Interventions to increase personal protective behaviours to limit the spread of respiratory viruses: A rapid evidence review and meta-analysis.

PURPOSE
Increasing personal protective behaviours is critical for stopping the spread of respiratory viruses, including SARS-CoV-2: We need evidence to inform how to achieve this. We aimed to synthesize evidence on interventions to increase six personal protective behaviours (e.g., hand hygiene, face mask use, maintaining physical distancing) to limit the spread of respiratory viruses.


METHODS
We used best practice for rapid evidence reviews. We searched Ovid MEDLINE and Scopus. Studies conducted in adults or children with active or passive comparators were included. We extracted data on study design, intervention content, mode of delivery, population, setting, mechanism(s) of action, acceptability, practicability, effectiveness, affordability, spill-over effects, and equity impact. Study quality was assessed with Cochrane's risk-of-bias tool. A narrative synthesis and random-effects meta-analyses were conducted.


RESULTS
We identified 39 studies conducted across 15 countries. Interventions targeted hand hygiene (n = 30) and/or face mask use (n = 12) and used two- or three-arm study designs with passive comparators. Interventions were typically delivered face-to-face and included a median of three behaviour change techniques. The quality of included studies was low. Interventions to increase hand hygiene (k = 6) had a medium, positive effect (d = .62, 95% CI = 0.43-0.80, p < .001, I2  = 81.2%). Interventions targeting face mask use (k = 4) had mixed results, with an imprecise pooled estimate (OR = 4.14, 95% CI = 1.24-13.79, p < .001, I2  = 89.67%). Between-study heterogeneity was high.


CONCLUSIONS
We found low-quality evidence for positive effects of interventions targeting hand hygiene, with unclear results for interventions targeting face mask use. There was a lack of evidence for most behaviours of interest within this review.


Statement of Contribution
What is already known on this subject?
Widespread adoption of personal protective behaviours (PPBs) is needed to block respiratory viral transmission.
Effective interventions that increase PPBs primarily target health care professionals.
The generalizability of such interventions to the general population remains unclear.
What does this study add? This rapid review extends findings from health care to community settings. We found evidence for a medium, positive effect of hand hygiene interventions.
Interventions targeting face mask use had unclear results. Few relevant studies were identified, and study quality was low.

Background
Respiratory viruses such as influenza, respiratory syncytial virus, parainfluenza, rhinovirus, coronavirus (including SARS-CoV-2), and adenovirus enter the body through the eyes, nose, and mouth (the 'T-Zone') (Killingley & Nguyen-Van-Tam, 2013;. Changing human behaviour is critical for stopping the spread of respiratory viruses in general and the SARS-CoV-2 virus in particular, and for supporting the easing of financially and psychologically costly physical distancing measures during viral epidemics (Ferguson et al., 2006;. Personal protective behaviours, including hand washing, disinfecting fomites such as clothes or furniture, and face mask wearing, are advocated for limiting the spread of SARS-CoV-2 (Lunn et al., 2020;World Health Organization, 2019). Simply advising people to adopt these behaviours has been found to be insufficient, just as has explaining what to do and why these behaviours are necessary (Bish & Michie, 2010). Directly relevant evidence on interventions to promote adherence to personal protective behaviours in community-dwelling children and adults is sparse but there is an urgent need to identify and synthesize what evidence does exist. Policymakers need evidence to inform the development of public health guidance and decide which interventions to prioritize. We adopted best practice for rapid evidence reviews to evaluate the acceptability, practicability, effectiveness, affordability, spill-over effects (i.e., unintended consequences), and equity impact (the 'APEASE' criteria (Michie, Atkins, & West, 2014)) of interventions to increase personal protective behaviours to limit the spread of respiratory viruses. During pandemics of respiratory viruses, multipronged approaches involving both pharmacological (e.g., vaccination) and behavioural measures (e.g., hand washing, physical distancing) are required to bring the reproductive number below 1 (Ferguson et al., 2006;. Vaccination of populations will take months, even years, to roll out, especially in low-and middle-income countries. Hence, physical distancing and other behavioural measures will be required, possibly permanently. Population-wide restrictions are costly from financial, social, and psychological perspectives: The world economy has been projected to shrink by approximately 4.9% in 2020 (International Monetary Fund, 2020), with an additional 88 million people globally being pushed into extreme poverty (i.e., living on less than $1.90/day) (Blake & Wadhwa, 2020), and prolonged periods of social isolation are associated with increases in domestic violence (SafeLives, 2020) and negative mental health effects, such as post-traumatic stress disorder, confusion, and anger (Brooks et al., 2020). Less costly, yet highly effective (Warren-Gash, Fragaszy, & Hayward, 2012) personal protective behaviours are thus important for supporting the easing of lockdown measures to ensure long-term suppression of viral transmission and preparedness for new viral waves and future pandemics . To successfully block the spread of respiratory viruses including (but not limited to) SARS-CoV-2which are transmitted via droplets, aerosols, and direct physical contact (Killingley & Nguyen-Van-Tam, 2013; several personal protective behaviours must be adopted across the population (see Figure 1). Although systematic reviews of interventions to change hand hygiene in health care professionals are available (Edwards et al., 2012;Huis et al., 2012;Luangasanatip et al., 2015;Mbakaya, Lee, & Lee, 2017;Olena Doronina, Jones, Martello, Biron, & Lavoie-Tremblay, 2017), generalizability to community settings is limited. There also appears to be little evidence about interventions to change behaviours such as not touching the T-Zone (eyes, nose, and mouth), which would have a significant effect if adopted (Kwok, Gralton, & McLaws, 2015) and carry little or no costs to people or society. If adopted at scale across the population including disadvantaged communities, such interventions have the potential to reduce health inequalities. Here, we aimed to conduct a rapid evidence review to evaluate the acceptability, practicability, effectiveness, affordability, spill-over effects, and Figure 1. Map of personal protective behaviours relevant for blocking transmission of respiratory viruses, including SARS-CoV-2: hand washing and use of hand sanitizers; avoiding touching the 'T-Zone'; catching droplets in tissues and discarding these; face mask use; disinfecting surfaces; and maintaining physical distancing. Reproduced with permission from the authors . equity of interventions to increase personal protective behaviours that limit the spread of respiratory viruses.

Study design
The study protocol was pre-registered on the Open Science Framework (https://osf.io/ 7cphy/). During ongoing pandemics, the World Health Organization recommends the use of rapid evidence reviews for swift knowledge generation (World Health Organization, 2017). We adopted acknowledged best practice for rapid evidence reviews, which involved completing the review in a timely fashion, limiting the search to main databases and the published literature, and having one reviewer extract data and another verify (Haby et al., 2016;Tricco et al., 2015).

Criteria for considering studies for this review Population
We included studies that recruited as participants community-dwelling children or adults (as opposed to qualified or trainee health care professionals in hospital or care home settings) across any type of study setting (e.g., schools, primary care).

Intervention
We included reports of evaluations of any type of intervention (e.g., mass media, face-toface, technology-mediated) designed to change at least one of six personal protective behaviours to block transmission of respiratory viruses, such as influenza, respiratory syncytial virus, parainfluenza, rhinovirus, coronavirus, or adenovirus, which have a shared route of transmission (i.e., droplets, aerosols, direct physical contact) (Killingley & Nguyen-Van-Tam, 2013). Although the relative importance of different personal protective behaviours depends on properties of the specific respiratory virus in addition to the clinical and/or environmental contextfor example, fomite transmission may be more pronounced for respiratory syncytial virus compared with coronaviruses (Boone & Gerba, 2007)at the time of planning this rapid review, little was known about the properties of SARS-CoV-2, and we therefore opted for an inclusive scope.

Comparison
We included studies with an active or passive (e.g., wait-list control, baseline) comparator.

Outcomes
We included studies that reported as outcome either the acceptability, practicability, effectiveness, affordability, spill-over effects, and/or equity of interventions, provided that these were measured at the individual level via self-report or direct observation.

Study designs
We included primary research studies that used experimental (e.g., randomized controlled trial) or quasi-experimental (e.g., pre-and post-test) study designs, with individuals or clusters as the unit of randomization, providing that they were conducted under free-living (as opposed to laboratory) conditions. In line with rapid review guidelines, we only included studies that were published in peer-reviewed journals and written in English (Tricco et al., 2015;World Health Organization, 2017).

Search methods for identification of studies Electronic searches
We searched Ovid MEDLINE and Scopus. Search terms for each behaviour (e.g., 'hand hygiene', 'hand washing', 'face mask') were piloted and refined to achieve balance between sensitivity and specificity (see Appendix S1).
Searching for other sources Expertise within the review team and consultation with topic experts was used to identify additional articles of interest. We had specified the use of reference chaining in the review protocol; however, given the rapid focus of the review and large number of identified studies, reference chaining was not performed.

Data collection and analysis
Selection of studies Two reviewers (OP and DS) independently screened (1) titles and abstracts and (2) full texts against the inclusion criteria. Discrepancies were discussed and resolved through consulting with a third reviewer (EC) if necessary.

Data extraction and management
A data extraction form was developed on the basis of the Behaviour Change Intervention Ontology (www.humanbehaviourchange.org) and Cochrane's PICO ontology (https:// linkeddata.cochrane.org/pico-ontology). Ontologies are classification systems which enable researchers to specify entities (e.g., behaviours, interventions) and their interrelationships. The use of ontologies in systematic reviews can help ensure that a comprehensive set of entities are considered and defined in standardized ways, thus facilitating systematic knowledge synthesis (Norris, Finnerty, Hastings, Stokes, & Michie, 2019). We extracted data from relevant sections of published articles and available Appendix S1 on study design, intervention content (i.e., behaviour change techniques (BCTs), coded against the BCT Taxonomy v1 (Michie et al., 2013)), mode of delivery, population, setting, and mechanism(s) of action (Carey et al., 2019;Moore & Evans, 2017). As a validated taxonomy of mechanisms of action is, to our knowledge, not yet available, authors' own definitions of mechanisms of action were extracted if they explicitly discussed how the selected intervention components/BCTs were expected to influence the target behaviour. Due to limited resources, we did not contact study authors for more detail on intervention descriptions. As an intervention may be effective but have negative spill-over effects to other behaviours, or be impracticable and/or unacceptable to key stakeholders, we also deemed it important to extract data from relevant sections of published articles and available supplementary materials pertaining to the APEASE criteria (see Table 1) (Michie et al., 2014). Although criteria such as affordability or practicability are arguably closely tied to the context in which the intervention was/will be implemented, APEASE intends to capture higher-order criteria against which to rate interventions, broadly applicable irrespective of the specific context. Therefore, authors' own descriptions of, for example, acceptability or spill-over effects (as opposed to reviewers' ratings) were extracted. Spill-over effects were broadly defined as any unintended consequences (positive or negative) reported by the authors, including but not limited to other behaviours that were changed by the intervention that it was not designed to target (e.g., teachers' improved hand hygiene positively or negatively influencing children's hand hygiene). Data were extracted by one reviewer (OP or DS). In the review protocol, we had specified that extracted data would be verified by a second reviewer to assess accuracy and completeness. However, given the large number of identified studies, a second reviewer (EC) verified 10% of studies.

Quality appraisal
The methodological rigour of included evaluation reports was assessed by one reviewer (OP or DS) using Cochrane's risk-of-bias tool (The Cochrane Collaboration, 2011). A second reviewer (EC) verified 10% of studies.

Stakeholder involvement
We solicited input from key stakeholders, including patient and public representatives recruited via panels convened by Public Health England (n = 282) and the University of East Anglia (n = 3), and UK policymakers and academic researchers contacted via a mailing list on the research objectives, target behaviours, and outcomes assessed. Feedback from patient and public representatives (n = 20) was incorporated into the review protocol; we did not receive any suggestions for improvement or clarification from the policymakers and academic researchers. The rapid review results will be disseminated to stakeholders via an infographic.

Data synthesis
A narrative (descriptive) synthesis was conducted for each of the personal protective behaviours. We had specified in the review protocol that meta-analyses would be Table 1. APEASE criteria for evaluating intervention approaches or components (Michie et al., 2014) Criterion To what extent . . .
Acceptability . . .is the intervention judged to be acceptable by all key stakeholders Practicability . . .can the intervention be delivered as intended at the scale intended and in the context intended Effectiveness . . .will the intervention deliver the desired outcome in the target population Affordability . . .can the intervention be afforded within an acceptable budget Spill-over effects . . .is the intervention likely to have additional negative or positive consequences Equity . . .is the intervention likely to increase or decrease inequalities in society conducted if practicable and appropriate (i.e., >5 studies with homogeneous study designs and outcome variables). After inspection of study designs and outcome variables, however, we deemed it useful to conduct a meta-analysis with k = 4 studies. Randomeffects meta-analyses to estimate a pooled odds ratio (OR) or standardized mean difference (d) were conducted in RStudio v.1.2.5033 with the metafor package (Viechtbauer, 2010). Cohen's conventions for small (d = . 2), medium (d = .5), and large (d = .8) effects were used in the interpretation of the results (Cohen, 1988). In studies with more than two arms (e.g., three-arm RCTs), we compared the 'most active' (i.e., the arm with the greatest number of intervention components) and the 'most passive' arms (i.e., the arm with the lowest number of intervention components, typically labelled the 'control' arm by study authors). Where studies reported more than one hand hygiene outcome (e.g., hand washing and hand sanitizer use), only the first reported outcome was included in the metaanalysis, so as not to violate the assumption of independence (Cheung, 2019). Where studies did not report sufficient detail to calculate effect sizes, authors' own description/ interpretation of results were grouped into 'positive' effects (i.e., a significant difference between intervention and control groups, favouring the intervention group, was detected), 'no difference' (i.e., a significant difference between groups was not detected), 'negative' effects (i.e., a significant difference between groups, favouring the control group, was detected), or 'indeterminate' (i.e., differences between groups were not reported or could not be computed given the study design). To aid interpretation, for behaviours where a majority of positive or negative results were observed, overall results were categorized as either 'positive' or 'negative', respectively. If consistent results were not observed or could not be determined, overall results were categorized as 'mixed'.

Study selection
After removing duplicates, 5,595 records were identified, with 159 studies carried forward to the full text screening. Of the 39 studies included in the narrative evidence synthesis, 10 were included in meta-analyses (see Figure 2).

Practicability
One study (MacIntyre et al., 2009) considered the practicability of scaling up mask fit testing outside the study setting and decided against including routine fit testing as part of the intervention (see Table 4).

Effectiveness
Studies relied on self-report (25/39; 64%), direct observation (7/39; 18%), a combination of self-report and direct observation (5/39; 13%), or photographs/video (2/39; 5%) to examine intervention effectiveness (see Table 4). Outcome variables were heterogeneous across studies (e.g., the frequency or amount of hand sanitizer/soap use per day, the rate of compliance with hand hygiene, or the rate of compliance with face mask use).
A random-effects meta-analysis (k = 4) found a large, positive effect of interventions on the odds of compliance with face mask use, OR = 4.14, 95% CI = 1.24-13.79, p < .001 (see Figure 6). However, between-study heterogeneity was high (I 2 = 89.67%) and the confidence interval for the pooled effect was wide.

Continued
Rapid review of interventions to increase personal protective behaviours 27 of hand sanitizer and mask use Participants in the hand sanitizer group used a mean of 12.1 ounces/month and those in the hand sanitizer and face mask group used a mean of 11.6 ounces/month (p = .36) (/). Half of the households with a case of infection reported using masks within 48 hrs of symptom onset. Those who used masks reported a mean of two masks/day/episode (range: 0-9) (@)

Continued
Rapid review of interventions to increase personal protective behaviours 33 Note. + = positive effect; À = negative effect; / = no difference; @ = indeterminate; and NA = not applicable.

Rapid review of interventions to increase personal protective behaviours 37
Affordability Two studies considered the affordability of interventions, with one study discussing the cost of face masks (Bundgaard et al., 2020), which may act as a barrier for wider roll-out, and a second study (Zomer et al., 2016) reporting that although they wanted to provide hand hygiene products to all daycare centre groups, they could only afford to do so for a maximum of two groups per centre due to budget restrictions, thus indicating that the selected intervention was not affordable at scale (see Table 4).

Spill-over effects
Three studies reported on secondary behaviour change (i.e., positive or negative spill-over to other behaviours), with one study (Zomer et al., 2016) assessing teachers' supervision of children's hand washing (in addition to their own hand washing), a second (Mott et al., 2007) assessing the impact of the intervention on military leaders' hand sanitizer use (in addition to trainees'), and a third (Larson et al., 2009) examining multiple hand hygiene behaviours in the same group of participants. The first study reported no change, the second a significant increase in hand sanitizer use, and the third a negative impact on hand washing with soap (see Table 4).

Equity
Four studies reported on the equity of interventions, with one study (Yardley et al., 2011) reporting that the intervention was equally effective for participants from high and low socio-economic status groups and three studies (Azman et al., 2013;Azor-Martinez et al., 2016;Larson et al., 2009) reporting differential intervention effectiveness by educational attainment, parental income, or ethnicity, with better outcomes reported in those with high educational attainment, high parental income, and from a Black ethnic background (see Table 4).

Quality of included studies
One study received an overall rating of 'low risk of bias', with 16 studies rated as 'some concern', 18 as 'high risk of bias', and for four studies, an overall rating could not be applied (see Table 5).

Discussion
This rapid review of interventions to increase personal protective behaviours to limit the spread of respiratory viruses identified 39 studies conducted across 15 countries. The majority of interventions targeted hand hygiene and/or face mask use, with one intervention targeting the catching of droplets in tissues in addition to hand hygiene. None of the identified interventions focused on avoiding touching the T-Zone, disinfecting surfaces or maintaining physical distancing. Interventions were typically delivered in participants' own homes or in nurseries/schools, targeting children/adult household members or pre-or school children/teachers. Two-or three-arm study designs with passive comparators were typically used. The overall quality of included studies was low, with only one study rated as 'low risk of bias'. The majority of interventions had a faceto-face component and delivered a median of three BCTs; the most frequent were 'Adding objects to the environment', 'Instruction on how to perform the behaviour' and 'Information about health consequences'. Where investigated, interventions were considered acceptable by participants, with a minority reporting issues with mask wear discomfort or skin irritation from hand hygiene products. Few studies reported the practicability, affordability, spill-over effects or equity of interventions. In a narrative synthesis, interventions targeting hand hygiene behaviour were found to have positive effects and those targeting face mask use had a mixture of positive and negative effects. Random-effects meta-analyses of a small number of studies found positive effects of interventions targeting hand hygiene behaviour and face mask use. However, betweenstudy heterogeneity was high and the confidence interval for the pooled effect of interventions targeting face mask use was wide, partly due to the small number of studies included in the comparison.

Strengths and limitations
This review was conducted rapidly (July-December 2020) with input on the research questions and review scope from public health and behavioural science experts and lay members as part of a written stakeholder consultation. However, the pragmatic nature of this review, conducted during an ongoing pandemic, also means that it has several important limitations. First, given the expected large number of hand hygiene studies related to gastrointestinal infections, we limited the review to studies explicitly studying behaviour change in relation to respiratory viruses. However, data from interventions targeting personal protective behaviours to prevent gastrointestinal illness are likely to add to our understanding of the acceptability, effectiveness, and equity of hand hygiene interventions. Second, evidence indicates that the relative importance of different personal protective behaviours may depend on properties of the specific respiratory virus and context (e.g., fomite transmission may be more pronounced for respiratory syncytial virus compared with coronaviruses) (Boone & Gerba, 2007). However, at the time of planning this rapid evidence review, little was known about SARS-CoV-2. We therefore opted for a broad scope and included interventions targeting personal protective behaviours to limit the spread of any respiratory viral infection. It was also not possible to group the results based on the specific viral infections studied as the majority of studies targeted multiple (as opposed to single) respiratory viral infections and there was little variability in the viral infections targeted. In addition, as the majority of studies targeted a host of different viruses within their interventions, this further limits the conclusions that can be drawn: The perceived susceptibility to different viruses likely differs between, for example, age groups (Rosenstock, 1974), and tailored intervention strategies may therefore be needed for younger (vs. older) adults. However, the current review was unable to address such nuanced questions due to the limited design and reporting of extant studies and the need to synthesize evidence quickly during an ongoing pandemic. Third, our electronic search was restricted to two databases, which may have limited the results, and data extraction was performed by a single reviewer, with a proportion verified by a second reviewer. Fourth, although most of identified studies were two-or three-arm RCTs, they were typically designed to study rates of respiratory infection as their primary outcome, with behaviour change outcomes less clearly reported. This hindered Figure 5. Forest plot for the standardized mean difference (d) in the frequency of hand hygiene behaviour in intervention and control or pre-and post-study comparisons. The comparison in Chan (Chan et al., 2007) pertains to a pre-and post-study comparison; the remaining studies were two-or three-arm RCTs.
quantitative synthesis, with only a small number of included studies contributing to metaanalyses. Future studies specifically designed to examine the effectiveness of interventions on behavioural outcomes are needed. Fifth, in line with guidelines (Wood et al., 2015), we only coded BCTs when there was clear evidence of their presence; interventions may have included additional BCTs not documented in this review. Sixth, as this was a rapid review with limited resources, we limited our analyses of intervention content to information presented in the published papers and/or available supplementary materials, and no attempts were made to contact study authors for access to detailed intervention descriptions. Seventh, as most interventions targeting multiple behaviours (e.g., hand hygiene and face mask use) did not clearly distinguish BCTs that targeted one (but not the other) behaviour, and none of the outcome evaluations considered potential behavioural dependencies (or statistical interactions), it was not possible to consider the extent of BCT overlap and/or behavioural interactions in the present review. Finally, due to the small number of studies available for meta-analysis, we were unable to group studies by, for example, population type, study setting, type of virus, type of outcome assessment, etc., which would have further improved our understanding of intervention effectiveness.

Implications for policy and practice
Although we caution against drawing firm conclusions due to the low quality of the evidence, positive effects of interventions targeting hand hygiene behaviour and face mask use were observed, with the majority of interventions providing free hand hygiene products and/or face masks to participants in addition to instructions on how to perform the behaviour and information about health consequences. As far as is practicably feasible, authorities should aim to provide free products to staff, clients, and visitors during respiratory viral epidemics. The limited range of BCTs detected in published intervention descriptions may suggest a missed opportunity for harnessing techniques indicated by relevant behaviour change theory and evidence. We encourage policymakers and health care practitioners to work collaboratively with behavioural scientists to incorporate   techniques that theory or evidence predicts are effective for enabling personal protective behaviours (Warren-Gash et al., 2012), such as techniques targeting motivational or selfregulatory processes. For example, while hand hygiene is a well-established, often private and widely accepted protective behaviour that most people have long experience with, face mask wearing is a relatively new (at least in some countries), public behaviour, where there is more debate and uncertainties about the impact of the behaviour among the public and scientists (Cowling, Zhou, Ip, Leung, & Aiello, 2010). It is therefore important to involve behavioural scientists in the development of any new interventions targeting personal protective behaviours in the context of respiratory viral infections to help map out potential influences (e.g., social, self-regulatory) that may help or hinder the target behaviour, acknowledging that different interventions are likely needed for different behaviours.
Avenues for future research Findings highlight the need for evaluations of interventions to support people to avoid touching the T-Zone, disinfect surfaces, maintain physical distancing, and ensure efficient ventilation. Ventilation is increasingly seen as an important personal protective behaviour but was missed from the present review as it was planned during an earlier epidemic phase when the emphasis was on viral transmission via droplets rather than aerosols (Anderson, Turnham, Griffin, & Clarke, 2020;Morawska & Milton, 2020). In addition, we need studies designed to detect effects on behavioural outcomes and data on the affordability and equity of interventions to increase personal protective behaviours, particularly in low-and middle-income countries. Although the provision of hand hygiene products and face masks may offset costs related to primary and secondary care or work absenteeism for those with severe respiratory viral illness, the provision of free products at scale may be prohibitively costly. Future research involving health and social care economists should evaluate the cost-effectiveness of different types of interventions to enable personal protective behaviours, including those targeting motivational and selfregulatory processes. We also need further evidence from studies evaluating interventions to improve adherence to face mask use, with unclear results observed at present. Finally, due to the small number of studies with data suitable for meta-analysis, we did not conduct moderator analyses to examine whether, for example, particular BCTs, broader content categories, or the unit of randomization (e.g., individual vs. cluster) were related to intervention effectiveness; this would be important to examine in future meta-analyses with larger sample sizes. We did not consider here the use of, for example, the Theoretical Domains Framework (Cane, O'Connor, & Michie, 2012) when coding the mechanisms of action of interventions; this may be useful to consider in future empirical studies and evidence reviews. Finally, although evidence generation during ongoing pandemics is challenging (with a need to balance a pragmatic approach and limited resources with scientific rigour), drawing primarily on studies conducted outside the pandemic setting to inform what behavioural interventions to implement is suboptimal. We therefore recommend that experimental studies of behavioural interventions are prioritized during future respiratory viral pandemics.

Conclusions
This rapid review identified 39 studies across 15 countries with interventions targeting hand hygiene and/or face mask use. Positive effects of interventions targeting hand hygiene were observed, with unclear results for interventions targeting face mask use. There was a lack of evidence for interventions targeting most behaviours of interest within this review.