Hacker News Reader

> Because he's being generous: the 11% difference was on a baseline seroprevalence of 0.76%.

Wait a second.

> In other words: you'd have to get 1300 people to wear masks to prevent one seropositive (in case you're wondering, the confidence interval on that value overlaps zero.)

You are assuming 0.76% prevalence. This is, I assume, in specific point of time, but somehow, you extend this to long-term intervention.

If you spin COVID-19 "but only 0.5% of population is infected _right now_", any intervention will seem extremely ineffective. "Oh, you want to vaccinate 300 millions of people to prevent 2000 deaths" (without saying it's 2000 deaths _per day_).

> You are assuming 0.76% prevalence.

I am not. I'm using the numbers from the paper to illustrate a point.

Even if seroprevalence was as high as 6% at any given time (which is about as high as it ever was in NYC, for example), an 11% difference is 0.66%, or 1 case per 152 people wearing masks. And that's in a fully unvaccinated population.

We're simply not talking about huge differences here. Certainly nothing close to a statistically meaningful difference in R0.

To be fair, it's not necessarily the case that an unvaccinated population is the best case for masks. If vaccination reduces infectiousness, adding a mask on top could push a small exposure down to a noninfection, whereas a one-time large exposure would overwhelm it.

> In other words: you'd have to get 1300 people to wear masks to prevent one seropositive. (Oh, and by the way? The confidence interval on that overlaps zero.)

That's not accurate unless I'm missing something... sticking with just the surgical mask arm -- the effect size of 0.08% was based on a binary mask vs. non-mask intervention. It has no relation to the number of people masking. The intervention was getting more people to mask -- they succeeded mildly, where 6/10 people still hadn't masked. But that was sufficient for a 10% reduction in seropositive cases.

Edit;

They actually address this specifically;

Our estimates suggest that mask-wearing increased by 28.8 percentage points, corresponding to an estimated 51,347 additional adults wearing masks in intervention villages, and this effect was persistent even after active mask promotion was discontinued. The intervention led to a 9.3% reduction in symptomatic SARS-CoV-2 seroprevalence (which corresponds to a 103 fewer symptomatic seropositives) and an 11.9% reduction in the prevalence of COVID-like symptoms, corresponding to 1,587 fewer people reporting these symptoms.

So for 51,000 people additional people wearing masks -- 1,600 people were spared Covid symptoms and 103 people were spared symptoms + seropostive diagnosis. This with only a total of 42% of people wearing masks and a relatively low level of underlying Covid.

It's an especially obvious intervention based on the exponential potential of Covid infection and early R0 of 6 or whatever it was, plus how effective the surgical masks were (especially in older age groups via Figure 3).

> That's not accurate unless I'm missing something... sticking with just the surgical mask arm -- the effect size of 0.08% was based on a binary mask vs. non-mask intervention.

The RCT was a binary test of each kind of mask. Any other outcome is imaginary. There's no mathematical adjustment that can tell you what would have happened in the hypothetical alternative universe where everyone wore a surgical mask. That's just storytelling with numbers.

> But that was sufficient for a 10% reduction in seropositive cases.

No. It was a an 11% relative reduction, on a 0.76% seropositivity rate. The actual effect size was about 0.09% (I wrote 0.08% earlier, because I just divided the two numbers, and didn't use the figures from the paper). When you compare the cost of an intervention to the benefits, you use the absolute effect size, because you don't pay relative costs for your interventions (e.g. is it worth forcing every toddler in the world to wear a mask while sleeping for a "huge" 66% relative risk reduction from 0.0000003% to 0.0000001%? Probably not unless you're a crazy person. The costs vastly outweigh the benefits, even though the "relative" benefit is strong.)

> Our estimates suggest that mask-wearing increased by 28.8 percentage points, corresponding to an estimated 51,347 additional adults wearing masks in intervention villages, and this effect was persistent even after active mask promotion was discontinued. The intervention led to a 9.3% reduction in symptomatic SARS-CoV-2 seroprevalence (which corresponds to a 103 fewer symptomatic seropositives) and an 11.9% reduction in the prevalence of COVID-like symptoms, corresponding to 1,587 fewer people reporting these symptoms.

This is garbage. They're combining the two mask arms and claiming a combined effect size, even though the cloth mask arm was shown to have no effect. You can't do that. They're spinning like crazy to hide a weak outcome. They also don't report the raw numbers for the arms, and they don't dwell too much on the fact that "distancing" increased in the mask villages over the controls. That could easily swamp the measured effects. But I digress.

The surgical mask villages had 106,201 people, and an 11.2% relative reduction in seroprevalence, from 0.76% to 0.67%. The cloth mask villages saw essentially no difference at all: 0.76% vs 0.74%

There's simply no way to look at this data as anything than incredibly weak evidence for masks. I honestly can't believe we're fighting about something that makes, at best, a tiny fraction of a percent difference in outcome. Mass hysteria.

> I honestly can't believe we're fighting about something that makes, at best, a tiny fraction of a percent difference in outcome. Mass hysteria.

I whole-heartedly don't understand this.. in the places with universal masking, there is essentially no Covid (most of Asia, even cities like SF). Surgical masks are obviously sufficient to prevent most infection in hospital and other clinical settings. There's evidence that even a small portion of people wearing them in public reduces the incidence of disease. And your takeaway is that it's mass hysteria to encourage an extraordinarily cheap NPI?

> in the places with universal masking, there is essentially no Covid (most of Asia, even cities like SF).

Just completely, utterly wrong. Japan had a huge spike this winter. Korea is currently at all-time highs. Vietnam is currently near all-time highs. Thailand peaked at 20,000 cases per day, and is currently at 10,000.

These numbers are lower than the US in absolute terms, but claiming "essentially no Covid" means only that you aren't informed of the reality of the situation. Covid is running rampant across Asia, even now.

> There's evidence that even a small portion of people wearing them in public reduces the incidence of disease.

There is literally no evidence of this. The few RCTs that have performed suggest the opposite: you have to put masks on a huge number of people (hundreds to thousands) to prevent even a single case.

> And your takeaway is that it's mass hysteria to encourage an extraordinarily cheap NPI?

No, my takeaway is that there's very little evidence supporting the use of masks. Because there isn't. And most of what there is, is of exceptionally low quality.

Spike in Japan this winter? lol ... yes it's really scary here 73 cases a day nation wide. The only spikes came during summer due to Olympics and Paralympics. Please stop making stuff up.

The major difference to Germany I see between the two countries in terms of measures, Japanese were masks (vaccinated) and Mask Wearing in Germany has become political as it did in the US.