Predictors of Depression and Anxiety Symptoms in Brazil during COVID-19

The COVID-19 pandemic in Brazil is extremely severe, and Brazil has the third-highest number of cases in the world. The goal of the study is to identify the prevalence rates and several predictors of depression and anxiety in Brazil during the initial outbreak of COVID-19. We surveyed 482 adults in 23 Brazilian states online on 9–22 May 2020, and found that 70.3% of the adults (n = 339) had depressive symptoms and 67.2% (n = 320) had anxiety symptoms. The results of multi-class logistic regression models revealed that females, younger adults, and those with fewer children had a higher likelihood of depression and anxiety symptoms; adults who worked as employees were more likely to have anxiety symptoms than those who were self-employed or unemployed; adults who spent more time browsing COVID-19 information online were more likely to have depression and anxiety symptoms. Our results provide preliminary evidence and early warning for psychiatrists and healthcare organizations to better identify and focus on the more vulnerable sub-populations in Brazil during the ongoing COVID-19 pandemic.


Introduction
We launched an online survey to study the mental health of adults in Brazil during 74 the COVID-19 pandemic on 9-22 May 2020. The survey was conducted through a link to 75 Google Forms to preserve social distancing protocols and to reach people across Brazil's 76 large and diverse regions. We used the non-probabilistic sampling technique of quota 77 sampling to approximate a representative sample of Brazilian adults. Quota sampling is 78 one of the most popular sampling methods and a viable method [34] to conduct online 79 surveys across all regions of Brazil without access to a probabilistic panel. The use of 80 quota sampling by age, gender and social class was effective and viable in our case to 81 obtain a sample that represents the population in Brazil. The study sampled adults aged 82 18 years or older by unclustered systematic random samples from all 23 states in Brazil. 83 The survey, in Brazilian Portuguese, contained a cover page, which explained the pur-84 pose of the study and all the participants consented before starting the survey. Questionnaire-9 (PHQ-9) is one of the most established depression scales, which cap-96 tures the frequency and severity of depression related symptoms in the past two weeks, 97 with a total of 9 items. In this study, the internal consistency coefficient of PHQ-9 is 0.902.

98
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 3, 2021. ; https://doi.org/10.1101/2021.06.28.21259409 doi: medRxiv preprint or did you overeat (0 = no day 1 = less than a week 2 = a week 3 = almost every day)? 100 The Generalized Anxiety Disorder scale (GAD-7) is a simple and effective way to 101 evaluate generalized anxiety disorder (GAD), a mental disorder with long-term persis-102 tent anxiety and excessive anxiety as the core symptoms. In this study, GAD-7 has 7 103 items, and the internal consistency coefficient is 0.937. A sample item is: in the past 2 104 weeks, how often did you feel nervous, anxious or very tense (0 = rarely 1 = some days 2 105 = more than half the days 3 = almost every day)?

118
The descriptive statistics are shown in Table 1. In this sample, 45.9% of the adults 119 were male and 54.1% were female, and the average age was 36.69 years old (SD=13.50). 120 Over 70% (72.5%) were doing or had college degrees or above, and just under half

131
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 3, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2021.  The results of the ordinal multi-class logistic regression are in Table 3. 158 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2021. ;  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2021. ; . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2021. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2021.  CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 3, 2021. ; https://doi.org/10.1101/2021.06.28.21259409 doi: medRxiv preprint