Ethics statement

This paper draws on data from the FOCUS – France Canada Observatory on COVID-19, Youth Health and Social Well-being – study which is a repeated cross-sectional online research study that explores the social and health experiences of young adults living in Canada and France during the COVID-19 pandemic. All procedures of the FOCUS study comply with the ethical standards of the relevant international and national regulations on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. This study received ethical approval from the University of British Columbia Behavioural Research Ethics Board (H20-02053). Implied informed consent was obtained from all participants.

Study design and settings

The first FOCUS survey was conducted from October to December 2020 with a focus on young adult’s experiences with COVID-19 public health measures [47], and their access to mental health services [17]. Given that a significant proportion of FOCUS participants reported not being able to access the mental health services they needed (i.e., one of three young adults in Canada and one of four in France) [17], the second survey was designed to investigate which coping strategies young adults employed in the context of the COVID-19 pandemic. For the present analysis, we used data from the second online survey conducted during the COVID-19 pandemic’s second year, which was launched between July 4 and December 13 2021, a period during which France and Canada were facing the fourth wave of the pandemic.

Recruitment and data collection procedures

Young adult survey participants were recruited with a convenience sampling approach through two main strategies (see Fig 1). First, an email invitation to participate in the FOCUS 2021 survey was sent to all participants who completed the first online FOCUS survey in Fall 2020 and who provided their consent to be contacted (Canada; n = 1700, France; n = 1235), allowing us to recruit 576 youth in Canada and 324 in France. Second, we promoted the survey through online posts and advertisements on social media (e.g., Facebook, Instagram) and via university partners websites, press articles, and word of mouth to recruit new participants in both countries (Canada; n = 2300, France; n = 2817). To participate in the FOCUS survey, eligible participants were young adults who had reached the age of majority in their jurisdiction of residence (18 or 19 years old, according to the Canadian province or territory, and 18 years old in France) and were not older than 29 years old (30 for those who took part in the first survey); lived in Canada or France; were able to fill out the online questionnaire in English (Canada) or French (either country). The anonymous online questionnaire collected data on socio-demographics, COVID-19 experiences, healthcare access and health outcomes (including mental health) using Qualtrics. The online questionnaire was first made in English and then converted to French by two bilingual researchers, one of whom was an English-French translator. The validated mental health scales for assessing depressive and anxiety symptoms (described below), as well as the questions about the coping strategies, were all available in both languages.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. A flow chart of sample selection from the FOCUS 2021 survey.

Given that ethno-racial identity is a sensitive and new topic for young adults in the French context, and based on the suggestions of the young adults’ group who conducted the pilot test, the questions about the immigration background (i.e., country of birth of their parents/grand-parents) were asked at the end of the questionnaire in France.

https://doi.org/10.1371/journal.pmen.0000261.g001

In each country, a pre-test questionnaire involving five voluntary young adults was conducted to ensure that the language and wording used in the questionnaire was appropriate for a young adult sample. Details about the study’s objectives and potential risks and benefits were outlined on the first page of the survey. Before they could access the questionnaire, all participants were notified that completing the survey implied informed consent. Participants were able to withdraw from the survey at any time. To ensure the integrity of our online survey and prevent fraudulent submissions (i.e., duplicates, bot infiltrations), we implemented multiple security measures. These included utilizing advanced security features available in our survey platform, Qualtrics, including a Captcha verification question before accessing the survey and detection capabilities to identify false IP addresses, survey duplicates and potential bots. Additional monitoring protocols were also implemented during the data collection period to assess survey completion times and durations and to verify consistency across responses.

Study population

Our analysis sample included young adults of the FOCUS 2021 survey who had completed the sociodemographic section and questions about their perceived level of coping and use of coping strategies. Participants who indicated that they had not experienced any COVID-19-related stress, and those who had missing information about use of coping strategies were excluded from this analysis (see Fig 1).

Measures

At the beginning of the coping section, we first questioned participants to express their level of stress in the context of the pandemic: “Thinking about the amount of stress in your life, would you say that for the last 6 months within the context of COVID-19 your life has felt: not very, a bit, quite or very stressful?” If participants reported that they did not experience any stress related to COVID-19 they did not complete questions about coping. Because the responses to this question indicated a high prevalence of stress, this was used as a descriptive variable in our analysis.

Second, participants were asked to specify which strategies they used to cope with stress related to COVID-19 in the last 6 months that were effective for managing stress using a multiple-choice question, as follows: Which of the following have helped you to cope with stress related to the COVID-19 pandemic in the past 6 months? Participant’s responses included a pre-selected list of 17 coping strategies (e.g., connecting in-person with friends or family, exercising at home, limiting exposure to COVID-19 news), an open-ended question to list other strategies used (these responses were recoded into pre-defined strategies wherever possible), and an option for those who did not use any strategies to cope with stress. Given that the context of the COVID-19 pandemic has placed young adults in specific conditions that have affected their mental health and coping abilities (e.g., lockdown periods) [48], our survey question and list of coping strategies were designed from pre-existing measures used in a large survey assessing the mental health impacts of COVID-19 on the Canadian population [49]. The questionnaire of this Canadian survey was initially designed for a UK longitudinal survey on COVID-19 and mental health in 2020 [50], using a participatory-based approach involving individuals with previous mental health conditions, and was then adapted to reflect the Canadian context through a collaboration with researchers with expertise in mental health (including co-authors CR and EJ) and the Canadian Mental Health Association, a national mental health advocacy organization [49]. This initial list of coping strategies developed by Jenkins et al., 2021 (available here [49]) was then reviewed by a group of volunteers who are young adults via pre-testing of the questionnaire. Based on their feedback, we decided to remove some coping strategies (e.g., connecting with those in my household, accessing government supports) and combine some strategies together (e.g., exercising in my home, doing a hobby, learning or doing something new) to reduce the number of survey responses and facilitate completion of the coping question. The survey questions and list of coping strategies used in the FOCUS survey (both French and English versions) are available in Box A in S1 Text.

Third, participants were asked to assess how well they had coped with COVID-19-related stress using a 4-item scale with the response options: “very well”, “fairly well”, “not very well” to “not well at all”. We dichotomized participants’ responses into two groups to compare those who perceived coping well (“very well” or “fairly well”) and those who did not (“not very well” or “not well at all”). This was used as a first mental health-related outcome to examine the association between our coping classes and the perception of the positive effect of their coping strategies on their level of stress. Three other mental health-related outcomes were included in this analysis. Depressive symptoms were measured using the Patient Health Questionnaire-9 (PHQ-9) [51], which assesses the occurrence and severity of nine depressive symptoms over the past two weeks on a 4-point Likert scale from “not at all” (=0) to “nearly every day” (=3). Total scores range from 0 to 27. Scores of 0–4 indicate minimal, 5–9 mild, 10–14 moderate, 15–19 moderately severe, and 20–27 severe depressive symptoms. To reduce the likelihood of overestimating depression prevalence, a cut-off score of 15 was used to identify participants with major depressive symptoms [52]. The PHQ-9 scale showed good internal consistency in our samples (Canada: α = 0.89, France: α = 0.87). Anxiety symptoms were assessed using the 7-item Generalized Anxiety Disorder scale (GAD-7) [53], which reflects the frequency of symptoms of generalized anxiety in the past two weeks. For each symptom, participant respond on a 4-point Likert scale from “not at all” (=0) to “nearly every day” (=3). As defined in the original instrument, total scores range from 0 to 21, with scores of ≥5, ≥10, and ≥15 indicating mild, moderate, and severe anxiety levels, respectively [53]. In line with the depression scale, we included GAD-7 score as a binary outcome to identify participants with severe anxiety symptoms (cut-off score of 15). The GAD-7 scale demonstrated excellent internal consistency in our study samples (Canada: α = 0.93, France: α = 0.91). Suicidal ideation in the past 6 months was assessed by asking participants: “Have you considered suicide or taking your own life in the last 6 months?”.

Sociodemographic characteristics

We included in this analysis the following characteristics: age, area of residence, gender identity, sexual orientation, ethno-racial identity, living arrangements, and employment status. The age variable was dichotomized based on the median age in our total sample (i.e., 24 years) to differentiate between youth (18–24) and young adults (25–30). Ethno-racial identity data were gathered in different ways in the two countries. In Canada, we used the Canadian Institute for Health Information standards to collect ethno-racial identity [54]. Participants who identified with any ethno-racial identity (one or more) other than white were considered as “ethno-racial minority”. Given that collecting ethno-racial identity data is not allowed in France, we asked participants to provide the birth countries of their parents and grand-parents from both sides to estimate their cultural background. This approach has already been used as a proxy for ethno-racial identity in previous demographic studies conducted in France, including studies on mental health [55,56]. French participants were classified as “descendants of immigrants” if at least one parent or two grand-parents from the same side were born outside Metropolitan France or Europe, according to the French National Institute of Statistics and Economic Studies [57]. To evaluate the impact of the pandemic-induced economic hardship on mental health [9], participants were asked to indicate whether or not the pandemic had impacted their capacity to meet their financial essential needs (e.g., rent, utilities, groceries). Participants who reported a “moderate” or “major impact” were compared with those who selected “none” or “minor impact”. We included these covariates in our analysis based on existing research indicating that these sociodemographic characteristics are associated with coping behaviors and mental health challenges among youth [42,43,49].

Statistical analysis

First, we conducted an exploratory data analysis to reduce the high number of coping strategies listed in our questionnaire in order to improve interpretability and ensure that the most relevant and meaningful strategies were included. This preliminary step was a constant back-and-forth process between brainstorming sessions with our research team (i.e., inductive analytical approach) and descriptive statistical analysis (i.e., data-driven approach) [58]. We then decided to exclude strategies for accessing mental health and peer support services (i.e., “receiving in-person mental health support”, “virtually connecting with a mental health worker”, “contacting a support group”). While important, these strategies reflect a more complex and dynamic interplay between health services and young adults featuring concepts of help-seeking behaviors, mental health literacy and structural determinants (such as availability of services, administrative and financial barriers) [59]. We also excluded strategies that consisted of “keeping up to date with information” and “increasing use of social media” because they were inversely correlated with “limiting exposure to COVID-19 news” and “limiting exposure to social media”, respectively (see Table A in S1 Text). To guide our selection process of coping variables, we also performed a multiple correspondence analysis (MCA) using the PROC CORRESP procedure in SAS on Demand for Academics (SAS Institute Inc.) to graphically represent the relationships between variable categories [60]. In the MCA plot, variables that were positioned close to each other were considered as a similar set of strategies and were then grouped together (see Fig A S1 Text). Specifically, two combined groups of coping strategies were created: self-care and exercising (“maintaining a healthy lifestyle”, “going for a walk/exercise outside”, “exercising in my home/doing a hobby/learning or doing something new”) and community engagement (“having a supportive employer”, “volunteering to help”, “going to local businesses”). Finally, this exploratory analytical approach allowed us to identify six distinct coping strategies that were used in the LCA (see Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Description of the set of coping strategies selected after the MCA analysis.

https://doi.org/10.1371/journal.pmen.0000261.t001

Second, we conducted a LCA to identify classes of participants with similar coping strategies using the R package poLCA [61] and Mplus version 7.4 [62]. LCA is a widely used statistical procedure to identify qualitatively different subgroups (referred to as latent groups or classes) within populations that share similar characteristics based on observed categorical variables [63]. We tested 2-, 3-, 4-, 5, and 6-class models and allocated participants to the groups for which they had the highest membership probability. The number of latent coping classes was selected based on the overall interpretability and criteria including the Bayesian information criterion (BIC), sample-size adjusted Bayesian information criterion (SABIC), consistent Akaike information criterion (CAIC), and Lo-Mendell-Rubin adjusted (LMR-A) likelihood ratio test. Smaller values for the BIC, SABIC and CAIC indicate a better relative fit [64]. Although not used to select a final model, we also reported the Relative Entropy (RE) to provide additional information on model quality [65]. Furthermore, adequate latent class size (i.e., no less than 5% and no more than 50% of the total sample), parsimony, and the meaningfulness of the combination of coping strategies for each latent class were also considered when selecting the optimal model. To test the reliability of the identified classes, we carried out LCA in the French and Canadian sample separately to assess whether the same classes were found in different samples (see Fig B in S1 Text).

Third, we analyzed the association between sociodemographic characteristics and latent class membership using multinomial logistic regression to investigate predictors of membership in the class of participants that had the most difficulties adopting coping strategies. Lastly, we conducted multivariable logistic regression to examine the association between latent coping classes and mental health outcomes (i.e., perceiving coping well, major depressive symptoms, severe anxiety symptoms, and suicidal ideation in the last 6 months prior the survey) after adjusting for sociodemographic characteristics. These regression analyses were performed using a complete-case approach; participants with missing data for socio-demographics and participants who reported “prefer not to say” for sexual minority status and “other employment status” were excluded. Given that the Canadian and French samples differed significantly in terms of sociodemographic characteristics (see Table 2), all of these regression models were conducted separately using R (version 4.2.3) to identify context-specific differences between young adult’s experiences in Canada and France. This analysis plan was not preregistered.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Sociodemographic characteristics and coping strategies of young adults in Canada and France, FOCUS 2021 survey (N=4179).

https://doi.org/10.1371/journal.pmen.0000261.t002

Participants’ characteristics

Of the 2876 young adults in Canada and 3141 in France who started the FOCUS 2021 survey, more than 90% had complete sociodemographic data in each country (see Fig 1). Of these, we excluded those who did not respond to the question about the level of stress (Canada: n = 313, 11.8%; France: n = 598, 20.8%), those who selected the “I don’t know” response option (Canada: n = 12, 0.4%; France: n = 21, 0.7%), and those who reported having not experienced any stress related to COVID-19 (Canada: n = 36, 1.3%; France: n = 133, 4.6%). These sub-groups of participants excluded were compared to the participants with completed data on COVID-19 related stress (Canada: n = 2293, 86.4%; France: n = 2119, 73.8%) to describe differences in sociodemographic characteristics (see Table B in S1 Text). We then excluded participants who had missing data for perceived level of coping (n = 2) and for coping strategies (n = 12). French participants who did not complete the questions to assess their ethno-cultural origins were then excluded (n = 219, 10.4%). Our analysis sample included 4179 young adults with 2288 (54.7%) residing in Canada and 1891 (45.3%) in France.

As described in Table 2, most participants in both countries were women (Canada: 65%, France: 65.7%), resided in urban areas (Canada: 78.6%, France: 70%), and reported living with family/friends (Canada: 83.8%, France: 70.2%). A higher proportion of participants self-identified as a sexual minority in Canada (47.3%) compared to France (28.5%). In Canada, 18.4% self-identify as an ethno-racial minority, while in France, 37.3% were classified as descendants of immigrants. Regarding the employment status, a higher proportion of unemployed young adults was part of the French sample compared to the Canadian sample (11% versus 6.3%). The vast majority (71% in each setting) reported feeling very stressed and over 40% indicated that the pandemic had a moderate or major impact on their financial obligations in both countries. With regards to coping strategies, more than two-thirds in both settings reported connecting in-person with family/friends (76% in each setting), self-care and exercise (Canada: 80.2%, France: 74.3%), and about half indicated connecting virtually (55% in each setting), limiting exposure to COVID-19 news (Canada: 54.6%, France: 57.3%), and engaging in community activities to cope with COVID-19-related stress (Canada: 58.6%, France: 43.4%). In Canada, one third (35.2%) reported limiting their use of social media while they were 22.3% in France.

Latent coping classes

Table 3 presents the fit statistics for the LCA models. Whilst CAIC indice was slightly better for the three-class model compared to the four-class model, the BIC and SABIC indices showed convergence with the lowest values found for the model with four classes. The results of the LMR-A likelihood ratio test also supported the selection of the four-class model, as adding more classes did not significantly enhance the model fit. Compared to the three-class model, the four class-model identified a specific class of participants (class 3 in Fig 2) who reported moderate social connection and community engagement while using self-care and exercising and limiting exposure to social media and news. This class also provides a more nuanced picture to further explore the role of exposure to COVID-19 news and social media as coping strategies. However, it is important to note that the relative entropy of all models, including the four-class model, was lower than the standard cut-off of 0.6. Overall, given the interpretability of the different latent classes and the values of the information criteria (i.e., BIC, SABIC), which have been considered as the most reliable fit statistics in LCA, we selected the four-class model. When running LCA separately by country, similar classes were also obtained across our study samples (see Fig B in S1 Text).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Fit statistic for latent class models.

https://doi.org/10.1371/journal.pmen.0000261.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Class-specific probabilities from four-class model of coping strategies, FOCUS 2021 survey (n= 4179).

https://doi.org/10.1371/journal.pmen.0000261.g002

The graphic representation of the class-specific probabilities of reporting coping strategies for the four-class model is presented in Fig 2. Class 1 (n = 1435, 31.6%) was characterized by participants who had the highest probabilities of reporting social connection, self-care and exercise, and community engagement compared to all other classes. Class 1 had high probability of limiting exposure to COVID-19 news and social media, and was labeled the “high coping” class. Class 2 contained the largest latent class (n = 2134, 47%), and was characterized as the “socially engaged” class as it had high probabilities of connecting with family/friends (in-person or virtually) and engaging in community activities. However, half of participants in Class 2 reported limiting exposure to COVID-19 news, and almost none of them reduced their use of social media. Class 3 (n = 503, 11.1%) had high probabilities of self-care and exercise as well as limiting exposure to both news and social media. This was combined with moderate probabilities of endorsing social connection and community engagement. This class was interpreted as the “self-care and healthy lifestyle” class. Finally, Class 4, the smallest group (n = 469, 10.3%) was characterized by participants with lower probabilities of all six coping strategies compared to all other classes and was therefore labeled the “low coping” class.

A statistically significant difference was observed in the distribution of latent coping classes by country (p < 0.001). A greater number of participants in the Canadian sample were in the high coping class compared to the French sample (37.8% versus 27.2%). In France, more than half (52.5%) were classified in the socially engaged class while they were 42.6% in Canada. Similar proportions of participants were found in the self-care and healthy lifestyle class (Canada: 10.7%; France: 9.4%) and in the low coping class (Canada: 8.9%, France: 10.9%) in both countries.

Association between latent coping classes and sociodemographic characteristics

Tables 4 and 5 show the sociodemographic characteristics of the four latent coping classes and the results of the multinomial logistic regression analysis in each country. Compared to the high coping class, young men had increased odds of belonging to the low coping class (in both countries: Canada: Adjusted Odd Ratio (AOR) [95% Confidence Interval]: 1.63 [1.13–2.35]; France: 2.82 [1.96–4.05]) and the socially engaged class (only in France only: AOR: 1.33 [1.03–1.72]). In addition, living in rural area was associated with an increased likelihood of belonging to the self-care and healthy lifestyle class (in both countries: Canada; AOR: 1.46 [1.02–2.08], France; AOR: 1.58 [1.09–2.28]) and the low coping class (only in Canada; AOR: 1.88 [1.30–2.71]).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Sociodemographic characteristics associated with latent classes: results of multinomial logistic regression in the Canadian sample (N = 2174).

https://doi.org/10.1371/journal.pmen.0000261.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Sociodemographic characteristics associated with latent classes: results of multinomial logistic regression in the French sample (N = 1803).

https://doi.org/10.1371/journal.pmen.0000261.t005

Some associations differed by countries. In Canada, reporting a moderate or major financial impact due to COVID-19 was a predictor of membership of the self-care and healthy lifestyle (AOR: 1.41 [1.04–1.90]) and low coping classes (AOR: 1.41 [1.02–1.96]). Student participants (compared to employed youth) had an increased likelihood of belonging to the socially engaged class in Canada (AOR: 1.25 [1.01–1.54). In France, young adults who were classified as descendants of immigrants were more likely to belong to the low coping class (AOR: 1.57 [1.11–2.22]) while those who self-identified as a sexual minority were less likely to belong to the socially engaged class (AOR: 0.76 [0.60–0.96]).

Association between latent coping classes and mental health outcomes

The results of the logistic regression analyses (Table 6) show that in both countries, participants in the self-care and healthy lifestyle class had higher odds of experiencing major depressive symptoms (Canada: AOR 2.03 [1.45–2.83], France: AOR 1.84 [1.17–2.87]), and reporting suicidal thoughts (Canada: AOR 1.56 [1.13–2.16], France: AOR 1.88 [1.27–2.79]) while they were less likely to perceive coping well with COVID-19-related stress (Canada: AOR 0.65 [0.48–0.88], France: AOR 0.54 [0.38–0.78]). Similar associations were found among participants belonging to the low coping class in Canada. In France, participants in the low coping class had increased odds of reporting major depressive symptoms (AOR: 2.87 [1.89–4.36]) and suicidal thoughts (AOR: 1.60 [1.08–2.34]) than those in the high coping class. No significant association between perceiving coping well and belonging to the low coping class was observed in France. In Canada, young adults in the self-care and healthy lifestyle class were more likely to report severe anxiety symptoms than those in the high coping class (AOR: 1.41 [1.03–1.94]).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Results of separate multivariate logistic regression analyses examining the association between latent classes membership and mental health outcomes in the Canadian and French samples, FOCUS 2021 survey.

https://doi.org/10.1371/journal.pmen.0000261.t006