According to the International Olympic Committee (IOC) consensus statement on mental health in elite athletes (Reardon et al., 2019), sports practice should protect against psychological disorders, given exercise’s antidepressant properties. Interestingly, it has been suggested that when athletes cease to exercise, its antidepressant effect disappears and mood could therefore worsen (Reardon and Factor, 2010). Meta-analytical accounts also support the general benefits of sports practice over health status (Oja et al., 2015). In the COVID-19 context, reports on the athletes’ population consistently showed decreased training activity, unhealthy nutritional and sleep trends during the lockdown, which correlated with poorer mental health (Mon-López et al., 2020a,b; Pillay et al., 2020; Şenışık et al., 2020; Facer-Childs et al., 2021). Less is known about perceived physical health during the pandemic, with only one study investigating individual differences in this variable in a sample of adolescent athletes (McGuine et al., 2021).

Identifying psychological variables related to PA and quality of life (i.e., mental and physical health) during a pandemic is crucial to sustaining athletes at higher risk of experiencing adverse mental and physical consequences under particularly stressful situations.

Self-determination theory (Deci and Ryan, 1985; Ryan and Deci, 2000) identifies three basic psychological needs (i.e., autonomy, competence, and relatedness) that must be satisfied to achieve self-motivation and higher wellbeing. Motivation is defined as the ways in which individuals regulate any of their behavior. More specifically, motivation toward PA is conceptualized as a continuum ranging from amotivation (i.e., lack of intention to engage in PA, non-self-determined regulation) to intrinsic motivation (i.e., practice PA for its own sake, self-determined). Intermediate forms of regulation include external regulation, introjection, identification, and integration (see Markland and Tobin, 2004). Intrinsic motivation toward PA has been associated with a higher commitment to sports practice during COVID-19 lockdown (Leyton-Román et al., 2021) and may be therefore positively related to self-reported PA during the lockdown.

Other factors contributing to understanding athletes’ quality of life under a stressful and uncertain situation such as a pandemic may include general distress, intolerance of uncertainty, and body dissatisfaction (Buckley et al., 2021; Li et al., 2021).

General distress is a state measure of overall negative emotionality, characterized by symptoms of anxiety, depression, and stress (Lovibond and Lovibond, 1995). It has been associated positively with mental and negatively with physical health (Sinclair et al., 2011) and has been recently suggested to be a solid construct in athletes as well (Vaughan et al., 2020).

Intolerance of uncertainty represents a general disposition to interpret uncertain situations as discomforting and has been associated with several psychopathologies, including health anxiety (Gentes and Ruscio, 2011). Under the COVID-19 pandemic, it has been suggested to positively relate to test anxiety in student-athletes (Li et al., 2021).

Body dissatisfaction describes all the concerns with one’s body size and shape and has been indicated as the strongest predictor of eating disorders in athletes (Reardon et al., 2019). Only one study has examined its role in athletes during the pandemic and found higher concerns in female current and former athletes compared to males (Buckley et al., 2021).

Italy was the first Western country to face the dramatic spread of the virus. For what concerns sports, the course of events was characterized by a high uncertainty and rapidity: on February 23rd, 2020, sports events and gatherings were suspended in the two most affected Italian regions (i.e., Lombardy and Veneto), but it was still possible to train and compete. On March 4th, 2020, sports events and competitions of all levels and disciplines in any public or private locations were suspended in specific areas (the so-called “red zones”). Elsewhere it was still allowed to hold events and competitions, as well as training sessions for elite athletes only, within closed sports facilities or outdoors without spectators; all sports societies and associations had to perform tests aimed at containing contagion among athletes, coaches, managers, etc. At that time, recreational sports and PA could be performed under sanitary recommendations. Starting on March 10th, 2020, these measures were extended to the entire national territory: due to nationwide lockdown measures, all, but vital businesses were closed and leaving one’s house restricted to job/health-related or other very serious reasons. Pertaining to the sport context, sports facilities could be still utilized by athletes of national interest training for the Olympics or other international competitions, behind closed doors; therefore, the only competitions allowed were those organized by international committees and could only be held behind closed doors or outdoors without spectators. On March 24th, 2020, the IOC announced the postponement of the 2020 Olympics to 2021. Then, on April 1st, 2020, suspension of all sports events, competitions, and training sessions was declared on the whole Italian national territory. Meanwhile, Italian federations were issuing their own public statements: the Italian Volleyball Federation (FIPAV) announced the suspension of the B series championship (March 4th); so did the Italian Soccer Federation for all championships (FIGC; March 10th), the Italian Track and field Federation (FIDAL) for all competitions (March 27th), and the Basketball Federation (FIP) for the A series championship (April 7th). Finally, on April 26th, Phase 2 was declared, with the possibility to train for athletes of individual disciplines recognized as being of national interest and for all individuals (team sports included) to train alone.

In this context, the aim of the present study was to investigate changes in PA from pre- to during lockdown, and to examine the relationships between PA, quality of life, and psychological variables (motivation to PA, psychological distress, intolerance of uncertainty, and body dissatisfaction) experienced during the lockdown. More precisely, based on previous literature, we formulated three sets of hypotheses:

A priori power analysis was conducted using G^∗Power (Faul et al., 2007) and the Shiny app MCpowrMed (Schoemann et al., 2017). As for mixed ANOVAs (i.e., considering Time as within factor, Level of sport, Type of sport, and Gender as between factors), a total sample size of 42 was sufficient to obtain a power of 0.80, with a medium f effect size (0.25). For 2 × 2 × 2 univariate ANOVAs (considering Level of sport, Type of sport, and Gender) the sample size required equaled 159. Finally, for mediation models, Montecarlo simulations (n = 5,000) conducted with MCpowrMed starting from a theory-based covariance matrix with small-to-medium correlations indicated that 174 participants were enough to obtain a power of 0.80 in Model 1 (i.e., testing the effect of Vigorous MET on SF-12 Mental component, with DASS-21 and IUS-R as parallel mediators), while for Model 2 (i.e., testing the effect of Vigorous MET on SF-12 Physical component, with SF-12 Mental component and IUS-R as parallel mediators) the suggested sample size was 140 participants. Therefore, the sample size achieved in the study (n = 204) over the timeframe of interest (i.e., Phase 1 of the lockdown, before Phase 2 was announced and measures lessened a bit) resulted satisfactorily. Indeed, participants were 204 athletes (94 males, 91 elite athletes), aged 18–58 (M = 22.55, SD = 5.76), practicing different disciplines [93 individual sports (42 track and field, 8 cycling, 7 martial arts, 5 horseback riding, 4 swimming, 4 artistic gymnastics), 111 team sports (44 volleyball, 21 soccer, 11 basketball)]. Participants were considered “elite” athletes if they self-reported competing at the national level or above, and “non-elite” if they competed below the national level. Inclusion criteria were being over 18 of age and practicing a sports discipline at a competitive level. All available data were considered in the analyses, with missing data referring mostly to SF-12 Mental and Physical components (17.15%) and IPAQ-S (26.9% for the total MET score pre-lockdown; 17.6% for the total MET score during lockdown). Table 1 displays the socio-demographic features of the two groups.

To investigate participants’ socio-demographic characteristics (i.e., age, gender, self-reported height and weight) and sport variables (i.e., type of sport, experience, number and hours of training pre- and during lockdown) a socio-demographic information schedule was employed.

To measure PA before and during the pandemic, we asked participants to complete the International Physical Activity Questionnaire–Short version (IPAQ-S; Craig et al., 2003; Italian version by Mannocci et al., 2010). This decision stemmed from the need to have a measure of PA that was short, easily implementable online, and validated in Italian. The IPAQ-S is a self-report questionnaire composed of seven items measuring five domains of PA (i.e., job-related, transportation, housework, sport and leisure time, and sitting). PA is expressed in terms of energy expenditure (MET), intensity (walking, moderate, vigorous), and duration (minutes/day, days/week). The IPAQ-S displayed acceptable reliability indices both in the 12-country analysis of the instrument (Spearman’s ρ∼0.80; Craig et al., 2003) and the Italian validation (Cronbach’s α = 0.60; Mannocci et al., 2010). This was the only instrument administered twice.

To investigate motivation, we used the Behavioral Regulation in Exercise Questionnaire-2 (BREQ-2; Markland and Tobin, 2004; Italian version by Costa et al., 2013). This is a self-report questionnaire comprising 19 items investigating four kinds of motivation toward PA: amotivation (e.g., “I don’t see why I should have to exercise”), external (e.g., “I exercise because other people say I should”), introjected (e.g., “I feel guilty when I don’t exercise”), identified (e.g., “I value the benefits of exercise”), and intrinsic (e.g., “I exercise because it’s fun”) on a 5-point Likert scale. The subscales displayed satisfactory internal consistencies in both the original version (α = 0.73–0.86; Markland and Tobin, 2004) and the Italian validation study (α = 0.70–0.87; Costa et al., 2013).

To address psychological distress during the lockdown, the Depression Anxiety Stress Scales-21 (DASS-21; Lovibond and Lovibond, 1995; Italian version by Bottesi et al., 2015) was administered. This is a 21-item self-report questionnaire that investigates symptoms related to depression (e.g., “I could not feel any positive emotion”), anxiety (e.g., “I felt I was having a panic attack”), and stress (e.g., “I felt stressed”) experienced over the past week on a 4-point Likert scale. Findings on the Italian version recommended the use of the total score, assessing “general distress” (α = 0.90; Bottesi et al., 2015), rather than calculating the three subscale scores separately. Therefore, given this suggestion and the purpose of the present study, we focused only on the total score of the DASS-21.

The Intolerance of Uncertainty Scale-Revised (IUS-R; Carleton et al., 2007; Italian version by Bottesi et al., 2019) was used to measure the tendency to interpret uncertain situations as stressful and threatening. This is a self-report questionnaire made of 12 items on a 5-point Likert scale (e.g., “Unforeseen events upset me greatly”). The scale showed good internal consistency, both in the original version (α = 0.91; Carleton et al., 2007) and the Italian validation study (α = 0.90; Bottesi et al., 2019).

The Italian version of the Short Form Health Survey-12 (SF-12; Ware et al., 1996; Italian version by Apolone, 2001) was used to evaluate athletes’ quality of life. This comprises twelve items (e.g., “During the past 4 weeks, have you had any of the following problems with your work or other regular daily activities as a result of your physical health?”); mental and physical health components were derived, as suggested by both the original (Ware et al., 1996) and the Italian validation study (Apolone, 2001) as being reliable.

Lastly, the Body Dissatisfaction subscale of the Eating Disorders Inventory-3 (BD; Garner, 2004; Italian version by Giannini et al., 2008) was used to investigate body dissatisfaction. This contains 10 items that measure the degree of body dissatisfaction (e.g., “I think my stomach is too big”) on a 6-point Likert scale. The EDI-3 questionnaire displayed good internal consistency both in the original version (α = 0.63–0.97; Garner, 2004) and the Italian validation study (α = 0.70–0.92; Giannini et al., 2008).

A convenience sample of athletes was recruited via email and personal contacts from April 10th, 2020 (1 month into the nationwide lockdown) to April 29th, 2020 (soon after Phase 2 was announced). Participants accessed a link to the survey implemented on Qualtrics and, after giving their informed consent, filled in the self-reported questionnaires: first, they provided socio-demographic and sport-related information, then answered the IPAQ-S (twice: first, referring to the period before the beginning of the lockdown; then, referring to the past 7 days), and finally the other self-report questionnaires in randomized order. The survey took around 30 min to be completed, and participants did not receive any compensation for their participation. The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethical Committee of the Biomedical Sciences Department of the local university (HEC-DSB/01-20).

All statistical analyses were performed using RStudio Team (2020). Mixed ANOVAs were used to evaluate pre- to lockdown changes in PA (IPAQ-S scores; number and hours of training per week; Hypothesis 1). Time (pre- vs. lockdown) was considered as the within-subject variable, while Level of sport (elite vs. non-elite), Type of sport (individuals vs. team), and Gender (males vs. females) as between-subjects variables. If a principal effect (Time, Level of sport, Type of sport, and Gender) or an interaction were significant, Tukey’s post hoc tests were conducted to inspect those differences.

Then, 2 × 2 × 2 univariate ANOVAs were run to compare elite and non-elite athletes by type of sport and gender in their scores in the self-report questionnaires during lockdown (Hypotheses 2a and 2b). Conventional significance levels were adopted for the DASS-21, the IUS-R, and the BD subscale of the EDI-3 (p < 0.05), whereas Bonferroni’s correction for multiple comparisons was applied to the BREQ subscales (p < 0.01), SF-12 components (p < 0.03), and IPAQ-S intensity of PA (p < 0.02) to decrease Type 1 error.

Pearson’s correlations were calculated to first examine the relationships among PA during lockdown and psychological variables (BREQ-2, DASS-21, IUS-R, BD, SF-12). Correlations were interpreted as small when r = 0.20–0.39, medium when r = 0.40–0.59, and strong when r > 0.60 (Zhu, 2012).

Two mediation models (Hypotheses 3a–3d) were tested using the package psych (Revelle, 2021). This method allows evaluating the direct effect of a predictor on a dependent variable (c), as well as the direct effect corrected for the effect that one or more mediating variables may have on the dependent variable (c’). Moreover, the indirect effect of the predictor (ab) can be computed as the effect of the predictor on the mediator (a) multiplied by the effect of the mediator on the dependent variable (b). All effects can be corrected for one or more covariates (e.g., socio-demographic variables). When the 95% bootstrap confidence intervals for 5,000 resamples do not include the 0, the effect is considered significant. If an indirect effect is not significant, it means that the effect of the predictor is independent of the effect of possible mediators.

The mixed ANOVA for the total MET showed a significant main effect for Time [F₍₁,₃₀₁₎ = 65.80, p < 0.001] and Level of sport [F₍₁,₃₀₁₎ = 15.67, p < 0.001], as well as a significant Level of sport × Time interaction [F₍₁,₃₀₁₎ = 5.81, p = 0.017]. Post hoc analyses indicated that there was a significant decrease in total MET from pre- to lockdown (p < 0.001) and that elite athletes had higher scores compared to non-elite athletes (p < 0.001); scores significantly decreased in both groups and elite athletes had higher total MET scores pre-lockdown only (p = 0.001). As for Walking MET, there was a significant main effect for Time [F₍₁,₂₆₀₎ = 55.42, p < 0.001] and Type of sport [F₍₁,₂₆₀₎ = 6.85, p = 0.009]. Post hoc analyses indicated that there was a significant decrease in Walking MET from pre- to lockdown (p < 0.001) and that individual sports had higher scores compared to team sports (p = 0.011). Concerning Moderate MET, a significant main effect for Time [F₍₁,₃₀₈₎ = 9.56, p = 0.002], Level of sport [F₍₁,₃₀₈₎ = 8.52, p = 0.004], and Level of sport × Time interaction [F₍₁,₃₀₈₎ = 7.98, p = 0.005] emerged. Post hoc analyses indicated a significant decrease in scores from pre- to lockdown (p = 0.002) and that elite athletes showed higher scores compared to non-elite (p = 0.004); the decrease in Moderate MET from pre- to lockdown was significant for elite athletes only (p < 0.001) and they had higher scores compared to non-elite pre-lockdown only (p = 0.001). With regards to Vigorous MET, results evidenced a significant main effect for Time [F₍₁,₃₀₂₎ = 55.93, p < 0.001] and Level of sport [F₍₁,₃₀₂₎ = 17.56, p < 0.001]. Post hoc analyses indicated a significant decrease in Vigorous MET from pre- to during lockdown (p < 0.001) and that elite athletes had higher scores compared to non-elite (p < 0.001).

Regarding the number of weekly training, it emerged a significant main effect for Type [F₍₁,₃₄₈₎ = 27.54, p < 0.001] and Level of sport [F₍₁,₃₄₈₎ = 41.40, p < 0.001], plus a significant Level of sport × Time interaction [F₍₁,₃₄₈₎ = 6.55, p = 0.011]. Post hoc analyses indicated that, independently of Time, individual athletes did a higher number of trainings per week compared to team athletes (p < 0.001). It also emerged that elite athletes did a higher number of trainings per week compared to non-elite only pre-lockdown (p < 0.001). Lastly, with respect to weekly hours of training, there were a significant main effect for Time [F₍₁,₃₉₂₎ = 95.07, p < 0.001], Level of sport [F₍₁,₃₉₂₎ = 47.30, p < 0.001], and Type of sport [F₍₁,₃₉₂₎ = 5.80, p = 0.016], together with a significant Level of sport × Time interaction [F₍₁,₃₉₂₎ = 8.10, p = 0.005] and Level of sport × Type of sport interaction [F₍₁,₃₉₂₎ = 9.02, p = 0.003]. Post hoc analyses indicated a significant decrease in the number of hours of training from pre- to during lockdown (p < 0.001); moreover, it emerged that elite athletes trained for more hours than non-elite (p < 0.001) and individual athletes trained more than team athletes (p = 0.020). The decrease in the number of hours of training per week from pre- to lockdown was significant in both elite and non-elite (p < 0.001) and elite athletes had higher scores only pre-lockdown (p = 0.001); moreover, elite individual athletes trained for more hours per week compared to non-elite individual athletes, and both non-elite and elite team athletes (ps < 0.02). Supplementary File 1 displays means and standard deviations for IPAQ measures and training according to Time, Level, Type of sport, and Gender.

The results of the 2 × 2 × 2 ANOVA for the BREQ-2 subscales showed no significant main effects or interactions (ps > 0.01) for Amotivation, External regulation, and Introjected regulation subscales. For what concerns Identified regulation, a significant main effect for Type of sport [F₍₁,₁₉₃₎ = 8.08, p = 0.005] and Gender [F₍₁,₁₉₃₎ = 7.40, p = 0.007] emerged. Post hoc analyses indicated that athletes in individual sports displayed higher scores compared to team sports athletes (p = 0.006) and that females showed higher scores than males (p = 0.008). Results for the Intrinsic regulation subscale revealed a significant main effect for Level [F₍₁,₁₉₄₎ = 12.23, p = 0.001] and Type of sport [F₍₁,₁₉₄₎ = 18.48, p < 0.001]. Post hoc analyses indicated that elite athletes had higher scores compared to non-elite (p = 0.008), and that athletes of individual sports displayed higher levels compared to team sports athletes (p < 0.001).

Results for the DASS-21 total score indicated a main significant effect for Gender [F₍₁,₁₉₆₎ = 12.72, p < 0.001]: females showed higher scores than males in general distress (p = 0.001).

As for the IUS-R, there was a significant Level × Type of sport interaction [F₍₁,₁₉₆₎ = 4.52, p = 0.035], but post hoc analyses showed no significant differences among elite and non-elite, individual and team athletes.

Results for the SF-12 showed no significant main or interaction effects for the Physical component (ps > 0.03), while for the Mental component a main significant effect for Gender [F₍₁,₁₆₁₎ = 12.72, p < 0.001] emerged; at post hoc it emerged that females showed lower scores, indicating poorer self-reported mental health than males (p = 0.013).

Lastly, with regards to BD, there were main significant effects for Type of sport [F₍₁,₁₉₆₎ = 4.62, p = 0.033] and for Gender [F₍₁,₁₉₆₎ = 38.06, p < 0.001]; at post hoc it emerged that team sports displayed higher scores, indicating higher body dissatisfaction, than individual sports (p = 0.039) and that females showed higher scores than males (p < 0.001).

Table 2 displays means and standard deviations for all the psychological variables in the entire sample, elite, and non-elite athletes. Supplementary File 2 displays means and standard deviations for all the psychological variables according to Time, Level, Type of sport, and Gender.

The results of the correlation analyses among all the variables showed mostly small correlations between during lockdown PA, quality of life, and psychological measures, as well as small-to-medium correlations between psychological self-report measures. For example, total MET scores during lockdown showed a small positive correlation with intrinsic motivation toward PA (r = 0.28), and Vigorous MET scores during lockdown displayed small positive correlations with intrinsic motivation toward PA (r = 0.33), mental (r = 0.20), and physical health (r = 0.21). Furthermore, intolerance of uncertainty and psychological distress were weakly (r = −0.22), and strongly (r = −0.66) negatively correlated with mental health, respectively. Table 3 displays all the correlations among MET scores during lockdown and the other self-report questionnaires (BREQ-2, DASS-21, IUS-R, SF-12, BD).

To test Hypotheses 3c and 3d, we assessed two mediation models, one in which DASS-21 and IUS-R mediated the relationship between Vigorous MET and SF-12 Mental component (Model 1), and the other in which SF-12 Mental component and IUS-R mediated the relationship between Vigorous MET and SF-12 Physical component (Model 2). In both models, Level, Type of sport, and Gender were included as covariates.

Results of Model 1 (Figure 1) showed a significant direct effect (c’ = 0.11, p = 0.035) of Vigorous MET on SF-12 Mental component, removing the effects of DASS-21 and IUS-R; the indirect effect was not significant (ab = 0.06, CI [−0.04, 0.15]) for neither DASS-21 (a₁b₁ = 0.06, CI [−0.04, 0.16]) nor IUS-R (a₂b₂ = 0, CI [−0.04, 0.15]).

Results of Model 2 (Figure 2) revealed a significant direct effect (c’ = 0.26, p < 0.001) of Vigorous MET on SF-12 Physical component, removing the effects of SF-12 Mental component and IUS-R; the indirect effect was not significant (ab = −0.03, CI [−0.11, 0.02]) for neither SF-12 Mental component (a₁b₁ = 0.06, CI [−0.04, 0.16]) nor IUS-R (a₂b₂ = 0, CI [−0.04, 0.15]).