The present paper is a cross-sectional study conducted in Denmark between April 3 and 15, 2020, during the COVID-19 shutdown. April 15 was chosen as the end date for the study because several restrictions from the Danish government were eased from that day.

On April 3, 2020, the study announced a press release, including a link to an online survey to recruit participants. The survey was also distributed through Facebook by posting a link to the survey on the university department's Facebook site and afterward sharing that link in different Facebook groups, on LinkedIn, and through the university's work mail. The Facebook post was boosted 1 time after 1 week to gain further reach. The study was advertised as a survey on people's movement behavior during the COVID-19 shutdown. It was explicitly mentioned that everyone, no matter how much or how little previous or current PA experience they had, could participate in the study. Inclusion criteria were people over the age of 15 years.

The online survey was created in SurveyXact based on a standardized Danish national PA survey repeated every fourth year (Pilgaard and Rask, 2016), which asks citizens about their PA behavior at work/school, during leisure time, and about active transportation. For this study, only questions about leisure PA were assessed. The survey started and ended with a range of background information on age, gender, education, ethnicity, job situation, and living situation (living in a small or large city, alone or with family). This was followed by a range of questions on the participants' general leisure PA behavior before and during the COVID-19 shutdown: “How much time did you usually spend on PA per week before the shutdown?” “How much time per week do you spend during the shutdown on PA (please write the number of minutes in total for 1 week)?” “How often were you physically active before the shutdown?” “How often are you physically active during the shutdown (5 or more times per week, 2–4 times per week, once a week, 2–3 times a month, rarely/never)?” It was clearly stated not to include active transportation, housework, or strenuous work. The participants were also asked, “What types of PA have you done on a regular basis during the last 12 months before the shutdown/during the shutdown?” The survey included examples of leisure PA (e.g., team ball games, outdoor activities, and gymnastics). Other questions were “Where have you primarily been physically active during the last 12 months before the shutdown/during the shutdown (e.g., sport associations, fitness/private center, at home, in public)?” “Who where you the most physically active with before and during the shutdown (alone, with friends, with colleagues, with family, with people I don't know)?” “Have you gained new PA habits during the shutdown?” Last, the participants were asked to identify perceived factors important for whether they are physically active during the shutdown to assess contextual factors that may explain a possible change in PA behavior. All questions came from the original national PA survey and were adapted to the specific COVID-19 shutdown.

The survey took, on average, 10 min to complete. The survey was tested on different age groups prior to its release, to ensure that the adaptions were understandable. The survey was found understandable by all age groups, but feedback from the test respondents resulted in the inclusion of additional response categories to the question on important factors for PA. The survey was only available in Danish.

Participants were given information about the study in the introduction to the survey. At the end of the survey, participants were able to read additional information on (1) how data were used and stored, (2) that they accepted that their data would be used for research purposes upon completion of the survey, and (3) that they were able to withdraw from the study at any time. The study and its data-management procedures were ethically approved by the Research & Innovation Organization of the University of Southern Denmark (10.975).

Data were downloaded from SurveyXact and uploaded in IBM SPSS Statistics version 25. Paired t-tests were performed to test for significant overall changes in weekly minutes of PA between, before, and during the shutdown as well as PA changes for several demographic groups (e.g., gender differences or differences in PA by education) between, before, and during the shutdown (Table 2). Pearson Chi-squared tests were performed for specific subgroup analyses on the four age groups: teens (15–18 years), young adults (19–29 years), adults (30–59 years), and older adults (60+ years) to test for independence at a 0.05 significance level between the four age groups before and during the shutdown (Table 3).

A total of 1,347 participants answered the full survey, and an additional 455 partly answered the survey. All 1,802 participants were included in the analysis. Participants' demographics are presented in Table 1. The oldest participant was 85 years old, and the youngest was 15 years old. Mainly females (75.8%) and Danish-speaking citizens (98.3%) participated in the survey. Two thirds of all participants (74.5%) lived in large towns and mainly worked at home during the shutdown (43.4%).

Table 2 displays demographic differences in mean minutes of PA per week before and during the shutdown. Overall mean minutes of PA dropped 16.1% (p < 0.001) from before to during the shutdown. Of the 215 participants who did not do PA before the shutdown, 51.2% started doing PA during the shutdown, and 7.1% of those doing PA before the shutdown stopped doing PA during the shutdown. Comparing participants' educational level, the highest drop in PA was found among those with the lowest education (−31.5%, p < 0.001), whereas the smallest drop was registered among those with the highest education (−1.5%) although it was not significant. Those living in a large city had the smallest decrease in PA (−13.5%, p < 0.001) compared with those living in a village or countryside (−30.9%).

Table 3 presents additional analyses made for the four specific age groups.

Teens experienced the largest decrease in mean minutes of PA (36.6%). At the same time, there was a substantial increase in individuals who rarely or never did PA during the shutdown (71.9%). Not doing PA for a longer time in this stage of life is alarming as this is the time when individuals begin to formulate their healthy habits for life (Telama, 2009). A study by Maugeri et al. (2020) finds similar results. They use an adapted version of the IPAQ questionnaire on a sample of 2,524 individuals and find a 34.4% decrease in mean METs min/week for younger participants (<21 years old) (Maugeri et al., 2020). The age range of the <21 age category is, however, not listed in the study, and it is, therefore, unknown whether they included 15–18-year-old adults. The teens were most often active in sports associations together with friends before the shutdown. When these facilities were closed, more than half did PA at home and alone. In fact, a third missed what they used to do, felt that they did not have access to the right facilities, and lacked someone with whom to do PA. For young people, social interactions with friends significantly influences PA behavior (Macdonald-Wallis et al., 2012), a factor that may have been negatively impacted by social distancing mandates as a result of COVID-19.

Younger adults had a decrease of 21.3% in mean minutes of PA per week from before to during the shutdown. This decrease may be due to their limited abilities to perform their usual PA routines. The younger adults were not able to do their usual PA in sport associations and fitness centers together with friends during the shutdown. Moreover, almost half of the younger adults reported that they missed what they used to do and felt that they did not have access to the right facilities. This suggests that restricting younger adults' abilities to perform their usual PA routines has negative consequences for their PA. In line with this, a previous Danish study found it important for younger adults to have fixed appointments and a fixed timetable when doing PA (Pilgaard and Rask, 2016). The study also shows that younger adults to a greater extent seek stability in their everyday life when they experience uncertainties. As such, younger adults might have felt a lack of certainty during the shutdown, making it impossible to do their usual PA routines.

Adults experienced the smallest decrease in PA (6.3% per week), and the number of adults who did PA 5 or more times per week almost doubled. Being forced to work from home or not work at all may have increased adults' flexibility and ability to do PA at different hours. Previously, adults were found to be the age group that, at most, prefer flexibility to be able to start or continue an active lifestyle (Pilgaard and Rask, 2016). Additionally, adults might not have been forced to make major adaptations to their usual PA habits during the shutdown. Adults primarily did outdoor PA alone before the shutdown, which they continued to do during the shutdown, combined with PA with their family. In fact, those living with their partner and children had the smallest decrease in minutes of PA (4.8%). This is somewhat supported by a study conducted in the UK during the shutdown. Rogers et al. (2020) find that families with school-aged children were associated with doing more intensive PA. Social support from family members and doing PA together is found to positively influence adults PA behavior (Zimmermann et al., 2008), and being “locked up” together may support doing PA in the family.

Globally, half of 80+-year-old and almost a third of 60–79-year-old adults do not reach the PA recommendations (Bauman et al., 2016) and represent the least active population group in Denmark (Jensen et al., 2018). This highlights the severity of the large decrease found in this study (24.9%) and supported by other studies (Maugeri et al., 2020; Yamada et al., 2020). This age group also experienced a 71.4% increase in participants rarely or never being physically active. This suggests that many older adults went from being regularly active during the week to almost no PA. As old age is associated with functional limitations and physical health problems (de Groot et al., 2004), maintaining good physical health is important. The observed decrease in PA may be due to several factors related to older adults. Older adults were the largest group that did not gain any new PA habits during the shutdown (32.1%). This suggests that older adults may have difficulties in adapting their PA habits as they rely more on social support to change and maintain their PA habits (Lindsay Smith et al., 2017). Furthermore, older adults primarily did PA together with friends in public or in sports associations before the shutdown, which, partly, was not possible during the shutdown. This is supported by previous research arguing that social interaction is one of the main motivators for older adults to leave their home and participate in activities (Schmidt et al., 2019) and that social isolation is associated with higher inactivity (Shankar et al., 2011).

This study, conducted in a unique situation during the COVID-19 outbreak, offers important insights into the research field of PA behavior. This study is the first of its kind conducted in Scandinavia and the first to provides contextual data on PA behavior for different age groups. A strength of the study is that the survey is based on a standardized Danish national survey on PA behavior repeated every fourth year (Pilgaard and Rask, 2016). Although the national survey has not been validated, it has been used numerous times in the Danish context throughout the last 20 years, making it highly relevant and comparable for this specific study focusing on PA in a Danish context. It is, therefore, argued that using an international validated questionnaire for this specific study would be less feasible although using a validated questionnaire may limit measurement error (Aggio et al., 2016).

A limitation is the recruitment method, which may have limited the representativeness of the study population as the study found an overrepresentation of highly educated females and adults (30–59 years). Nevertheless, 53% of the Danish population uses Facebook every day with relatively equal distributions of gender (Statistik, 2018), and almost all other similar studies have used online recruitment methods, such as social media (Cheval et al., 2020; López-Bueno et al., 2020; Maugeri et al., 2020; Rogers et al., 2020; Smith et al., 2020). These factors suggest that Facebook may be a reliable recruitment method. On the contrary, because relevance of the survey topic is shown to influence response rates (Groves et al., 2000), using social media platforms may have caused self-selection bias with an overrepresentation of respondents who may find PA important (Lavrakas, 2008). Additionally, the survey was only conducted in Danish and, therefore, excluded a range of ethnic minority groups living in Denmark.

The study builds on retrospective and self-reported data on PA, which is valuable in obtaining people's views of their own PA behavior, but can be threatened by recall bias (Althubaiti, 2016) with great variance in type, frequency, and duration of PA. Although survey questions were on routine PA behavior, which may have minimized the recall bias (Althubaiti, 2016), the length of the recall period was “within the last 12 months,” which may be difficult for some to recall. We suggest that future studies should obtain data on citizens' PA behavior collected prior to the COVID-19 shutdown to better compare and assess any consequences of the shutdown on people's PA behavior. More so, objective (device-based) data on PA rather than self-reported data on PA prior to, during, and after the COVID-19 shutdown would be preferable. Last, because this study included Danish citizens over 15 years old, caution should be made when interpreting the results of the low-educated participants. Participants in the age category 15–29 years may not have finished a higher education yet. As such, the low-educated group may not reflect a low socioeconomic group.