Public Utilization of Face Masks in Brazil During COVID-19 Pandemic Outbreak: Temporal Trend Analysis

Background: The coronavirus disease 2019 (COVID-19) has recorded approximately 8.6 million conrmed cases and more than 450.000 deaths worldwide. As of today, Brazil remains the second most affected country, with more than 1 million conrmed cases and more than 50,000 related deaths. This study aimed to evaluate the temporal trend of the frequency of face mask use among Brazil’s general population. Method: A cross-sectional survey method was adopted in this study. Online survey regarding sociodemographic and protective personal measures against COVID-19 was used to collect data. Data were collected from April 17 to May 15, 2020. The frequency of face mask use was divided into ve categories: “never,” “rarely,” “sometimes,” “frequently,” and “always.” Joinpoint regression model was employed to analyze the daily percentage change (DPC) of each category. Ethics aspects were considered. Results: A total of 14,756 volunteers comprised the study sample. The “frequently” and “always” categories represented 71% of the sample, with the former being superior to the latter (39.4% vs. 31.6%, respectively). Temporal trend analysis showed two trend periods for all categories. The “never,” “rarely,” and “sometimes” categories decreased in the rst period and stabilized in the second period. Conversely, the “frequently” and “always” categories showed an upward trend. Like other categories, the “frequently” category had stabilized in the second period. However, the “always” category still showed an increasing trend in the second period. The association analysis results showed a decrease in the use of face masks in women, postgraduate people, those aged above 35 years, and those earning more than seven times the federal monthly income minimal wage. However, the use of masks increased among young people and those with elementary to graduate degrees. The “always” use of face masks increased in the south and decreased in the southeast of Brazil. Conclusion: The preponderance of the “frequently” category and the slow DPC in the second trend period of the “always” category suggest the necessity of further enforcement of the use of face masks in Brazil.

to health professionals and surgical or cloth masks to the general population [3]. The e cacy level of N95, medical, and homemade masks regarding human-to-human virus transmission prevention is 99.98%, 97.14%, and 95.15%, respectively [4].
Simulation models for the impact of the use of face masks in the pandemic in New York and Washington (US) have suggested that community-wide utilization of face masks could reduce the risks of COVID-19, including transmission, hospitalizations, and deaths [5]. Another study supported this claim by showing an extremely lower incidence of COVID-19 in an administrative region of Hong Kong that implemented community-wide utilization of face masks than the countries doing differently, such as Spain, France, Italy, and Germany [6]. In addition to personal protection against COVID-19, the utilization of face masks reminds the people of the severity of the disease and highlights the importance of social distancing [7].
Given the internal political con icts in Brazil and the insu cient evidence on face mask e ciency meanwhile, the public use of face masks started to implement on April 2, 2020, three weeks after the rst 100 cases [8][9].
However, the number of COVID-19 cases and deaths are still increasing even with this implementation.
The noti cation of new COVID-19 cases has increased rapidly up to 1,329.04% from April 2 to May 15 (1,071 and 15,305 cases, respectively). On June 29, 2020, a total of 24,052 new con rmed cases were reported to the Ministry of Health, corresponding to an increase of 2,145.75% since April 2, 2020 [2]. This continuously increasing number suggests a low adherence of the public to the use of face masks.
Therefore, this study aims to analyze the changes in the frequency of public utilization of face masks during this pandemic, within the period after federal implementation.

Study design and settings
A cross-sectional survey study was conducted and participated by Brazilians adult.

Study sample and variables
Convenience sampling was used to select the study sample, which was composed of adult people aged above 18 years. The speci c adult population projection in 2019 (159,095,000) was used as the basis to calculate the sample size, considering 1.2% margin of error and 99% of con dence interval. Although the result provided a sample size of 11,556 adult people, a total of 14,756 volunteers had participated in this study.
The following variables were collected from the online survey: dates of answering the survey, age, sex, province of residency, monthly income, schooling, occupations, marital status, and the frequency of face mask use. The frequency of face mask use was divided into ve categories: never, rarely, sometimes, frequently, and always.
The respondents were also asked if they consider their occupation at risk for SARS-CoV-2 transmission/infection, if they had close contact with someone diagnosed with COVID-19, if they consider themselves vulnerable to the virus infection, and if they developed feelings of fear toward it.

Data collection procedure
Data were collected online from April 17 to May 15, 2020 via social media (Facebook, Twitter, Instagram, WhatsApp, and e-mail), employing online survey through Google forms. The survey was validated by a committee of experts. Afterwards, a pilot test participated by 20 people was conducted on each region.
This study was widely disseminated in all regions in Brazil. All volunteers read and accepted the free and informed consent terms included in the Google Forms. All ethics aspects were considered.

Statistical analysis
Data were rst analyzed descriptively considering absolute (n) and relative (%) frequencies. Then, the joinpoint regression model was employed to analyze changes in relative frequency of face mask use during the period of this study. The daily frequency for each category was calculated according to the total of people that responded to the survey in that day. In the temporal trend analysis, the categories of frequency were taken as dependent variables and the days as independent variable. The best tting joinpoint regression model was accessed by Monte Carlo permutation test, which employed 4,999 permutations. The daily percentage change (DPC), 95% con dence interval (95% CI) and p-value were considered for each category. Trends were considered increasing or decreasing if the DPC is positive or negative, respectively, and p < 0.05. Otherwise, they were considered stationary.
We employed the chi-squared test to verify the level of association among outcome variables with the periods identi ed in joinpoint regression for the category of "always." We grouped the outcome variables in each one of the trend periods before applying the association test (April 17-25 and April 25 to May 15). Then, we applied binary logistic regression to all association with p < 0.20 to estimate the probabilities associated with the frequency of face masks use in each trend period. We considered Odds Ratio (OR), 95% con dence intervals, and p < 0.05 as statistically signi cance.

Results
The study sample was composed by 14,756 people. Most of them were female (75.4%), single (47.0%), postgraduate (43.5%), from the northeast region (39.2%), and had jobs out of the health eld (58.0%) ( Table 1). The sample averaged age was 35 years old (SD = 13.0), with minimal and maximum ages of 18 and 88 years, respectively. For the frequency of face mask use, the following values were the absolute and relative (%) frequencies for each category: never = 700 (4.7%), rarely = 902 (6.1%), sometimes = 2,674 (18.1%), frequently = 5,814 (39.4%), and always = 4,666 (31.6%). Figure 1 and Table 2 show the joinpoint analysis results. All categories had two trend periods. The "never," "rarely," and "sometimes" categories had downward trends in the rst period and became stabilized in the second one. Conversely, the "frequently" and "always" categories had an upward trend in the rst period. However, the "frequently" category had a stable trend in the second period, and the "always" category had a slow increasing trend.  Table 3 shows the association analysis results among the outcome variables with the two trend periods identi ed in the "always" category. In the second period, an increase in the frequency of face mask use was observed among males, singles, young people, and those living in the north, midwest, northeast, and south regions. The same phenomenon was observed in people with elementary to graduate degrees and those with low rental income. An increase in the frequency of face mask use was observed among people who have low-risk jobs, those did not experience close contact with a sick person, and those who did not express vulnerability to the virus or fear about acquiring the disease. All variables were submitted to binary logistic regression to estimate the association probabilities among them with both periods in the "always" category. The variables associated to both trend periods were gender, age, region, monthly income, occupation considered at risk for virus infection, and feelings of vulnerability to virus. In the second period, an increased frequency of face mask usage was observed in ages 18-34 years (1.5x) and in the south region (2x). A decreased chance was observed to the following predictors: women (1.8x), those aged above 35 years (1.5x), those living in the southeast region (8.0x), those who had a postgraduate degree (7.0x), those who earned seven times the monthly minimal income (2.2x), those who worked in job positions considered at risk for COVID-19 (1.9x), and those expressing vulnerability and feelings of fear toward the disease (1.7x) ( Table 5).  in Malaysia and China [11][12]. In our study, the "frequently" and "always" categories have represented 71% of the sample. However, the ideal "always" category is only 31.6%.
The temporal trend analysis results show that women did not always wear face masks in the second trend period. This result corroborates with previous studies [12][13]. In China, a study with 10,304 volunteers showed that although women practice more protective measures against COVID-19 during the pandemic, only males have shown an upward trend [13]. This nding can be explained by the quarantine restrictions implemented by the Health Ministry in Brazil. After the school activities were interrupted, most women stopped their jobs to take care of their children, thereby decreasing their exposure to the virus. Given that women are less vulnerable to the virus, their frequency of face mask use is also lesser to that of men. Discrepancies among regional implementations of face mask use were also observed. In the second period, an increasing trend was observed in the south region, whereas a decreasing trend was observed in the southeast region. The north region had the lowest frequency in both trend periods. Among all the regions, the south and north regions were the least and the most affected by COVID-19, respectively. The regional discrepancies in COVID-19 incidence are still evident today [2]. The following facts are also considered: 1) regional developmental differences: the south and southeast regions are the richest, The continuous use of face masks increased among the youth in the second period and decreased in people aged above 35 years. A study in Pakistan showed the same result and suggested that young people have higher knowledge regarding prevention practices than the adults [14]. However, 42% of our sample is composed of health professionals, and 55.02% of them assumed older than 35 years old. Like other countries, Brazil has experienced scarcity in face masks and other personal protective equipment (PPE). Consequently, health workers are often working without proper medical gear and attire. Health professionals' long and stressful work and the scarcity of PPE in the national level have increased their feelings of vulnerability and fear of acquiring the virus [15].
Health workers could also have in uenced the faster DPC observed in the rst period than that of the second one (6.95% vs. 1.23%, respectively). By contrast, the slower DPC in the second period suggests a low gradual adherence of the population in general to face masks. This hypothesis corroborates with the increase of face mask use in people earning low monthly income and those with lower degrees compared with the ones in the rst period.
The internal political con icts in Brazil and the spread of fake news still in uence the willingness of the public to adopt the promoted protective measures against COVID-19. The general population in Brazil remains clueless about the pandemic situation they are currently experiencing, and no speci c guidelines are given to them to stop further transmission of the virus [18][19][20]. Negative assumptions and fake news regarding the risks of wearing face masks were widely disseminated during the second period (April 25 to May 15) [21][22]. In addition, utilization of face masks is not a cultural habit in Brazil, and it is never needed that much before. Strategies must be implemented with congruences among policymakers and local authorities to incorporate this new habit in the daily routine of the general population and reinforce the importance of face masks [19][20][21][22].
The study sample is limited by the convenience method used in this study and by the limitation on response rate. Finally, the participants may have also provided socially desirable responses.

Conclusion
Policymakers and local authorities need to reinforce further the frequent use of face masks to the public in addition to other protective personal measures to combat COVID-19. Although 71% of the sample fell under the "frequently" and "always" categories, the ideal "always" category was still lower than the "frequently" category. In addition, more than 29% were not using face masks at all. This nding was alarming given that SARS-CoV-2 could be easily transmitted from one person to another.
Providing PPE and better work conditions to health workers were extremely important and long overdue.