Herman Cain's Death, RCTs, and the Mask Debate

Herman Cain (1945-2020)

Source: Gage Skidmore/Wikimedia Commons

Ahead of the Tulsa Rally on June 20th, Herman Cain tweeted, “Masks will not be mandatory for the event… People are fed up.” Nine days letter, on June 29th, he tested positive for COVID-19. On July 1st, he was hospitalized. On July 30th, tragically, he died.

Despite his spokesperson raising the possibility that he contracted COVID while traveling to Arizona (where mask mandates are also optional), given the timeline and the high-risk environment the rally presented (few were wearing masks, in a closed environment, where social distancing posters had been pulled down), a likely explanation is that he contracted it at the rally—but not, mind you, because he wasn’t wearing a mask. Cloth/surgical masks, while they do offer some, offer only limited protection to their wearer. The way masks primarily help prevent the spread of COVID is by keeping those who don’t know they are infected from spreading the virus to others. They help contain infected droplets and aerosols. So it was others not wearing masks at the event that likely increased Cain’s chance of infection. Combined with the spike in COVID cases that Tulsa itself subsequently saw, the aftermath of the maskless Tulsa rally seems to be supporting evidence that masks, indeed, do help mitigate the spread of COVID-19.

It’s not confirming evidence, however. For that, you look to studies indicating that masks can block the majority of virally loaded droplets and aerosols an infected person spreads into the air. Of particular importance is their significant ability to block droplets between 50 to 100 microns, which will carry more potent virus loads, but which can evaporate and become aerosols before they hit the ground.

See notes for sources

Source: Rosieredfield/Wikimedia Commons (modified)

For more confirming evidence, you look to studies in hospitals, cities, large areas, states, and countries where transmission rates dropped after mask mandates went into effect. You compare that to how they did not drop in places that did not mandate masks. You look to modeling, which suggests that if masks are just 50% effective, they could bring down infection rates to non-epidemic levels. You look at reviews of observational and comparative studies, as well as collections of collaborating experts. With all that, cases like Cain’s, or those infected hairstylists in Missouri who didn’t pass it on to 140 of their clients because they were wearing a mask, just tell you what you already know.

But in the anti-mask movement, none of this means anything. All that counts is randomized controlled trials (RCTs). And RCTs, they argue, show that masks don’t work. In reality, however, this is false—and, indeed, RCTs are largely irrelevant to the mask mandate debate. Why?

First, most of the studies anti-maskers cite are about whether masks offer adequate protection to healthcare workers in high-risk environments—not whether they help protect the public by keeping public areas from becoming high-risk environments. In other words, these studies are about whether masks can protect the wearer, not protect others.

Other studies are about whether N95 masks offer more protection than surgical masks—which, by the way, they do. But even if they didn’t, that would not mean masks don’t work. Again, such studies are about whether they protect the wearer, not others—and all protective equipment reaches a point where “adding more” offers less protective value. Take Kevlar vests for example. 7mm of Kevlar stops all the same bullets as 14mm, so an RCT would show no protective difference between the two. But that doesn’t mean Kevlar vests don’t protect their wearer.

And that brings me to the primary reason RCTs on the effectiveness of masks are irrelevant to the debate. Because, while they are necessary and useful for some scientific questions (like drug efficacy), RCTs are useless for others (and others). Indeed, they can even be counterproductive. They are not the gold standard of all scientific evidence that trumps all others, in all situations. Let me explain this little bit of philosophy of science by continuing the analogy about Kevlar vests.

Suppose you wanted to know whether Kevlar vests offered protection to soldiers during wartime. What kind of explanation would you seek out? You’d want a physical explanation of how Kevlar can stop bullets and some lab experiments showing that it does. Shoot some manikins wearing and not wearing Kevlar and see how they fare—that sort of thing. And once you see mortality and injury rates drop in platoons where vests were issued, you’d know they work. No RCTs would be needed. In fact, they would be pretty useless. Why?

Since you can’t shoot live ammunition at people wearing and not wearing Kevlar vests in a lab, the best way to do a RCT on vests would be to give vests to certain groups of wartime soldiers but not others, and compare death and injury rates for each over time. (Even the ethics of this is questionable.) But because the real world is fraught with all kinds of chaotic randomness, such tests could not really be well controlled. If they showed less injury in the vest groups, they would just tell you what you already knew. But if they were unable to find a difference between the groups, that wouldn’t mean vests offer no protection. It would be much more likely that real-world randomness threw off the trial. Those without vests faced easier battles, for example, or those with them didn’t wear them properly. No flawed real-world RCT is going to overturn the evidence you already have that Kevlar vests protect their wearer. Maybe such results would entail that, during wartime, soldiers need thicker vests; but no such study could prove that Kevlar vests offer no protection at all.

This is how it is with masks. We understand physically how they keep infected droplets and aerosols out of the environment, and we have lab evidence that they do exactly that. We have many examples of infection rates slowing after their use becomes widespread, and increasing when they are not. That’s all we need to know in order to infer that they work. And no flawed field RCT, subject to real-world randomness, is going to be able to overturn that. Of course, masks are not perfect. Neither are vests. After all, just one bullet can kill you and vests don’t cover your head. But both masks and vests reduce risk.

Coronavirus Disease 2019 Essential Reads

Update on COVID-19 and How It Changes the Brain

How Political Partisanship Can Be Deadly

To be clear, if it is careful enough, such a study might be able to show which masks (or vests) offer the most protection, and point towards how to best allocate resources. (“Medical workers in these environments need N95s.” “Soldiers in these conditions need 7mm vests.”) Likewise, they might show that the public should be wearing masks of a certain kind during a pandemic. (Multilayered cloth masks that mix cotton and non-woven fabric, like flannel, seem to be about as good as surgical masks.) But such studies might also be counterproductive. If they are unable to find an effect, because they can’t be properly controlled, people in the “anti-vest” lobby might capitalize on that by quoting the one line in the study about not being able to find an effect, but then leave out the part where the authors admit the limits of the study, and clearly state the fact that its results can’t be generalized.

So while Herman Cain’s death, at the hands of COVID-19, after attending a rally at which few wore a mask, is hardly proof of anything, it might actually teach us more than most RCTs on masks do. At the least, it supports what we already know: Mask-wearing helps to prevent the spread of COVID-19.