What was studied in the review?
In order to reduce the spread of the virus that causes COVID-19, many governments and societies put mitigation measures in place in schools. However, we do not know whether these measures work with regards to reducing the spread of the virus, or how these measures affect other aspects of life, such as education, the economy or society as a whole.
What are measures implemented in the school setting?
Measures in the school setting can be grouped into the following four broad categories.
1. Measures reducing the opportunity for contacts: by reducing the number of students in a class or a school, opening certain school types only (for example primary schools) or by creating a schedule by which students attend school on different days or in different weeks, the face-to-face contact between students can be reduced.
2. Measures making contacts safer: by putting measures in place such as face masks, improving ventilation by opening windows or using air purifiers, cleaning, handwashing, or modifying activities like sports or music, contacts can be made safer.
3. Surveillance and response measures: screening for symptoms or testing sick or potentially sick students, or teachers, or both, and putting them into isolation (for sick people) or quarantine (for potentially sick people).
4. Multicomponent measures: measures from categories 1, 2 and 3 are combined.
What is the aim of the review?
We aimed to find out which measures implemented in the school setting allow schools to safely reopen, stay open, or both, during the COVID-19 pandemic.
What did we do?
We searched for studies that looked at the impact of these types of measures in the school setting on the spread of the virus that causes COVID-19, the impact on the healthcare system (i.e. how many hospital beds are needed), as well as important social aspects (i.e. how often students attended school). The studies could focus on students, teachers and other school staff, as well as on families and the whole community. They could use real-life data (observational studies) or data from computer-generated simulations (modelling studies).
What are the main results of the review?
We found 38 relevant studies. Most of these were modelling studies (33 studies). Five studies used real-world data. Twenty studies were conducted in North or South America, 16 in Europe and two in China.
Below we summarise the main findings by category.
1. Measures reducing the opportunity for contacts
We found 23 modelling studies assessing measures to reduce the opportunity for contacts. All studies showed reductions in the spread of the virus that causes COVID-19 and the use of the healthcare system. Some studies also showed a reduction in the number of days spent in school due to the intervention.
2. Measures making contacts safer
We found 11 modelling studies and two real-world studies looking at measures, such as mask wearing in schools, cleaning, handwashing, and ventilation. Five of these studies combined multiple measures, which means we cannot see which specific measures worked and which did not. Most studies showed reductions in the spread of the virus that causes COVID-19; some studies, however, showed mixed or no effects.
3. Surveillance and response measures
We found 13 modelling studies and one real-world study assessing surveillance and response measures. Twelve studies focused on mass testing and isolation measures, while two looked specifically at symptom-based screening and isolation. Most studies showed results in favour of the intervention, however some showed mixed or no effects.
4. Multicomponent measures
We found three studies that looked at multicomponent interventions, where it was not possible to determine the effect of each individual intervention. These included one modelling study and two real-world studies. These studies assessed physical distancing, modification of activities, cancellation of sports or music classes, testing, exemption of high-risk students, handwashing, and face masks. Most studies showed reduced transmission of the virus that causes COVID-19, however some showed mixed or no effects.
How confident are we in the findings of this review?
Our confidence in these results is limited. Most studies used models, that is, they estimated the effects of the interventions rather than observing outcomes. As the models are built on assumptions about how the virus spreads and how people behave, we lack real‐world evidence. Many studies were published as 'preprints' without undergoing rigorous checks of published studies, which further limits our confidence. Also, the studies were very different from each other (for example, with regards to the levels of transmission in the community).
What are the key messages?
Reopening schools or keeping schools open while having a broad range of measures in place can reduce transmission of the virus that causes COVID-19. Such measures can also reduce the number of people who will need to go to hospital due to developing COVID-19. We still know very little about other consequences of these measures, such as those linked to education, resources, and physical or mental health, as this knowledge is mostly based on studies modelling the real world. More studies set in the real world using real-world data are needed.
How up to date is this evidence?
The evidence is up-to-date to December 2020.
Our review suggests that a broad range of measures implemented in the school setting can have positive impacts on the transmission of SARS-CoV-2, and on healthcare utilisation outcomes related to COVID-19. The certainty of the evidence for most intervention-outcome combinations is very low, and the true effects of these measures are likely to be substantially different from those reported here. Measures implemented in the school setting may limit the number or proportion of cases and deaths, and may delay the progression of the pandemic. However, they may also lead to negative unintended consequences, such as fewer days spent in school (beyond those intended by the intervention). Further, most studies assessed the effects of a combination of interventions, which could not be disentangled to estimate their specific effects. Studies assessing measures to reduce contacts and to make contacts safer consistently predicted positive effects on transmission and healthcare utilisation, but may reduce the number of days students spent at school. Studies assessing surveillance and response measures predicted reductions in hospitalisations and school days missed due to infection or quarantine, however, there was mixed evidence on resources needed for surveillance. Evidence on multicomponent measures was mixed, mostly due to comparators. The magnitude of effects depends on multiple factors. New studies published since the original search date might heavily influence the overall conclusions and interpretation of findings for this review.
In response to the spread of severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) and the impact of coronavirus disease 2019 (COVID‐19), governments have implemented a variety of measures to control the spread of the virus and the associated disease. Among these, have been measures to control the pandemic in primary and secondary school settings.
To assess the effectiveness of measures implemented in the school setting to safely reopen schools, or keep schools open, or both, during the COVID-19 pandemic, with particular focus on the different types of measures implemented in school settings and the outcomes used to measure their impacts on transmission-related outcomes, healthcare utilisation outcomes, other health outcomes as well as societal, economic, and ecological outcomes.
We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, and the Educational Resources Information Center, as well as COVID-19-specific databases, including the Cochrane COVID-19 Study Register and the WHO COVID-19 Global literature on coronavirus disease (indexing preprints) on 9 December 2020. We conducted backward-citation searches with existing reviews.
We considered experimental (i.e. randomised controlled trials; RCTs), quasi-experimental, observational and modelling studies assessing the effects of measures implemented in the school setting to safely reopen schools, or keep schools open, or both, during the COVID-19 pandemic. Outcome categories were (i) transmission-related outcomes (e.g. number or proportion of cases); (ii) healthcare utilisation outcomes (e.g. number or proportion of hospitalisations); (iii) other health outcomes (e.g. physical, social and mental health); and (iv) societal, economic and ecological outcomes (e.g. costs, human resources and education). We considered studies that included any population at risk of becoming infected with SARS-CoV-2 and/or developing COVID-19 disease including students, teachers, other school staff, or members of the wider community.
Two review authors independently screened titles, abstracts and full texts. One review author extracted data and critically appraised each study. One additional review author validated the extracted data. To critically appraise included studies, we used the ROBINS-I tool for quasi-experimental and observational studies, the QUADAS-2 tool for observational screening studies, and a bespoke tool for modelling studies. We synthesised findings narratively. Three review authors made an initial assessment of the certainty of evidence with GRADE, and several review authors discussed and agreed on the ratings.
We included 38 unique studies in the analysis, comprising 33 modelling studies, three observational studies, one quasi-experimental and one experimental study with modelling components.
Measures fell into four broad categories: (i) measures reducing the opportunity for contacts; (ii) measures making contacts safer; (iii) surveillance and response measures; and (iv) multicomponent measures. As comparators, we encountered the operation of schools with no measures in place, less intense measures in place, single versus multicomponent measures in place, or closure of schools.
Across all intervention categories and all study designs, very low- to low-certainty evidence ratings limit our confidence in the findings. Concerns with the quality of modelling studies related to potentially inappropriate assumptions about the model structure and input parameters, and an inadequate assessment of model uncertainty. Concerns with risk of bias in observational studies related to deviations from intended interventions or missing data. Across all categories, few studies reported on implementation or described how measures were implemented. Where we describe effects as 'positive', the direction of the point estimate of the effect favours the intervention(s); 'negative' effects do not favour the intervention.
We found 23 modelling studies assessing measures reducing the opportunity for contacts (i.e. alternating attendance, reduced class size). Most of these studies assessed transmission and healthcare utilisation outcomes, and all of these studies showed a reduction in transmission (e.g. a reduction in the number or proportion of cases, reproduction number) and healthcare utilisation (i.e. fewer hospitalisations) and mixed or negative effects on societal, economic and ecological outcomes (i.e. fewer number of days spent in school).
We identified 11 modelling studies and two observational studies assessing measures making contacts safer (i.e. mask wearing, cleaning, handwashing, ventilation). Five studies assessed the impact of combined measures to make contacts safer. They assessed transmission-related, healthcare utilisation, other health, and societal, economic and ecological outcomes. Most of these studies showed a reduction in transmission, and a reduction in hospitalisations; however, studies showed mixed or negative effects on societal, economic and ecological outcomes (i.e. fewer number of days spent in school).
We identified 13 modelling studies and one observational study assessing surveillance and response measures, including testing and isolation, and symptomatic screening and isolation. Twelve studies focused on mass testing and isolation measures, while two looked specifically at symptom-based screening and isolation. Outcomes included transmission, healthcare utilisation, other health, and societal, economic and ecological outcomes. Most of these studies showed effects in favour of the intervention in terms of reductions in transmission and hospitalisations, however some showed mixed or negative effects on societal, economic and ecological outcomes (e.g. fewer number of days spent in school).
We found three studies that reported outcomes relating to multicomponent measures, where it was not possible to disaggregate the effects of each individual intervention, including one modelling, one observational and one quasi-experimental study. These studies employed interventions, such as physical distancing, modification of school activities, testing, and exemption of high-risk students, using measures such as hand hygiene and mask wearing. Most of these studies showed a reduction in transmission, however some showed mixed or no effects.
As the majority of studies included in the review were modelling studies, there was a lack of empirical, real-world data, which meant that there were very little data on the actual implementation of interventions.