When the news makes you say "Huh?!?"

The statement certainly seems to be totally at odds with everything public health experts have been telling us for months. And besides, you probably know lots of people who wear masks—maybe even in your own family!—and you would have noticed if almost every single one had become ill. So the statement is certainly startling. But having assured yourself that you did indeed hear it right, how would you determine whether it’s true?

Tracking down the evidence behind scientific claims usually means going in search of the original study. As we’ve seen several times during the pandemic, what ends up in the headlines sometimes goes a bit—or a lot—beyond what the study actually found. We saw that with Gaitergate, the mask study that was said to show that gaiters are less effective than no mask at all (the study didn’t measure mask effectiveness, and the conclusion was mere speculation). We saw it again with the study asking “do eyeglasses lessen the risk of catching COVID?” So what do we find when we go after the source of the statement that 85% of people who wear masks get COVID?

The original study can be found in the Center for Disease Control and Prevention’s publication, Morbidity and Mortality Weekly Report (MMWR), a reputable source relied on by epidemiologists and physicians for decades. So what’s the study about? Pro tip: you can usually get the main gist of a scientific article just from the title; in fact, very often, the main conclusion of the study can be found there. In this case, the title is “Community and Close Contact Exposures Associated with COVID-19 Among Symptomatic Adults ≥18 Years in 11 Outpatient Health Care Facilities—United States, July 2020.” Indeed, just as advertised, the study found that symptomatic people who test positive for the virus that causes COVID-19 are more likely than people who test negative to have had close contacts with other positive individuals and to have engaged in activities where they were more likely to encounter the virus.

Gosh, really? People who caught COVID were more likely than people who didn’t catch COVID to have been around other people with COVID? Yes. That’s one conclusion, and that conclusion is consistent with lots of other studies and not surprising. But what about people who test positive who have no known exposure to the virus? Can we say anything about what activities or practices seem to be most risky? As our communities open up, what is it safe to do and what should we avoid? This study offers some clues.

Briefly, this is how the study was done. The authors compared two groups of people who showed up for testing at 11 testing sites. All of these people had symptoms, but only some of them—called “case-patients”—tested positive for SARS-CoV-2. The symptomatic people whose tests came back negative were called “control-participants.” The authors surveyed these two groups about their contacts and activities in the two weeks before their tests.

As expected, a lot more of the people who tested positive reported having been in contact with someone who had COVID-19: 42% of the case-patients compared with just 14% of the control-participants. Most of those contacts were members of their households. We know that this is the main way that COVID-19 spreads, and the authors emphasize that anyone with a sick family member should be extra careful about isolating that person and maintaining strict hygiene. So far, so good, right? No new news.

But what about the case-patients who reported no known contact with anyone who’d tested positive? This is where the report offers some new information. Let’s dig into the actual data—it’s pretty much all contained in just one table:

Reading tables is a core science skill that can seem a bit daunting at first, but let’s break this down. In the left-hand column are all the kinds of information the authors gathered about the study participants: age, sex, race, education, chronic medical conditions, and types of community exposure. The second and third columns show the number of people in the case-patient group and the control-participant group who have those characteristics, followed (in parentheses) by the percentage of the total that number represents.

You can ask your students questions to see whether they can interpret the table correctly. How many subjects over sixty years of age were in the study (42)? How many tested positive (18) and how many tested negative (24)? What percent of subjects had less than a high school education (12.4%)? What percentage of subjects had less than a high school education and tested positive (10.5%) and negative (1.9%)? When you’re sure your students are confident navigating their way around the table, you can turn to the last column: the p-value.

Some high school students may have learned some basic statistics and understand how p-values are calculated, but for the purposes of understanding this study, it’s really only important that they know this: the smaller the p-value, the more likely it is that the difference you see between the two groups is not due to chance. A p-value of 0.05 means that there is a 95% chance that the difference is real, not random. A p-value of 0.01 means that there’s only a 1% chance that the differences are due to chance.

Remember that in this case, the difference between the two groups is whether or not the subjects tested positive for coronavirus. Scanning down that right-hand column and looking for low p-values, what do you find? One thing that jumps out right away is that testing positive correlates very strongly with race and ethnicity. Hispanic/Latino and Black subjects are much more likely to test positive (p<0.01). This result is consistent with lots of evidence that Black and Latino Americans are less likely to have jobs that can be done at home and are more likely to live in multi-generational households, and therefore are more likely to be exposed to the virus. Education levels also are predictive of who will test positive; people with college degrees are more likely to be able to work at home and reduce their exposure—and sure enough, testing negative is correlated with higher education level at a p-value of <0.01.

I don’t need to say that these results do not mean that Black and Latino people, or people with less education, are innately more vulnerable to the virus, do I? I hope not. But it might be good to check in with your students and make sure they don’t come away with any misconceptions along those lines.

But now, finally, we get to the heart of the paper. Look down the p-value column next to the various community activities. What do you see? That’s right: a p-value of 0.01 for having been to a restaurant in the 14 days before testing. The next lowest p-value (0.22) shows up for having been to a bar or coffee shop. Lots of other activities, such as going shopping, or to a salon, don’t appear to increase the likelihood of testing positive for the virus.

So what does the experience of going to restaurants or bars have in common compared to those other activities? The authors suggest that the difference is that most of the other activities can be done while wearing a mask, reducing the risk of spread. But in a restaurant or a bar, you must remove your mask at least part of the time. This study suggests that these activities, therefore, are riskier than going shopping or even going to a gym.

Remember, way back in May 2020, we discussed how to decide what activities are riskiest and how to spend your “risk budget” wisely? Well, what this study tells you is that eating in restaurants seems to be riskier than shopping or going to a salon (assuming, of course, that you wear your mask and so does everyone else). You can adjust your risk budget accordingly.

Hey, wait a minute. What about the role of mask-wearing in catching COVID? You know, that statement that 85% of people who wear masks catch the virus? You sort of have to dig to find out where that particular statistic comes from. It’s at the very, very end of the table. Participants were asked how often they wear masks: never, rarely, sometimes, often, or always. You can see that the p-value for that section is 0.86, which means that there isn’t a significant difference between the two groups’ mask-wearing habits. About 89% of the subjects who tested negative reported wearing masks often or always, compared to (Bweep! Bweep! Pay attention! Here it comes!) 85% of people who tested positive. Clearly this does not mean that 85% of people who wear masks catch COVID-19!

To think otherwise is to succumb to a well-known fallacy in probabilistic reasoning, sometimes called the inverse fallacy. Formally, the fallacy is to assume that P(A|B) = P(B|A), but if you’re formula-shy, let me give an example where the fallacy is clear. The probability that you’re hearing loud clopping sounds given that there’s a herd of zebras galloping through your neighborhood is pretty high: zebras are not known to pussyfoot. But the probability that there’s a herd of zebras galloping through your neighborhood given that you’re hearing loud clopping sounds is pretty low (unless you live in the middle of the Serengeti): a herd of horses or maybe a too-loud Western movie on your neighbor’s television is more plausible.

In just the same way, the fact that 85% of the subjects who wore masks caught COVID-19 does not mean that the probability of catching COVID-19 if you wear a mask is 85%. Nothing of the sort.

It’s true, of course, that wearing a mask, even most of the time, will not protect you perfectly. If you have to work outside the home, live with a lot of other people who also have high exposure rates, or go to restaurants or bars where people aren’t wearing masks and you aren’t either, you still may contract the virus.

So be careful. Be mindful of the people who because of their jobs or their housing circumstances face greater risks and deserve our special attention to maintaining our social distance, hand-washing hygiene, and mask compliance. If someone in your household tests positive, be careful and keep that person as isolated as possible. Wear your mask whenever you can. Avoid places where people aren’t wearing masks, especially indoors. Maybe take a pass on eating inside restaurants until we’ve reduced the amount of virus circulating in our communities.

And don’t be afraid to go looking for the sources of statements that make you go “huh?” You have the power to figure out whether they’re worth taking seriously.