Cloth Masks Suck
Refusing to speculate got us into a whole heap of trouble.
During the first few months of the pandemic, a shortage of medical masks left the United States scrambling to find workarounds. CDC authorities advised citizens to protect themselves and others by using cloth masks — and newspaper headlines were soon trumpeting heartwarming stories about grandmothers pitching in to sew stylish masks for desperate healthcare workers. In early 2020, cloth masks seemed like a good-faith attempt to do the best we could in a difficult situation. Hindsight tells a darker story. A large randomized controlled trial (RCT) in 2022 showed that while medical masks offer highly effective protection against Covid infection, cloth masks offered no protection whatsoever. In other words, it’s now clear that the only thing cloth masks were offering us in 2020 was a false sense of security. The most tragic aspect of the hindsight is that the 2022 RCT was simply confirming an older RCT showing that use of cloth masks slightly increased the risk of respiratory infections in a hospital setting. The abstract of MacIntyre and colleagues’ 2015 report concluded, “cloth masks should not be recommended.”
How did we blunder into mandating the use of a defective medical device? To understand how the disaster happened, we need to look back at the thinking of public health officials in the difficult early days of the pandemic. It’s well known that very small water droplets can pass through cloth. To block small-droplet aerosol transmission, proper medical masks use a sophisticated melt-blown inner layer that induces electric charge interactions to entrap virus-containing droplets. Unfortunately, the general policy consensus in early 2020 was that the melt-blown layer wasn’t really necessary because the virus was assumed to be transmitted via larger droplets that can be blocked by ordinary cloth. Public health officials concluded there was too little evidence to support the conclusion that SARS-CoV-2 is transmitted via small-droplet aerosols.
With hindsight, we can clearly see that the large-droplet consensus was itself an inconclusive guess that ignored available scientific evidence. I remember reading the 2015 MacIntyre study in March of 2020 and immediately thinking that telling everybody to wear cloth masks was unbelievably terrible advice. The writing was on the wall. My family ignored the cloth mask advice and instead made do with some old respirators we had kicking around in the basement from home maintenance and crafts projects.
The popularity of ineffective cloth fashion masks delayed the use of proper medical masks that actually do offer highly effective protection against Covid transmission. I watched in horror while respected public health officials continued to wear festive cloth logo masks in television interviews through late 2020, long after proper medical masks had become widely available from online retailers. This morning, I bought a loose-weave cloth mask in the gift shop of one of the nation’s top research hospitals. Three years into the pandemic it remains an ongoing fiasco.
Policymakers made a big show of rejecting reckless speculation about aerosol transmission while recklessly embracing more comforting and politically convenient lines of speculation about large-droplet transmission. We used the battlecry “follow the science” to rationalize status quo bias. In the future, it will be important to remember that the scientific method begins with speculation about which ideas best comport with existing scientific knowledge. If policymakers had actually used the scientific method, cloth masks wouldn’t have passed this test.
In traditional statistics, one first develops a null hypothesis that essentially reflects status quo thinking and then sets out to see whether experimental evidence can convincingly reject the null hypothesis. A common blunder in traditional statistics is to fallaciously imagine that a failure to disprove the null hypothesis constitutes affirmative evidence in favor of the null hypothesis. Recognizing this problem, Thomas Bayes developed statistical methods that seek to estimate the probability that any given hypothesis is true. My theory is that a more Bayesian approach could have saved us from the cloth mask disaster. Rather than asking whether the status quo large-droplet null hypothesis had been conclusively disproven, we would have instead asked whether the large-droplet hypothesis or the aerosol hypothesis was a better fit for the available data. We would have accepted the 2015 MacIntyre conclusion that cloth masks don’t work as more in line with small-droplet aerosol transmission. It might not have taken us so long to wake up to Li and colleagues’ April 2020 inference, “Aerosol transmission of SARS-CoV-2 due to poor ventilation may explain the community spread of COVID-19.” The lesson of the tragic saga seems to be that rather than deciding whether status quo thinking has been disproven, we should instead judge which available hypotheses best fit the available evidence.
This article was originally posted on Medium in March 2023