Universal masking during COVID-19 outbreaks in aged care settings: A systematic review and meta-analysis

meta-analysis


Introduction
Aged care facilities (ACF) are a high-risk setting for SARS-COV-2 transmission (Hashan et al., 2021).There have been over 771 million cases and nearly seven million deaths confirmed worldwide as of 4 October 2023 (World Health Organization, 2023), with approximately 40% of COVID-19 deaths believed to be in residential ACF (Beiting et al., 2021;Bernabeu-Wittel et al., 2021).Residents in ACF are vulnerable and often require complex care, due to older age, co-morbidity, dementia or behavioural disorders, or functional impairments (Hashan et al., 2021).Older patients are at greater risk of severe COVID-19 disease and hospitalisation, and higher attack and mortality rates (Hashan et al., 2021).Since COVID-19 is highly contagious, frequent close contact between ACF residents, staff and visitors increases transmission risk and can prolong outbreaks (Kunasekaran et al., 2022).Reports of COVID-19 outbreaks in ACF have been common worldwide (Hashan et al., 2021;Kunasekaran et al., 2022).A systematic review and meta-analysis of COVID-19 cases in 757 facilities in 11 countries found a pooled attack rate (AR) for SARS-CoV-2 in residents of 42.0% (95% CI: 38.0-47.0%)(Kunasekaran et al., 2022).
Face masks are a simple, effective and widely accepted infection prevention and control (IPC) measure for reducing COVID-19 transmission risk (Ueki et al., 2020), with multiple countries mandating staff and visitor mask-wearing during periods of rising infections (Mitze et al., 2020).Masks reduce virus uptake in uninfected individuals, and act as source control to prevent outward transmission from infected people (Ueki et al., 2020).Few studies have quantitatively assessed the impact of universal masking on SARS-CoV-2 attack rates in ACF.A systematic review of outbreaks in ACF in the pre-vaccine period (1 January to 30 June 2020) found that use of personal protective equipment (PPE) by staff, including masks, gowns, gloves or eye protection, reduced the probability of COVID-19 infection by 45% (Natolanda et al., 2021).Another study assessed IPC measures in 12 ACF and found that delayed masking and lack of mask availability were associated with larger COVID-19 outbreaks (Reyne et al., 2021).Vaccines are another effective measure for reducing COVID-19 transmission and improving disease outcomes (Zheng et al., 2022).Many countries began their vaccine rollouts, in late 2020 or early 2021, by prioritising older population groups, including residents in ACF (Bailly et al., 2022;Teran et al., 2021;Williams et al., 2021).A meta-analysis assessing vaccine effectiveness (VE) across 51 studies found that even partial vaccination provided protection against SARS-CoV-2 infection and severe COVID-19 disease outcomes, though VE was lower among older adults (>60 years) compared to the general population (Zheng et al., 2022).Little data exists on the combined effects of masks and vaccines in ACF.
Understanding the effectiveness of universal masking and vaccination on SARS-CoV-2 attack rates during COVID-19 outbreaks in ACF could help inform IPC policies for ACF staff to reduce the COVID-19 disease burden in residents in ACF.
Aims: To investigate the effect of universal masking and vaccination on SARS-CoV-2 attack rates in aged care settings, with the hypothesis that universal masking of staff, and full vaccination of residents, would serve to reduce attack rates during outbreaks.

Search strategy
Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed (Page et al., 2021).The search for articles was conducted on 4-5 March 2022, in databases of Medline, Embase, PubMed, Scopus, Web of Science and Cumulative Index to Nursing and Allied Health Literature (CINAHL), using a date range of 1 December 2019-28 February 2022.A combination of Medical Subject Headings (MeSH) terms and keywords were used, including: ('Aged care facilities', 'Nursing homes', 'Long-term care facilities', 'Residential care facilities', 'Skilled nursing facilities', 'Assisted living facilities', 'Homes for the aged') AND ('COVID-19', 'SARS-COV-2', 'severe acute respiratory syndrome coronavirus 2', '2019-nCoV', 'coronavirus disease', 'new coronavirus, 'novel coronavirus', 'Wuhan coronavirus').Snowball searching of the references of relevant papers was also conducted to capture any additional suitable articles.This study was not registered with PROSPERO.

Selection criteria
Studies about COVID-19 outbreaks in ACF were included.These included prospective and retrospective cohort studies, case reports and outbreak investigations.Outbreaks were defined as ≥1 COVID-19 case in a resident or staff member of an ACF detected following identification of an index case.Articles were limited to English-language, peerreviewed, and with publicly available primary data.Commentaries, editorials, letters, pre-prints, conference proceedings, and non-English or not peer-reviewed articles were excluded.

Data extraction
After removing duplicate articles, titles and abstracts were screened to exclude articles that did not meet selection criteria.Full-text screening was conducted for remaining articles.Three reviewers (RC, SK & MK) were involved in the review process, with overlap in reviewing between the first two authors, and discussion with the third author where articles were ambiguous, to reach consensus regarding inclusion or exclusion of the article.Article data was extracted and recorded in an Excel file, including publication details (first author's name, year of publication), facility details (location, resident/staff capacity), outbreak details (index case details, outbreak onset and duration, causative variant, and IPC measures).If SARS-CoV-2 variant was not reported, the variant was assumed to be the predominant variant circulating in the country at the time of outbreak onset, using genomic surveillance data from Global Initiative on Sharing All Infuenza Data (GISAID) (Nextstrain, 2023).Data were extracted on implementation of universal masking, and on mask or PPE non-compliance and shortages, where available.Universal masking was defined as the wearing of face masks by ACF staff at all times in all areas of the facility during an outbreak (World Health Organization, 2020), and further categorized as implemented before outbreak onset (due to existing work policies or government mandates), after outbreak onset (delayed implementation following identification of index case), or not implemented.Studies where masks were worn only in limited facility areas were categorized as not having universal masking.Data on SARS-CoV-2 testing strategy were also extracted.Facility-wide testing (FWT) was defined as the testing of all consenting residents and/or staff in the facility, by rapid antigen test or polymerase chain reaction (PCR), at least once, following identification of the index case (Centers for Medicare & Medicaid Services, 2020).For facilities which held COVID-19 vaccination clinics, information was extracted regarding vaccine type, number of vaccine doses received by residents and staff prior to outbreak onset, and the date(s) of the vaccination clinic(s).Vaccination dose was defined as the total number of vaccine doses received by each resident or staff member, categorized as 0, 1 or 2 doses of an eligible COVID-19 vaccine.Doses provided < 14 days before outbreak onset were not considered full vaccination (i.e.<14 days post-dose 1 was considered unvaccinated, and <14 days post-dose 2 was considered as one dose).If such timing was not provided, it was assumed that the doses reported were given 14 days or more prior to outbreak onset.
Articles meeting the outbreak definition, with sufficient data for calculating resident and/or staff SARS-CoV-2 attack rates, were included for quantitative analysis.For articles which reported multiple outbreaks, and where sufficient data was available to do so, each outbreak was included in the analysis individually.Data were extracted regarding the total number of residents and staff undergoing COVID-19 testing during the outbreak, the number of COVID-19 cases and SARS-CoV-2 attack rates.

Risk of bias assessment
Two authors, RC and MK, reviewed each article to conduct a risk of bias assessment, using relevant Joanna Briggs Institute critical appraisal checklists (Munn et al., 2020;Joanna Briggs Institute, 2020;Joanna Briggs Institute, 2020).Agreement in risk of bias rating between authors was calculated, and articles were discussed to reach consensus in case of differing ratings.

Analysis
Meta-analysis of single proportions was performed using R 4.1.3,using the 'metafor' and 'meta' packages (Viechtbauer, 2010;Balduzzi et al., 2019), using a similar approach taken by Kunasekaran et al (Kunasekaran et al., 2022).A random effects model was used throughout.Heterogeneity was assessed using I 2 and χ 2 statistics.The restricted maximum-likelihood estimator method was used to estimate between-study variance (τ 2 ).Statistical significance was defined as a two-sided P-value of < 0.05.
Subgroup analysis was performed to compare pooled SARS-CoV-2 attack rates in residents by levels of universal masking implementation.Subgroup analysis was also performed to compare SARS-CoV-2 attack rates in residents and staff by number of vaccine doses R. Chen et al. received.For outbreaks where universal masking was implemented before outbreak onset and facility-wide testing was conducted, pooled overall and asymptomatic attack rates in residents and staff were estimated.Forest plots were generated.A multivariate mixed-effects metaregression of universal masking (yes vs no), facility-wide testing, vaccination status (unvaccinated vs vaccinated), and SARS-CoV-2 variant, on attack rate was also conducted using the 'metafor' package (Viechtbauer, 2010).Residents from ACF with outbreak onset before vaccine rollouts in their respective countries were assumed to be unvaccinated.

Results
A total of 99 articles were included for descriptive analysis.S3).Resident SARS-CoV-2 attack rates were available for 139 outbreaks across 86 studies, and staff SARS-CoV-2 attack rates were available for 78 outbreaks across 49 studies (Supplementary Materials, Table S5).Of 68 outbreaks with index case  information, 40 (58.8%)index cases were residents and 28 (41.2%) were staff.Where symptom information was provided for index cases, most were symptomatic (33/35 [94.3%] residents and 18/21 [85.7%] staff).Five articles were deemed to have a high risk of bias and were excluded from quantitative analysis, while 47 articles were considered moderate and low risk each (Supplementary Materials, Table S1).Agreement between authors was 45.5%.Fifty articles were ultimately included for quantitative analysis (Fig. 1), with four reporting on the same two outbreaks (Dora et al., 2021(Dora et al., , 2020;;Roxby et al., 2020aRoxby et al., , 2020b)).

Discussion
Our study demonstrates the potential for high attack rates in ACF, which underscores the importance of preventing SARS-CoV-2 transmission using pharmaceutical and non-pharmaceutical interventions.Universal masking and vaccination of ACF staff was associated with a statistically significant decrease in SARS-CoV-2 attack rate in residents and staff respectively.Our findings are consistent with existing evidence regarding the effectiveness of mask-wearing in reducing COVID-19 transmission in general populations.A comparative case study found that introducing mask mandates in Germany nearly halved the daily growth rate of reported COVID-19 infections (Mitze et al., 2020).Considering time delays in receiving test results, and potentially variable antigen test sensitivity (Dinnes et al., 2022;Del Vecchio et al., 2022), universal masking may help contain SARS-CoV-2 spread early during an outbreak or in case of false negative COVID-19 test results.Implementation of universal masking should also be considered in the context of ensuring high quality respiratory protection for staff (such as with P2 or N95 respirators) (Ueki et al., 2020), adequate supplies, fit-testing, staff training and compliance.Once an index case in an ACF has been identified, the capacity to urgently implement IPC measures, such as universal masking, in a responsive and tailored approach is crucial.Our study suggests timing of universal masking is not significant, and that even implementation after an outbreak is detected is effective.
The rationale for universal masking is that infection with SARS-CoV-2 may be asymptomatic.Notably, facilities that conducted universal testing showed that asymptomatic or presymptomatic infection may be present, and detection of asymptomatic infection (enhanced case ascertainment) may explain the lack of difference in attack rate between facilities with and without FWT.Asymptomatic staff members who become infected in the community could work across multiple facilities (Ueki et al., 2020) and potentially inadvertently introduce SARS-CoV-2 into ACF (Lyn Gilbert AO AL, 2021).
We also found a decreasing trend in SARS-CoV-2 attack rate with increasing number of vaccine doses in residents, but these differences were not statistically significant.This may be influenced by small subgroup sample sizes, virus mutations, and differences in elapsed time from vaccination to outbreak onset.Due to the limited number of studies, we were also not able to account for masking in the vaccination subgroups.Vaccines have proven efficacy in randomized controlled clinical trials (Zheng et al., 2022) but were made against the ancestral SARS-CoV-2 strain which has been superseded by a range of variants of concern since late 2020, each with varying degrees of vaccine escape (Chen et al., 2022).Studies have demonstrated that neutralising antibodies peak within 4-30 days following receipt of two Pfizer-BioNTech (BNT162b2) vaccine doses (Levin et al., 2021).Thus, outbreak onset may have occurred before or after peak vaccine effectiveness in residents was reached.The high intensity of transmission in aged care settings, waning of vaccine immunity (Levin et al., 2021) and vaccine escape caused by new variants of concern may all contribute to the lack of apparent effectiveness.However, vaccines do prevent against severe disease, and this protection is preserved longer than protection against infection (Ferdinands et al., 2022).Therefore vaccination remains an important protective measure in aged care.Staff vaccination was significantly protective, and is also key to outbreak mitigation.Our study showed that the D614G and Alpha variants were associated with higher SARS-CoV-2 attack rate compared to the ancestral strain, though the latter was not statistically significant, possibly due to small sample size.However, the ongoing potential for novel SARS-CoV-2 mutations to cause increased virus transmissibility and immune evasion (for example, the Omicron variant is about 10 and 2.8 times more infectious compared to the original virus and Delta respectively (Chen et al., 2022)), coupled with waning vaccine immunity, underscores the need for updated vaccine boosters, non-pharmaceutical interventions and therapeutic measures.
This study had several limitations.Firstly, many studies were excluded from quantitative analysis because authors did not mention whether masks were used during outbreaks.There was also little information regarding mask-wearing by residents.Were this data available, pooled SARS-CoV-2 attack rates may be different than calculated.However, the prevalence of dementia is high in most aged care facilities, and masking of residents is uncommon.Secondly, masking was generally implemented alongside other IPC measures (e.g.visitor restrictions or bans, cohorting of COVID-19 cases, PPE use), the extent of which differed between facilities and was not reported reliably in all studies.Some ACF housed residents in single rooms while others provided shared rooms, potentially affecting transmission.Follow-up periods and testing strategies also differed between studies: not all studies performed FWT, either due to existing testing guidelines or equipment shortages.Timing of repeat testing in facilities that conducted serial FWT also varied, with differing follow-up periods, thus we could not fully ascertain the effect of FWT strategy, when combined with masking, on SARS-CoV-2 attack rate.Our study also did not account for facility size and infrastructure, though studies have shown that larger facility size and older building designs are associated with greater frequency of COVID-19 cases (Kunasekaran et al., 2022).There was also substantial heterogeneity between studies, possibly due to varied reporting standards.

R. Chen et al.
Variations included the methods and interventions used in studies, case-ascertainment in facilities, extent of testing and follow-up, resident health characteristics and sex differences (male to female ratio).In some studies, total resident and staff numbers were unclear, and so were derived from provided percentages of infected residents or staff members.Only English-language studies were included.Due to the assumptions made regarding SARS-CoV-2 variants and vaccination status, and small sample size, results of the meta-regression model should be interpreted with caution.
Areas of future research could involve delineating the impact of masking from other IPC measures on SARS-CoV-2 attack rates and assessing the effects of masking in vaccinated ACF residents as more data becomes available.As most outbreaks in this study occurred before Delta and Omicron predominance, future research could also investigate the effect of universal masking and vaccination during outbreaks caused by these and newer variants.
Despite the above limitations, our findings indicate the protective effect of universal masking of staff during SARS-CoV-2 outbreaks.Though many governments have now relaxed their pandemic restrictions, including mandatory masking in ACF in certain countries, COVID-19 epidemics continue to occur in aged care and cause substantial morbidity, mortality and economic impacts (Hashan et al., 2021).To limit disease burden, IPC measures form an important line of defence (Lyn Gilbert AO AL, 2021).Particular attention should also be given to monitoring community transmission as a predictor of SARS-CoV-2 outbreaks in ACF (Lyn Gilbert AO AL, 2021).

Conclusions
This study supports using universal, precautionary masking in ACF as an effective IPC strategy for lowering the risk of SARS-CoV-2 infection, but even masking implemented after an outbreak starts is effective.Vaccination is also important for staff and residents, as it protects against severe disease and may mitigate outbreaks.Policies and guidelines should outline protocols for rapid implementation of universal masking and facility-wide testing in case of outbreaks in ACF.
Fig. depicts a flowchart showing the article selection and screening process based on PRISMA guidelines.Data were available for 798 ACF across countries, for 54,856 residents and 33,380 staff members.The earliest outbreak began in February 2020 and the latest in July 2021 (Supplementary Materials, Table
R.Chen et al.

Fig. 2 .
Fig. 2. Pooled SARS-CoV-2 attack rates in residents by universal masking comparison groups.Blue squares represent the median attack rate and black horizontal lines represent the 95% confidence interval for each study.The centre of the red diamond indicates the median pooled attack rate, while the ends of the red diamond indicate the 95% confidence interval.
R.Chen et al.

Fig. 3 (
Fig. 3(.A-B).Pooled SARS-CoV-2 attack rates in outbreaks with universal masking implemented before outbreak onset and facility-wide testing in residents (A) and staff (B).Blue squares represent the median attack rate and black horizontal lines represent the 95% confidence interval for each study.The centre of the black diamond indicates the median pooled attack rate, while the ends of the black diamond indicate the 95% confidence interval.
R.Chen et al.

Fig. 4 .
Fig. 4. Pooled SARS-CoV-2 attack rates in residents comparing by 0, 1 or 2 COVID-19 vaccine doses received.Blue squares represent the median attack rate and black horizontal lines represent the 95% confidence interval for each study.The centre of the red diamond indicates the median pooled attack rate, while the ends of the red diamond indicate the 95% confidence interval.