Background
COVID-19 is a highly contagious disease related to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Once the outbreak of the disease was categorized as a global health pandemic in March, 2020 [
1], extensive social distancing and isolation policies (eg, lockdowns) were employed by governments to reduce the strain on health services. Lockdowns have severely limited opportunities for physical activity (PA) [
2]. The health benefits of PA include reduced risk of cardiovascular disease, hypertension, diabetes, and some cancers [
3]. The closure of businesses, schools, and community facilities (eg, public parks) has encouraged sedentary behavior during the pandemic [
4]. High levels of sedentary behavior, typically assessed via daily sitting and screen time, are associated with greater risk for all-cause mortality, cardiovascular disease, type 2 diabetes, and some cancers [
5,
6].
It has been estimated that 970 million people worldwide suffer from mental health problems, such as depressive and anxiety disorders [
7]. Accordingly, the stressors associated with COVID-19 (eg, inability to see loved ones, job uncertainty) are likely to augment pre-existing psychological distress in modern society [
8]. This is exacerbated by limited opportunities for PA during lockdown, given that such behavior has a positive impact on mental health [
9].
Early findings indicated that lockdowns led to a decrease in PA coupled with an increase in sedentary behavior [
10,
11], albeit that researchers rarely take direct or objective measures of PA behavior (eg, daily step counts) [
12]. In the USA, Meyer et al. conducted a cross-sectional study in which they reported a 32% decrease in PA during the COVID-19 pandemic among adults who had been physically active [
13]. Moreover, the researchers detailed that the largest increases in sedentary behavior were associated with those who had been compelled to self-isolate.
Recent findings also illustrate detriments in mental health during the COVID-19 pandemic [
14,
15]. For example, Banna et al. administered an online survey to adults in Bangladesh and found that the prevalence of anxiety symptoms and depressive symptoms were 33.7% and 57.9%, respectively [
16]. Furthermore, 59.7% of participants reported mild-to-extremely severe levels of stress. Women reported higher anxiety, depression, and stress when compared to men [
16]. Similar findings were reported in China and the UK [
17,
18].
It is plausible that changes in PA, sedentary behavior, and mental health are not evenly distributed within populations and across nations [
8]. Groups of interest include women [
18] as well as younger (18–29 years) and older (≥ 60 years) adults [
19]. The policies of national governments varied considerably during the COVID-19 pandemic, resulting in disproportionate effects on subgroups of populations within individual nation states (eg, gig economy workers, people of color, and health professionals) [
19‐
21]. Nonetheless, there is limited research that incorporates data spanning multiple nations [
22].
The primary aim of the present study was to examine the effects of initial COVID-19 lockdowns across four Western nations with a focus on PA levels, sedentary behavior, and mental health. A secondary aim was to examine age as a moderator but in the case of mental health, we examined sex as a moderator [
18,
20]. In each of the analyses, nation membership was included for exploratory purposes.
We hypothesized that there would be reductions in PA dimensions and steps per day, from pre- to during lockdown, and that age would not moderate this trend (
H1). For sedentary behavior, we hypothesized increases from pre- to during lockdown and that age would not moderate this trend (
H2). For mental health, we hypothesized a decline from pre- to during lockdown, with a greater decline among women (ie, a significant lockdown × sex interaction;
H3). The knowledge derived from the present investigation might enable public health practitioners to develop interventions targeted toward the promotion of PA and mental health, coupled with a reduction in sedentary behavior, all of which have been identified as public health priorities [
19,
23,
24].
Discussion
The primary aim of the present study was to investigate effects of COVID-19 lockdown across four Western nations with specific reference to PA, sedentary behavior, and mental health. The planned and unplanned dimensions of PA, as well as recorded steps per day, showed a significant reduction from pre- to during lockdown (Figs.
1b and
2b). A lockdown × age group interaction emerged for unplanned PA, which can be attributed to a steep decline among young adults, and so
H1 is only partially accepted. It is notable that across the four nations, daily step count decreased by ~ 2000 steps.
An increase in sedentary behavior (ie, sitting and screen time) was observed from pre- to during lockdown. Lockdown × age group interactions emerged for both variables, which can be attributed to marked increases among young adults and older adults (Fig.
3b), and so
H2 is not accepted. A decrement in mental health was observed from pre- to during lockdown, represented by increases in GHQ-12 scores. Women exhibited a more pronounced decline in their mental health during lockdown when compared to men (ie, a lockdown × sex interaction emerged), providing support for
H3.
The present data indicate clearly that the phenomenon of lockdown in Western nations had negative consequences for planned and unplanned PA, as well as steps per day. Starting with planned PA, it is worth stressing that many exercise facilities were forced to close during lockdown. Consequently, some Western governments, such as those of the UK and Australia strongly encouraged regular daily exercise [
35,
36]. Moreover, technology-mediated exercise programs gained huge popularity during the course of the pandemic [
37]. A decline in planned PA emerged in the USA, UK, and Australia; this being least pronounced in the UK where strong encouragement was given for daily, socially distanced exercise [
35].
It is notable that no difference emerged from pre- to during lockdown in the France sample. However, the French started with a much lower base of planned PA, which remained stable when lockdown was imposed (Fig.
1a). The implications for the French are that their nation’s base levels of PA, particularly among young and lower middle-aged adults, are so low that there is a case for far stronger public messaging to promote PA. Nonetheless, France has among the lowest levels of obesity among European nations [
38], hinting that the population espouses relatively healthy eating habits [
39]. To further investigate this notion, we conducted an a posteriori analysis into BMI scores across nations, which showed that the France sample exhibited the lowest scores (
F3, 2520 = 58.70; ƞ
p2 = .07; France [
M = 23.90,
SD = 4.29] < UK [
M = 25.46,
SD = 4.96] < USA [
M = 26.63,
SD = 5.46] < Australia [
M = 27.64,
SD = 5.96]).
For unplanned PA, the expected decline during lockdown was moderated by nation and age-group membership in combination (ie, a three-way interaction; Fig.
1b). Generally, young adults engaged in less unplanned PA during lockdown than other age groups, with the most marked decline evident in Europe (see UK and France stacked dotplots in Fig.
1b). An interesting aspect of the unplanned PA findings was that no differences emerged from pre- to during lockdown in the France and Australia samples for older adults. In the UK sample, the smallest decline was evident in older adults. This trend suggests that, during lockdown, people of working age were far less able to take advantage of the health benefits associated with incidental activity, such as ascending a flight of stairs in an office building. Given the lack of opportunity for unplanned PA among people of working age during lockdown, they would be well advised to schedule additional PA (ie, planned PA) into their daily routine [
40].
Steps per day data indicated a reduction in steps equating to a daily energy expenditure of ~ 100 kcal across the four nations; roughly equivalent to a weight gain of 1.50 kg over lockdown, which has been reported elsewhere [
41]. Other recent pandemic-related studies from Italy and Spain have reported lockdown-related weight gains of ~ 2 kg [
41,
42]. The significant lockdown × age group interaction (Table
1) indicated a sharp decline in the daily step count of young adults from pre- to during lockdown (
Mdiff = 4185.67). This is just under half the number of steps that this age group would be recommended to take for optimal physical health [
43]. It is notable that France and Australia showed the steepest declines in steps per day (
Mdiff = 2586.26,
Mdiff = 2554.21, respectively). This is unsurprising given government messages in densely populated areas, such as Paris, requiring people not to leave their homes except for “essential purposes”. Twenty-three participants from the France sample were from the Paris area and another 215 were from large metropolitan areas.
As expected, when people are forced to stay inside their homes, lockdown resulted in a ~ 2-h increase in reported sitting time (Fig.
3a). This finding is consistent with other studies conducted in Western nations [
44,
45], albeit some studies report as much as a ~ 3-h increase [
46]. The significant lockdown × age group interaction showed that the increase in sitting time was greatest among young and lower middle-aged adults. This suggests that these groups might be more active in non-pandemic times (eg, through walking to work, moving around their workplace, and dancing at social events).
The lockdown × nation interaction showed the increase in sitting time in the USA and Australia to be more pronounced than that in the UK and France. Notably, there is much less of a culture of walking or cycling as a mode of transport in the USA and Australia—vast countries in which per capita car ownership is high [
47]—when compared to European countries such as the UK and France, which have a long-established culture of active travel [
48]. Interestingly, the statistical trends found in screen time mirror those found in sitting time (Fig.
3b). Accordingly, much of the time spent sitting entailed the use of screen-based technology.
Findings for sedentary behavior are among the most compelling when juxtaposed against the backdrop of dangerously high levels of sedentary behavior in the pre-pandemic era [
5]. The implications for cardiometabolic health are manifold and hint at the importance of encouraging the public to engage in regular bouts of PA during periods of lockdown. Failure to do so will increase the number of lives claimed by COVID-19 with many additional lives lost to obesity, type 2 diabetes, and heart disease [
4]. The most worrisome findings come in examining jointly the PA and sedentary behavior data for young adults, as this group appears to have been most adversely affected by lockdown [
10,
18].
The combined findings for PA and sedentary behavior paint a picture of large swaths of young adults in Western countries who give insufficient attention to their PA needs. It is worth highlighting that our findings mirror those of other recent multination studies [
22,
49]. The pandemic has served to shine a light on underlying attitudes toward PA that will need to be addressed in the post-pandemic era. It has become clear from a welter of epidemiological studies that regular PA can have a prophylactic effect in the face of COVID-19 (ie, in terms of the most severe symptoms), as well as many other infections, such as influenza and pneumonia [
50,
51]. Thus, siting health and PA as a centerpiece of school curricula—with appropriate theoretical and practical content—is a societal imperative [
52].
The GHQ-12 provides insight into common mental disorders (CMDs) such as anxiety, depression, and social dysfunction. Although usually less disabling than major psychiatric disorders, CMDs are more prevalent (eg, one in six adults in England) [
53], and are thus likely to have greater societal impact. We predicted that conditions of lockdown would elicit declines in mental health that would be greater among women, which is precisely what is shown in the lockdown × sex interaction (Fig.
4b), and replicated in many similar studies [eg, [
18,
54]]. The findings are also notable for variations in mental health across Western nations. Pre-lockdown, a small difference was evident between the USA and Australia, with poorer mental health scores in Australia (Fig.
4a). During lockdown, the USA sample reported the largest decline in mental health scores, with differences emerging between the USA and both the UK and France.
Notably, the French data show the greatest stability in mental health from pre- to during lockdown (Fig.
4a). This might be attributed, in part, to 40.6% of French respondents residing in rural locations. Although lockdowns are immediately apparent and perhaps anxiety provoking in urban environments, there are fewer noticeable changes in rural environments, where population density is much lower. The decrease in mental health in the USA sample is of particular concern (
Mdiff = 4.43); perhaps the uncertainties associated with an impending presidential election coupled with the lack of an economic safety net for large segments of the population, contributed to this finding [
14]. It is vital that further work is conducted into the association between population density and mental health; times of crisis such as war, famine, and now a pandemic, bring the mental well-being of urban populations into sharp focus [
55].
The decline in women’s mental health during lockdown, regardless of the nation in which they resided, is also worrisome (Fig.
4b). There was little that state governments could do to mitigate against the competing demands of full-time work, home schooling, and domestic responsibilities that many women faced [
56]. It should be added that the demands of looking after children or elderly relatives may have prevented many women from engaging in exercise activities, which are known to contribute positively to mental health [
9]. A clear implication is that state governments should consider women-friendly policies pertaining to flexible working, childcare, and mental health helplines in any future pandemic [
18,
57].
The present study employed an online survey method to collect data across four Western nations. Although the way in which the data were collected was standardized and entailed adjustments to render each survey culturally specific, it should be noted that a self-selection bias does pervade scientific work of this nature. In examining our demographic data (Additional file
1: Table S1), it is clear that (a) relatively few men chose to complete the survey (23.5%), (b) it attracted relatively few responses from those living in rural areas (34.2%), with the notable exception of French respondents (40.6%), and (c) it largely failed to reach individuals representing lower socio-economic groups (4.7%). The self-selection bias should thus be borne in mind when attempting to generalize the present findings to the populations that were sampled, as well as to the populations of other Western nations. A further limitation pertains to the retrospective recall of planned/unplanned PA, sedentary behavior, and mental health in relation to the period prior to lockdown. Future online studies of this nature might use incentives for the hard-to-reach contingents of the population, as well as adopt a longitudinal approach to circumvent the need for retrospective recall.
Participant incentives and a longitudinal approach were not possible in the present study, as the research team responded nimbly to the initial spate of national lockdowns and the circumstances for a ‘natural experiment’ [
26]. In the limited time window open to the research team, we aimed to collect as much data as we could, rather than be guided by an a priori power analysis. Given the wealth of data now published [
10,
11,
13], investigators of future lockdowns will have effect sizes from several nations to inform their estimations of sample size. We put ethical approval applications through our respective institutions immediately after national lockdowns were declared, without having time to apply for funding that would have provided incentives for participants, and without knowing that there would be multiple lockdowns in the months ahead. At the time that our study was initiated, the general consensus was that lockdowns would be a relatively short-lived phenomenon [
58]. A further limitation concerns the low alpha estimates for the three-item unplanned PA scale (0.52 pre-lockdown and 0.64 during lockdown).
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.