"Hell is other people"


The esteemed Cochrane Library published the results of their study of mask studies and their conclusion was unambiguous:

Wearing masks in the community probably makes little or no difference to the outcome of influenza‐like illness (ILI)/COVID‐19 like illness compared to not wearing masks (risk ratio (RR) 0.95, 95% confidence interval (CI) 0.84 to 1.09; 9 trials, 276,917 participants; moderate‐certainty evidence. Wearing masks in the community probably makes little or no difference to the outcome of laboratory‐confirmed influenza/SARS‐CoV‐2 compared to not wearing masks (RR 1.01, 95% CI 0.72 to 1.42; 6 trials, 13,919 participants; moderate‐certainty evidence). Harms were rarely measured and poorly reported (very low‐certainty evidence).

This touched off another fiery outburst of Mask v No Mask hostilities.

An exasperated Derek Thompson asks, “Why are we still arguing over masks?”

I don’t understand why we’re still arguing, either. A reasonable person should be displaying humility in the face of so much ambiguity. Covid outcomes were pretty similar across different locations with wildly different covid public health policies. Taking into account age stratification, population densities and the myriad other factors at play only makes analysis murkier.

The Bangladesh Study

Why do so many mask advocates, stung by the Cochrane Study, shout and point to the Bangladesh study? Because it is one of the few outlier studies which demonstrates an (albeit statistically slight) advantage to community masking.

Here is the famous Bangladesh Study which concludes community education and increased masking decreased incidence of covid:

Impact of community masking on COVID-19: A cluster-randomized trial in Bangladesh

The intervention increased proper mask-wearing from 13.3% in control villages (N = 806,547 observations) to 42.3% in treatment villages (N = 797,715 observations) (adjusted percentage point difference = 0.29; 95% confidence interval = [0.26, 0.31]). 

In villages randomized to surgical masks (N = 200), the relative reduction was 11.1% overall (adjusted prevalence ratio = 0.89 [0.78, 1.00]). The effect of the intervention is most concentrated among the elderly population; in surgical mask villages, we observe a 35.3% reduction in symptomatic seroprevalence among individuals ≥60 years old (adjusted prevalence ratio = 0.65 [0.45, 0.85]).

Criticism of the Bangladesh Study

Here is a criticism of the study from conservative Heartland Institute, which attempts to completely discredit the study:

Famed Bangladesh Mask Study Excluded Critical Data

The main criticism is that the authors of the study omitted mortality data, and a squabble over whether the results are “statistically significant.” It is worth noting that if the results were mathematically overwhelming, there would be no argument over statistical significance.

For a friendly treatment of statistical significance, Tom Chivers offers a helpful article on understanding p value and its role in scientific papers: The Right Way to Look at the Bangladesh Mask Study. He also refers us to the related article, The Scent of Bad Psychology.

Another study: The Bangladesh study was flawed…

… and this study brings the numbers and analysis:

Re-analysis on the statistical sampling biases of a mask promotion trial in Bangladesh: a statistical replication

Here is the damning charge, which the study then backs up with data:

Upon reanalysis, we find a large, statistically significant imbalance in the size of the treatment and control arms evincing substantial post-randomization ascertainment bias by unblinded staff.

In layman’s terms, they found evidence the researchers by their methodology, may have tipped the scales in favor of their thesis that masks work. (See: Blinding in Randomized Controlled Trials)

Defense of the Bangladesh Study

The Bangladesh study authors respond by criticizing the studies included in the Cochrane analysis:

Abaluck and Jimenez said, these studies don’t really ask the question Do masks work? Instead, they ask: When you hand out masks and information to an intervention group without much enforcement, does it make them healthier?

Fair enough, and that is a distinction lost on the various warring factions.

If you can understand the following statement, you are a long way on the path of humble understanding:

Masks may be effective at reducing the spread of COVID-19 at the individual level, but their effectiveness at the community level depends on several factors, many of which are beyond human control.

I agree with Derek Thompson’s conclusion:

My advice in navigating this mess is: Do not trust people who, in their handling of complex questions with imperfect data, manufacture simplistic answers with perfect confidence. Instead, trust people who allow for complexity and uncertainty.

The smartest and most educated among us can fall into epistemic arrogance (aka the Fauci Factor). Caveat emptor. Following others off a cliff makes you look unserious.

Unrelated Note: No more weekend posts here. I’m trying to post something big and sticky on Fridays so people can argue over it all weekend. See you Monday!

%d bloggers like this: