Design and procedure
This study is a survey with the inclusion criteria being 65 years or older and living at home in Norway. The study was approved by the Central Regional Committee for Medical and Health Research Ethics (REC Central—10418/REK midt). Participants were given written information about the project and gave an active consent by checking of a box (“I have read the information and consent to participate”) before they obtained access to the questionnaire. If a participant checked the box “I do not wish to participate”, no other questions were available, and the questionnaire was ended.
Participants completed a questionnaire either digitally or on paper in the period from June to August 2020. Recruitment to the digital questionnaire was conducted through snowball sampling, using social media and newsletters. A link to the digital questionnaire was distributed via various Facebook pages: the Norwegian Pensioners’ Association (NPA), “Sterk og stødig” (Strong and Steady, a national preventive group exercise concept for older people), and the author’s personal pages. In addition, the NPA included information on the study and the link to the survey in their monthly newsletter. The paper version of the questionnaire was distributed and collected through the NPA’s local offices in three counties (Møre og Romsdal, Agder, and former Østfold in Viken). All paper versions were kept in sealed envelopes during transport and then stored in a locked cabined in a locked room. The NPA hired someone (who were registered as a project employee at REC Central) to manually plot the data from the paper questionnaire to a digital format.
The digital questionnaire was created using www.nettskjema.no, a survey solution developed and hosted by the University of Oslo (firstname.lastname@example.org). The data were stored on the TSD (Tjeneste for Sensitive Data) facilities, owned by the University of Oslo and operated and developed by the TSD service group at the University of Oslo’s IT Department (email@example.com).
The questionnaire included items on general health, mental health, and physical activity. Items on background, the HUNT 1 Physical Activity Questionnaire (HUNT 1PA-Q)  and Cohort NORway Mental Health Index (CONOR-MHI)  were taken from the HUNT survey, a large population-based longitudinal study involving 240,000 participants since 1984 . (The HUNT questionnaires can be found at https://www.ntnu.edu/hunt/data/que.) Most questions were asked for both before the 13th of March and after/present-day at the time of the survey. The questionnaire used in the current study consisted of three parts.
Part One included 13 questions on background information, such as age, gender, living conditions, the use of walking aids, falls in the previous 12 months (0/1/2/3/ > 3), fear of falling (not at all/yes a little/yes pretty worried/yes quite worried), changes in cognition in the last five years (none/mild/moderate/severe), and ADL. ADL was assessed by the mobility subscale plus three items from the domestic subscale of the Nottingham Extended ADL Index (Nottingham EADL) .
Part Two included 10 items on PA and exercise, including the amount of PA and exercise in general along with the amount of different specific activities (e.g., walking, biking, gym, and group exercise). The amount of PA was measured with the HUNT 1 PA-Q which includes questions on frequency (0.0–5.0 points), intensity (1–3 points) and duration (0.10–1.00 points) of PA. These three subscales together gives an index with a score ranging from 1.1–15, with 15 as the best score . (Example item: “Before the 13th of March, how often did you do physical exercise?”).
Part Three included six items on general and mental health. (Example item: “Today, most of the time, how would you say your health is?”) For general health, we used the first question from the SF-36 , and for mental health, we used the seven questions from the CONOR-MHI. Each of the CONOR-MHI items are rated on a Likert Scale (1–4), with total scores ranging from 7–28, where higher scores indicate worse mental health .
The questionnaire also included items on what motivates participants to perform physical activity and exercise, but this was not the focus of this paper, thus these items are not reported here.
Data processing and analysis
Descriptive analyses include participants’ demographic characteristics, presented as numbers and percentages for the total sample and by gender. Some categories were merged because of few answers or practical means; for education “high school” and “apprenticeship”, for living conditions “with husband” and “with others”, for number of falls 2, 3 and > 3 falls, for fear of falling “pretty worried” and “quite worried” and for change in cognition “moderate change” and “severe change” were merged. The Chi-square test was used to examine differences in background variables for gender.
We analyzed changes in PA, general health, and mental health before versus after the lockdown using the Wilcoxon Signed-Rank Test. Logistic regression was used to analyze how maintaining one’s activity level affected their general and mental health status during lockdown, as well as how the type of activity before lockdown (organized offer exercise or non-organized exercise) affected activity levels during lockdown. Independent variables with a bivariate Spearman’s correlation of > 0.1 with the dependent variable were included as covariates in the logistic regression analyses: these were gender, general health, and type of activity before lockdown.
For the regression analyzes, based on the participants’ answers on HUNT 1 PA-Q from before 13th of March and today a new variable was created and labeled as change in PA. From this new variable the participants were divided into two groups; those who reported a reduced levels of PA, and those who reported the same or increased levels of PA. Based on the participants’ answers related to general health and CONOR-MHI from pre-lockdown and at the time of the questionnaire, two new variables were created and labeled as change in general health and change in mental health. For these two new variables, the participants were divided into two groups: those who reported having worse general or mental health currently compared to pre-lockdown, and those who reported the same or better general and mental health currently compared to pre-lockdown.
We also divided the participants based on how frequently they attended organized offers of exercise before the lockdown. The two new groups were labeled as “organized offer exercisers” (those attending the gym or group exercise classes once or more per week) and “non-organized exercisers” (those attending the gym or group exercise classes less than once a week).
All statistical analyses were computed using IMB SPSS 27 (IBM Corp. Released 2020. Armonk, NY: IBM Corp), and the significance level was set at p < 0.05.We did not perform any power calculation as we included participants within a limited timeframe, to ensure similar conditions across participants regarding restrictions.