*2.2. Measures*

All instruments used in the larger European study have been previously validated in English and French. The instruments were administered in French.

*World Health Organization Quality of Life brief version (WHOQOL-BREF)* [34]. It includes 26 items measuring physical, psychological, social, and environmental aspects of QoL using 5-points Likert scales. It also includes two items measuring overall QoL and state of health, respectively. The questionnaire, as well as its French translation, have proven to be reliable and valid [34,35]. As recommended by the authors of the questionnaire, mean scores were transformed to range from 0 (poor QoL) to 100 (good QoL). As we were interested in measuring QoL as a global construct, we used the mean score of all items. Cronbach's α revealed good internal consistency (.91) for the total QoL score.

*Perceived Stress Scale (PSS-14)* [36]. The French translation of this scale has been shown to have good psychometric properties [37]. It includes 14 items rated from 1 (low perceived stress) to 5 (high perceived stress) that we averaged to compute the perceived stress score. Cronbach's α revealed good internal consistency (.90) for the total stress score.

*Connor-Davidson Resilience Scale (CD-RISC)* [24]. It contains 10 items rated from 1 (low resilience) to 5 (high resilience) and has been translated and validated in French [38]. Cronbach's α revealed good internal consistency (.88) for the total resilience score.

*Multidimensional Scale of Perceived Social Support (MSPSS)* [28]. It measures perceived social support from family, friends, and significant other (i.e., any one person to whom the individual feels especially close) and consists of 12 items rated from 1 (low social support) to 7 (high support). The French translation exhibited good psychometric properties [39]. Cronbach's α revealed good internal consistency (.94) for the total social support score.

*Coping Orientation to Problems Experienced Inventory (Brief-COPE)* [40]. This scale measures individuals' favored coping strategies. It includes 14 dimensions, each measured by two items, that represent a coping strategy. The items are rated from 1 (rare use of that coping strategy) to 4 (frequent use). We calculated the mean of the two items for each dimension, and as Cronbach's alpha may underestimate the reliability of two-items scales, we evaluated their reliability by using the Spearman-Brown coefficient rs as recommended by Eisinga et al. [41].

The 14 dimensions and their brief descriptions are as follows: Active coping refers to individuals actively attempting to suppress their problems or their effects (rs = 0.56); Planning consists of devising steps to best manage problems (rs = 0.69); Seeking instrumental support refers to seeking advice or help (rs = 0.78); Seeking emotional support refers to seeking moral support or sympathy (rs = 0.73); Venting refers to expressing emotions about problems (rs = 0.71); Positive reframing involves reassessing problem situations as positive (rs = 0.77); Acceptance is acknowledging the existence of problems (rs = 0.69); Denial is refusing to acknowledge the existence of problems (rs = 0.60); Self-blame is reproaching oneself for problems (rs = 0.56); Humor is not taking problems seriously (rs = 0.75); Religion is seeking solace in religious beliefs (rs = 0.84); Self-distraction is diverting one's attention away from problems by focusing on something else (rs = 0.36); Substance use is escaping reality by consuming alcohol or drugs (rs = 0.94); Behavioral disengagement is abandoning goals prevented by problems (rs = 0.62) [42]. The French translation showed acceptable psychometric properties [42].

*Sociodemographic Variables*. Participants were asked to indicate their gender (male, female, 'I define myself otherwise'), age category (18–29 years old, 30–39, 40–49, and 50 or more), marital status (married, single, other), and having children (yes, no). The question

on 'marital status other' was open-ended, and most who responded indicated they were divorced (73.3%). Other questions asked were as follows: how long they have had their nursing diploma (less than 5 years, 5 to 10 years, or more than 10 years); if they had been reassigned to another service other than their usual one during the pandemic (yes, no); if at any point, they had been exposed to COVID-19 during their work (direct exposure: worked in a COVID-specific unit; indirect exposure: worked in a non-COVID-specific unit but that they received some COVID patients; and no exposure: no COVID patient was admitted to the unit); and whether they were engaged in private practice (NPPs) or a hospital/institutional care setting (NHCIs).

## *2.3. Data Analysis*

All variables were treated as continuous except for sociodemographic variables that were treated either as dichotomized variables or as dummy variables (more than two categories). Descriptive analyses were first used to describe the sample and independent samples *t*-tests and Chi-square tests of independence were used to compare the two groups of nurses (NHCIs vs. NPPs). We also computed Pearson's correlations and, after checking for linearity and normality assumptions, we conducted multiple linear regressions with QoL as the outcome variable: on (a) the full sample, and (b) stratified by the type of practice samples. We checked for multicollinearity among the predictive variables using the variance inflation factor (VIF) index, which revealed no problematic collinearity (all VIFs < 3) among the predictor variables [43]. As our sample was very large, we used listwise deletion for handling missing values and we lowered the significance threshold to 0.005 to minimize type I error [44]. All analyses were performed using R 4.1.1.
