Abstract
Objective To review the evidence on the effectiveness of public health measures in reducing the incidence of covid-19, SARS-CoV-2 transmission, and covid-19 mortality.
Design Systematic review and meta-analysis.
Data sources Medline, Embase, CINAHL, Biosis, Joanna Briggs, Global Health, and World Health Organization COVID-19 database (preprints).
Eligibility criteria for study selection Observational and interventional studies that assessed the effectiveness of public health measures in reducing the incidence of covid-19, SARS-CoV-2 transmission, and covid-19 mortality.
Main outcome measures The main outcome measure was incidence of covid-19. Secondary outcomes included SARS-CoV-2 transmission and covid-19 mortality.
Data synthesis DerSimonian Laird random effects meta-analysis was performed to investigate the effect of mask wearing, handwashing, and physical distancing measures on incidence of covid-19. Pooled effect estimates with corresponding 95% confidence intervals were computed, and heterogeneity among studies was assessed using Cochran’s Q test and the I2 metrics, with two tailed P values.
Results 72 studies met the inclusion criteria, of which 35 evaluated individual public health measures and 37 assessed multiple public health measures as a “package of interventions.” Eight of 35 studies were included in the meta-analysis, which indicated a reduction in incidence of covid-19 associated with handwashing (relative risk 0.47, 95% confidence interval 0.19 to 1.12, I2=12%), mask wearing (0.47, 0.29 to 0.75, I2=84%), and physical distancing (0.75, 0.59 to 0.95, I2=87%). Owing to heterogeneity of the studies, meta-analysis was not possible for the outcomes of quarantine and isolation, universal lockdowns, and closures of borders, schools, and workplaces. The effects of these interventions were synthesised descriptively.
Conclusions This systematic review and meta-analysis suggests that several personal protective and social measures, including handwashing, mask wearing, and physical distancing are associated with reductions in the incidence covid-19. Public health efforts to implement public health measures should consider community health and sociocultural needs, and future research is needed to better understand the effectiveness of public health measures in the context of covid-19 vaccination.
Systematic review registration PROSPERO CRD42020178692.
Introduction
The impact of SARS-CoV-2 on global public health and economies has been profound.1 As of 14 October 2021, there were 239 007 759 million cases of confirmed covid-19 and 4 871 841 million deaths with covid-19 worldwide.2
A variety of containment and mitigation strategies have been adopted to adequately respond to covid-19, with the intention of deferring major surges of patients in hospitals and protecting the most vulnerable people from infection, including elderly people and those with comorbidities.3 Strategies to achieve these goals are diverse, commonly based on national risk assessments that include estimation of numbers of patients requiring hospital admission and availability of hospital beds and ventilation support.
Globally, vaccination programmes have proved to be safe and effective and save lives.45 Yet most vaccines do not confer 100% protection, and it is not known how vaccines will prevent future transmission of SARS-CoV-2,6 given emerging variants.789 The proportion of the population that must be vaccinated against covid-19 to reach herd immunity depends greatly on current and future variants.10 This vaccination threshold varies according to the country and population’s response, types of vaccines, groups prioritised for vaccination, and viral mutations, among other factors.6 Until herd immunity to covid-19 is reached, regardless of the already proven high vaccination rates,11 public health preventive strategies are likely to remain as first choice measures in disease prevention,12 particularly in places with a low uptake of covid-19 vaccination. Measures such as lockdown (local and national variant), physical distancing, mandatory use of face masks, and hand hygiene have been implemented as primary preventive strategies to curb the covid-19 pandemic.13
Public health (or non-pharmaceutical) interventions have been shown to be beneficial in fighting respiratory infections transmitted through contact, droplets, and aerosols.1415 Given that SARS-CoV-2 is highly transmissible, it is a challenge to determine which measures might be more effective and sustainable for further prevention.
Substantial benefits in reducing mortality were observed in countries with universal lockdowns in place, such as Australia, New Zealand, Singapore, and China. Universal lockdowns are not, however, sustainable, and more tailored interventions need to be considered; the ones that maintain social lives and keep economies functional while protecting high risk individuals.1617 Substantial variation exists in how different countries and governments have applied public health measures,18 and it has proved a challenge for assessing the effectiveness of individual public health measures, particularly in policy decision making.19
Previous systematic reviews on the effectiveness of public health measures to treat covid-19 lacked the inclusion of analytical studies,20 a comprehensive approach to data synthesis (focusing only on one measure),21 a rigorous assessment of effectiveness of public health measures,22 an assessment of the certainty of the evidence,23 and robust methods for comparative analysis.24 To tackle these gaps, we performed a systematic review of the evidence on the effectiveness of both individual and multiple public health measures in reducing the incidence of covid-19, SARS-CoV-2 transmission, and covid-19 mortality. When feasible we also did a critical appraisal of the evidence and meta-analysis.
Methods
This systematic review and meta-analysis were conducted in accordance with PRISMA25 (supplementary material 1, table 1) and with PROSPERO (supplementary material 1, table 2).
Eligibility criteria
Articles that met the population, intervention, comparison, outcome, and study design criteria were eligible for inclusion in this systematic review (supplementary material 1, table 3). Specifically, preventive public health measures that were tested independently were included in the main analysis. Multiple measures, which generally contain a “package of interventions”, were included as supplementary material owing to the inability to report on the individual effectiveness of measures and comparisons on which package led to enhanced outcomes. The public health measures were identified from published World Health Organization sources that reported on the effectiveness of such measures on a range of communicable diseases, mostly respiratory infections, such as influenza.
Given that the scientific community is concerned about the ability of the numerous mathematical models, which are based on assumptions, to predict the course of virus transmission or effectiveness of interventions,26 this review focused only on empirical studies. We excluded case reports and case studies, modelling and simulation studies, studies that provided a graphical summary of measures without clear statistical assessments or outputs, ecological studies that provided a descriptive summary of the measures without assessing linearity or having comparators, non‐empirical studies (eg, commentaries, editorials, government reports), other reviews, articles involving only individuals exposed to other pathogens that can cause respiratory infections, such as severe acute respiratory syndrome or Middle East respiratory syndrome, and articles in a language other than English.
Information sources
We carried out electronic searches of Medline, Embase, CINAHL (Cumulative Index to Nursing and Allied Health Literature, Ebsco), Global Health, Biosis, Joanna Briggs, and the WHO COVID-19 database (for preprints). A clinical epidemiologist (ST) developed the initial search strategy, which was validated by two senior medical librarians (LR and MD) (supplementary material 1, table 4). The updated search strategy was last performed on 7 June 2021. All citations identified from the database searches were uploaded to Covidence, an online software designed for managing systematic reviews,27 for study selection.
Study selection
Authors ST, DG, SS, AM, ET, JR, XL, WX, IME, and XZ independently screened the titles and abstracts and excluded studies that did not match the inclusion criteria. Discrepancies were resolved in discussion with the main author (ST). The same authors retrieved full text articles and determined whether to include or exclude studies on the basis of predetermined selection criteria. Using a pilot tested data extraction form, authors ST, SS, AM, JR, XL, WX, AM, IME, and XZ independently extracted data on study design, intervention, effect measures, outcomes, results, and limitations. ST, SS, AM, and HW verified the extracted data. Table 5 in supplementary material 1 provides the specific criteria used to assess study designs. Given the heterogeneity and diversity in how studies defined public health measures, we took a common approach to summarise evidence of these interventions (supplementary material 1, table 6).
Risk of bias within individual studies
SS, JR, XL, WX, IME, and XZ independently assessed risk of bias for each study, which was cross checked by ST and HW. For non-interventional observational studies, a ROBINS-I (risk of bias in non-randomised studies of interventions) risk of bias tool was used.28 For interventional studies, a revised tool for assessing risk of bias in randomised trials (RoB 2) tool was used.29 Reviewers rated each domain for overall risk of bias as low, moderate, high, or serious/critical.
Data synthesis
The DerSimonian and Laird method was used for random effects meta-analysis, in which the standard error of the study specific estimates was adjusted to incorporate a measure of the extent of variation, or heterogeneity, among the effects observed for public health measures across different studies. It was assumed that the differences between studies are a result of different, yet related, intervention effects being estimated. If fewer than five studies were included in meta-analysis, we applied a recommended modified Hartung-Knapp-Sidik-Jonkman method.30
Statistical analysis
Because of the differences in the effect metrics reported by the included studies, we could only perform quantitative data synthesis for three interventions: handwashing, face mask wearing, and physical distancing. Odds ratios or relative risks with corresponding 95% confidence intervals were reported for the associations between the public health measures and incidence of covid-19. When necessary, we transformed effect metrics derived from different studies to allow pooled analysis. We used the Dersimonian Laird random effects model to estimate pooled effect estimates along with corresponding 95% confidence intervals for each measure. Heterogeneity among individual studies was assessed using the Cochran Q test and the I2 test.31 All statistical analyses were conducted in R (version 4.0.3) and all P values were two tailed, with P=0.05 considered to be significant. For the remaining studies, when meta-analysis was not feasible, we reported the results in a narrative synthesis.
Public and patient involvement
No patients or members of the public were directly involved in this study as no primary data were collected. A member of the public was, however, asked to read the manuscript after submission.
Results
A total of 36 729 studies were initially screened, of which 36 079 were considered irrelevent. After exclusions, 650 studies were eligible for full text review and 72 met the inclusion criteria. Of these studies, 35 assessed individual interventions and were included in the final synthesis of results (fig 1) and 37 assessed multiple interventions as…
Effectiveness of public health measures in reducing the incidence of covid-19, SARS-CoV-2