Effects of quarantine on Physical Activity prevalence in Italian Adults: a pilot study

Participants and study design

The current study is a retrospective observational design, which used a questionnaire to record and gather data. The surveys were shared on the Web from 2^nd December 2020 up to 10^th January 2021 and involved 749 participants (female = 552 (73.7%), male = 197 (26.3%), Table 1). During this period, the Italian Government declared the end of many social restrictions, allowing citizens to practice all types of physical activity in outdoor spaces, and few physical activities in indoor spaces, in respect of every emergency measure (wearing a mask, social distancing of almost one meter, body-temperature measurement, and hand sanitation).

The questionnaire selection criteria included adult people who were at least 18 years old, who lived in Italy during the COVID-19 pandemic, and who spoke the Italian language. No country or region limitations were explained. Written informed consent was provided by the participants before the study began. The study was approved by the Bioethics Committee of the University of Bologna (Approval code: 224254, October 9, 2020).

Questionnaire

We created the questionnaire with Google Forms and shared it on one global social network (Facebook®; Meta Platforms, Inc, Cambridge, MA, USA) and WhatsApp® (Meta Platforms, Inc, Cambridge, MA, USA). All participants were informed and gave us privacy consent to treat their data. They could fill out the survey with no Google sign-in request. They could manually enter all general information or allow social networks to report them. The questionnaire was self-administered in the Italian language. Each completed survey was saved on a Google database, and we gathered all data as an Excel spreadsheet (Microsoft Office®; Microsoft Corporation, Redmond, WA, USA).

The questionnaire was of two parts: the first part included general and demographic information such as age (in years), sexes (male or female), educational status (primary school, secondary school, high school, degree, master’s degree, PhD, or other), occupational status (student, employed, freelance, unemployed, homemade, retired, other), information on PA prevalence over the quarantine (example: “Did you practice PA and/or sports before the COVID-19 quarantine?”), and information about motivations that negatively or positively affected citizens’ PA habits (example: “If you did not practice any PA during the lockdown, why?”); the last part included 13 questions that investigated whether the sample believed physical inactivity could negatively impact people’s health and whether the sample thought that socio-political strategies affected the citizens’ habits.

Several types of data were collected and transformed before the analysis. Questions on PA prevalence and gender were recorded as binary data (yes or no, male or female) and transformed in Bernoulli data (0 or 1) where 0 was not and female, and 1 was yes and male. The citizens’ age was collected as years and successively grouped into four categories (a) 18–39 years (young adults), (b) 40–49 years (early middle-aged adults), (c) 50–64 years (late middle-aged adults), and (d) ≥65 years (old adults). A previous study reported four age intervals stratified into 12 age subsets (Horng, Lee & Chen, 2001); however, we divided the middle-aged category into two groups and we settled on 65 years as the cut-off point for the elder category (Italian retired age). Information that investigated why citizens did not practise PA were registered as nominal data with five options for the period during and after the lockdown respectively: “COVID-19 problems”, “Environmental problems”, “Health problems”, “Psycho-social problems”, “Technological problems”; “COVID-19 problems”, “Environmental problems”, “Health problems”, “Psychological problems”, “Socio-economics problems”. Differently, the reasons for which people did PA and/or sport during lockdown were: “General well-being”, “Physical well-being”, “Psychological well-being”, “Physical and General well-being”, “Psychological and General well-being”, and “Psychological and Physical well-being”. Finally, the 13 items were recorded as ordinal data such as “disagree, partially disagree, partially agree and agree”, and then transformed into discrete items (Likert-type scale) from 1 (disagree) to 4 (agree).

Statistical analysis

We reported the frequencies of occurrence as the number of observations for categorical and binary data and mean ± standard deviation (SD) and the minimum and maximum observed values for numerical (continuous and discrete) data. The prevalence of PA was analysed among sexes and age groups: the McNemar’s test ( ${\chi }^2$) was assessed with the exact McNemar significance probability (P) (Brunner & Giannini, 2011). In addition, the Risk Ratio (RR) was calculated as the previous study suggested (Ranganathan, Aggarwal & Pramesh, 2015).

The reliability of the questionnaire items was tested through homogeneity and internal consistency. To identify the dimensions of our questionnaire, the exploratory factor analysis (EFA) of tetrachoric correlations was assessed (Schreiber, 2021). To measure the sampling adequacy, the Kaiser–Meyer–Olkin (KMO) value was calculated; values >0.80 were considered meritorious. To test the null hypothesis that variables were not intercorrelated, Bartlett’s test of sphericity was performed, and the determinant of matrix correlation value, ${\chi }^2$, degree of freedom (df), and the p-value (P) were reported. The choice of the number of factors was based on the eigenvalues, and we used the unweighted least-squares method and the Kaiser rule to extract only factors with an eigenvalue ≥1. Finally, the orthogonal Varimax rotation was used and the related ${\chi }^2$ and p-values (P) were settled. Items with a loading value <0.40 were dropped, and the final model included only items with loadings of ≥0.40 on their specific factors. In addition, to give a measure of the common variance proportion of the variable not associated with the factors, the Uniqueness (1–commonality) was calculated (Lawley & Maxwell, 1962).

Then, to provide the internal consistency of our questionnaire, we calculated Cronbach’s alpha on all items and each factor, respectively. We considered the alpha value acceptable ranging from 0.70 to 0.95 (Tavakol & Dennick, 2011), and we reported the average interitem correlation and alpha values (α). Finally, the test-retest reliability was assessed through Spearman’s rank correlation coefficient (ρ), where a value equal to 1 was considered perfect reliability, a value between 1 and ≥0.90 was considered excellent, a value ≥0.80 and ≤0.90 was considered good, a value ≥0.70 and ≤0.80 was considered acceptable, a value ≥0.60 and ≤0.70 was considered questionable, and all values <0.60 was considered poor reliable (Shou, Sellbom & Chen, 2022).

The ANOVA methods were used to compare the mean differences among age groups, sexes, and age-sexes interactions. The Snedecor-Fisher test value (F) and p-value (P) were reported.

We selected a priori hypothesis test significance levels as equal to 0.05 (α). In addition, we used G*Power 3.1.9.7 for Windows 10 (Heinrich-Heine-Universitat Düsseldorf, Universitätsstraße 1, 40225 Düsseldorf, Germany) to compute an a priori sample size analysis. which equalled to 731 with the following parameters: type I error = 0.05, Effect Size = 0.15, Statistical Power = 0.80 and number of groups = 8 (gender*Age). All statistical analyses were performed by STATA® software (version 17, StataCorp., College Station, TX, USA, StataCorp LP).

Sample characteristics

We collected 751 interviews. Two of these were excluded for missing values and we finally analysed 749 interviews.

Table 1 shows participants’ characteristics of gender type, educational level, occupational kind, and age. Most of the sample was female and younger than 50 years old, and approximately half participants had high school graduation. Just 3.87% of the sample was unemployed.

Physical activity prevalence

Table 2 shows the prevalence of PA before, during, and after the forced lockdown, among age groups and sexes. Generally, a decreasing prevalence of PA trend was observed with significant differences between the period before lockdown and periods during and after. When stratified for gender, no significant differences emerged between the periods before and during the lockdown in both sexes, while the number of women and men who practised PA before lockdown decreased after quarantine drastically. As regards the age classes, a negative trend was observed for the elder categories (≥40 years), which decreased both during and after the lockdown. The number of younger individuals (18–39) who remained active during lockdown did not decrease significantly, but a greater reduction was observed after the quarantine.

In sexes analysis, female people showed a negative trend with lower prevalence post the lockdown, whereas male subjects did not report significant differences.

In age comparisons, a significant reduction of people who practised PA before the lockdown was observed post the lockdown in subjects who were 40–64. Also, a lower prevalence was recorded post-lockdown than during lockdown in the youngest group.

Successively, the motivations for which individuals chose to practice (or not) PA and/or sport during lockdown were investigated. Generally, most of the sample who did not practice PA and/or sport experienced Psycho-social problems (64.57%), while 16.54% reported Environmental problems. Only 10 individuals (3.94%) had problems with technology. When gender differences were analysed, most of both male and female samples were inactive due to Psycho-social problems (M = 35, 59.32%; F = 129, 66.15%); also, the percentage of women and men who did not practice PA and/or sport for Health problems was similar, 8.72% and 8.47% respectively. In addition, Fig. 1A shows the percentage of responses divided by gender and age. Differently, 74.95% of the sample who were active during the lockdown selected answers with psychological benefits, while just 3.23% practised PA and/or sport for physical well-being alone. This prevalence in psychological responses remained same in gender analysis too, despite was more prevalent in women (F = 276, 77.31%; M = 95, 68.84%). Figure 1B shows the percentage of responses divided by gender and age.

In addition, the reasons why individuals chose to not practice PA and/or sport after the lockdown was investigated. Generally, 189 people (62.17%) did not practice PA and/or sport due to Psychological problems, while 12.83% experienced Socio-Economic problems. Just 18 individuals (5.92%) were influenced by Environmental problems. When gender differences were analysed, 62.08% of women and 62.5% of men experienced Psychological problems, while the minor problems for both females and males were Environmental related (F = 16, 6.67%; M = 2, 3.12%). Finally, Fig. 2 shows the percentage of responses divided by gender and age.

Questionnaire characteristics

We included 11 items in the model (Table 3). In the beginning, 18 questions were considered, of which five were excluded because data was not on the Likert scale and two were excluded due to poor loading values. Each item was labelled as PAA#, which indicates Physical Activity Aspects and the number (order) of the question. The questions were the following:

1. PAA2: “Do you think the lack of practice of physical activity and/or sport during lockdown has negatively affected your health?”

2. PAA3: “Do you think the lack of practice of physical activity and/or sport in older adults during lockdown has negatively affected their health?”

3. PAA5: “Do you think the lack of practice of physical activity and/or sport during lockdown could cause problems for the national health system and public health?”

4. PAA6: “Do you think the lack of practice of physical activity and/or sport could have affected emotional/psychological wellbeing during the lockdown?”

5. PAA7: “Do you think the lack of practice of physical activity and/or sport could have affected emotional/psychological growth in children/adolescents during the lockdown?”

6. PAA8: “Do you think children/adolescents have practised enough amount of physical activity and/or sport during the lockdown?”

7. PAA9: “Do you think the lack of practice of physical activity and/or sport could have affected emotional/psychological status in older adults (>65 years old) during the lockdown?”

8. PAA10: “Do you think the Italian government adequately considered the practice of physical activity and/or sport during the lockdown?”

9. PAA11: “Do you think the usage of tech devices to practice physical activity and/or sport during lockdown could have affected social interaction negatively?”

10. PAA12: “Do you think sports federations and societies adequately promoted the practice of physical activity and/or sport during the lockdown?”

11. PAA13: “Do you think public and private schools adequately promoted the practice of physical activity and/or sport during the lockdown?”

Firstly, we analysed the sample adequacy (KMO = 0.81), the correlation matrix determinant (Det = 0.09), and the Bartlett’s test of sphericity ( ${\chi }^2$ = 1,743.66; dfs = 55; P < 0.001). Figure 3 shows the scree plot of eigenvalues with the Kaiser rule. Two factors met our criteria and the Varimax rotation reported a LR test of significant results ( ${\chi }^2$ = 1,746; P < 0.001; n = 749). Table 3 shows that factor 1 (how participants thought physical inactivity impacted people’s health) included seven items and factor 2 (what participants thought about socio-political strategies on PA habits) included four items. In addition, we assessed Cronbach’s alpha on 11 items. The last row of Table 3 shows all Cronbach’s alpha values. We used the mean test scale on standardized items, deleting missing values from the analysis. The average interitem correlation on 749 observations was 0.097 and the scale reliability coefficient (α) was 0.724. In addition, we reported each consistency factor data: factor one α was equal to 0.779, whereas factor two α equalled 0.603.

After assessing EFA computations, test-retest reliability was assessed in a sample of 149 citizens (Table 4). The retest period was of 1 month. The lowest value of ρ was 0.738 (acceptable), while the highest value was equal to 0.914 (excellent).

Physical activity aspects

Table 5 shows the ANOVA outcomes. As regards factor 1, both female and younger subjects exhibited higher value than men, which means they believed that physical inactivity due to COVID-19 social restrictions impacted the psychological status of adolescents and adults and will negatively affect global health. About factor 2, male and elder people reported significantly higher values than women and youths on items 8 and 10, which indicates that younger women thought of Italian government adopted inadequate solutions to promote PA during the lockdown, and adolescents did an insufficient amount of PA.