key: cord-0779503-z4pi0rf6
authors: Ng, Tommy KY; Kwok, Chris KC; Ngan, Gabriel YK; Wong, Horace KH; Zoubi, Fadi Al; Tomkins-Lane, Christy; Yau, Suk Ki; Samartzis, Dino; Pinto, Sabina M; Fu, Siu-Ngor; Li, Heng; Wong, Arnold YL
title: Differential impacts of COVID-19 pandemic on physical activity involvements and exercise habits in people with and without chronic diseases: A systematic review and meta-analysis
date: 2022-04-10
journal: Arch Phys Med Rehabil
DOI: 10.1016/j.apmr.2022.03.011
sha: c8cdd50c74237ad53187cbbc9484386e02f49d05
doc_id: 779503
cord_uid: z4pi0rf6

Objective To conduct a systematic review and meta-analysis to summarize evidence regarding differential changes in PA involvements and exercise habits in people with/without chronic diseases during the COVID-19 outbreak. Data Sources MEDLINE, Embase, SPORTDiscus, CINAHL, PsycINFO, Cochrane Library, PEDro were searched from November 2019 to May 2021. Study selection Two reviewers independently screened cross-sectional and longitudinal studies that investigated changes in PA-related outcomes in people with/without chronic diseases during the pandemic. Data Extraction PA-related outcomes and sedentary time were extracted from the included studies. Relevant risk of bias were assessed. Meta-analyses were conducted for each PA-related outcome, if applicable. Quality of evidence of each PA-related outcome was evaluated by GRADE. Data Synthesis Of 1,226 identified citations, 36 articles (28 with and 8 without chronic diseases) with 800,256 participants were included. Moderate evidence from wearable sensors supported a significant reduction in pooled estimates of step count (standardized mean differences (SMD)=-2.79, p<0.01). Very limited-to-limited evidence substantiated significant decreases in self-reported PA-related outcomes and significant increases in sedentary behaviours among people with/without chronic diseases. Specifically, pooled estimates of metabolic equivalent-minute per week (SMD=-0.16, p=0.02), and PA duration (SMD=-0.07, p<0.01) were significantly decreased, while sedentary time (SMD=0.09, p=0.04) showed significant increases in the general population (small- to large-effects). Very limited evidence suggested no significant PA changes among people in a country without lockdown. Conclusion During the pandemic, objective and self-reported assessments showed significant reductions in PA in people with/ without chronic diseases globally. This mainly occurred in countries with lockdowns. Although many countries have adopted the “live with the coronavirus” policy, authorities should implement population-based strategies to revert the potential lockdown-related long-term deleterious impacts on people's health.

Conclusion During the pandemic, objective and self-reported assessments showed significant reductions in PA in people with/ without chronic diseases globally. This mainly occurred in countries with lockdowns. Although many countries have adopted the "live with the coronavirus" policy, authorities should implement population-based strategies to revert the potential lockdown-related long-term deleterious impacts on people's health. [1] , the disease has rapidly plagued the globe, inflicting unprecedented negative impacts on the global socioeconomic and healthcare systems. As of September 2021, 221 countries had been struck by COVID-19, resulting in more than 248 million infected cases and over 5 million deaths [2] .

Countries with lower national income and suboptimal medical services are more vulnerable to the negative consequences of the COVID-19 pandemic including changes in health behaviors such as physical activity (PA) participation [3] .

Given the escalating number of confirmed COVID-19 cases and overburdened healthcare systems, the World Health Organization (WHO) declared COVID-19 outbreak as a pandemic [2] .

Most governments implemented stringent measures including travel ban, nationwide quarantine, social distancing and lockdowns to suppress the outbreak [2] . Approximately four billion people were confined to their homes, while more than 90 countries or regions had imposed lockdowns by April 2020 [4] .

Prolonged lockdowns have a negative impact on people's physical, psychological, and social health . Reduced PA or exercise participation, alongside increased sedentary behaviors, could compromise physical and mental health of many individuals [40, 41] . The WHO recommends adults to perform 150-300 minutes of moderate-intensity or 75-150 minutes of vigorous-intensity aerobic PA every week [42] . People with chronic diseases, who are recommended to do regular exercises to delay their disease progression [43, 44] , may be more susceptible to the adverse effect of reduced PA. Reduced PA in these patients not only may affect their disease progression, but also increases their risk of developing additional inactivityrelated diseases. Importantly, regular moderate-to-vigorous PA (MVPA) can boost immunity against community-acquired infectious diseases, and increase potency of vaccination. Although an earlier systematic review has summarized the preliminary effects of the COVID-19 pandemic lockdown on PA changes of the public [45] , it was limited by small representative samples, and lack of assessments of evidence or meta-analyses regarding the impacts of the pandemic on various PA-related outcomes among people with/without chronic diseases in countries with or without lockdowns. Since PA changes measured by wearable sensors may differ from those collected from self-reported PA questionnaires, comprehensive meta-analyses of various PArelated outcomes can better inform policy makers in developing tailored strategies to revert the adverse effects of physical inactivity in vulnerable subgroups during and after the pandemic. The current systematic review and meta-analysis addressed this gap to summarize the evidence regarding impacts of the COVID-19 pandemic on PA-related outcomes in the people with/without chronic diseases who did not contract COVID-19.

The study protocol was registered on PROSPERO (CRD42021234936). The Preferred Reporting Items of Systematic Reviews and Meta-analyses (PRISMA) guidelines [46] were adopted to report this review.

A systematic literature search was conducted on seven databases (MEDLINE, EMBASE, SPORTDiscus, CINAHL, PsycINFO, Cochrane Libraries, and PEDro) to identify articles published between 1 November 2019 and 31 May 2021 without any language restrictions. We searched these databases using a combination of two sets of keywords: ['COVID' OR 'cov*' OR 'corona*' OR 'severe acute respiratory syndrome coronavirus 2' OR 'SARS*'] AND ['physical activit*' OR 'activity level' OR 'exercise habit*' OR 'exercise routine*' OR 'lifestyle'] (Appendix 1). Additional relevant articles were searched from the reference lists of the included studies. Forward citation tracking was conducted using Scopus. The corresponding authors of the included articles were contacted by emails to identify any additional relevant publications.

Cross-sectional and longitudinal studies that investigated PA-related outcomes during the COVID-19 pandemic were included. Articles were excluded if the participants were actively or previously infected with COVID-19. Commentaries, letters to editors, reviews, conference proceedings, and qualitative studies were also excluded.

All citations identified from database searches were exported to EndNote X9 (Clarivate, Pennsylvania, the USA). After removing duplicates, two reviewers (K.Y.N, and K.H.W.) independently screened the titles and abstracts following the selection criteria. They piloted on 100 abstracts to align discrepancy. They then independently screened the remaining references.

Abstracts deemed relevant were included for full-text screening. The process was repeated for the full-text screening. Reviewers met to reach a consensus about the eligible articles. If disagreements persisted, a third reviewer (K.C.K.) arbitrated the disagreements. The interrater agreement was calculated using Cohen's kappa [47] .

Two independent reviewers (K.Y.N., and K.H.W) used a standardized form to extract data related to authors, year of publication, study location, study design, data collection methods, response and attrition rate, participants' demographics, definitions of PA and sedentary behaviors, changes in PA-related outcomes, and the corresponding statistics.

Two independent reviewers (K.Y.N., and K.H.W) used two separate tools to assess the quality of the included studies. Specifically, the methodological quality of cross-sectional studies was assessed by the 20-item Appraisal tool for Cross-Sectional Studies (AXIS) [48] . The tool only provides descriptive assessments without numeric scores. It is flexible for researchers to use it based on their priorities. Therefore, we rearranged the items into six domains: objectives and design, study participation, handling of non-respondents, outcome measures, statistical analysis, and reporting [49] . Similar to our previous reviews [49, 50] , each domain was ranked as low, moderate, or high based on the criteria listed in Appendix 2. The Quality In Prognosis Studies (QUIPS) tool was used to assess the methodological quality of longitudinal studies [51, 52].

QUIPS tool comprises six domains: study participation, study attrition, prognostic factor measurement, outcome measurement, study confounding, and statistical analysis and reporting [52] . Each domain was rated as low, moderate or high [52] . From the quality of each domain, the overall methodological quality was graded as low, moderate, or high [52] (Appendix 2).

PA-related outcomes extracted from the included studies were categorized into two pairs of subgroups: (1) the people with/without chronic diseases; and (2) countries with and without lockdown. If two or more included studies reported changes in a particular PA-related outcome during the pandemic in a given subgroup, the respective standardized mean differences (SMD)

were pooled for a meta-analysis using a random effects model. All meta-analyses were performed using the Comprehensive Meta-analysis Version 3.3 software (Biostat, Englewood, NJ, US). Statistical heterogeneity of the included studies was assessed by I² statistics, and classified as low (I²<40%), moderate (I²=40% to 59%), substantial (I²=60% to 74%) and considerable (I²>75%) heterogeneity [53] . The potential sources of heterogeneity of each metaanalysis were explained if substantial or considerable statistical heterogeneity was observed [53] .

The quality of evidence of each PA-related outcome was rated by the Grading of Recommendations Assessment, Development and Evaluation (GRADE) [54] . The GRADE framework consists of seven domains. Five of which could downgrade the quality of evidence regarding the estimated effect size, while the other two domains could increase the confidence in the estimated effect size. The synthesized data was ranked as very limited, limited, moderate, or high quality of evidence regarding how the true effect lays close to the estimated effect (Appendix 2) [55] .

The literature search identified 1,226 publications, while 13 records were identified through other sources (Figure 1 ). After removing 103 duplicates, 1,136 studies were eligible for the title and abstract screening. Of the 95 screened full-text articles, 36 articles were included. 56] Fifty-nine full-text articles were excluded due to: not investigating PA changes (n=40); inclusions of confirmed COVID-19 cases (n=6); or commentaries, letters or reviews (n=13). Our Kappa coefficients showed substantial (κ=0.75) and almost perfect (κ =0.93) agreements between the two reviewers (K.Y.N, and K.H.W.) during the title/abstract screening, and full-text screening, respectively [47] .

The 36 included studies recruited 800,256 participants from Asia, Africa, Australia, Europe, and North and South America. Thirty-five studies were conducted in countries/regions with lockdown , while a Swedish study was conducted without lockdown [56] . The participants' mean ages ranged from 7.3 to 74.0 years. Table 1 summarizes participants' demographics in the included studies. Twenty-three studies adopted a cross-sectional design [6, 7, 9-12, 14, 15, 17, 18, 21-25, 27, 28, 30-32, 37, 38, 56] , while 13 adopted a retrospective design [5, 8, 13, 16, 19, 20, 26, 29, [33] [34] [35] [36] 39] .

Twenty-eight studies investigated changes in PA in people without chronic diseases [6-8, 10-22, 24-26, 28, 30-32, 35-38, 56] . Notably, 20 of them focused on adults [6-8, 10, 12-15, 18, 22, 24-26, 28, 31, 32, 35, 36, 38, 56] , two on older people [21, 37] , four on students [11, 17, 19, 20] , and two on children and adolescents [16, 30] . Eight studies investigated changes in PA among people with chronic diseases [5, 9, 23, 27, 29, 33, 34, 39] . Specifically, four focused on patients with cardiovascular diseases [5, 9, 33 , 34], one on type 2 diabetes mellitus (T2DM) [29] one on musculoskeletal pain [27] , one on adults with obesity [23] , and one on children with chronic respiratory diseases [39] .

One study was rated as having low risk of bias [29] , 31 as moderate [6-26, 28, 32-39, 56] , and 4 as high [5, 27, 30, 31] . Of the 23 cross-sectional studies, the most common risks of bias were no sample size justification [6, 7, 9-12, 14, 15, 18, 21-25, 27, 30-32, 37, 38, 56] , and no description of non-responders' characteristics [6, 7, 9-12, 14, 15, 17, 18, 21-25, 27, 28, 30-32, 37, 38, 56] (Appendix 3). All the 13 retrospective studies did not provide information regarding the drop-out participants nor accounted for potential confounders (Appendix 4) [5, 8, 13, 16, 19, 20, 26, 29, [33] [34] [35] [36] 39] .

The included studies used diverse definitions of PA (Appendices 5 to 7, Table 2 ). Five studies used accelerometers/pedometers to quantify PA levels [13, 26, 29, 34, 35] . Eight studies used the International Physical Activity Questionnaire (IPAQ) to categorize PA into different levels [6, 7, 10, 18, 19, 24, 36, 37] . Fourteen studies had miscellaneous definitions of PA (e.g., regular exercise for different durations or different leisure time PA) [5, 8, 11-13, 16, 18, 21, 22, 28, 30, 31, 33, 56] , whereas nine studies did not clearly define PA [9, 14, 15, 17, 20, 23, 25, 27, 39] . For the 12 studies that investigated sedentary behaviors [9, 10, 16, 18, 19, 27, 29, 31, 33, 36, 38, 56] , one used accelerometers to record sedentary time [29] , four used self-reported screen time [9, 10, 18, 19] , and four used self-reported sitting/couch time [10, 16, 27, 36] . However, three studies 

The included studies investigating changes in PA of people without chronic diseases were conducted in countries with and without [56] lockdown. These changes in PA-related variables are summarized in Figure 2 , and Table 2 ).

Step counts (per day/week) Moderate evidence from four studies consistently showed significant decreases in step counts after the outbreak as measured by accelerometers, pedometers or a self-reported questionnaire [13, 26, 35, 38] . The meta-analysis showed a significant reduction in step counts with a large effect size (pooled SMD= -2.79; p<0.01, I²= 100%).

Very limited evidence from one Canadian study revealed a significant reduction (12.6%) in the duration of light PA as measured by accelerometers during the pandemic [13] . Likewise, very limited evidence from two studies suggested reduced durations of MVPA [13, 19] . Specifically, one study used accelerometers to detect significant decreases in the duration of MVPA (9.3%) after the outbreak [13] . Another study used IPAQ to reveal a significant reduction in time spent on MVPA after a lockdown [19] .

Very limited evidence substantiated decreases in self-reported weekly PA duration [14, 32, 37] .

However, because one included study did not present the relevant statistical data [32], it was excluded from the meta-analysis. The meta-analysis showed a significant reduction in PA duration per week with a small effect size (pooled SMD= -0.07; p< 0.01; I²= 0%) [14, 37] .

There was inconsistent evidence regarding the proportion of people reporting changes in PA levels during the outbreak. Five studies used customized questionnaires to evaluate changes in PA during lockdowns although PA was not defined [15, 17, 20, 21, 25] . They found that 20.7% to 61.5% of participants reported decreases; 13.6% to 53.2% reported no change; and 5.3% to 48 .6% reported increases in PA levels [15, 17, 20, 21, 25] . Similarly, three included studies used validated questionnaires to evaluate changes in PA levels during lockdowns [10, 22, 32] . They showed that 20.7% to 33.9% of participants reported decreases; 25.1% to 30.5% reported no change; and 35.7% to 49.1% reported increases in PA levels [10, 22, 32] . A Swedish study (without lockdown) used a customized questionnaire to reveal that 63.0% of participants reported no change in PA levels, and only 26.0% reported decreases in PA levels during the outbreak [56] .

There was very limited evidence supporting the adoption of an inactive lifestyle during lockdowns [28, 30, 31] . Although three studies revealed that more people were classified into the low PA category after the outbreak [28, 30, 31] , one of them reported approximately 30% increase in the number of people being categorized into "never performed PA" or "frequently performed PA" during the COVID-19 outbreak [28] .

Limited evidence from two included studies supported that people living in countries with lockdowns reported either no change or decreases in their exercise duration, while only 17.8% to 20.0% of people reported increases in their exercise duration [11, 18] . Conversely, up to 26% of participants reported increases in exercise in a country without lockdown [56] .

There was conflicting evidence regarding the proportion of people participating in regular exercises [8, 12] . One included study reported no significant changes in the proportion of participants involved in regular exercise training after the outbreak [12] , while another study revealed a 19% drop in the number of participants who exercised regularly [8] . However, both studies used unvalidated questionnaires.

Very limited evidence supported reduced estimated MET-minute per week as measured by IPAQ [7, 24, 36] . The PA reduction ranged from 23.8% to 69.8% [7, 24, 36] . The meta-analysis from three studies showed a significant reduction in MET-minute per week with a small effect size (pooled SMD= -0.16; p= 0.02; I²= 77%) [7, 24, 36] .

Very limited evidence from one study reported a 24.0% decrease in IPAQ scores although it was described in MET-minute [6] .

Very limited evidence suggested approximately 7.4% to 24.5% increases in sedentary time (defined by screen time, sitting, and sedentary activities) in the general public as measured by IPAQ [19, 36] or a self-developed questionnaire [38] . Our meta-analysis showed a significant increase in sedentary time with a small effect size (pooled SMD=0.09; p=0.04; I²=84%) [19, 36, 38 ].

There was limited evidence that a relatively larger proportion of participants reported increases in sedentary time in countries imposing lockdowns [10, 16, 18, 31] . Notably, approximately 41.0% to 68.3% of participants reported increases in their sedentary time 10, 16, 18. One study also found that the proportion of participants being classified as the "sedentary" category increased from 12.2% to 25.0% [31] . All these studies adopted validated questionnaires to quantify sedentary time [10, 16, 18, 31] . Conversely, very limited evidence suggested that Swedish (without lockdown) were less likely to adopt a sedentary lifestyle [56] . Nobably, only 26 .0% of Swedish participants reported increased sitting time, while 66.0% reported no change [56] .

Step counts (per day/week) Very limited evidence suggested significant decreases in step counts by 7.6% in patients with T2DM [29] . Similarly, there was very limited evidence that patients with cardiovascular diseases had approximately 16.4% reduction in step counts [34] .

Very limited evidence supported no significant change in the duration of MVPA in patients with T2DM as quantified by accelerometers [29] . Conversely, very limited evidence substantiated 25.0% increases in MVPA among patients with cardiovascular diseases as measured by a validated questionnaire [33].

There was very limited evidence that patients with heart failure displayed an average of 0.8 hours reduction in daily duration of PA as measured by cardiac implantable electronic devices [5] .

Very limited evidence suggested significant decreases in PA levels among approximately 41% of patients with congestive heart failure 9 or adults with obesity [23] .

Very limited evidence showed an increased number of participants classified as no activity in patients with musculoskeletal pain [27] or low PA categories in children with chronic respiratory diseases [39] . Specifically, one study showed a 82.5% increase in the number of participants being classified as "no PA" [27] , while another study revealed a 237.5% increase in the number of participants being classified as having less than 1-2 hour(s) of PA per day [39] .

Very limited evidence supported an increase in the frequency of continuous PA in patients with T2DM as recorded by an accelerometer [29] . Rowlands et al. revealed that the average frequency of 30-and 60-minute continuous PA in patients with T2DM significantly increased from 0.65 day/week to 1.0 day/week and from 0.24 day/week to 0.44 day/week, respectively [29] .

Very limited evidence supported a 3.0% increase in sedentary time in patients with T2DM as measured by accelerometers [29] and a 14.4% increase in patients with cardiovascular disease documented by a validated questionnaire [33] .

Very limited evidence suggested that 46% of participants with congestive heart failure [9] and 152% of participants with musculoskeletal pain [27] reported increases in sedentary time. people from performing PA [28] . This is attested by the fact that most respondents in a country without lockdown reported no change in their PA and sedentary behaviours [56] .

It is well-known that physical inactivity adversely affect physical and mental health, as well as quality of life [57] . Insufficient PA heightens the risk of developing non-communicable diseases, (e.g., 24%, 16%, and 42% increase in the risk of having coronary heart disease, cardiovascular accident and T2DM, respectively) (56). Since 23.3% and 27.5% of the global population had insufficient PA in 2010 and 2016, respectively [57] , the WHO implemented an action plan between 2018 and 2030 to counteract physical inactivity [58] , and to reduce the global physical Lockdown-related physical inactivity may increase the incidence of non-communicable diseases and related healthcare burdens [60] . It is well known that regular MVPA boosts immunity against infectious diseases. An average energy expenditure of 500 to 1,000 MET-minute per

week is associated with a lower risk of severe acute respiratory syndrome coronavirus 2 infection [61] . PA can also improve an individual's depression and mood by the augmented release of endorphins [62] . Thereby, health authorities should implement new strategies to promote active lifestyle (especially MVPA) during and after lockdowns. Since some people may be fear of going out even after lockdown lifted, governments should run proper campaigns and/or use mobile applications to promote indoor PA and exercises [58] among people who hesitate to exercise outdoor during or after lockdowns.

Most of the included studies reported decreases in PA among people with chronic diseases during the pandemic. Physical inactivity may have greater detrimental effects on people with chronic diseases. Increased physical inactivity in patients with chronic heart conditions could heighten their morbidity [63] and mortality rates [64] . A 30-minute reduction of daily PA in any given month over 4 years in patients after implanting cardioverter-defibrillators was associated with 48% increased hazard for death as compared to active patients in the same month [65] . Similarly, physical inactivity and suspended face-to-face physiotherapy treatments during lockdowns led to increased symptoms in patients with musculoskeletal pain as compared to the pre-pandemic period [27] . Increased frequency and intensity of pain in patients with osteoarthritis [66] or chronic low back pain [67] in turn may result in the adoption of a sedentary lifestyle. If this vicious cycle persists, these patients may experience other pain-related comorbidities (e.g., depression). Imperatively, decreased PA levels [8] and the suspension of routine medical care in children with chronic respiratory diseases may lead to weight gain and mental health issues during the pandemic [39].

While reduced PA during the COVID-19 pandemic is prevalent, some people with chronic diseases reported increased PA during lockdowns. One included study reported no significant change in the duration of MVPA, and even increases in the frequency of 30-and 60-minute exercise sessions among people with T2DM [29] . These findings might be attributed to the success of the British government in promoting exercise for health maintenance, and permitting outdoor exercises during lockdowns. Similarly, increased population-interest regarding the impacts of PA and screen time on health might inspire some patients with chronic heart diseases to increase their MVPA during the pandemic [68] . Physical activities may increase the levels of adiponectin that can dampen the proinflammatory pathway of T2DM [69] and reduce plaque formation in patients with heart diseases [69] . These results underscore the importance of proper public health policies/strategies to minimize the negative impact of lockdowns on PA.

During lockdowns, exercise formats have shifted from outdoors to indoors [70], and from onfield team sports to home-based individual exercises (e.g., yoga) [28] . Additionally, tele-exercise has gained its popularity. Some governments produced online exercise videos by physiotherapists to promote home-based training to the general public [71] . Similarly, nongovernment organizations (e.g., National Centre of Health, Physical Activity and Disability) launched different campaigns (e.g., #MoveInMay, online toolkits, or workout videos) via social media to engage and educate people to perform PA [72] . While some private companies (e.g., ParticipACTION) used website and/or mobile applications to provide users with exercise demonstration videos and guidelines, interactive team challenges, and rewarding schemes to counteract the lockdown-related physical inactivity [13] , other companies embedded a body positional tracking system in a mirror to provide individualized home exercise training [73] .

Although telerehabilitation/telemedicine may facilitate home-based disease management, older people or underprivileged individuals may have difficulty in using telehealth [27, 74] . Future studies should investigate the optimal strategies for delivering telerehabilitation/tele-exercise to older people, or people in low income countries.

Wearable devices (e.g., smart watches) allow objective PA measurements. All included studies [13, 26, 29, 34, 35] using wearable devices consistently showed reduced PA in people with/without chronic diseases during the pandemic, except for one study investigating patients with diabetes [29] . However, PA levels quantified by wearable devices rely on participants' compliance [13, 75] . Studies that used wearable devices to collect PA data might underestimate the negative impacts of the pandemic on PA because people using wearable sensors might be more health conscious and intended to monitor their PA levels to stay active [34].

Most included studies used self-reported questionnaires to estimate PA during the pandemic, which was common in epidemiological studies to investigate the prevalence of diseases or behaviours [76] . However, self-reported PA levels may be subjected to recall bias. Therefore, PA levels estimated by online questionnaires might not be related to those measured by accelerometers [77] . Further, while 30 included articles used different self-reported questionnaires to estimate PA [6-12, 14-25, 27, 28, 30-33, 36-39, 56 ], only 15 studies used selfreported PA questionnaires with reported reliability and validity [6, 7, 10, 16, 18, 19, 22, 24, 28, 30-33, 36, 37] . The estimated impacts of the COVID-19 pandemic on PA might have been more accurate had validated questionnaires been used. Therefore, an international consortium should be formed to determine a core set of PA questionnaires (e.g., the global physical activity questionnaire) [58] to allow comparisons of PA levels across studies in the future.

Given the sudden onset of the pandemic, most of the included studies adopted a cross-sectional design. Twenty-three included studies evaluated the current PA levels. The remaining studies used questionnaires (n = 7) or wearable/implanted devices to retrospectively evaluate the changes in PA levels before and after the outbreak. Cross-sectional studies are needed during a pandemic to cost-effectively garner relevant information from large samples [78] in order to inform policy-making, and help plan further prospective studies [79] . However, retrospective studies using wearable/implanted devices to quantify changes in PA-related variables following the pandemic are also important because they help reveal causation [78] .

The current review had several limitations. First, the searched keywords might not be comprehensive enough to capture all relevant articles although most key papers have been included. Second, since diverse PA-related variables were used in the included studies, some variables were only used in one included study, which prevented the conduction of meta-analysis.

Third, the included studies covered a range of chronic diseases, which prevented the conduction of meta-analysis of PA for each disease.

For limitations of the included studies, only one article reported PA-related variables in a country without lockdown, which limited its generalizability. Further, people's PA levels may change over time. Their PA levels showed the greatest decline at the beginning of lockdowns, but PA levels increased toward the end of lockdown [13, 26] . Since some studies did not report the time of data collection, people's PA levels at a given time point might not illustrate changes in PA levels throughout the pandemic. Moreover, many included studies recruited participants by convenience sampling or recruiting from a single center, which might affect the generalizability of results. Finally, although three PA-related variables were pooled for meta-analyses, they showed substantial heterogeneity (I²>60%). The high heterogeneity might be attributed to differences in the sampling methods, delivery mode of questionnaires, durations spent on completing questionnaires, and socioeconomic status. 

Origin, transmission, diagnosis and management of coronavirus disease 2019 (COVID-19)

A brief review of socio-economic and environmental impact of Covid-19

Half of humanity now on lockdown as 90 countries call for confinement 2020 03/04/2020

Remotely monitored inactivity due to COVID-19 lockdowns Potential hazard or heart failure patients

Effects of home confinement on mental health and lifestyle behaviours during the COVID-19 outbreak: insights from the ECLB-COVID19 multicentre study

Physical activity levels in physical education teachers before and during school suspension brought by the COVID-19 quarantine

Does Sars-Cov-2 threaten our dreams? Effect of quarantine on sleep quality and body mass index

Impact of lockdown on patients with congestive heart failure during the coronavirus disease 2019 pandemic

Self-weighing Practices and Associated Health Behaviors during COVID-19

Association of webbased physical education with mental health of college students in Wuhan during the COVID-19 outbreak: Cross-sectional survey study

Eating habits and lifestyle changes during COVID-19 lockdown: An Italian survey

Don't walk so close to me: physical distancing and adult physical activity in Canada

The effect of COVID-19 lockdown on lifestyle and mood in Croatian general population: a cross-sectional study

Perceived change in physical activity levels and mental health during COVID-19: Findings among adult twin pairs

Early effects of the COVID-19 pandemic on physical activity and sedentary behavior in children living in the U.S

Understanding knowledge and behaviors related to COVID−19 epidemic in italian undergraduate students: The epico study

Impact of the COVID-19 epidemic on lifestyle behaviors and their association with subjective well-being among the general population in Mainland China: cross-sectional study

Impact of COVID-19 lockdown on activity patterns and weight status among youths in China: the COVID-19 Impact on Lifestyle Change Survey (COINLICS)

The Quarantine 15:" perceived versus observed weight changes in college students in the wake of COVID-19

Changes in lifestyle, behaviors, and risk factors for cognitive impairment in older persons during the first wave of the coronavirus disease 2019 pandemic in Finland: results from the FINGER study

The impact of COVID-19 on physical activity behavior and well-being of canadians

Managing obesity in lockdown: survey of health behaviors and telemedicine

Gender differences in the impact on physical activity and lifestyle in Italy during the lockdown, due to the pandemic

Beyond lockdown: the potential side effects of the SARS-CoV-2 pandemic on public health

Wearable activity trackers for monitoring adherence to home confinement during the COVID-19 pandemic worldwide: data aggregation and analysis

The Aftermath of COVID-19 Lockdown on Daily Life Activities in Orthopaedic Patients

Musculoskeletal pain and teleworking in times of the COVID-19: Analysis of the impact on the workers at two Spanish universities

The impact of COVID-19 restrictions on accelerometer-assessed physical activity and sleep in individuals with type 2 diabetes

The PRISMA 2020 statement: an updated guideline for reporting systematic reviews

The measurement of observer agreement for categorical data

Development of a critical appraisal tool to assess the quality of cross-sectional studies (AXIS)

Prevalence, incidence, and factors associated with non-specific chronic low back pain in communitydwelling older adults aged 60 years and older: A systematic review and meta-analysis

Prevalence/Incidence of low back pain and associated risk factors among nursing and medical students: A systematic review and meta-Analysis

Evaluation of the quality of prognosis studies in systematic reviews

Cochrane Handbook for Systematic Reviews of Interventions version 6

Judging the quality of evidence in reviews of prognostic factor research: adapting the GRADE framework

Grading quality of evidence and strength of recommendations for diagnostic tests and strategies

Lifestyle habits and mental health in light of the two COVID-19 pandemic waves in Sweden, 2020

Worldwide trends in insufficient physical activity from 2001 to 2016: a pooled analysis of 358 population-based surveys with 1· 9 million participants

World Health Organization. Global action plan on physical activity 2018-2030: more active people for a healthier world: at-a-glance

Sanchis-Gomar F. Physical inactivity and cardiovascular disease at the time of coronavirus disease 2019 (COVID-19)

Physical activity and the risk of SARS-CoV-2 infection, severe COVID-19 illness and COVID-19 related mortality in South Korea: a nationwide cohort study

Effects of exercise and physical activity on depression

Patient activity decreases and mortality increases after the onset of persistent atrial fibrillation in patients with implantable cardioverter-defibrillators

Associations of leisure-time physical activity with cardiovascular mortality: A systematic review and meta-analysis of 44 prospective cohort studies

Patient activity and survival following implantable cardioverter-defibrillator implantation: the ALTITUDE activity study

Sedentary behavior and physical function: objective evidence from the Osteoarthritis Initiative

Longer sitting time and low physical activity are closely associated with chronic low back pain in population over 50 years of age: a cross-sectional study using the sixth Korea National Health and Nutrition Examination Survey

Is the COVID-19 lockdown nudging people to be more active: a big data analysis

Exercise Increases Adiponectin and Reduces Leptin Levels in Prediabetic and Diabetic Individuals: Systematic Review and Meta-Analysis of Randomized Controlled Trials

Elderly Health Service Depaetment of Health

National Center of Health Physical Activity and Disability. NCHPAD COVID Response. 2020. 73. The Hong Kong Trade Development Council RESEARCH. Navigating Covid-19: a smart fitness app

COVID-19 and its Impact on Back Pain

Walk this way: validity evidence of iphone health application step count in laboratory and free-living conditions

Information bias in health research: Definition, pitfalls, and adjustment methods

A simultaneous evaluation of 10 commonly used physical activity questionnaires

Advantages and disadvantages : Longitudinal vs. repeated crosssection surveys

Longitudinal studies