The use of face masks among the public has been a recommended mitigation measure during the COVID-19 pandemic [1]. A May 2022 study, using data from 92 regions on 6 continents, determined community mask wear reduced SARS-CoV-2 transmission [2]. Most studies estimating mask wear relied on a survey design [3–5], which is subject to response bias, including social desirability bias [6, 7]. The observational study design allows for direct, impartial, and rapid assessment of easily observed behaviors like mask wear in public areas, but few observational studies have measured adherence to mask wear recommendations in communities. Of 2 published observational studies, persons observed wearing masks ranged widely: 67.7% in Portland, Oregon, (during June–August 2020) and 41.2% in Wisconsin (May–June 2020) [8, 9].

Ada and Canyon Counties, Idaho’s 2 most populous counties, are similar in many measures, including percentages of population ≥65 years (14.9% Ada, 14.0% Canyon), female (49.9%, 50.4%), and White (91.9%, 93.2%) [10–12]. These neighboring counties are designated urban, with population densities one order of magnitude larger than the state average (Ada: 470 individuals per square mile, Canyon: 393, Idaho: 22) [10–12]. As of April 12, 2022, the percentage of the population that is fully vaccinated is lower in both Ada (53%) and Canyon (41%) counties than the percentage nationwide (65.8%) [13, 14]. These counties differ by overall political affiliation (43% of voters are registered with the Republican party in Ada County; 55.9% Canyon County) [15] and by percentage of residents aged >25 years with at least a bachelor’s degree (40.7% Ada County; 20.7% Canyon County) [11, 12]. Neither county had a mask mandate during our study, although Boise (Ada County’s largest city) previously had a mandate from August 2020 through May 2021 [16, 17].

Under the direction of Idaho public health leadership, we observed mask usage to assess contemporary mask wear during the COVID-19 surge driven by the Delta variant in Fall 2021. Both Ada and Canyon counties were experiencing high community transmission, with 7-day average of >50 cases/100,000 people during the entire study period [18]. During this period, CDC recommended everyone, regardless of vaccination status, wear a mask in indoor public settings in communities with high transmission [19]. As the number of patients with COVID-19 overwhelmed hospitals, Idaho public health encouraged residents to get vaccinated and wear a mask [20]. We conducted observations in Boise and Nampa (Canyon County’s largest city) as these are the 2 largest cities in each of Idaho’s 2 most populous counties, are geographically adjacent, and have similar demographic profiles by race, sex, and proportion of population ≥65 years of age. We observed persons exiting different stores and public venues to measure mask wear, including where, when, and how people wore masks. We conducted this investigation to guide public health response efforts during a COVID-19 surge and while crisis standards of care were enacted for hospitals (statewide: September 16–November 22; northern public health districts: September 6–December 20).

Of >3000 persons directly observed exiting 5 retail chain stores in Idaho’s 2 largest cities during November 8–December 5, 2021, only 22.0% of individuals wore a mask. Although both cities were experiencing a surge in COVID-19 cases and neither city had a mask mandate, percentage of mask wear among observed individuals was more than double at Boise store locations (31.3%) than Nampa locations (14.3%). The percentage of persons wearing masks in both cities during our study were lower than percentages observed in similar studies conducted in Portland, Oregon, and Toronto, Canada, (93.5% at locations with mandates; observed during June–August 2020) and Wisconsin (41.2% at locations without mandates; May–June 2020), with differences in adherence to recommended COVID-19 mitigation efforts likely indicating behavioral shifts during different pandemic phases [8, 9, 30]. Boise previously had a mask mandate in effect that ended in May 2021, and the possibility exists that lingering effects of the mandate influenced mask use. However, a study conducted at community locations in Tennessee (observations during February–April 2021) found no statistically significant difference in mask use between the period when a county-wide mandate was in effect compared to immediately after the mandate was lifted [31].

During the equivalent study conducted December 2020 through February 2021 when Boise’s mandate was in effect, >93% of Boise patrons were observed wearing masks at 5 stores, whereas the range of patrons wearing masks at Nampa stores was 49.6%–79.6% by store [16, 17, 21, 22]. In our study conducted during November–December 2021, neither city had a mask mandate and the percentage of persons wearing masks was lower in all stores in both cities, compared to the study conducted the previous winter (Figure 2). The largest decrease in mask wear was among persons observed at Store E in Boise (previous study: 96.9%; our study: 12.0%). On average, observed mask use decreased by 67.1 percentage points at Boise store locations and 54.9 points at Nampa locations. An observational study conducted outside grocery stores in Wisconsin in 2020 reported similar effects of mandates on face coverings: mask wear was initially 41% in mandate’s absence but increased to 93% with store mandates and then to 96% after a statewide mandate [32]. In Idaho, cloth masks were less common (Boise: 50.5%; Nampa: 46.2%) during November–December 2021 than during December 2020–February 2021 (Boise: 56.7%; Nampa: 54.7%) (Graff, unpublished data, 2021), possibly reflecting wider availability of surgical masks and the shift towards encouraging surgical masks over cloth mask.

Mask use is supported by CDC and World Health Organization and remains one of the most cost-effective COVID-19 mitigation measures available [33, 34]. During this study period, both Ada and Canyon counties were classified as high transmission areas, and federal and state public health recommended all persons wear masks indoors in high transmission areas. Without mandates, alternative measures to increase uptake to public health recommendations should be considered where possible and might have been helpful in both cities, and likely statewide, during this study. Personalization of masks has been shown to increase mask wear, as has distribution of free masks and frequent reminders by community leaders and event organizers [35, 36]. One study reported emphasizing how masks protect other community members led to an increase in mask wear among surveyed Americans, whereas no increase was reported in mask wear when emphasizing how masks protect the wearer [37]. In other studies, researchers found messages focused on empathy, especially toward community members most at risk for severe disease, were successful in increasing participants’ motivation to undertake COVID-19 preventive measures [38, 39].

Across all stores, mask wear was lower on weekend days, compared with weekdays in Nampa. Proportion of persons wearing masks was higher in certain locations during the morning. Persons who are older or have conditions that might put them at higher risk for severe COVID-19 outcomes might shop at times that are less crowded [40]. Earlier in the pandemic, many stores offered reserved shopping hours for persons at higher risk; no store locations were advertising reserved hours, but a residual effect is possible. Increased work flexibility resulting from pandemic policies might have enabled more persons to shop during traditionally less crowded hours [41]. Certain stores, such as farm supply Store C, have senior discount days or hours, which might also affect shopping behavior. Published studies have shown higher mask wear among vaccinated persons compared with unvaccinated persons [42], and among registered Democrats compared to registered Republicans [43, 44]. Higher education levels, female gender, and older age are all associated with mask use [45]. Vaccination rates and the percentage of voters registered as Democrat were higher in Ada County than Canyon County, the percentage of residents with at least a bachelor’s degree is higher in Ada than Canyon County, and the average of median age is slightly higher in Boise than Nampa (37.2 years vs. 34.5 years) [11, 12]. These factors might explain some of the difference in observed mask wear between the 2 cities.

Observational studies are a rapid, nonconfrontational way to evaluate use of mitigation measures among the public, but are limited to the locations and times assessed. We captured a sample of shoppers in the community by selecting a range of retail chain companies; demographic information was not ascertained and ability to assess representativeness was not possible. One limitation of our study is not knowing why people chose to wear a mask or not because we did not conduct interviews. Possible reasons for not wearing a mask include believing protection from a mask was not needed after COVID-19 vaccination, doubting mask effectiveness, forgetting a mask, or believing the pandemic to be exaggerated [46–48]. Observations varied by chain, with the number of individuals observed lowest at the gas station convenience store in both cities; only 5 individuals were observed wearing masks at the Nampa location, leading to a large CI and less precise RR estimate. The other store types had more shoppers, which allowed us to collect more data and provided more insight into mask wear in the community. Another limitation was observers might have been unable to distinguish between mask types or determine correct placement from a distance. Unclear observations were recorded as ‘unknown’, but misclassification is possible. To minimize effects of reduced visibility after sunset and to ensure observer safety, sessions ended by 6 pm; we did not observe those shopping from 6 pm–8 am, a period which may include shift workers.

Another limitation is mask-wearing behavior might have shifted during the month-long study. The Delta surge peaked on September 16, 2021, and reported cases were declining during the study period, which ended before Omicron was identified in Idaho [49]. Persons who wore masks during the peak might have stopped doing so as reported cases decreased, despite the counties continuing to experience high transmission. Decreasing community cases could lower an individual’s perceived risk for contracting COVID-19, therefore mask use might be less prevalent [50]. Pandemic fatigue can lead to complacency and a disregard to mitigation measures, including mask wear [51, 52].

Community-level data collected through observational studies can inform public health officials and guide future studies [53, 54]. Studies that observe public health-related behaviors at retail locations also might provide unique opportunities for future messaging strategies. For example, locations with a small percentage of persons observed wearing masks might be selected as locations for promoting alternative mitigation measures, such as pop-up vaccination or testing sites. This would be based on further studies that included collection of shopper demographics, values, and motivations. Research supports multipronged approaches for increasing community vaccination, and mobile sites at places persons are already visiting would decrease cost and physical barriers and increase vaccination opportunity [55]. None of the store locations had mask requirements in place at the time of our study, but efforts to work with stores to encourage mask wear or to host vaccination clinics might be beneficial, especially in areas with low vaccination rates [56]. Future work could build on observation studies such as ours to leverage marketing products such as demographic profiles created by retail chain companies, or the use of innovative marketing techniques, to identify store locations with customer demographics that might be at higher risk for severe outcomes and reach those customers with tailored messaging.

Despite both cities experiencing a COVID-19 surge, only 22.0% of >3000 persons were observed wearing a mask after exiting retail stores. Proportion of persons wearing masks was higher at all Boise locations, compared with Nampa locations, possibly indicating residual effects of Boise’s earlier mask mandate or sociodemographic differences such as age, education, or political affiliation between the 2 cities. Vaccination rates are low in both counties, but lower in Canyon County, perhaps indicating the need for broader promotion of mitigation efforts. Among people wearing masks, >94% wore masks correctly in each city. Variations in mask wear by weekend vs weekday and time of day indicate observational studies should include a variety of observational periods. Interviews with observed persons or recording of identifiers would be needed in future studies to learn more about shopping behaviors and mask wear.

Public health should consider benefits of developing an infrastructure that can be used to rapidly conduct similar observation studies statewide in future extended outbreaks or pandemics. Data collected might be helpful for allocating resources or focusing communications. Protocols for conducting observations could readily be adapted for other behaviors. A statewide collaboration with colleges and universities, such as training select faculty to train and deploy teams of student observers under the direction of public health, could be a viable option. Periodic refresher trainings could be conducted, in much the same way as preparedness exercises, so that observation studies could be quickly initiated when the need arises.