The Lancet Digital Health  2021.01.19

Policies that increase prevalence of mask-wearing by 10% in a community would more than triple its likelihood of hitting a key COVID-19 control target, a modeling study suggested.


After controlling for physical distancing, population demographics, and other confounders, the odds ratio for achieving an instantaneous reproductive number (R1) less than 1 for SARS-CoV-2 virus transmission was 3.53 (95% CI 2.03-6.43) for each 10% increment in mask-wearing, reported John Brownstein, PhD, of Boston Children’s Hospital, and colleagues in The Lancet Digital Health.

“Past evidence on the effectiveness of mask use against COVID-19 transmission is mixed and setting up [randomized] controlled trials to investigate this is challenging,” Brownstein said in a statement. “Our findings, based on observational data, suggest a community benefit for wearing face masks for slowing the transmission of COVID-19.”


An accompanying editorial by Hannah Clapham, PhD, and Alex Cook, PhD, both of National University of Singapore, characterized the research as “an elegant ecological analysis.” They acknowledged mask wearing has been divisive, but said this study helps to bring some clarity to the debate about whether masks protect the wearer from infection or prevent the wearer from transmitting the virus themselves.


“An ecological analysis … measures the overall effect of face mask wearing on transmission, and thus obviates the need to disentangle the two modes of effect,” the editorialists wrote.

And while observational and laboratory studies are generally considered “weak evidence” for an intervention versus the “gold standard” randomized trial, Clapham and Cook noted any trial trying to prove mask wearing protects others from SARS-CoV-2 would be “logistically challenging, bordering on infeasible and potentially unethical.”

“Well conducted, real-world, observational studies … probably provide the strongest evidence to inform policy,” they wrote.


Brownstein and colleagues examined data from serial cross-sectional surveys administered via SurveyMonkey to randomly selected individuals ages 13 or older in the U.S. from June 3 to July 27. They were asked about mask usage on a four-point scale ranging from “very likely” to “not likely at all” on how likely they were to wear masks “while grocery shopping” or “while visiting with family and friends in their homes,” as well as a range of questions about mask usage.


They analyzed self-reported data with R1 estimates, or the number of secondary cases arising from a single case for a given day, from both the COVID Tracking Project and the open COVID-19 data working group, to estimate transmission control (defined as R1<1). They then estimated the association between mask wearing from the survey data and community transmission control.

Data from 378,207 survey responses were included. About 85% of individuals said they were very likely to wear a mask to the grocery store, while 40% said they would wear one with friends and family. Self-reported mask wearing was higher among women, respondents with a lower income, and Black, Hispanic, and other racial/ethnic groups versus white individuals. The highest percentage of mask wearing occurred along the U.S. coasts and southern border and in large urban areas.


Distancing played an important role in the modeling results as well, showing that states with high mask wearing — but no change from baseline in physical distancing — had a 22% predicted probability of community transmission control, albeit with a wide confidence interval (3%-76%).

“Self-reported mask-wearing is shown to increase the odds of transmission control across all levels of physical distancing, suggesting that any intervention to improve this community-based behavior might be worthwhile,” Brownstein and colleagues wrote.