> While no cause-effect conclusions could be inferred from this observational analysis, the lack of negative correlations between mask usage and COVID-19 cases and deaths suggest that the widespread use of masks at a time when an effective intervention was most needed, i.e., during the strong 2020-2021 autumn-winter peak, was not able to reduce COVID-19 transmission. Moreover, the moderate positive correlation between mask usage and deaths in Western Europe also suggests that the universal use of masks may have had harmful unintended consequences.
Dangerous conclusions, attempting to convince people that wearing masks is harmful.
So many assumptions are made that this 'correlation' gives little insight. For example, place of case infection is not even considered. If most people become infected at gatherings where their mask is left off, but they wear their mask in public, then how is this factored? Or perhaps those who wore masks more frequently were at greater risk of infection?
I've been to the hospital recently, and have witnessed the extreme use of masks there. They would laugh at this article.
The one sentence summary: “These findings indicate that countries with high levels of mask compliance did not perform better than those with low mask usage.“
Should this source be trusted or not? If it’s trusted, does that mean that our efforts to mask were basically a waste of time, or not?
1. It's a publication with post-publication peer review. Meaning: It's not been peer reviewed before publication. That's not invalidating its findings, but it's evidence one should rate as potentially of lower quality until other experts have reviewed it.
2. You should never look at one study, but at all the evidence. It is generally difficult to study the effects of something like mask wearing, as you can't easily do something like an RCT (or if you try it may not tell you what you wanted to learn, i.e. it's even harder to test if others are protected by mask wearing). There are no easy answers, but it's been my impression that experts reviewing all the evidence in the past came to the conclusion that mask wearing likely works quite well.
A single study generally isn't conclusive, especially in a field with as many confounding factors as public health policies.
It'll take a lot of time and hard work to decisively say if there's an effect and to what degree. The various studies and meta-analysis I've read appear to indicate that masks somewhat slowed the transmission rate, but overall didn't greatly reduce final case counts. There's enough uncertainty that I believe we'll need solid RCTs to get better answers. There was on large RCT slated to happen in west African country (Rwanda maybe?) where covid-19 hadn't hit.
Seems legitimate, though that doesn’t mean it’s a particularly valuable analysis.
> If it’s trusted, does that mean that our efforts to mask were basically a waste of time, or not?
It doesn't say.
In a world where everyone wakes up each day, checks the daily cases nearby, and only puts on a mask if it’s over 1000… you'd get near perfect correlation between mask use & covid cases, even if the masks prevent every single transmission.
And that’s not all that out of line with reality.
High rates of Covid absolutely cause increased levels of mask compliance, so seeing a correlation between the two doesn't really tell us much about mask effectiveness.
You could maybe do some time-series analysis to try and get a better idea. The decision to mask today would be influenced by cases in the recent past, and any actual effect of mask wearing would only affect diagnoses 3-5 days later. But I’ve not done that math, and neither has this paper.
While this is a very interesting study, it absolutely cannot show that masks are or are not effective at reducing covid transmission. There are so many confounding factors, the results are basically meaningless.
The most obvious confounding factor is of course that as covid rates rise, people are more likely to wear a mask. This by itself explains the positive correlation between mask compliance and covid rates.
The paper says that the positive correlation is surprising. It is not.
Of course you can’t always see direct causality in a complex system, but aren’t there already controlled experiments that showed that face masks reduce risk of transmission? Those were the basis for health experts to recommend the measure in the first place.
The fact that we've had prior "events" and didn't /don't understand the strengths and weaknesses of the public health tools (e.g., various types of masks) is worrisome.
Looking at what medical professional said during the beginning on the pandemic, the general understanding of the strengths and weaknesses of mask seemed to have been fairly good. Mask are classified and tested based on medical criteria, and hospital and other industries use them during high risk situations since it has been proven effective in such scenarios. The lack of understanding was primarily when mask got taken outside the context of high risk situations and into general use in outdoor and indoor environments, and for purposes which the masks classifications don't consider. Everything from an airplane with circulated air, to people outside on a beach in the fresh breeze. On top of that we have politicians making the decision rather than medical professionals.
There are however fairly low understanding of other public health tools that is outside the scope of what medical professionals can do. Shutting down mass transportation, restricting travel, shutting down social activities. Finland as an example fully closed down their capital during the start of the pandemic, preventing almost all movement in and out of the city. Looking at the graphs, that alone might have been what attribute the low case and death rates, and its not something that the medical profession has many studies on.
It wasn't until deep into the pandemic (18+ months) that, in the USA, the CDC emphasized the KN95 masks or better.
That doesn't say there was a clear understanding. So not only were many effectively not wearing masks, we now can't make reliable heads or tails of the "post match" data.
If that's not negligence, then it was intentional. Neither is a good explanation.
Maybe it was because of unavailability of that much masks of suitable properties (KN95/FFP2) for all people, and inertia to realize that situation?
And then, saying at least Duck&Cover! in case of a thermonuclear explosion, because SDI/Iron Dome/Someshinyshield failed. Better late than never? Or nothing at all?
So they knowingly "exaggerated" the effectiveness of the masks most ppl were using, and to cover their ass there has been - sans the CDC announcement - zero effort to update that narrative? And these are the same ppl and entities insisisting "trust us" and "trust the science"?
This ^^^ isn't a conspiracy theory. This is what happened and is still happening. How am I, or anyone, supposed to truth that architecture?
Masks reduce the risk of transmission a lot that is pretty clear as far as I know but
* Are they used correctly by the majority?
* Are the situations where masks are/were used actually the main contributors to infections overall?
* How well do masks work in situations with long time of contact and wearing (public transport, on the job site, ...)?
* + Probably a lot of other questions
So it can very well be that the impact of masks overall and in comparison with other measures is lower than controlled experiments suggest. If that's the case it does not mean wearing a mask is a bad idea but knowing this could influence decisions on mandating doing it.
I’m not aware of any RCT that could control for the variables to measure transmission. That would be very hard. But two RCT studies have been done to measure infection on the individual wearing it and they show cloth and surgical masks as not significant. Of course there’s a lot of nuance to the results but it’s obvious the signal isn’t there unlike say the vaccine during its RCT.
it absolutely cannot show that masks are or are not effective at reducing covid transmission. There are so many confounding factors, the results are basically meaningless.
It can show that. If masks were effective there would be visible inflection points in the data that correlate with mandates being added and removed, as those make a huge difference to mask wearing. Where I lived before the first mask mandate, almost nobody was wearing them. The day the law changed, nearly 100% were wearing them and it remained that way until the mandates were removed.
We're not talking tiny effect sizes here. These were massive, overnight changes. To justify the impositions on people, there should have been correspondingly massive effects. There were none.
The most obvious confounding factor is of course that as covid rates rise, people are more likely to wear a mask.
Why people wear them is irrelevant to the question of whether they work or not.
There should be an inflection in the exponent of the rate (eg changing only the magnitude of the peak), not necessarily in the direction of the graph, and not necessarily even changing the timing (aka flatten the curve). The difficulty of analyzing the data is what makes studies and articles like this still interesting and relevant to consider.
There's also a long delay between any actions taken and the actual effect on infection rates, and especially hospitalisation and death rates, which makes analysing it even more tricky.
The serial interval of SARS-CoV-2 is/was claimed to be at the start ~3 days. So the delay is very short and this is one reason why we can be so sure masks had no effect. If they did, you would expect to see a sharp and visible inflection point in the graph three days later.
However no such inflection points are visible and this happens repeatedly.
Why do you expect an inflection point? At steady-state, there should be a change of that sort described. But the hypothetical intent was to establish these masking policies while still in the early, exponential portion of the graph. When the spread is doubling every 3 days, shouldn’t we expect it to be hard to visually observe a change to the number of days it takes to double the cases? That thus rhetorically leaves us stuck with a significant difference between “no effect” and “moderate-but impossible-to-quantify effect” however
The serial interval is/was three days, that's not the same thing as doubling time. Serial interval is how long it takes someone who just got infected to become infectious and pass it on.
Regardless, the answer is no. Inflection points should be clearly visible because the underlying dynamic is organic, and mass-scale human intervention is inorganic/synchronized. Remember - the policy is justified exclusively on the basis of creating those inflection points. Nothing else. If no such points are visible the policy is a failure and we're left with only the downsides (of which there are many!).
I think you are thinking of the term discontinuities? Inflection points are notoriously not easily visible in graphs. Here’s an inflection point for example: https://www.calculushowto.com/inflection-point/
I sincerely suggest you to read the history instead, starting from the scandals who have followed all recent pandemic declaration by the WHO: H2N2 (1957), H3N2 (1968), SARS-CoV (2003), H5N1 (2004), H1N1 (2009), MERS-CoV (2012) ALL OF THEM ended with inquiries about their nature, all of those conclude that NO SUCH MEASURE suggested by WHO have had positive effects except for the private lenders of the WHO.
I think that's enough to conclude that on one side peoples are to fragile to Bernaysian propaganda, on the others there are criminals who need to be evicted for the sake of humanity. Sorry for the rude tone, but that's is.
I avoided that on purpose: we are so trained to find answers on click that we forgot how and why search. If you are interested and you look for you'll see that some filters bubble around you have made you only see a certain narrative. Looking for something else, you'll break the bubble discovering on your own something new, something you do not expect.
Is it enough to stimulate your appetite and dig yourself? If you do for all pandemic declaration I cited you'll find the same press coverage just-before, during, after, same patterns and same scandals. Again from OFFICIAL, governmental sources, of course.
If you dig a bit more you'll discover various WHO publications about how flu vaccines, vaccines in general are the future of drug BUSINESS because they target healthy people, so the vastest cohort of consumers of drugs possible concluding that those will the the future earning for Big Pharma.
You really have to dig yourself though because you'll see how hard is finding things without the right keywords and so how easy is to convince even skilled and acculturated people of something thanks to modern way of "consuming" news. Only with that discovering you'll start to dig in anything else.
>The most obvious confounding factor is of course that as covid rates rise, people are more likely to wear a mask.
Another confounding factor is that the West has many densely populated cities AND good record-keeping regarding mortality. As you get into poorer and poorer countries, you must also take into account the likelihood of error in their mortality statistics.
You must also consider that those who are less likely to wear a mask (even with a mandate) may also be otherwise more likely to contract and spread covid in the first place, and as long as there is a sufficiently large network of those people for the virus to operate in, mask mandates may make no overall difference in the mortality even if it were 100% effective in preventing death. Ie, if the virus was always primarily circulating among people unlikely to wear a mask, a mask mandate would make little difference. Mask mandates affect public behavior, but it may very well be that what people do in private is much more significant for transmission.
That seems like a fundamentally flawed counting approach to me.
Suppose a person is infected, but he's showing no symptoms, and doesn't even realize he's infected.
He feels just fine, and goes for a long-distance bike ride. Near the end of it, he's hit by a transport truck that runs a red light, and he suffers severe head injuries.
He's taken to the hospital, where he is tested as a precaution, and his test returns a positive result. Due to the severity of his head injuries, he dies a few hours later.
If that happened in Toronto, then he'd apparently be counted as a "COVID-19 death", despite the infection having absolutely nothing to do with his death.
Of course, that's just one of many ways that somebody infected could die in a way that has absolutely nothing to do with an infection.
Any count that includes things completely unrelated to that which are meant to be counted is a count that shouldn't be trusted.
> Suppose a person is infected, but he's showing no symptoms, and doesn't even realize he's infected ... hit by a transport truck
It's easy to invent a scenario that wouldn't get counted correctly.
Here's one that wouldn't get counted correctly with your method:
"Suppose a person is hospitalized for diabetes and would have a 95% chance of surviving, except that she also has covid which reduces her survival probability to 40%."
There are two or three orders of magnitude more hospitalizations for [diabetes + covid] than [massive head trauma when hit by a truck + covid].
So maybe we should use the counting method that is least inaccurate overall, and not try to optimize for rare events?
In this case, a systematic error wouldn't change the conclusion as long as the same method was used for all countries.
As for the broader argument about people dying with COVID vs people dying of COVID, the excess mortality rates are consistent with the assumption that the reported death rate from COVID in Western nations is pretty accurate, and may even undercount the true figure (see e.g. https://www.nature.com/articles/d41586-022-00104-8).
Suppose a person is your average overweight middle aged sedentary American and they die of a stroke and happen to test positive for the virus.
There's a pretty good chance they died of COVID-19-associated immune thrombocytopenia, even though they may not have been symptomatic when they died.
A lot of those cardiac events are going to go uncounted if a country isn't testing for the virus, or isn't attributing CVAs to the virus when the person didn't really have any respiratory complications / pneumonia.
It's really hard to make conclusion based on mask usage not accounting for many other factors such as population density.
You have places like Finland with 5 mil people and minimum contact vs. Netherlands with 1/8 of the area and 4 times the population. Then you have France, Italy where all people kiss to greet each other and Germany where they barely shake hands etc.
So don't think this is very productive study and I'm generally not in favour of mask mandates (sentiment-wise for we have yet to see reliable data) especially for recent variants.
The alternative to mask mandates was selective use of mask in situation of high risk. If Netherlands is generally a high risk area while Finland is a low risk area, then that would illustrate a failure of mask mandates in favor of selective use. If that is what this study demonstrate then that is an important finding.
To make it even harder, population density is itself not easy to interpret correctly. For instance, a naïve comparison on the population densities of England (432 people/km²) and Spain (94 people/km²) could lead one to believe that England would have a much tougher time than Spain at containing the spread of a contagious disease. However, the population density surrounding the average Spaniard is much higher than the density surrounding the average English person, as Spanish population tends to be much more densely urbanised into flats than the population of England.
Thanks, I didn't know there was an explicit term for this (though I should have guessed). It is a pity though that the map that they use to illustrate the data is divided into administrative divisions that are not really comparable: England, Wales and Scotland for Great Britain, but all 17 autonomous communities for Spain.
Population density doesn’t work that way. Viruses don’t care about how much empty area there is in a country.
Take Greenland for example. Population density is.. zero.
But zoom in and you’ll see they live quite densely.
The number of contacts matter.
This comes down to many factors like urbanization, how advanced the economy is (how large fraction can isolate and work?), family sizes/structures etc.
Raw density is correlated to this, but on a country level it’s almost useless because some countries have all their people in a few areas. Netherlands is pretty special in not having vast wilderness and that’s why it’s dense. But that doesn’t change the fact that a Dutch city looks and functions like almost other European cities.
If I were to guess I’d say a bigger differentiator between (say) Amsterdam and most other comparable cities in regards to diseases would be the bike culture that means thousands get to work without having to use public transport.
Thanks, I thought the same. Population density/distribution have a huge impact on the transmission. Without adjusting for that, the analysis needs to assume good uniformity either within each single country or that the error is the same for each country.
Hold up. Who the heck thought it was a good idea to put masking policy on the Y axis and deaths on the X axis. Also what the heck does Y% mask compliance mean? Does that mean Y% of people wore it 100% of the time and (1-Y%) wore it never? Or does it mean if you picked a random person at a random time, there was a Y% chance they were wearing a mask? There's a distinct lack of clarity in what exactly is being tested here.
I agree with the ranking the paper has gotten so far in the side bar. Its weak point is its study design and methods. Frankly, that part reads like someone who furiously googled some data to win an online argument, not like a scientific paper.
Germany is currently a good testing ground for mask effectiveness. At the beginning of this month, the mask mandate has been dropped in most states, with some exceptions like public transport and hospitals. However, Mecklenburg Vorpommern and Hamburg declared themselves hotspots and kept the full indoor mask mandate, even though their infection rate was about the same as other states'. In MV it had been overturned by a court last week, but in Hamburg the mask mandate is still in effect.
So far, infection rates in MV and Hamburg have been about the same as in the other states (actually a bit higher).
As a caveat, there are still a lot of people wearing masks voluntarily. My guess is about 50% in super markets still wear them, with regional differences.
> My guess is about 50% in super markets still wear them, with regional differences.
My personal gut feeling (being German, but also having been influenced by a "privacy-consciousness bubble" ;-) ) is that there exist people who like the situation that it has become socially acceptable to wear a mask in public (without being concerned about COVID-19 in the slightest), for example because
- this enables you to say a big "f... you" to surveillance cameras
- this might possibly be even a socially acceptable and semi-legal way to "circumvent" the "Vermummungsverbot" on rallys (concerning "Vermummungsverbot": in Germany, you are not allowed to wear face coverings on public rallys; the police often uses this as a pretense to take action against protestors when it does not like the goal of the rally)
In this spirit: Let's continue the situation that disguising yourself in public stays socially accepted. Thus, I will continue wearing a mask in public and simply give the pretense that I am still concerned about COVID-19, or that I believe to belong to a group that is more vulnerable regarding COVID-19.
TLDR: Do not believe that everybody who now still wears a mask in public in Germany does this because he is afraid of COVID-19.
I think people wearing masks for privacy are a very small minority. This is the first time I heard that. People I know just feel uneasy not to wear them, especially indoors, either thinking it will protect them or feel that it's the right thing to do given the still very high infection rates. Some also have to wear them in public transport and just keep them on because it's easier.
I've actually thought about that before, and came to the same conclusions :-)
Can even be combined with sunglasses, though I don't like hoodies, basecaps, or other stuff on my head. I guess I'm black wavy ponytail for any recognition algorithms.
I liked masks. Privacy, warmth during winter, anecdotally had no cold infection in the past 2 years (usually have one every year, and I wasn't stuck at home in the worst period of Covid), protects against each others' bad breath, everyone is quieter since it's harder to talk.
Everyone complaining about masks ignored all the other, more pressing problems around restrictions/Covid. It was kind of like complaining about losing a rare banknote by getting shot in a mugging.
Now we're "Covid free" and no one wears one, and it's really awkward to be the only one with a mask...
I like masks too. I wear a blue surgical mask when I'm confined indoors (e.g. a supermarket) with a bunch of maskless wonders. The blue mask may have poor protective properties (for me), but that's not why I wear it; I wear it to protect my fellow shoppers. In fact the purpose of those surgical masks is to prevent your surgeon or dentist infecting you - not vice-versa. It seems obvious that wearing a mask will reduce the number of droplets released during exhalation.
I'm a vulnerable person; I have a pre-existing lung condition. I therefore try to maintain "social distance" in addition to wearing a mask. Around here, only older people are still wearing masks, even though there's a recent flare-up locally. Impatient people in supermarket checkout queues often press up behind me, as if pushing forward is going to get them through the checkout faster. Those people annoy me more than maskless wonders.
I find I like social distancing more than mask-wearing. Pandemic or not, I find I like having my personal space respected. I don't want to inhale some girl's miasma of cheap perfume, not some homeless man's stink of B.O. and booze.
I never wear a mask in the fresh air. And I've received hostile comments when walking into a pub with a mask on; so I don't wear a mask in pubs (I find it's hard to drink through a mask anyway).
Edit: in addition to the part below, I now see major flaws:
Of course there is a positive correlation between masks and deaths. But the causality is important: when I see more deaths in my country, of course will I wear my mask more often.
This study would have been much better if changepoints would have been looked at.
In this Form there are many unknown von founders to draw conclusions in my opinion.
For now I keep my doubts on this study.
Before edit:
Can someone shed light on the trustworthiness of that publication website as well as the author?
Is “cureus” a trusted resource? I’m always doubtful if there is only one author on a paper - unless it’s a really well known person.
COVID-19 was used by far too many as a vehicle to push their profile or agendas. This result seems to be quite in contrast to the investigations famous labs and institutions did (like RKI and the likes)
At this point the famous „correlation is not causaution“ quote is so abused that its on par with „goto is bad“. Studying correlation does not dismiss the research itself
But here it's really important, because people might mistakenly believe that this research proves that (a) you shouldn't wear a mask (b) you shouldn't tell people to wear masks.
No, that comment does not apply here at all. The saying is "Correlation does not imply Causation". This saying is correct, you can have correlation for lots of reasons without a causal relationship.
However, causation does imply correlation. You cannot have causation without correlation. The contra-positive of this is no-correlation implies no-causation.
This study shows a lack of correlation in the data that they have, which does, within the limitations of that data, imply a lack of causation. There could be confounding factors here, but it isn't completely irrational to think that wearing a cotton mask that only filters out about 10% of the particles in question (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7185834/) might not be very effective at preventing transmission. If not, it is important that governments start stockpiling effective masks to prepare for future virus outbreaks.
You are mistaken. You can have causation without correlation.
For example, suppose my umbrella keeps me dry. But I only take my umbrella out when it's raining. The correlation between my taking my umbrella and my being wet is zero: I either don't get rained on because I have an umbrella, or because it isn't raining; I'm only wet from other causes, unrelated to umbrellas or rain. However, my umbrella causes me to be dry. If I didn't take my umbrella out, holding everything else constant, I would get wet.
In this case, it's quite possible that (a) mask-wearing keeps you safe but (b) mask-wearing is only brought in when Covid is bad; and that the two countervailing effects lead to an estimated correlation of zero.
Yes, if you very carefully set up the data to avoid one entire class of data then you don't have any examples of that class.
When you say `mask-wearing is only brought in when Covid is bad;` that is a hypothesis that can be tested. If that is true then data rolled up at the country level would not be sufficient to determine whether masks are effective, this is a confounding effect.
That is a lack of ability to find correlation in the data. The correlation is still there, your data is just biased in a way that cancels it out or has entire classes of scenarios missing.
My point is that umbrellas ARE correlated with staying dry in the rain, otherwise we shouldn't bother with them as they don't work. The data collected in your example is just extremely biased. The data in this study could be biased, and I call that out. However, at a logical level, you MUST have correlation if there is causation, that is very very nearly a tautology. What does it mean to cause something if the outcome is independent?
My example wasn't set up to avoid one class of data. I said that I might get wet for other, non-rain-related reasons. So there is data with me being wet and with me being dry. And in the data, the observed correlation is zero, even though there is causation. That is because there is a confound. Also, in my toy example, the data collection process is not biased. You can collect the entire population of my days out, and you will still find no correlation. The problem is not sampling bias, it is an unobserved confound.
That was just a toy example. But the linked paper may be an important real world example! You're right that "mask-wearing is only brought in when Covid is bad" is testable. Does the paper test it? Does it attempt to control for endogeneity of mask-wearing? If not, then it may be reporting a spurious null result.
You're right that causation implies correlation in the following sense. "X causes Y" means that if we hold everything else equal, and change X, then Y changes. But that requires holding everything else equal. It's entirely possible that in the data we observe, we see no correlation, because an unobserved confound is biasing our estimates towards 0. That is why serious social scientists spend so much time looking for random variation, and worrying about unobserved confounds. In particular, it is just false to say "they found no correlation, so that proves no causation".
The fact that there is no CLEAR evidence that masks are effective at improving COVID outcomes tells you that if they do have an effect, the size must be small. If they worked well, anywhere near say preventing 50% of infections, then you would not need complicated statistical analysis to tease out marginal hazard ratios from massive data sets.
I think the biggest problem is putting _all_ masks in the same bucket. The selection criteria is too wide.
I have used 'duck beak' ffp3 masks through several flights during peak pandemic, and I had to physically workout my chest muscles to take a full breath of air in. I have also bought disposable masks online that were paper mache level.
Both cases are 'people wearing masks' but I can guarantee that these two masks are not having the same effect.
We can argue logistics and feasibility of everyone using ffp3 masks, but claiming that masks don't have clear evidence is to ignore the whole physics of filtration.
And then you rub your eyes as you nod off for a nap on the plane.
We know that airborne transmission has been a small piece of the puzzle. But masks are visible and so the only thing we could do for security theater. And we turned the response into this virtue signaling debate.
Even this thread has people bending over backwards to defend their views...that's not how science works! I do not believe that masks _increase_ mortality. But I think it's pretty clear that masks have minimal, if any, impact on transmission.
> We know that airborne transmission has been a small piece of the puzzle.
Do you have a quote for this claim?
Everything I read is claiming the exact opposite, that authorities overstated the contact surface transmission and that the majority of SARS-COV2 infections occur through droplets or aerosol.
> And we turned the response into this virtue signaling debate.
By mentioning 'virtue signaling' you are virtue signaling that you share the views of a good chunk of the HN readership. Everyone now knows what side you lean towards, and can assume what affiliated views that you might have.
For what it is worth now, there a number of mask comparative analyses out there now (eg wirecutter) which measured the filtration and give recommendations on which masks are both easy to breathe through and good at filtering. It is not always the ones that are hardest to breathe through (though the counterfactual is generally true: being harder to breathe through generally corresponds to higher filtration)
Maybe, maybe not. First, efficacy below 50%, even at 10%, can be argued to be worthwhile from both an epidemiological point of view and from the purpose of individual protection. Second, I'm not convinced that your conclusion follows from the premise. I don't see why this must be so and you haven't given a reason.
Confounding factors in other comments here are used as an argument against using the findings as evidence for mask efficacy. Note that if those arguments have any merit at all, they are also valid against your conclusion.
It's really hard to prove some things scientifically. But that doesn't imply the effects are small, just that to move from correlation to causality is not easy at all.
It's really hard to prove some things scientifically. But that doesn't imply the effects are small, just that to move from correlation to causality is not easy at all.
In this case it's easy because there's no effect.
The sort of argument you're making above has become common in threads like these, but it's based on what looks a bit like template/pattern matching rather than logic. See also the comment below by ltbarcly3 which is correcting another instance of this. I noticed this problem crops up in COVID threads a lot.
Confounding factors matter when there is a correlation and you're trying to explain causation, because it means you might infer causality wrongly when in reality there's a confounder that explains it differently. Confounding factors are irrelevant when there's no correlation to explain, unless you posit that there actually is an effect but a confounder perfectly balances it out in the opposite direction. If you go there you'd better have really good evidence of it because (a) that's quite unlikely a priori and (b) that evidence is all that would stand between you and hallucinating things that aren't real.
In this case there is no effect to explain. Neither masks nor lockdowns have any observable effect on outcomes and this has been demonstrated six ways from Sunday, from the start of COVID. A lot of people desperately struggle to accept this because of what it implies about the honesty and competence of public health authorities, but the fact remains that these interventions had no effect and there are many studies showing this. You don't really need studies of course. Just looking at case graphs is enough to see this for yourself, because mask mandates were justified on the basis that they'd have a big impact on those graphs. When they don't do that reliably, it automatically means the policy is a failure. It's still nice to rigorously characterize the lack of effect, though.
Because there's no effect that means "correlation doesn't imply causation" is an invalid argument, it means "but real world data can be confounded" is irrelevant because there's nothing to be confounded, and it means a whole lot of other arguments that crop up in discussions of bad science are irrelevant. All these things are common objections to bad science because academics are so keen to announce effects in data that aren't real. They're therefore possibly valid objections to studies where masks or lockdowns are claimed to create real effects, but they aren't valid objections to studies supporting the null hypothesis.
> unless you posit that there actually is an effect but a confounder perfectly balances it out in the opposite direction.
No need for an exact balance; ther study found that there was a positive correlation between mask-wearing and infection. That suggests there was a confounder of some kind, and and that it was more than sufficient to overcompensate for any protective effect.
The only explanation I've seen offered for the positive correlation is that people wore masks more when infection rates were known to be high; that explanation is obviously a confounder.
It found such a correlation only sometimes, and at any rate such a correlation is either irrelevant or harmful to the actual claim of interest (that mask mandates have a negative effect on COVID). The correlation between masking and reduction of COVID is zero.
It certainly doesn't imply the existence of a confounder that takes real effectiveness and reduces it somehow past zero. Once you try to explain the (inconsistent) positive correlation between masks and COVID you're back in the territory where "correlation != causation" is a valid counter argument. As you note, such a correlation could be driven entirely by the false perception of effectiveness, without any actual effectiveness. But it can also be for other reasons e.g. places that were tougher about enforcing mask mandates had generically less competent governments that were more likely to do harmful things like flushing hospital patients into care homes.
More likely: such correlations are spurious and there's no real causality there at all. The fact it doesn't show up consistently would certainly imply that. Also possible that the IHME mask usage data is bogus in some countries, or potentially lots of other things. We have no evidence that can support any explanation over any other.
I agree with most of that, I think. Especially that the correlation between masking and protection isn't supported (I don't agree that it's zero; it looks to me like there's scant evidence either way). I agree that a counfounder can't reduce effectiveness below zero.
> These findings indicate that countries with high levels of mask compliance did not perform better than those with low mask usage.
That could be because the majority of masks (e.g., the ultra-cheap blue ones) are - per the CDC - all but useless. That is, wear a mask (that's no mask at all) and you're all but unmasked.
Oddly enough, currently, (e.g.) Philadelphia reinstated an indoor mask mandate, but made no requirement for an effective mask (according to the data / science).
Two years later. Actual imerial data have shown zero impact on mask wearing and transmission rates among the public.
And still forums cling on to subpar research like this.
Masks among the public have done a fantatic job of filling the oceans with garbage. Nothing positive came out of the mask mandates.
Not a single positive aspect.
Mask, visors and goggles among healthcare personell are completely different. Takes a few minutes to suit up. And can be so painful that the people wearing them have bruises on their faces.
But they protect the wearer and zero slogans out there have yelled masks on all hospital workers to protect the patients.
Never was the case.
Jesus the obsession with forcing one another to cover our faces ( in US even small toddler for crying out loud ), is closer to saudi or taliban religious dogma than science.
Crazy world. Masks in restaurants. Cafees. Masked servent around maskless famous people. During strenous excercise. On small children where they are absolutely a choking hazard, so masks during playtime but masks off during collective naptime.
All the masks on the streets, in bushes and trees and flying about especially around bus stations and stores.
Disposable pink chinese surgical mask on a hipster wearing a save the wales tshirt where there are no madates is about as cringe as it gets.
I assume this was a phone typo, but now I want to know if “save the wales” shirts get sold in Great Britain in any tourist shops, or only to the native hipsters there?
How to drown any effect in the random sea of confounding factors but still claim correlation an potential causality in the end.
> the moderate positive correlation between mask usage and deaths in Western Europe also suggests that the universal use of masks may have had harmful unintended consequences.
Or people in countries with higher mortality were slightly more likely to wear masks.
One argument I've seen against the use of bike helmets is that people naturally gravitate to a certain risk level. Thus helmet wearing, it is argued, leads directly to a slightly more risk-taking style of cycling which cancels out the safety benefit of the helmet.
Agree with that or not (I personally argue that my head is disproportionately in need of protection compared to other body parts - I had a crash that broke the helmet into three pieces and broke my collarbone - but no harm done to my head) - but the same could be said about masks. I've seen many ineffective masks - simple cloth, worn with big gaps around the nose, even ski masks at the ski hill that have a vent for the nose. But now that we're masked, we're doing our thing, we can relax, right? So the usual mob scene at Costco with not even token attempts to keep distance. Would absence of masks cause people to be more cautious?
>One argument I've seen against the use of bike helmets is that people naturally gravitate to a certain risk level. Thus helmet wearing, it is argued, leads directly to a slightly more risk-taking style of cycling which cancels out the safety benefit of the helmet.
This is a nonsensical argument. Do you drive your car 100mph because you have airbags and a seatbelt? If you wore a helmet in the car would you do so?
Masks do not encourage riskier behavior either. It makes current behavior safer that's all.
You are certainly in the minority though, given that millions of people drove much faster than 30mph daily for many many years with neither. Seatbelts weren't even required in cars in the US until 1968, airbags in the 90's. Anecdotally, neither I nor anyone I know thinks this way, the availability of a seatbelt or airbag isn't the determining factor in my driving speed by any measure.
I see a lot of comments conflating the mask types, and exactly who is being protected.
Just to put us all on a common footing, here is a reminder of what the public messaging (at least, here in the UK) has been for two years:
Any mask is essential for us to protect others.
At no point has it been shown that a mask-like face covering of unspecified material provides any significant degree of protection to the wearer against airbourne viruses.
It's still fairly controversial to suggest that mass mask-wearing of cloth masks may not have done much public good anyway.
It's pretty clear that a mask that is actually rated to filter to a certain level will protect the wearer, but then those masks typically have a valve to relieve the extra load on the wearer's respiratory system. Obviously, these masks will also not "protect others', as they dump all of your exhaled air right back into the vicinity.
In the studies I thought they were mainly quoting for that, I thought they usually showed there was similar reductions in both directions, but a greater reduction for source masking. That message somehow then got turned into there being no filtering for inhaling and that a masks only purpose is to protect others
All that being strongly dependent on picking a mask that has decent filtration though, since many don’t and only some do.
One problem with the data is that they have distinct entries for United Kingdom and Northern Ireland, yet the latter is part of the former.
So does United Kingdom as used in the table actually mean Great Britain, or are the NI figures actually also included in the UK figures?
If they were/are distinct, non sub-setted, data sets, then they'd possibly make a useful comparison, but without that extra info it is impossible to interpret those numbers.
What this study shows is that when infection rates are high people wear masks, and conversely, that when people wear masks infection rates are high. Or put another way, it either shows that high infection rates cause people to wear masks, that wearing masks causes high infection rates, or that neither causes the other but some tertiary value not part of the study causes both to rise and fall in tandem.
It does not effectively demonstrate the amount that wearing a mask raises, or reduces, infection rates.
85 comments
[ 5.4 ms ] story [ 114 ms ] thread> While no cause-effect conclusions could be inferred from this observational analysis, the lack of negative correlations between mask usage and COVID-19 cases and deaths suggest that the widespread use of masks at a time when an effective intervention was most needed, i.e., during the strong 2020-2021 autumn-winter peak, was not able to reduce COVID-19 transmission. Moreover, the moderate positive correlation between mask usage and deaths in Western Europe also suggests that the universal use of masks may have had harmful unintended consequences.
So many assumptions are made that this 'correlation' gives little insight. For example, place of case infection is not even considered. If most people become infected at gatherings where their mask is left off, but they wear their mask in public, then how is this factored? Or perhaps those who wore masks more frequently were at greater risk of infection?
I've been to the hospital recently, and have witnessed the extreme use of masks there. They would laugh at this article.
Should this source be trusted or not? If it’s trusted, does that mean that our efforts to mask were basically a waste of time, or not?
2. You should never look at one study, but at all the evidence. It is generally difficult to study the effects of something like mask wearing, as you can't easily do something like an RCT (or if you try it may not tell you what you wanted to learn, i.e. it's even harder to test if others are protected by mask wearing). There are no easy answers, but it's been my impression that experts reviewing all the evidence in the past came to the conclusion that mask wearing likely works quite well.
It'll take a lot of time and hard work to decisively say if there's an effect and to what degree. The various studies and meta-analysis I've read appear to indicate that masks somewhat slowed the transmission rate, but overall didn't greatly reduce final case counts. There's enough uncertainty that I believe we'll need solid RCTs to get better answers. There was on large RCT slated to happen in west African country (Rwanda maybe?) where covid-19 hadn't hit.
Seems legitimate, though that doesn’t mean it’s a particularly valuable analysis.
> If it’s trusted, does that mean that our efforts to mask were basically a waste of time, or not?
It doesn't say.
In a world where everyone wakes up each day, checks the daily cases nearby, and only puts on a mask if it’s over 1000… you'd get near perfect correlation between mask use & covid cases, even if the masks prevent every single transmission.
And that’s not all that out of line with reality.
High rates of Covid absolutely cause increased levels of mask compliance, so seeing a correlation between the two doesn't really tell us much about mask effectiveness.
You could maybe do some time-series analysis to try and get a better idea. The decision to mask today would be influenced by cases in the recent past, and any actual effect of mask wearing would only affect diagnoses 3-5 days later. But I’ve not done that math, and neither has this paper.
The most obvious confounding factor is of course that as covid rates rise, people are more likely to wear a mask. This by itself explains the positive correlation between mask compliance and covid rates.
The paper says that the positive correlation is surprising. It is not.
There are however fairly low understanding of other public health tools that is outside the scope of what medical professionals can do. Shutting down mass transportation, restricting travel, shutting down social activities. Finland as an example fully closed down their capital during the start of the pandemic, preventing almost all movement in and out of the city. Looking at the graphs, that alone might have been what attribute the low case and death rates, and its not something that the medical profession has many studies on.
That doesn't say there was a clear understanding. So not only were many effectively not wearing masks, we now can't make reliable heads or tails of the "post match" data.
If that's not negligence, then it was intentional. Neither is a good explanation.
And then, saying at least Duck&Cover! in case of a thermonuclear explosion, because SDI/Iron Dome/Someshinyshield failed. Better late than never? Or nothing at all?
At some point they had to say something.
This ^^^ isn't a conspiracy theory. This is what happened and is still happening. How am I, or anyone, supposed to truth that architecture?
* Are they used correctly by the majority?
* Are the situations where masks are/were used actually the main contributors to infections overall?
* How well do masks work in situations with long time of contact and wearing (public transport, on the job site, ...)?
* + Probably a lot of other questions
So it can very well be that the impact of masks overall and in comparison with other measures is lower than controlled experiments suggest. If that's the case it does not mean wearing a mask is a bad idea but knowing this could influence decisions on mandating doing it.
It can show that. If masks were effective there would be visible inflection points in the data that correlate with mandates being added and removed, as those make a huge difference to mask wearing. Where I lived before the first mask mandate, almost nobody was wearing them. The day the law changed, nearly 100% were wearing them and it remained that way until the mandates were removed.
We're not talking tiny effect sizes here. These were massive, overnight changes. To justify the impositions on people, there should have been correspondingly massive effects. There were none.
The most obvious confounding factor is of course that as covid rates rise, people are more likely to wear a mask.
Why people wear them is irrelevant to the question of whether they work or not.
However no such inflection points are visible and this happens repeatedly.
Regardless, the answer is no. Inflection points should be clearly visible because the underlying dynamic is organic, and mass-scale human intervention is inorganic/synchronized. Remember - the policy is justified exclusively on the basis of creating those inflection points. Nothing else. If no such points are visible the policy is a failure and we're left with only the downsides (of which there are many!).
I think that's enough to conclude that on one side peoples are to fragile to Bernaysian propaganda, on the others there are criminals who need to be evicted for the sake of humanity. Sorry for the rude tone, but that's is.
Anyway, few starting points:
- the famous, for those who have searched official sources and collected evidence along years to grab that when needed in the future, intervention to the European Council from the May 26, 2010 just an year after the H1N1 pandemic declaration: https://assembly.coe.int/CommitteeDocs/2010/20100126_Stateme... or if you prefer the press https://www.pharmatimes.com/news/eu_to_probe_pharma_over_fal... you might also like some other sources like the British Medical Journal https://doi.org/10.1136/bmj.c2912 for the pdf https://www.bmj.com/bmj/section-pdf/186584?path=/bmj/340/775...
- something from a bit more past https://www.ema.europa.eu/en/documents/product-information/p... while just in the meantime something have already emerged https://doi.org/10.1086/652719 and just check the press at that time https://www.nytimes.com/2006/02/12/weekinreview/greetings-ki...
Is it enough to stimulate your appetite and dig yourself? If you do for all pandemic declaration I cited you'll find the same press coverage just-before, during, after, same patterns and same scandals. Again from OFFICIAL, governmental sources, of course.
If you dig a bit more you'll discover various WHO publications about how flu vaccines, vaccines in general are the future of drug BUSINESS because they target healthy people, so the vastest cohort of consumers of drugs possible concluding that those will the the future earning for Big Pharma.
You really have to dig yourself though because you'll see how hard is finding things without the right keywords and so how easy is to convince even skilled and acculturated people of something thanks to modern way of "consuming" news. Only with that discovering you'll start to dig in anything else.
For instance just to add something "hot" right now: try digging about Ukraine, you can start from https://www.osce.org/files/f/documents/e/7/233896.pdf and https://www.rand.org/pubs/research_briefs/RB10014.html after having read the old DNI report 2004 (who also talk about pandemics) https://www.dni.gov/files/documents/Global%20Trends_Mapping%...
Another confounding factor is that the West has many densely populated cities AND good record-keeping regarding mortality. As you get into poorer and poorer countries, you must also take into account the likelihood of error in their mortality statistics.
You must also consider that those who are less likely to wear a mask (even with a mandate) may also be otherwise more likely to contract and spread covid in the first place, and as long as there is a sufficiently large network of those people for the virus to operate in, mask mandates may make no overall difference in the mortality even if it were 100% effective in preventing death. Ie, if the virus was always primarily circulating among people unlikely to wear a mask, a mask mandate would make little difference. Mask mandates affect public behavior, but it may very well be that what people do in private is much more significant for transmission.
Consider the counting approach used in Toronto, which is Canada's most populous city, and the fourth most populous city in North America.
According to Toronto Public Health:
"Individuals who have died with COVID-19, but not as a result of COVID-19 are included in the case counts for COVID-19 deaths in Toronto."
https://twitter.com/TOPublicHealth/status/127588839006028596...
That seems like a fundamentally flawed counting approach to me.
Suppose a person is infected, but he's showing no symptoms, and doesn't even realize he's infected.
He feels just fine, and goes for a long-distance bike ride. Near the end of it, he's hit by a transport truck that runs a red light, and he suffers severe head injuries.
He's taken to the hospital, where he is tested as a precaution, and his test returns a positive result. Due to the severity of his head injuries, he dies a few hours later.
If that happened in Toronto, then he'd apparently be counted as a "COVID-19 death", despite the infection having absolutely nothing to do with his death.
Of course, that's just one of many ways that somebody infected could die in a way that has absolutely nothing to do with an infection.
Any count that includes things completely unrelated to that which are meant to be counted is a count that shouldn't be trusted.
It's easy to invent a scenario that wouldn't get counted correctly.
Here's one that wouldn't get counted correctly with your method:
"Suppose a person is hospitalized for diabetes and would have a 95% chance of surviving, except that she also has covid which reduces her survival probability to 40%."
There are two or three orders of magnitude more hospitalizations for [diabetes + covid] than [massive head trauma when hit by a truck + covid].
So maybe we should use the counting method that is least inaccurate overall, and not try to optimize for rare events?
As for the broader argument about people dying with COVID vs people dying of COVID, the excess mortality rates are consistent with the assumption that the reported death rate from COVID in Western nations is pretty accurate, and may even undercount the true figure (see e.g. https://www.nature.com/articles/d41586-022-00104-8).
There's a pretty good chance they died of COVID-19-associated immune thrombocytopenia, even though they may not have been symptomatic when they died.
A lot of those cardiac events are going to go uncounted if a country isn't testing for the virus, or isn't attributing CVAs to the virus when the person didn't really have any respiratory complications / pneumonia.
You have places like Finland with 5 mil people and minimum contact vs. Netherlands with 1/8 of the area and 4 times the population. Then you have France, Italy where all people kiss to greet each other and Germany where they barely shake hands etc.
So don't think this is very productive study and I'm generally not in favour of mask mandates (sentiment-wise for we have yet to see reliable data) especially for recent variants.
Take Greenland for example. Population density is.. zero. But zoom in and you’ll see they live quite densely.
The number of contacts matter.
This comes down to many factors like urbanization, how advanced the economy is (how large fraction can isolate and work?), family sizes/structures etc.
Raw density is correlated to this, but on a country level it’s almost useless because some countries have all their people in a few areas. Netherlands is pretty special in not having vast wilderness and that’s why it’s dense. But that doesn’t change the fact that a Dutch city looks and functions like almost other European cities.
If I were to guess I’d say a bigger differentiator between (say) Amsterdam and most other comparable cities in regards to diseases would be the bike culture that means thousands get to work without having to use public transport.
I agree with the ranking the paper has gotten so far in the side bar. Its weak point is its study design and methods. Frankly, that part reads like someone who furiously googled some data to win an online argument, not like a scientific paper.
My personal gut feeling (being German, but also having been influenced by a "privacy-consciousness bubble" ;-) ) is that there exist people who like the situation that it has become socially acceptable to wear a mask in public (without being concerned about COVID-19 in the slightest), for example because
- this enables you to say a big "f... you" to surveillance cameras
- this might possibly be even a socially acceptable and semi-legal way to "circumvent" the "Vermummungsverbot" on rallys (concerning "Vermummungsverbot": in Germany, you are not allowed to wear face coverings on public rallys; the police often uses this as a pretense to take action against protestors when it does not like the goal of the rally)
In this spirit: Let's continue the situation that disguising yourself in public stays socially accepted. Thus, I will continue wearing a mask in public and simply give the pretense that I am still concerned about COVID-19, or that I believe to belong to a group that is more vulnerable regarding COVID-19.
TLDR: Do not believe that everybody who now still wears a mask in public in Germany does this because he is afraid of COVID-19.
That is why I wrote 'having been influenced by a "privacy-consciousness bubble" ;-)'. But I actually do know quite a few such people.
Can even be combined with sunglasses, though I don't like hoodies, basecaps, or other stuff on my head. I guess I'm black wavy ponytail for any recognition algorithms.
shrug
Everyone complaining about masks ignored all the other, more pressing problems around restrictions/Covid. It was kind of like complaining about losing a rare banknote by getting shot in a mugging.
Now we're "Covid free" and no one wears one, and it's really awkward to be the only one with a mask...
I'm a vulnerable person; I have a pre-existing lung condition. I therefore try to maintain "social distance" in addition to wearing a mask. Around here, only older people are still wearing masks, even though there's a recent flare-up locally. Impatient people in supermarket checkout queues often press up behind me, as if pushing forward is going to get them through the checkout faster. Those people annoy me more than maskless wonders.
I find I like social distancing more than mask-wearing. Pandemic or not, I find I like having my personal space respected. I don't want to inhale some girl's miasma of cheap perfume, not some homeless man's stink of B.O. and booze.
I never wear a mask in the fresh air. And I've received hostile comments when walking into a pub with a mask on; so I don't wear a mask in pubs (I find it's hard to drink through a mask anyway).
Before edit: Can someone shed light on the trustworthiness of that publication website as well as the author? Is “cureus” a trusted resource? I’m always doubtful if there is only one author on a paper - unless it’s a really well known person. COVID-19 was used by far too many as a vehicle to push their profile or agendas. This result seems to be quite in contrast to the investigations famous labs and institutions did (like RKI and the likes)
:-(
However, causation does imply correlation. You cannot have causation without correlation. The contra-positive of this is no-correlation implies no-causation.
This study shows a lack of correlation in the data that they have, which does, within the limitations of that data, imply a lack of causation. There could be confounding factors here, but it isn't completely irrational to think that wearing a cotton mask that only filters out about 10% of the particles in question (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7185834/) might not be very effective at preventing transmission. If not, it is important that governments start stockpiling effective masks to prepare for future virus outbreaks.
For example, suppose my umbrella keeps me dry. But I only take my umbrella out when it's raining. The correlation between my taking my umbrella and my being wet is zero: I either don't get rained on because I have an umbrella, or because it isn't raining; I'm only wet from other causes, unrelated to umbrellas or rain. However, my umbrella causes me to be dry. If I didn't take my umbrella out, holding everything else constant, I would get wet.
Here are more details and examples: https://theincidentaleconomist.com/wordpress/causation-witho...
In this case, it's quite possible that (a) mask-wearing keeps you safe but (b) mask-wearing is only brought in when Covid is bad; and that the two countervailing effects lead to an estimated correlation of zero.
When you say `mask-wearing is only brought in when Covid is bad;` that is a hypothesis that can be tested. If that is true then data rolled up at the country level would not be sufficient to determine whether masks are effective, this is a confounding effect.
That is a lack of ability to find correlation in the data. The correlation is still there, your data is just biased in a way that cancels it out or has entire classes of scenarios missing.
My point is that umbrellas ARE correlated with staying dry in the rain, otherwise we shouldn't bother with them as they don't work. The data collected in your example is just extremely biased. The data in this study could be biased, and I call that out. However, at a logical level, you MUST have correlation if there is causation, that is very very nearly a tautology. What does it mean to cause something if the outcome is independent?
That was just a toy example. But the linked paper may be an important real world example! You're right that "mask-wearing is only brought in when Covid is bad" is testable. Does the paper test it? Does it attempt to control for endogeneity of mask-wearing? If not, then it may be reporting a spurious null result.
You're right that causation implies correlation in the following sense. "X causes Y" means that if we hold everything else equal, and change X, then Y changes. But that requires holding everything else equal. It's entirely possible that in the data we observe, we see no correlation, because an unobserved confound is biasing our estimates towards 0. That is why serious social scientists spend so much time looking for random variation, and worrying about unobserved confounds. In particular, it is just false to say "they found no correlation, so that proves no causation".
I have used 'duck beak' ffp3 masks through several flights during peak pandemic, and I had to physically workout my chest muscles to take a full breath of air in. I have also bought disposable masks online that were paper mache level.
Both cases are 'people wearing masks' but I can guarantee that these two masks are not having the same effect.
We can argue logistics and feasibility of everyone using ffp3 masks, but claiming that masks don't have clear evidence is to ignore the whole physics of filtration.
We know that airborne transmission has been a small piece of the puzzle. But masks are visible and so the only thing we could do for security theater. And we turned the response into this virtue signaling debate.
Even this thread has people bending over backwards to defend their views...that's not how science works! I do not believe that masks _increase_ mortality. But I think it's pretty clear that masks have minimal, if any, impact on transmission.
Do you have a quote for this claim?
Everything I read is claiming the exact opposite, that authorities overstated the contact surface transmission and that the majority of SARS-COV2 infections occur through droplets or aerosol.
By mentioning 'virtue signaling' you are virtue signaling that you share the views of a good chunk of the HN readership. Everyone now knows what side you lean towards, and can assume what affiliated views that you might have.
Confounding factors in other comments here are used as an argument against using the findings as evidence for mask efficacy. Note that if those arguments have any merit at all, they are also valid against your conclusion.
It's really hard to prove some things scientifically. But that doesn't imply the effects are small, just that to move from correlation to causality is not easy at all.
In this case it's easy because there's no effect.
The sort of argument you're making above has become common in threads like these, but it's based on what looks a bit like template/pattern matching rather than logic. See also the comment below by ltbarcly3 which is correcting another instance of this. I noticed this problem crops up in COVID threads a lot.
Confounding factors matter when there is a correlation and you're trying to explain causation, because it means you might infer causality wrongly when in reality there's a confounder that explains it differently. Confounding factors are irrelevant when there's no correlation to explain, unless you posit that there actually is an effect but a confounder perfectly balances it out in the opposite direction. If you go there you'd better have really good evidence of it because (a) that's quite unlikely a priori and (b) that evidence is all that would stand between you and hallucinating things that aren't real.
In this case there is no effect to explain. Neither masks nor lockdowns have any observable effect on outcomes and this has been demonstrated six ways from Sunday, from the start of COVID. A lot of people desperately struggle to accept this because of what it implies about the honesty and competence of public health authorities, but the fact remains that these interventions had no effect and there are many studies showing this. You don't really need studies of course. Just looking at case graphs is enough to see this for yourself, because mask mandates were justified on the basis that they'd have a big impact on those graphs. When they don't do that reliably, it automatically means the policy is a failure. It's still nice to rigorously characterize the lack of effect, though.
Because there's no effect that means "correlation doesn't imply causation" is an invalid argument, it means "but real world data can be confounded" is irrelevant because there's nothing to be confounded, and it means a whole lot of other arguments that crop up in discussions of bad science are irrelevant. All these things are common objections to bad science because academics are so keen to announce effects in data that aren't real. They're therefore possibly valid objections to studies where masks or lockdowns are claimed to create real effects, but they aren't valid objections to studies supporting the null hypothesis.
No need for an exact balance; ther study found that there was a positive correlation between mask-wearing and infection. That suggests there was a confounder of some kind, and and that it was more than sufficient to overcompensate for any protective effect.
The only explanation I've seen offered for the positive correlation is that people wore masks more when infection rates were known to be high; that explanation is obviously a confounder.
It certainly doesn't imply the existence of a confounder that takes real effectiveness and reduces it somehow past zero. Once you try to explain the (inconsistent) positive correlation between masks and COVID you're back in the territory where "correlation != causation" is a valid counter argument. As you note, such a correlation could be driven entirely by the false perception of effectiveness, without any actual effectiveness. But it can also be for other reasons e.g. places that were tougher about enforcing mask mandates had generically less competent governments that were more likely to do harmful things like flushing hospital patients into care homes.
More likely: such correlations are spurious and there's no real causality there at all. The fact it doesn't show up consistently would certainly imply that. Also possible that the IHME mask usage data is bogus in some countries, or potentially lots of other things. We have no evidence that can support any explanation over any other.
That could be because the majority of masks (e.g., the ultra-cheap blue ones) are - per the CDC - all but useless. That is, wear a mask (that's no mask at all) and you're all but unmasked.
Oddly enough, currently, (e.g.) Philadelphia reinstated an indoor mask mandate, but made no requirement for an effective mask (according to the data / science).
It makes no sense.
Possibly even worse: you feel better protected as thus behave more riskily with respect to COVID-19 transmission.
Keyword: risk compensation
> https://en.wikipedia.org/wiki/Risk_compensation
And still forums cling on to subpar research like this.
Masks among the public have done a fantatic job of filling the oceans with garbage. Nothing positive came out of the mask mandates. Not a single positive aspect.
Mask, visors and goggles among healthcare personell are completely different. Takes a few minutes to suit up. And can be so painful that the people wearing them have bruises on their faces.
But they protect the wearer and zero slogans out there have yelled masks on all hospital workers to protect the patients. Never was the case.
Jesus the obsession with forcing one another to cover our faces ( in US even small toddler for crying out loud ), is closer to saudi or taliban religious dogma than science.
Crazy world. Masks in restaurants. Cafees. Masked servent around maskless famous people. During strenous excercise. On small children where they are absolutely a choking hazard, so masks during playtime but masks off during collective naptime. All the masks on the streets, in bushes and trees and flying about especially around bus stations and stores. Disposable pink chinese surgical mask on a hipster wearing a save the wales tshirt where there are no madates is about as cringe as it gets.
> the moderate positive correlation between mask usage and deaths in Western Europe also suggests that the universal use of masks may have had harmful unintended consequences.
Or people in countries with higher mortality were slightly more likely to wear masks.
Agree with that or not (I personally argue that my head is disproportionately in need of protection compared to other body parts - I had a crash that broke the helmet into three pieces and broke my collarbone - but no harm done to my head) - but the same could be said about masks. I've seen many ineffective masks - simple cloth, worn with big gaps around the nose, even ski masks at the ski hill that have a vent for the nose. But now that we're masked, we're doing our thing, we can relax, right? So the usual mob scene at Costco with not even token attempts to keep distance. Would absence of masks cause people to be more cautious?
This is a nonsensical argument. Do you drive your car 100mph because you have airbags and a seatbelt? If you wore a helmet in the car would you do so?
Masks do not encourage riskier behavior either. It makes current behavior safer that's all.
Just to put us all on a common footing, here is a reminder of what the public messaging (at least, here in the UK) has been for two years:
Any mask is essential for us to protect others.
At no point has it been shown that a mask-like face covering of unspecified material provides any significant degree of protection to the wearer against airbourne viruses.
It's still fairly controversial to suggest that mass mask-wearing of cloth masks may not have done much public good anyway.
It's pretty clear that a mask that is actually rated to filter to a certain level will protect the wearer, but then those masks typically have a valve to relieve the extra load on the wearer's respiratory system. Obviously, these masks will also not "protect others', as they dump all of your exhaled air right back into the vicinity.
All that being strongly dependent on picking a mask that has decent filtration though, since many don’t and only some do.
>All that being strongly dependent on picking a mask that has decent filtration though
Like I said, masks rated for the job will do the job (of filtering the air you are breathing into your lungs).
So does United Kingdom as used in the table actually mean Great Britain, or are the NI figures actually also included in the UK figures?
If they were/are distinct, non sub-setted, data sets, then they'd possibly make a useful comparison, but without that extra info it is impossible to interpret those numbers.
It does not effectively demonstrate the amount that wearing a mask raises, or reduces, infection rates.