With the rapid emergence of COVID-19, public health officials, companies, and individuals
in the public have scrambled to find solutions to prevent and treating this fast-spreading
virus, but because of the novelty of the virus, few data are available on the effectiveness
of possible solutions. Consequently, we have sometimes had to rely on studies done
on the prevention and treatment of other upper respiratory infections, such as influenza.
One popular practice for reducing the transmission of COVID-19, which has been widely
recommended, is for people to wear face masks in public places and other venues where
they will be in close proximity to other people. The equipoise about the effectiveness
of face masks in these circumstances is in contrast to the well-established role of
face masks to prevent the spread of viruses in hospital settings, especially with
the use of N-95 face masks. MacIntyre and Chughtai (MacIntyre and Chughtai, 2020)
present their findings of a systematic review of randomized controlled trials assessing
the effect of face masks in the community to prevent the transmission of influence
and other influenza-like illnesses. They concluded that face masks could be effective
at preventing COVID-19 transmission, based on the results of eight randomized studies.
However, upon closer inspection, their conclusion might be a bit optimistic. Here,
we review issues of concern regarding the information they used to support their conclusion.
These issues of concern fall primarily into several categories: favoring the results
of a (possibly biased or spurious) sub-analysis over the null results of the primary/overall
randomized analysis; relying on results of multiple testing with the possibility of
experiment-wise error; and failure to separate the effects of hand hygiene from the
effects of face masks.
In the MacIntyre et al., (2009) study, there was no beneficial effect of wearing face
masks in the intention-to-treat population. The review reports that in the subgroup
of people who were adherent, face masks were beneficial, but adherence was low (<30%
by day five), resulting in a biased sample. Any difference seen may be because people
who wore the face masks were notably different in other ways than those who did not
wear them.
The two studies by Cowling et al., (2008/2009) (Cowling et al., 2009, Cowling et al.,
2008) are actually preliminary and final results of the same study, which did not
find any difference between masks plus hand hygiene and any of the other groups (hand
hygiene or control group) in the primary analysis. They did do a sub-analysis, and
they did find that both the hand hygiene group and the masks plus hand hygiene had
less influenza (confirmed by PCR) than doing nothing (no difference between the two
intervention groups), but when considering clinical symptoms, there was more influenza-like
illness in the face mask group than in the hand hygiene only group. This suggests
that face masks could actually increase virus transmission.
In another study by Aiello et al., (2010) (Aiello et al., 2010) there was no difference
in rates of influenza-like illness between the face mask only group and the control
group. However, there was a trend towards lower influenza-like illness with the addition
of handwashing (in addition to masks), suggesting that it is the handwashing and not
the masks that are lowering influenza rates.
In the study by Aiello et al., (2012) (Aiello et al., 2012) the use of face masks
plus hand hygiene was no better at reducing rates of influenza-like illness than hand
hygiene alone in the cumulative analysis. If looking at individual weeks, only the
face mask plus hand hygiene group had significantly lower rates of illness in the
later weeks, not the face mask only group, suggesting that it is the handwashing and
not the masks that are lowering influenza rates.
Larson et al., (2010) (Larson et al., 2010) did not find a significant benefit among
the face mask and hand hygiene groups (there was no face mask only group) in incident
rates of influenza. They did find a benefit in reducing the secondary attack rate,
but they also reported that the hand hygiene group (not the face mask group) had fewer
people in the household with influenza symptoms. Simmerman et al., (2011) (Simmerman
et al., 2011) likewise, did not find any reduction in influenza from the use of face
masks or handwashing. In fact, they found a higher rate of influenza-like illness
among both the handwashing group and the handwashing group plus face masks than in
the control group.
Sues et al., (2012) (Suess et al., 2012) did not find any differences in incident
influenza or secondary attack rates in the overall analysis. They did do many sub-analyses,
but caution should be practiced when interpreting the results, considering the experiment-wise
error, which can lead to spurious results when multiplicity of testing is being done.
The Sues et al., study is a good example of this when some of the sub-analyses indicate
that face masks are beneficial in reducing incident influenza, but in other sub-analyses,
it is the hand hygiene and not face masks that appear to reduce incident influenza.
Of the seven separate randomized studies MacIntyre and Chughtai include in their review,
none of them show a benefit of face mask use in the community to prevent the incidence
of respiratory viruses, and the results of the sub-analyses are equivocal, sometimes
suggesting that masks prevent virus transmission but other times increase transmission.
Based upon these studies, there is little evidence to show that face masks prevent
influenza and influenza-like illnesses. It may be that face masks for COVID-19 will
have different outcomes, but we cannot know for sure until an adequately powered,
well-conducted randomized trial is performed, one of which is set to be completed
in July 2020 (https://clinicaltrials.gov/ct2/show/NCT04337541). In evaluating whether
face masks work in the community in reducing respiratory viruses, there is biologic
plausibility, which has led to wide-spread popularity. However, this enthusiasm does
not justify incorrect/biased summaries of published articles.