Can physical activity protect against depression and anxiety during the COVID-19 pandemic? A rapid systematic review

The Covid-19 pandemic is affecting the entire world population. During the rst spread, most governments have implemented quarantine and strict social distancing procedures. Similar measures during recent pandemics resulted in an increase in post-traumatic stress, anxiety and depression symptoms. The development of novel interventions to mitigate the mental health burden are of outmost importance. In this rapid review, we aimed to provide a systematic overview of the literature with regard to associations between physical activity (PA) and depression and anxiety during the COVID-19 pandemic. We searched major databases (PubMed, EMBASE, Sportdiscus, Web of Science) and preprint servers (MedRxivs, SportRxiv, ResearchGate and Google scholar), for relevant papers up to 25/07/2020. We identied a total of 21 observational studies (4 longitudinal, one cross-sectional with retrospective analysis and 16 cross-sectional), including information of 42,293 (age range = 6-70 years, median female = 68%) participants from 5 continents. The early evidence suggests that people who performed PA on a regular basis with higher volume and frequency and kept the PA routines stable, showed less symptoms of depression and anxiety. activity routines even under pandemic-specic barriers. However, there is a clear need for more systematic research for effectively and safely usable apps or web-based programs to prevent psychiatric disorders through physical activity.


Introduction
With 23,057,288 con rmed cases all over the world (up to August 23 th , 2020) [1], COVID-19 is a global public health emergency. The World Health Organization (WHO) has issued recommendations to implement social distancing measures for the general public as well as quarantining procedures, for people infected with the COVID-19.
Quarantine and social distancing measures had already been enforced during earlier pandemics, such as the 2003 outbreak of SARS and the 2014 outbreak of Ebola [2]. Studies on the effects of these measures have reported increased symptoms of anxiety, post-traumatic-stress, and depressive disorders, as well as a 30% increase in suicide rates in populations impacted by these measures [3,4]. These ndings are being replicated during the Covid-19 pandemic with multiple studies reporting an increased prevalence of depression and anxiety [e.g. 5, [6][7][8][9].
Notably psychiatric disorders result in a considerable burden of disease, accounting for 6.7% of overall disability-adjusted life years [10] and being attributable to 14.3% of death worldwide [11]. Despite the high burden of psychiatric disorders, there is a severe gap between people in need and people receiving mental health care [12]. This general treatment gap is especially severe in low-and middle-income countries, where 76% to 85% of people with mental disorders do not receive any treatment [13]. The latest WHO "mental health Atlas" indicates that only 95,6 out of 100 000 depressed cases worldwide receive any professional mental health care, whereas the treatment prevalence in high-income countries is 16-times higher compared to low-income countries [14]. Although there is no current global data available, the treatment gap is assumed to be much higher during or after the Covid-19 pandemic. Access to general mental health care might be restricted for several reasons, including supply priorities that being focused on Covid-19 infections, medication shortages, prohibition of face-to-face psychotherapeutic sessions of psychological treatment, closing of inpatient facilities to mention only some reasons.
To mitigate the negative mental health consequences of pandemics, evidence suggests that policymakers should ensure quarantine measures to be as short as possible, to provide adequate general supplies for basic needs, give people as much information as possible and strengthen social support and communication among people affected by the pandemic [2]. A recently published position paper on research priorities for mental health science regarding COVID-19 [15] demands the interdisciplinary development of novel interventions to protect mental wellbeing by mechanistically based approaches to strengthen altruism and prosocial behavior. Among others, physical activity (PA) interventions are highlighted as a promising approach. PA is de ned as any bodily movement produced by skeletal muscles that results in energy expenditure and exercise is de ned as PA, that is planned, structured, and repetitive, with the primary aim to improve or maintain physical tness [16]. International PA guidelines recommend 150 minutes of moderate or 75 minutes of vigorous intensity PA per week for optimal physical and mental health bene ts [17]. Indeed, in pre-pandemic times PA has been identi ed as a protective factor against incident depression [18] and anxiety [19]. However decreased levels of PA were observed in the general population in multiple countries during the pandemic [e.g. 20,21,9]. This rapid systematic review aims to outline current evidence regarding the associations of PA and exercise with depression and anxiety during the Covid-19 pandemic.

Methods
In this rapid review, we sought for observational studies examining the associations of PA and depression and anxiety during the COVID-19 pandemic. inclusion criteria were: 1) Observational studies in any population, including cross-sectional and longitudinal designs. Longitudinal studies could be either prospective or retrospective; 2) Studies have tested the association, of PA with depression or anxiety, using linear or logistic regressions; 3) depression and anxiety were assessed using validated screening or diagnostic tools. We excluded opinion pieces, systematic reviews, and studies addressing other viruses.
We searched the electronic databases PubMed, EMBASE, Sportdiscus, and Web of Science using the following strategy: (physical activity OR exercise OR sport) AND (coronavirus OR sars-cov-2 OR COVID* OR severe acute respiratory syndrome OR pandemic) AND (depression OR anxiety OR mental health). Preprints were searched in MedRxiv, SportRxiv, and Scielo preprints using the following strategy: "(physical activity OR exercise) AND (coronavirus OR sars-cov-2 OR COVID* OR severe acute respiratory syndrome OR pandemic)".
Additional hand searches were performed on COVID-19 platform on ResearchGate and Google scholar. Searches were made by an experienced reviewer (FS) on 29th July, 2020. Study selection was conducted in three steps: 1) duplicates removal; 2) screening at the title and abstract level; and 3) assessment based on full-text. The selection was made by one reviewer (FS). Data extraction of selected studies was then performed by three researchers (FS, BS, SW). Data extracted were: Author and year, country of the included sample, study design, sample size, age group of the sample included, when possible, mean or range of age sample, % of women, instrument/question used to assess PA levels, instruments used to assess depression and anxiety, publication type and statistical outcomes (regression standardized beta coe cients and odd's ratios). If they were indicated in the report, fully adjusted coe cients and odd's ratios were extracted. As studies included in this review used very heterogeneous statistical approaches, a meta-analysis could not be conducted. Instead, we summarized the evidence and presented effect sizes (betas and odds ratios (OR)) with con dence intervals and indicated signi cant associations between PA and depression or anxiety, separately (see table 2). In case the study just reported the unstandardized betas, we requested the standardized betas by email. If standardized effects could not be obtained, unstandardized effects were presented and indicated. The risk of bias of individual studies was assessed using the National Institutes of Health (NIH) study quality assessment tool for observational cohort and cross-sectional studies [22]. The NIH tool assessment is composed by 14 questions the risk of potential selection bias, information bias measurement bias or confounding bias. There are 3 options (yes, no, other) for each question. Each "no" or "other" is suggestive of the presence of some risk of bias. Questions #6 (exposure prior outcome), #7 (su cient time to see an effect), #10 (repeated exposure assessment), and #13 (follow-up rate) were disregarded for cross-sectional studies. Due to the self-reported nature of the assessments, question #12 (blinding of outcome assessors) was also disregarded for all studies.

Results
Searches on PubMed, EMBASE, Sportdiscus, and Web of science resulted in 592 potentially relevant studies. Preprint databases identi ed additional 572 potentially relevant studies. A ow-chart of the selection process is provided in gure 1. Of the identi ed studies, 21 studies meet the criteria [9, 23-34, 5, 35-37, 6, 38-40]. Four studies had a prospective longitudinal design [26,29,39, 36], one was a cross-sectional study with a retrospective measure of the exposure factor (henceforth treated as retrospective) [5], and 16 were crosssectional studies [9, 23-25, 27, 29- Data form a total of 42,293 (median = 68% of women) participants were included. Only one study was exclusively composed by older adults (over 50 years), 4 were in children, adolescents, or young adults, while 13 studies were in adults (over 18). Only 7 studies used validated measures to assess PA levels. A wide range of scales to measure depression or anxiety were used, the most used scales being the Beck Depression and Anxiety inventory and the DASS-21. Most studies (n=14) were not per-reviewed (pre-prints).
A summary of studies is provided in Table 1.
Results are summarized and presented in table 2. Out of 9 studies reporting analyses on the association between the overall volume of PA and depression, 6 studies showed that more PA is signi cantly associated with less depression symptoms [23,25,32,33,35,37], and 3 out of 8 studies investigating the association between the overall volume of PA and anxiety symptoms showed that more PA is signi cantly associated with less anxiety symptoms [25,32,35]. 3 out of 5 studies reported higher frequencies of PA to be signi cantly associated with less depression [27,29,36] and 2 out of 4 studies to be signi cantly associated with less anxiety [27,29]. One study showed that vigorous but not moderate PA is signi cantly associated with less depression and anxiety symptoms [6] and another study indicated that light and vigorous PA is signi cantly correlated with less depression, but moderate intensity was not [23]. Out of 5 studies assessing an association between regular and guideline-consistent PA less depression and anxiety symptoms, two studies demonstrate that regular PA (compared to not regular) is signi cantly associated with less depression and anxiety symptoms [27,30] and 1 study demonstrated that guideline conforming moderate to vigorous PA is associated with lower odds of depression and anxiety [6]. 5 out of 6 studies showed that a decreasing PA during the pandemic was signi cantly associated with more depression symptoms [9,27,5,38, 40] and 3 out of 6 studies showed that a decrease in PA was signi cantly associated with more anxiety symptoms [9,27,38]. 1 study reported that an increase in PA was associated with less depressive symptoms The risk of bias of individual studies is presented in table 2. All studies clearly de ned their research questions and used valid tools to assess main outcomes. Among the cross-sectional studies, 11 (68.75%) studies did not report the participation rate or included less than 50% of eligible participants, and 13 (81.25%) did not use valid tools to assess the exposure measure. A total of 3 out of 5 (60%) longitudinal studies are in risk of bias in the evaluating the de nition of the study population, the participation rate, the validity of the exposure measure and in the retention of the sample.

Discussion
The present study is, to the best of our knowledge, the rst study to summarize the evidence on the associations of PA with depression and anxiety during the COVID-19 pandemic. The majority of studies included in the present review showed that those who performed PA on a regular basis with higher volume and frequency and kept the PA routines stable, showed less symptoms of depression and anxiety. There was consistent evidence that those who could not keep their PA routine stable during the pandemic showed more depression and anxiety symptoms [9,27,5,[38][39][40]. However, the association was more consistent regarding depressive compared to anxiety symptoms. Those reporting a higher total time spent in moderate to vigorous PA had 12% to 32% lower chances of presenting depressive symptoms and 15% to 34% of presenting anxiety. These ndings are in line with results of recent meta-analyses showing that those with higher PA levels were 17% less likely of developing depression [18] and 26% less likely to develop anxiety [19], independently of the the COVID-19 pandemic.
Indeed, the found reduction of PA behavior during COVID-19 speci c conditions is highly expected. For example, due to social distancing, exercising in a group setting was limited or completely prohibited. However high social support is associated with more engagement in PA [42]. Indeed, social support was one of the strongest factors associated with adherence to PA in effective exercise interventions [43]. Furthermore, the COVID-19 pandemic impaired opportunities to be physically active due to the closure of sports clubs, gyms, or common indoor and outdoor places for PA. While some people were still allowed to do exercises like jogging on the streets, others were not [44]. In general, a lack of sporting opportunities seems to be associated with reduced PA [45]. Further negative consequences of the pandemic such as nancial insecurities might have caused stress in individuals and stress, in turn, may differentially impact individuals' level of PA. Whereas habitually active individuals might even increase their level of PA, those who had not yet integrated exercise as a part of daily life, reduce their level of PA [46]. Thus, habitually active individuals might have built PA-related health competence and learned to utilize PA as a strategy to cope with negative feelings, such as stress, that may arise with sudden adaptions [47,48]. Therefore, in order to prevent an increase in psychiatric disorders during the current or further pandemics, factors that facilitate the integration of PA into daily life routines, such as motivational and volitional skills, need to be identi ed and encouraged [49,50]. One of the best ways to promote motivation and volition is the application of behavior change techniques (BCTs) [51]. During the COVID-19 pandemic, some BCTs appear to be particularly important for the maintenance of regular PA. For instance, the knowledge about the bene ts of PA on symptoms that accompany lock-down procedures, such as lowered mood or anxiety [2][3][4]. Furthermore, individuals need the strong ability of coping planning to anticipate barriers that could discourage them to engage in PA (e.g., closed facilities) and nd strategies to overcome them (e.g., engage in home training).
A web-based tool, e.g. a smartphone application could be a low-threshold and cost-effective option to train, supervise, apply, and adopt such BCTs, especially in terms of COVID-19. First empirical evidence showed preliminary e cacy of apps in promoting PA. Users of such apps are more likely to meet recommendations on PA than non-users [52][53][54]. Furthermore, a meta-analysis showed that Internet-delivered interventions, which are able to use different BCTs, were effective in increasing PA [55]. A major advantage of such webbased tools is the possibility to overcome some of the COVID-19 speci c barriers. For instance, it is possible to become physically active online with friends or a virtual community, which might work against the lack of social support. In addition, limited sporting opportunities may be expanded through tness technology and the provision of structured programs, as they can be used both indoors (e.g. through tness videos) and outdoors (e.g. through running apps) and therefore be adapted to the speci c situation.

Limitations
Most of the studies included in this review used cross-sectional research designs. A causal nature of these associations therefore remains unclear. There are notably differences in effect sizes which point at a high heterogeneity of the effects. Several studies further showed methodological shortcomings, e.g. not reporting the participation rate, including less than 50% of eligible participants, no validated tools to assess PA and failure to report standardized coe cients. Heterogeneity in research designs and statistical analyses hindered meta-analytic approaches, which would have provided a more sophisticated overall effect estimate. Finally, several included studies were published as preprints and are currently in review processes for nal publications. It is therefore planned to update this review in the future.

Conclusions, Future Research Directions And Implications
This rapid review shows promising evidence that higher volume and frequency of PA and the keeping of regular PA habits during the Covid-19 pandemic is associated with less symptoms of depression and anxiety. For instance, those reporting a higher total time spent in moderate to vigorous PA had 12% to 32% lower chances of presenting depressive symptoms and 15% to 34% of presenting anxiety. Thus, the promotion of PA habits and routines might be a cost-effective and comprehensive worldwide applicable strategy to overcome the severe gap between people in need and people receiving mental health care, especially in low-income countries with even non-existing mental health supplies. Particularly, web-based technologies, could be an easily accessible way to increase motivation and volition for PA and maintain daily PA routines even under pandemic-speci c barriers. However only very few apps or websites have been tested in RCTs with high methodological standards [56]. Thus, there is a clear need for more systematic research for effectively and safely usable apps or web-based programs to prevent psychiatric disorders through PA.

Declarations
Funding. No funding has been received for the conduct of this review and preparation of this manuscript.
Con ict of Interest. All authors declare that they have no con ict of interest.
Availability of data and material. Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.
Author Contributions. SW and FS devised the project and the main conceptual ideas. FS 32. Jacob L, Tully MA, Barnett Y, Lopez-Sanchez GF, Butler L, Schuch F et al.    PRISMA ow-chart of the screening and selection of studies.