Background: On 11 March 2020, the World Health Organization declared coronavirus disease 2019 (COVID-19) to be a global pandemic (
1). To curb the spread of the disease, various regional and national governments advocated for social distancing measures with varying degrees of enforcement, ranging from unenforced recommendations to quarantine and business closures. Physical activity is an important determinant of health (
2) and is likely affected by social distancing measures. Daily step count, a proxy for physical activity, has been associated with all-cause mortality (
3). Beyond physical activity, regional step count trends may also provide a proxy for adherence to social distancing, providing real-time insights to inform public policy decisions. Because prolonged social distancing is considered to contain infection, it will be important to gauge adherence to these measures and their effect on other aspects of health, such as physical activity.
Objective: To examine worldwide changes in step count before and after the announcement of COVID-19 as a global pandemic.
Methods and Findings: In this descriptive study, we used deidentified, individual-level data from 19 January to 1 June 2020 that were collected from a convenience sample of users of the free, popular health and wellness smartphone app Argus (Azumio). Daily step counts were determined using smartphone accelerometers and Apple or Android algorithms for step counting (
4). User location was determined by smartphone IP address. The COVID-19 pandemic declaration date used was 11 March 2020. Regional mean steps were calculated daily, and percentage of change in steps was calculated daily as a percentage of the regional mean from 19 January to 11 March 2020. Displayed figure regions were selected to achieve half less-affected and half more-affected regions with regard to both COVID-19 and social distancing and greater than 1000 and 700 users at the country and city levels, respectively. This study was exempted by the University of California, San Francisco Institutional Review Board.
A total of 19 144 639 daily step count measurements were provided by 455 404 unique users from 187 unique countries during the study period; 92% of smartphones were Apple, and 8% were Android. Worldwide, within 10 days of the pandemic declaration, there was a 5.5% decrease in mean steps (287 steps), and within 30 days, there was a 27.3% decrease in mean steps (1432 steps). There was wide regional variation in average step count change and in the timing and rate of that change (Figures
1 and
2). For example, Italy declared a nationwide lockdown on 9 March 2020 and exhibited a 48.7% maximal decrease, whereas Sweden, to date, has primarily advocated for social distancing and limitations on gatherings and showed a 6.9% maximal decrease. Samples from countries such as Italy and Iran, which had earlier regional COVID-19 outbreaks, exhibited earlier step count decreases from their relative baselines. Samples from different countries varied widely in the number of days after pandemic declaration that a 15% step count decrease was seen: Italy (5 days), Spain (9 days), France (12 days), India (14 days), the United States (15 days), the United Kingdom (17 days), Australia (19 days), and Japan (24 days). Step count trends in samples from U.S. cities exhibited similarities, although there was wide international variability (
Figure 2).
Discussion: Step counts decreased worldwide in the period after COVID-19 was declared a global pandemic. Differences were seen between regions, likely reflecting regional variation in COVID-19 timing, regional enforcement, and behavior change. Countries that, to date, have had relatively low COVID-19 infection rates and have therefore not instituted lockdowns, such as South Korea, Taiwan, and Japan, have still exhibited decreases in overall step count. Within-region step count trends likely reflect a combination of changes to physical activity (for example, walking and exercising) and activities of daily living (for example, commuting and shopping) due to social distancing efforts. Assuming no regulatory changes that affect engaging in physical activity within a region, we suspect that sustained population-level trends over time may reflect changes to social distancing adherence (for example, many regions showed increases from their regional step count nadir before orders were lifted). Observed variation in step counts is also likely influenced by socioeconomic inequalities among regions and disparities in the ability to engage in or access to recreational physical activity within a region (
4).
Limitations of this study include sampling bias due to the reliance on smartphone and app ownership, measurement error from smartphone-measured step counts, variability in smartphone carry and use habits, no assessment of activity intensity, and inability to capture nonstepping exercise (
5). Our data set is a nonrepresentative convenience sample with a variable number of contributing daily users. It also lacks participant characteristics beyond IP address, limiting comparisons among regions.
Rapid worldwide step count decreases have been seen during the COVID-19 pandemic, with regional variability. Within-region step count trends may reflect social distancing measures and changes to social distancing adherence; however, more formal analytic studies are required. The effect of social distancing measures on overall physical activity, an important determinant of health, should be considered, particularly if prolonged social distancing is required.
Physical activity behaviors during the COVID-19-related confinement: general trend or individual adaptations?
The recent propagation of the COVID-19 virus led the World Health Organization (WHO) to officially declare an international global pandemic on March 11 2020, plunging most the world into an unseen period of general confinement. Progressively, governing bodies instituted a lockdown leading populations to remain at home, limiting their displacements and definitely changing their daily habits and routine. This unique situation definitely changed everyone’s overall physical activity and sedentary behaviors. Recently, Tison and colleagues proposed a great analysis of the population’s physical activity level during this period of confinement, based on the data collected through the smartphone app Argus (Azumio). According to their results and based on the record of about 19 144 639 daily step count measurements provided by 455 404 users of this app, representing 187 countries, there has been a clear and collective decrease in the daily step counts of individuals (reaching 27.3% after 30 days of confinement). The authors also point out an important variability between regions with Italy whose daily step count decrease reached 48.7% (ref). Although the authors point that these variations in step counts might be influenced by socioeconomic inequalities among regions and disparities in the ability to engage in or access to recreational physical activity within a region, their main results clearly suggest a global and worrying decline in the population’s daily physical activity (Tison et al., 2020).
Tison and colleagues provide us with a terrific international evaluation of the effect of the recent COVID-19-related confinement on our physical activity and must be commended for their work (Tison et al., 2020) that creates a great opportunity to better analysis, understand and discuss these human adaptations to such a situation that might not remain unique and isolated in the future. It seems however important to us that their results must be interpreted in light of their sample characteristics and specificities. First, we believe that the national sample used might not be representative of the overall situation. As the French National Observatory for Physical Activity and Sedentary Behaviors (ONAPS), we focused our interest on their French sample that was composed of 4,114 individuals living in Paris. While their results clearly underline an overall decline in physical activity (using daily step counts as an indicator), our recent national survey gathering information from about 28,000 individuals indicates that about 40% of French adults increased their overall physical activity level, particularly those who did not reach the recommendations before the confinement (ref Genin et al., 2020). Importantly, our results regarding sedentary time also suggest different changes during this period depending on the pre-confinement behaviors with 74% of adults who spent less than 6 hours seated per day before confinement who increased this sedentary time while a decrease was observed in 16% of those who initially spent more than 6 hours seated per day (ref Genin et al., 2020). This clearly suggests the necessity to consider the initial physical activity and sedentary profile of individuals that might have oriented their active behaviors during the lockdown.
We also collectively believe that the sample used in the discussed paper (Tison et al., 2020), that entirely represents a urban population living in Paris for France or other main cities in the other studied countries, limit the interpretation of the results. Indeed, while physical activity and sedentary behaviors have been shown different between individuals living in rural and urban areas (Sygit et al., 2019; Cohen et al., 2018), our recent national survey also point out different modifications of these behaviors during the lockdown depending on this living area (Genin etv al., 2020). While we acknowledge that this difference remains slight regarding physical activity in our study (with 39% of people living in rural area declaring a reduction of their PA level against about 45% in urban and peri-urban zones), it is definitely more pronounced when it comes to sedentary time with 37% versus 27% of individuals respectively living in urban versus rural areas who declare an increase of their sitting time.
Altogether, we believe that the results presented by Tison and colleagues are of main importance and describe a global international trend for a decline in physical activity level during the COVID-19-related confinement period. Other markers than steps should be used/added as our study has shown that new home-based physical activities have been adopted during the lockdown (home-trainer, muscle strengthening) that are not identify by pedometers or equivalent. This should be taken into consideration for future recommendations. Similarly, their analysis, according to us, should be deepen particularly considering the characteristics and profile of the individuals under study to better understand the effect of such a period on their active behaviors and then propose adapted post-confinement recommendation as well as potential recommendation in the likely case of a new lockdown in a more or less near future.
References.
Tison GH, Avram R, Kuhar P, Abreau S, Marcus GM, Pletcher MJ, Olgin JE. Worldwide Effect of COVID-19 on Physical Activity: A Descriptive StudyFREE. Annals of Internal Medicine, 2020. In press.
Genin PM, Lambert C, Larras B, Pereira B, Thivel D, Duclos M. How did the COVID-19 confinement period affect our on physical activity level and sedentary behaviors? Methodology and first results from the French National ONAPS Survey. Journal of Physical Activity and Health. 2020. In press.
Sygit KM, Sygit M, Wojtyła-Buciora P, Lubiniec O, Stelmach W, Krakowiak J. Physical activity as an important element in organizing and managing the lifestyle of populations in urban and rural environments. Ann Agric Environ Med. 2019 Mar 22;26(1):8-12.
Cohen JA, Greaney ML, Sabik NJ. Assessment of Dietary Patterns, Physical Activity and Obesity From a National Survey: Rural-urban Health Disparities in Older Adults. PLoS One 2018 Dec 5;13(12):e0208268.
Disclosures:
The authors have no conflict of interest to declare