Skip to main content

Imprecise assessment of mask use may obscure associations with SARS-CoV-2 positivity

Our take —

In an online survey of 3,058 adults in three US states, general self-reported mask use consistency was not associated with SARS-CoV-2 PCR test positivity in the past two weeks; however, more frequent mask removal and higher activity participation were both associated with recent COVID-19 infection. These findings are subject to caution since respondents to this internet-based survey may systematically differ from non-respondents, and respondents who tested positive may have been more likely to recall social activities and instances of mask removal. Nonetheless, the study suggests that more nuanced measures of mask use behavior are needed to accurately capture risks of SARS-CoV-2 exposure in population-based surveys. The senior author of this scientific article is also a member of the Hopkins NCRC; therefore, this expert review was written by an NCRC member without current or past collaborations with Wesolowski.

Study design


Study population and setting

Between September and October 2020, investigators administered an online survey to 3,058 adults in three US states (Florida, Maryland, and Illinois). Respondents were asked whether they wore a mask (always, sometimes, or never) when indoors, outdoors, and within 6 feet of others. They were also asked whether they had participated in one of six types of activities (visiting with friends, relatives, or neighbors [indoor/outdoor]; going to a bar or restaurant [indoor/outdoor]; going to a gym or fitness class [indoor/outdoor]). Finally, respondents were asked if they always, sometimes, or never removed their masks while participating in these activities. The investigators created summary scores reflecting respondents’ frequency of activity participation and mask removal, respectively. The investigators then measured associations between self-reported SARS-CoV-2 positivity in the past two weeks and mask use measures.

Summary of Main Findings

Self-reported consistent mask use in both indoor (73%) and outdoor (73%) settings was high and did not vary considerably across states. Among these respondents, a majority (78%) reported taking their mask off when outside their homes, either in indoor or outdoor spaces. Nearly one-fourth (23%) of respondents who self-reported SARS-CoV-2 PCR testing in the past two weeks had a positive test. SARS-CoV-2 test positivity was significantly associated with participation in more activities outside the home (OR per additional activity: 2.03, 95% CI: 1.68–2.59) and more frequent mask removal when participating in indoor and outdoor activities (OR for a one-unit increase in mask removal score: 9.92, 95% CI: 1.16–85.1). When not accounting for frequency of mask removal or activity participation, consistency of mask use indoors and outdoors was not significantly associated with SARS-CoV-2 test positivity.

Study Strengths

The study used novel measures of mask use behavior, including removal of masks when participating in specific indoor and outdoor activities. Additionally, investigators restricted reporting periods to the two weeks prior to survey participation to mitigate possible recall biases and strengthen temporal inferences in the observed associations between mask use behaviors, activity participation, and recent COVID-19 infection.


Individuals may recall prior exposures differently depending on their SARS-CoV-2 infection status; for example, those testing positive within the past two weeks may be particularly apt to recall instances when they removed their masks. If this recall bias occurred systematically, the association between mask removal and SARS-CoV-2 test positivity would be overestimated. Reliance on self-reported SARS-CoV-2 test positivity is likely to result in underestimation of COVID-19 infection prevalence in the study population, as decisions to seek COVID-19 testing and access to testing are likely to be imbalanced among survey respondents across residence and symptomatology. Additionally, while measures of mask use behaviors used in the survey were nuanced, these measures did not include specific dimensions of mask use behavior (i.e., appropriate use, mask type, duration of mask removal) that are essential for exposure surrogacy measures. Internet-based recruitment and administration of the survey limits generalizability of results, as respondents are likely to be substantially different from those who chose not to participate or are excluded from participating. As with other cross-sectional online surveys, results from this study are limited by the simultaneous assessment of mask use and SARS-CoV-2 positivity (i.e., causation cannot be inferred in the mask behavior and SARS-CoV-2 positivity relationship). Responses may also be subject to social desirability bias (i.e., respondents may have been more likely to self-report behavior that they deemed to be desirable), which may bias measures of behavior and observed associations in analysis.

Value added

This study is among the first to find that more nuanced measures of mask use behaviors, specifically frequency of mask removal when participating in activities outside the household, explain SARS-CoV-2 test positivity more so than general self-reported adherence to mask use recommendations.

This review was posted on: 3 February 2021