Three people pose for a photo while in masks during an indoor holiday event in Taipei, Taiwan.

Spending long stretches of time with a person with COVID-19 increases the chance of catching the disease, but masking and staying outdoors reduces the risk.Credit: Ceng Shou Yi/NurPhoto via Getty

An analysis of hundreds of COVID-19 cases suggests that face masks are most protective in specific circumstances, such as exposure to a person with COVID-19 that lasts for more than three hours or that takes place indoors1.

The study shows that several of the measures that are collectively known as non-pharmaceutical interventions — such as physical distancing, keeping interactions outdoors and wearing masks — “are in fact helpful” for preventing SARS-CoV-2 transmission, says study co-author Joseph Lewnard, an epidemiologist at the University of California, Berkeley. Previous studies2,3 provided evidence that masking helps to protect against infection, but the latest work shows that it is beneficial even when other measures, such as distancing, aren’t in use.

Close encounters with COVID-19

Although vaccinations and medical treatments are key to controlling the pandemic, non-pharmaceutical interventions remain important public health measures. But it’s difficult to measure the effectiveness of these interventions in real-world settings.

To address this challenge, Lewnard, Seema Jain, a medical epidemiologist at the California Department of Public Health in Richmond, and their colleagues studied cases from roughly 1,280 people in California who tested positive for SARS-CoV-2 between February and September 2021. For each person with COVID-19, the researchers sought out at least one control participant: someone who matched them for factors such as age and sex but who tested negative during the same time period. Participants who’d been exposed to someone known to have COVID-19 provided details about the encounter, such as the setting and duration.

The study found that participants who were not fully vaccinated had the greatest risk of infection when they reported an exposure to someone with COVID-19 that occurred indoors or that lasted for more than three hours. Participants exposed to someone with COVID-19 had lower odds of infection if masks were worn at the encounter than if they weren’t. “This protection is especially important for people who were not yet vaccinated,” says Lewnard. But encounters where masks were worn were linked with additional protection for vaccinated participants, too.

Jain says that the analysis also suggests that masks provide the greatest benefit during high-risk exposures — those lasting for more than three hours, occurring indoors or involving a person from another household. Masking did not show a clear benefit when the participant made direct physical contact with a person known to have COVID-19 or when that person was a member of the participant's household.

The results, which have not yet been peer reviewed, were published on the preprint server medRxiv.

Some scientists are not entirely convinced by the findings. Natalie Dean, a biostatistician at Emory University in Atlanta, Georgia, thinks that non-pharmaceutical interventions are beneficial, but she is hesitant to accept the paper’s estimates of the size of the benefits. That is in part, she says, because of potential biases introduced by the study’s case-control design.

Grant Brown, a biostatistician at the University of Iowa in Iowa City, is also cautious about the study’s precise figures on masking’s benefits, because of the study’s matching of case and control individuals. “Even so, it’s a reasonable approach to a hard problem,” he says. He also notes that the results are supported by studies on the mechanisms of viral shedding.

Kirsten Bibbins-Domingo, an epidemiologist and physician at the University of California, San Francisco, notes that the study began before the rise of the highly transmissible Delta variant, and she, too, agrees that matching infected people with control participants is a challenge. But she says that the authors went to great lengths to overcome this limitation, and the result is a “well designed and well executed” study. The findings, she says, fill a gap in knowledge about the effectiveness of non-pharmaceutical interventions — and could therefore help to inform policies to control viral spread.