COVID-Inconfidentes: How did COVID-19 and home office influence the prevalence of leisure-time physical inactivity? An analysis of before and during the pandemic


 Background

The COVID-19 pandemic has brought drastic changes to the lives of the global population. The restrictions imposed by government agencies impacted the daily lives of citizens, influencing several health behaviors, such as physical activity (PA). Thus, the present study aimed to assess the prevalence of physical inactivity (PI) and its associated factors before and during the COVID-19 pandemic.
Methods

A population-based household seroepidemiological survey was conducted in two Brazilian municipalities located in the state of Minas Gerais, in which 1750 volunteers were interviewed between October and December 2020. Face-to-face interviews were conducted using a structured questionnaire in an electronic format. The moments considered for the PI analysis were M0 (before the pandemic), M1 (from March to July 2020), and M2 (from October to December 2020). Descriptive statistics and univariate and multivariate logistic regression were used to examine the factors associated with PI before (M0) and during the pandemic (M1 and M2).
Results

The prevalence of PI was higher in the first months of the pandemic (M1) (67.3%; 95% confidence interval (CI): 62.4-71.9) than in the months from October to December 2020 (M2) (58.7%; 95% CI: 52.8-64.3); however, at both times (M1 and M2), PI was more prevalent than in the period before the pandemic started (M0) (39.7%; 95% CI: 35.6-43.8). Individuals who were overweight, obese, and had low educational levels were more likely to be physically inactive. At both M1 and M2, individuals who worked at a home office were less likely to have PI.
Conclusions

The results suggest that the COVID-19 pandemic negatively influenced PA, substantially increasing the prevalence of PI. The determinants associated with PI were education, body mass index, and home office work.


Results
The prevalence of PI was higher in the rst months of the pandemic (M1) (67.3%; 95% con dence interval (CI): 62.4-71.9) than in the months from October to December 2020 (M2) (58.7%; 95% CI: 52. 8-64.3); however, at both times (M1 and M2), PI was more prevalent than in the period before the pandemic started (M0) (39.7%; 95% CI: 35. 6-43.8). Individuals who were overweight, obese, and had low educational levels were more likely to be physically inactive. At both M1 and M2, individuals who worked at a home o ce were less likely to have PI.

Conclusions
The results suggest that the COVID-19 pandemic negatively in uenced PA, substantially increasing the prevalence of PI. The determinants associated with PI were education, body mass index, and home o ce work.

Background
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent of COVID-19, is part of a group of viruses responsible for causing acute respiratory syndrome. Infection caused by this virus has a clinical spectrum, ranging from asymptomatic to severe, and is associated with signi cant morbidity and mortality rates [1]. Due to the speed of spread of the virus and the high rate of infection, concomitant with the lack of knowledge of speci c therapies, strategies have been established to mitigate the spread of the virus and reduce its impact [2]. Thus, one of the main prevention measures recommended by the World Health Organization (WHO) and adopted by governments is social restriction [3,4], aimed at minimizing the exponential growth of infected people, avoiding deaths, and not generating a collapse of health systems [2].
Despite the bene ts related to the decrease in viral circulation, social restriction promoted a sudden and drastic impact, both in the economy, with a reduction in household income, and in aspects related to health, such as eating habits, sleep quality, sedentary behavior, and physical activity (PA) practice [5].
Speci cally, regarding PA practice, in the rst months of the COVID-19 pandemic, the closure of several commercial establishments, such as gyms, and of public spaces open for PA practice, such as squares and parks, may have led to an increase in the prevalence of physical inactivity (PI) in the population.
According to the current recommendations (≥150 min of moderate PA or ≥75 min of vigorous activity), PI refers to the performance of insu cient amounts of moderate to vigorous intensity activities [6,7] and may generate a negative impact on physical and mental health [8] and favor an increase in the prevalence of chronic diseases, such as coronary heart disease, colon and breast cancer, hypertension, stroke, type 2 diabetes, and osteoporosis [9,10].
Thus, in the face of this critical and unprecedented scenario, a better understanding of how the COVID-19 pandemic in uenced the population's PA practice becomes essential to produce important information for decision-making aimed at promoting quality of life and health of the population. Therefore, this study aimed to assess the prevalence of PI and its associated factors before and during the COVID-19 pandemic.

Design and sample
The COVID-Incon dentes study is a population-based seroepidemiological household survey conducted between the months of October and December 2020 in two Brazilian municipalities (Ouro Preto and Mariana), located in the state of Minas Gerais, in the Iron Quadrangle region, which is an area with one of the largest iron ore reserves in the world and is economically important. Face-to-face interviews were conducted, using a structured questionnaire in an electronic format [11].
The sample size calculation considered the 2010 population estimate for the urban area of the municipal headquarters of each municipality, 95% con dence level, estimated infection by SARS-CoV-2 from 3-10%, design effect equal to 1.5, and 20% of recomposition, considering losses due to refusals, absence of the resident drawn, and the existence of closed households during the visit. The sample was calculated using the OpenEpi program (https://www.openepi.com/Menu/OE_Menu.htm), totaling 732 interviews for each municipality. We used conglomerate sampling in three stages: census sector (selected with probability proportional to the number of households), household (selected from a systematic sampling), and resident (randomly selected through the application Sorteador de Nomes®) [11].
The sample weight of each selected unit (census sector, household, and individual) was calculated to correlate with the 2019 population projections (DATASUS) [12]. Adjustments were applied in this calculation to compensate for interview losses due to non-response. Further details on the sample calculation and eld logistics are described by Meireles et al. [11].
Outcome Variable: Leisure-time Physical Inactivity Outcome variable: Leisure-time physical inactivity Participants were asked about PA during leisure time at different times related to the pandemic. For the moment before the pandemic (Moment 0 [M0]), we asked, "Before the pandemic (March/2020) did you practice physical exercise? (1) No; (2) Yes." Then, they were asked about the moment referring to the rst months of the pandemic, referred to as Moment 1 (M1), "During the rst months of the pandemic (March to July/2020), did you participate in physical exercise? (1) No; (2) Yes." Finally, they were asked about the moment of data collection, referred to as Moment 2 (M2), "Do you currently (October-December 2020) participate in any type of physical exercise? (1) No; (2) Yes." Individuals who self-reported participating in PA during leisure time were classi ed as physically active, and those who said they did not participate in PA during leisure time were classi ed as physically inactive.

Explanatory Variables
We considered socioeconomic, anthropometric, and COVID-19-related variables as explanatory variables.
The questions related to the way of working were, "Currently, how is your work routine regarding location? (1) All work activities are being performed at my home (home o ce); (2) Part of the activities are performed in the traditional work environment, i.e., some days at home and others at the workplace; (3) All work activities are being performed in my work environment." This was categorized as normal routine (all activities at the workplace), home o ce (partial and/or total home o ce work), and no work (individuals who were not working during the pandemic).
Questions related to the COVID-19 pandemic were also evaluated, such as presenting at least one symptom in the last 15 days (fever, feeling feverish, palpitation, diarrhea, sore throat, cough, di culty breathing, vomiting, skin rashes, unusual tiredness, ageusia, and anosmia). Responses were categorized into the presence (one or more symptoms) and absence of symptoms.
Social withdrawal was assessed using the question, "Are you currently socially withdrawn? (1) No; (2) Yes." Finally, for seroepidemiological evaluation of anti-SARS-CoV-2 antibodies, serum samples were obtained by venipuncture using a 7.5 mL S-Monovette® (Sarstedt) serum gel tube. The samples diagnosed by qualitative immunochromatographic method using the One Step COVID 2019® test (Guangzhou Wondfo Biotech, China), according to the manufacturer's protocol. The seropositive results for this method are de ned by the presence of two bands indicating, respectively, its performance and the non-differentiated presence of IgM/IgG anti-SARS-COV-2 antibodies [11].

Ethical Declarations
This study was approved by the Research Ethics Committee of the Universidade Federal de Ouro Preto, under protocol number 32815620.0.1001.5149. All procedures adopted in this study followed the Declaration of Helsinki and the Brazilian guidelines and norms for research involving humans.

Data analysis
The statistical analyses considered the complex sample design using the svy command of Stata® software, version 15.0. Descriptive analysis was performed, with calculation of frequency and 95% con dence interval (95%CI) for all variables of interest. McNemar's paired test was used to verify the change in the prevalence of PI during the three evaluations. Pearson's χ 2 test was used to assess the possible relationship between home o ce and sociodemographic factors.
Univariate logistic regression was performed to assess the factors associated with PI before the pandemic and at the two time points of the COVID-19 pandemic. Based on the results of this analysis, variables with p ≤ 0.20 were included in the multivariate logistic regression. A stepwise backward approach was used to choose the nal model, and variables with p ≤ 0.05, were retained. The odds ratios (ORs) and 95% CIs were calculated.
It is noteworthy that the variables income and education showed high collinearity in this study, so we chose education, instead of income, in the multivariate model.

Results
The survey included 1750 individuals, with the majority being female (52.4%; 95% CI: 40.5-54. 8  In Table 2, we observe the prevalence of PI according to the moments investigated and the crude OR of the explanatory variables and PI. No associations were observed between PI and the variables gender and related to COVID-19 (social withdrawal, symptomatology, and testing positive for COVID-19). On the other hand, age, education level, income, and BMI were associated with PI at different time points. It is noteworthy that self-reported skin color and marital status variables were associated only with M2. Furthermore, working at a home o ce was associated with PI at M1 and M2.

Discussion
This study presents important evidence on PI during the COVID-19 pandemic, suggesting that social restriction, although necessary, may have contributed to a higher prevalence of PI at the two time points assessed after the pandemic onset. The results revealed that before and during the pandemic, leisuretime PI was associated with lower educational attainment, overweight, and obesity. Moreover, working in a home o ce was a protective factor for PI in both moments evaluated during the pandemic.
PA practice bene ts different aspects of life, whether physical, mental, and social [16]. However, the pandemic caused abrupt changes, which contributed to the increase in the prevalence of PI [17]. Several studies have already shown evidence of a PI pandemic, even before COVID-19 [18], leading the WHO to launch a global action plan to encourage PA practice in June 2018, seeking a 15% reduction in PI rates by 2030 [19]. However, with the advent of the COVID-19 pandemic, there has been a worsening of this scenario due to social restriction and reduced urban mobility [17].
Our ndings are corroborated by recent literature, which highlights the considerable increase in PI during the COVID-19 pandemic. The results further indicate that the rates of PI found in the study population during the pandemic are more than double the global prevalence (27.5%), as well as is higher than the rate found in other countries during social restriction [20,21]. In China, a population-based study conducted in the early pandemic period (January and February 2020) with 12,107 participants found that 60% of the population did not meet the 150 min per week of PA in the four domains (leisure, commuting, home, and work) and were, therefore, considered physically inactive [20]. A similar condition was reported in Spain, in which self-reported PA in the four domains decreased signi cantly across the population during the lockdown, resulting in a reduction of 16.8% (p<0.001) and 58.2% (p<0.001) in vigorous PA and walking time, respectively [22]. Moreover, corroborating our results, an online survey conducted in Brazil between April and May 2020 with 43,995 adults found a prevalence of leisure-time PI of 66.6%, with 21.3% of individuals becoming inactive after the onset of the pandemic [23].
It is important to highlight that PA is determined by several individual, social, environmental, and political factors [24], and, in general, it may vary according to age, gender, income, and education level [25]. In the present study, we found an association between leisure-time PI and lower educational level, overweight status, and obesity, which can be considered to hinder the participation of leisure-time PA [26,27].
It is well documented that lower education is associated with PI in general. Our ndings reinforce data from the literature in which low education was reported as a factor that increases the likelihood of PI [28]. In a study by Kari et al. [25], the authors stated that a higher level of education is related to making healthier lifestyle decisions, including PA. Similar ndings were found in the study by Park and Kang [29], which showed that an increase in the years of schooling in adulthood induces individuals to participate in PA more regularly. Furthermore, a direct relationship is perceived between schooling and PA that may be permeated by income, which, in turn, offers more opportunities to invest in PA [25,30]. Furthermore, a peer-reviewed systematic review identi ed education as a positive determinant of PA [31].
The higher probability of occurrence of PI in overweight or obese individuals found in our study is a relevant factor because, in addition to complicating the overweight condition, it may contribute to the emergence and worsening of chronic diseases [32]. Studies prior to the COVID-19 pandemic have stated that a higher level of PA is more prevalent in individuals with lower BMI, and the reverse is also true, that is, subjects with higher BMI are less likely to meet the guidelines for moderate to vigorous PA [33][34][35][36].
Although the results of our study cannot demonstrate causal relationships, it is well-established in the literature that participating in PA prevents obesity and metabolic diseases arising from overweight status [32,37,38]. These ndings are alarming and require the implementation of actions and public policies to encourage PA engagement, as the restrictions imposed to mitigate the circulation of SARS-CoV-2 contributed to weight gain in the population and, in the current scenario, the control of overweight status or obesity becomes more challenging. Moreover, the high prevalence of overweight status and obesity, along with reduction in PA practice during the pandemic, increases vulnerability to several diseases, including COVID-19, leading to increased costs and a possible crisis in the health care system [39,40].
Remote work was an important change implemented in the routine of the population due to the COVID-19 pandemic, and home o ce work was identi ed as a protective factor for FI during the period of social restriction. To date, very few studies have aimed at understanding the health implications of home o ces. Changes in physical behaviors, such as increased sitting and lying down time, and less time spent on PA were expected [41]. However, our study found an association between home o ce work as a protective factor for PI. This is in contrast to the ndings of a study recently conducted in Brazil, in which o ce workers were found to have reduced their PA practice in leisure time when comparing the periods before (o ce workday, September 2019) and during the pandemic (home o ce workday, July 2020), [41].
However, these results were contradictory. In a study conducted in Sweden that compared the days workers stayed at home vs. the days they commuted to the o ce, it was shown that physical behaviors during work and leisure did not change signi cantly. The authors suggested that individuals tend to work fewer hours when they perform their function at home, which may be an indication of additional time for PA in leisure time [42].
Additionally, a longitudinal study of 112 o ce workers in the United States, initiated in January 2018, immediately before (i.e., February 2020) and during the detachment of COVID-19 (i.e., at the time of survey data collection), recorded no signi cant changes in PA during leisure [43]. Another study by Martins et al. [44] observed a higher prevalence of PI among people who work than among those who did not work. This was due to the lack of time available for leisure, which is a common report regarding physical exercise or sports [44]. This may add to our ndings as subjects in a home o ce were able to have a exible schedule and, therefore, dedicate themselves to some PAs in their leisure time.
The abovementioned ndings indicate that working at home can create opportunities for individuals to become physically active. It reduces the time spent commuting to work and/or school, in addition to reducing time spent on the workday, providing exible schedules and additional time to engage in PA. This association is also in uenced by the pro le of individuals who worked at home o ces, with a higher proportion having higher income and more education.
Finally, this study has some limitations. The outcome variable was evaluated only in a binary way if individuals engaged in PA during leisure time. Thus, we did not evaluate the frequency, duration, and type of PA modality, limiting the obtainment of information about the individuals' level of engagement, according to the world recommendations on PA. However, it is important to point out that, to the best of our knowledge, our study is one of the rst investigations on PI at different times of the COVID-19 pandemic in Brazil. This outcome variable is among the top ve risk factors for increased chronic disease occurrence and mortality worldwide, accounting for up to 10% of the global burden of coronary heart disease, type 2 diabetes, breast and colon cancer, and premature mortality [45]. Therefore, information about the impacts of the COVID-19 pandemic on PI is valuable and contributes to the planning and targeting of population health promotion actions. Therefore, it is important to understand the changes in PI during the period of restricted social movement, so that the negative effects can be reversed, from the creation and implementation of effective and feasible policies to increase PA at all population levels.

Conclusions
These results suggest that the disruption of daily routine due to the COVID-19 pandemic may have negatively in uenced the practice of leisure-time PA, substantially increasing the prevalence of FI in the rst months of the pandemic (M1), a condition that remained at M2. We also found that during the COVID-19 pandemic, individuals with lower education, overweight status, and obesity were more likely to be physically inactive, and home o ce work favored PA.

Consent for publication
Informed consent was obtained from all individual participants included in the study.
Availability of data and materials