Face Masks and COVID – A Failed Technology

This post is co-authored with Dr Colin Axon, senior lecturer in mechanical engineering, Brunel University London, UK.

Whenever new evidence is produced demonstrating the ineffectiveness of masks, whether cloth, surgical, or N95/FP2, in preventing community transmission of COVID and other respiratory viruses, a commentator can be guaranteed to claim that different standards of evaluation should be applied to laboratory and real-world studies. A section of the biomedical and public health community has latched on to this claim, that evidence from highly-controlled ‘perfect-world’ laboratory experiments shows that masks really do work. The challenge, as they see it, is to make the real-world conform to the experimental conditions by sufficient training and enforcement to achieve compliance. This is a standard, if rather old-fashioned, response to health and safety issues.

What is involved in the typical experiment? Early on, in 2020, many researchers cobbled together ‘quick and dirty’ experiments with masks (or their components) clamped flat into sealed units so air could only flow through the mask or material. All manner of proxies or substitutes for virus particles were used under a very wide range of air-flow conditions, which were not necessarily representative of the range of what people exhale. Often the size of the substitutes was too large or too uniform or the flow rates too low, and frequently poorly reported. Yet the claimed filtering efficiencies – often as high as 30 percent or more – gained traction and became accepted figures. Later, more sophisticated, experiments used standardized, life-size, models of a human head. The model’s nose and mouth are coupled to a tube that blows air, to which particles have been added to simulate the virus. These particles usually have properties that make them easy to photograph, often by their size or by flaring under ultra-violet light. The simulated breath goes into a chamber, where the distribution of particles can be visualized. The air flow is then interrupted by a mask of the chosen material placed over the model’s nose and mouth.

All experiments involve some degree of simplification of the real world. From this description, however, we can see that the experimenter has considerable scope to vary the conditions to achieve a preferred outcome. This is very like what happens with observational studies in epidemiology, where the problems of cherry-picking comparisons in time, location and socio-cultural context are well-recognized.

Many variables can, then, be adjusted. Human faces are unique, flexible and dynamic, not uniform and rigid. Under laboratory conditions, masks can be fitted perfectly, and remain so, which may mean taping their edges so that air is forced through them rather than escaping at the sides. The experiments test the materials, not how people use them. The volume and velocity of exhalation of real people is not standard; reproducing the exact profile, and variation, of how humans breathe in and out is tricky. Comparative experiments show that these manikins are poor at reproducing human breathing patterns. The particles exhaled by humans vary greatly in size and density. To get clear images, the air in the chamber receiving the simulated breaths must be perfectly still so that the only air flow is from the experiment. Whilst such simplifications are justified in trying to understand the underlying science in a systematic way, they do not – and are not designed to – represent the real-world.

What humans exhale ranges from atoms and molecules to small droplets of spittle and may include microscopic pieces of nose and throat lining. Rather than being exhaled as singular particles, it is usually assumed that a virus is either hitching a ride on a water droplet or a tiny piece of biological material, or as a cluster of viral particles, or some combination of these. The amount of these objects at each size – the distribution – varies from breath to breath, and person to person It turns out that the physics of small particles (called aerosols) – larger than atoms and small molecules, but smaller than droplets – makes them particularly hard to count, measure, and manipulate. As breath leaves the human body, it always encounters air currents that rapidly disperse it.

Masks must be porous for wearers to be able to breathe. The question, then, is what size of object can pass through? For cloth masks the spaces between the woven fibers, and for surgical masks the perforations in the active layer, are large enough to pass virus- and aerosol-sized particles in both directions, as well as leaking around the edges. Although N95/FP2 masks are less porous they still pass smaller aerosols and particles, which, it seems, is not sufficient to block transmission. In part, this is why health and safety agencies do not recommend them for workplaces or laboratories handling viruses. If workers are dealing with hazardous pathogens, then either the workplace should be designed to isolate them from each other or the workers should be using much higher-grade respirators with a proper fit-test procedure in place.

In the laboratory it is also difficult to deal with the fact that viruses start to dry out as soon as they leave the body. Human breath is relatively warm and humid compared with its, indoor or outdoor, surroundings. As warm breath rises, it disperses and the virus dries, rapidly deactivating. Ironically, catching droplets on a damp mask surface may extend the viability of the virus. Virus particles may be detectable in the air with PCR tests but those tests do not establish that the virus is actually infective. Sadly, there are still many who assume that surgical masks are used to protect patients from viruses carried by the surgical team, instead of protecting the team against splashes of bodily fluids from the patient.

It is not inconsistent to accept both that airborne transmission is important for the SARS-Cov-2 virus and that face masks, of any kind, are an ineffective intervention. Successful laboratory studies do not tell us much about the world outside. Feeding laboratory results into equally simplified computational models does not improve their validity, even when the models are used to generate pretty videos. A model is only as good as its underlying simplifying assumptions and data – and in this case those data are at best fragile.

Of course, physicists and engineers are concerned about the usability of their knowledge in the real world. However, it is worth remembering, as noted by Henry Petroski, that failures are as important as successes in advancing these disciplines. It is time to accept that face masks are a failed technology for preventing community transmission.