Facemasks catch viruses – so why is the evidence on their effectiveness so confusing?

When it comes to assessing the efficacy of facemasks some types of trial are better than others, experts say.

  • 13 November 2023
  • 6 min read
  • by Linda Geddes
Protective masks. Photo by Mika Baumeister on Unsplash
Protective masks. Photo by Mika Baumeister on Unsplash


When the next pandemic strikes, policymakers will face tough decisions about how best to protect the public’s health. Mask-wearing is a straightforward intervention that could be recommended alongside other measures, assuming the pathogen spreads through respiratory droplets as COVID-19 and influenza do.

Yet ongoing disagreement over the effectiveness of masks could undermine policymakers’ ability to take decisive action when new threats appear, researchers warn.

Overall, mask use was associated with a 9% reduction in symptomatic COVID-19 infections, and a 35% reduction among people aged 60 years or older.

A new study by former US Centers for Disease Control and Prevention (CDC) Director Dr Tom Frieden and colleagues tries to cut through the misinformation and summarise what the COVID-19 pandemic has taught us about masks and their ability to prevent disease transmission.

Gold standard

Randomised clinical trials (RCTs) are often regarded as the gold standard for evaluating the effectiveness of interventions or treatments. They involve randomly assigning people to either receive the intervention of interest or a control intervention, like no treatment or a placebo, and then following what happens to them.

Yet, while a well-designed RCT can supply valuable information in many areas of public health and medical practice – from speed limits on highways to seatbelt and motorcycle helmet laws – they are difficult to conduct, Frieden and colleagues said.

Similarly, during a pandemic caused by a lethal respiratory virus, it is difficult to find a setting in which it is ethical and workable to randomise people to masking vs. no masking. The time and funding needed to properly design and implement such trials further limits their feasibility. Also, different types of masks and contexts in which people might be recommended to wear them, with no guarantee that participants will do so correctly, may all affect study results.

Just two RCTs on community mask use have been published in the context of COVID-19. The first study randomised 3,030 people in Denmark to either receive surgical masks and a recommendation to wear them outside the home, or no recommendation. The results were inconclusive: although COVID-19 infections were 20% lower in the mask group, the study was too small to be sure this wasn’t down to chance.

The second study involved more than more than 340,000 people living in 6 villages in Bangladesh, randomised to receive cloth or surgical masks, plus information promoting their use, or no such intervention. Mask use was three times more common (42% vs. 13%) in villages assigned to receive masks – particularly when people visited mosques. Overall, mask use was associated with a 9% reduction in symptomatic COVID-19 infections, and a 35% reduction among people aged 60 years or older. 


Another type of study that is sometimes used to probe areas of medical uncertainty is meta-analysis. These do not create new data, but statistically combine the results from existing RCTs to draw conclusions based on a larger set of data. However, if the studies differ in their methods, populations, contexts or measurements, their results and conclusions may be unreliable.

“As with any intervention, they are just one of several measures that are needed – but masks are an important tool to prevent infections, protect healthcare workers and the public, and save lives.”

Dr Tom Frieden, former US Centers for Disease Control and Prevention (CDC) Director

Frieden and colleagues pointed to a recent Cochrane review on interventions to reduce the spread of respiratory viruses as an example of this. It pooled the results of 78 RCTs examining the impact of various physical interventions, including hand hygiene and masks, for preventing respiratory illnesses – mostly influenza – during non-epidemic flu seasons when viral circulation and transmission was lower than during the COVID-19 pandemic.

The results did not show a clear reduction in respiratory viral infection associated with mask wearing, which some people have seized upon as evidence that masks don’t work. However, the authors clearly said that the high risk of bias in the trials, variation in outcome measurements and relatively low adherence to the interventions meant that they couldn’t be certain as to whether wearing masks helps slow respiratory virus spread. They called for large, well-designed RCTs to address the effectiveness of these interventions in multiple settings and populations.

Observational studies

Observational studies can help researchers to investigate relationships that can’t be tested in RCTs and understand what happens in real life situations. They involve following groups of people without trying to change who is or isn’t exposed to the intervention of interest. Frieden and colleagues reviewed a number of observational studies conducted during the COVID-19 pandemic and said that high quality data showed the effectiveness of masks to prevent COVID-19 transmission on airplanes, in schools and among household and community contacts of infected people.

An outbreak on the USS Theodore Roosevelt aircraft carrier – which occurred early in the pandemic when few people would have had immunity to SARS-CoV-2 – was particularly instructive, they said. In this high-risk environment, more than 80% of people who reported not masking were infected, while the odds of infection were 30% lower for those who wore masks.

Frieden’s team also reviewed data from other types of study – including laboratory studies that showed masks reduced the spread of infectious respiratory particles, and that N95 respirators offered better protection than surgical or cotton masks.

Frieden and colleagues stressed that whether masks work is a different question from whether mask mandates work. If adherence is low, mandates are unlikely to have an impact. They also acknowledged that it can be difficult for young children to wear well-fitting masks, and that it was possible masking may impede cognitive and social development, meaning the risks and benefits should be carefully weighed up.

However, they concluded that, taken together, the available evidence strongly suggests that masks can reduce the spread of COVID-19, and that high-quality masks should play an important role in future respiratory pandemics.

“As with any intervention, they are just one of several measures that are needed – but masks are an important tool to prevent infections, protect healthcare workers and the public, and save lives,” Frieden told VaccinesWork.

He also stressed that the protection they supply isn’t perfect, and is more important for certain people, in certain places and at certain times. “People who are elderly, immunosuppressed, or concerned about reducing their risk of infection, can wear a mask, particularly an N95 or KN95 mask, indoors, particularly in crowded spaces during times when COVID-19 is spreading widely, to reduce risk,” he said. “For people who are ill or who have been exposed to COVID-19 and may have asymptomatic infections, wearing a mask protects others.”