**3. Results**

A total of 558 respondents, 347 participants in group N and 211 in group S, accepted to participate in the study and completed the survey. A detailed description of the study population is summarized in Table 1.


**Table 1.** Description of study population: sociodemographic characteristics and work-related factors.


**Table 1.** *Cont.*

The study population consisted of 399 women (71.5%) and 159 men (28.5%) aged 18–65 years. We found statistically significant differences between the two groups in all the considered sociodemographic characteristics: the number of women in group N was higher than in group S (75.2% and 65.4%, respectively); less than one-third of subjects in group N (27.4%) and the majority in group S (56.9%) were aged under 40 years; most participants in group S were graduated (55%), while in group N the percentages were more equally distributed among the different educational degree. Regarding marital status, in group S, single (not married and divorced) and in pairs (married and unmarried partners) were similarly represented, whilst in group N, the majority had a partner (72.6%) and parenthood was more frequent in group N than in group S (59.7% and 45.5% had children, respectively).

Considering work-related factors, most of the participants were nurses in group N and doctors in group S; in both groups, there were no statistical differences in relation to the employment in COVID wards and the number of contacts per week with COVID patients. Moreover, 68 subjects (42 in group S and 26 in group N) were employed in remote working during the pandemic. In addition, we observed a higher length of employment in group N than in group S, with a statistically significant difference.

European Quality of life–5 Dimensions (Index and VAS), Athens Insomnia Scale and Brief COPE scores are reported in Table 2. The reliability assessment showed the following Chronbach's alpha: EQ–5 D Index 0.59; Athens Insomnia Scale 0.86; while for the different coping strategies we found Active 0.70; Planning 0.74; Positive Reframing 0.70; Acceptance 0.54; Humor 0.65; Religion 0.88; Emotional Support 0.81; Instrumental Support 0.79; Self Distraction 0.50; Denial 0.55; Venting 0.58; Substance Use 0.89; Disengagement 0.50; Self Blame 0.42.

Despite the two groups showing high values of self-reported quality of life, group S showed better scores than group N both in Index and VAS of EQ-5D questionnaire with statistically significant differences. Moreover, we stratified the sample into different subgroups according to sociodemographic and work-related variables, comparing the two groups. Subsequently, we found the highest values of EQ-5D-Index in the stratified group S, with statistically significant differences among women, graduated subjects, participants with no children, workers not employed in COVID wards. Moreover, a similar trend was observed in EQ-VAS, except for gender, for which statistical significance was found among men but not among women. Furthermore, in order to identify possible predictors of better scores, we used a generalized linear model for EQ-5D-Index as reported in Table 3.

**Table 2.** Mean scores of validated questionnaires assessing health-related and perceived quality of life, insomnia, and coping strategies in healthcare personnel during the first wave of COVID-19 pandemic (*n* = 558).


**Table 3.** Generalized linear model for EQ-5D-Index, assessing quality of life in healthcare workers during the first wave of COVID-19 pandemic (*n* = 558).


In the total sample, male gender, high education levels, and lower seniority were positive predictors of a better perceived quality of life according to EQ-5D-Index. Having a partner and lower seniority were considered predictors of a better quality of life respectively in group N and group S. For EQ-VAS (Table 4), male gender and high education levels in the total sample represented significant predictors of better perceived quality of life. High education degree was identified as a positive predictor both in group N and S; while in group S male gender and lower seniority were considered predictors of more excellent scores in the European Quality of life questionnaire.


**Table 4.** Generalized linear model for EQ-VAS, assessing perceived wellbeing in healthcare workers during the first wave of COVID-19 pandemic (*n* = 558).

Differently, the Athens Insomnia Scale questionnaire revealed insomnia in 162 out of 247 subjects (46.7% in group N) and 91 out of 211 (43.1% in group S), without statistically significant differences. Nevertheless, after stratifying the sample as described above, we found statistically significant differences among not married subjects and participants with no children, showing worse outcomes in group N after stratification. Moreover, in the distribution of the Athens Insomnia Scale, we considered the score 6 as pathological cut-off (such as proposed by Soldatos et al. [32]); consequently, we used univariate and multivariate logistic regression (Table 5) in order to individuate significant predictors of insomnia symptoms.

Accordingly with univariate logistic regression, female subjects (OR 2.09, 95% CI 1.42–3.07) and nurses (OR 1.62, 95% CI 1.09–2.42), both male and female, showed a high risk of suffering from insomnia in the total sample, while multivariate approach showed only women as the category at high risk (OR 2.20, 95% CI 1.48–3.28), in the overall sample as well as in both groups N and S. In group N, single subjects (not married and divorced)

showed a higher risk of suffering from insomnia (OR 1.76, 95% CI 1.09–2.83) in univariate regression. In group S univariate approach showed that the number of contacts per week with COVID patients was also a work-related factor determining a high risk of insomnia (OR 1.29, 95% CI 1.00–1.66); moreover, in the multivariate logistic regression, nurses showed a lower risk of insomnia when compared to physicians (OR 0.99, 95% CI 0.98–0.99).

**Table 5.** Univariate and multivariate logistic regression for Athens Insomnia Scale in healthcare workers during the first wave of COVID-19 pandemic (*n* = 558).


Considering the mean scores of the Brief COPE questionnaire (Table 2), the coping strategies with the highest values were Active, Planning and Acceptance, while Substance Use and Disengagement reported the lowest scores in both groups. Moreover, group S reported higher values than group N in Humor, Religion, Denial, and Self-blame, showing statistically significant differences. Additionally, we applied a generalized linear model for each one of the 14 coping strategies. In the overall sample, we found different predictive variables as illustrated in Table 6A,B, for sociodemographic and work-related features of the study population, respectively. Male gender was revealed to be the most frequently described negative predictor in our statistical models, showing that being a woman is related to almost all the analyzed coping strategies. An age of >40 y acted as a predictor of Acceptance and Religion; education positively predicted Emotional Support, while a lower educational level was in relation with Denial and Venting. Being part of group S predicted Religion and Denial, while group N participants were related to Instrumental Support. As regards work-related factors, the employment in COVID wards was related to Emotional and Instrumental Support. On the other hand, remote working predicted Religion, Denial, and Disengagement. No predictive variables were found for the coping strategies Positive reframing, Humor, and Substance use. While Disengagement was not predicted from any

sociodemographic characteristics, no work-related variables were found as predictors of Acceptance, Self-distraction, Venting, and Self-blame.

**Table 6.** (**A**). Generalized linear model for Brief-COPE in relation to sociodemographic predictors in healthcare workers (*n* = 558). (**B**). Generalized linear model for Brief-COPE in relation to work-related predictors in healthcare workers (*n* = 558).


Table reports B-values; 95% CI (in brackets); <sup>T</sup> = Total sample; <sup>N</sup> = Group N; <sup>S</sup> = Group S; \* = *p*-value < 0.05; \*\* = *p*-value < 0.01; \*\*\* = *p*-value < 0.001. No predictive variables were found for the coping strategies Positive reframing, Humor, Substance use and Disengagement. Acceptance, Humor, Self-distraction, Venting, Substance use and Self-blame.
