Original quantitative research – Risk and protective factors of solitary cannabis use among Canadian secondary school students: findings from the COMPASS study

Abstract

Introduction: Solitary cannabis use has become increasingly prevalent among youth in Canada and is linked to both mental and behavioural concerns. This study seeks to fill current gaps in the literature by examining the factors associated with solitary cannabis use to understand the complexities of specific contexts in which use occurs and improve prevention efforts targeted towards behaviours that are associated with greater health impacts and riskier behaviours.

Methods: This study examined solitary cannabis use among a cross-sectional sample (n = 2899) of Canadian youth (11–18 years) who reported past 30-day cannabis use and participated in the COMPASS study in 2020–21. Using a stepwise approach, generalized linear mixed models were run to examine risk and protective factors associated with solitary use. A Bayesian imputation framework was used to address missing data.

Results: Sixty-six percent of students reported having used cannabis alone in the past 30 days. All substance use variables examined were strongly associated with solitary use. A preliminary interaction between school connectedness and flourishing was observed, suggesting school connectedness may offer modest protection against solitary cannabis use and vary across flourishing scores.

Conclusion: Solitary cannabis use is common among Canadian youth and is associated with several other risk factors examined. These findings suggest that comprehensive prevention strategies that promote both flourishing and equitable school environments may be relevant to reduce solitary cannabis use and its associated risks. Future research should explore how behaviours may cluster with solitary use and employ longitudinal designs to understand how behaviours change over time.

Keywords: cannabis, substance use, youth, mental health, school health services

Introduction

Cannabis use among Canadian youth is widespread and has become increasingly normalized since the federal legalization of cannabis (the Cannabis Act) in 2018.^{Footnote 1} Adolescents are uniquely vulnerable to cannabis-related harms, and such behaviours, including frequent use, early initiation, poly-use, and consumption context, have been linked to greater risks of problematic substance use, dependency, and negative psychosocial outcomes.^{Footnote 2Footnote 3} One risk behaviour that has become increasingly prevalent among North American youth is solitary cannabis use (i.e. using cannabis alone rather than in a social context).^{Footnote 4Footnote 5} In Canada, it has been estimated that over one-third of youth who have used cannabis in the past year use alone.^{Footnote 6} Solitary cannabis use (SCU) may pose its own risks due to fewer social precautions or be an important proxy for evaluating related risk. Reflecting on Cooper’s motivational model of substance use, non-social use is often internally driven and coping-motivated,^{Footnote 7} aligning with problematic use patterns and maladaptive regulation. Beyond coping motives, solitary substance use has also been linked to poorer mental health, other risky substance use indicators, and sustained substance use problems into adulthood.^{Footnote 8Footnote 9Footnote 10Footnote 11Footnote 12Footnote 13Footnote 14Footnote 15Footnote 16} Recent literature examining cannabis use disorder risk among young adults in Canada found that participants at high-risk for cannabis use disorder had particularly elevated SCU prevalence rates compared to low-risk participants.^{Footnote 17} While various risks and long-term health outcomes have been identified, research on youth SCU remains extremely limited.

Social factors have a central role in influencing adolescent cannabis use and function across various levels of influence. A systematic review examining the multi-level influences of cannabis demonstrates that peer disapproval and parental monitoring emerged as strong protective factors, while peer norms and parental acceptance of use were associated with increased risk of cannabis use; similarly, school connectedness served as a protective factor.^{Footnote 18} However, COVID-19-related restrictions in 2020–2021, specifically school closures, altered how youth interacted with their home and school environments. Notably, social and environmental supports that youth typically rely on, including friends, teachers, and school programs, were disrupted, resulting in increased internalizing symptoms and substance use reported by youth.^{Footnote 19} As cannabis is typically consumed in social settings or at school,^{Footnote 1} using alone deviates from more typical or experimental patterns seen in adolescence, representing a fundamentally different behavioural pattern. The extent to which these social influences similarly shape non-social use remains poorly understood. Therefore, characterizing SCU is critical to informing future research and surveillance initiatives, as well as targeted approaches to mitigate related risks, especially for vulnerable populations such as youth.

Since the enactment of the Cannabis Act, Canadian youth have reported increased normalization of cannabis, greater access, and reduced risk perceptions.^{Footnote 1Footnote 20} Further, COVID-19-related restrictions introduced additional contextual and social changes that may have influenced youth substance use behaviours. While research indicates the prevalence of cannabis use remained stable during COVID-19,^{Footnote 21} youth more commonly reported using cannabis to cope and manage their mental health,^{Footnote 22Footnote 23} suggesting that self-isolation may have had an impact on how and why youth consume cannabis. The COMPASS study (COMPASS)^{Footnote 24} is a prospective cohort study (2012–2027) that collects longitudinal health behaviour data from Canadian secondary school students, and was designed to evaluate ongoing real-world changes in programs, policies, and resources for schools, communities, and provinces. Using cross-sectional data from COMPASS, this study examined SCU and associated risk and protective factors among a large sample of Canadian secondary school students during the COVID-19-related school closures. Specifically, the objectives were to 1) estimate the prevalence of SCU among participating youth who reported cannabis use in the prior 30 days; and 2) examine the demographic, psychological, behavioural, and social support correlates of SCU.

Methods

Study design

COMPASS collects data annually from a rolling cohort of students attending a convenience sample of secondary schools in Alberta, British Columbia, Ontario, and Quebec, Canada. Ethical approval was obtained from the University of Waterloo Human Research Ethics Committee (ORE#30118), Brock University Research Ethics Committee (REB#23-087), CIUSSS de la Capitale-Nationale–Université Laval (#MP-13-2017-1264), University of Alberta Research Ethics Office (Project #00040729), University of British Columbia (H17-00167), and participating school boards. This study utilized cross-sectional student data from the 2020–21 wave of COMPASS (2020–21 host sample: n = 45 329).^{Footnote 24} Student questionnaires were administered online with an initial survey link and one reminder emailed by school administrators. All students were invited to participate using active-information passive-consent parental permission protocols^{Footnote 25} and could decline or withdraw at any time (student response rate: 58%).

Study sample

Secondary school students (11–18 years) reporting past-month cannabis use during the 2020–21 school year were included. This subsampling is necessary as SCU was assessed based on past 30-day cannabis use (N = 3153). Students with undefined data were removed, including those unsure about solitary cannabis use (N = 159), those who declined to report their gender (N = 64), and Quebec students who were in non-traditional academic streams (N = 48). The final analytical sample size was 2899 students.

Study measures

Dependent variable

To measure SCU, students were asked: “In the last 30 days, how many times have you: Used marijuana or cannabis when you were all by yourself?” with response options of “never,” “once,” “twice,” “three or more times,” and “I don’t know.” Given the low proportion of students who used solitary “once” or “twice,” a binary measure of SCU was used based on students reporting having used alone at least once in the past 30 days (yes = 1) or “never” (no = 0). Students reporting “I don’t know” were removed from the analytical sample (n = 159).

Independent variables

Independent variables were examined for their association with SCU. Variables were selected based on existing literature examining youth cannabis use^{Footnote 6Footnote 18Footnote 26Footnote 27Footnote 28Footnote 29} and organized hierarchically based on multi-level behavioural frameworks.^{Footnote 30}

Sociodemographic characteristics

Sociodemographic characteristics included in the analyses are as follows: gender (“female,” “male,” “I describe my gender in a different way,” or “I prefer not to say”), grade (9,10,11,12, and “Secondary I–V” and “other” [for schools in Quebec only; Secondary I and II are equivalents to Ontario Grades 7 and 8, “other” represents students who fall within a class with no official grade equivalent]), and, race (White, Black, Asian, First Nations (status and non-status), Métis, Inuit, Latin American or Hispanic, and/or “other”), and perceived family affluence compared to their peers (“more comfortable,” “as comfortable,” and “less comfortable”) as a proxy for socioeconomic status (SES).

Within the analytical models, province was excluded due to being highly correlated with grade (e.g. Secondary I–II is unique to Quebec) and students who described their gender “in a different way” were excluded due to small cell counts. Race was recoded into racialized (e.g. Black, Asian, Indigenous, Latin American, other) vs. non-racialized (White).

Psychological characteristics

Depressive symptoms were assessed using the Centre for Epidemiologic Studies Depression Scale (Revised)-10 (CESD-R10;^{Footnote 31}). Students responded to questions about depressive symptoms such as feelings of sadness, loss of interest, and difficulties sleeping over the past week.^{Footnote 31Footnote 32Footnote 33} Anxiety symptoms were measured using the Generalized Anxiety Disorder 7-item Scale (GAD-7;^{Footnote 34}). Students responded to questions about having symptoms of anxiety such as feelings of worry, fear, and irritability over the past 2-week period. Scores for CESD-R-10 and GAD-7 ranged from 0 to 30 and 0 to 21, respectively, where higher sum scores indicated greater frequency of symptoms. Consistent with other research, this study used scores greater than or equal to 10 for both depression^{Footnote 33} and anxiety^{Footnote 34} scales to indicate clinically relevant symptoms. Shortened versions of the depression^{Footnote 33} and anxiety^{Footnote 34} scales have been validated against the original scales for use in adolescent populations. Internal consistency of the CESD-R-10 (α = 0.99) and GAD-7 (α = 0.98) scales were high.

Student psychosocial well-being was assessed using Diener’s Flourishing Scale.^{Footnote 35Footnote 36} Students were asked to report on their perceived successes in various areas of psychosocial well-being. The revised and validated^{Footnote 37} five-point response scale from strongly disagree to strongly agree was used to rate perceived psychosocial strengths and resources for support, with higher sum scores (range: 8–40) indicating greater flourishing.^{Footnote 36} Internal consistency of the Flourishing Scale was high (α = 0.98).

Emotional dysfunction was assessed using a shortened version of the Difficulties in Emotion Regulation Scale (DERS).^{Footnote 38} Students were asked to report on their perceived emotional regulation on six items, including one item from each subscale: nonacceptance of difficult emotions, difficulties engaging in goal-directed behaviours, impulse control difficulties, emotional awareness, access to emotion regulation strategies, and emotional clarity. A five-point Likert scale from strongly disagree to strongly agree was used to rate perceived emotional dysregulation with higher sum scores (range: 6–30) indicating greater emotional dysfunction.^{Footnote 37} Internal consistency of the DERS was high (α = 0.99).

Substance use characteristics 

To measure cannabis use frequency, students were asked: “In the last 12 months, how often did you use marijuana or cannabis? (a joint, pot, weed, hash).” Response options appropriate for this study included: “once a month,” “2 or 3 times a week,” “4 to 6 times a week,” and “every day.” The frequency of use among students who report current cannabis use was assessed as “monthly,” “weekly,” and “daily” use.

To measure increases in cannabis use during COVID-19, students were asked: “How has your life changed because of COVID-19? My cannabis use has…:” where the response options included: “increased,” “stayed the same/not applicable,” and “decreased.” Increased cannabis use due to COVID-19 was assessed as a binary measure [increased = 1, not increased = 0].

To measure mode of cannabis use, students were asked: “If you have used marijuana or cannabis in the last 12 months, how did you use it?” Response options included: “I have used it by smoking it,” “I have used it by vaping it,” and “I have used it by eating or drinking it.” Students were asked to select all that applied. All three modes of use were examined as separate binary variables (i.e. mode of use (smoking, vaping, eating [yes = 1, no = 0]).

To measure binge drinking, students were asked to answer: “In the last 12 months, how often did you have 5 drinks of alcohol or more on one occasion?” Response options ranged from “never” to “daily or almost daily.” Student responses were assessed as a binary measure based on past-30-day binge drinking [yes = 1, no = 0].

To measure solitary alcohol use, students were asked: “In the last 30 days, how many times have you: Drank alcohol when you were all by yourself?” with response options of “never, once, twice, three or more times, and I don’t know.” Student responses were coded as a binary measure based on past-30-day solitary alcohol use [yes = 1, or no = 0] and “I don’t know” was categorized with missing data.

Driving under the influence (of alcohol and/or cannabis) was assessed with two questions about riding in a vehicle in the past 30 days with a driver (self or others) who had used alcohol or cannabis. Responses (“never” to “three or more times” or “I don’t know”) were coded as: 1 = yes to either substance, 0 = no to both, and 2 = mixed/uncertain responses (including “I don’t know”).

Social characteristics

A six-item version of the National Longitudinal Study of Adolescent Health School Connectedness Scale was used to assess student-perceived school connectedness.^{Footnote 39Footnote 40} Students were asked to report their perceived connection to school (e.g. feeling close to people at school, membership, feeling happy and safe at school). A four-point Likert scale from strongly disagree to strongly agree was used to rate perceived school connectedness, with higher sum scores (range: 6–24) indicating stronger school connectedness.^{Footnote 39} Internal consistency of the School Connectedness Scale was high (α = 0.99).

Two items from the Multidimensional Scale of Perceived Social Support and an additional item to assess the home environment were used to examine the self-perceived adequacy of supportive relationships and home environment.^{Footnote 41} Students were asked to report agreement with the following statements? “I can talk about my problems with my family,” “I can talk about my problems with my friends,” and “I have a happy home life” on a five-point Likert scale from strongly disagree to strongly agree. Responses were dichotomized (agree = 0, disagree = 1). Students reporting neutrality were categorized as “agree” to reduce potential misclassification and to err on the side of inclusion, as research on Likert scales suggests that neutral responses reflect a lack of strong feeling in either direction, rather than the absence of the attribute being measured.^{Footnote 42}

Analysis

Descriptive statistics were used to estimate the proportion of students who reported SCU. Univariate tests (chi-square and one-way ANOVA) were used to examine differences in SCU by student sociodemographic, psychological, behavioural, and social characteristics. Using a stepwise approach, four generalized linear mixed models were run with random intercepts at the school-level. Model 1 examined the association between sociodemographic characteristics (gender, grade, ethnicity, SES) and SCU. The following models involved the stepwise addition of psychological (Model 2), behavioural (Model 3), and social (Model 4) characteristics. This approach demonstrated the evolution of associations between variables as new variables were added, and changes in strength and statistical significance. Post-hoc testing for product-term interactions was conducted based on Model 4 results, and a significant interaction was observed, resulting in a final interaction model, Model 5. Data cleaning and formatting were completed in SAS Studio.^{Footnote 43} Generalized linear mixed models were run within a Bayesian imputation framework^{Footnote 44} in R Studio^{Footnote 45} using the JointAI package,^{Footnote 46} and imputed missing data on dependent and independent variables. Sample sizes between Table 1 and Table 2 are discordant, as descriptive results utilized a complete case sample (n = 2822), and linear models used imputation (n = 2899).

Results

Sample descriptives and SCU

Within the analytic sample of 108 participating schools, 2822 students reported using cannabis in the past month, and 1853 of these students (66%) reported SCU. Half of the students identified as female, the majority were non-racialized (71%), and reported being as financially comfortable compared to the average student in their class (59%). The largest proportion of students (48%) resided in Ontario. The intraclass correlation coefficient (ICC) for SCU (ICC: 4.8%) suggests low between-school variation in student SCU.

Of students reporting SCU, 30% reported having used alone once or twice in the past 30 days (n = 552), while 70% reported having used alone three or more times (n = 1303). Student characteristics by SCU were examined descriptively (Table 1). The proportion of students reporting SCU increased with grade. Among students reporting solitary use, 50% were female, 40% were male, and 10% described their gender differently. Thirty-three percent of people who used cannabis alone were racialized, while 68% were non-racialized. SCU was most common among students with perceived average SES (58%), compared to those with perceived higher SES (24%) and lower SES (18%).

Students who reported SCU demonstrated worse psychological characteristics compared to students who did not report SCU across all measures. Compared to students who did not report SCU, a greater proportion of students who reported SCU indicated clinically relevant depression scores (71.8% vs. 60.2%) and anxiety scores (53.5% vs. 42.5%), as well as lower mean flourishing (27.5 vs. 30.1) and higher emotional dysregulation (18.0 vs. 16.7). These unadjusted differences were statistically significant across all psychological characteristics. Related to other substance use behaviours, 40% of students who reported SCU also used cannabis on a weekly basis, 34% reported daily use, and 73% reported increased cannabis use during COVID-19. Students reporting SCU also reported high rates of solitary alcohol use (61%), driving under the influence of cannabis and/or alcohol (46%), and cannabis consumption by smoking (90%), vaping (49%), and eating/drinking (50%).

Students who reported SCU had poorer perceptions of the various social supports available across all measures. Mean school connectedness was notably lower among students reporting SCU (15.8 vs. 17.4) and were also more likely to disagree that they had a happy home life (20.0% vs. 12.9%), report lacking family support (38.0% vs. 30.2%) and insufficient peer support (16.0% vs. 9.4%), compared to students who did not report SCU.

Model results

Stepwise model results predicting students’ SCU are presented in Table 2. In the final model (Model 5), students in Secondary I and II (OR: 0.50 [95% CI: 0.33–0.77]) had a lower likelihood of SCU (due to inclusion of Secondary I and II, the Quebec sample is younger compared to other provinces). Gender-diverse students had a higher likelihood of SCU than females (OR: 2.08 [1.26–3.50]), and racialized students had a higher likelihood of SCU (OR: 1.43 [1.13–1.81]) compared to non-racialized students. Students who reported being as financially comfortable (OR: 1.27 [1.02–1.59]) or less financially comfortable (OR: 1.55 [1.10–2.19]) had a higher likelihood of SCU compared to their peers who were more financially comfortable. All substance use variables were strongly and positively associated with SCU, except for binge drinking. None of the mental health variables examined were independently associated with SCU in the final model (Model 5). An interaction between flourishing and school connectedness was selected for post-hoc exploration based on both theoretical and empirical grounds. Theoretically, flourishing and school connectedness represent complementary protective influences, suggesting a plausible moderating relationship, which was further supported by the diminished effect size of flourishing after school connectedness was added in Model 4. A small but significant interaction was observed between flourishing and school connectedness (OR: 1.00 [0.99–1.00]) (Figure 1). Figure 1 demonstrates that higher school connectedness is linked to a lower likelihood of SCU, varying by flourishing quantiles. The negative estimate (β_(FLRSH*SC) = −.00431) implies that as flourishing increases, the slope of school connectedness becomes more negative, resulting in a greater reduction of SCU probability. Flourishing and school connectedness were moderately and positively correlated (r = 0.54, 95%CI [0.51, 0.56], p < .001). While effects were significant, the correlation between both variables may explain the small effect size observed.

Figure 1 - Text description

Figure 1 illustrates the predicted likelihood of solitary cannabis use (SCU), based on a two-way interaction between school connectedness and flourishing. School connectedness (ranging from 6 to 24) is plotted on the x-axis, and predicted probability of SCU is on the y-axis. Three lines represent the interaction at low, median and high flourishing score quartiles (corresponding to the 25th [score: 21.5], 50th [score: 28.5], and 75th [score: 35.5] percentiles), each with a shaded 95% confidence band. The linear equations for each line illustrated are as follows: Flourishing at 21.5: y = 0.195 − 0.0019x (R² = 0.94); Flourishing at 28.5: y = 0.230 − 0.0053x (R² = 0.98); and Flourishing at 35.5: y = 0.260 − 0.0080x (R² = 0.98). This translates into the following data:

School connectedness	Flourishing = 21.5 predicted (95% CI)	Flourishing = 28.5 predicted (95% CI)	Flourishing = 35.5 predicted (95% CI)
6	0.18 (0.09, 0.33)	0.20 (0.11, 0.35)	0.22 (0.11, 0.39)
7	0.18 (0.09, 0.32)	0.20 (0.11, 0.33)	0.21 (0.11, 0.37)
8	0.18 (0.09, 0.32)	0.19 (0.11, 0.32)	0.20 (0.11, 0.35)
9	0.18 (0.10, 0.31)	0.18 (0.10, 0.31)	0.19 (0.10, 0.33)
10	0.18 (0.10, 0.30)	0.18 (0.10, 0.29)	0.18 (0.10, 0.31)
11	0.17 (0.10, 0.29)	0.17 (0.10, 0.28)	0.17 (0.09, 0.29)
12	0.17 (0.10, 0.29)	0.17 (0.10, 0.27)	0.16 (0.09, 0.27)
13	0.17 (0.10, 0.28)	0.16 (0.09, 0.26)	0.15 (0.08, 0.25)
14	0.17 (0.10, 0.28)	0.15 (0.09, 0.25)	0.14 (0.08, 0.24)
15	0.17 (0.09, 0.28)	0.15 (0.09, 0.24)	0.13 (0.08, 0.22)
16	0.16 (0.09, 0.27)	0.14 (0.08, 0.24)	0.13 (0.07, 0.21)
17	0.16 (0.09, 0.27)	0.14 (0.08, 0.23)	0.12 (0.07, 0.20)
18	0.16 (0.09, 0.27)	0.13 (0.08, 0.22)	0.11 (0.06, 0.19)
19	0.16 (0.09, 0.27)	0.13 (0.07, 0.21)	0.10 (0.06, 0.18)
20	0.16 (0.09, 0.27)	0.12 (0.07, 0.21)	0.10 (0.06, 0.17)
21	0.15 (0.08, 0.27)	0.12 (0.07, 0.20)	0.09 (0.05, 0.16)
22	0.15 (0.08, 0.27)	0.12 (0.06, 0.20)	0.09 (0.05, 0.15)
23	0.15 (0.08, 0.27)	0.11 (0.06, 0.19)	0.08 (0.04, 0.14)
24	0.15 (0.07, 0.28)	0.11 (0.06, 0.19)	0.08 (0.04, 0.14)

These approximate linear equations are visual/descriptive approximations only, intended to characterize the general trend of each line as plotted. Because the underlying model is a logistic regression, the relationship is linear on the log-odds scale but curved on the probability scale shown in the figure; the linear equations on the probability scale only approximates this relationship.

Note: Figure 1 was created for illustrative purposes, where values are based on a complete case analysis and not the imputed values reported in Table 2, Model 5. However, the direction and strength of association are consistent between both outputs.

Discussion

This study found that, during COVID-19 and related school closures, SCU was highly prevalent among Canadian secondary school students who reported past-30-day cannabis use. Notably, 64% of students in this sample reported engaging in SCU, which is consistent with existing literature.^{Footnote 6Footnote 8Footnote 9} During this time, nearly three-quarters of students who reported SCU reported that they had used cannabis alone three times or more in the last month, indicating a higher frequency pattern of use in isolation. Substance use behaviours examined in this study were consistently and strongly associated with SCU. Preliminary evidence for a moderating effect of flourishing on the protective relationship between school connectedness and SCU was observed, despite the unique context of school closures, where conditions may have been conducive to diminished connection and solitary use. These findings fill a gap in current research, reinforcing the importance of schools even during COVID-19-related restrictions, and warrant the need for future research to move beyond general cannabis use patterns among adolescents and examine the specific contexts in which use occurs, especially those that are associated with greater health impacts and riskier behaviours.

Who is at risk of SCU?

Consistent with existing research, SCU clustered with other substance use behaviours, indicating broader patterns of risk.^{Footnote 5Footnote 9Footnote 15Footnote 47} SCU was more likely among students who increased their cannabis use during COVID-19, reflecting how social isolation, school closures and disrupted routines may influence substance use. Prior research has highlighted the protective role of structured environments,^{Footnote 48} and these findings are aligned in demonstrating that stronger school connectedness is associated with reduced SCU. These findings are particularly relevant given that school closures may have altered student connectedness that would typically be protective against risky substance use behaviours, possibly influencing SCU observed in this study.

SCU was more prevalent among marginalized youth—specifically racialized students, gender-diverse students, and those with perceived lower SES—highlighting important equity considerations in school-based substance use protection strategies. These populations have historically faced exclusion from supportive school environments, which may compound their risk for solitary and problematic cannabis use.^{Footnote 49} These findings reinforce the importance of equity-informed approaches when implementing prevention/intervention strategies.^{Footnote 50}

What is protective against SCU?

These findings point to a potentially nuanced but protective relationship between school connectedness, flourishing, and SCU. Students with higher levels of flourishing experienced stronger protective effects from school connectedness, with the steepest decline in solitary use observed in the top quartile of flourishing scores. This suggests that students who report higher flourishing may receive greater benefits from the positive social and relational aspects of school environments, and in turn are less likely to use cannabis alone. Importantly, these preliminary findings suggest that the protective association between school connectedness and SCU may vary by flourishing level, and interventions that solely focus on increasing school connectedness may be insufficient for some youth. Instead, integrated approaches, such as school-based well-being programs, may be necessary to support these students and reduce SCU. This interaction offers preliminary evidence of a complex interplay between individual, interpersonal, and environmental factors in shaping youth substance use behaviour, and indicates a need for further research to clarify the nature and magnitude of these relationships.

Non-predictive correlates of SCU

Contrary to previous research,^{Footnote 10Footnote 11Footnote 12Footnote 16} SCU was not independently significantly associated with symptoms of depression or anxiety. These null findings may suggest that generalized measures of depression, anxiety, and emotional regulation may not adequately capture the motivations or psychosocial contexts that underlie SCU. In contrast, flourishing, was predictive in earlier models and significantly interacted with school connectedness. This aligns with prior research in the present sample, demonstrating that positive well-being measures are more closely associated with youth substance use behaviours compared to traditional mental illness scales.^{Footnote 51} Future SCU research should incorporate more targeted mental health measures (e.g. social anxiety, coping motives) to better understand unique drivers that may explain these behaviours and motivations more clearly.

Recommendations for future research and prevention

Consistent with existing research,^{Footnote 52} our findings suggest that school connectedness can mitigate health risks during adolescence and appears most impactful for vulnerable students (e.g. less resourced/unsupportive home environments). Yet, these same populations often report feeling less connected to their schools, highlighting a critical gap for equitable prevention strategies. While enhancing school connectedness is necessary, it is insufficient to reduce SCU on its own; comprehensive approaches that support student well-being and flourishing, especially for marginalized or at-risk youth, are essential.^{Footnote 53Footnote 54}

From a population-health lens, the ICC (4.84%) indicates school-level factors play a small but meaningful role in shaping SCU patterns. Where data is available, future research may explore specific school-level characteristics, such as inclusivity policies, staff-student relationships, extracurricular offerings, or available supports, to explain possible between-school differences in student SCU. Additionally, longitudinal research is needed to understand how SCU patterns evolve, and if the frequency of using cannabis alone presents with different risk factors, especially in response to COVID-related restrictions and changes in students’ perceived connection with their school and substance use behaviours over time. Given that SCU behaviours may be partially driven by isolating conditions of COVID-19, it remains unclear whether SCU prevalence and associated characteristics observed in this study reflect a behavioural pattern that will persist in post-COVID-19 conditions, or if results are shaped by these unique circumstances. As such, continued monitoring and surveillance for these high-risk behaviours is required in a post-pandemic era.

Strengths and limitations

A key strength of this study is the use of a large, population-based sample that enabled robust analysis of multiple risk and protective factors for SCU. The breadth of the dataset also allowed for consideration of equity-relevant variables and complex interactions between student characteristics and school environment. Further, this study managed missing data using a Bayesian imputation framework to mitigate biases common in complete case analyses.

However, limitations remain. SCU data were unavailable for the period immediately before COVID-19, and SCU was dichotomized due to response distributions; future research may consider how predictors of SCU vary by frequency where data are available. Additionally, COMPASS does not collect information on intentions or motivations for using alone, nor the context of isolation (e.g. with friends online). These data challenges are especially relevant for students reporting “I don’t know” to SCU, potentially reflecting uncertainty about product content (e.g. tetrahydrocannabinol [THC]/cannabinol [CBD]) or ambiguity in interpreting the question. Focus group interviews would help clarify participant understanding, reasons for response, and inform improvements to this measure for future use. Notably, data were collected during COVID-19, when social restrictions and school closures may have structurally changed the way in which youth used cannabis; SCU patterns observed in this study may be a result of mandated isolation, thus limiting the generalizability of these findings to non-pandemic contexts.

Because these data are cross-sectional, directionality between SCU and correlates cannot be inferred. However, the current cross-sectional data from COMPASS (2020–21) represent the first wave of SCU available from the full sample of COMPASS students. This study had a response rate of 58%, which may be due to data collection challenges during COVID-19, when school-based research participation rates declined. Nevertheless, selection bias is possible as non-responding students may differ systematically from respondents on key variables such as school connectedness and cannabis use. Results from this exploratory cross-sectional study identify the variables associated with SCU and provide a foundation for future longitudinal research. These findings are not generalizable to all Canadian youth, although it is worth noting that COMPASS was not designed to be nationally representative, and evidence suggests its large convenience sample aligns with other similar nationally representative data.^{Footnote 55}

Conclusion

This study, which takes place during the COVID-19-mandated school closures, highlights the prevalence of SCU among Canadian secondary school students, with the majority of those who report current use reporting frequent SCU. Solitary use was more common among students engaging in other risky behaviours, those who increased use during COVID-19, and marginalized youth. Further, school connectedness emerged as a key protective factor and appeared to vary across flourishing levels. These findings highlight the potential value of considering comprehensive prevention strategies and warrant the need for additional research on interventions that promote both flourishing and equitable school environments for reducing SCU and its associated risks.

Acknowledgements

The COMPASS study has been supported by a bridge grant from the CIHR Institute of Nutrition, Metabolism and Diabetes (INMD) through the “Obesity – Interventions to Prevent or Treat” priority funding awards (OOP-110788; awarded to SL), an operating grant from the CIHR Institute of Population and Public Health (IPPH) (MOP-114875; awarded to SL), a CIHR project grant (PJT-148562; awarded to SL), a CIHR bridge grant (PJT-149092; awarded to KP/SL), a CIHR project grant (PJT-159693; awarded to KP), and by a research funding arrangement with Health Canada (#1617-HQ-000012; contract awarded to SL), a project grant from the CIHR Institute of Population and Public Health (IPPH) (PJT-180262; awarded to SL and KP). The COMPASS-Quebec project additionally benefits from funding from the Ministère de la Santé et des Services sociaux of the province of Quebec, and the Direction régionale de santé publique du CIUSSS de la Capitale-Nationale. KAP is the Canada Research Chair in Child Health Equity and Inclusion. The authors would like to thank the schools, school boards, and students that have participated in the COMPASS study, and all COMPASS team members, staff, partners, and youth engagement committee members.

Conflicts of interest

None.

Authors’ contributions and statement

• AB: Conceptualization, methodology, writing—original draft, review and editing, formal analysis.
• AD: Writing—review and editing, formal analysis. 
• KS: Writing—review and editing.
• KAP: Writing—review and editing, funding acquisition.
• CP: Writing—review and editing.
• SL:  Writing—review and editing, supervision, investigation, funding acquisition, data curation.

The content and views expressed in this article are those of the authors and do not necessarily reflect those of the Government of Canada.

Previous | Table of Contents