Ahmed et al.

BMC Public Health (2018) 18:518

https://doi.org/10.1186/s12889-018-5446-1

RESEARCH ARTICLE Open Access

Effectiveness of workplace social distancing

measures in reducing influenza
transmission: a systematic review
Faruque Ahmed*, Nicole Zviedrite and Amra Uzicanin

Abstract
Background: Social distancing is one of the community mitigation measures that may be recommended during
influenza pandemics. Social distancing can reduce virus transmission by increasing physical distance or reducing
frequency of congregation in socially dense community settings, such as schools or workplaces. We conducted
a systematic review to assess the evidence that social distancing in non-healthcare workplaces reduces or slows
influenza transmission.
Methods: Electronic searches were conducted using MEDLINE, Embase, Scopus, Cochrane Library, PsycINFO,
CINAHL, NIOSHTIC-2, and EconLit to identify studies published in English from January 1, 2000, through May 3,
2017. Data extraction was done by two reviewers independently. A narrative synthesis was performed.
Results: Fifteen studies, representing 12 modeling and three epidemiological, met the eligibility criteria. The
epidemiological studies showed that social distancing was associated with a reduction in influenza-like illness
and seroconversion to 2009 influenza A (H1N1). However, the overall risk of bias in the epidemiological studies
was serious. The modeling studies estimated that workplace social distancing measures alone produced a median
reduction of 23% in the cumulative influenza attack rate in the general population. It also delayed and reduced
the peak influenza attack rate. The reduction in the cumulative attack rate was more pronounced when workplace
social distancing was combined with other nonpharmaceutical or pharmaceutical interventions. However, the
effectiveness was estimated to decline with higher basic reproduction number values, delayed triggering of
workplace social distancing, or lower compliance.
Conclusions: Modeling studies support social distancing in non-healthcare workplaces, but there is a paucity of
well-designed epidemiological studies.
Systematic review registration number: PROSPERO registration # CRD42017065310.
Keywords: Influenza, Distancing, Community mitigation, Non-pharmaceutical, Systematic review, Telework,
Workplace

Background in the United States, in the absence of vaccination and

Influenza pandemics occur when new influenza A viruses other mitigation measures, has been estimated to be $71
emerge that spread from person to person in an efficient to $166 billion [2]. A vaccine against a new pandemic
and sustained way. Over the past 100 years, the clinical virus might not be widely available for up to 6 months
severity of influenza pandemics has ranged from moderate given current vaccine production technology [3]. The
for the 2009 pandemic to very high for the 1918 pandemic 2017 Update of the Pandemic Influenza Plan published by
[1]. The economic impact of the next influenza pandemic the US Department of Health and Human Services
includes community mitigation measures, known also as
* Correspondence: fahmed@cdc.gov nonpharmaceutical interventions (NPIs), that can be used
Community Interventions for Infection Control Unit, Division of Global before a pandemic vaccine is widely available [4]. The
Migration and Quarantine, Centers for Disease Control and Prevention (CDC), goals of community mitigation are to delay the influenza
Atlanta, GA, USA

© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Ahmed et al. BMC Public Health (2018) 18:518 Page 2 of 13

peak to buy time for the development and administration (workplace closure is not a recommended NPI [3]).
of a well-matched pandemic vaccine; reduce the peak Studies on generic social distancing that did not spe-
number of daily influenza cases to decrease stress on the cifically mention workplace social distancing were also
health-care system and to protect critical infrastructure excluded. The primary outcomes of interest were the
(by reducing daily absenteeism rates); and reduce the following: cumulative influenza attack rate (percentage
overall number of influenza cases in order to decrease of individuals in a given population who will get in-
morbidity and mortality [3]. NPIs include personal pro- fluenza illness); peak influenza attack rate; time to
tective measures, environmental measures, and commu- peak; lost workdays; and harms.
nity measures aimed at increasing social distancing. Social
distancing can reduce virus transmission from infected Literature search strategy and study selection
persons to susceptible individuals by increasing physical Electronic searches of the published and grey literature
distance between people or reducing frequency of congre- were conducted using MEDLINE, Embase, Scopus,
gation in socially dense community settings, such as Cochrane Library, PsycINFO, Cumulative Index to Nurs-
schools or workplaces [3]. The US Pandemic Influenza ing and Allied Health Literature (CINAHL), NIOSHTIC-
Plan as well as the World Health Organization Public 2, and EconLit to identify studies published in English
Health Research Agenda for Influenza have called for during the period January 1, 2000, through May 3, 2017.
more research on the effectiveness, timing, and optimal The search terms are provided in PROSPERO [10].
implementation of social distancing measures in different Two reviewers (FA and NZ) independently identified
community settings [4, 5]. Research on the effectiveness of eligible articles by screening titles and abstracts and
social distancing has focused on schools, most notably on reviewing full-text articles. The reference lists of in-
pre-emptive school closures, for which systematic reviews cluded studies were examined to search for add-
have been published [6, 7]. itional studies.
Of the US civilian noninstitutionalized population
aged ≥16 years, about two-thirds participate in the Data extraction and risk of bias assessments
labor force [8]. The influenza illness attack rate in the Two reviewers (FA and NZ) extracted data independ-
workplace in a severe pandemic can be over 20% [1]. ently from all included studies using a standard form
Contacts made in the workplace represent 20–25% of that was piloted. Variables for which data were sought
all weekly contacts, and influenza transmission in the included the following: study dates, study design, pre-
workplace represents on average 16% (range 9–33%) dominant influenza strain, threshold for triggering so-
of all transmissions [9]. Social distancing measures in cial distancing, basic reproduction number (R0),
non-healthcare workplaces can include increased use population characteristics, type of intervention (in-
of telecommuting and remote-meeting options, stag- cluding duration of intervention), comparator, type of
gered work hours, and spacing workers further apart outcome measures, setting, publication status, and
[3]. The objective of this systematic review is to as- funding source. Two reviewers (FA and NZ) inde-
sess the evidence that social distancing interventions pendently assessed the quality of epidemiological
in non-healthcare workplaces, compared to no inter- studies using the Risk of Bias in Non-randomized
vention, reduce or slow influenza transmission among Studies of Interventions (ROBINS-I) tool [12]. Risk of
workers and in the general population. bias for each domain is classified into four categories:
low (study is comparable to a well performed ran-
Methods domized trial), moderate (study is sound for a non-
The protocol for this systematic review was registered randomized study but cannot be considered compar-
on PROSPERO, an international prospective register able to a well performed randomized trial), serious
of systematic reviews (ID # CRD42017065310) [10]. (study has some important problems), and critical
The systematic review was conducted following the (study is too problematic to provide any useful evi-
Preferred Reporting Items for Systematic Reviews and dence on the effects of intervention). A particular
Meta-Analysis (PRISMA) statement (see Add- level of risk of bias for an individual domain means
itional file 1) [11]. The inclusion criteria included that the overall risk of bias for the study is at least
randomized controlled trials, epidemiological studies, this severe. Any disagreements were resolved through
and modeling studies reporting results of social discussion or a third reviewer (AU). The quality of
distancing interventions in non-healthcare workplaces. modeling studies was not assessed.
The exclusion criteria included the following: review
articles, commentaries, and editorials; studies in ani- Data synthesis
mals; studies conducted in health-care, school, or uni- Percentage reductions were calculated using the following
versity settings; and studies on workplace closure formula: Percentage reduction = ((Attack rate in the
Ahmed et al. BMC Public Health (2018) 18:518 Page 3 of 13

absence of intervention – Attack rate with intervention) epidemiological [27–29] and 12 modeling [30–40], met
/Attack rate in the absence of intervention) × 100 [6]. A the inclusion criteria. Nine studies were from North
narrative synthesis was performed [13]. America, four from Asia, one from Europe, and one from
It was decided a priori to present results by basic Australia (Appendix). The funding sources of the studies
reproduction number (R0), a measure of virus trans- were government (10 studies), university (2 studies),
missibility. R0 is defined as the average number of research council (1 study), industry (1 study), and none/
secondary cases produced by a typical infectious case unknown (1 study).
in a fully susceptible population [14]. A reproduction Social distancing measures in the epidemiological
number greater than 1 indicates that the infection studies included segregation of persons into small
will grow in the population, whereas a value less than subgroups and working from home (Appendix).
1 indicates that the infection will decline [14]. Higher These studies showed reductions in seroconversion
R0 values are associated with higher cumulative attack to 2009 influenza A (H1N1), occurrence of
rates [15]. Factors that affect R0 include the popula- influenza-like illness (ILI), and workplace attendance
tion contact rate, the probability of infection per con- with severe ILI (which would result in reduced
tact, and the duration of illness. The results are transmission) (Table 1). The overall risk of bias in
presented using three R0 categories: ≤ 1.9; 2.0–2.4; the epidemiological studies was serious in two stud-
and ≥ 2.5 [6]. The R0 of the 1918 influenza pandemic ies and critical in one study (Table 2). All three
was somewhat higher than those of the 1957, 1968, studies had moderate or serious risk of bias in the
and 2009 pandemics, but the R0 values of all four confounding domain, and two studies had moderate
pandemics were estimated to be less than 1.9 [14]. risk of bias in the outcome measurement domain. In
The characteristics of influenza pandemics are unpre- addition, because the outcomes used in these studies
dictable, and the higher R0 categories provide esti- were surrogates for influenza illness, the evidence
mates for an atypical pandemic. was indirect [41].
Among the modeling studies, the most frequent workplace
Results social distancing measure assessed was reduction in work-
The selection of eligible studies is shown in the PRISMA place contacts by 50% for the entire duration of the outbreak
flow diagram (Fig. 1). The database search identified 4743 (Appendix). One study assessed the effect of extending the
records. After removal of duplicates, 3421 records were weekend. Several studies assessed the effect of combining
screened. Among the excluded studies, 10 included work- workplace social distancing measures with other interven-
place closure and one did not include a “no intervention” tions, including school closure, community contact reduc-
comparator [16–26]. Fifteen studies, representing three tion, antiviral treatment and prophylaxis, and vaccination.

Fig. 1 PRISMA flow diagram of study selection

Ahmed et al. BMC Public Health (2018) 18:518 Page 4 of 13

Table 1 Percentage reduction in cumulative influenza attack rate, 2000–2017

First author, year published Country Influenza strain Interventionb Percentage reductiona
R0 ≤ 1.9 R0 = 2.0–2.4 R0 ≥ 2.5
Epidemiological studiesc
Rousculp, 2010 [27] USA Seasonal influenza A(H3N2), 2007– Single 20 – –
2008
Kumar, 2012 [28] USA 2009 influenza A(H1N1) pandemic Single 36 – –
Lee, 2010 [29] Singapore 2009 influenza A(H1N1) pandemic Multiple 61 – –
Modeling studies
Timpka, 2016 [30] Sweden Future pandemic strain Single 12d – –
Zhang, 2012 [31] Singapore Not reported Single 18 – –
Mao, 2011 [32] USA Seasonal scenario (R0 = 1.4) and Single 82 23 –
a pandemic scenario (R0 = 2.0)
Xia, 2013 [33] China 2009 influenza A(H1N1) pandemic Single – – –
Single + VAC – – –
Milne, 2008 [34] Australia Pandemic strain Single 28 13 7
Multiple 94 96 95
Milne, 2013 [35] Papua New Pandemic strain Multiple 63e – –
Guinea
Miller, 2008 [36] USA Influenza A(H3N2) in population Multiple 88 – –
with no prior immunity
Andradottir, 2011 [37] Canada 2009 influenza A(H1N1) pandemic Multiple 30 – –
Multiple + 61 – –
VAC
Multiple + AV 73 – –
Perlroth, 2010 [38] USA Not reported Multiple 77 38 –
Multiple + AV 90 71 –
Halloran-Imperial/Pitt model, 2008 [39] USA Future pandemic strain Multiple 73 – –
Multiple + AV 83 70 53
Halloran-UW/LANL model, 2008 [39] USA Future pandemic strain Multiple 89 – –
Multiple + AV 94 92 86
Halloran-VBI model, 2008 [39] USA Future pandemic strain Multiple 72 – –
Multiple + AV 91 81 64
Abbreviations: ILI influenza-like illness, R0 basic reproduction number, Imperial/Pitt Imperial College and the University of Pittsburgh, UW/LANL University of Washington
and Fred Hutchinson Cancer Research Center in Seattle and the Los Alamos National Laboratories, VBI Virginia Bioinformatics Institute of the Virginia Polytechnical
Institute and State University
a
Percentage reduction = ((Attack rate in the absence of intervention – Attack rate with intervention) / Attack rate in the absence of intervention) × 100. Unless
otherwise stated, percentage reduction applies to the intervention group in the epidemiological studies and to the general population in the modeling studies
b
Single: Workplace social distancing (e.g., working from home, reduction in workplace contacts by 50%); Multiple: Workplace social distancing and other
nonpharmaceutical interventions; AV: Antiviral treatment and prophylaxis; VAC: Vaccination
c
Outcomes are surrogates for influenza: Rousculp - Attended work with severe ILI; Kumar - ILI; Lee - Seroconversion to 2009 influenza A(H1N1)
d
Reduction in cumulative influenza attack rate in the workplace = 58%
e
Reduction in cumulative influenza attack rate in the workplace = 81%

For studies modeling R0 ≤ 1.9, workplace social Subgroup analysis reported in two studies indicated that
distancing measures alone (single intervention) showed the percentage reduction was higher in workplaces than
a median reduction of 23% in the cumulative influenza in the general population (Table 1 footnote).
attack rate in the general population (Table 1). The modeling studies reported that percentage
Workplace social distancing measures combined with reduction in cumulative influenza attack rate in the gen-
other nonpharmaceutical interventions showed a median eral population declined with higher R0 values (Table 1).
reduction of 75% in the general population. Adding anti- The percentage reduction declined with increasing
viral treatment and prophylaxis further reduced the in- threshold for triggering interventions or with delayed
fluenza attack rate (median reduction = 90%) (Table 1). implementation of interventions (Table 3). The
Ahmed et al. BMC Public Health (2018) 18:518 Page 5 of 13

Table 2 Risk of bias in epidemiological studies of workplace social distancing, 2000–2017*

First author, Outcome Confounding Selection Intervention Intervention Missing Outcome Reported Overall
year published classification deviations data measurement results
Rousculp, 2010 [27] Attend work with Moderatea Low Low Low Low Moderateb Seriousc Serious
severe ILI
Kumar, 2012 [28] ILI Seriousd Low Moderatee Low Moderatef Moderateb Low Critical
d
Lee, 2010 [29] Seroconversion to Serious Low Low Low Low Low Low Serious
2009 influenza A(H1N1)
Abbreviations: ILI influenza-like illness
*Assessed using the Risk of Bias in Epidemiological Studies of Interventions (ROBINS-I) tool. Risk of bias for each domain is classified into four categories: low
(study is comparable to a well performed randomized trial), moderate (study is sound for a non-randomized study but cannot be considered comparable to a well
performed randomized trial), serious (study has some important problems), and critical (study is too problematic to provide any useful evidence on the effects
of intervention)
a
A nonrandomized study is rarely at low risk of bias for confounding
b
Subjective outcome self-reported by participants who were aware of the intervention group
c
Results for attending work with ILI symptoms of any severity are not reported
d
Inadequate or no adjustment
e
Intervention status was determined retrospectively
f
Response rate was 56%

percentage reduction also declined with lower compli- to reduce the effective reproduction number to below
ance to workplace social distancing interventions one if R0 is higher [6]. The lower effectiveness with de-
(Table 4). layed triggering or lower compliance may be due to sev-
The percentage reduction in the peak daily attack eral factors. Delayed triggering of workplace social
rate was reported in five modeling studies (Table 5). distancing precludes the opportunity to impact cases
These studies showed substantial effects in reducing that have already occurred and represents a missed op-
the peak rate (median reduction for workplace social portunity to diminish further transmission. Lower com-
distancing alone = 45%). The time to influenza peak pliance increases the opportunity for person-to-person
was reported in one epidemiological and four model- transmission.
ing studies (Table 6). These studies reported later This systematic review has several potential limita-
peaks with intervention compared to no intervention tions. First, most of the included studies were based on
(median delay to peak for workplace social distancing modeling and few were in actual settings. Models can fill
alone = 6 days). gaps when decisions must be made when there is a pau-
city of information [47]. However, more epidemiological
Discussion studies are needed on social distancing in actual settings.
Epidemiological and modeling studies indicated that Second, we did not assess the quality of the modeling
workplace social distancing reduced the overall number studies. Input parameters used in simulation models in-
of influenza cases. It also reduced and delayed the influ- clude the population characteristics that describe expos-
enza peak. The modeling studies reported that the re- ure points (e.g., households, schools, workplaces); the
duction in influenza cases was more pronounced when population’s behaviors that represent exposure frequen-
workplace social distancing was combined with other in- cies (e.g., contact rates and durations); and disease trans-
terventions. However, the effectiveness was estimated to mission parameters [15, 48]. There are few empirical
decline with higher R0 values, delayed triggering of studies on contact rates at workplaces [30]. No studies
workplace social distancing, or lower compliance. provided empirical information regarding the impact of
Droplets, and possibly aerosols, generated by coughs workplace social distancing measures on changing work-
and sneezes are a major source of influenza transmission place contact rates. Third, the studies included did not
[42–45]. Social distancing in workplaces can decrease report the effects of workplace social distancing on two
the risk of person-to-person influenza transmission by of our primary outcomes of interest (lost workdays,
reducing droplet transmission that occurs within 3–6 ft harms). The impact on lost workdays would represent
[43]. Workplace social distancing and other nonpharma- the balance between potential work loss associated with
ceutical or pharmaceutical interventions implemented social distancing (which can be mitigated by the ability
together can act in complementary ways to reduce virus to work from home) and sick days averted by reduction
transmission [3, 46]. Social distancing was estimated to in influenza transmission and illness. One study reported
be less effective for higher R0 values. The lower effective- that a lower proportion of Hispanic and African Ameri-
ness could be because social distancing may be less likely can workers than of white workers are able to work at
Ahmed et al. BMC Public Health (2018) 18:518 Page 6 of 13

Table 3 Percentage reduction in cumulative influenza attack rate in the general population, by threshold for triggering intervention,
modeling studies, 2000–2017
First author, year published Interventiona Threshold (%)b Percentage reductionc
R0 ≤ 1.9 R0 = 2.0–2.4 R0 ≥ 2.5
Zhang, 2012 [31] Single 0.02 18 – –
0.25 18 – –
1.5 18 – –
5.0 17 – –
Halloran-Imperial/Pitt model, 2008 [39] Multiple + AV 0.0001 99 96 64
0.001 99 95 64
0.01 99 94 64
0.1 97 88 62
1.0 83 70 53
10.0 31 27 23
Halloran-UW/LANL model, 2008 [39] Multiple + AV 0.0001 99 99 99
0.001 99 99 99
0.01 99 99 99
0.1 99 99 98
1.0 94 92 86
10.0 57 54 47
Halloran-VBI model, 2008 [39] Multiple + AV 0.0001 96 89 67
0.001 96 89 67
0.01 96 89 67
0.1 96 88 66
1.0 91 81 64
10.0 55 49 50
Milne, 2008 [34, 40] Single Prior to first case 28 – –
2 weeks after 1st case 27 – –
4 weeks after 1st case 25 – –
6 weeks after 1st case 19 – –
Multiple Prior to first case 94 – 95
2 weeks after 1st case 94 – 89
4 weeks after 1st case 86 – 29
6 weeks after 1st case 73 – 1
Milne, 2013 [35] Multiple Immediately after 1st case 63 – –
2 weeks after 1st case 63 – –
4 weeks after 1st case 48 – –
Abbreviations: R0 basic reproduction number, Imperial/Pitt Imperial College and the University of Pittsburgh, UW/LANL University of Washington and Fred
Hutchinson Cancer Research Center in Seattle and the Los Alamos National Laboratories, VBI Virginia Bioinformatics Institute of the Virginia Polytechnical Institute
and State University
a
Single: Workplace social distancing; Multiple: Workplace social distancing and other nonpharmaceutical interventions; AV: Antiviral treatment and prophylaxis
b
Threshold percent: Cumulative influenza illness attack rate in the general population that will trigger intervention
c
Percentage reduction = ((Attack rate in the absence of intervention – Attack rate with intervention) / Attack rate in the absence of intervention) × 100

home [28]. This observation indicates the need to consider magnitude of impact in a future pandemic. Finally, only one
the potential for racial or ethnic disparities. Fourth, because of the included studies represented a lower-income country
the effectiveness of workplace social distancing would de- setting [35]. The findings of our evidence synthesis may not
pend on many factors, including R0, timing of implementa- be generalizable to lower-income countries that differ in
tion, and compliance, it is difficult to estimate the likely demography and contact patterns.
Ahmed et al. BMC Public Health (2018) 18:518 Page 7 of 13

Table 4 Percentage reduction in cumulative influenza attack rate in the general population, by compliance with intervention,
modeling studies, 2000–2017
First author, Interventiona Compliance Percentage reductionb
year published (%)
R0 ≤ 1.9 R0 = 2.0–2.4 R0 ≥ 2.5
Mao, 2011 [32] Single 100 82 23 –
90 61 20 –
75 41 16 –
50 22 9 –
Milne, 2008 [34] Single 100 28 – 7
90 26 – 7
75 25 – 5
50 17 – 2
Abbreviation: R0 basic reproduction number
a
Single: Workplace social distancing
b
Percentage reduction = ((Attack rate in the absence of intervention – Attack rate with intervention) / Attack rate in the absence of intervention) × 100

There were several strengths. We conducted a com- workplace closure and two included workplace social
prehensive search of the literature that focused on work- distancing), concluded that combination strategies in-
place social distancing. The studies included in our creased the effectiveness of individual strategies.
review assessed the effect of workplace social distancing Other systematic reviews that have assessed the ef-
measures alone or combined with other interventions, fectiveness of interventions in reducing pandemic in-
allowing assessment of the relative effectiveness of single fluenza transmission did not examine workplace
and combination interventions. social distancing measures [50].
To our knowledge, our systematic review is the first An increasing trend in the ability to telework aligns
one that focuses on workplace social distancing. A with recommendations for social distancing in a pan-
previous systematic review of modeling studies pub- demic, but teleworking is less feasible for many occu-
lished during 1990 to 2009 assessed the effectiveness pations [51]. About 24% of employed persons did
of pharmaceutical (vaccines and antiviral agents) and some or all of their work at home in 2015, ranging
nonpharmaceutical (case isolation, quarantine, per- from about 35% in managerial and professional occu-
sonal hygiene measures, social distancing, and travel pations to only 6% in production occupations. It is
restrictions) strategies for pandemic influenza re- estimated that it is possible for 50% of the US work-
sponse [49]. This previous review, which was based force to telework at least partially [52]. Studies that
on 19 articles (five of these articles included examine feasible and acceptable workplace social

Table 5 Percentage reduction in peak influenza attack rate in the general population, modeling studies, 2000–2017
First author, Country Influenza strain Interventiona Percentage reductionb
year published
R0 ≤ 1.9 R0 = 2.0–2.4 R0 ≥ 2.5
Zhang, 2012 [31] Singapore Not reported Single 28 – –
Mao, 2011 [32] USA Seasonal scenario Single 97 53 –
(R0 = 1.4) and a
pandemic scenario
(R0 = 2.0)
Xia, 2013 [33] China 2009 Influenza Single 51 – –
A(H1N1) pandemic
Single + VAC 91 – –
Milne, 2008 [34] Australia Pandemic strain Single 39 25 18
Multiple 97 99 99
Milne, 2013 [35] Papua Pandemic strain Multiple 91 – –
New Guinea
Abbreviations: R0 basic reproduction number, VAC vaccination
a
Single: Workplace social distancing; Multiple: Workplace social distancing and other nonpharmaceutical interventions; VAC: Vaccination
b
Percentage reduction = ((Attack rate in the absence of intervention – Attack rate with intervention) / Attack rate in the absence of intervention) × 100
Ahmed et al. BMC Public Health (2018) 18:518 Page 8 of 13

Table 6 Time to influenza peak, epidemiological and modeling studies, 2000–2017a

First author, year Country Influenza strain Interventionb Days to peak
published
R0 ≤ 1.9 R0 = 2.0–2.4 R0 ≥ 2.5
Lee, 2010 [29] Singapore 2009 influenza A Multiple Peak later (unspecified) with – –
(H1N1) pandemic intervention
Zhang, 2012 [31] Singapore Not reported Single Peak 1 day later with intervention – –
Mao, 2011 [32] USA Seasonal scenario (R0 = 1.4) Single Peak 89 days later with Peak 18 days later –
and a pandemic scenario intervention with intervention
(R0 = 2.0)
Xia, 2013 [33] China 2009 influenza A(H1N1) Single Peak 6 days later with intervention – –
pandemic
Milne, 2013 [35] Papua New Pandemic strain Multiple Peak 13 days later with intervention – –
Guinea
Abbreviations: R0 basic reproduction number
a
For the modeling studies (Zhang [31], Mao [32], Xia [33], Milne [35]), time to influenza peak is reported for the general population
b
Single: Workplace social distancing; Multiple: Workplace social distancing and other nonpharmaceutical interventions

distancing strategies in a variety of work settings are definition (laboratory testing of all ill patients may
needed to improve pandemic preparedness. Because not be possible), the representativeness of the re-
contact patterns differ in different types of indus- ported cases, and the completeness of reporting. Al-
tries and workplaces, the impact of social distancing gorithms for estimating the total number of cases in
in various settings needs to be assessed. Further a community based on cases detected by local surveil-
research is needed to facilitate development of lance systems, or use of proxy measures, may be
higher-fidelity models of influenza transmission in needed to guide triggering decisions. Decision-makers
the workplace for model-based evaluation of NPI ef- should weigh the benefits versus disruptions of imple-
fects in different industries. Cross-sectional epi- menting workplace social distancing measures in the
demiological studies can be used to assess the context of pandemic severity [46]. Third, effectiveness
prevalence of workplace social distancing measures, declined with lower compliance. Triggering social dis-
but this design is not optimal to assess effectiveness tancing too early can contribute to lower compliance
because of inherent biases [53]. Because randomiz- because of intervention fatigue [54]. Finally, effective-
ing employers or employees to social distancing or ness was reported to be greater when workplace
control arms may not be feasible, prospective co- social distancing was combined with other nonphar-
hort studies may provide the best available evidence maceutical or pharmaceutical interventions. The
on effectiveness. Employers that have implemented findings underscore the importance for coordination
workplace social distancing measures could be com- between employers and state/local health departments
pared to those that have not, particularly during a to potentially enhance impact using a combination of
pandemic. However, attention should be paid to col- measures.
lecting data on potential confounding variables and
using outcomes that are defined in an objective man- Conclusions
ner (e.g., laboratory-confirmed influenza illness, sensi- Our systematic review shows that there are few epi-
tive and specific case definitions using electronic demiological studies in actual settings. More research
medical records data). is needed to assess the effectiveness of social distan-
Our findings have several implications. First, the ef- cing measures in a variety of industries and work
fectiveness of workplace social distancing was esti- settings. The included epidemiological and modeling
mated to decline with higher R0 values. This finding studies indicate that social distancing in non-
has ramifications regarding the intensity of commu- healthcare workplaces reduces the overall as well as
nity mitigation measures that may be needed in atyp- the peak number of influenza cases. It also delays
ical pandemics with higher R0 values. Second, the influenza peak. The finding that reduction in in-
effectiveness declined with delayed triggering. The fluenza cases is more pronounced when workplace
ability of local surveillance systems to accurately de- social distancing is combined with other nonpharma-
tect influenza circulation in the community to inform ceutical or pharmaceutical interventions highlights
triggering decisions will depend on several factors, in- the importance of using a combination of measures
cluding the sensitivity and specificity of the case to reduce the transmission of pandemic influenza.
Appendix
Table 7 Description of studies included in a review of effectiveness of workplace social distancing to reduce influenza transmission, 2000–2017
First author, year Study design Influenza strain and Population, setting, and Intervention(s) and comparatora Relevant outcomes
published transmissibility (R0) number of people (n)
Epidemiological studies
Rousculp, 2010 [27] Cohort (participants surveyed Seasonal influenza Employees of three large Single: Can work from home Attended work for ≥1
at baseline and monthly from A(H3N2), 2007–2008 US companies (national Comparator: Cannot work from home day while ILI symptoms
October 2007 to April 2008) retail chain, transportation, were most severe
and manufacturing) (n = 793 (which would result
employees with ILI) in transmission to
co-workers)
Kumar, 2012 [28] Cross-sectional (survey completed 2009 influenza Random sample of US adults Single: Can work from home ILI during April 2009
from January 22 to February 1, 2010) A(H1N1) pandemic from the Knowledge Networks Comparator: Cannot work from home to date survey
online research panel (n = 2079) completed
Ahmed et al. BMC Public Health (2018) 18:518

Lee, 2010 [29] Cohort (June 22 to October 9, 2009) 2009 influenza Singapore military personnel Multiple: Standard pandemic plan Seroconversion to
A(H1N1) pandemic (n = 907) (provided general health education on 2009 influenza
respiratory and hand hygiene and A(H1N1); time to
advised to seek medical care if ill) + peak (based on
segregation of units into subgroups as onset of symptoms
small as 20 individuals (including among those
having different activity and meal times, who seroconverted)
and times of entry and exit from camp)
+ daily temperature and symptom
monitoring with provision of home
medical leave
Comparator: Standard pandemic plan
Modeling studies
Timpka, 2016 [30] Model based on an ontology Future pandemic strain General population in Single: Social distancing that decreases Cumulative influenza
system. Mean duration of Linkoping municipality, workplace influenza virus transmission attack rate
outbreak in the reference Sweden (n = 136,000) probability by 50%
model = 92 days Comparator: No intervention
Zhang, 2012 [31] Agent-based model. Simulated R0 = 1.9 General population in Single: Team-based rotational workforce Cumulative and peak
for 200 days. Intervention is Singapore (n = 480,000, shift for 6 weeks (i.e., each company or influenza attack rates;
triggered at a threshold of 0.02% representing a 10% sample) institution splits its employees into two peak attack day
work teams and minimizes contacts
between the teams through 7-day
rotations of staying at home or attending
work)
Comparator: No intervention
Mao, 2011 [32] Agent-based model. Simulated Seasonal scenario General population in Single: Weekend extension by 3 days Cumulative influenza
for 200 days. Intervention is (R0 = 1.4) and a pandemic Buffalo, New York, USA (Sat, Sun, Mon, Tues, Wed). attack rate; daily new
triggered at a threshold of 0.1% scenario (R0 = 2.0) (n = 985,001) Comparator: No weekend extension cases
(Sat, Sun)
Xia, 2013 [33] Compartmental model. Simulated 2009 influenza A(H1N1) General population in Hong Single: Workplace contact reduction Peak infectious
for 200 days pandemic. R0 = 1.5 Kong, China (n = 7 million) Single + Vaccination: Workplace contact population; days to
reduction + vaccination with a coverage peak
of 2.5%
Page 9 of 13
Table 7 Description of studies included in a review of effectiveness of workplace social distancing to reduce influenza transmission, 2000–2017 (Continued)
First author, year Study design Influenza strain and Population, setting, and Intervention(s) and comparatora Relevant outcomes
published transmissibility (R0) number of people (n)
Comparator: No intervention
Milne, 2008 [34, 40] Agent-based model. Intervention Pandemic strain. General population in Albany, Single: Workplace nonattendance Cumulative and peak
is triggered before the introduction R0 = 1.5, 2.0, 2.5 Australia (n = 30,000) (each person attending a workplace influenza attack rates
of the first infected case has a 50% chance each day of staying
home instead of attending the workplace)
Multiple: Workplace nonattendance +
school closure + case isolation + community
contact reduction
Comparator: No intervention
Milne, 2013 [35] Agent-based model. Intervention Pandemic strain. General population in Madang, Multiple: Workplace nonattendance (each Cumulative influenza
is triggered 2 weeks after the R0 = 1.88 Papua New Guinea (n = 35,000) person attending a workplace has a 50% attack rate; daily
first case chance each day of staying home instead incident cases
Ahmed et al. BMC Public Health (2018) 18:518

of attending the workplace) + community

contact reduction (50%) + school closure
Comparator: No intervention
Miller, 2008 [36] Discrete event model. Intervention Influenza A(H3N2) General population in San Multiple: Reducing the number of Cumulative influenza
is triggered after the first infection in population with Antonio, Texas, USA (n = 1.4 community contacts to 50% by day 21 attack rate
in city no prior immunity million) after first infection (representing closing
schools and churches, banning public
gatherings, and encouraging people to
work from home)
Comparator: No intervention
Andradottir, 2011 [37] Individual-based compartmental 2009 influenza A General population in Hamilton, Multiple: Social distancing (20% reduction Cumulative influenza
model. Simulated for 180 days. (H1N1) pandemic. Canada (n = 649,565) in workplace and general community attack rate
Intervention is triggered at a R0 = 1.4 contacts) + school closure
threshold of 0.01% Multiple + Vaccination: Social distancing
+ school closure + vaccination of up to 35%
of the population with a vaccine with 40%
vaccine efficacy (60-day delay in vaccination
after trigger threshold is reached)
Multiple + Antiviral: Social distancing +
school closure + antiviral treatment and
prophylaxis of up to 10% of the population
Comparator: No intervention
Perlroth, 2010 [38] Agent-based model. Intervention R0 = 1.6, 2.1 General population in a small Multiple: Adult social distancing (work Cumulative influenza
is triggered at a threshold of 0.1% US town (n = 10,000) contacts reduced by 50%) + child social attack rate
distancing + school closure
Multiple + Antiviral: Adult social distancing
+ child social distancing + school closure +
antiviral treatment and prophylaxis
Comparator: No intervention
Halloran (Imperial-Pitt Agent-based model. Results at day Future pandemic General population in Chicago, Multiple + Antiviral: Workplace social Cumulative influenza
model), 2008 [39] 180 of the epidemic are reported. strain. R0 = 1.9, 2.4, 3.0 Illinois, USA (n = 8.6 million) distancing (workplace contacts reduced attack rate
Intervention is triggered at a by 50%) + community social distancing +
threshold of 1% school closure + interventions within the
households of ascertained influenza cases
Page 10 of 13
Table 7 Description of studies included in a review of effectiveness of workplace social distancing to reduce influenza transmission, 2000–2017 (Continued)
First author, year Study design Influenza strain and Population, setting, and Intervention(s) and comparatora Relevant outcomes
published transmissibility (R0) number of people (n)
(antiviral treatment and prophylaxis, home
isolation of cases, and quarantine of
household contacts)
Multiple: Above, excluding antiviral treatment
and prophylaxis
Comparator: No intervention
Halloran (UW/LANL Agent-based model. Results at day Future pandemic General population in Chicago, Multiple + Antiviral: Workplace social Cumulative influenza
model), 2008 [39] 180 of the epidemic are reported. strain. R0 = 2.1, 2.4, 3.0 USA (n = 8.6 million) distancing (workplace contacts reduced by attack rate
Intervention is triggered at a 50%) + community social distancing + school
threshold of 1% closure + interventions within the households
of ascertained influenza cases (antiviral treatment
and prophylaxis, home isolation of cases, and
Ahmed et al. BMC Public Health (2018) 18:518

quarantine of household contacts)

Multiple: Above, excluding antiviral treatment
and prophylaxis
Comparator: No intervention
Halloran (VBI model), Agent-based model. Results at day Future pandemic strain. General population in Chicago, Multiple + Antiviral: Workplace social distancing Cumulative influenza
2008 [39] 180 of the epidemic are reported. R0 = 2.1, 2.4, 3.0 USA (n = 8.6 million) (workplace contacts reduced by 50%) + community attack rate
Intervention is triggered at a social distancing + school closure + interventions
threshold of 1% within the households of ascertained influenza
cases (antiviral treatment and prophylaxis, home
isolation of cases, and quarantine of household
contacts)
Multiple: Above, excluding antiviral treatment and
prophylaxis
Comparator: No intervention
Abbreviations: ILI influenza-like illness, R0 basic reproduction number, Imperial/Pitt Imperial College and the University of Pittsburgh, UW/LANL University of Washington and Fred Hutchinson Cancer Research Center in
Seattle and the Los Alamos National Laboratories, VBI Virginia Bioinformatics Institute of the Virginia Polytechnical Institute and State University
a
Unless otherwise stated, the modeling studies assumed that the duration of social distancing in workplaces was for the entire influenza outbreak period
Page 11 of 13
Ahmed et al. BMC Public Health (2018) 18:518 Page 12 of 13

Additional files Received: 18 January 2018 Accepted: 11 April 2018

Additional file 1: Table S1. PRISMA 2009 checklist. (DOC 63 kb)

Additional file 2: Table S2. Data for percentage reduction in References
cumulative influenza attack rate. (XLSX 13 kb) 1. Reed C, Biggerstaff M, Finelli L, Koonin LM, Beauvais D, Uzicanin A, et al.
Novel framework for assessing epidemiologic effects of influenza epidemics
Additional file 3: Table S3. Data for percentage reduction in
and pandemics. Emerg Infect Dis. 2013;19:85–91.
cumulative influenza attack rate in the general population, by threshold
2. Meltzer MI, Cox NJ, Fukuda K. The economic impact of pandemic influenza
for triggering intervention. (XLSX 14 kb)
in the United States: priorities for intervention. Emerg Infect Dis. 1999;5:659–
Additional file 4: Table S4. Data for percentage reduction in 71.
cumulative influenza attack rate in the general population, by compliance 3. Qualls N, Levitt A, Kanade N, Wright-Jegede N, Dopson S, Biggerstaff M,
with intervention. (XLSX 11 kb) et al. Community mitigation guidelines to prevent pandemic influenza -
Additional file 5: Table S5. Data for percentage reduction in peak United States, 2017. Morbidity & Mortality Weekly Report Recommendations
influenza attack rate in the general population. (XLSX 11 kb) & Reports. 2017;66:1–34.
4. U.S. Department of Health and Human Services. Pandemic Influenza Plan:
Additional file 6: Table S6. Data for time to influenza peak. (XLSX 9 kb)
2017 Update. 2017. https://www.cdc.gov/flu/pandemic-resources/pdf/pan-
flu-report-2017v2.pdf. Accessed 12 Oct 2017.
5. World Health Organization. WHO Public Health Research Agenda for
Abbreviations Influenza: 2017 Update. 2017. http://www.who.int/influenza/resources/
AV: Antiviral treatment and prophylaxis; CINAHL: Cumulative Index to research/publication_research_agenda_2017/en/. Accessed 6 Mar 2018.
Nursing and Allied Health Literature; ILI: Influenza-like illness; Imperial/ 6. Jackson C, Mangtani P, Hawker J, Olowokure B, Vynnycky E. The effects of
Pitt: Imperial College and the University of Pittsburgh; school closures on influenza outbreaks and pandemics: systematic review of
NPI: Nonpharmaceutical intervention; PRISMA: Preferred Reporting Items for simulation studies. PLoS One. 2014;9:e97297.
Systematic Reviews and Meta-Analysis; R0: Basic reproduction number; 7. The Community Guide. Emergency preparedness and response: school
US: United States; UW/LANL: University of Washington and Fred Hutchinson dismissals to reduce transmission of pandemic influenza. 2012. https://www.
Cancer Research Center in Seattle and the Los Alamos National Laboratories; thecommunityguide.org/findings/emergency-preparedness-and-response-
VAC: Vaccination; VBI: Virginia Bioinformatics Institute of the Virginia school-dismissals-reduce-transmission-pandemic-influenza. Accessed 29 Mar
Polytechnical Institute and State University 2018.
8. Bureau of Labor Statistics, U.S. Department of Labor. The Employment
Situation - August 2017. 2017. https://www.bls.gov/news.release/archives/
Acknowledgements empsit_09012017.htm. Accessed 12 Oct 2017.
The authors thank Joanna Taliano, librarian at the Stephen B. Thacker CDC 9. Edwards CH, Tomba GS, de Blasio BF. Influenza in workplaces: transmission,
Library, for her expert contribution to developing the syntax for the searches workers’ adherence to sick leave advice and European sick leave
and conducting electronic database searches. recommendations. Eur J Pub Health. 2016;26:478–85.
The findings and conclusions in this study are those of the authors and do 10. Ahmed F, Uzicanin A, Zviedrite N. Effectiveness of workplace social
not necessarily represent the official position of the Centers for Disease distancing measures in reducing influenza transmission: a systematic review.
Control and Prevention. PROSPERO: International prospective register of systematic reviews. 2017.
CRD42017065310. http://www.crd.york.ac.uk/PROSPERO/display_record.
Availability of data and materials asp?ID=CRD42017065310. Accessed 12 Oct 2017.
All data generated or analyzed during this study are included in this 11. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, et al.
published study and its Additional files (see Additional files 2, 3, 4, 5 and 6). The PRISMA statement for reporting systematic reviews and meta-analyses
of studies that evaluate health care interventions: explanation and
elaboration. PLoS Med. 2009;6:e1000100.
Authors’ contributions 12. Sterne JA, Hernan MA, Reeves BC, Savovic J, Berkman ND, Viswanathan M,
FA participated in all steps of the research and wrote the first draft of the et al. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of
manuscript. NZ participated in screening records, data extraction, quality interventions. BMJ. 2016;355:i4919.
assessment of epidemiological studies, and revising the manuscript. AU 13. Centre for Reviews and Dissemination. Systematic Reviews: CRD’s Guidance
participated in conceptualizing the review, resolving any questions about for Undertaking Reviews in Health Care. University of York. 2009. https://
data extraction, and revising the manuscript. All authors read and approved www.york.ac.uk/media/crd/Systematic_Reviews.pdf. Accessed 13 Oct 2017.
the final manuscript. 14. Biggerstaff M, Cauchemez S, Reed C, Gambhir M, Finelli L. Estimates of the
reproduction number for seasonal, pandemic, and zoonotic influenza: a
systematic review of the literature. BMC Infect Dis. 2014;14:480.
Author's information 15. Andradottir S, Chiu W, Goldsman D, Lee ML. Simulation of influenza
NZ is a fellow appointed through the Research Participation Program at the propagation: model development, parameter estimation, and mitigation
Centers for Disease Control and Prevention administered by the Oak Ridge strategies. IIE Transactions on Healthcare Systems Engineering. 2014;4:27–48.
Institute for Science and Education through an interagency agreement 16. Carrat F, Luong J, Lao H, Salle AV, Lajaunie C, Wackernagel H. A ‘small-
between the U.S. Department of Energy and the Centers for Disease Control world-like’ model for comparing interventions aimed at preventing and
and Prevention. controlling influenza pandemics. BMC Med. 2006;4:26.
17. Ciofi degli Atti ML, Merler S, Rizzo C, Ajelli M, Massari M, Manfredi P,
Furlanello C, et al. Mitigation measures for pandemic influenza in Italy: an
Ethics approval and consent to participate individual based model considering different scenarios. PLoS One. 2008;3:
Not applicable. e1790.
18. Ferguson NM, Cummings DA, Cauchemez S, Fraser C, Riley S, Meeyai A,
et al. Strategies for containing an emerging influenza pandemic in
Competing interests
Southeast Asia. Nature. 2005;437:209–14.
The authors declare that they have no competing interests.
19. Ferguson NM, Cummings DA, Fraser C, Cajka JC, Cooley PC, Burke DS.
Strategies for mitigating an influenza pandemic. Nature. 2006;442:448–52.
20. Halder N, Kelso JK, Milne GJ. Cost-effective strategies for mitigating a future
Publisher’s Note influenza pandemic with H1N1 2009 characteristics. PLoS One. 2011;6:e22087.
Springer Nature remains neutral with regard to jurisdictional claims in 21. Mao L. Predicting self-initiated preventive behavior against epidemics with
published maps and institutional affiliations. an agent-based relative agreement model. JASSS. 2015;18(4)
Ahmed et al. BMC Public Health (2018) 18:518 Page 13 of 13

22. Merler S, Jurman G, Furlanello C, Rizzo C, Bella A, Massari M, et al. Strategies 46. Barrios LC, Koonin LM, Kohl KS, Cetron M. Selecting nonpharmaceutical
for containing an influenza pandemic: the case of Italy. In: 2006 1st bio- strategies to minimize influenza spread: the 2009 influenza a (H1N1)
inspired models of network, information and computing systems, pandemic and beyond. Public Health Rep. 2012;127:565–71.
BIONETICS: 2006; 2006. 47. Fischer LS, Santibanez S, Hatchett RJ, Jernigan DB, Meyers LA, Thorpe PG,
23. Rizzo C, Lunelli A, Pugliese A, Bella A, Manfredi P, Tomba GS, et al. Scenarios et al. CDC grand rounds: modeling and public health decision-making.
of diffusion and control of an influenza pandemic in Italy. Epidemiol Infect. MMWR Morb Mortal Wkly Rep. 2016;65:1374–7.
2008;136:1650–7. 48. Prieto DM, Das TK, Savachkin AA, Uribe A, Izurieta R, Malavade S. A
24. Roberts MG, Baker M, Jennings LC, Sertsou G, Wilson N. A model for the systematic review to identify areas of enhancements of pandemic
spread and control of pandemic influenza in an isolated geographical simulation models for operational use at provincial and local levels. BMC
region. J R Soc Interface. 2007;4:325–30. Public Health. 2012;12:251.
25. Xia H, Nagaraj K, Chen J, Marathe MV. Synthesis of a high resolution social 49. Lee VJ, Lye DC, Wilder-Smith A. Combination strategies for pandemic
contact network for Delhi with application to pandemic planning. Artif Intell influenza response - a systematic review of mathematical modeling studies.
Med. 2015;65:113–30. BMC Med. 2009;7:76.
26. Martinez DL, Das TK. Design of non-pharmaceutical intervention strategies 50. Saunders-Hastings P, Reisman J, Krewski D. Assessing the state of
for pandemic influenza outbreaks. BMC Public Health. 2014;14:1328. knowledge regarding the effectiveness of interventions to contain
27. Rousculp MD, Johnston SS, Palmer LA, Chu BC, Mahadevia PJ, Nichol KL. pandemic influenza transmission: a systematic review and narrative
Attending work while sick: implication of flexible sick leave policies. J Occup synthesis. PLoS One. 2016;11:e0168262.
Environ Med. 2010;52:1009–13. 51. Bureau of Labor Statistics, U.S. Bureau of Labor. The Economics Daily, 24
28. Kumar S, Quinn SC, Kim KH, Daniel LH, Freimuth VS. The impact of percent of employed people did some or all of their work at home in 2015.
workplace policies and other social factors on self-reported influenza-like 2016. https://www.bls.gov/opub/ted/2016/24-percent-of-employed-people-
illness incidence during the 2009 H1N1 pandemic. Am J Public Health. did-some-or-all-of-their-work-at-home-in-2015.htm. Accessed 17 Feb 2017.
2012;102:134–40. 52. Global Workplace Analytics. Latest telecommuting statistics. 2017. http://
29. Lee V, Yap J, Cook AR, Chen M, Tay J, Barr I, et al. Effectiveness of public globalworkplaceanalytics.com/telecommuting-statistics. Accessed 12 Oct 2017.
health measures in mitigating pandemic influenza spread: a prospective 53. Levin KA. Study design III: cross-sectional studies. Evid Based Dent. 2006;7:24–5.
sero-epidemiological cohort study. J Infect Dis. 2010;202:1319–26. 54. Collinson S, Khan K, Heffernan JM. The effects of media reports on disease
30. Timpka T, Eriksson H, Holm E, Stromgren M, Ekberg J, Spreco A, et al. Relevance of spread and important public health measurements. PLoS One. 2015;10:
workplace social mixing during influenza pandemics: an experimental modelling e0141423.
study of workplace cultures. Epidemiol Infect. 2016;144:2031–42.
31. Zhang T, Fu X, Ma S, Xiao G, Wong L, Kwoh CK, et al. Evaluating temporal
factors in combined interventions of workforce shift and school closure for
mitigating the spread of influenza. PLoS One. 2012;7:e32203.
32. Mao L. Agent-based simulation for weekend-extension strategies to
mitigate influenza outbreaks. BMC Public Health. 2011;11:522.
33. Xia S, Liu J, Cheung W. Identifying the relative priorities of subpopulations
for containing infectious disease spread. PLoS One. 2013;8:e65271.
34. Milne GJ, Kelso JK, Kelly HA, Huband ST, McVernon J. A small community
model for the transmission of infectious diseases: comparison of school
closure as an intervention in individual-based models of an influenza
pandemic. PLoS One. 2008;3:e4005.
35. Milne GJ, Baskaran P, Halder N, Karl S, Kelso J. Pandemic influenza in Papua
New Guinea: a modelling study comparison with pandemic spread in a
developed country. BMJ Open. 2013;3(3)
36. Miller G, Randolph S, Patterson JE. Responding to simulated pandemic
influenza in San Antonio, Texas. Infection Control & Hospital Epidemiology.
2008;29:320–6.
37. Andradottir S, Chiu W, Goldsman D, Lee ML, Tsui KL, Sander B, et al.
Reactive strategies for containing developing outbreaks of pandemic
influenza. BMC Public Health. 2011;11(Suppl 1):S1.
38. Perlroth DJ, Glass RJ, Davey VJ, Cannon D, Garber AM, Owens DK. Health
outcomes and costs of community mitigation strategies for an influenza
pandemic in the United States. Clin Infect Dis. 2010;50:165–74.
39. Halloran ME, Ferguson NM, Eubank S, Longini IM Jr, Cummings DA, Lewis B,
et al. Modeling targeted layered containment of an influenza pandemic in
the United States. Proc Natl Acad Sci U S A. 2008;105:4639–44.
40. Kelso JK, Milne GJ, Kelly H. Simulation suggests that rapid activation of
social distancing can arrest epidemic development due to a novel strain of
influenza. BMC Public Health. 2009;9:117.
41. Guyatt GH, Oxman AD, Kunz R, Woodcock J, Brozek J, Helfand M, et al.
GRADE guidelines: 8. Rating the quality of evidence–indirectness. J Clin
Epidemiol. 2011;64:1303–10.
42. Brankston G, Gitterman L, Hirji Z, Lemieux C, Gardam M. Transmission of
influenza A in human beings. Lancet Infect Dis. 2007;7:257–65.
43. Bischoff WE, Swett K, Leng I, Peters TR. Exposure to influenza virus aerosols
during routine patient care. J Infect Dis. 2013;207:1037–46.
44. Cowling BJ, Ip DK, Fang VJ, Suntarattiwong P, Olsen SJ, Levy J, et al. Modes of
transmission of influenza B virus in households. PLoS One. 2014;9:e108850.
45. Yan J, Grantham M, Pantelic J, Bueno de Mesquita PJ, Albert B, Liu F, et al.
Infectious virus in exhaled breath of symptomatic seasonal influenza cases
from a college community. Proc Natl Acad Sci U S A. 2018;115:1081–6.