Few issues during this evolving pandemic have elicited as much debate as the effectiveness of mask wearing.
Public-health officials originally told people not to wear them, because they feared shortages for health-care workers.
Then they told us that cloth masks had limitations, and debated whether they protected the mask wearer.
In contrast, countries like Japan, South Korea and Taiwan simply masked up to great effect. They recognized the absence of evidence is not the same as evidence of absence.
Meanwhile, medical authorities and journalists pointed to existing studies on masks and concluded the statistical evidence wasn’t great.
And that assumption brings us a recent study in the Journal of Travel Medicine that examined its veracity.
Turns out, like so many things in this forever pandemic, we weren’t reading the data properly.
“There weren’t any studies that conclusively showed masks were not effective,” said Alex Siegenfeld, a PhD student at the Massachusetts Institute of Technology. “Yet common sense just got undervalued.”
Most of us now understand the logic of mask wearing.
Masks filter out viral-containing particles of various sizes. Some masks filter out more particles than others. In any case, the amount of viral particles transmitted between an infected and uninfected citizen will be greatly reduced if one or both are wearing masks. Masks just slow down disease transmission.
Yet proving this efficacy has been something of a conundrum. Siegenfeld and his colleagues, including Yaneer Bar-Yam, Nassim Nicholas Taleb and Pratyush Kollepara, looked at 23 different studies on mask effectiveness to find out why.
Some of the studies were surveys, and others were randomized clinical trials in which some people used masks and the control groups did not. Because the studies were drawn from a review of existing evidence at the beginning of the pandemic, most involved a variety of respiratory viruses, but not COVID-19.
What Siegenfeld and company found is rather instructive about modern-day thinking and the human propensity to make assumptions.
The studies — there were about 10 — that did not find masks to be effective, for example, didn’t have a large enough sample size to prove anything. In other words, they had no statistical power.
But all the studies that had sufficient statistical power found that mask wearing dramatically reduced transmission.
So there was strong evidence to support mask wearing, but it wasn’t generally recognized in the studies themselves because of faulty assumptions about sample size.
The reason why many of the studies were underpowered seems almost comical. In many trials participants wore masks for less than half of the time, while the researchers assumed high rates of compliance, which then deflated their results when determining sample size.
A study on the Hajj pilgrimage to Mecca, where respiratory infection is always a concern, assumed that were masks effective, there would be a 50-per-cent reduction in infection probability between the control and mask group.
But data in the study indicated that mask adherence was only 3.2 per cent. That meant mask wearing would only result, at most, in a 3.2-per-cent reduction in the probability of infection.
As a consequence, many randomized trials that compared mask wearers to unmasked people frequently found only small effects in curtailing infection because people weren’t wearing the masks regularly.
Not surprisingly, low adherence to mask-wearing protocols simply undermined the reliability of many studies.
“In a lot of the studies the researchers didn’t go back and check the validity of their assumptions,” Siegenfeld told The Tyee.
So when medical folks took a look at the existing evidence they found it either mixed, unreliable or not convincing enough.
But when Siegenfeld and colleagues identified and separated out those studies with a high enough combination of sample size and mask adherence, they found that masks repeatedly made a significant difference in stopping infections. They not only prevented the wearer from spreading the virus but protected the wearer from getting infected.
As the study itself plainly noted, “The studies that did not find statistically significant effects prove only that masks cannot offer protection if they are not worn.”
But the study was flawed. It tested all participants after exposure with an antibody test without acknowledging the test’s weaknesses. Antibody tests have a tendency to find false positives and are not as reliable as PCR tests. But these reliability effects weren’t accounted for in the study.
If they had been, the study would have found that mask-wearing recommendations did indeed reduce infections in a statistically significant manner, said Siegenfeld.
Another issue not addressed by the studies were non-linear effects. In other words, if a few people in a room wear masks, transmission is reduced by a small amount.
But if everyone wears a mask, there is an enormous effect way beyond what might have been expected. In other words, multiple behavioural changes can have unexpected power.
As a result of our collective misinterpretation of flawed data, we acted slowly and offered confused messages about masks.
This delay accelerated transmission of the virus and death in many jurisdictions.
In the face of uncertainty, the status quo undervalued common sense.
Authorities didn’t want to endorse masks because they couldn’t find convincing evidence.
And now nobody wants to admit that just maybe they should have acted in the interests of common sense.
“There is a feeling that there is a lack of accountability,” said Siegenfeld. “We expect science to get it right, but there is a lack of a feedback loop. Nobody wants to say this is how we got it wrong and this is how we are going to prevent this kind of mistake in the future.”
Read more: Coronavirus