key: cord-1040901-lh0nwfqh authors: Ford, Nathan; Holmer, Haley K.; Chou, Roger; Villeneuve, Paul J.; Baller, April; Van Kerkhove, Maria; Allegranzi, Benedetta title: Mask use in community settings in the context of COVID-19: A systematic review of ecological data date: 2021-07-19 journal: EClinicalMedicine DOI: 10.1016/j.eclinm.2021.101024 sha: 1b9979a2954babcbcd18ede7663086cfc241bb1f doc_id: 1040901 cord_uid: lh0nwfqh BACKGROUND: The wearing of medical and non-medical masks by the general public in community settings is one intervention that is important for the reduction of SARS-CoV-2 transmission, and has been the subject of considerable research, policy, advocacy and debate. Several observational studies have used ecological (population-level) data to assess the effect of masks on transmission, hospitalization, and mortality at the region or community level. METHODS: We undertook this systematic review to summarize the study designs, outcomes, and key quality indicators of using ecological data to evaluate the association between mask wearing and COVID-19 outcomes. We searched the World Health Organization (WHO) COVID-19 global literature database up to 5 March 2021 for studies reporting the impact of mask use in community settings on outcomes related to SARS-CoV-2 transmission using ecological data. FINDINGS: Twenty one articles were identified that analysed ecological data to assess the protective effect of policies mandating community mask wearing. All studies reported SARS-CoV-2 benefits in terms of reductions in either the incidence, hospitalization, or mortality, or a combination of these outcomes. Few studies assessed compliance to mask wearing policies or controlled for the possible influence of other preventive measures such as hand hygiene and physical distancing, and information about compliance to these policies was lacking. INTERPRETATION: Ecological studies have been cited as evidence to advocate for the adoption of universal masking policies. The studies summarized by this review suggest that community mask policies may reduce the population-level burden of SARS-CoV-2. Methodological limitations, in particular controlling for the actual practice of mask wearing and other preventive policies make it difficult to determine causality. There are several important limitations to consider for improving the validity of ecological data. The wearing of medical and non-medical masks by the general public in community settings for prevention of respiratory virus transmission has been the subject of considerable debate, research, policy and advocacy. Since January 2020, WHO has recommended the use of masks as part of a comprehensive control strategy to suppress transmission of SARS-CoV-2 for health workers and the general public and this advice has been updated regularly [1] . At the start of the pandemic, much of the evidence supporting the protective benefit of mask wearing was derived from studies of other respiratory pathogens, in particular seasonal influenza, severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERs) [2À4] . A limited number of studies have reported outcomes on the wearing of face masks by the community to prevent SARS-CoV-2 transmission, including one randomized trial [5] and several comparative and non-comparative observational studies [6À8] . A larger number of observational studies have used ecological (population-level) data to assess the effect of masks on virus transmission at the region or group level [9] . Ecological studies are often conducted using routinely collected data, such as country or regional data and can provide some evidence on the effectiveness of policies on disease progression and other outcomes. The findings of ecological studies are often used to make strong causal inferences about individual and contextual effects. However, data aggregation can lead to cross-level bias À that average characteristics of the group apply to individuals (also known as the ecological fallacy) [10] . Due to known limitations, ecological studies have generally been excluded from systematic reviews of studies assessing community mask use [2, 8, 11] . However, ecological studies of mask policies may provide useful information, regardless of potential limitations, in an emergency event such as COVID-19, when there is low certainty evidence and end users seek to make rapid decisions based on all available evidence. We undertook this systematic review to summarize the study designs, outcomes, and key quality indicators of using ecological data to evaluate the association between mask wearing and COVID-19 outcomes. To our knowledge, this is the first systematic review of these studies. This study has been designed and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [12] . The protocol for this study, including search strategy, is available in the Supplementary appendix. Using a highly sensitive search strategy developed by an information specialist, we searched the WHO COVID-19 global literature database for studies reporting the impact of mask use in community settings on outcomes related to SARS-CoV-2 transmission using ecological data. The WHO COVID- 19 Research Database (https://search.bvsalud.org/global-liter ature-on-novel-coronavirus-2019-ncov/) was created in January 2020 in response to the need for a centralized repository of citations relevant to the current pandemic. The database is a compilation of over 40 searches conducted in resources such as PubMed, Embase, CINAHL, GIM. A full list of resources is listed here (https://www.who. int/docs/default-source/coronaviruse/who-covid-19-database/whocovid-19_sources_searchstrategy_20210105.pdf? sfvrsn=480292c0_9). Outcomes of interest included incidence of SARS-CoV-2, disease severity, and mortality. Studies that evaluate the effects of a masking policy on the population and do not assess both exposure and health outcomes at the individual level are considered ecological studies for this review. Searches were conducted by a single reviewer (NF) and verified by a second review (HH). Reference lists of published reviews were also screened. From a broad screen of all articles published on the use of masks to prevent SARS-CoV-2 transmission in community settings, we searched for studies that reported on the association between mask wearing and incidence, disease severity, and mortality at the ecological (population or aggregate) level; any aggregate level was included (e.g. state, region or country). The search was conducted from database inception (05 January 2020) to 05 March 2021. No language restrictions were applied. Pre-prints were excluded from review. Ecological measures can be classified into three types: aggregate (population-level) measures of health outcomes, environmental measures (i.e., physical characteristics of the place in which members of each group live or work) and global measures (attributes of groups or places for which there is no distinct analogue at the individual level, such as population density, level of social disorganization, or the existence of a specific law) [14] . For this review, all studies are considered as aggregate studies. Information on study characteristics, outcomes, and indicators of study quality were extracted and summarized. All extractions were sent to the original study authors for verification. There is no agreed checklist for assessing the methodological quality of ecological studies. Several papers have outlined issues with study designs commonly used to assess ecological data [15] , common sources of bias [14, 16, 17] , and reporting issues [18] . Considering these issues, we assessed risk of bias using an adapted version of the Newcastle Ottawa scale in order to evaluate eight key domains for evaluating ecological studies: representativeness of exposure group, ascertainment of exposure (including measurement of compliance with mask wearing), population exposed, comparability of groups, adjustment for confounders (including policy-level and/or community-level factors), outcome assessment, appropriateness of time lag between mask intervention and outcome assessment, and statistical methodology. There was no funding source for this study. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. From an initial screen of 4082 citations, we identified 37 published articles reporting outcomes on the use of masks for the Evidence before this study The wearing of medical and non-medical masks by the general public in community settings is an important intervention to reduce SARS-CoV-2 transmission. Much of the evidence supporting the protective benefit of mask wearing is derived from studies of other respiratory pathogens in health care settings, with only a limited number of studies reporting direct outcomes on the protective wearing of face masks by the community to prevent SARS-CoV-2 transmission. A larger number of studies have been published based on ecological data, but such studies are subject to important methodological limitations. This is the first review to systematically assess the available ecological (country or regional) data evaluating the association between mask wearing and COVID-19 outcomes and assess the quality of this body if evidence. All studies identified by this review reported a protective benefit in terms of either reduced incidence, mortality, hospitalization, or a combination of these outcomes. However, few studies provided any information about where masks were worn and by whom, type of mask (medical or non-medical), rate of mask wearing and level of compliance, and studies were limited in their ability to control for other infection control measures and confounders. The results of ecological studies in this review provide supportive evidence about the protective effect of mask wearing in community settings, and this body of evidence suggests that mask policies may reduce the population-level burden of SARS-CoV-2 in non-vaccinated populations. Future research should consider approaches to improving the validity of ecological data to inform policy and practice, in particular to be able to draw conclusions about the relative effectiveness of different nonpharmaceutical public health measures. prevention of COVID-19 in community settings. From this set of studies, 21 articles were identified that analysed ecological data to assess the protective effect of masks ( Fig. 1) [9,17,19À36] . All studies evaluated the effect of mask policies at the population level; only one study also used individual-level data to assess associations between self-reported maskwearing and SARS-CoV-2 incidence, assessed through a cross-sectional survey [29] . Five studies compared different countries [21, 23, 25, 28, 36] , six compared different regions within a single country [20, 22, 26 ,30À32], eight undertook a before-after comparison [17,24,25,27,30,34À37] , and one compared health workers and the general population [19] . Eleven studies compared the incidence of SARS-CoV-2 across different US States; four studies compared SARS-CoV-2 incidence over time within single US States; one study compared regions in Germany, and the rest compared different countries. Study characteristics are summarized in Table 1 . Studies used various study designs and analytic methods to assess associations between mask-wearing and COVID-19 outcomes. The majority of studies reported the influence of mask wearing policies on SARS-CoV-2 incidence; four studies reported incidence and mortality [20, 24, 30, 35] , and one study reported incidence, hospitalization and mortality [27] . Analytical approaches included univariate and multivariable analysis, synthetic control method, and interrupted time series analysis. All studies reported a rapid and significant reduction in incidence associated with mask wearing policies. A study from Jena city, Germany, reported a 15À75% reduction in SARS-CoV-2 infections over 20 days after mandatory introduction [9] . A study from 16 US jurisdictions estimated 230,000À450,000 averted cases over a 7 week period [17] . Another study, from Arizona, USA, reported a 75% reduction in the 7-day moving average of daily cases over a 4 week period [24] . Similarly, large reductions in adverse clinical outcomes were reported: a study from Delaware, USA, reported an 88% reduction in hospitalizations and a 100% reduction in mortality associated with stay-at-home orders, public mask mandates and contact tracing [27] . Key findings are provided in Table 1 . The methodological quality of the included studies was variable, and overall study quality was considered acceptable for contributing evidence to the broad question of protective efficacy of community mask wearing. The majority of studies carried out either no or limited adjustment for area-level confounders such as the use of other control measures in addition to the initiation of mask policies; the few studies that did attempt to perform statistical adjustment were limited in their ability to measure other infection control measures and confounders. Eleven studies defined an appropriate time interval between intervention (e.g. implementation of mask policy) and outcomes. Only three studies noted the limitation of cross-level bias, and that it is not possible to determine the effects of masking policies on individuals based on ecological data [18] ( Table 2) . The majority of studies provided no information about ascertainment of exposure (rate of mask wearing and level of compliance) or the population affected by the policies (where masks should be worn and by whom). This prevented any conclusions to be drawn about the effectiveness of different policies. Details of the quality assessment for the individual studies are provided in Table 3 . Masks have long been a key intervention for infection prevention and control. For SARS-CoV-2, WHO has consistently given advice on the use of masks for the general public. Guidance has evolved from advising mask wearing for people who were unwell and their care givers in January 2020 [38] , to guidance in April 2020 establishing policies of the use of masks by decisions makers, such as the use by the general public when individuals cannot physically distance [39] with additional guidance provided on specific settings in which to wear masks, and information on mask compositions in June [40] and December 2020 [1] . A limited number of systematic reviews have summarized the effectiveness of community mask wearing to protect against transmission of respiratory diseases at the individual level [2, 8, 11, 41] . A systematic review published in June 2020 concluded that face mask use could result in a large reduction in risk of infection but did not include any studies of community mask wearing for the prevention of COVID-19 [41] . A second systematic review, updated to April 2021, identified one randomized trial and 3 observational studies assessing community mask wearing for the prevention of COVID-19, and rated the strength of the evidence as low for a small reduction in risk for infection with any mask use [8] . This systematic review identified 21 studies that used ecological data to assess the effectiveness of mask wearing to protect against adverse health outcomes related to SARS-CoV-2 infections in community settings. All studies reported a protective benefit in terms of either reduced incidence, mortality, hospitalization, or a combination of these outcomes. This review summarized the literature up to March 2021, since which time a number of additional studies have reported a protective benefit impact of community mask wearing by the general public in community settings using ecological data [42À46] . While the results of these individual studies support the findings of this review, the large amount of research being done to assess community mask wearing warrants future systematic assessment. Studies of mask usage can try to measure two different effects: reduction of general population transmission or specific reduction of risk to an individual. The ecological fallacy is less of a concern in the former because many of these ecological studies focus on estimating population level effects regardless of individual risk. As such, the use of population level data is a feature as much as it is a limitation. WHO Guidelines Development Groups evaluate all forms of relevant available evidence. While ecological studies have been cited as evidence to advocate for the adoption of universal masking policies [47] , there are a number of important limitations for making causal inferences, in particular the ecological fallacy, or cross-level bias (Table 4 ) [16] . The GRADE framework, which is widely used to support guideline development [48] , is not well-suited to assess such studies. Due to their observational design and inherent limitations, [36] Colombia, Chile Region Country mask mandates Pre-post mandate comparison Incidence Mask wearing in public spaces reduced confirmed cases as indicated by difference between pre-and post-intervention slope aPR, adjusted prevalence ratio; aIRRR, adjusted ratio of incidence rate ratios. Mitze et al. [9] * À * * * * * * * * Lan et al. [19] * À * À À * * * Chernozhukov et al. [20] * À À * * * * * * Chiang et al. [21] * À À À À * À À Cheng et al. [22] * * * À * À À À À Bo et al. [23] * À À À ** * ** * Lyu and Wehby [17] * À * À ** * ** * Gallaway et al. [24] * À À À À * À À Leffler et al. [25] * À À À * * À * Van Dyke et al. [26] * À À À À * À * Kanu et al. [27] * À À À À * À À Zhang et al. [28] * À À À À * À À Zhang and Warner [33] * À À * * * * * À Rader et al. [29] * ** * * ** * ** * Li et al. [30] * À À * À * * * * Rebeiro et al. [31] * À À À * À * * Krishnamachari et al. [32] * À À À À * À * Joo et al. [37] * À * * * * ** * Dasgupta et al. [34] * À À * * * * À * Guy et al. [35] * À * * ** * * * Poppe [36] * À À * À * * * *Satisfactory. ** Good. a If studies chose a sample which were truly or somewhat representative of the average in the target population, we assigned 1 star. b If rate of mask wearing was assessed within the population, we assigned 1 star. If this included a measure of level of compliance, we assigned 2 stars. c If details about where masks should be worn and by whom, we assigned 1 star. d Where a comparison was made, the comparison group was appropriate (ie similar risk of outcome) or statistical adjustments were made, we assigned 1 star. e If other policy-level factors were controlled for (such as physical distancing, stay at home order, closure of public venues, restriction of gatherings), we assigned 1 star. If community level factors were controlled for (such as community prevalence and population size) we assigned 2 stars. f If study used case data corresponding to the target population, we applied 1 star. g If an appropriate lag time was incorporated to account for timing of effects of mask introduction and assessment outcome, we assigned 1 star. If a sensitivity analysis was conducted using a range of time lags, we assigned 2 stars. h If the statistical tests used to analyze the data was clearly described and appropriate, and the measurement of the association was presented with confidence intervals, we assigned 1 star. applying GRADE to such studies would result in very low strength of evidence assessments. An important limitation of the evidence base reviewed here is that in all but two studies [22, 29] , mask policies were assessed as the exposure variable, without providing information about compliance to these policies. Moreover, policies requiring people to wear masks in the community are variable in terms of where and when and in whom mask wearing is required. Few studies assessed the possible influence of concurrent implementation of other preventive measures such as hand hygiene, physical distancing, working from home policies, closures of businesses, and policies limiting gatherings. Where attempts were made to control for such confounders this was based on the existence of policies supporting such measures, with little information about compliance to these policies. Considering these limitations, it is challenging to disentangle the effectiveness of a single policy and draw conclusions about the superiority of one policy compared to another [49] . Ecological studies are also unable to account for rapid changes that impact transmission dynamics, such as the appearance of new variants of concern or the phased introduction of vaccines with unknown effectiveness at scale. To address these limitations, future research should consider approaches to improving the reliability of ecological data to inform policy and practice. The time interval between changes in mask policies/masking rates and assessment of outcomes is another important limitation. Trends may have already been observed at the time the policies were implemented and some findings may have been sensitive to the time periods selected for analysis. Future studies should account for the time element, for example by pre-defining the time periods assessed using plausible assumptions about the expected time that effects of mask policies would be expected, account for trends in infection rates when the mask policies are implemented, and perform sensitivity analyses on the periods selected for analysis to determine robustness of findings to assumptions regarding the temporal relationship. Finally, the existence of a policy alone provides no information about levels of compliance to the intervention, and future studies are encouraged to include an assessment of levels of compliance. It has been proposed that cross-level bias can be reduced by supplementing ecological with individual-level information [50] ; individual-level studies on masking generally support an association between mask use and decreased risk of infection in those wearing masks [8] . Certain study designs should be considered to improve the validity of assessing the effect of mask policies on COVID-19 outcomes. Longitudinal studies with individual-level data are an appropriate study design, as has been suggested for investigating the relationship between air pollution and increased risk of COVID-19 infection or adverse clinical outcomes [52] . Case-control studies could provide important information about risk estimates at an individual level, while controlling for the other relevant risk factors at both regional-and individual levels. Notwithstanding these limitations, where direct evidence based on individual-level data exist, the results of ecological studies can be considered to provide low certainty evidence about the protective effect of mask wearing at the community level. In conclusion, the studies summarized by this review suggest that mask policies may reduce the population-level burden of SARS-CoV-2. The appropriate and safe use of suitable masks as well as proper storage and cleaning or disposal of masks are essential to make them as effective as possible; and that the use of masks be supplemented by a comprehensive set of measures. All data are available in the original published articles included in this review, and are available upon request. There was no funding source for this study. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. All authors have nothing to declare. Table 4 Challenges in interpreting findings from ecological mask studies Diagnostic capabilities Accounting for changes in SARS-CoV-2 infection diagnostic capacity and types of diagnostic tools used is critical for interpreting changes in incidence associated with policy implementation [53] . Testing capacity was only assessed by one study included in this review [17] . More affluent regions may have a greater ability to detect COVID-19 cases than less affluent regions, including among asymptomatic individuals, and the practice of face mask wearing, including availability and quality, may also be related to area-level affluence (meaning that both exposure and outcome are influenced by affluence). Policies mandating community mask wearing are accompanied by policies promoting other nonpharmaceutical interventions known to prevent SARS-CoV-2 transmission À notably physical distancing, schools and workplace closures, and hand hygiene. Seventeen studies considered one or more nonpharmaceutical interventions in addition to mask wearing, with statistical adjustments made in nine of these studies [9, 17, 20, 23, 25, 29, 31, 33, 37] ; the remainder did not consider the influence of other nonpharmaceutical interventions. It is not possible to reliably estimate the benefit of face-mask policies over and above the benefits that are brought about by the other measures. Furthermore, the contribution of vaccines will need to be accounted for as vaccination programmes reach scale. Mask policy vs mask wearing Mask quality varies by type of mask fabric [54, 55] , and when and how they are worn [56] . Only 2 studies reported mask type [19, 21] and two considered adherence to mask policies [22, 29] . None of the studies evaluated individual-level adherence to and appropriate wearing of masks. The ability to categorize mask policies is relatively crude (i.e. requiring people to wear masks in the community could encompass a lot of variability in terms of where and when and in whom masks are required) None of these studies accounted for the potential contribution of superspreading events [57] or the emergence of new SARS-CoV-2 variants of concern that have higher transmissibility [58] on incidence trends. While most studies were completed prior to the emergence of variants of concern, and it is uncertain that the contribution of such variants on incidence trends could be assessed through ecological analyses. Timing from policy to outcome There is a time lag between policy implementation and possible impact on incidence, followed by hospitalization, and then mortality. Interpreting studies on the effect of mask and other policies requires appropriate consideration of timing. There is also variability in the requirements of mask policies, including differences in exemptions for certain age groups, places of worship, and other specific settings [59] . Mask use in the context of COVID-19: interim guidance Physical interventions to interrupt or reduce the spread of respiratory viruses Uptake and effectiveness of facemask against respiratory infections at mass gatherings: a systematic review Community use of face masks and similar barriers to prevent respiratory illness such as COVID-19: a rapid scoping review Effectiveness of adding a mask recommendation to other public health measures to prevent SARS-CoV-2 infection in Danish mask wearers Case-Control Study of Use of Personal Protective Measures and Risk for SARS-CoV 2 Infection Reduction of secondary transmission of SARS-CoV-2 in households by face mask use, disinfection and social distancing: a cohort study in Beijing, China Update alert 5: masks for prevention of respiratory virus infections, including SARS-CoV-2, in health care and community settings Face masks considerably reduce COVID-19 cases in Germany Ecologic versus individual-level sources of bias in ecologic estimates of contextual health effects Community use of face masks and similar barriers to prevent respiratory illness such as COVID-19: a rapid scoping review Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement Ecologic studies in epidemiology: concepts, principles, and methods Policy evaluation in COVID-19: a guide to common design issues Invited commentary: ecologic studiesÀbiases, misconceptions, and counterexamples Community use of face masks and COVID-19: evidence from a natural experiment of state mandates in the US The quality of modern cross-sectional ecologic studies: a bibliometric review Effects of universal masking on Massachusetts healthcare workers' COVID-19 incidence Causal impact of masks, policies, behavior on early covid-19 pandemic in the U.S The practice of wearing surgical masks during the COVID-19 pandemic The role of community-wide wearing of face mask for control of coronavirus disease 2019 (COVID-19) epidemic due to SARS-CoV-2 Effectiveness of non-pharmaceutical interventions on COVID-19 transmission in 190 countries from 23 Trends in COVID-19 Incidence After Implementation of mitigation measures -Arizona Association of country-wide coronavirus mortality with demographics, testing, lockdowns, and public wearing of masks Trends in county-level COVID-19 incidence in counties with and without a mask mandate -Kansas Declines in SARS-CoV-2 transmission, hospitalizations, and mortality after implementation of mitigation measures-Delaware Identifying airborne transmission as the dominant route for the spread of COVID-19 Mask-wearing and control of SARS-CoV-2 transmission in the USA: a cross-sectional study Evaluating the impact of New York's executive order on face mask use on COVID-19 cases and mortality: a comparative interrupted times series study The impact of state mask-wearing requirements on the growth of COVID-19 cases in the United States Online ahead of print) The role of mask mandates, stay at home orders and school closure in curbing the COVID-19 pandemic prior to vaccination COVID-19 policy differences across US States: shutdowns, reopening, and mask mandates Differences in rapid increases in county-level COVID-19 incidence by implementation of statewide closures and mask mandates -United States Association of state-issued mask mandates and allowing on-premises restaurant dining with county-level COVID-19 case and death growth rates -United States Impact of the healthcare system, macro indicator, general mandatory quarantine, and mask obligation on COVID-19 cases and death in six Latin American countries: an interrupted time series study Decline in COVID-19 hospitalization growth rates associated with statewide mask mandates -10 States Advice on the use of masks in the context of COVID-19: interim guidance. WHO, Geneva. 29 Advice on the use of masks in the context of COVID-19: interim guidance. World Health Organization; 2020 Advice on the use of masks in the context of COVID-19: interim guidance 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 Differential impact of non-pharmaceutical public health interventions on COVID-19 epidemics in the United States COVID-19 transmission in Hong Kong despite universal masking Use of stay-at-home orders and mask mandates to control COVID-19 transmission -Blackfeet tribal reservation Effectiveness of face coverings in mitigating the COVID-19 pandemic in the United States Mask adherence and rate of COVID-19 across the United States An evidence review of face masks against COVID-19 GRADE: an emerging consensus on rating quality of evidence and strength of recommendations Do face masks help? Is not the question Ecologic studies revisited Methodological considerations for epidemiological studies of air pollution and the SARS and COVID-19 coronavirus outbreaks. Environ Estimating effects of intervention measures on COVID-19 outbreak in Wuhan taking account of improving diagnostic capabilities using a modelling approach Efficacy of cloth face mask in prevention of novel coronavirus infection transmission: a systematic review and meta-analysis A cluster randomised trial of cloth masks compared with medical masks in healthcare workers Face mask use and control of respiratory virus transmission in households Phylogenetic analysis of SARS-CoV-2 in Boston highlights the impact of superspreading events Increased transmissibility and global spread of SARS-CoV-2 variants of concern as at Local, state and federal face mask mandates during the COVID-19 pandemic We would like to thank Landon Myer for critical input at the initial stage of this review and Tomas Allen for his support in designing the search strategy for this review. Supplementary material associated with this article can be found in the online version at doi:10.1016/j.eclinm.2021.101024.