42 Citations (Scopus)
50 Downloads (Pure)


Background: Evidence for face-mask wearing in the community to protect against respiratory disease is unclear.

Aim: To assess effectiveness of wearing face masks in the community to prevent respiratory disease, and recommend improvements to this evidence base.

Methods: We systematically searched Scopus, Embase and MEDLINE for studies evaluating respiratory disease incidence after face-mask wearing (or not). Narrative synthesis and random-effects meta-analysis of attack rates for primary and secondary prevention were performed, subgrouped by design, setting, face barrier type, and who wore the mask. Preferred outcome was influenza-like illness. Grading of Recommendations, Assessment, Development and Evaluations (GRADE) quality assessment was undertaken and evidence base deficits described.

Results: 33 studies (12 randomised control trials (RCTs)) were included. Mask wearing reduced primary infection by 6% (odds ratio (OR): 0.94; 95% CI: 0.75–1.19 for RCTs) to 61% (OR: 0.85; 95% CI: 0.32–2.27; OR: 0.39; 95% CI: 0.18–0.84 and OR: 0.61; 95% CI: 0.45–0.85 for cohort, case–control and cross-sectional studies respectively). RCTs suggested lowest secondary attack rates when both well and ill household members wore masks (OR: 0.81; 95% CI: 0.48–1.37). While RCTs might underestimate effects due to poor compliance and controls wearing masks, observational studies likely overestimate effects, as mask wearing might be associated with other risk-averse behaviours. GRADE was low or very low quality.

Conclusion: Wearing face masks may reduce primary respiratory infection risk, probably by 6–15%. It is important to balance evidence from RCTs and observational studies when their conclusions widely differ and both are at risk of significant bias. COVID-19-specific studies are required.
Original languageEnglish
Article number2000725
Issue number49
Publication statusPublished - 10 Dec 2020


  • coronavirus
  • facemask
  • influenza-like-illness
  • Hajj
  • respiratory infection

Cite this