when social interactions wereunrestricted.

Nishiura et al (2020) present a very short,two page, unreviewed paper that does notprovidedetailsofprimarydata,norofhowthedatawas searchedandextracted. It concludesthat the likelihood of transmission in a closedenvironmentis18.7timeshigherthanoutdoors,butitisnotpossibletovalidatethebasisofthisconclusion from the paper. It appears from

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

5

Figure1

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

6

achartprovidedwiththepaperthat,ofthe110cases,11(circa10%)resultedintransmissiontoat least one other person in an outdoorenvironment, but only two of these createdmultiple cases, and none transmitted to morethanthreepeople. Whilstwidelyreferencedinother papers, the findings of this paper shouldbetreatedwithcaution.

Of the 7,324 cases reviewed by Qian et al(2020), only one transmission incident,comprising two cases (a single transmission)tookplaceinanoutdoorenvironment,andwasthe result of a conversation between twoindividuals. The case reports were extractedfrom the health commissions of 320municipalitiesinChina(excludingthoseinHubeiprovince). Although this paper is unreviewed,and the data comprises only case reports,reasonable confidence can be placed in itsfindings, particularly as China has acomprehensivetrackandtracesystem.

The publications from the London School ofHygeine and Tropical Medicine (LSHTM)databasecompriseapeer-reviewedpublication(Leclerc et al, 2020) and a report to SAGE (theUK government’s emergency scientific advisorygroup) (Knight, Leclerc & Kucharski, 2020) inJune, and an unreviewed update published inJuly(Lakha,Rudge&Holt,2020).Thedatabaseis the result of a systematic search of peer-reviewed scientific reports of cases andmediareports of cases. Following an earlierpublicationof theLeclercetal (2020)paperon1st May, the authors sought to crowd sourcereports of further examples of transmission oroutbreaksthrougha‘suggestedupdates’linkontheirpublicly availabledatabase. The resultingupdated publication (Lakha, Rudge & Holt,2020)consideredanestimated20,471reportedcases across 616 clusters, and refined thecategorisation of settings, which resulted infewer settings being considered outdoor, orhaving an outdoor element. For example, inLeclerc et al (2020) and Knight, Leclerc andKucharski (2020) religious settings wereconsidered to have outdoor elements, but inLakha, Rudge and Holt (2020) this had beenrefined to be indoor only. While the LSHTMdatabasecannotbeconsideredcomprehensive,atover20,000casesitisextensive,andthereisnoreasontoassumeanybiasinthereportingof

indoorversusoutdoorcases.Consequently,thefindingscanbetreatedwithsomeconfidence.

Of the20,471cases in theLSHTMdatabase,only 461 in 11 clusters were associated solelywithoutdoorenvironments.Afurther628casesin 34 clusters were associated withenvironmentsthathadsomeoutdoorelements(Lakha,Rudge&Holt,2020). As such,only6%of cases, and 7% of clusters were associatedwith outdoor environments or environmentswith an outdoor element. 179 cases in 19clusters were associated with sportenvironments, and a further 179 cases in 6clusters were associated with parties, but ineach case a significant majority of these wereindoor.100casesin6clusterswereassociatedwithmusicvenues,allofwhichwereindoors.

Lakha, Rudge and Holt (2020) note that,despite lower risk of outdoor transmission,outdoor settings linked to crowding (e.g.marketsandrallies)arelinkedtorelativelylargeclusters (min 25 cases,max 163 cases), and inthese settings people tend to circulate withinthe crowd. Similarly, numerous clusters wereassociated with close range interaction, andloudconversations,shoutingorsinging.Leclercet al (2020) also note that duration in thesecontexts is important. Together this suggeststhatcautionandfurthermitigationislikelytoberequired in relation to gathering density,circulation and size, as well as the duration ofgatherings.Knight,LeclercandKucharski(2020)suggestthatitistheinteractionofenvironment,activityanddurationthatisimportant.

Twoothersourcesreferredtotransmissioninoutdoor environments. One reported anoutbreakatasummercampwhere, inadditionto outoor activities, participants shareddormitories and participated in singing(Szablewski et al, 2020). Another was a rapidreview for consideration at SAGE in April(UNCOVER, 2020), which could find no highquality studiesdirectly addressing thequestionofoutdoorversusindoortransmission. Neitherofthesourcesaddtotheinsightsabove.

Across sources, there is limited evidence oftransmission of COVID-19 in outdoorenvironments during the natural course ofeveryday life. Despite this limited evidencebeingfoundbyonlythreesources,itcollectivelyrelatestoover25,000caseswith,inthecasesof

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

7

Qianetal (2020),Knight, LeclercandKucharski(2020), Leclerc et al (2020) and Lakha, RudgeandHolt(2020),extensivesearcheshavingbeenundertaken. Consequently, it is reasonable toassume that absence of evidence of extensiveoutdoor transmission of COVID-19 in everydaylifecanbetakentobeevidenceofabsenceofameaningfulriskofoutdoortransmission.

However, there is someevidence to suggestthat there is a higher risk of outdoortransmissioninenvironmentswherethenaturalsocial distancing that takes placewhen ‘millingaround’ in everyday life is breached, andgathering density, circulation and size isincreased, particularly where this involves anextended duration. This could include aspectsofoutdoorconcerts,festivalsandsometypesofphysicalactivityandsportingevents.Has weather impacted transmission throughencouragingindoororoutdooractivity?

Five included sources refer to a link betweenweather conditions and transmission, all ofwhichassociatelowertemperatureswithhighertransmission. All sources suggest at least apartial role for a behavioural effect, in whichlowertemperaturesencouragepeopletospendmore time indoors. Four sources presentindependent analyses (Alvarez-Ramireez &Meraz,2020;Carleton&Meng,2020;Corripio&Raso, 2020; Newell, 2020), whilst one(UNCOVER, 2020) comments on the analysespresented by the first two. None of thesesourceshavebeenpeer-reviewed.

Carleton and Meng (2020) correlate globaltemperature data with a sample of 166,686confirmed COVID-19 cases between 22/1/20and15/3/20across134countries.Theirresultsshow a clear correlation between increasedtemperatures and reduced transmission, andthey suggest three mechanisms: first, abiological effect on the virus itself; second, abehavioural effect, where people spend moretime outdoors in less dense interactions; third,anincreaseinco-morbidities.Similarly,Alvarez-Ramireez and Meraz (2020) correlated dailycases in Wuhan, China between 29/1/20 and6/3/20 with temperature trends, lagged for 6daystoaccountfortheincubationperiodofthevirus. This showed an increase in cases at

temperaturesbelow8C(c.1,650perday)andadecrease above 10C (c.350 per day), with thisbeing attributed to human behaviour and an‘indoor crowding effect’. UNCOVER (2000)concludes that the transmission effectsdemonstrated in these two papers areattributabletocooltemperaturesdrivingpeopleindoors.However,afurtherexacerbatingfactormaybethatpeopletendtohaveclosercontactwhengathering indoors thanoutdoors (‘indoorcrowding’).

Corripio and Raso (2020) correlatemeteorological data on temperature andhumidity (lagged by 8 days) with COVID caserecords from the European Centre for DiseaseProtectionandControl(forItaly)andtheCOVID-19 data Repository at John Hopkins University(for the USA) between 1/1/20 and 7/4/20,representing the time up to lockdown. Theirdata for Italy show a decrease in cases attemperaturesover11C,aligningwiththatfoundfor Wuhan (Alvarez-Ramireez & Meraz, 2020).However,theUSAdatawasunclear,althoughavery weak association between temperatureand transmission was found. This led theauthors to conclude that their data wasconsistentwithpooroutdoortransmission,andthatanyeffectwasmorelikelybehaviouralthanbiological.

Perhaps the most interesting evidencerelatedtoweather,whichmayalsoexplain theunclear data from the USA found by Corripioand Raso (2020), is data correlatingtemperatures lagged by one week in six USstates (California, Minnesota, Michigan, Ohio,Illinois,NewYork)witharatioofthegrowthofcases between 10/3/20 to 18/6/20 (duringperiods of which some COVID-19 restrictionsand mitigations were in place) (Newell, 2020).Newell’s (2020) analysis shows that infectionrates in all six states fell distinctly at circa 50F(c.10C) and grew again from circa 70F (c.21C).Newell (2020)alsopresentedtwoyearaveragedata showing that energy consumption ofheating and airconditioning systems is lowestbetween 50F and 70F (c. 10C–21C). Together,thissuggeststhatpeopleswitchofftheirheatingand go outside when temperatures rise above50F(10C),andcomebackinsideagaintoutilisetheirairconditioningwhentheheatrisesabove70F (21C). Newell (2020) concludes that lower

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

8

infectionratesarecorrelatedwithperiodswhenpeople spendmore time outdoors, and itmaybe that the secondeffectat70F (20C)explainsthe lack of clear correlations in Corripio andRaso’s(2020)data.

Similar correlations between temperatureandCOVIDcaseshavebeenfoundineachofthestudies described, in which data sources andcalculationsareclear,andthusgoodconfidencecanbeplacedinthesecorrelations,despitethelackofpeer-review. Italsoappearsreasonableto conclude that the correlations suggest atleast a partial behavioural effect, where timespentoutdoorsleadstoafall incases,andthatthismaybeexacerbatedbyanindoorcrowdingeffect where people tend to gather moredenselyinindoorsettings.

A final possible effect to note is that iftemperatures below 10C encourage people tospend more time indoors, this may havesupressed,althoughlikelynomorethanslightly,the numbers of cases of outdoor transmissionfound in thepapers from Japan (Nishiuraetal,2020) and China (Qian et al, 2020) in theprevious section, both of which experiencedmeteorological winter (December to February)duringtheperiodstudied.WhatistheevidenceforoutdoortransmissionofCOVID-19atmassgatherings?

‘Mass gatherings’, such as Champions Leaguefootballmatches in Italy,SpainandtheUK,theCheltenham Festival (A major UK horse racingevent),anda rockconcert inCardiff (UK),haveall featuredinthemediaaspotentialdriversofCOVID-19 transmission clusters that pre-datelockdowns.Inaddition,aslockdownrestrictionshave eased, media concerns have also beenexpressed about gatherings on beaches onsunnydaysandcrowdsgeneratedbyBlackLivesMatterprotests.However,onlyfoursourcesofevidence relating to outdoor transmission atmass gatheringswere found (Dave et al, 2020;Lazer et al, 2020; Sassano et al, 2020; and theLSHTM database, comprising Knight, Leclerc &Kucharski,2020,Lakha,Rudge&Holt,2020,andLeclercetal,2020),ofwhichtwo(Sassanoetal,2020;Leclercetal,2020)werepeer-reviewed.

Sassano et al (2020) present evidence onCOVID-19 cases and deaths in Bergamo (Italy)

followingtheChampionsLeaguefootballmatchbetween Bergamo’s Atalanta BC (the hometeam) and Valencia CF from Spain on 19/2/20.Although Atalanta were the home team, thegamewasheld35milesawayinMilan’sSanSirostadiumduetoitsgreatercapacity-circa45,000fansattendedthegame,anditisestimatedthatcirca40,000travelledfromBergamo.

TherehadbeennorecordedCOVID-19casesinBergamobeforethematch,butSassanoetal(2020) present data showing that there were1,815 cases three weeks after the game, and8,803casesand2,060deathssixweeksafterthegame. During March 2020, daily deaths inBergamowere568%higherthantheaverageforthe four years previous, compared to 187%higherinthewiderLombardyregion.However,whileitispossiblethattheoccasionofthegamemayhavecontributedtothissignificantlyhighernumber of ‘excess deaths’, attendance at thestadium is unlikely to have done so. The SanSiro is an all seater stadium which, by design,holdseach fan inasingleplacethroughoutthegame,thuspreventingthemfromcirculatingtoanygreatextentwithotherfans. Sassanoetal(2020) conclude that the interactions ofBergamo fans with each other and thoseoutside their home town on transport, and inbars, clubs and other venues, includinggatherings in homes, was likely to have beenmoresignificantforCOVID-19transmissionthananyoutdoortransmissionaspartofthestadiumcrowdduringthegameitself.

Daveetal(2020)andLazeretal(2020)eachaddress the impact of Black Lives Matterprotests in the USA, finding that the protestswere either associated with no effect, or areduction in the growth of COVID-19 cases.Lazer et al (2020) analyse 37,325 responsesfromtwowaves(12-28/6/20and10-26/7/20)ofamonthly COVID-19 omnibus survey across 50states, and show that there was a clear andsignificant negative correlation between thepercentageofastate’spopulationthatreportedprotesting, and subsequent COVID-19 cases.That is, the growth of COVID-19 cases wasslower in those states where more peoplereported protesting. Lazer et al (2020)speculated that other mitigating behavioursmayhavebeenresponsibleforthis,andshowedthat in states where more people reported

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

9

protesting, adherence to mask wearingguidelineswashigher.

AdifferentexplanationisofferedbyDaveetal (2020), who present data for 315 cities ofmore than 100,000 population that shows nodifference in COVID-19 growth rates betweencitiesthatdid(286)anddidnot(29)experienceBlack Lives Matter protests between 26/5/20and7/7/20.

However, Dave et al (2020) also draw onanonymised GPS data from mobile phonerecordstoshowanet increase instayathomebehaviourincitiesthatexperiencedprotests.Itisassumedthatnon-protestorsstayedathometo avoid the perceived risk posed by theprotests, and the data shows that the netimpact was to increase the volume of socialisolation,andthusprevent increasedgrowthofCOVID-19cases,acrossthecities’populationsasa whole. Of course, it is possible that amongprotestorscasesincreased,butthereisnodataavailabletoconfirmorconfoundthis.

The implications of the above in relation tooutdoor transmissionofCOVID-19 is that thereis no evidence that has meaningfully testedoutdoor transmission at mass gatherings, andconsequently an evaluationof evidencequalityissuperfluous.Ineachofthecasesconfoundingfactors are as likely, if not more likely, thanoutdoor transmission to explain upward,downward or neutral trends in COVID-19 casedata.Inaddition,Leclercetal(2020)notethat,while a concert in Cardiff (UK), a horse-racingfestival in Cheltenham (UK) and ChampionsLeague footballmatches in Italy, Spainand theUK that were the subjects of media concerncould potentially have been connected toCOVID-19 clusters, the absence of surveillancesystemsandrigoroustestingmeansthatnodatais available and thus such connections remainspeculation.Thesameistrueofwhatappearedtobeacrowdedgatheringoflocalsandtouristson Boscombe Beach (Bournemouth, UK) on25/6/20, during which Bournemouth Councildeclareda‘majorincident’,askedpeopletostayaway, and requested thatpolice forcesoutsidethe area be put on alert to send re-inforcements.Again,despitemediainterest,notestingorsurveillancedataisavailable.

Itisclear,therefore,thatthereisanabsenceof evidence regarding outdoor transmission of

COVID-19 at mass gatherings. However,outdoor mass gatherings are heterogenous,particularly in relation to crowd density,circulationandsize,which largelycomprise theactivity element in the key interaction ofenvironment, activity and duration that hasbeen suggested to determine risk (Knight,Leclerc&Kucharski,2020).

It is therefore also clear that absence ofevidence in relation tomass gatherings cannotbe assumed to be evidence that there is anabsence of outdoor transmission at massgatherings – this is likely to vary dependingonthe features and characteristics of differentmassgatherings.

ContextandCaveats:ExtantScienceandLiterature

Multiple sources exploring the science of

transmission conclude that the risk oftransmission of COVID-19 is low outdoors(Contini&Constable,2020;Dominski&Brandt,2020; TWEG, 2020; Redacted Author, 2020).Others conclude that aerosol transmission islikely to play a negligible role outdoors (AlHuraimel et al, 2020; EMG/NERVTAG, 2020),andwith the riskof surface transmissionbeingclearly identifiable from the featuresof venues(as well as being easily mitigatable), it istransmissionby respiratorydroplets that is themajorareaofconcernforoutdoortransmission(EMG,2020a,2020b).

InassessingtheriskofoutdoortransmissiononCOVID-19aslow,thescienceoftransmissiongenerallyassumesthatthisriskisinthecontextof social distancing at or around 2m (ECDPC,2020; EMG, 2020a; TWEG, 2020; NERVTAG,2020), facilitated either by specific mitigationmeasures (Chu et al, 2020), or by the normalconventions of personal space and the moreopenoutdoorenvironmentthatgenerallykeepsgroups of people further apart than indoors(Alvarez-Ramireez and Meraz, 2020; Slater,Christiana&Gustat,2020).Iftheseconventionsarebreached, then theextentof thebreach interms of proximity, duration, and circulation issignificant. In this respect, TWEG (2020)conclude that the risk of outdoor droplettransmission in close, face-to-face contact in

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

10

crowded areas is likely to be similar to that insomeindoorsettings.

Various studies suggest atmosphericconditions, such as air pollution, temperatureand humidity (Contini & Constable, 2020;Sanchez-Lorenzoetal,2020;Zoranetal,2020),haveapotentialbiological impactonCOVID-19transmission.However,asthesestudiesdidnotconsider any explanation other than biologicalmechanisms,wedonotbelievetheycontradictour review conclusions that there is at least apartialbehaviouralexplanationrelatedtolowerCOVID-19 transmission in temperatures whenoutdooractivityincreases.

Reviews relating to the impact of massgatherings on infectious disease transmission(McCloskeyetal,2020;Nunan&Brassey,2020;Rainey,Phelps&Shi,2016),which largelydrawon pre-COVID-19 evidence, generally concludethat elevated risk comes with longer duration,crowdedness (particularly in communalovernight accommodation), and indoorenvironments. In terms of restrictions, NunanandBrassey(2020)suggestthatrestrictingmassgatherings closer to theepidemicpeakmaybemoreeffective than restrictionsapplied furtherout. However, McCloskey et al (2020) andNunanandBrassey(2020)eachemphasisethat,for COVID-19 transmission in particular, massgatheringsarenothomogenous,andrisksmustbeassessedonacase-bycasebasis.

Finally, some studies have explored thespecificimpactofparticularactivities.Followinganumberof reportedoutbreaksamongchoirs,NERVTAG(2020)concludedthattheproductionof dropets during singing may be akin to acough,althoughtheynotedthereisnoevidencedescribing the distance travelled by dropletsduring singing. They did recommend, though,thatparticularcautionshouldbeexercisedwiththeenvironment(outdoorpreferred),distancing(2mpreferred),sizeanddensity(smallergroupspreferred), arrangement (side-to-side ratherthan face-to-face) and duration (shorterpreferred)when singing. These cautions applynot just to choirs, but to football crowds, andcarnival, concert and festival go-ers, aswell asto any other setting inwhich singingmay takeplace.

In addition, some authors have speculatedthat exertion, such as that during physical

activity and sport, may be similar to singing(Arias, 2020), while others have suggestedsimilarcautions(Dominski&Brandt,2020).But,while this seems intuitive, there are as yetneither incidents nor modelling studies thatprovideevidenceforthis.Similarly,asimulationstudy (Broken et al, 2020) has suggested aslipstreameffect,inwhichwalkers,runnersandcyclists positioned directly behind the bodywidthofanothermayneedtodistancefor5m,10m and 20m respectively, ormove slightly tothe side, to avoid droplets. However, whenSAGE considered evidence of extended carrydue to downwind flow conditions (RedactedAuthor, 2020) they concluded that this wasunlikely to be a significant route for infectionunlesspeopleareinpositionforalongperiodoftime and presumably unable to move. Thissuggests that the general considerations forduration and density set out in other contextsarethekeymitigationsthatshouldapply.

These insights from the science oftransmissionandtheextantliteraturealignwithour review evidence that the risk of outdoortransmission is low, unless the natural socialdistancing that takes place when ‘millingaround’ in everyday life is breached. In suchcircumstances,cautionandfurthermitigationislikely to be required in relation to gatheringdensity, circulation, size and duration ofcontact(s). Knight, Leclerc & Kucharski (2020)suggest that the interaction of environment,activity and duration is a useful framework toassess risk and relevant mitigations which,importantly, will not be homogenous acrossdifferent outdoor activities, gatherings, eventsand environments. However, outdoor eventsand activities are heterogenous, particularly inrelation to gathering density, circulation andsize,whichlargelycomprisetheactivityelementin the key interaction of environment, activityand duration that has been suggested todeterminerisk.Considerations for Events andActivities that Generate OutdoorGatherings

The World Health Organisation describes fourtransmission scenarios for COVID-19: no cases,

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

11

sporadic cases, clusters of cases, andcommunity transmission (WHO, 2020). Whencountriesorareasenterlockdowns,itisbecausecases have progressed to communitytransmission,wherelargenumbersofcasesnotlinkabletotransmissionchainsexist. However,as countries or areas exit lockdowns, it isassumedtobebecausecommunitytransmissionhasbeenbroughtundercontrol,andthatcasescan now be traced to transmission chains orclusters. In this scenario, given that achievingnocasesisunlikelywithoutavaccine,continuedisolated sporadic cases of transmission areinevitable, and so the key goal is to preventtransmission through clusters of cases focusedinaparticularlocationand/oractivity.Itisthislatter risk that should be considered by hostsand organisers of events and activities thatgenerateoutdoorgatherings.

A framework for assessing this risk thatfocuses on the environment, activity andduration of a gathering has been suggested byKnight, LeclercandKucharski (2020). Both thisreview and the extant science and literatureagree that an outdoor environment presents alowriskduetothenaturalsocialdistancingthathappens through the normal conventions ofpersonal space in everyday life, as well as thewideropenspaceoftheoutdoorsthatgenerallykeeps groups of people further apart thanindoors.Consequently,thekeyriskstofocusonare those related to the activities and theirduration that take place at outdoor gatheringswhere this natural social distancing might bebreached.

In both the review and the extant scienceand literature, density, circulation, size andduration of the gathering generated by theactivity or event have been highlighted as keyriskfactors.

The density of the gathering relates to thecloseness of contact between individuals, andincludes theirarrangement (e.g. face-to-faceorside-by-side). NHS Track and Trace defines‘close contact’1 as spending more than 151 https://www.gov.uk/government/publications/guidance-for-contacts-of-people-with-possible-or-confirmed-coronavirus-covid-19-infection-who-do-not-live-with-the-person/guidance-for-contacts-of-people-with-possible-or-confirmed-coronavirus-covid-19-infection-who-do-not-live-with-the-person (accessed 30/8/20)

minuteswithintwometresofsomeonewhohastestedpositiveforCOVID-19,orhavingface-to-face contact within one metre, or having skincontact, or being within one metre of aninfected person for more than one minutewithoutface-to-facecontact.

Circulationwithinthegatheringreferstohowfar individuals move around and bringthemselvesintocontactwithalargernumberofother people. For example, in an all-seaterfootball stadium or at the start of a runningrace, people ususallymaintain their position inrelation to those around them: at a festival orcarnival, people oftenmove aroundwithin thegathering and come into contact with a largernumberofpeople.

The size of the gathering relates to thelikelihood that infected individuals or groupsmight be present. Neither the review nor theexplorationofextantscienceand literaturehasconsidered factors such as the underlyinginfection rate in a community, or case fatalityrates, which were beyond the scope of thereview. However, the baseline underlying riskwill be lower in countries and communitieswhere the underlying rate of infection is low,and the risk of the size of gathering should beconsideredinrelationtothis.

Thedurationofthegatheringrelatesbothtothe gathering as a whole (e.g. a multi-dayfestival; an afternoon football match) and toindividual elements of it (e.g. standing in acrowd, queing at the bar, visiting the toilet;before,atthestart,during,andafterarunningrace).

No one risk factor presents an inherentlylarger risk than any other, and all of the riskfactors mitigate each other. For example: amoredensegathering ismitigated ifcirculationor duration is low; a larger gathering ismitigated if it is less dense, or if less time isspent in it; a longer duration is mitigated atlower levels of gathering density, size orcirculation;andsoon. Importantly,riskfactorsmustbeconsideredinrelationtothesizeoftheunderlying risk, comprising elements such asinfectionrates in thecommunity, theextent towhichvulnerableorsusceptiblegroupsmightbepresent, and the indoor or outdoor nature ofthe gathering. If the density of an outdoorgathering is such that the natural social

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

12

distancing that happens through the normalconventionsofpersonalspaceineverydaylifeisnotbreached,thentheunderlyingriskislargelymitigated. However, if density is increasedbeyond the norms of natural social distancingand personal space, then risk will need to bemitigated across the other risk factors ofcirculation,sizeand/orduration. Riskmightbemitigatedextensivelybyoneortworiskfactors,or moderately across three or four, either tooffset any increases in other risk factors, or tomitigatealargerunderlyingrisk.Ultimately,thetask for hosts and organisers is to achieve abalanceacrossriskfactorsthatwillmitigatetheunderlyingrisk,andavoidaclusteroutbreak.

Complicatingmattersfurther,theriskfactorsshouldbeconsideredinrelationtoeachaspectof an event or activity that has differentelements. Risk should be considered for eachindividualaspect,andinaggregatefortheeventoractivityasawhole.

Inadditiontotheabove,therearetwootherfactors that shouldbeconsideredbyhostsandorganisersofeventsandactivitiesthatgenerateoutdoor gatherings of people. Firstly, thereview highighted the potential for events andactivitiestopromptotherbehavioursthatmightbe higher risk than the gathering itself. Theseinclude communal travel to the event oractivity, indoor congregation in bars, cafes orother venues, and collective stays in overnightaccommodation. Secondly, a keymitigation topreventclusteroutbreaksgeneratingsecondarycases (i.e. subsequent transmission away fromtheeventoractivity)andheighteningtheriskofelevationtocommunitytransmission,willbetheabilitytorapidlytraceattendeesandcontactsatthe event or activity. These issues areintegrated with the discussion of risk factorsabove,andsummarisedinFigure2.

The considerations presented here relate tohosting or organising an event or activity.However, therearea threeotherwider factorsthatmight be considered. Firstly, perhaps thelargest risks from gatherings come fromspontaneousorunregulatedeventsoractivitieswhichdonotconsideranyoftheissuesoutlinedin this report (WHO, 2020). While attemptscouldbemadetobanorregulatesucheventsoractivities, providing gatherings that have beenproperly risk assessed would be a mitigating

factor against spontaneous or unregulatedgatherings. Secondly, and related, given thatmanyeventsandactivitieshavecultural,social,economicandevenpoliticalimplications(WHO,2020), the opportunity cost of not hosting ororganising the activity or event should beconsidered. This could be a rise in riskierunregulated activity, or a fall in physical andmental health and wellbeing that could havewidermorbidityandmortalityimpacts.Thirdly,the public appetite or aversion for events andactivities that generate outdoor gatheringsshould be considered, as should the mediacoveragethatdrivesthis.Thiswillalsorelatetowhatothereventsandactivitiesareandarenottaking place. Honey-Roses et al (2020) andSlater et al (2020) each address aspectsof thisissue,withHoney-Rosesetal(2020)notingthatthere may be a longer term impact on ourperceptions of outdoor spaces, their carryingcapacity, and their design. For example, willpublic perceptions ofwhat is crowded change,and will health criteria (both to preventinfection and promote healthy behaviours andactivities–e.g.wideningfootpathsandtrails)bemainstreamed into the design of outdoorspaces? And will such infrastructural changesbeexpectedaspartofthecontinuedre-openingofsociety,orinthelongerterm?

ConclusionandLimitations

The protocol for this review was designed toallowrapidcompletionin15days.Thisisclearlya limitation. However, the review was alsodesignedtoscopetheavailableevidence.Bothelectronic searches were saturated within 100returns,andthesamerootsourcesofevidenceunderpinned the majority of sources reviewedandevaluatedforinclusion. Therefore,whileitis possible that individual sources of relevantevidence may not have been captured by thereview, we consider it highly unlikely that anyfurther significant bodies of evidence relevanttothisreviewwereavailablewhenthesearchesconcluded on 16/8/20. A more substantialreview, with a more detailed and extensivesearch strategy, is therefore unlikely to be aproductiveenterprise.

Available evidence of outdoor transmission

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

13

Figure2

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

14

of COVID-19 has been reviewed, and thecontext and caveats provided by the extantscienceandliteratureconsidered.Thisleadstothe conclusion that the outdoor environmentpresentsalowriskoftransmissionofCOVID-19due to the natural social distancing thathappens through the normal conventions ofpersonal space in everyday life. However, theareas in which risk increaseswhen the normalconventions of personal space are breachedinoutdoor environments have also beendiscussed and outlined, and these have beentranslated into considerations for hosts andorganisersofeventsandactivitiesthatgenerateoutdoorduetothenaturalsocialdistancingthathappens through the normal conventions ofpersonal space in everyday life. However, theareas in which risk increaseswhen the normalconventions of personal space are breached inoutdoorenvironmentshavealsobeendiscussedand outlined, and these have been translatedinto considerations for hosts and organisers ofevents and activities that generate outdoorgatherings, and discussed how risks can bebalanced. It is, of course, the balance of suchrisks that must be considered as, almost bydefinition, no gathering (even that withinhouseholds)canbelowriskinallareas.ConflictofInterestsMW is a member of the parkrun InternationalResearchBoard,andregularlytakespartinparkrun.FundingThisworkisfundedbyparkrun.References*=papersincludedinthereview(14)**=paperssourcedfromreviewsearchthatinformcontextandcaveats(21)

**Ahammer,A.,Halla,M.&Lackner,M.(2020).

MassgatheringcontributedtoearlyCOVID-19spread:EvidencefromUSsports.CovidEconomics,30,44-62.

**AlHuraimel,K.,Alhosani,M.,Kunhabdulla,S.&Stietiya,M.H.(2020).SARS-CoV-2intheenvironment:Modesoftransmission,earlydetectionandpotentialroleofpollutions.ScienceoftheTotalEnvironment.DOI:10.1016/j.scitotenv.2020.140946.

*Alvarez-Ramirez,J.&Meraz,M.(2020).RoleofmeteorologicaltemperatureandrelativehumidityintheJanuary-February2020propagationof2019-nCoVinWuhan,China.medRxivpreprint.DOI:10.1101/2020.03.19.20039164.

**Arias,F.J.(2020).AreRunnersMorePronetoBecomeInfectedwithCOVID‐19?AnApproachfromtheRaindropCollisionalModel.JournalofScienceinSportandExercise.DOI:10.1007/s42978-020-00071-4

**Blocken,B.,Malizia,F.,vanDruenen,T.&Marchal,T.(2020).TowardsaerodynamicallyequivalentCOVID191.5msocialdistancingforwalkingandrunning.Retrievedfromhttp://www.urbanphysics.net/Social%20Distancing%20v20_White_Paper.pdf(accessed10/8/20).

*Carleton,T.&Meng,K.C.(2020).CausalempiricalestimatessuggestCOVID-19transmissionratesarehighlyseasonal.medRxivpreprint.DOI:10.1101/2020.03.26.20044420.

**Chu,D.K.,Duda,S.,Solo,K.etal.(2020).Physicaldistancing,facemasks,andeyeprotectiontopreventperson-to-persontransmissionofSARS-CoV-2andCOVID-19:asystematicreviewandmeta-analysis.TheLancet,395:1973-87.

**Contini,D.&Costabile,F.(2020).DoesAirPollutionInfluenceCOVID-19Outbreaks?Atmosphere.DOI:10.3390/atmos11040377.

*Corripio,J.G.&Raso,L.(2020).WeathervariablesimpactonCOVID-19incidence.medRxivpreprint.DOI:10.1101/2020.06.08.20125377.

*Dave,D.M.,Friedson,A.I.,Matsuzawa,K.,Sabia,J.J.&Safford,S.(2020).BlackLivesMatterprotests,SocialDistancing,andCOVID-19.NationalBureauofEconomicResearch:WorkingPaper27408.Retrievedfromhttp://www.nber.org/papers/w27408(accessed10/8/20).

**Dominski,F.H.&Brandt,R.(2020).DothebenefitsofexerciseinindoorandoutdoorenvironmentsduringtheCOVID‐19pandemicoutweightherisksofinfection?SportSciencesforHealth.DOI:10.1007/s11332-020-00673-z.

**ECDPC(EuropeanCentreforDiseasePreventionandControl).(2020).TransmissionofCOVID-19.Retrievedfromhttps://www.ecdc.europa.eu/en/covid-19/latest-evidence/transmission(accessed11/8/20).

**EMG(EnvironmentalandModellingGroup).(2020a).EnvironmentalInfluenceonTransmission.ReporttoSAGE.Retrievedfromhttps://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/887618/EMG_Environmental_transmission-_02052020__1_.pdf(accessed13/8/20)

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

15

**EMG(EnvironmentalandModellingGroup).(2020b).TransmissionofSARS-CoV-2andMitigatingMeasures.ReporttoSAGE.Retrievedfromhttps://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/892043/S0484_Transmission_of_SARS-CoV-2_and_Mitigating_Measures.pdf(accessed13/8/20)

**EMG/NERVTAG(EnvironmentalandModellingGroup/NewandEmergingRespiratoryVirusThreatsAdvisoryGroup).(2020).RoleofAerosolTransmissioninCOVID-19.ReporttoSAGE.Retrievedfromhttps://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/907587/s0643-nervtag-emg-role-aerosol-transmission-covid-19-sage-48.pdf(accessed13/8/20)

Ghebreyesus,T.A.(2020).WHODirector-General'sopeningremarksatthemediabriefingonCOVID-19(11March2020).Retrievedfromhttps://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---11-march-2020(accessed29/6/20)

**Honey-Roses,J.,Anguelovski,I.,Chireh,V.K.etal.(2020).TheimpactofCOVID-19onpublicspace:anearlyreviewoftheemergingquestions–design,perceptionsandinequities.Cities&Health.DOI:10.1080/23748834.2020.1780074.

*Knight,G.,Leclerc,Q.&Kucharski,A.(2020).AnalysisofSARS-CoV-2transmissionclustersandsuperspreadingevents.ReporttoSAGE.Retreivedfromhttps://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/897562/S0473_SPI-M_Superspreading_and_clusters.pdf(accessed13/8/20)

*Lakha,F.,Rudge,J.W.&Holt,H.(2020).RapidsynthesisofevidenceonsettingswhichhavebeenassociatedwithSARS-CoV-2transmissionclusters.Retrievedfromhttps://superspreadingdatabase.github.io/Evidence_on_clusters_final.pdf(accessed10/8/20).

*Lazer,D.,Santillana,M.,Perlis,R.H.etal.(2020).TheStateoftheNation:A50-stateCOVID-19Survey–Report#10:ThePandemicandtheProtests.COVID-19ConsortiumforUnderstandingthePublic’sPolicyPreferencesAcrossStates.Retrievedfromhttps://kateto.net/covid19/COVID19%20CONSORTIUM%20REPORT%2010%20PROTEST%20AUGUST%202020.pdf(accessed12/8/20).

*Leclerc.,Q.,Fuller,N.M.,Knight,L.E.,Funk,S.&Knight,G.M.(2020).WhatsettingshavebeenlinkedtoSARS-CoV-2transmissionclusters?WellcomeOpenResearch,5(38).DOI:10.12688/wellcomeopenres.15889.1.

**McCloskey,B.,Zumla,A.,Ippolito,G.etal.(2020).MassgatheringeventsandreducingfurtherglobalspreadofCOVID-19:apoliticalandpublichealthdilemma.TheLancet,395,1096-8.

**NERVTAG(NewandEmergingRespiratoryVirusThreatsAdvisoryGroup).(2020).AssessmentoftransmissionofCOVID-19insingingandmusicevents.ReporttoSAGE.Retrievedfromhttps://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/905062/S0628_NERVTAG_Assessment_of_transmission_of_COVID-19_through_musical_events.pdf(accessed13/8/20)

*Newell,T.(2020).LinkingOutdoorAirTemperatureandSARS-CoV-2TransmissionintheUSUsingaTwoParameterTransmissionModel.medRxivpreprint.DOI:10.1101/2020.07.20.20158238.

*Nishiura,H.,Oshitani,H.,Kobayashi,T.etal.(2020).Closedenvironmentsfacilitatesecondarytransmissionofcoronavirusdisease2019(COVID-19).medRxivpreprint.DOI:10.1101/2020.02.28.20029272.

**Nunan,D.&Brassey,J.(2020).Whatistheevidenceformassgatheringsduringglobalpandemics?Arapidsummaryofbest-availableevidence.OxfordCOVID-19EvidenceServiceTeamCentreforEvidence-BasedMedicine.Retrievedfromhttps://www.cebm.net/wp-content/uploads/2020/03/Mass-gatherings-and-sporting-events-during-a-pandemic_PDF-template-4.pdf(accessed12/8/20).

Pope,C.,&Mays,N.(2006).Synthesisingqualitativeresearch.InC.Pope&N.Mays(Eds.),Qualitativeresearchinhealthcare(pp.142–152).London:Blackwell.

*Qian,H.,Miao,T.,Liu,L.,Zheng,X.,Luo,D.&Li,Y.(2020).IndoortransmissionofSARS-CoV-2.medRxivpreprint.DOI:10.1101/2020.04.04.20053058.

**Rainey,J.J.,Phelps,T.&Shi,J.(2020).MassGatheringsandRespiratoryDiseaseOutbreaksintheUnitedStates–ShouldWeBeWorried?ResultsfromaSystematicLiteratureReviewandAnalysisoftheNationalOutbreakReportingSystem.PLOSOne,11(8):e0160378.

**RedactedAuthor.(2020).Evidenceofenvironmentaldispersionfordifferentmechanisms,andtherisksandpotentialmitigations/measuresofcontrolwithindifferentenvironmentsfromwhatweknowaboutCOVID-19.EvidenceforSAGE.Retrievedfromhttps://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/887

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint

16

556/SAGE_paper_Apr_2020_Final-redacted.pdf(accessed13/8/20)

**Sanchez-Lorenzo,A.,Vaquero-Martinez,J.,Calbo,J.etal.(2020).AnomalousatmosphericcirculationfavoredthespreadofCOVID-19inEurope.medRxivpreprint.DOI:10.1101/2020.04.25.20079590.

*Sassano,M.,McKee,M.,Ricciardi,W.&Boccia,S.(2020).TransmissionofSARS-CoV-2andOtherInfectionsatLargeSportsGatherings:ASurprisingGapinOurKnowledge.FrontiersinMedicine.DOI:10.3389/fmed.2020.00277.

**Slater,S.J.,Christiana,R.W.&Gustat,J.(2020).RecommendationsforKeepingParksandGreenSpaceAccessibleforMentalandPhysicalHealthDuringCOVID-19andOtherPandemics.PreventingChronicDisease:PublicHealthResearch,PracticeandPolicy,17(E59).

*Szablewski,C.M.,Chang,K.T.,Brown,M.M.etal.(2020).SARS-CoV-2TransmissionandInfectionAmongAttendeesofanOvernightCamp—Georgia,June2020.CentersforDiseaseControlandPreventionMorbidityandMortalityWeeklyReport.Vol.69:31July.

**TWEG(TransmissionofSARS-CoV-2intheWiderEnvironnmentGroup).(2020).EvidenceofWiderEnvironmentalTransmissionofSARS-CoV-2.EvidenceSummaryforSAGE.Retrievedfromhttps://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/899611/s0546-tweg-evidence-wider-environmental-transmission-200612.pdf(accessed13/8/20)

*UNCOVER(UsherNetworkforCOVID-19EvidenceReviews).(2020).Review:WhatistheevidencefortheimportanceofoutdoortransmissionandofindoortransmissionofCOVID-19?ReporttoSAGE.Retrievedfromhttps://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/891989/S0116_Review_-_What_is_the_evidence_for_the_importance_of_outdoor_transmissions_and_of_indoor_transmission_of_Covid-19.pdf(accessed13/8/20)

UKAthletics.(2020).PrinciplesforStagingRoadandMTRunningEventsDuringCovid-19Restrictions.Retrievedfromhttps://www.runbritain.com/shared/get-file.ashx?itemtype=document&id=6438(accessed7/8/20)

HMGovernment.(2020a).Ourplantorebuild:TheUKGovernment’sCOVID-19recoverystrategy.London:HMGovernment.

HMGovernment.(2020b).GuidanceforprovidersofoutdoorfacilitiesonthephasedreturnofsportandrecreationinEngland.London:HMGovernment.

**WHO(WorldHealthOrganisation).(2020).KeyplanningrecommendationsformassgatheringsinthecontextofCOVID-19.Retrievedfromhttps://www.who.int/publications/i/item/10665-332235(accessed12/8/20).

Worldometer.(2020).COVID-19CoronavirusPandemic:TotalDeaths.RetrievedAugust6,2020,fromhttps://www.worldometers.info/coronavirus/(accessed4/9/20)

**Zoran,M.A.,Savastru,R.S.,Savastru,D.M.&Tautan,M.N.(2020).AssessingtherelationshipbetweensurfacelevelsofPM2.5andPM10particulatematterimpactonCOVID-19inMilan,Italy.ScienceoftheTotalEnvironment.DOI:10.1016/j.scitotenv.2020.139825.

. CC-BY 4.0 International licenseIt is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 10, 2020. ; https://doi.org/10.1101/2020.09.04.20188417doi: medRxiv preprint