In a recent post on Sensible Medicine, written on my phone while traveling, I criticized a randomized trial of masking in the BMJ that found Norweigans told to wear masks for 2 weeks had a lower self-reported rate of cold symptoms from 12% to 9%. A reduction of 3%.
In a follow up post, Adam Cifu, my friend and colleague, took me to task. Let me highlight his criticism, and why he is wrong both about the trial, and my Kurt Cobain shirt. AC: “I do not like Vinay’s black Kurt Cobain t-shirt.”
Photo via instagram (you can follow me there).
The parsimonious explanation for Adam’s evidence-based and aesthetic errors is that he has been wearing his mask too tightly.
Adam nicely summarized the trial in his post— but he was wrong in the emailed version (since corrected online) about the mask used—it wasn’t “an FFP3 mask (an N-99)”— it was a surgical mask. Those are the loose, plastic ones that end up in the ocean, strangling turtles and other sea creatures, after being used for armed robberies or college protests.
My original criticism
The big problem with the BMJ study, I argue, is that it could all be placebo effect. The endpoint is self reported. Cold symptoms are vague. Sometimes allergies can mimic them. Just yesterday I woke up with a sore throat, headache and fatigue. Should I report that? Might that be a cold or too many drinks the night before? If you think masks work, and you wore a mask, all things being equal, you might be less likely to report symptoms. In fact, you would be scored as having a ‘respiratory infection’ in this trial if you said you had a stuffy nose, headache and fatigue. (see BMJ methods). That could be a hangover.
I argued investigators should have swabbed participants to ensure that symptoms were actually caused by viruses and not merely something they perceived. (I will return to this).
My second argument was this is a short time horizon. 14 days. What happens if you mask over the entire winter season, or multiple winter seasons? That matters because what is the point of wearing a mask at Whole Foods in December each year if by April you were sick the exact same number of days? Compliance and placebo effects will both diminish with time.
My third point is maybe it’s a good thing to get colds, and avoiding them is silly. No one likes colds but they are a part of life. My grandfather would laugh, and rightly so, if he saw me wearing a mask to go to the mall in January. He would tell me to toughen up, and he would be correct.
Asking for PCR proven virus reduction is moving the goalposts; Reducing symptoms is what matters.
Adam argues that I am “moving of the goalpost” by asking for evidence that masks reduce viruses and not just symptoms of a cold. But, I don’t think he actually believes that, and I offer 2 arguments.
First, prior mask researchers know that symptoms alone is not a good endpoint. In the Bangladesh randomized trial, the authors collected symptoms and indeed mask wearers reported fewer symptoms, but the primary endpoint was not symptoms, but serology proven COVID19. The majority of people with symptoms (75%) did not have COVID in that study.
Second, imagine I take people who are devout catholics. None of them wear a necklace. I randomize them to wear a Virgin Mary necklace for 2 weeks, which I tell them will ward off infections, or not to wear one. Report back how often you get a stuffy nose, headache and fatigue. The result, 9% of necklace wearers have ‘colds’, and 12% of those who don’t wear one. Adam would be the last person to say ‘this shows the Virgin Mary necklace’ prevents colds, but rather, this is a bias susceptible endpoint, and this could all be placebo effect. There is evidence of such effects in medicine. Stenting has a huge effect on anginal symptoms and treadmill time when you know you are stented, and no effect on angina when you don’t (Orbita 1 - on top of meds; It did have an effect in orbita-2 when you stopped all meds… like a lunatic).
More evidence it is placebo effect
Adam says in his piece that “these were not committed mask wearers” implying that the placebo effect is unlikely, but look at this subgroup analysis.
The benefit in the trial was ONLY seen in people who believed masks work (p value for interaction is significant). Only those who believe in the Virgin Mary have fewer colds. Also look at the participants. They aren’t in equipoise. Thousands think masks work and only hundreds are skeptical.
The study proves Adam and David Rind wrong.
Even if you believe the study’s result is true, masking has a modest 25% RRR effect size. Not a 100% effect size. In prior posts on Sensible Medicine, David Rind, a thoughtful internist, said his clinical experience was that masks work. He wore a surgical mask when seeing patients and never got COVID. Adam has a similar experience. But this study shows that at best, only 1 in 33 people in this study didn’t have symptoms because of the mask. The study shows why clinical experience is not useful. I never got sick when I wore my lucky socks in high school.
Conclusion
Overall, I agree with Adam that this is a good trial and that is why I complimented the investigators. I want to reiterate just how much of a failure Anthony Fauci was. He went on TV and said so many ridiculous things about masking and double masking, and, even though he controlled a 5 billion dollar budget, ran 0 RCTs. These European researchers said little about masking, and with ~0 budget ran a study. They are to be celebrated.
Most importantly, the investigators are giving Sensible Medicine 4 posts (one more to come— Ben Recht says we are both wrong— Coming next week.)
Ad hominem: Like Adam, I will end my post with an ad hominem comment. Adam only likes masking because he looks better with one on. ;)
Random asides
Adam criticizes me for not linking to the paper. I was writing it on my phone.
Some commenters criticize me for not summarizing the paper before criticizing it. I assume you read the paper.
Do I contradict myself between Tamiflu vs. masking. This is a longer aside. Obviously ultimately what matters is what people feel, but one purpose of science is understanding the mechanism. Tamiflu or any drug might lower PCR positive (asymptomatic) swabs and I would care less because that is not a clinical endpoint. A Virgin Mary necklace might alter symptomatic colds but I could care less because I think it is just a placebo. Symptom driven swabbing and random swabs are meant to establish scientific validity— the way a placebo controlled trial does. Darrel Francis put it well with stenting. Orbita is meant to test if stenting is actually better than a placebo, but if it works, and you stent obviously you get a bigger effect because there is real effect + placebo effect in real life.
The phrase is 'couldn't care less', not 'could care less'. As is, you are implying the opposite of what you actually mean.
Do the socks need to worn and unwashed or unworn and washed?