key: cord-346310-venpta28
authors: Filgueiras, A.; Stults-Kolehmainen, M.
title: Factors linked to changes in mental health outcomes among Brazilians in quarantine due to COVID-19
date: 2020-05-16
journal: nan
DOI: 10.1101/2020.05.12.20099374
sha: 
doc_id: 346310
cord_uid: venpta28

The 2020 COVID-19 pandemic is a crisis of global proportions with a significant impact on the country of Brazil. The aims of this investigation were to track changes and risk factors for mental health outcomes during state-mandated quarantine. Adults residing in Brazil (n = 360, 37.9 years of age, 68.9% female) were surveyed at the start of quarantine and 1 month later. Outcomes assessed included perceived stress, state anxiety and depression. Aside from demographics, behaviors and attitudes assessed included exercise, diet, use of tele-psychotherapy and number of COVID-19 related risk factors, such as perceived risk of COVID-19, information overload, and feeling imprisoned. Overall, all mental health outcomes worsened from Time 1 to time 2, although there was a significant gender x time interaction for stress. 9.7% of the sample reported stress above the clinical cut-off (2 SD above mean), while 8.0% and 9.4% were above this cutoff for depression and anxiety, respectively. In repeated measures analysis, female gender, worsening diet and excess of COVID-19 information was related to all mental health outcomes. Changes in diet for the worse were associated with increases in anxiety. Exercise frequency was clearly related to state anxiety (0 days/week > 6 days/week). Those who did aerobic exercise did not have any increase in depression. Use of tele-psychotherapy predicted lower levels of depression and anxiety. In multiple regression, anxiety was predicted by the greatest number of COVID-19 specific factors. In conclusion, mental health outcomes worsened for Brazilians during the first month of quarantine and these changes are associated with a variety of risk factors.

The 2020 COVID-19 pandemic is a crisis of global proportions with a significant impact on the 23 country of Brazil. The aims of this investigation were to track changes and risk factors for mental 24 health outcomes during state-mandated quarantine. Adults residing in Brazil (n = 360, 37.9 years 25 of age, 68.9% female) were surveyed at the start of quarantine and 1 month later. Outcomes 26 assessed included perceived stress, state anxiety and depression. Aside from demographics, 27 behaviors and attitudes assessed included exercise, diet, use of tele-psychotherapy and number of 28 COVID-19 related risk factors, such as perceived risk of COVID-19, information overload, and 29 feeling imprisoned. Overall, all mental health outcomes worsened from Time 1 to time 2, 30 although there was a significant gender x time interaction for stress. 9.7% of the sample reported 31 stress above the clinical cut-off (2 SD above mean), while 8.0% and 9.4% were above this cutoff 32 for depression and anxiety, respectively. In repeated measures analysis, female gender, 33

worsening diet and excess of COVID-19 information was related to all mental health outcomes. 34 Changes in diet for the worse were associated with increases in anxiety. Exercise frequency was 35 clearly related to state anxiety (0 days/week > 6 days/week). Those who did aerobic exercise did 36 not have any increase in depression. Use of tele-psychotherapy predicted lower levels of 37 depression and anxiety. In multiple regression, anxiety was predicted by the greatest number of 38 Introduction 44 Mental health comprises the set of emotions, thoughts and behaviours that enable individuals to 45 work, cope and deal with problems in everyday tasks (WHO, 2004) . Historically, although 46 researchers from the biomedical sciences dedicated more time and resources in the study of 47 physical health, findings from the last 50 years have slowly captured the interest of scientists 48 from diverse fields to look upon mental health to explain somatic diseases, physical functioning, 49 quality-of-life, well-being and work productivity, (Christensen et al., 1999; Prince et al., 2007; 50 Stults-Kolehmainen, Tuit & Sinha, 2014) . For instance, mental health is associated with 51 disability-adjusted life years (DALYs) and premature mortality (Vigo, Kestel, Pendakur et al., 52 2019) with 17% of DALYs attributable to mental health in Brazil and 22% in the United States. 53

Those with worse mental health, such as higher levels of chronic stress, have a greater risk for 54 physical health problems, such as cardiovascular disease (Stults-Kolehmainen, 2013). Poor 55 mental health costs society a great deal of money, in terms of lost productivity, strain on 56 healthcare systems, loss of income and other consequences (Trautman, Rehm, Wittchen, 2016). 57

On the other hand, recent research from the World Health Organization suggests that every one 58

American-dollar spent in mental health care is equivalent to a return of four American-dollars in 59 better well-being and ability to work (WHO, 2016).Thus, a person who has good mental health 60 entails someone who is physically healthy, happy and productive for themselves and the greater 61 functioning of society (Prince et al., 2007; WHO, 2016) . 62

The recent outbreak of the Corona Virus Disease 2019 (COVID-19 or SARS-CoV-2) around the 63 world at the end of 2019 and the beginning of 2020 led to a series of guidelines to avoid mass 64 contamination and limit its lethality (WHO, 2020). Among these recommendations are 65 quarantine, confinement and social distancing (Wilder-Smith & Freeman, 2020). These 66 impositions mean that people cannot walk freely from their homes; they need to keep a 2-meter 67 physical distance from one another on the streets and sick people are obliged to be confined in 68 hospitals or their own homes without any kind of physical proximity to others. These restrictions 69 are intended to benefit the physical health and safety of all people and must be adopted to save 70 lives. Unfortunately, such directives come at a cost to the mental health and well-being a 71 substantial proportion of the population (Rubin & Wessely, 2020 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 May 16, 2020. .

An updated systematic review on the effects of social distancing and quarantine on mental health 78 revealed that anxiety, depression, stress, anger, insomnia, hopelessness, and sadness were all 79 increased during those conditions (Brooks et al., 2020) . A recent study (Hu, Su et al., 2020 ) from 80 a cross-national sample (n = 992) in China found that levels of anxiety increased, and 9.6% of 81 the population was anxious at clinically relevant levels. Other behavioural problems also appear 82 during this period; participants in a nationwide survey recently published in China reported 83 nutritional issues, lack of ability to exercise and numerous changes in daily routines and habits 84 (Qiu et al., 2020) . Accordingly, psychosocial and behavioural dimensions seem associated under 85 quarantine conditions (Filgueiras & Stults-Kolehmainen, 2020 Unfortunately, resources are scarce in every field of the health system, including those for mental 107 health (Qiu et al., 2020) . Therefore, it is pivotal to establish a priori where and how to invest 108 those scarce resources. This is a difficult task because the current stressor is highly unique. 109

Quarantine is due to a pandemic of truly global proportions that has reached every level of 110 society, with a long duration and remarkable social upheaval (WHO, 2020). There is no research 111 on the association between psychological, demographic and behaviour variables in the general 112 population during society-wide social isolation. Furthermore, it is a consensus that psychological 113 phenomena, such as stress and depression, are multifactorial with a large amount of variables to 114

consider (WHO, 2004; 2016) . In order to help governments, service providers and scientists to 115 . CC-BY 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) comparing before and during quarantine (options were "no changes"; "increased exercise 146 frequency" and "decreased exercise frequency") and (xiii) types of exercise (aerobic, anaerobic, 147 both, no exercise). It also collected data regarding diet and nutritional habits: (xiv) possible 148 changes on diet by comparing before and during quarantine; whether the person (xv) gained or 149 (xvi) lost more than 5 kilograms since the beginning of the quarantine. Finally, attitudinal 150

questions were also computed. One question (xvii) asked about the amount of information the 151 participant felt he/she was receiving and the answers were provided in three possible categories 152 to choose from: "Too much information", "Enough information" and "Little information". 153 . CC-BY 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.

The copyright holder for this preprint this version posted May 16, 2020. .

Another three items were informed in a five-point Likert-type scale ranging from 1 "Totally 154 agree" to 5 "Totally disagree"; the items were: (xviii) "Do you feel imprisoned due to this 155 quarantine?", (xix) "Do you feel you are able to understand what is happening?", (xx) "Do you 156 trust your own ability to differentiate good from bad sources of information?". 157

The PSS-10 (Cohen & Williamson, 1988 ) is a 10-item questionnaire that asks individuals about 158 their perception regarding stress-like symptoms. It is answered in a five-point Likert-type scale 159 ranging from 0 "Never" to 4 "Very often" (scores range from 0-40). The population mean is 17.0 160 (SD = 5.02) with a score over 27 indicating excessive stress (Cacciari, Haddad, Dalmas, 2016) . 161

The FDI (Filgueiras et al., 2014 ) is a 20-item scale that asks individuals to grade the level of 162 association between the respondent's own self-perception and one-word items extracted from 163 depression symptoms listed in the DSM-V in the last fortnight. It is rated in a six-point Likert-164 type scale ranging from 0 "not related to me at all" to 5 "totally related to me" (scores range 165 from 0-100 of the respondent who answers questions about own feelings in a four-point Likert-type scale 170 ranging from 1 "not at all" to 4 "very much so" (scores range from 0-80). Gender-specific 171 reference means are 36.5 (SD = 21.4) for men and 43.7 (12.6) for women, with cut-offs being 66 172

for men and 69 for women (Pasquali, Pinelli Jr, Soha, 1994) . 173

Volunteers of the present research answered the questionnaires in the Google Forms online 174 platform that was configured in the same order of presentation: 1) Term of Consent, 2) 175 demographic and attitudinal questionnaire, 3) PSS-10, 4) FDI, 5) S-STAI, 6) Thank you page. 176

Those participants who answered "no" to the Term of Consent were addressed to the Thank you 177 page without having any contact with the other questionnaires. First round of data collection 178 (time 1) took place between March 20 th and March 25 th , 2020, whereas the second round (time 2) 179 happened between April 15 th and April 20 th , 2020. 180

After data collection, Google Spreadsheets were utilized to consolidate the database and to 181 export it in the format .csv. Then, researchers used SPSS (IBM, version 21.0) to run the analyses. 182

Descriptive statistics of PSS-10, FDI and S-STAI were calculated for each categorical 183

(demographic) variable with exception of those that were answered in Likert-type scales. Due to 184 the large amount of variables collected in an online platform, Cronbach's alpha (α) was 185 calculated for the three scales in time 1 and time 2; results were expected to show α > .70. 186

Pairwise t-test comparisons between groups were computed to identify significant differences 187 between the first round (time 1) and second round (time 2) of data collection for the whole 188 sample. A repeated-measures ANOVA was performed to compare within and between groups 189

for each demographic independent variable. Furthermore, prevalence of stress, depression and 190 anxiety-like symptoms were calculated in percentage of participants above the means and cut-off 191 . CC-BY 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.

The copyright holder for this preprint this version posted May 16, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 16, 2020. For perceived stress, 237 (65.8%) and 269 (74.7%) of participants scored above the population 259 mean at time 1 and 2, respectively. Prevalence of excessive stress (>2 SD above reference mean) 260 was 6.9% (IC 95 5.2%-8.6%) in the first round and 9.7% (IC 95 8.2%-11.2%) in the second 261 round. Of the 34 individuals in this category, 94% of these individuals were women. 82% did no 262 exercise at all, but the remaining 18% complete 6 days a week of exercise. Also, 0% utilized 263 tele-psychotherapy. Regarding depression, 224 (62.2%) and 260 (72.2%) of participants were 264 above the reference mean at Time 1 and 2, respectively. High depression (>2 SD above reference 265 mean) had a prevalence of 4.2% (IC 95 3.6%-4.8%) at time 1 and 8.0% (IC 95 7.1%-8.9%) at 266 . CC-BY 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.

The copyright holder for this preprint this version posted May 16, 2020. . time 2. Participants > 2 SD (n = 24) were mostly women (88%) and did not utilized tele 267 psychotherapy (88% as exercise frequency and perceived stress (r = -.28); whereas, moderate correlations were found 278 between the same variable between time 1 and time 2 (intertemporal correlations). Tables 3 and  279 Supplemental 2 provide the correlation matrix of the psychological variables. to understand what is happening, level of education and gender respectively. Independent 291 variables explained 33% of the variance of depression in the second round of data collection. 292

Finally, the state anxiety LMR depicted that the dependent variable (S-STAI time 2) was 293 predicted, in order of association, risk for COVID-19, feeling safe, the score of S-STAI time 1, 294 weight loss, changes on diet, amount of information, feeling imprisoned and age. Independent 295 variables of this LMR explained cumulatively 42% of the variance. Table 3 presents the  296 coefficient β , the t-test statistics, effect-size and coefficient of determination for the three LMR. The current investigation provides a unique glimpse into the mental health of Brazilians in the 302 midst of quarantine from the COVID-19 pandemic, a novel, disruptive and society-wide stressor. 303

Findings indicate that a substantial portion of respondents were distressed at both time points, 304 . CC-BY 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.

The copyright holder for this preprint this version posted May 16, 2020. . https://doi.org/10.1101/2020.05.12.20099374 doi: medRxiv preprint with worsening mental health from the initiation of quarantine to a point one month later. More 305 specifically, increases in perceived stress, depression and state anxiety were observed, with a 306 gender x time interaction recorded for stress. Men experienced increases in depression and 307 anxiety over time, but not for perceived stress. Across genders, the number of days in quarantine 308

was linearly related to worse perceptions of perceived stress. Repeated measures ANOVA 309 revealed that 3 factors were all related to worse levels of stress, depression and anxiety: female 310 gender, worsening diet and excess of COVID-19 information. In regression analyses, however, 311 mental health outcomes were influenced by a variety of other demographic, COVID-19 specific, 312

and behavioural factors, such as use of tele-psychotherapy. Exercise-related factors, such as 313 exercise frequency, were the predominate predictors of perceived stress. 314

A substantial portion of the participants reported levels of stress, depression and anxiety above 315 established means for the population. At time 2, greater than 70% of the sample was above the 316 normative mean for both stress and depression. For anxiety, >60% of both men and women were 317 above the normative mean. More importantly, some participants scored very high for mental 318 health disturbances, especially at time 2. For stress, 9.7% of the sample was above 2 SD at time 319 2, whereas the prevalence according to the Brazilian norms is 6.8% (Cacciari, Haddad & 320 Dalmas, 2016) . This was an increase from 6.9% at time 1. Similar trends were seen for 321 depression (4.2% at time 1, 8.0% at time 2; versus a norm of 4.1%) (Filgueiras et al., 2014 ) and 322 state anxiety (8.7% increasing to 14.9%; versus a norm of 9.4%) (Pasquali, Pinelli Jr & 323 Solha,1994) . This is similar to anxiety levels observed in a large sample during quarantine in 324

China (Hu, Su et al., 2020) . While the percentage of individuals scoring at these extremes is still 325 relatively low, it potentially represents a huge increase in burden to society when multiplied 326 across the entire population. Mental health initiatives on the national level would have to be 327 scaled up to meet new demand (WHO, 2008) . Key to this endeavour would be a) identifying 328 those most at risk and b) properly assessing their condition. 329

In the effort to identify those most at risk, pertinent predictors of mental health outcomes were 330

analysed. Interestingly, each mental health indicator was predicted by a varying set of factors. condition of regrets about the past (Buechler, 2015) , was understandably not predicted by 338 COVID-19 related factors. Only "understanding what is happening" was a significant inverse 339

predictor. Stress was predicted by feelings of being imprisoned, days in quarantine and risk for 340

COVID-19 and also by a number of exercise factors. 341

In general, exercise was associated with mental health outcomes in the expected manner -more 342 frequent exercise and aerobic exercise being related to the lowest levels of distress. For all 3 343 . CC-BY 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.

The copyright holder for this preprint this version posted May 16, 2020. . mental health outcomes, those with no exercise (0 days per week) had the highest average levels 344

of stress (22.9 at time 1 to 26.4 at time 2), depression (69.0 to 74.6) and anxiety (48.2 to 54.7). 345

These seems to support the previous findings that "something is better than nothing" ( slightly different, with changes in exercise not being significant, but use of online fitness 360 coaching reaching significance. An interaction was observed in that those who performed 361 aerobic exercise had the lowest levels of depression at both time points. In fact, those who did 362 aerobic exercise did not have any increase in depression. However, the clearest association of 363 exercise frequency and mental health was for anxiety. Those at the highest levels of exercise had 364 the lowest anxiety and each day less was associated with more anxiety. 365

Aside from exercise, there were notable findings for dietary habits and use of tele-366 psychotherapy. Those who rated their dietary habits as becoming worse also had the highest 367 levels of stress, depression and anxiety. Those with the highest levels of anxiety were those with 368 worsening diet at the second time point (effect size for interaction was .37). Those who used 369 online nutrition services had lower levels of depression, but there was no difference for stress or 370

anxiety. Those who utilized online psychotherapy reported lower levels of depression and 371

anxiety. While there is no income data to explain use of online resources, those using online 372 resources were more educated. Thus one might surmise that those from better off demographic 373 groups are less affected partly because of greater access to resources. Given the limited quantity 374 of resources to mitigate mental health impairments during crises, such as pandemic and 375 quarantine, it is crucial to identify the risk factors that may predispose individuals for worsening 376 outcomes. 377

Despite the progress this study makes in tracking changes in mental health and identifying risk 378 factors, the current research does demonstrate some limitations. First of all, there was no pre-379 quarantine baseline and assessments spanned just a single month. Furthermore, this was a 380

relatively well-off population with higher-educated individuals being over-represented in the 381

sample. There was no measure of adherence to quarantine guidelines. It is possible that those 382

with higher compliance to regulations could be of either higher or lower distress. To lessen 383 . CC-BY 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 May 16, 2020. . survey fatigue for participants, validated measures of exercise and dietary habits, which can be 384 very lengthy, were not utilized. More importantly, the current data needs interpreted with some 385 caution because factors other than quarantine could contribute to changes in the mental health 386 outcomes observed, such as growing political and economic unrest in Brazil (THE LANCET, 387 2020). Also, it should be noted that effect sizes for changes over 1 month were small (Cohen's d 388

were .25 -stress, .30 -depression, and .38 -anxiety), possibly because in some cases 389 individuals had improved mental health (n = 31; 8.6%) due to quarantine conditions, such as 390 being closer to loved ones throughout the day or being removed from dangerous work 391 environments. Lastly, correlations between instruments at time 1 or time 2 were small -possibly 392

indicating the uniqueness of the quarantine as a stressor, particularly given the rapidly changing 393 circumstances during this time period (Main, Zhou et al., 2011) . 394 395

This study provides crucial data needed to understand how pandemic, state-mandated quarantine 397

is related to changes in mental health outcomes. From the time point when quarantine was 398 decreed until 1 month later, worsening perceived stress, depression and anxiety was observed in 399 this sample of the Brazilian population. Moreover, many individuals in the sample reported very 400 high levels of distress (> 2 SD). At the time of writing of this study, the quarantine is still being 401 enforced and cases of COVID-19 and associated deaths on rising rapidly (THE LANCET, 2020; 402

Imperial College COVID-19 Response Team, 2020). Future research should continue to track 403 these trends as the crisis unfolds. Analyses from this study identified several risk factors for 404 mental health, including gender (being female), lower education, less exercise, worsening diet 405 and a lack of resources, such as access to tele-psychotherapy. COVID-19 related factors 406 predicted anxiety and stress more so than depression. The implications of these data is clear; 407 mental health worsens with great change, requiring more resources to improve the experience of 408 life in quarantine. The extent to which these can be diligently developed and allocated will 409 depend on a data-driven process such as described here. 410 411

Everyday stressors and gender differences in daily distress

Survey of stress reactions among health care workers 422 involved with the SARS outbreak

The psychological impact of quarantine and how to 424 reduce it: rapid review of the evidence

La alegría, la vergüenza, el arrepentimiento y la tristeza en la terapia1

Nível de estresse em trabalhadores readequados 428 e readaptados em universidade estadual pública. Texto & Contexto-Enfermagem

Hope, hopelessness, and anxiety: A pilot 431 instrument comparison study

Perceived stress in a probability sample of the United States

Age differences in depression and anxiety 436 symptoms: a structural equation modelling analysis of data from a general population sample. 437 Psychological medicine

The effect of exercise on clinical depression and depression 439 resulting from mental illness: A meta-analysis

Walking in (affective) circles: can 442 short walks enhance affect?

Development and psychometric 444 properties of a novel depression measure

The Relationship Between Behavioural and

Psychosocial Factors Among Brazilians in Quarantine Due to COVID-19. SSRN, e3566245

Feeling safe and content: 448 A specific affect regulation system? Relationship to depression, anxiety, stress, and self-449 criticism

SARS control and psychological effects of 451 quarantine

Countrywide quarantine only mildly increased 453 anxiety level during COVID-19 outbreak in China. medRxiv

Short-term forecasts of COVID-19 deaths in 455 multiple countries

Mapa brasileiro da COVID-19 [Brazilian Map for COVID-19

Mental health status of people isolated due to Middle 460 East respiratory syndrome

Physical activity and mental health

Relations of SARS-related stressors and 464 coping to Chinese college students' psychological adjustment during the 2003 Beijing SARS 465 epidemic

Diet quality and depression risk: a 467 systematic review and dose-response meta-analysis of prospective studies

Gender differences in depression. Current directions in psychological 470 science

The association between 472 physical activity and depression among individuals residing in Brazil. Social psychiatry and 473 psychiatric epidemiology

Relationship between diet and mental health in 475 children and adolescents: a systematic review

Contribuiçäo à validaçäo e normalizaçäo da escala de 478 ansiedade-traço do IDATE

No health without mental health

A nationwide survey of psychological distress among 482

Chinese people in the COVID-19 epidemic: implications and policy recommendations

The psychological effects of quarantining a city

Decreto 64.881: Decreta quarentena no Estado de São Paulo, no contexto da 486 pandemia do COVID-19 (Novo Coronavírus), e dá providências complementares

STAI: manual for the State-Trait Anxiety 490 Inventory

Education level, income level and mental health 492 services use in Canada: Associations and policy implications

The interplay between stress and physical activity in the 494 prevention and treatment of cardiovascular disease

The effects of stress on physical activity and exercise. 496 Sports medicine

Lower cumulative stress is associated with 498 better health for physically active adults in the community

Factors influencing psychological distress during a 500 disease epidemic: data from Australia's first outbreak of equine influenza

COVID-19 in Brazil

The economic costs of mental disorders

Disease burden and government 506 spending on mental, neurological, and substance use disorders, and self-harm: cross-sectional, 507 ecological study of health system response in the Americas

Physical activity and depression: is 150 min/week 510 of moderate to vigorous physical activity a necessary threshold for decreasing risk of depression 511 in adults? Different views from the same data. Social psychiatry and psychiatric epidemiology

Isolation, quarantine, social distancing and community 514 containment: pivotal role for old-style public health measures in the novel coronavirus (2019-515 nCoV) outbreak

Out of the shadows: making mental health a global development priority

World Bank Group and World Health Organization

Promoting mental health: concepts, emerging evidence, practice (Summary 521 Report). Geneva: World Health Organization

Considerations for quarantine of individuals in the context of containment for 524 coronavirus disease (COVID-19): interim guidance

The anxiolytic effects of exercise: a meta-analysis 528 of randomized trials and dose-response analysis

The present research is funded by the Coordenacao de Aperfeicoamento de Pessoal de Nivel 413Superior (CAPES) of the Ministry of Education of Brazil under the PROAP program.