Content uploaded by Guang Chen
Author content
All content in this area was uploaded by Guang Chen on Jun 05, 2020
Content may be subject to copyright.
Content uploaded by Thomas Piggott
Author content
All content in this area was uploaded by Thomas Piggott on Jun 03, 2020
Content may be subject to copyright.
Available via license: CC BY 3.0
Content may be subject to copyright.
Articles
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
1
Physical distancing, face masks, and eye protection to
prevent person-to-person transmission of SARS-CoV-2 and
COVID-19: a systematic review and meta-analysis
Derek K Chu, Elie A Akl, Stephanie Duda, Karla Solo, Sally Yaacoub, Holger J Schünemann, on behalf of the COVID-19 Systematic Urgent Review
Group Effort (SURGE) study authors*
Summary
Background Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes COVID-19 and is spread person-
to-person through close contact. We aimed to investigate the eects of physical distance, face masks, and eye
protection on virus transmission in health-care and non-health-care (eg, community) settings.
Methods We did a systematic review and meta-analysis to investigate the optimum distance for avoiding person-to-
person virus transmission and to assess the use of face masks and eye protection to prevent transmission of viruses.
We obtained data for SARS-CoV-2 and the betacoronaviruses that cause severe acute respiratory syndrome, and
Middle East respiratory syndrome from 21 standard WHO-specific and COVID-19-specific sources. We searched
these data sources from database inception to May 3, 2020, with no restriction by language, for comparative studies
and for contextual factors of acceptability, feasibility, resource use, and equity. We screened records, extracted data,
and assessed risk of bias in duplicate. We did frequentist and Bayesian meta-analyses and random-eects meta-
regressions. We rated the certainty of evidence according to Cochrane methods and the GRADE approach. This study
is registered with PROSPERO, CRD42020177047.
Findings Our search identified 172 observational studies across 16 countries and six continents, with no randomised
controlled trials and 44 relevant comparative studies in health-care and non-health-care settings (n=25 697 patients).
Transmission of viruses was lower with physical distancing of 1 m or more, compared with a distance of less than 1 m
(n=10 736, pooled adjusted odds ratio [aOR] 0·18, 95% CI 0·09 to 0·38; risk dierence [RD] –10·2%, 95% CI
–11·5 to –7·5; moderate certainty); protection was increased as distance was lengthened (change in relative risk
[RR] 2·02 per m; pinteraction=0·041; moderate certainty). Face mask use could result in a large reduction in risk of
infection (n=2647; aOR 0·15, 95% CI 0·07 to 0·34, RD –14·3%, –15·9 to –10·7; low certainty), with stronger
associations with N95 or similar respirators compared with disposable surgical masks or similar (eg, reusable
12–16-layer cotton masks; pinteraction=0·090; posterior probability >95%, low certainty). Eye protection also was associated
with less infection (n=3713; aOR 0·22, 95% CI 0·12 to 0·39, RD –10·6%, 95% CI –12·5 to –7·7; low certainty).
Unadjusted studies and subgroup and sensitivity analyses showed similar findings.
Interpretation The findings of this systematic review and meta-analysis support physical distancing of 1 m or more
and provide quantitative estimates for models and contact tracing to inform policy. Optimum use of face masks,
respirators, and eye protection in public and health-care settings should be informed by these findings and contextual
factors. Robust randomised trials are needed to better inform the evidence for these interventions, but this systematic
appraisal of currently best available evidence might inform interim guidance.
Funding World Health Organization.
Copyright © 2020 World Health Organization. Published by Elsevier Ltd. This is an Open Access article published
under the CC BY 3.0 IGO license which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited. In any use of this article, there should be no suggestion that WHO
endorses any specific organisation, products or services. The use of the WHO logo is not permitted. This notice
should be preserved along with the article’s original URL.
Introduction
As of May 28, 2020, severe acute respiratory syndrome
coronavirus 2 (SARS-CoV-2) has infected more than
5·85 million individuals worldwide and caused more than
359 000 deaths.1 Emergency lockdowns have been initiated
in countries across the globe, and the eect on health,
wellbeing, business, and other aspects of daily life are felt
throughout societies and by individuals. With no eective
pharmacological interventions or vaccine available in
the imminent future, reducing the rate of infection
(ie, flattening the curve) is a priority, and prevention of
infection is the best approach to achieve this aim.
SARS-CoV-2 spreads person-to-person through close
contact and causes COVID-19. It has not been solved if
Published Online
June 1, 2020
https://doi.org/10.1016/
S0140-6736(20)31142-9
See Online/Comment
https://doi.org/10.1016/
S0140-6736(20)31183-1
*Study authors are listed in the
appendix and at the end of the
Article
Department of Health Research
Methods, Evidence and Impact
(D K Chu MD, S Duda MSc,
K Solo MSc, Prof E A Akl MD,
Prof H J Schünemann MD),
and Department of Medicine
(D K Chu, Prof H J Schünemann),
McMaster University,
Hamilton, ON, Canada;
The Research Institute of
St Joe’s Hamilton, Hamilton,
ON, Canada (D K Chu);
Department of Internal
Medicine (Prof E A Akl), and
Clinical Research Institute
(Prof E A Akl, S Yaacoub MPH),
American University of Beirut,
Beirut, Lebanon; and
Michael G DeGroote Cochrane
Canada and GRADE Centres,
Hamilton, ON, Canada
(Prof H J Schünemann)
Correspondence to:
Prof Holger J Schünemann,
Michael G DeGroote Cochrane
Canada and McMaster GRADE
Centres, McMaster University,
Hamilton, ON L8N 3Z5, Canada
schuneh@mcmaster.ca
See Online for appendix
Articles
2
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
SARS-CoV-2 might spread through aerosols from
respiratory droplets; so far, air sampling has found virus
RNA in some studies2–4 but not in others.5–8 However,
finding RNA virus is not necessarily indicative of repli-
cation-competent and infection-competent (viable) virus
that could be transmissible. The distance from a patient
that the virus is infective, and the optimum person-to-
person physical distance, is uncertain. For the currently
foreseeable future (ie, until a safe and eective vaccine or
treatment becomes avail able), COVID-19 prevention will
con tinue to rely on non-pharmaceutical interventions,
including pandemic mitigation in community settings.9
Thus, quantitative assessment of physical distancing is
relevant to inform safe interaction and care of patients
with SARS-CoV-2 in both health-care and non-health-care
settings. The definition of close contact or potentially
exposed helps to risk stratify, contact trace, and develop
guidance docu ments, but these definitions dier around
the globe.
To contain widespread infection and to reduce
morbidity and mortality among health-care workers
and others in contact with potentially infected people,
jurisdictions have issued conflicting advice about
physical or social distancing. Use of face masks with or
Research in context
Evidence before this study
We searched 21 databases and resources from inception to
May 3, 2020, with no restriction by language, for studies of any
design evaluating physical distancing, face masks, and eye
protection to prevent transmission of the viruses that cause
COVID-19 and related diseases (eg, severe acute respiratory
syndrome [SARS] and Middle East respiratory syndrome
[MERS]) between infected individuals and people close to them
(eg, household members, caregivers, and health-care workers).
Previous related meta-analyses have focused on randomised
trials and reported imprecise data for common respiratory
viruses such as seasonal influenza, rather than the pandemic and
epidemic betacoronaviruses causative of COVID-19 (severe
acute respiratory syndrome coronavirus 2 [SARS-CoV-2]),
SARS (SARS-CoV), or MERS (MERS-CoV). Other meta-analyses
have focused on interventions in the health-care setting and
have not included non-health-care (eg, community) settings.
Our search did not retrieve any systematic review of information
on physical distancing, face masks, or eye protection to prevent
transmission of SARS-CoV-2, SARS-CoV, and MERS-CoV.
Added value of this study
We did a systematic review of 172 observational studies in
health-care and non-health-care settings across 16 countries and
six continents; 44 comparative studies were included in a
meta-analysis, including 25 697 patients with COVID-19, SARS,
or MERS. Our findings are, to the best of our knowledge, the first
to rapidly synthesise all direct information on COVID-19 and,
therefore, provide the best available evidence to inform optimum
use of three common and simple interventions to help reduce the
rate of infection and inform non-pharmaceutical interventions,
including pandemic mitigation in non-health-care settings.
Physical distancing of 1 m or more was associated with a much
lower risk of infection, as was use of face masks (including
N95 respirators or similar and surgical or similar masks
[eg, 12–16-layer cotton or gauze masks]) and eye protection
(eg, goggles or face shields). Added benefits are likely with even
larger physical distances (eg, 2 m or more based on modelling)
and might be present with N95 or similar respirators versus
medical masks or similar. Across 24 studies in health-care and
non-health-care settings of contextual factors to consider when
formulating recommendations, most stakeholders found these
personal protection strategies acceptable, feasible, and reassuring
but noted harms and contextual challenges, including frequent
discomfort and facial skin breakdown, high resource use linked
with the potential to decrease equity, increased difficulty
communicating clearly, and perceived reduced empathy of care
providers by those they were caring for.
Implications of all the available evidence
In view of inconsistent guidelines by various organisations
based on limited information, our findings provide some
clarification and have implications for multiple stakeholders.
The risk for infection is highly dependent on distance to the
individual infected and the type of face mask and eye
protection worn. From a policy and public health perspective,
current policies of at least 1 m physical distancing seem to be
strongly associated with a large protective effect, and distances
of 2 m could be more effective. These data could also facilitate
harmonisation of the definition of exposed (eg, within 2 m),
which has implications for contact tracing. The quantitative
estimates provided here should inform disease-modelling
studies, which are important for planning pandemic response
efforts. Policy makers around the world should strive to
promptly and adequately address equity implications for
groups with currently limited access to face masks and eye
protection. For health-care workers and administrators,
our findings suggest that N95 respirators might be more
strongly associated with protection from viral transmission
than surgical masks. Both N95 and surgical masks have a
stronger association with protection compared with
single-layer masks. Eye protection might also add substantial
protection. For the general public, evidence shows that physical
distancing of more than 1 m is highly effective and that face
masks are associated with protection, even in non-health-care
settings, with either disposable surgical masks or reusable
12–16-layer cotton ones, although much of this evidence was
on mask use within households and among contacts of cases.
Eye protection is typically underconsidered and can be effective
in community settings. However, no intervention, even when
properly used, was associated with complete protection from
infection. Other basic measures (eg, hand hygiene) are still
needed in addition to physical distancing and use of face masks
and eye protection.
Articles
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
3
without eye protection to achieve additional protection is
debated in the mainstream media and by public health
authorities, in particular the use of face masks for the
general population;10 moreover, optimum use of face
masks in health-care settings, which have been used for
decades for infection prevention, is facing challenges
amid personal protective equipment (PPE) shortages.11
Any recommendations about social or physical
distancing, and the use of face masks, should be based on
the best available evidence. Evidence has been reviewed
for other respiratory viral infections, mainly seasonal
influenza,12,13 but no comprehensive review is available of
information on SARS-CoV-2 or related betacoronaviruses
that have caused epidemics, such as severe acute
respiratory syndrome (SARS) or Middle East respiratory
syndrome (MERS). We, therefore, systematically reviewed
the eect of physical distance, face masks, and eye
protection on transmission of SARS-CoV-2, SARS-CoV,
and MERS-CoV.
Methods
Search strategy and selection criteria
To inform WHO guidance documents, on March 25, 2020,
we did a rapid systematic review.14 We created a large
international collaborative and we used Cochrane meth-
ods15 and the GRADE approach.16 We prospectively sub-
mitted the systematic review protocol for registration
on PROSPERO (CRD42020177047; appendix pp 23–29).
We have followed PRISMA17 and MOOSE18 reporting
guidelines (appendix pp 30–33).
From database inception to May 3, 2020, we searched
for studies of any design and in any setting that included
patients with WHO-defined confirmed or probable
COVID-19, SARS, or MERS, and people in close contact
with them, comparing distances between people and
COVID-19 infected patients of 1 m or larger with smaller
distances, with or without a face mask on the patient, or
with or without a face mask, eye protection, or both on
the exposed individual. The aim of our systematic review
was for quantitative assessment to ascertain the physical
distance associated with reduced risk of acquiring
infection when caring for an individual infected with
SARS-CoV-2, SARS-CoV, or MERS-CoV. Our definition of
face masks included surgical masks and N95 respirators,
among others; eye protection included visors, faceshields,
and goggles, among others.
We searched (up to March 26, 2020) MEDLINE (using
the Ovid platform), PubMed, Embase, CINAHL (using
the Ovid platform), the Cochrane Library, COVID-19 Open
Research Dataset Challenge, COVID-19 Research
Database (WHO), Epistemonikos (for relevant systematic
reviews addressing MERS and SARS, and its COVID-19
Living Overview of the Evidence platform), EPPI Centre
living systematic map of the evidence, ClinicalTrials.gov,
WHO International Clinical Trials Registry Platform,
relevant documents on the websites of governmental and
other relevant organisations, reference lists of included
papers, and relevant systematic reviews.19,20 We hand-
searched (up to May 3, 2020) preprint servers (bioRxiv,
medRxiv, and Social Science Research Network First
Look) and coronavirus resource centres of The Lancet,
JAMA, and N Engl J Med (appendix pp 3–5). We did not
limit our search by language. We initially could not obtain
three full texts for evaluation, but we obtained them
through interlibrary loan or contacting a study author. We
did not restrict our search to any quantitative cuto for
distance.
Data collection
We screened titles and abstracts, reviewed full texts,
extracted data, and assessed risk of bias by two authors
and independently, using standardised prepiloted forms
(Covidence; Veritas Health Innovation, Melbourne, VIC,
Australia), and we cross-checked screening results using
artificial intelligence (Evidence Prime, Hamilton, ON,
Canada). We resolved disagreements by consensus. We
extracted data for study identifier, study design, setting,
population characteristics, intervention and comparator
characteristics, quantitative outcomes, source of funding
Figure 1: Study selection
10
222 records identified through additional sources
8859 COVID-19 specific databases
870 clinical trials registries
9 hand-searching
4 screening references of included studies
480 other
17
678 records identified through traditional
database searching
3314 MEDLINE
975 PubMed
11
115 Embase
567 CINAHL
43 Cochrane Library
1664 Chinese databases
20
013 records after duplicates removed
604 full-text articles assessed for eligibility
19
409 records excluded
172 studies included in systematic review
44 comparative studies included in
meta-analysis
20
013 records screened against title and abstract
432 studies excluded
166 wrong study design (eg, editorial,
narrative review, guideline,
commentary, letter, modelling
without primary clinical data)
118 wrong outcomes
88 wrong or no intervention
52 wrong patient population
6 duplicates
2 news articles
Articles
4
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
Population
size (n)
Country Setting Disease
caused by
virus
Case definition
(WHO)
Adjusted
estimates
Risk of bias*
Alraddadi et al (2016)34 283 Saudi Arabia Health care MERS Confirmed Yes ∗∗∗∗∗∗∗∗
Arwady et al (2016)35 79 Saudi Arabia Non-health care
(household and family
contacts)
MERS Confirmed No ∗∗∗∗∗∗
Bai et al (2020)36 118 China Health care COVID-19 Confirmed No ∗∗∗∗∗
Burke et al (2020)37 338 USA Health care and
non-health care
(including household
and community)
COVID-19 Confirmed No ∗∗∗∗
Caputo et al (2006)38 33 Canada Health care SARS Confirmed No ∗∗∗∗∗
Chen et al (2009)39 758 China Health care SARS Confirmed Yes ∗∗∗∗∗∗∗
Cheng et al (2020)40 226 China Non-health care
(household and family
contacts)
COVID-19 Confirmed No ∗∗∗∗∗∗
Ha et al (2004)42 117 Vietnam Health care SARS Confirmed No ∗∗
Hall et al (2014)43 48 Saudi Arabia Health care MERS Confirmed No ∗∗∗
Heinzerling et al (2020)44 37 USA Health care COVID-19 Confirmed No ∗∗∗∗
Ho et al (2004)45 372 Taiwan Health care SARS Confirmed No ∗∗∗∗∗∗∗∗
Ki et al (2019)47 446 South Korea Health care MERS Confirmed No ∗∗∗∗∗∗
Kim et al (2016)48 9 South Korea Health care MERS Confirmed No ∗∗∗∗∗
Kim et al (2016)49 1169 South Korea Health care MERS Confirmed No ∗∗∗∗∗∗
Lau et al (2004)50 2270 China Non-health care
(households)
SARS Probable Ye s ∗∗∗∗∗∗
Liu et al (2009)51 477 China Health care SARS Confirmed Yes ∗∗∗∗∗
Liu et al (2020)52 20 China Non-health care (close
contacts)
COVID-19 Confirmed No ∗∗∗∗∗∗∗
Loeb et al (2004)53 43 Canada Health care SARS Confirmed No ∗∗
Ma et al (2004)54 426 China Health care SARS Confirmed Yes ∗∗∗∗∗∗∗∗∗
Nishiura et al (2005)55 115 Vietnam Health care SARS Confirmed Yes ∗∗∗∗∗∗∗∗
Nishiyama et al (2008)56 146 Vietnam Health care SARS Confirmed Yes ∗∗∗∗∗∗
Olsen et al (2003)57 304 China Non-health care
(airplane)
SARS Confirmed No ∗∗∗∗∗∗
Park et al (2004)58 110 USA Health care SARS Confirmed No ∗∗∗∗∗∗∗∗∗∗
Park et al (2016)59 80 South Korea Health care MERS Confirmed and
probable
No ∗∗∗
Peck et al (2004)60 26 USA Health care SARS Confirmed No ∗∗∗∗∗∗∗∗∗
Pei et al (2006)61 443 China Health care SARS Confirmed No ∗∗∗∗∗∗∗∗
Rea et al (2007)62 8662 Canada Non-health care
(community contacts)
SARS Probable No ∗∗∗∗
Reuss et al (2014)63 81 Germany Health care MERS Confirmed No ∗∗∗∗∗
Reynolds et al (2006)64 153 Vietnam Health care SARS Confirmed No ∗∗∗
Ryu et al (2019)65 34 South Korea Health care MERS Confirmed No ∗∗∗∗∗∗
Scales et al (2003)66 69 Canada Health care SARS Probable No ∗∗
Seto et al (2003)67 254 China Health care SARS Confirmed Yes ∗∗∗∗∗∗∗∗
Teleman et al (2004)68 86 Singapore Health care SARS Confirmed Yes ∗∗∗∗∗∗∗∗
Tuan et al (2007)69 212 Vietnam Non-health care
(household and
community contacts)
SARS Confirmed Yes ∗∗∗∗∗∗
Van Kerkhove et al
(2019)46
828 Saudi Arabia Non-health care
(dormitory)
MERS Confirmed Yes ∗∗∗∗∗∗∗∗
Wang et al (2020)41 493 China Health care COVID-19 Confirmed Yes ∗∗∗∗
(Table 1 continues on next page)
Articles
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
5
and reported conflicts of interests, ethics approval, study
limitations, and other important comments.
Outcomes
Outcomes of interest were risk of transmission (ie, WHO-
defined confirmed or probable COVID-19, SARS, or
MERS) to people in health-care or non-health-care settings
by those infected; hospitalisation; intensive care unit
admission; death; time to recovery; adverse eects of
interventions; and contextual factors such as acceptability,
feasibility, eect on equity, and resource considerations
related to the interventions of interest. However, data
were only available to analyse intervention eects for
transmission and con textual factors. Consistent with
WHO, studies generally defined confirmed cases with
laboratory confirmation (with or without symptoms) and
probable cases with clinical evidence of the respective
infection (ie, suspected to be infected) but for whom
confirmatory testing either had not yet been done for any
reason or was inconclusive.
Data analysis
Our search did not identify any randomised trials of
COVID-19, SARS, or MERS. We did a meta-analysis of
associations by pooling risk ratios (RRs) or adjusted odds
ratios (aORs) depending on availability of these data from
observational studies, using DerSimonian and Laird ran-
dom-eects models. We adjusted for variables including
age, sex, and severity of source case; these variables were
not the same across studies. Because between-study
heterogeneity can be misleadingly large when quantified
by I² during meta-analysis of obser vational studies,21,22
we used GRADE guidance to assess between-study hetero-
geneity.21 Throughout, we present RRs as unadjusted
estimates and aORs as adjusted estimates.
We used the Newcastle-Ottawa scale to rate risk of bias
for comparative non-randomised studies corresponding
to every study’s design (cohort or case-control).23,24 We
planned to use the Cochrane Risk of Bias tool 2.0 for
randomised trials,25 but our search did not identify any
eligible randomised trials. We synthesised data in both
narrative and tabular formats. We graded the certainty of
evidence using the GRADE approach. We used the
GRADEpro app to rate evidence and present it in GRADE
evidence profiles and summary of findings tables26,27
using standardised terms.28,29
We analysed data for subgroup eects by virus type,
intervention (dierent distances or face mask types), and
setting (health care vs non-health care). Among the studies
assessing physical distancing measures to prevent viral
transmission, the intervention varied (eg, direct physical
contact [0 m], 1 m, or 2 m). We, therefore, analysed
the eect of distance on the size of the associ ations
by random-eects univariate meta-regressions, using
restricted maximum likelihood, and we present mean
eects and 95% CIs. We calculated tests for interaction
using a minimum of 10 000 Monte Carlo random
permutations to avoid spurious findings.30 We formally
assessed the credibility of potential eect-modifiers using
GRADE guidance.21 We did two sensitivity analyses to test
the robustness of our findings. First, we used Bayesian
meta-analyses to reinterpret the included studies
considering priors derived from the eect point estimate
and variance from a meta-analysis of ten randomised
trials evaluating face mask use versus no face mask use to
prevent influenza-like illness in health-care workers.31
Second, we used Bayesian meta-analyses to reinterpret
the ecacy of N95 respirators versus medical masks
on preventing influenza-like illness after seasonal viral
(mostly influenza) infection.13 For these sensitivity
analyses, we used hybrid Metropolis-Hastings and Gibbs
sampling, a 10 000 sample burn-in, 40 000 Markov chain
Monte Carlo samples, and we tested non-informative
and sceptical priors (eg, four time variance)32,33 to inform
n Country Setting Disease
caused by
virus
Case definition
(WHO)
Adjusted
estimates
Risk of bias*
(Continued from previous page)
Wang et al (2020)70 5442 China Health care COVID-19 Confirmed No ∗∗∗∗∗
Wiboonchutikul et al
(2016)71
38 Thailand Health care MERS Confirmed No ∗∗∗∗∗
Wilder-Smith et al
(2005)72
80 Singapore Health care SARS Confirmed No ∗∗∗∗∗∗∗∗
Wong et al (2004)73 66 China Health care SARS Confirmed No ∗∗∗∗∗
Wu et al (2004)74 375 China Non-health care
(community)
SARS Confirmed Yes ∗∗∗∗∗∗∗∗
Yin et al (2004)75 257 China Health care SARS Confirmed Yes ∗∗∗∗∗∗
Yu et al (2005)76 74 China Health care SARS Confirmed No ∗∗∗∗∗∗∗
Yu et al (2007)77 124 wards China Health care SARS Confirmed Yes ∗∗∗∗∗∗∗
Across studies, mean age was 30–60 years. SARS=severe acute respiratory syndrome. MERS=Middle East respiratory syndrome. *The Newcastle-Ottawa Scale was used for
the risk of bias assessment, with more stars equalling lower risk.
Table 1: Characteristics of included comparative studies
For more on the GRADEpro app
see https://www.gradepro.org
Articles
6
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
mean estimates of eect, 95% credibility intervals (CrIs),
and posterior distri butions. We used non-informative
hyperpriors to esti mate statistical heterogeneity. Model
convergence was confirmed in all cases with good mixing
in visual inspection of trace plots, autocorrelation plots,
histo grams, and kernel density estimates in all scenarios.
Parameters were blocked, leading to acceptance of
approximately 50% and eciency greater than 1% in all
cases (typically about 40%). We did analyses using Stata
version 14.3.
Role of the funding source
The funder contributed to defining the scope of the
review but otherwise had no role in study design and
data collection. Data were interpreted and the report
drafted and submitted without funder input, but
according to contractual agreement, the funder provided
review at the time of final publication. The corresponding
author had full access to all data in the study and had
final responsibility for the decision to submit for
publication.
USA
South Korea
USA
USA
China
South Korea
Canada
Singapore
Vietnam
Germany
Canada
South Korea
Vietnam
China
China
Country
Saudi Arabia
Vietnam
China
Canada
USA
China
China
China
Thailand
Saudi Arabia
China
China
China
Taiwan
Saudi Arabia
USA
China
1
1
0
1
0
0
0
1
0
1
0
1
0
0
Respirator
(0=no)
1
0
0
1
0
0
1
1
1
0
0
0
0
1
0
0
0
0
Distance
(m)
0
2
2
1
1
2
0
1
1
2
1
2
0
1·5
1
0
0
2*
0
1
0
2
1
1
1
1*
2
0*
1*
0
1·8
1
0·30 (0·20–0·44)
(Not calculable)
0·23 (0·04–1·20)
0·72 (0·14–3·70)
(Not calculable)
(Not calculable)
0·33 (0·24–0·47)
0·15 (0·03–0·73)
0·59 (0·04–8·77)
0·35 (0·05–2·57)
1·07 (0·49–2·33)
0·22 (0·09–0·54)
(Not calculable)
0·35 (0·23–0·52)
0·08 (0·05–0·14)
0·18 (0·09–0·38)
(Not calculable)
0·23 (0·03–1·57)
0·34 (0·16–0·75)
RR (95% CI)
(Not calculable)
0·23 (0·06–0·89)
0·92 (0·52–1·64)
0·20 (0·01–3·24)
0·04 (0·003–0·68)
0·36 (0·19–0·70)
0·02 (0·001–0·37)
0·48 (0·29–0·81)
(Not calculable)
0·25 (0·04–1·73)
0·04 (0·003–0·76)
0·63 (0·41–0·96)
0·20 (0·01–3·00)
0·18 (0·07–0·50)
0·55 (0·19–1·58)
0·05 (0·02–0·12)
0·97 (0·06–16·14)
0·11 (0·01–1·63)
% weight
(random)
100·0
0
12·9
3·2
0
0
7·1
10·9
1·6
2·6
5·8
5·5
0
76·1
6·6
0
2·7
5·8
0
3·9
6·5
1·6
1·6
6·2
1·5
6·6
0
2·6
1·5
6·9
1·6
5·0
4·8
5·5
1·5
1·7
Events, shorter
distance (n/N)
641/4163
0/76
28/302
4/42
0/37
0/38
136/1124
26/229
5/25
6/19
32/77
17/29
0/69
587/3632
41/647
0/27
26/73
11/35
0/43
6/57
39/341
8/40
2/2
139/382
12/42
13/20
0/22
8/20
2/3
63/445
4/8
43/294
7/36
11/54
3/33
3/3
167/6573
0/50
11/856
2/29
0/41
0/3
39/965
5/248
0/3
1/12
4/9
5/38
0/12
151/5469
18/3493
0/7
1/12
9/84
0/5
3/123
14/133
0/11
0/13
8/61
0/76
17/54
0/16
1/10
0/17
28/314
0/4
4/149
5/47
8/774
0/4
0/4
Events, further
distance (n/N)
Favours further distance Favours shorter distance
0·1 0·5 12 10
Interaction by type of virus p=0·49
Unadjusted estimates, overall (I2=73%)
Burke et al (2020)37
Random, subtotal (I2=75%)
Ki et al (2019)47
Burke et al (2020)37
Peck et al (2004)60
Lau et al (2004)50
Random, subtotal (I2=59%)
Park et al (2016)59
COVID-19
Scales et al (2003)66
Teleman et al (2004)68
Reynolds et al (2006)64
Reuss et al (2014)63
Random, subtotal (I2=75%)
Rea et al (2007)62
Adjusted estimates, overall (1 MERS, 8 SARS)
Ryu et al (2019)65
Nishiyama et al (2008)56
Olsen et al (2003)57
Hall et al (2014)43
Tuan et al (2007)69
Liu et al (2009)51
SARS
Loeb et al (2004)53
Burke et al (2020)37
Pei et al (2006)61
MERS
Bai et al (2020)36
Yu et al (2005)76
Wiboonchutikul et al (2019)71
Arwady et al (2016)35
Liu et al (2020)52
Chen et al (2009)39
Wong et al (2004)73
Ma et al (2004)54
Cheng et al (2020)40
Van Kerkhove et al (2019)46
Heinzerling et al (2020)44
Wong et al (2004)73
0·20 (0·10–0·41)
aOR
aRR
Figure 2: Forest plot showing the association of COVID-19, SARS, or MERS exposure proximity with infection
SARS=severe acute respiratory syndrome. MERS=Middle East respiratory syndrome. RR=relative risk. aOR=adjusted odds ratio. aRR=adjusted relative risk. *Estimated values; sensitivity analyses
excluding these values did not meaningfully alter findings.
Articles
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
7
Results
We identified 172 studies for our systematic review from
16 countries across six continents (figure 1; appendix
pp 6–14, 41–47). Studies were all observational in nature;
no randomised trials were identified of any interventions
that directly addressed the included study populations. Of
the 172 studies, 66 focused on how far a virus can travel by
comparing the association of dierent distances on virus
transmission to people (appendix pp 42–44). Of these
66 studies, five were mechanistic, assessing viral RNA,
virions, or both cultured from the environment of an
infec ted patient (appendix p 45).
44 studies were comparative34–77 and fulfilled criteria for
our meta-analysis (n=25 697; figure 1; table 1). We used
these studies rather than case series and qualitative
studies (appendix pp 41–47) to inform estimates of eect.
30 studies34,37,41–45,47–51,53–56,58–61,64–70,72,74,75 focused on the asso-
ciation between use of various types of face masks and
respirators by health-care workers, patients, or both with
virus transmission. 13 studies34,37–39,47,49,51,54,58,60,61,65,75 addressed
the association of eye protection with virus transmission.
Some direct evidence was available for COVID-19
(64 studies, of which seven were comparative in
design),36,37,40,41,44,52,70 but most studies reported on SARS
(n=55) or MERS (n=25; appendix pp 6–12). Of the
44 comparative studies, 40 included WHO-defined
confirmed cases, one included both confirmed and
probable cases, and the remaining three studies included
probable cases. There was no eect-modification by case-
definition (distance pinteraction=0·41; mask pinteraction=0·46; all
cases for eye protection were confirmed). Most studies
reported on bundled interven tions, including dierent
components of PPE and distancing, which was usually
addressed by statistical adjustment. The included studies
all occurred during recurrent or novel outbreak settings of
COVID-19, SARS, or MERS.
Risk of bias was generally low-to-moderate after
con sidering the observational designs (table 1), but both
within studies and across studies the overall findings
were similar between adjusted and unadjusted estimates.
We did not detect strong evidence of publication bias
in the body of evidence for any intervention (appendix
pp 15–18). As we did not use case series data to inform
estimates of eect of each intervention, we did not
systematically rate risk of bias of these data. Therefore, we
report further only those studies with comparative data.
Studies and
participants
Relative effect
(95% CI)
Anticipated absolute effect (95% CI),
eg, chance of viral infection or
transmission
Difference
(95% CI)
Certainty* What happens (standardised GRADE
terminology)29
Comparison
group
Intervention group
Physical distance
≥1 m vs <1 m
Nine adjusted studies
(n=7782); 29 unadjusted
studies (n=10 736)
aOR 0·18 (0·09 to 0·38);
unadjusted RR 0·30
(95% CI 0·20 to 0·44)
Shorter distance,
12·8%
Further distance,
2·6% (1·3 to 5·3)
–10·2%
(–11·5 to –7·5)
Moderate† A physical distance of more than 1 m
probably results in a large reduction in
virus infection; for every 1 m further
away in distancing, the relative effect
might increase 2·02 times
Face mask vs no face
mask
Ten adjusted studies
(n=2647); 29 unadjusted
studies (n=10 170)
aOR 0·15 (0·07 to 0·34);
unadjusted RR 0·34
(95% CI 0·26 to 0·45)
No face mask,
17·4%
Face mask,
3·1% (1·5 to 6·7)
–14·3%
(–15·9 to –10·7)
Low‡ Medical or surgical face masks might
result in a large reduction in virus
infection; N95 respirators might be
associated with a larger reduction in
risk compared with surgical or similar
masks§
Eye protection
(faceshield, goggles)
vs no eye protection
13 unadjusted studies
(n=3713)
Unadjusted RR 0·34
(0·22 to 0·52)¶
No eye
protection,
16·0%
Eye protection,
5·5% (3·6 to 8·5)
–10·6%
(–12·5 to –7·7)
Low|| Eye protection might result in a large
reduction in virus infection
Table based on GRADE approach.26–29 Population comprised people possibly exposed to individuals infected with SARS-CoV-2, SARS-CoV, or MERS-CoV. Setting was any health-care or non-health-care setting.
Outcomes were infection (laboratory-confirmed or probable) and contextual factors. Risk (95% CI) in intervention group is based on assumed risk in comparison group and relative effect (95% CI) of the
intervention. All studies were non-randomised and evaluated using the Newcastle-Ottawa Scale; some studies had a higher risk of bias than did others but no important difference was noted in sensitivity
analyses excluding studies at higher risk of bias; we did not further rate down for risk of bias. Although there was a high I2 value (which can be exaggerated in non-randomised studies)21 and no overlapping CIs,
point estimates generally exceeded the thresholds for large effects and we did not rate down for inconsistency. We did not rate down for indirectness for the association between distance and infection because
SARS-CoV-2, SARS-CoV, and MERS-CoV all belong to the same family and have each caused epidemics with sufficient similarity; there was also no convincing statistical evidence of effect-modification across
viruses; some studies also used bundled interventions but the studies include only those that provide adjusted estimates. aOR=adjusted odds ratio. RR=relative risk. SARS-CoV-2=severe acute respiratory
syndrome coronavirus 2. SARS-CoV=severe acute respiratory syndrome coronavirus. MERS-CoV=Middle East respiratory syndrome coronavirus. *GRADE category of evidence; high certainty (we are very
confident that the true effect lies close to that of the estimate of the effect); moderate certainty (we are moderately confident in the effect estimate; the true effect is probably close to the estimate, but it is
possibly substantially different); low certainty (our confidence in the effect estimate is limited; the true effect could be substantially different from the estimate of the effect); very low certainty (we have very
little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect). †The effect is very large considering the thresholds set by GRADE, particularly at plausible
levels of baseline risk, which also mitigated concerns about risk of bias; data also suggest a dose–response gradient, with associations increasing from smaller distances to 2 m and beyond, by meta-regression;
we did not rate up for this domain alone but it further supports the decision to rate up in combination with the large effects. ‡The effect was very large, and the certainty of evidence could be rated up, but we
made a conservative decision not to because of some inconsistency and risk of bias; hence, although the effect is qualitatively highly certain, the precise quantitative effect is low certainty. §In a subgroup analysis
comparing N95 respirators with surgical or similar masks (eg, 12–16-layer cotton), the association was more pronounced in the N95 group (aOR 0·04, 95% CI 0·004–0·30) compared with other masks (0·33,
0·17–0·61; pinteraction=0·090); there was also support for effect-modification by formal analysis of subgroup credibility. ¶Two studies54,75 provided adjusted estimates with n=295 in the eye protection group and
n=406 in the group not wearing eye protection; results were similar to the unadjusted estimate (aOR 0·22, 95% CI 0·12–0·39). ||The effect is large considering the thresholds set by GRADE assuming that ORs
translate into similar magnitudes of RR estimates; this mitigates concerns about risk of bias, but we conservatively decided not to rate up for large or very large effects.
Table 2: GRADE summary of findings
Articles
8
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
Across 29 unadjusted and nine adjusted
studies,35–37,39,40,43,44,46,47,50–54,56,57,59–66,68,69,71,73,76 a strong association
was found of proximity of the exposed individual with
the risk of infection (unadjusted n=10 736, RR 0·30,
95% CI 0·20 to 0·44; adjusted n=7782, aOR 0·18, 95% CI
0·09 to 0·38; absolute risk [AR] 12·8% with shorter
distance vs 2·6% with further distance, risk dier ence
[RD] –10·2%, 95% CI –11·5 to –7·5; moderate certainty;
figure 2; table 2; appendix p 16). Although there were
six studies on COVID-19, the association was seen
irrespective of causative virus (pinteraction=0·49), health-care
setting versus non-health-care setting (pinteraction=0·14),
and by type of face mask (pinteraction=0·95; appendix pp 17, 19).
However, dierent studies used dierent distances for
the intervention. By meta-regression, the strength of
association was larger with increasing distance (2·02
change in RR per m, 95% CI 1·08 to 3·76; pinteraction=0·041;
moderate credibility sub group eect; figure 3A; table 2).
AR values with increasing distance given dierent
degrees of baseline risk are shown in figure 3B, with
potential values at 3 m also shown.
Across 29 unadjusted studies and ten adjusted
studies,34,37,41–45,47–51,53–56,58–61,64–70,72,74,75 the use of both N95 or
similar respirators or face masks (eg, disposable surgical
masks or similar reusable 12–16-layer cotton masks) by
those exposed to infected individuals was associated
with a large reduction in risk of infection (unadjusted
n=10 170, RR 0·34, 95% CI 0·26 to 0·45; adjusted studies
n=2647, aOR 0·15, 95% CI 0·07 to 0·34; AR 3·1% with
face mask vs 17·4% with no face mask, RD –14·3%,
95% CI –15·9 to –10·7; low certainty; figure 4; table 2;
appendix pp 16, 18) with stronger associ ations in health-
care settings (RR 0·30, 95% CI 0·22 to 0·41) compared
with non-health-care settings (RR 0·56, 95% CI
0·40 to 0·79; pinteraction=0·049; low-to-moderate credibility
for subgroup eect; figure 4; appendix p 19). When
dierential N95 or similar respirator use, which was
more frequent in health-care settings than in non-
health-care settings, was adjusted for the possibility that
face masks were less eective in non-health-care
settings, the subgroup eect was slightly less credible
(pinteraction=0·11, adjusted for dierential respirator use;
figure 4). Indeed, the association with protection from
infection was more pronounced with N95 or similar
respirators (aOR 0·04, 95% CI 0·004 to 0·30) compared
with other masks (aOR 0·33, 95% CI 0·17 to 0·61;
pinteraction=0·090; moderate credibility subgroup eect;
figure 5). The interaction was also seen when addit-
ionally adjusting for three studies that clearly reported
aerosol-generating procedures (pinteraction=0·048; figure 5).
Supportive evidence for this interaction was also seen in
within-study comparisons (eg, N95 had a stronger
protective association compared with surgical masks or
12–16-layer cotton masks); both N95 and surgical masks
also had a stronger association with protection versus
single-layer masks.38,39,51,53,54,61,66,67,75
We did a sensitivity analysis to test the robustness of
our findings and to integrate all available information
on face mask treatment eects for protection from
COVID-19. We reconsidered our findings using ran-
dom-eects Bayesian meta-analysis. Although non-
informative priors showed similar results to frequentist
approaches (aOR 0·16, 95% CrI 0·04–0·40), even using
informative priors from the most recent meta-analysis
on the eectiveness of masks versus no masks to
prevent influenza-like illness (RR 0·93, 95% CI
0·83–1·05)31 yielded a significant association with
protection from COVID-19 (aOR 0·40, 95% CrI
0·16–0·97; posterior probability for RR <1, 98%).
Minimally informing (25% influence with or without
four-fold smaller mean eect size) the most recent and
rigorous meta-analysis of the eectiveness of N95
exp(b)=2·02 per m, 95% CI 1·08–3·76; p=0·041
0 1 2 3
0
1
5
10
15
Absolute risk (%)
Distance (m)
B
95% CI
Cut points, mean
Out of sample predictions
95% CI
Regression slope
Study
High baseline risk for infection (eg, 50%)
Intermediate baseline risk for infection (eg, 10%)
Low baseline risk for infection (eg, 1%)
0 0·5 1·0 2·01·5
–4
–2
–3
–1
0
Log risk ratio
Distance (m)
A
Figure 3: Change in relative risk with increasing distance and absolute risk with increasing distance
Meta-regression of change in relative risk with increasing distance from an infected individual (A). Absolute risk of
transmission from an individual infected with SARS-CoV-2, SARS-CoV, or MERS-CoV with varying baseline risk and
increasing distance (B). SARS-CoV-2=severe acute respiratory syndrome coronavirus 2. SARS-CoV=severe acute
respiratory syndrome coronavirus. MERS-CoV=Middle East respiratory syndrome coronavirus.
Articles
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
9
respirators versus medical masks in randomised
trials (OR 0·76, 95% CI 0·54–1·06)13 with the eect-
modification seen in this meta-analysis on COVID-19
(ratio of aORs 0·14, 95% CI 0·02–1·05) continued to
support a stronger association of protection from
COVID-19, SARS, or MERS with N95 or similar respi-
rators versus other face masks (posterior probabi lity for
RR <1, 100% and 95%, respectively).
In 13 unadjusted studies and two adjusted
studies,34,37-39,47,49,51,54,58,60,61,65,75 eye protection was associated
with lower risk of infection (unadjusted n=3713,
RR 0·34, 95% CI 0·22 to 0·52; AR 5·5% with eye
protection vs 16·0% with no eye protection, RD –10·6%,
95% CI –12·5 to –7·7; adjusted n=701, aOR 0·22,
95% CI 0·12 to 0·39; low certainty; figure 6; table 2;
appendix pp 16–17).
Across 24 studies in health-care and non-health-care
settings during the current pandemic of COVID-19,
previous epidemics of SARS and MERS, or in general
use, looking at contextual factors to consider in
recom mendations, most stakeholders found physical
distancing and use of face masks and eye protection
acceptable, feasible, and reassuring (appendix pp 20–22).
However, challenges included frequent discomfort,
high resource use linked with potentially decreased
equity, less clear communi cation, and perceived
reduced empathy of care providers by those they were
caring for.
10·1 0·5 2 10
Favours face mask Favours no face mask
Interaction by setting, p=0·049; adjusted for N95 and distance, p=0·11
Unadjusted estimates, overall (I2=48%)
Ki et al (2019)47
Yin et al (2004)75
Park et al (2016)59
Lau et al (2004)50
Adjusted estimates, overall (1 COVID-19, 1 MERS, 8 SARS)
Random subtotal (I2=0%)
Ho et al (2004)45
Kim et al (2016)49
Heinzerling et al (2020)44
Random subtotal (I2=50%)
Burke et al (2020)37
Non-health-care setting
Nishiyama et al (2008)56
Teleman et al (2004)68
Wilder-Smith et al (2005)72
Ha et al (2004)42
Health-care setting
Loeb et al (2004)53
Ryu et al (2019)65
Wu et al (2004)74
Peck et al (2004)60
Wang et al (2020)70
Wang et al (2020)41
Scales et al (2003)66
Liu et al (2009)51
Tuan et al (2007)69
Park et al (2004)58
Reynolds et al (2006)64
Nishiura et al (2005)55
Pei et al (2006)61
Kim et al (2016)48
Alraddadi et al (2016)34
Hall et al (2014)43
Seto et al (2003)67
Country
South Korea
China
South Korea
China
Singapore
South Korea
USA
USA
Vietnam
Singapore
Singapore
Vietnam
Canada
South Korea
China
USA
China
China
Canada
China
Vietnam
USA
Vietnam
Vietnam
China
South Korea
Saudi Arabia
Saudi Arabia
China
Respirator
(0=no)
1
0
0
0
1
1
0
1
0
1
1
1
1
1
0
1
1
1
0
0
0
1
0
0
0
0
1
1
1
Infection
MERS
SARS
MERS
SARS
SARS
MERS
COVID-19
COVID-19
SARS
SARS
SARS
SARS
SARS
MERS
SARS
SARS
COVID-19
COVID-19
SARS
SARS
SARS
SARS
SARS
SARS
SARS
MERS
MERS
MERS
SARS
RR (95% CI)
0·34 (0·26–0·45)
0·08 (0·005–1·43)
0·40 (0·29–0·57)
0·25 (0·06–1·06)
0·53 (0·28–0·99)
0·15 (0·07–0·34)
0·18 (0·08–0·38)
aOR
aRR
0·56 (0·40–0·79)
0·16 (0·03–1·02)
0·04 (0·01–0·33)
0·03 (0·002–0·54)
0·30 (0·22–0·41)
(Not calculable)
0·36 (0·22–0·58)
0·21 (0·07–0·62)
0·40 (0·19–0·84)
(Not calculable)
0·23 (0·07–0·78)
(Not calculable)
0·57 (0·38–0·85)
(Not calculable)
0·03 (0·004–0·19)
0·04 (0·002–0·63)
0·70 (0·19–2·63)
0·54 (0·26–1·11)
1·03 (0·06–16·83)
(Not calculable)
0·34 (0·17–0·69)
0·79 (0·37–1·67)
0·21 (0·12–0·38)
0·13 (0·01–2·30)
0·44 (0·17–1·12)
(Not calculable)
0·15 (0·01–2·40)
Events,
face mask
(n/N)
163/3686
0/218
46/202
3/24
12/89
37/244
2/62
1/444
0/31
126/3442
0/64
17/61
3/26
6/27
0/61
3/23
0/24
25/146
0/13
1/1286
0/278
3/16
8/123
0/9
0/60
8/42
8/43
11/98
0/7
6/116
0/42
0/51
Events, no
face mask
(n/N)
2/4
3/6
0/1
5/9
1/2
0/6
546/6484
6/230
31/55
25/98
101/481
2/10
16/308
445/6003
0/13
14/18
33/60
39/71
0/10
69/229
0/19
119/4036
10/215
4/15
43/354
7/154
0/45
14/25
17/72
61/115
12/101
13/203
% weight
(random)
3·2
100·0
0·8
10·3
2·8
7·5
18·1
1·9
1·6
0·9
81·9
0
9·0
4·2
6·5
0
3·6
0
9·7
0
1·7
0·9
6·7
0·9
0
6·7
6·5
7·9
0·8
5·0
0
0·9
Figure 4: Forest plot showing unadjusted estimates for the association of face mask use with viral infection causing COVID-19, SARS, or MERS
SARS=severe acute respiratory syndrome. MERS=Middle East respiratory syndrome. RR=relative risk. aOR=adjusted odds ratio. aRR=adjusted relative risk.
Articles
10
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
Discussion
The findings of this systematic review of 172 studies
(44 comparative studies; n=25 697 patients) on COVID-19,
SARS, and MERS provide the best available evidence
that current policies of at least 1 m physical distancing
are associated with a large reduction in infection, and
distances of 2 m might be more eective. These data also
suggest that wearing face masks protects people (both
health-care workers and the general public) against
infection by these coronaviruses, and that eye protection
could confer additional benefit. However, none of these
interventions aorded complete protection from infection,
and their optimum role might need risk assessment and
several contextual considerations. No randomised trials
were identified for these interventions in COVID-19,
SARS, or MERS.
Previous reviews are limited in that they either have not
provided any evidence from COVID-19 or did not use
direct evidence from other related emerging epidemic
betacoronaviruses (eg, SARS and MERS) to inform the
eects of interventions to curtail the current COVID-19
pandemic.13,19,31,78 Previous data from randomised trials are
mainly for common respiratory viruses such as seasonal
influenza, with a systematic review concluding low
certainty of evidence for extrapolating these findings to
COVID-19.13 Further, previous syntheses of available
randomised control led trials have not accounted for
cluster eects in analyses, leading to substantial
imprecision in treatment eect estimates. In between-
study and within-study comparisons, we noted a larger
eect of N95 or similar respirators compared with other
masks. This finding is inconsistent with conclusions of a
review of four randomised trials,13 in which low certainty
of evidence for no larger eect was suggested. However, in
that review, the CIs were wide so a meaningful protective
eect could not be excluded. We harmonised these
findings with Bayesian approaches, using indirect data
from randomised trials to inform posterior estimates.
Despite this step, our findings continued to support the
ideas not only that masks in general are associated with a
large reduction in risk of infection from SARS-CoV-2,
SARS-CoV, and MERS-CoV but also that N95 or similar
respirators might be associated with a larger degree of
protec tion from viral infection than disposable medical
masks or reusable multilayer (12–16-layer) cotton masks.
Nevertheless, in view of the limitations of these data, we
did not rate the certainty of eect as high.21 Our findings
accord with those of a cluster randomised trial showing a
potential benefit of continuous N95 respirator use over
medical masks against seasonal viral infections.79 Further
high-quality research, including randomised trials of
the optimum physical distance and the eectiveness of
dierent types of masks in the general population and
for health-care workers’ protection, is urgently needed.
Two trials are registered to better in form the optimum use
of face masks for COVID-19 (NCT04296643 [n=576] and
Figure 5: Forest plot showing adjusted estimates for the association of face mask use with viral infection causing COVID-19, SARS, or MERS
SARS=severe acute respiratory syndrome. MERS=Middle East respiratory syndrome. RR=relative risk. aOR=adjusted odds ratio. AGP=aerosol-generating procedures.
*Studies clearly reporting AGP.
100·0
7·7
10·4
34·5
12·8
11·2
12·0
65·5
10·8
6·2
8·9
11·0
9·0
% weight
(random)
0·15 (0·07–0·34)
0·08 (0·01–0·50)
0·41 (0·13–1·26)
0·40 (0·16–0·97)
0·04 (0·004–0·30)
0·78 (0·61–1·00)
0·30 (0·12–0·73)
0·32 (0·17–0·61)
0·33 (0·17–0·61)
0·22 (0·08–0·62)
0·002 (0·000–0·02)
0·01 (0·003–0·06)
0·29 (0·11–0·75)
0·08 (0·02–0·34)
aOR (95% CI)
Nishiyama et al (2008)56
Surgical face mask or similar (eg, 12–16-layer cotton) vs no face mask
Alraddadi et al* (2016)34
Random subtotal (I2=87%)
Yin et al (2004)75
Wu et al (2004)74
Lau et al (2004)50
Random subtotal (I2=76%)
Random overall (I2=88%)
Bayesian overall (Jefferson31 seasonal viruses)
Interaction p=0·090; adjusted for setting, p=0·17; adjusted for AGP, p=0·048
Liu et al* (2009)51
Wang et al (2020)41
Ma et al* (2004)54
Nishiura et al (2005)55
Seto et al (2003)67
N95 respirator or similar vs no face mask
Country
Vietnam
Saudi Arabia
China
China
China
China
China
China
Vietnam
China
Virus
SARS
MERS
SARS
SARS
SARS
SARS
COVID-19
SARS
SARS
SARS
Setting
Health care
Health care
Health care
Health care
Health care
Health care
Health care
Health care
Non-health care
Non-health care
10·10·5 210
Favours face mask Favours no face mask
Articles
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
11
NCT04337541 [n=6000]). Until such data are available, our
findings represent the current best estimates to inform
face mask use to reduce infection from COVID-19. We
recognise that there are strong, perhaps opposing,
sentiments about policy making during outbreaks. In one
viewpoint, the 2007 SARS Commission report stated:
“...recognize, as an aspect of health worker safety, the
precautionary principle that reasonable action to reduce
risk, such as the use of a fitted N95 respirator, need not
await scientific certainty”.80
“...if we do not learn from SARS and we do not make the
government fix the problems that remain, we will pay a
terrible price in the next pandemic”.81
A counter viewpoint is that the scientific uncertainty
and contextual considerations require a more nuanced
approach. Although challenging, policy makers must
carefully consider these two viewpoints along with our
findings.
We found evidence of moderate certainty that current
policies of at least 1 m physical distancing are probably
associated with a large reduction in infection, and that
distances of 2 m might be more eective, as implemented
in some countries. We also provide estimates for 3 m.
The main benefit of physical distancing measures is to
prevent onward transmission and, thereby, reduce the
adverse outcomes of SARS-CoV-2 infection. Hence, the
results of our current review support the implementation
of a policy of physical distancing of at least 1 m and, if
feasible, 2 m or more. Our findings also provide robust
estimates to inform models and contact tracing used to
plan and strategise for pandemic response eorts at
multiple levels.
The use of face masks was protective for both health-
care workers and people in the community exposed
to infection, with both the frequentist and Bayesian
analyses lending support to face mask use irrespective
of setting. Our unadjusted analyses might, at first
impression, suggest use of face masks in the community
setting to be less eective than in the health-care setting,
but after accounting for dierential N95 respirator use
between health-care and non-health-care settings, we did
not detect any striking dierences in eectiveness of
Figure 6: Forest plot showing the association of eye protection with risk of COVID-19, SARS, or MERS transmission
Forest plot shows unadjusted estimates. SARS=severe acute respiratory syndrome. MERS=Middle East respiratory syndrome. RR=relative risk. aOR=adjusted odds ratio.
aRR=adjusted relative risk.
Interaction by virus, p=0·75
Random overall (I2=43%)
Alraddadi
et al
(2016)34
Park
et al
(2004)58
Kim
et al
(2016)49
Liu
et al
(2009)51
Caputo
et al
(2006)38
COVID-19
Random subtotal (
I
2=62%)
Pei
et al
(2006)61
Ma
et al
(2004)54
Adjusted estimates, overall (2 studies, Yin75 and Ma54)
SARS
Yin
et al
(2004)75
Burke
et al
(2020)37
Ki
et al
(2019)47
Chen
et al
(2009)39
Ryu
et al
(2019)65
Random subtotal
Peck
et al
(2004)60
Random subtotal (I2=0%)
MERS
100·0
4·0
0
1·8
21·2
5·6
92·0
26·0
15·6
19·4
0
2·2
4·2
0
0
0
8·0
Saudi Arabia
USA
South Korea
China
Canada
China
China
China
USA
South Korea
China
South Korea
USA
Country
1
1
1
0
1
0
1
0
1
1
0
1
1
Respirator
(0=no)
0·34 (0·22–0·52)
0·21 (0·03–1·51)
(Not calculable)
0·13 (0·01–2·76)
0·58 (0·33–1·01)
0·35 (0·07–1·79)
0·34 (0·21–0·56)
0·53 (0·36–0·77)
0·27 (0·12–0·59)
0·22 (0·12–0·39)
0·25 (0·14–0·43)
aOR
aRR
0·17 (0·09–0·32)
(Not calculable)
0·50 (0·03–8·21)
0·17 (0·02–1·22)
(Not calculable)
(Not calculable)
(Not calculable)
0·24 (0·06–0·99)
RR (95% CI
)%
weight
(random)
62/1335
1/47
0/30
0/443
17/221
2/46
61/770
24/120
7/175
10/120
0/9
1/45
0/24
0/42
0/42
0/13
1/523
Events, eye
protection
(n/N)
383/2378
17/165
0/72
2/294
34/256
4/32
358/1811
123/323
40/269
67/137
6/64
90/703
0/10
0/34
0/34
0/19
25/533
Events, no
eye protection
(n/N)
10·1 0·5 2 10
Favours eye protection Favours
no eye protection
Articles
12
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
face mask use between settings. The credibility of eect-
modification across settings was, therefore, low. Wearing
face masks was also acceptable and feasible. Policy
makers at all levels should, therefore, strive to address
equity implications for groups with currently limited
access to face masks and eye protection. One concern is
that face mask use en masse could divert supplies from
people at highest risk for infection.10 Health-care workers
are increasingly being asked to ration and reuse PPE,82,83
leading to calls for government-directed repurposing of
manufacturing capacity to overcome mask shortages84
and finding solutions for mask use by the general
public.84 In this respect, some of the masks studied in
our review were reusable 12–16-layer cotton or gauze
masks.51,54,61,75 At the moment, although there is consensus
that SARS-CoV-2 mainly spreads through large droplets
and contact, debate continues about the role of
aerosol,2–8,85,86 but our meta-analysis provides evidence
(albeit of low certainty) that respirators might have a
stronger protective eect than surgical masks. Biological
plausi bility would be supported by data for aerosolised
SARS-CoV-25–8 and preclinical data showing seasonal
coronavirus RNA detection in fine aerosols during tidal
breathing,87 albeit, RNA detection does not necessarily
imply replication and infection-competent virus.
Nevertheless, our findings suggest it plausible that
even in the absence of aerosolisation, respirators might
be simply more eective than masks at preventing
infection. At present, there is no data to support viable
virus in the air outside of aerosol generating procedures
from available hospital studies. Other factors such as
super-spreading events, the subtype of health-care set-
ting (eg, emergency room, intensive care unit, medical
wards, dialysis centre), if aerosolising proce dures are
done, and environmental factors such as ventilation,
might all aect the degree of protection aorded by
personal protection strategies, but we did not identify
robust data to inform these aspects.
Strengths of our review include adherence to full
systematic review methods, which included artificial intel-
ligence-supported dual screening of titles and abstracts,
full-text evaluation, assessment of risk of bias, and no
limitation by language. We included patients infected
with SARS-CoV-2, SARS-CoV, or MERS-CoV and searched
relevant data up to May 3, 2020. We followed the GRADE
approach16 to rate the certainty of evidence. Finally, we
identified and appraise a large body of published work
from China, from which much evidence emerged before
the pandemic spread to other global regions.
The primary limitation of our study is that all studies
were non-randomised, not always fully adjusted, and
might suer from recall and measurement bias (eg, direct
contact in some studies might not be measuring near
distance). However, unadjusted, adjusted, frequentist, and
Bayesian meta-analyses all supported the main findings,
and large or very large eects were recorded. Nevertheless,
we are cautious not to be overly certain in the precise
quantitative estimates of eects, although the qualitative
eect and direction is probably of high certainty. Many
studies did not provide information on precise distances,
and direct contact was equated to 0 m distance; none of the
eligible studies quanti tatively evaluated whether distances
of more than 2 m were more eective, although our meta-
regression provides potential pre dictions for estimates of
risk. Few studies assessed the eect of interventions in
non-health-care settings, and they primarily evaluated
mask use in households or contacts of cases, although
beneficial associations were seen across settings.
Furthermore, most evidence was from studies that
reported on SARS and MERS (n=6674 patients with
COVID-19, of 25 697 total), but data from these previous
epidemics provide the most direct information for
COVID-19 currently. We did not specifically assess the
eect of duration of exposure on risk for transmission,
although whether or not this variable was judged a risk
factor considerably varied across studies, from any
duration to a minimum of 1 h. Because of inconsistent
reporting, information is limited about whether aerosol-
generating procedures were in place in studies using
respirators, and whether masks worn by infected patients
might alter the eectiveness of each intervention, although
the stronger association with N95 or similar respirators
over other masks persisted when adjusting for studies
reporting aerosol-generating medical procedures. These
factors might account for some of the residual statistical
heterogeneity seen for some outcomes, albeit I² is com-
monly inflated in meta-analyses of observational data,21,22
and nevertheless the eects seen were large and probably
clinically important in all adjusted studies.
Our comprehensive systematic review provides the
best available information on three simple and com-
mon interventions to combat the immediate threat of
COVID-19, while new evidence on pharmacological treat-
ments, vac cines, and other personal protective strategies is
being generated. Physical distancing of at least 1 m is
strongly associated with protection, but distances of up to
2 m might be more eective. Although direct evidence is
limited, the optimum use of face masks, in particular N95
or similar respirators in health-care settings and 12–16-layer
cotton or surgical masks in the community, could depend
on contextual factors; action is needed at all levels to
address the paucity of better evidence. Eye protection
might provide additional benefits. Globally collaborative
and well conducted studies, including randomised trials,
of dierent personal protective strategies are needed
regardless of the challenges, but this systematic appraisal
of currently best available evidence could be considered to
inform interim guidance.
Contributors
DKC, EAA, SD, KS, SY, and HJS designed the study. SY, SD, KS,
and HJS coordinated the study. SY and LH designed and ran the
literature search. All authors acquired data, screened records, extracted
data, and assessed risk of bias. DKC did statistical analyses. DKC and
HJS wrote the report. All authors provided critical conceptual input,
analysed and interpreted data, and critically revised the report.
Articles
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
13
COVID-19 Systematic Urgent Review Group Effort (SURGE) study authors
Argentina—German Hospital of Buenos Aires (Ariel Izcovich);
Canada—Cochrane Consumer Executive (Maureen Smith); McMaster
University (Mark Loeb, Anisa Hajizadeh, Carlos A Cuello-Garcia,
Gian Paolo Morgano, Leila Harrison, Tejan Baldeh, Karla Solo,
Tamara Lotfi, Antonio Bognanni, Rosa Stalteri, Thomas Piggott,
Yuan Zhang, Stephanie Duda, Derek K Chu, Holger J Schünemann);
Southlake Regional Health Centre (Jerey Chan); University of British
Columbia (David James Harris); Chile—Pontificia Universidad Católica
de Chile (Ignacio Neumann); China—Beijing University of Chinese
Medicine, Dongzhimen Hospital (Guang Chen); Guangzhou University
of Chinese Medicine, The Fourth Clinical Medical College (Chen Chen);
China Academy of Chinese Medical Sciences (Hong Zhao); Germany—
Finn Schünemann; Italy—Azienda USL–IRCCS di Reggio Emilia
(Paolo Giorgi Rossi); Universita Vita-Salute San Raaele, Milan, Italy
(Giovanna Elsa Ute Muti Schünemann); Lebanon—American University
of Beirut (Layal Hneiny, Amena El-Harakeh, Fatimah Chamseddine,
Joanne Khabsa, Nesrine Rizk, Rayane El-Khoury, Zahra Saad,
Sally Yaacoub, Elie A Akl); Rafik Hariri University Hospital
(Pierre AbiHanna); Poland—Evidence Prime, Krakow (Anna Bak,
Ewa Borowiack); UK—The London School of Hygiene & Tropical
Medicine (Marge Reinap); University of Hull (Assem Khamis).
Declaration of interests
ML is an investigator of an ongoing clinical trial on medical masks
versus N95 respirators for COVID-19 (NCT04296643). All other authors
declare no competing interests.
Acknowledgments
This systematic review was commissioned and in part paid for by WHO.
The authors alone are responsible for the views expressed in this article
and they do not necessarily represent the decisions, policy, or views of
WHO. We thank Susan L Norris, April Baller, and Benedetta Allegranzi
(WHO) for input in the protocol or the final article; Xuan Yu (Evidence
Based Medicine Center of Lanzhou University, China), Eliza Poon,
and Yuqing (Madison) Zhang for assistance with Chinese literature
support; Neera Bhatnagar and Aida Farha (information specialists) for
peer-reviewing the search strategy; Artur Nowak (Evidence Prime,
Hamilton, ON, Canada) for help with searching and screening using
artificial intelligence; and Christine Keng for additional support. DKC is
a CAAIF-CSACI-AllerGen Emerging Clinician-Scientist Research Fellow,
supported by the Canadian Allergy, Asthma and Immunology
Foundation (CAAIF), the Canadian Society of Allergy and Clinical
Immunology (CSACI), and AllerGen NCE (the Allergy, Genes and
Environment Network).
Editorial note: the Lancet Group takes a neutral position with respect to
territorial claims in published maps and institutional aliations.
References
1 Worldometer. COVID-19 coronavirus pandemic. 2020. https://www.
worldometers.info/coronavirus/ (accessed May 28, 2020).
2 Guo ZD, Wang ZY, Zhang SF, et al. Aerosol and surface
distribution of severe acute respiratory syndrome coronavirus 2 in
hospital wards, Wuhan, China, 2020. Emerg Infect Dis 2020;
published online April 10. DOI:10.3201/eid2607.200885.
3 Chia PY, Coleman KK, Tan YK, et al. Detection of air and surface
contamination by severe acute respiratory syndrome coronavirus 2
(SARS-CoV-2) in hospital rooms of infected patients. medRxiv 2020;
published online April 9. DOI:10.1101/2020.03.29.20046557 (preprint).
4 Santarpia JL, Rivera DN, Herrera V, et al. Transmission potential of
SARS-CoV-2 in viral shedding observed at the University of
Nebraska Medical Center. medRxiv 2020; published online
March 26. DOI:10.1101/2020.03.23.20039446 (preprint).
5 Cheng V, Wong S-C, Chen J, et al. Escalating infection control
response to the rapidly evolving epidemiology of the coronavirus
disease 2019 (COVID-19) due to SARS-CoV-2 in Hong Kong.
Infect Control Hosp Epidemiol 2020; 41: 493–98.
6 Wong SCY, Kwong RT-S, Wu TC, et al. Risk of nosocomial
transmission of coronavirus disease 2019: an experience in a
general ward setting in Hong Kong. J Hosp Infect 2020; 105: 119–27.
7 Faridi S, Niazi S, Sadeghi K, et al. A field indoor air measurement
of SARS-CoV-2 in the patient rooms of the largest hospital in Iran.
Sci Total Environ 2020; 725: 138401.
8 Ong SWX, Tan YK, Chia PY, et al. Air, surface environmental,
and personal protective equipment contamination by severe acute
respiratory syndrome coronavirus 2 (SARS-CoV-2) from a
symptomatic patient. JAMA 2020; 323: 1610–12.
9 Qualls N, Levitt A, Kanade N, et al. Community mitigation
guidelines to prevent pandemic influenza: United States, 2017.
MMWR Recomm Rep 2017; 66: 1–34.
10 Feng S, Shen C, Xia N, Song W, Fan M, Cowling BJ. Rational use of
face masks in the COVID-19 pandemic. Lancet Respir Med 2020;
8: 434–36.
11 MacIntyre R, Chughtai A, Tham CD, Seale H. COVID-19:
should cloth masks be used by healthcare workers as a last resort?
April 9, 2020. https://blogs.bmj.com/bmj/2020/04/09/covid-19-
should-cloth-masks-be-used-by-healthcare-workers-as-a-last-resort/
(accessed May 12, 2020).
12 Loeb M, Dafoe N, Mahony J, et al. Surgical mask vs N95 respirator
for preventing influenza among health care workers: a randomized
trial. JAMA 2009; 302: 1865–71.
13 Bartoszko JJ, Farooqi MAM, Alhazzani W, Loeb M. Medical masks
vs N95 respirators for preventing COVID-19 in healthcare workers:
a systematic review and meta-analysis of randomized trials.
Influenza Other Respir Viruses 2020; published online April 4.
DOI:10.1111/irv.12745.
14 Schünemann HJ, Moja L. Reviews: rapid! Rapid! Rapid! . . . and
systematic. Syst Rev 2015; 4: 4.
15 Cochrane Training. Cochrane handbook for systematic reviews of
interventions, version 6. 2019. https://training.cochrane.org/
handbook/current (accessed May 12, 2020).
16 Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging
consensus on rating quality of evidence and strength of
recommendations. BMJ 2008; 336: 924–26.
17 Moher D, Liberati A, Tetzla J, Altman DG. Preferred reporting
items for systematic reviews and meta-analyses: the PRISMA
statement. J Clin Epidemiol 2009; 62: 1006–12.
18 Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of
observational studies in epidemiology: a proposal for reporting.
JAMA 2000; 283: 2008–12.
19 Jeerson T, Del Mar CB, Dooley L, et al. Physical interventions to
interrupt or reduce the spread of respiratory viruses.
Cochrane Database Syst Rev 2011; 7: CD006207.
20 Oeddu V, Yung CF, Low MSF, Tam CC. Eectiveness of masks and
respirators against respiratory infections in healthcare workers:
a systematic review and meta-analysis. Clin Infect Dis 2017;
65: 1934–42.
21 Guyatt GH, Oxman AD, Kunz R, et al. GRADE guidelines, 7:
rating the quality of evidence—inconsistency. J Clin Epidemiol 2011;
64: 1294–302.
22 Iorio A, Spencer FA, Falavigna M, et al. Use of GRADE for
assessment of evidence about prognosis: rating confidence in
estimates of event rates in broad categories of patients. BMJ 2015;
350: h870.
23 Moskalewicz A, Oremus M. No clear choice between Newcastle-
Ottawa Scale and Appraisal Tool for Cross-Sectional Studies to assess
methodological quality in cross-sectional studies of health-related
quality of life and breast cancer. J Clin Epidemiol 2020; 120: 94–103.
24 Wells GA, Shea B, O’Connell D, et al. The Newcastle-Ottawa Scale
(NOS) for assessing the quality of nonrandomised studies in
meta-analyses. 2019. http://www.ohri.ca/programs/clinical_
epidemiology/oxford.asp (accessed May 12, 2020).
25 Sterne JAC, Savović J, Page MJ, et al. RoB 2: a revised tool for
assessing risk of bias in randomised trials. BMJ 2019; 366: l4898.
26 Guyatt G, Oxman AD, Akl EA, et al. GRADE guidelines, 1:
introduction—GRADE evidence profiles and summary of findings
tables. J Clin Epidemiol 2011; 64: 383–94.
27 Guyatt GH, Thorlund K, Oxman AD, et al. GRADE guidelines, 13:
preparing summary of findings tables and evidence profiles—
continuous outcomes. J Clin Epidemiol 2013; 66: 173–83.
28 Santesso N, Carrasco-Labra A, Langendam M, et al. Improving
GRADE evidence tables part 3: detailed guidance for explanatory
footnotes supports creating and understanding GRADE certainty in
the evidence judgments. J Clin Epidemiol 2016; 74: 28–39.
29 Santesso N, Glenton C, Dahm P, et al. GRADE guidelines, 26:
informative statements to communicate the findings of systematic
reviews of interventions. J Clin Epidemiol 2020; 119: 126–35.
Articles
14
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
30 Higgins JP, Thompson SG. Controlling the risk of spurious
findings from meta-regression. Stat Med 2004; 23: 1663–82.
31 Jeerson T, Jones M, Al Ansari LA, et al. Physical interventions to
interrupt or reduce the spread of respiratory viruses, part 1:
face masks, eye protection and person distancing—systematic
review and meta-analysis. medRxiv 2020; published online April 7.
DOI:10.1101/2020.03.30.20047217 (preprint).
32 Sutton AJ, Abrams KR. Bayesian methods in meta-analysis and
evidence synthesis. Stat Methods Med Res 2001; 10: 277–303.
33 Goligher EC, Tomlinson G, Hajage D, et al. Extracorporeal membrane
oxygenation for severe acute respiratory distress syndrome and
posterior probability of mortality benefit in a post hoc Bayesian
analysis of a randomized clinical trial. JAMA 2018; 320: 2251–59.
34 Alraddadi BM, Al-Salmi HS, Jacobs-Slifka K, et al. Risk factors for
Middle East respiratory syndrome coronavirus infection among
healthcare personnel. Emerg Infect Dis 2016; 22: 1915–20.
35 Arwady MA, Alraddadi B, Basler C, et al. Middle East respiratory
syndrome coronavirus transmission in extended family,
Saudi Arabia, 2014. Emerg Infect Dis 2016; 22: 1395–402.
36 Bai Y, Wang X, Huang Q, et al. SARS-CoV-2 infection in health care
workers: a retrospective analysis and a model study. medRxiv 2020;
published online April 1. DOI:10.1101/2020.03.29.20047159
(preprint).
37 Burke RM, Balter S, Barnes E, et al. Enhanced contact
investigations for nine early travel-related cases of SARS-CoV-2 in
the United States. medRxiv 2020; published online May 3.
DOI:10.1101/2020.04.27.20081901 (preprint).
38 Caputo KM, Byrick R, Chapman MG, Orser BJ, Orser BA.
Intubation of SARS patients: infection and perspectives of
healthcare workers. Can J Anaesth 2006; 53: 122–29.
39 Chen WQ, Ling WH, Lu CY, et al. Which preventive measures might
protect health care workers from SARS? BMC Public Health 2009;
9: 81.
40 Cheng H-Y, Jian S-W, Liu D-P, Ng T-C, Huang W-T, Lin H-H.
High transmissibility of COVID-19 near symptom onset. medRxiv
2020; published online March 19. DOI:10.1101/2020.03.18.20034561
(preprint).
41 Wang X, Pan Z, Cheng Z. Association between 2019-nCoV
transmission and N95 respirator use. J Hosp Infect 2020;
105: 104–05.
42 Ha LD, Bloom SA, Hien NQ, et al. Lack of SARS transmission among
public hospital workers, Vietnam. Emerg Infect Dis 2004; 10: 265–68.
43 Hall AJ, Tokars JI, Badreddine SA, et al. Health care worker contact
with MERS patient, Saudi Arabia. Emerg Infect Dis 2014; 20: 2148–51.
44 Heinzerling A, Stuckey MJ, Scheuer T, et al. Transmission of
COVID-19 to health care personnel during exposures to a
hospitalized patient: Solano County, California, February 2020.
MMWR Morb Mortal Wkly Rep 2020; 69: 472–76.
45 Ho KY, Singh KS, Habib AG, et al. Mild illness associated with
severe acute respiratory syndrome coronavirus infection: lessons
from a prospective seroepidemiologic study of health-care workers
in a teaching hospital in Singapore. J Infect Dis 2004; 189: 642–47.
46 Van Kerkhove MD, Alaswad S, Assiri A, et al. Transmissibility of
MERS-CoV infection in closed setting, Riyadh, Saudi Arabia, 2015.
Emerg Infect Dis J 2019; 25: 1802–09.
47 Ki HK, Han SK, Son JS, Park SO. Risk of transmission via medical
employees and importance of routine infection-prevention policy in
a nosocomial outbreak of Middle East respiratory syndrome
(MERS): a descriptive analysis from a tertiary care hospital in
South Korea. BMC Pulm Med 2019; 19: 190.
48 Kim T, Jung J, Kim SM, et al. Transmission among healthcare
worker contacts with a Middle East respiratory syndrome patient in
a single Korean centre. Clin Microbiol Infect 2016; 22: e11–13.
49 Kim CJ, Choi WS, Jung Y, et al. Surveillance of the Middle East
respiratory syndrome (MERS) coronavirus (CoV) infection in
healthcare workers after contact with confirmed MERS patients:
incidence and risk factors of MERS-CoV seropositivity.
Clin Microbiol Infect 2016; 22: 880–86.
50 Lau JTF, Lau M, Kim JH, Tsui HY, Tsang T, Wong TW.
Probable secondary infections in households of SARS patients in
Hong Kong. Emerg Infect Dis 2004; 10: 235–43.
51 Liu W, Tang F, Fang LQ, et al. Risk factors for SARS infection
among hospital healthcare workers in Beijing: a case control study.
Trop Med Int Health 2009; 14 (suppl 1): 52–59.
52 Liu ZQ, Ye Y, Zhang H, Guohong X, Yang J, Wang JL. Analysis of
the spatio-temporal characteristics and transmission path of
COVID-19 cluster cases in Zhuhai. Trop Geogr 2020; published
online March 12. DOI:10.13284/j.cnki.rddl.003228.
53 Loeb M, McGeer A, Henry B, et al. SARS among critical care
nurses, Toronto. Emerg Infect Dis 2004; 10: 251–55.
54 Ma HJ, Wang HW, Fang LQ, et al. A case-control study on the risk
factors of severe acute respiratory syndromes among health care
workers. Zhonghua Liu Xing Bing Xue Za Zhi 2004; 25: 741–44
(in Chinese).
55 Nishiura H, Kuratsuji T, Quy T, et al. Rapid awareness and
transmission of severe acute respiratory syndrome in Hanoi French
Hospital, Vietnam. Am J Trop Med Hyg 2005; 73: 17–25.
56 Nishiyama A, Wakasugi N, Kirikae T, et al. Risk factors for SARS
infection within hospitals in Hanoi, Vietnam. Jpn J Infect Dis 2008;
61: 388–90.
57 Olsen SJ, Chang HL, Cheung TY, et al. Transmission of the severe
acute respiratory syndrome on aircraft. N Engl J Med 2003;
349: 2416–22.
58 Park BJ, Peck AJ, Kuehnert MJ, et al. Lack of SARS transmission
among healthcare workers, United States. Emerg Infect Dis 2004;
10: 244–48.
59 Park JY, Kim BJ, Chung KH, Hwang YI. Factors associated with
transmission of Middle East respiratory syndrome among Korean
healthcare workers: infection control via extended healthcare
contact management in a secondary outbreak hospital.
Respirology 2016; 21 (suppl 3): 89 (abstr APSR6-0642).
60 Peck AJ, Newbern EC, Feikin DR, et al. Lack of SARS transmission
and U.S. SARS case-patient. Emerg Infect Dis 2004; 10: 217–24.
61 Pei LY, Gao ZC, Yang Z, et al. Investigation of the influencing
factors on severe acute respiratory syndrome among health care
workers. Beijing Da Xue Xue Bao Yi Xue Ban 2006; 38: 271–75.
62 Rea E, Laflèche J, Stalker S, et al. Duration and distance of exposure
are important predictors of transmission among community
contacts of Ontario SARS cases. Epidemiol Infect 2007; 135: 914–21.
63 Reuss A, Litterst A, Drosten C, et al. Contact investigation for
imported case of Middle East respiratory syndrome, Germany.
Emerg Infect Dis 2014; 20: 620–25.
64 Reynolds MG, Anh BH, Thu VH, et al. Factors associated with
nosocomial SARS-CoV transmission among healthcare workers in
Hanoi, Vietnam, 2003. BMC Public Health 2006; 6: 207.
65 Ryu B, Cho SI, Oh MD, et al. Seroprevalence of Middle East
respiratory syndrome coronavirus (MERS-CoV) in public health
workers responding to a MERS outbreak in Seoul, Republic of
Korea, in 2015. Western Pac Surveill Response J 2019; 10: 46–48.
66 Scales DC, Green K, Chan AK, et al. Illness in intensive care sta
after brief exposure to severe acute respiratory syndrome.
Emerg Infect Dis 2003; 9: 1205–10.
67 Seto WH, Tsang D, Yung RWH, et al. Eectiveness of precautions
against droplets and contact in prevention of nosocomial
transmission of severe acute respiratory syndrome (SARS).
Lancet 2003; 361: 1519–20.
68 Teleman MD, Boudville IC, Heng BH, Zhu D, Leo YS. Factors
associated with transmission of severe acute respiratory syndrome
among health-care workers in Singapore. Epidemiol Infect 2004;
132: 797–803.
69 Tuan PA, Horby P, Dinh PN, et al. SARS transmission in Vietnam
outside of the health-care setting. Epidemiol Infect 2007;
135: 392–401.
70 Wang Q, Huang X, Bai Y, et al. Epidemiological characteristics of
COVID-19 in medical sta members of neurosurgery departments
in Hubei province: a multicentre descriptive study. medRxiv 2020;
published online April 24. DOI:10.1101/2020.04.20.20064899
(preprint).
71 Wiboonchutikul S, Manosuthi W, Likanonsakul S, et al. Lack of
transmission among healthcare workers in contact with a case of
Middle East respiratory syndrome coronavirus infection in
Thailand. Antimicrob Resist Infect Control 2016; 5: 21.
72 Wilder-Smith A, Teleman MD, Heng BH, Earnest A, Ling AE,
Leo YS. Asymptomatic SARS coronavirus infection among
healthcare workers, Singapore. Emerg Infect Dis 2005; 11: 1142–45.
73 Wong TW, Lee CK, Tam W, et al. Cluster of SARS among medical
students exposed to single patient, Hong Kong. Emerg Infect Dis
2004; 10: 269–76.
Articles
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31142-9
15
74 Wu J, Xu F, Zhou W, et al. Risk factors for SARS among persons
without known contact with SARS patients, Beijing, China.
Emerg Infect Dis 2004; 10: 210–16.
75 Yin WW, Gao LD, Lin WS, et al. Eectiveness of personal
protective measures in prevention of nosocomial transmission of
severe acute respiratory syndrome.
Zhonghua Liu Xing Bing Xue Za Zhi 2004; 25: 18–22.
76 Yu ITS, Wong TW, Chiu YL, Lee N, Li Y. Temporal-spatial analysis
of severe acute respiratory syndrome among hospital inpatients.
Clin Infect Dis 2005; 40: 1237–43.
77 Yu IT, Xie ZH, Tsoi KK, et al. Why did outbreaks of severe acute
respiratory syndrome occur in some hospital wards but not in
others? Clin Infect Dis 2007; 44: 1017–25.
78 Verbeek JH, Rajamaki B, Ijaz S, et al. Personal protective
equipment for preventing highly infectious diseases due to
exposure to contaminated body fluids in healthcare sta.
Cochrane Database Syst Rev 2019; 7: CD011621.
79 MacIntyre CR, Wang Q, Seale H, et al. A randomized clinical trial
of three options for N95 respirators and medical masks in health
workers. Am J Respir Crit Care Med 2013; 187: 960–66.
80 Campbell A. Chapter eight: it’s not about the mask: SARS
Commission final report, volume 3. December, 2006. http://www.
archives.gov.on.ca/en/e_records/sars/report/v3-pdf/Vol3Chp8.pdf
(accessed May 12, 2020).
81 Webster P. Ontario issues final SARS Commission report.
Lancet 2007; 369: 264.
82 Rimmer A. COVID-19: experts question guidance to reuse PPE.
BMJ 2020; 369: m1577.
83 Mackenzie D. Reuse of N95 masks. Engineering 2020; published
online April 13. DOI:10.1016/j.eng.2020.04.003.
84 Greenhalgh T, Schmid MB, Czypionka T, Bassler D, Gruer L.
Face masks for the public during the covid-19 crisis. BMJ 2020;
369: m1435.
85 Bahl P, Doolan C, de Silva C, Chughtai AA, Bourouiba L,
MacIntyre CR. Airborne or droplet precautions for health workers
treating coronavirus disease 2019? J Infect Dis 2020; published
online April 16. DOI:10.1093/infdis/jiaa189.
86 Schünemann HJ, Khabsa J, Solo K, et al. Ventilation techniques
and risk for transmission of coronavirus disease, including
COVID-19: a living systematic review of multiple streams of
evidence. Ann Intern Med 2020; published online May 22.
DOI:10.7326/M20-2306.
87 Leung NHL, Chu DKW, Shiu EYC, et al. Respiratory virus
shedding in exhaled breath and ecacy of face masks.
Nat Med 2020; 26: 676–80.
Comment
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31183-1
1
Physical distancing, face masks, and eye protection for
prevention of COVID-19
The choice of various respiratory protection mecha-
nisms, including face masks and respirators, has been
a vexed issue, from the 2009 H1N1 pandemic to the
west African Ebola epidemic of 2014,1 to the current
COVID-19 pandemic. COVID-19 guidelines issued by
WHO, the US Centers for Disease Control and Prevention,
and other agencies have been consistent about the
need for physical distancing of 1–2 m but conflicting on
the issue of respiratory protection with a face mask or a
respirator.2 This discrepancy reflects uncertain evidence
and no consensus about the transmission mode of severe
acute respiratory syndrome coronavirus 2 (SARS-CoV-2).
For eye protection, data are even less certain. Therefore,
the systematic review and meta-analysis by Derek Chu
and colleagues in The Lancet3 is an important milestone
in our understanding of the use of personal protective
equipment (PPE) and physical distancing for COVID-19.
No randomised controlled trials were available for the
analysis, but Chu and colleagues systematically reviewed
172 observational studies and rigorously synthesised
available evidence from 44 comparative studies on SARS,
Middle East respiratory syndrome (MERS), COVID-19,
and the betacoronaviruses that cause these diseases.
The findings showed a reduction in risk of 82% with
a physical distance of 1 m in both health-care and
community settings (adjusted odds ratio [aOR] 0·18,
95% CI 0·09–0·38). Every additional 1 m of separation
more than doubled the relative protection, with
data available up to 3 m (change in relative risk [RR]
2·02 per m; pinteraction=0·041). This evidence is important
to support community physical distancing guidelines
and shows risk reduction is feasible by physical
distancing. Moreover, this finding can inform lifting of
societal restrictions and safer ways of gathering in the
community.
The 1–2 m distance rule in most hospital guidelines
is based on out-of-date findings from the 1940s, with
studies from 2020 showing that large droplets can
travel as far as 8 m.4 To separate droplet and airborne
transmission is probably somewhat artificial, with both
routes most likely part of a continuum for respiratory
transmissible infections.4 Protection against presumed
droplet infections by use of respirators, but not masks,5
supports a continuum rather than discrete states of
droplet or airborne transmission. Both experimental
and hospital studies have shown evidence of aerosol
transmission of SARS-CoV-2.6–8 One study found viable
virus in the air 16 h after aerosolisation and showed
greater airborne propensity for SARS-CoV-2 compared
with SARS-CoV and MERS-CoV.6
Chu and colleagues reported that masks and respi-
rators reduced the risk of infection by 85% (aOR 0·15,
95% CI 0·07–0·34), with greater effectiveness in health-
care settings (RR 0·30, 95% CI 0·22–0·41) than in
the community (0·56, 0·40–0·79; pinteraction=0·049).
They attribute this difference to the predominant use
of N95 respirators in health-care settings; in a sub-
analysis, respirators were 96% effective (aOR 0·04,
95% CI 0·004–0·30) compared with other masks, which
were 77% effective (aOR 0·33, 95% CI 0·17–0·61;
pinteraction=0·090). The other important finding for health
workers by Chu and colleagues was that eye protection
resulted in a 78% reduction in infection (aOR 0·22,
95% CI 0·12–0·39); infection via the ocular route might
occur by aerosol transmission or self-inoculation.9
For health-care workers on COVID-19 wards, a
respirator should be the minimum standard of care.
This study by Chu and colleagues should prompt a
review of all guidelines that recommend a medical
mask for health workers caring for COVID-19 patients.
Although medical masks do protect, the occupational
health and safety of health workers should be the
highest priority and the precautionary principle should
be applied. Preventable infections in health workers
can result not only in deaths but also in large numbers
of health workers being quarantined and nosocomial
outbreaks. In the National Health Service trusts in
the UK, up to one in five health workers have been
infected with COVID-19,10 which is an unacceptable
risk for front-line workers. To address global shortages
of PPE, countries should take responsibility for scaling
up production rather than expecting health workers to
work in suboptimum PPE.11
Chu and colleagues also report that respirators
and multilayer masks are more protective than are
single layer masks. This finding is vital to inform the
Published Online
June 1, 2020
https://doi.org/10.1016/
S0140-6736(20)31183-1
See Online/Articles
https://doi.org/10.1016/
S0140-6736(20)31142-9
Tim Dirven/Panos Pictures
Comment
2
www.thelancet.com Published online June 1, 2020 https://doi.org/10.1016/S0140-6736(20)31183-1
proliferation of home-made cloth mask designs, many
of which are single-layered. A well designed cloth mask
should have water-resistant fabric, multiple layers,
and good facial fit.12 This study supports universal face
mask use, because masks were equally effective in both
health-care and community settings when adjusted for
type of mask use. Growing evidence for presymptom-
atic and asymptomatic transmission of SARS-CoV-213
further supports universal face mask use and distancing.
In regions with a high incidence of COVID-19, universal
face mask use combined with physical distancing
could reduce the rate of infection (flatten the curve),
even with modestly effective masks.14 Universal face
mask use might enable safe lifting of restrictions in
communities seeking to resume normal activities and
could protect people in crowded public settings and
within households. Masks worn within households
in Beijing, China, prevented secondary transmission
of SARS-CoV-2 if worn before symptom onset of the
index case.15 Finally, Chu and colleagues reiterate that
no one intervention is completely protective and that
combinations of physical distancing, face mask use,
and other interventions are needed to mitigate the
COVID-19 pandemic until we have an effective vaccine.
Until randomised controlled trial data are available, this
study provides the best specific evidence for COVID-19
prevention.
CRM and QW declare no competing interests. CRM is supported by a National
Health and Medical Research Council Principal Research Fellowship (grant
number 1137582).
Copyright © 2020 The Author(s). Published by Elsevier Ltd. This is an Open
Access article under the CC BY-NC-ND 4.0 license.
*C Raina MacIntyre, Quanyi Wang
r.macintyre@unsw.edu.au
The Kirby Institute, University of New South Wales, Sydney, NSW 2052, Australia
(CRM); and Beijing Center for Disease Prevention and Control, Beijing, China (QW)
1 MacIntyre CRC, Chughtai AA, Seale H, Richards GA, Davidson PM.
Respiratory protection for healthcare workers treating Ebola virus disease
(EVD): are facemasks sufficient to meet occupational health and safety
obligations? Int J Nurs Stud 2014; 51: 1421–26.
2 Chughtai AA, Seale H, Islam MS, Owais M, Macintyre CR. Policies on the use
of respiratory protection for hospital health workers to protect from
coronavirus disease (COVID-19). Int J Nurs Stud 2020; 105: 103567.
3 Chu DK, Akl EA, Duda S, et al. Physical distancing, face masks, and eye
protection to prevent person-to-person transmission of SARS-CoV-2 and
COVID-19: a systematic review and meta-analysis. Lancet 2020; published
online June 1.
https://doi.org/10.1016/S0140-6736(20)31142-9.
4 Bahl P, Doolan C, de Silva C, Chughtai AA, Bourouiba L, MacIntyre CR.
Airborne or droplet precautions for health workers treating COVID-19?
J Infect Dis 2020; published online April 16. DOI:10.1093/infdis/jiaa189.
5 MacIntyre CR, Chughtai AA, Rahman B, et al. The efficacy of medical masks
and respirators against respiratory infection in healthcare workers.
Influenza Other Respir Viruses 2017; 11: 511–17.
6 Fears AC, Klimstra WB, Duprex P, et al. Comparative dynamic aerosol
efficiencies of three emergent coronaviruses and the unusual persistence of
SARS-CoV-2 in aerosol suspensions. medRxiv 2020; published online
April 18. DOI:10.1101/2020.04.13.20063784 (preprint).
7 Guo Z-D, Wang Z-Y, Zhang S-F, et al. Aerosol and surface distribution of
severe acute respiratory syndrome coronavirus 2 in hospital wards, Wuhan,
China, 2020. Emerg Infect Dis 2020; published online April 10.
DOI:10.3201/eid2607.200885.
8 Santarpia JL, Rivera DN, Herrera V, et al. Transmission potential of
SARS-CoV-2 in viral shedding observed at the University of Nebraska
Medical Center. medRxiv 2020; published online March 26.
DOI:10.1101/2020.03.23.20039446 (preprint).
9 Lu CW, Liu XF, Jia ZF. 2019-nCoV transmission through the ocular surface
must not be ignored. Lancet 2020; 395: e39.
10 Hunter E, Price DA, Murphy E, et al. First experience of COVID-19 screening
of health-care workers in England. Lancet 2020; 395: e77–78.
11 Greenhalgh T, Schmid MB, Czypionka T, Bassler D, Gruer L. Face masks for
the public during the covid-19 crisis. BMJ 2020; 369: m1435.
12 MacIntyre R, Chughtai A, Tham CD, Seale H. Covid-19: Should cloth masks
be used by healthcare workers as a last resort? April 9, 2020. https://blogs.
bmj.com/bmj/2020/04/09/covid-19-should-cloth-masks-be-used-by-
healthcare-workers-as-a-last-resort/ (accessed May 14, 2020).
13 He X, Lau EHY, Wu P, et al. Temporal dynamics in viral shedding and
transmissibility of COVID-19. Nat Med 2020; 26: 672–75.
14 Ngonghala CN, Iboi E, Eikenberry S, et al. Mathematical assessment of the
impact of non-pharmaceutical interventions on curtailing the 2019 novel
Coronavirus. Math Biosci 2020; 325: 108364.
15 Wang Y, Tian H, Zhang L, et al. Reduction of secondary transmission of
SARS-CoV-2 in households by face mask use, disinfection and social
distancing: a cohort study in Beijing, China. BMJ Glob Health 2020;
published online May 28. DOI:10.1136/bmjgh-2020-002794.
