3.1 Socio-demographic section
Environmental Context (EC). Analyses of responses collected in this section of the protocol showed that, with regard to the place of residence (item 1), 73.6% of the participants experienced quarantine in a dwelling with four or more rooms (excluding bathrooms and closets, M = 4.3, SD = 1.3). Most of the adolescents did not have a room for themselves (item 2), but shared their own room with a brother/sister (53.4%), while 32.5% had their own room to study and sleep in. With regard to open spaces (item 3), 3.7% had no outdoor space available in the house (only windows) or in the condominium, 5% had only a condominium space available (courtyard or green area), 46.3% had a balcony, while 45.1% had a terrace or garden. Regarding the possibility to access the Internet (item 4), the data showed that 3.1% did not have access to the Internet, 9.5% accessed the Internet with their mobile phone or for a limited time, 3.1% accessed the Internet via Wi-Fi but for a limited time, 58.6% accessed the Internet via unlimited time Wi-Fi, while 25.8% accessed the Internet via fast and unlimited time Wi-Fi. With regard to the availability of devices to connect to the Internet (item 5), the data showed that 33.7% had a mobile phone only, 3.1% had a tablet, 36.2% a shared PC and 27.0% a PC for exclusive use. To get a single measure of environmental context (variable named EC), each of the five items of the section were recorded into three levels of comfort (low, medium and high) and a principal component analysis (PCA) was carried out by extracting one single component and computing the factor score. The PCA showed that the unidimensional solution explained 30.7% of the variance, with each item showing a saturation > .497. The higher was the score on this index, the better was the living context during the lockdown.
Changes in lifestyle (CL). Analyses of responses collected in this section of the protocol showed that there were changes in both eating habits and sleep-wake rhythms during the quarantine. In particular, with regard to eating habits, the data showed that 82% of adolescents stated that they had modified their diet from a quantitative point of view (item 1, 54.0% "a little", 28.2% "a lot"). 57.9% stated that they had modified their diet from a qualitative point of view (item 2, 42.9% "a little", 15.0% "a lot"). 68.4% declared to have modified the times (hours and frequency) of the alimentation (item 3, 43.2% "a little", 25.2% "a lot"). With regard to the analysis of changes in biological rhythms and sleep quality, more generalized changes emerge. In fact, 65.0% of adolescents declared that they have modified "a lot" the sleep-wake rhythms (item 4), 29.4% "a little", while only 5.5% "not at all". While, with regard to the quality of sleep (item 5), 40.5% of adolescents reported that they have modified the quality of sleep "very much", 37.7% "a little", while only 21.8% "not at all". To get a single measure of changes in lifestyle (variable named CL), a principal component analysis (PCA) was carried out on the five items of the sections by extracting one single component and computing the factor score. The PCA showed that the unidimensional solution explained 44.2% of the variance, with each item showing a saturation > .594. The higher was the score on this index, the higher were the changes in the lifestyle during the lockdown.
Worries about infection (WI). Analyses of responses collected in this section of the protocol showed that during the quarantine, adolescents were more worried about their families getting infected (item 1), M = 7.2, SD = 3.1, than they were worried about themselves (item 2), M = 4.3, SD = 3.6; the comparison between these two scores is significant, t(325) = 14.71, p < .001. In particular, 38.0% declared they were not worried about contracting the virus themselves, while only 10% reported they were not worried about a family member contracting the virus. In the latter case, 31% declared a maximum level of concern, i.e. "10". Finally, with regard to the time spent reading or listening to information related to contagion (item 3), the majority of adolescents (50%) reported they spend about one hour, 13.2% spend two or more hours, while 12.9% do not know anything at all. To get a single measure of worries about infection (variable named WI), a principal component analysis (PCA) was carried out on the three items of the sections by extracting one single component and computing the factor score. The PCA showed that the unidimensional solution explained 53.1% of the variance, with each item showing a saturation > .589. The higher was the score on this index, the higher were the worries about the infection during the lockdown.
3.2 Psychopathology section
State anxiety
the assessment of state anxiety symptoms during the COVID-19 revealed that adolescent had a mean score of 41.6 (SD = 10.8); considering the cut-off of 40, predictive of clinically relevant symptoms17, data showed that the 47.5% of the sample exceeded it; specifically, 27.0% showed “mild anxiety”, 14.1% showed “moderate anxiety” and 6.4 “severe anxiety”. A significant gender difference was observed, t(324) = 5.74, p < .001, with females showing higher state-anxiety (S-A) than males (see Table 1).
Depression
the assessment of depressive symptoms during the COVID-19 revealed that adolescent had a mean score of 6.5 (SD = 5.6); considering the cut-off of 12, predictive of clinically relevant symptoms18, data showed that 14.1% of the sample exceeded it. A significant gender difference was observed, t(324) = 6.89, p < .001, with females showing higher depression (MFQ-SF) than males (see Table 1).
General psychopathology
the assessment of the presence of general psychopathology symptoms referred to the six months (thus before the onset of pandemic) showed that adolescents had a mean total score of 11.4 (SD = 5.9); considering a cut-off score of 14, data indicate that 26.7% of the sample exceeded it; specifically, 9.2% showed a “slightly raised” score, 6.1% showed a “high” score, 11.3% showed a “very high” score. A significant gender difference was observed, t(324) = 5.80, p < .001, with females showing more symptoms (SDQ) than males (see Table 1).
3.3 Relationship between extracted indicators (EC, CL and WI) and emotional symptoms
Data from the hierarchical regression analysis are reported in Table 2. Results showed a similar pattern of effects for the two considered dependent variables. As regards the state anxiety, data showed that over and above the control variables (gender and age), the general psychopathology symptoms (SDQ) were uniquely associated with the anxiety scores, R2diff = .294, p < .001. As expected, the model-fit increased when the three indicators were entered into the model, R2diff = .019, p = .017, whereas the last step did not show significant two-way interaction effects. The parameters of the final model revealed that general psychopathology symptoms (SDQ), β = .556, p < .001 and worries about infection (WI), β = .110, p = .013 were both uniquely independent predictors of anxiety, R2 = .425, p < .001. No other significant effects were observed. That is, over and above the other variables in the model, the more were the general psychopathology symptoms before the COVID-19 the higher was the state anxiety during the lock-down, the higher was the worries about the infection the more was the anxiety.
As regards the depression, data showed that over and above the control variables (gender and age), the general psychopathology symptoms (SDQ) were uniquely associated with the depression scores, R2diff = .390, p < .001. As expected, the model-fit increased when the three stress indicators were entered into the model, R2diff = .020, p = .003, whereas the last step showed significant two-way interaction effects between indicators and the general psychopathology, R2diff = .023, p < .001. The parameters of the final model revealed that gender, β = − .103, p = .012, general psychopathology symptoms (SDQ), β = .625, p < .001, environmental context (EC), β = − .106, p = .005, and changes in lifestyle (CL), β = .108, p = .006 were all uniquely independent predictors of depression, R2 = .569, p < .001, and that the amount of changes in lifestyle (CL) moderated the relation between the general psychopathology and the depression scores (see Fig. 1). No other two-way interaction effects were observed. That is, over and above the other variables in the model, females showed a higher level of depression than males, the more were the general psychopathology symptoms before the COVID-19 the higher was the depression during the lock-down, the better was the living context the lower were the depressive symptoms, the higher were the changes in lifestyle the more was the depressive symptoms as well as the combination of the general psychopathology and changes in lifestyle increased the depressive symptoms.