35

There's a study that concluded that wearing face masks for extended periods of time (8h-10h) causes a hypoxia:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8152240/

COVID-19 virus has caused the world’s deadliest pandemic. Early April 2020, the Delhi Government made it compulsory for people to wear face masks while going outdoors to curb disease spread. Prolonged use of surgical masks during the pandemic has been reported to cause many adverse effects. Intermittent hypoxia has been shown to activate erythropoietin (EPO leading to increased hemoglobin mass.

This study including 19504 blood donors spanning over one and a half year shows that prolonged use of face mask by blood donors may lead to intermittent hypoxia and consequent increase in hemoglobin mass.

It seems peer reviewed and published in a real journal, is it credible and authentic? Are the conclusions valid?

2
81

This study has way too many problems to draw such a conclusion. It can only be taken as an indication that it may be worth further study, or may corroborate or contradict other, more carefully controlled, studies.

The main problem is simply that by the nature of the study, the authors are comparing two groups of people, at two different times and circumstances. They cant go back to the people in the second (2020/covid) group to check for individual data previously. Essentially they gamble that the profile of blood donors in Delhi did not change so that any change is likely due to mask-wearing.

But that's an assumption, not a fact. There have been other studies in other fields where it appeared two things were linked, and only more careful studies showed that the assumptions were incorrect or dubious. So we must tread very carefully.

A good paper will try to anticipate other factors ("confounds") that might have muddied the results, work around those it can, and at least highlight or try to estimate the effect of those it can't, so others are aware, and can either allow for it, form their own views, or design further experiments to address them. A bad paper will just conclude without such careful in-depth thought.

We don't know what factors might exist, or which factors might have had an effect (if any). Without reading the paper in-depth, I can guess at a few that I'd expect a gold-plated experiment would address. For example:

  • We need to consider how representative blood donors are, of society at large, to draw such a conclusion from blood-donor-only data. If you study only blood donors, you will potentially exclude or under-represent huge swathes of society, with other characteristics. Anyone under 18. Anyone over (say) 65. Ages with above or below average representation among donors (maybe the population is older, but donors tend to be younger or whatever). People with a huge range of specific medical conditions. Rural dwellers (fewer local centres). Probably some ethnicities/genders/intersections are perceived to be socially suspicious or dissuaded by their cultural norms, and won't be as represented. Probably much more. Perhaps fitter people have lowered Hb but someone with already lowered Hb doesn't have as much of a reduction. We just don't know. The criteria and characteristics of blood donors as a group is very different from society overall. So all we can say is these results apply to the kinds of people who donate blood. Not to "people generally".
  • The tests used for blood donor measurements are often designed to emphasise a pass/fail level, rather than an exact amount (which doesn't matter so much). As a result, how reliable and precise are the levels recorded?
  • Do biases get introduced in the donation filtering process? For example, suppose women donors or people with certain clothing indicating religious beliefs, tend to have the measuring device placed slightly differently to avoid touching their religious clothing (or to go over it), or actual contact if that's taboo. Or women are seen as more fragile and borderline results more likely to be rounded down than up. Or Sikhs/Muslims seen as more suspicious, or whites/the poor (stereotype promiscuous/more likely infected) as more likely to have undetected HIV, and such bias tends to subtly impact how these donors are talked to/assessed/deterred. Or the questions asked are problematic for some groups. Or maybe some policy/criterion change had the effect of excluding some poor health people in the first group, but this had been removed prior to covid, so the excluded poor health people were prone to return in the 2nd group. Perhaps staff had more time to investigate borderline donors' health in a bit of depth before covid, so that borderline-pass cases were more often rejected after review, raising the average in the first group.
  • As a variant on this, what if a more wary/cautious mindset existed in the medical services overall (including donor centre staff) due to the pandemic, and all measurements in 2020 were more likely to be rounded down than up, compared to normal where the staff are less looking for problems?
  • Was the centre materially the same? Are we sure? People typically wait a while before a nurse measures them. Plenty of time that if the centre itself has some change, it could have had an effect. Suppose the AC is off or windows closed in the 2nd group to reduce spread, or patients are asked to wait in a more stuffy area, or the hospital waiting area was cleaned every hour by some cleaning product that lingered in the air and in high ongoing uses could affect a subset of patients breathing. Suppose by the time of the 2nd group, they called patients upstairs to a different room, and rushed their stats gathering (hospital pressures under covid) so patients were more likely to have run upstairs and been in slight oxygen debt. Suppose they used less experienced staff for taking patient stats, because of covid staff pressures, or they had budget pressures or a new supplier and their measuring devices were less well-calibrated or had systemic differences between the groups.
  • Could time of year, pollen/asthma, air humidity (wetter or drier air), or other seasonal changes have had any effect? People may also pant if hot and the windows are closed, and CO2 could be higher, oxygen absorption may differ with moisture in some ways, or people may experience other changes to metabolism/physiology related to the time of year/weather.
  • Did they match records of those donors who were in both groups, and analyse statistically the variation between Hb levels for individuals who appeared in both groups, or only compare the overall distribution of results at a group level?
  • Did they compare also, say, a third or fourth group, taken at the same time of year as the 2020 (covid era) group but from say 2017 or 2018, to check if two or three groups before the covid era group, were comparable? In which case that would help emphasise if there was something truly different with the 2020 (covid) group. They didn't, and this seems a crucial control, to me.
  • Was the profile of blood donors affected by other events (causation not correlation)? For example, we might speculate that (1) a lot of people had low-grade health conditions due to say, asymptomatic covid, (2) vaccine reactions, (3) changes to daily life affecting stress levels hence low-level changes to usual heart rate, breathing or other autonomous functions, for many people, (4) changes to diet and exercise, (5) reduction in health and respiratory/heart/blood system efficiency due to lockdowns or lack of work, (6) changes resulting from changing behaviours/commerce/travel, affecting polluters such as vehicles and industry, affecting the air in Delhi which is hugely polluted, or (7) changing how people breathed, if your work and life change due to lockdown then you might breathe deeper or lighter, (8) or be more/less exposed to the air if you have to change between walking/bus/tuktuk.......
  • Also we can imagine what kind of people are more or less likely to donate blood during a lockdown or pandemic. Blood donation is unpaid in India, but we might speculate that people in less than excellent health or economic circumstances (which correlates strongly with poorer health and other conditions) are disproportionately deterred from donating normally, because there is not so huge pressure or reward, they work longer hours, or are less able to take time and travel to do so, and during a pandemic, they may have more time and opportunity, see news that hospitals need support, and see it as their duty to help others since they can now do so and it's urgently needed. So perhaps the health profile of those giving, will become poorer anyway. Or maybe it's the other way around. We can speculate that more affluent, economically secure, healthy, and educated, were more prone to stay at home and * not * go out for non-essential matters during the pandemic, because they were used to avoiding risks and watching their health, whereas poorer and less secure donors (on average having poorer health) were more likely to be fatalistic and carried on as normal.

In addition, there are other relevant studies to guide us, already. Did they use those or refer to them?

  • Some countries like Japan have long social histories of mask-wearing before covid. Are there studies of such groups? Do Japanese long term maskers, vs. new (2020) maskers vs non-maskers (pre-covid) vs. non-maskers (2020) show a consistent difference in oxygen levels?
  • Masks are also worn in many other contexts. Are there studies showing changes to blood Hb in Japan or elsewhere, when a mask is worn for a long time, or does a person's average mask-wearing times changes? For example, many jobs needed masks even before covid - forensic and lab technicians, "clean room" computer and technology workers, people working with hazardous chemicals or cement dust (construction).....
  • And in any case is it medically important or not, that's another issue. Are there studies showing actual effects? Like assessment tests of mental performance/ability, or long term vs short term effect, or effect of a few hours a day vs continual vs none, or recovery curves if any needed?

And so on.

The point is, without better comparisons and controls, we don't know what happened, literally, all we know is oxygen, as marked by Hb, is down among donors, but why? Could be any of these, or something else, or indeed, yes, could be masks.

Fortunately, other credible studies do exist. This one for example studies the impact of mask-wearing for doctors during surgery. This one is in progress for exercisers. This one studies the effects of N95 masks during pregnancy.

The overall conclusion of these seems to be that there is a moderate reduction in blood oxygen, and a slight elevation of demands on the body as a result. But not really enough to be a danger to health. More a level to be aware of and manage how hard you push yourself and your body while wearing a mask.

That's paraphrased. You can read the originals, or google for others - I stopped at 3.

13
  • 4
    That's probably a different question. Sadly science data and science communication are 2 different things, nuance and detail matter, and good communication is hard....?
    – Stilez
    Aug 8 at 20:30
  • 1
    A confounding factor you might have missed; Can resporatory infections cause hypoxia?
    – Taemyr
    Aug 10 at 0:00
  • 2
    Kinda covered under "low grade health conditions due to say, asymptomatic covid". If it was obvious covid they presumably would be bkocked from donating, if its any other respiratory infection I suspect we would probably expect to see fairly similar rates independent of covid, in both groups. But that could be wrong, so yes, maybe that too.
    – Stilez
    Aug 10 at 0:03
  • 1
    "A good paper will try to anticipate other factors ("confounds") that might have muddied the results, work around those it can, and at least highlight or try to estimate the effect of those it can't, so others are aware, and can either allow for it, form their own views, or design further experiments to address them." - Did you read sections 3.3, 5.1, 5.2? It seems like you're providing broad reasons why any one study isn't sufficient evidence of something, without actually looking at this one - I agree whole heartedly on the question of medical importance of the results though.
    – TCooper
    Aug 12 at 23:44
  • 2
    @Stilez I read your answer, and I read the entire paper before I commented. There's no point in listing "possible confounds" that are clearly not possible as they were controlled for in the study, as outlined in the paper. You let your imagination get ahead of your critical thinking in writing this answer.
    – TCooper
    Aug 13 at 15:31
24

The study gets off on the wrong foot with the unqualified and unsubstantiated claim that Covid-19 is the world's deadliest pandemic. There is some discussion about that in the previous question Did COVID-19 take the lives of 3,100 Americans in a single day, making it the third deadliest day in American history?

The study appears to make a correlation between prolongued mask use and hypoxia but has the rather weak

Conclusion

This study including 19504 blood donors spanning over one and a half year shows that prolonged use of face mask by blood donors may lead to intermittent hypoxia and consequent increase in hemoglobin mass.

The findings are further weakened by

5.2. Study limitations

Being retrospective and observational in nature, a potential limitation of the present study is that a complete blood count, arterial oxygen saturation and serum EPO level were not performed, which would help elicit a rise in EPO levels or erythrocytosis in these donors. Correlation with pre-COVID-19 Hb of donors with current Hb following exposure to intermittent hypoxia and duration for which the donors wore face masks could also not be elicited. However majority of the donors were from the working category which would involve moving outdoors and therefore use of face masks for longer periods.

Nonetheless, the present findings represent an important first step to future studies to investigate the effect of prolonged use of face masks causing intermittent hypoxia in blood donors and its possible consequences on the Hb mass of blood donors.

My bolding

Notice that the author seems to dissmiss their own objections with the assumptive

However majority of the donors were from the working category which would involve moving outdoors and therefore use of face masks for longer periods.

So the study seems to show a correlation, based only on previously filed statistics, and suggests that further study is needed, effectively admitting that the case is not proven.

2
  • I had to downvote this because the "it's not good because further study is needed" is the same kind of thing that antivaxxers are using. "It's not FDA approved, only EUAed, so it's no good! Further tests are needed!"
    – user253751
    Aug 10 at 9:44
  • 9
    @user253751 I am not arguing one way or the other. The question asks if the study proves what the question title says it does, and it does not, by its own admissions. That is what I am pointing out. Aug 10 at 9:56
8

I'm not a hematologist, but I do ponder whether the statistically significant difference observed between those two groups, of 0.52 g/dl of hemoglobin (Hb) is of actual/much clinical relevance, let alone an indicator of "intermittent hypoxia" and (solely) attributable to the "mask mandate" as Setia et al. conclude. Their observed difference is that

Mean Hb of blood donors in Group 2 (15.01 ± 1.1 g/dl) was higher than Group1 (14.49 ± 1.15 g/dl), (p < 0.0001).

For some measure of comparison, the difference in Hb (hemoglobin) is approximately an order of magnitude higher between high- (≥ 4500 meters) and "low"-altitude living (∼ 850 meters) Tibetans:

Mean hemoglobin concentrations in males from high and low altitudes were 20.01 ± 1.91 and 13.42 ± 1.75 g/dL, respectively. In females, these were 16.97 ± 1.74 and 12.56 ± 1.19 g/dL, respectively. [...] The hemoglobin concentration observed in our study for the low altitude Tibetans was on a par with another study with Tibetans at sea level.

Furthermore, in another study, 10g/l of Hb concentration was "equivalent" with a 3.5% BMI increase.

The increase of 10 g/liter in Hb corresponded to an increase of 3.5% in BMI, i.e., in an individual with a BMI of 25.0 kg/m2, this corresponds to a BMI increase of 25.9 kg/m2 [...] Hb levels were also positively associated with waist circumference, hip circumference, waist-hip ratio, body fat percentage, and visceral fat area, and negatively associated with body muscle mass ratio [...]. Thus, the subjects with lower Hb levels appeared to be less abdominally obese and to have more muscle tissue.

10 g/liter = 1g/dl, so 0.5g/dl from the Setia study correspond to a BMI increase of 1.75%. Which brings me to my final question mark: whether attributing the 0.52g/dl change in Hb to the "mask mandate" is reasonable given that other pandemic-related factors, like less physical exercise (due to "lockdown") leading to weight/fat gain might explain it as well. I'm not seeing much in the way of discussion on this angle in the Setia paper. Although they did not seem to have measured even the body weight of participants, there's one characteristic that perhaps stands out. "Group 2" had substantially more people in government service than "Group 1" (29.1% vs 17.8%) and substantially fewer "Semi-professional /self employed" (19.6% vs 31.1%).

Government service [n=]1935 17.8[%] [n=]2512 29.1[%]
Semi-professional /self employed [n=]3382 31.1[%] [n=]1695 19.6[%]

I hate to bring any cliches here (like "fat gov't bureaucrats"), but the type of job could possibly be related to the observed difference as well. In general, the BMI of people, even in India, is definitely job-correlated, but the way the employment is broken down in the Setia paper isn't too helpful in coming up with any group-level estimates, as their categorization isn't along the more clear lines of blue-collar/white-collar job used in BMI studies.

3

Looking at what other factors that cause hemoglobin concentration, I looked at smoking in particular to see how much that causes these levels to change. In doing so, I uncovered 2 major problems with the assumptions of the article in the Question.

That article states that "mask wearers" have a Hb count of "15.01 ± 1.1 g/dl" (150.1 g/L) and their "control group" has Hb levels of "14.49 ± 1.15 g/dl" (144.9 g/L). In one article I found it states that non-smoking women have an average level of 133 ±0.5 g/L and non-smoking men are at 152 ±0.5 g/L. This shows that more men donating blood during Covid could alone have raised the average levels of Hb, and the fact that smokers tend to have a higher count by around 3 g/L (smoking men having an average of 156 ±0.4 g/L) could also account for a much higher average in the "mask wearer" sampling.

Drinking alcohol also increases hemoglobin. This next article confirms that smoking increases hemoglobin and independently looks at non-smokers that drink doing the same. So drinking heavily will be a third reason why the hemoglobin average could have increased last year.

Heavy smokers (more than ten cigarettes/day) had significantly higher mean haemoglobin (1.4% higher in men, on average 3.5% higher in women) than non-smokers (p < 0.01). Smokers demonstrated a significant correlation between cigarettes/day and drinks/week in men (r = 0.24, p < 0.001) and women (r = 0.16, p < 0.001). Non-smokers displayed a significant positive correlation between drinks/week and haemoglobin concentrations in men (r = 0.14, p = 0.001) and women (r = 0.08, p = 0.05). In non-smokers, alcohol consumption >14 drinks/week and more than seven drinks/week for men and women, respectively, increased mean haemoglobin by 1.3% in men and by average 1.9% in women compared with those consuming < or =14 and less than or equal to seven drinks/week. Smokers displayed similar results.

https://pubmed.ncbi.nlm.nih.gov/19039534/

In fact, something as simple as dehydration can cause an increased hemoglobin concentration.

Trying to pin higher levels of hemoglobin on mask wearing without considering other options is just another example of confirmation bias. The article in the Question was written based on research that set out to prove that masks caused hypoxia and since that's the only thing they considered in their data analysis (as stated in their "Aim"), that's what they found.

Confirmation bias is the tendency to search for, interpret, favor, and recall information in a way that confirms or supports one's prior beliefs or values.[1] People display this bias when they select information that supports their views, ignoring contrary information, or when they interpret ambiguous evidence as supporting their existing attitudes. The effect is strongest for desired outcomes, for emotionally charged issues, and for deeply entrenched beliefs. Confirmation bias cannot be eliminated entirely, but it can be managed, for example, by education and training in critical thinking skills.

https://en.wikipedia.org/wiki/Confirmation_bias

2
  • I don't think you read the study before trying to disprove it. Section 3.3 lists inclusions and exclusions which clearly states they excluded people affected by the factors you mention here, and more.
    – TCooper
    Aug 12 at 23:33
  • @TCooper: Setia et al. excluded "chronic" smokers and people living at high altitude from their study. Alcohol consumption is not mentioned though, but from the data given above one of the groups would have to be entirely composed of heavy drinkers (and the other have none) to explain the data on that factor alone, which seems a stretch.
    – Fizz
    Aug 26 at 9:42

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .