Un-Masking Bias


Ask anyone about the wearing of masks, and most people will tell you that they protect us from the transmission of viruses. So strong is that opinion that the mere suggestion that masks may not be beneficial is likely to result in the perpetrator being locked in the public stocks and pelted with rotten fruit. The suggestion masks might not be beneficial appears to be met with the same ridicule and disdain as a Flat-Earther, perhaps more so.

In fact a recent poll I placed on a popular social media platform I asked the following question to a group of mainly healthcare professionals…

“On the subject of wearing face masks to prevent viral transmission – what do you believe?”

230 people believed “Masks Definitely Reduce Viral Transmission”, while only 18 answered “There is No Clinically Significant Evidence for Wearing Masks”

I also asked the same cohort about their concerns on the benefit/harm ratio of mask wearing, particularly in the young who have a long lifetime ahead to develop any complications. What was very interesting was the nature of the response. The majority of comments were scorning and patronising from two arguments, the first that masks were clearly beneficial, and the second that the idea they could be harmful was preposterous.

It would seem then, that the overall opinion of people in the public and the healthcare profession is; ‘not only are masks definitely beneficial but to consider otherwise is ridiculous’.

What was most striking how the definitive responses were, there seemed to be no middle ground. To many of my medical colleagues, masks are clearly beneficial and have no harms. Full Stop!

Bi·as (bī’ăs)

Epidemiology: Deviation of results or inferences from the truth, or processes leading to such systematic deviation; any trend in the collection, analysis, interpretation, publication, or review of data that can lead to conclusions that are systematically incorrect – McGraw-Hill Concise Dictionary of Modern Medicine. © 2002 


As someone who is a strong advocate of evidence based medicine I always worry when people are blind to potential harm. I cannot think of one intervention where there are no harms, even if those that exist are small.

We could take a lesson from Dr Alexander Spock, who in the 1958 version of his book ‘Baby & Childcare’ gave advice for mum’s to place their babies in the prone position (face down) when sleeping so they would not choke if they were to vomit. This sounded, on the face of it, very sensible advice. The arguments made seemed very compelling.

However, research done in 1989 by Peter Fleming found an association between babies sleeping prone and cot death (Sudden Infant Death Syndrome – SIDS). In 1991 the ‘Back to Sleep’ campaign advised parents to ignore Dr Spock’s advice, and to put babies to sleep in the supine (face up) position

In 1989 there were 1,500 cot deaths per year in the UK. Now there are less than 300. This is a valuable warning that sometimes the harms of interventions are not always immediately obvious, yet can also be significant. Unfortunately all medical interventions carry risk of harm, even wearing PPE.

Known Harm

Putting aside any potential medical harms that mask wearing ‘could’ cause, there are several harms that are already obvious. We know that masks cost money, and the provision of vast quantities of masks is sure to be consuming a lot of money from the healthcare budget. There’s also the cost to the environment, we now need to dispose of billions of masks and not all are making it to correct disposal. Masks are starting to appear in our streets, fields, and waterways. Masks are considered clinical waste in hospitals and clinical waste is expensive business. There is also the harm to culture, the harm to communication, and the harm to security.

Unknown Harm

Then of course there are the unknown harms of the chronic exposure on a much larger cohort of people than previously. The common retort here is that surgical and anaesthetic staff have worn masks for years, and this is very true however the risk profile is different. Surgical staff wear masks for short periods and change them frequently. Their whole day was not in theatre, and masks were generally clean when disposed of. I know that after several hours of wearing my mask my five o’clock shadow starts to pull fibres from the mask surface making it fluffy.

The worry about potential harms is that they are unknown, as is their magnitude. This in itself is not a reason to not wear masks (the appeal to fear fallacy) but all harms, both known and unknown must be considered when trying to balance risks against benefit.

Benefit/Risk Balance

So we are aware there are some known harms, and also some potential unknown harms. These should easily pale into insignificance in the light of strong benefit from masks right? After all, we don’t care about the cost of a few billion masks if they are saving lives. This is the risk/benefit balance, and is the crux of all evidence based medicine. The benefit of an intervention must also outweigh its harm to be worthwhile.

So is there clear evidence that masks are beneficial?
Contrary to mass opinion, no there isn’t!

But wait a minute! There are trials that show benefit to masks, and there have been several reviews, including the latest published this month in the BMJ, which suggests that mask wearing is beneficial!

What the evidence shows.

At the end of the influenza pandemic in 2010 six published randomised controlled trials (RCTs) focused on the benefits of different mask types  The results of these trials with respect to face mask for influenza-like illness (ILI) in healthcare workers, family units, and student groups; reported poor compliance, and revealed the pressing need for future trials, although no significant harms were found.

Ten years later in 2020, only six further RCTs had been published; five in healthcare workers and one in pilgrims. The addition of these trials to the evidence showed that masks alone had no significant effect in interrupting the spread of ILI or influenza in the general population, nor in healthcare workers

Interestingly, one of these Randomised Control Trials showed infection rates were three times higher in Vietnamese hospital workers who wore cloth masks than those who wore no masks at all.

Whether this data, that was performed on a different pathogen, can be applied to the transmission of SARs-CoV-2 is very much a topic for debate, but never-the-less this data (and the debate) remains.

Not surprisingly all the systematic reviews that used this evidence base of trials came to the same conclusion; that masks were not beneficial, however 2 earlier systematic reviews using weaker evidence (observational studies) showed that masks were beneficial, though the authors recognised the need for more robust randomised control studies.

It is interesting that the weaker (prone to bias and confounding) reviews of the observational studies demonstrated a different result to the more robust (Gold Standard) RCTs performed later.

Studies Vs Trials

Those of you whom remember the evidence based pyramid will remember, in the unfiltered information randomised control trials (RCTs) are gold standard and are more robust than studies, so systematic reviews that review RCTs are more robust than those that review observational studies (which are what you’ll use during a current pandemic).

Studies are prone to bias and confounding. Confounding to those less familiar with evidence based medicine can crudely be described as attributing an effect to the wrong cause (which is linked). For example, attributing mask wearing to the reduction of a virus, when it’s actually attributable to another associated intervention or behaviour.

The Evidence-Based Medicine Pyramid! - Students 4 Best Evidence

In 2020 another review was published that attempted to select its evidence by searching for SARs-CoV-2 pandemic related sources. This study’s search was only able to find 3 RCTs so also looked at 10 comparative studies, 13 predictive models, and 9 laboratory experimental studies.

Whether predictive models should be used in evidence is another debate, but regardless the majority of the evidence reviewed were relatively low valued studies, and even though the authors stated the study supports the use of masks they also admitted…

“Robust randomised trials on face mask effectiveness are needed to inform evidence-based policies.”

So the water grows murkier. We’ve some more robust trials done outside of the SARs-CoV-2 pandemic setting which show no benefit to masks, and some weaker studies that may be more applicable to SARs-CoV-2 , which show benefit. Either way nothing definitive was found.

A recent review this month in the BMJ looked at a variety of interventions from 35 studies to reduce viral transmission including hand-washing. Of these 35 studies only 6 were used to justify the position that masks have benefit. Again, these were observational and cross sectional studies which are heavily prone to confounding, and are not as robust as the much needed RCTs. The authors themselves had this to say…

Additionally, most studies that assessed mask wearing were prone to important confounding bias, which might have altered the conclusions drawn from this review (ie, effect estimates might have been underestimated or overestimated or can be related to other measures that were in place at the time the studies were conducted). Thus, the extent of such limitations on the conclusions drawn remain unknown.

While all evidence has value, I feel a heavy amount of critical analysis is required to see the true value of the current evidence.

Professor Carl Heneghan is an epidemiologist, a senior investigator for the NIHR, director of the Centre for Evidence Based Medicine at Oxford University (home to the Cochrane Library), and former editor for the BMJ’s EBM. He had this to say about the current evidence in his article Masking Lack of Evidence with Politics

Many countries have gone onto mandate masks for the public in various settings. Several others  – Denmark, and Norway – generally do not.  Norway’s Institute for Public Health reported that if masks did work then any difference in infection rates would be small when infection rates are low: assuming 20% asymptomatics and a risk reduction of 40% for wearing masks, 200 000 people would need to wear one to prevent one new infection per week.

What do scientists do in the face of uncertainty on the value of global interventions? Usually, they seek an answer with adequately designed and swiftly implemented clinical studies as has been partly achieved with pharmaceuticals. We consider it is unwise to infer causation based on regional geographical observations as several proponents of masks have done. Spikes in cases can easily refute correlations, compliance with masks and other measures is often variable, and confounders cannot be accounted for in such observational research.   

The small number of trials and lateness in the pandemic cycle is unlikely to give us reasonably clear answers and guide decision-makers. This abandonment of the scientific modus operandi and lack of foresight has left the field wide open for the play of opinions, radical views and political influence.  

Professor Carl Heneghan. CEBM, Oxford University.

Summary of Evidence

It seems then that there are 12 RCTs that looked at Influenza like illness that suggest masks are not beneficial, and a handful of observational studies that started part way through the pandemic that suggest masks are beneficial.

However, we also know during the Influenza pandemic ‘observational studies’ similarly showed benefit when the more robust RCTs done later showed no benefit. Is this pattern repeating itself?

The strength of evidence currently points to masks not being beneficial, but the jury is still out – even for me. There’s a wealth of more work to be done (RCTs), and it is unlikely we’ll have a definitive answers for many years. In the meanwhile I feel it is unwise to rely on the current observational studies given what we already learned about them during the Influenza pandemic.

Even if the position reversed and the weight of evidence started to fall in favour of masks, it’s unlikely to show anything more than low benefit else I feel the evidence would be more significant now. With this is mind, I wonder if we have the risk balance quite right?

Cognitive Bias

Given the weight of evidence currently suggests no benefit to the wearing of masks, what about the harm/benefit balance? My belief is that the wearing of masks is grossly overvalued, but cognitive bias prevents us from being able to balance the risks safely, and this may be at a cost.

The sum of the evidence base on masks currently suggests masks aren’t beneficial, yet the opinion of most healthcare workers and the public is the complete opposite, and the resistance to change this opinion is very high. When justification of the pro-mask opinion is offered, it is generally done so by citing only the current and ongoing ‘observational studies’ with no reference to any of the ILI RCTs done over the last 2 decades. This is classic selection and confirmation bias. (Confirmation bias is the tendency to search for, interpret, favour, and recall information in a way that confirms or supports one’s prior beliefs or values)

Whether you feel mask wearing is a good thing or not you cannot dispute that this is a subject still very much open to debate and is not settled, however, this is a debate people appear resistant to enter into due to the significant cognitive bias over the issue.


It’s almost 2 years since the start of the COVID pandemic, and I’m sure most can remember that initial out-roar from the public demanding increased PPE for healthcare professionals. I can see how this posed a difficult political dilemma.

One of the difficulties for a democratic government is if the technically correct thing to do is not what the public want you’re in political strife. If you do the best thing for your people, you’re a dictator, if you do what they want you’re incompetent.

Media driven hysteria made it very difficult for any government to not do what the public demanded.

We started to wear masks in the NHS from the 5th June 2020 and on July 24th 2020 it became compulsory to do so in shops and supermarkets. This was before many of the observational studies which all started part way through the pandemic, and after knowing the results of the RCTs which showed no benefit to mask wearing. I wonder then what the real motivation was. Was it evidence based, was it political, or was it public pressure?

Cognitive bias is a huge problem in healthcare, but it would seem it also has profound consequences for politics and society.



Dr Shaun Favell

Dr Shaun Favell is a Consultant in Emergency Medicine, and an Honorary University Lecturer. He has interests in Mathematics and Physics, and has previously undertaken research in 'Quantum Probability Theory'. He also has interests in Evidence Based Medicine, Over-testing, Over-diagnosis, Over-treatment, and Cognitive Bias.

Leave a Reply