The Canadian Journal of Psychiatry / La Revue Canadienne de Psychiatrie
Article reuse guidelines:
sagepub.com/journals-permissionsDOI: 10.1177/07067437221111368TheCJP.ca | LaRCP.ca
Background: Mental health concerns are common among university students and maybe elevated among those with specific risk exposures. The study examined the association between childhood adversities and mental health outcomes among undergraduate university students and assessed whether psychosocial and behavioral factors mediate those associations.
Methods: The Queen’s University Student Well-Being and Academic Success Survey identified two large cohorts of first-year undergraduate students entering university in Fall 2018 and 2019 (n = 5,943). At baseline, students reported sociodemographic information, family-related mental health history, childhood physical abuse, sexual abuse, peer bullying, and parental separation or divorce. Baseline and follow-up surveys in Spring 2019, Fall 2019, and Spring 2020 included validated measures of anxiety (7-item Generalized Anxiety Disorder) and depressive symptoms (9-item Patient Health Questionnaire), non-suicidal self-harm, and suicidality, along with psychological processes and lifestyle variables. Repeated measures logistic regression using Generalized Estimating Equations was used to characterize the associations between childhood adversities and mental health outcomes and examine potential mediation.
Results: Adjusting for age, gender, ethnicity, familial mental illness, and parental education, any childhood abuse (odds ratio: 2.89; 95% confidence interval, 2.58 to 3.23) and parental separation or divorce (odds ratio: 1.29; 95% confidence interval, 1.12 to 1.50) were significantly associated with a composite indicator of mental health outcomes (either 9-item Patient Health Questionnaire score ≥10 or 7-item Generalized Anxiety Disorderscore ≥10 or suicidality or self-harm). The association with childhood abuse weakened when adjusted for perceived stress, self-esteem, and insomnia (odds ratio: 2.05; 95% confidence interval, 1.80 to 2.34), and that with parental divorce weakened when adjusted for self-esteem (odds ratio: 1.17; 95% confidence interval, 1.00 to 1.36).
Conclusion: Childhood abuse and parental separation or divorce were associated with mental health concerns among university students. Childhood adversities may impact later mental health through an association with stress sensitivity, selfesteem, and sleep problems. The findings suggest that prevention and early intervention focusing on improving sleep, selfesteem, and coping with stress while considering the individual risk profile of help-seeking students may help support student mental health.
Les problèmes de santé mentale sont répandus chez les étudiants universitaires et peuvent être élevés chez ceux ayant des expositions à un risque spécifique. La présente étude a examiné l’association entre les adversités vécues dans l’enfance et les résultats de santé mentale chez les étudiants universitaires de premier cycle, et évalué si les facteurs psychosociaux et comportementaux servent de médiateurs dans ces associations.
Méthodes: L’enquête de l’Université Queen’s sur le bien-être et la réussite scolaire des étudiants a identifié deux vastes cohortes d’étudiants de première année du premier cycle commençant l’université à l’automne de 2018 et 2019 (n = 5,943). Au départ, les étudiants ont rapporté de l’information sociodémographique, des antécédents de santé mentale liés à la famille, la violence physique dans l’enfance, l’abus sexuel, l’intimidation par les pairs et la séparation ou le divorce des parents. Les sondages au départ et au suivi du printemps 2019, de l’automne 2019, et du printemps 2020 incluaient des mesures validées de l’anxiété (GAD-7) des symptômes dépressifs (PHQ-9), de l’autodestruction non suicidaire, et de la suicidabilité, de même que des processus psychologiques et des variables de style de vie. Des mesures répétées de régression logistique utilisant des Équations d’estimation généralisées ont servi à caractériser les associations entre les adversités vécues dans l’enfance et les résultats de santé mentale, et à examiner la médiation potentielle.
Résultats: Après ajustement pour l’âge, le sexe, l’ethnicité, la maladie mentale familiale, et l’éducation parentale, tout abus dans l’enfance (RC 2,89; IC à 95% 2,58 à 3,23) et la séparation ou le divorce des parents (RC 1,29; IC à 95% 1,12 à 1,50) étaient significativement associés à un indicateur composite des résultats de santé mentale (soit un score PHQ-9 ≥ 10 ou un score GAD-7 ≥ 10 ou la suicidabilité ou l’autodestruction). L’association avec l’abus dans l’enfance faiblissait quand elle était ajustée pour le stress perçu, l’estime de soi, et l’insomnie (RC 2,05; IC à 95% 1,80 à 2,34), et avec le divorce parental, l’association faiblissait lorsqu’ajustée pour l’estime de soi (RC 1,17; IC à 95% 1,00 à 1,36).
Conclusion: L’abus dans l’enfance et la séparation ou le divorce des parents étaient associés à des problèmes de santé mentale chez les étudiants universitaires. Lesadversités vécuesdansl’enfance peuvent toucher ultérieurement la santé mentale au moyen d’une association avec la sensibilité au stress, l’estime de soi et des problèmes de sommeil. Les résultats suggèrent que la prévention et l’intervention précoce axées sur l’amélioration du sommeil, l’estime de soi, et l’adaptation au stress en tenant compte du profil de risque individuel des étudiants qui recherchent de l’aide peut contribuer à soutenir la santé mentale des étudiants.
Keywordschildhood adversity, childhood abuse, parental separation or divorce, student mental health, prevention and early intervention
University students represent a significant and increasing sector of the young adult population,1 and mental health concerns are common among university students.2 In a recent study of university (majority undergraduate) students in Canada, 36.1% and 31.9% of students reported symptoms of anxiety and depression, respectively.3 A survey of multiple postsecondary institutions across Canada reported the prevalence of past-year self-injury, serious suicidal thoughts, and suicidal attempts to be 8.7%, 13%, and 2.1%, respectively.4 Entry to university may be an important time for identifying students at risk of developing distressing symptoms and mental disorders, so that timely interventions can be planned and implemented to address varied student needs.1,5
Various sociodemographic factors (such as young age, gender, ethnicity, and socioeconomic status) are consistently reported to be associated with poor mental health; however, socio-contextual factors such as childhood adversities are underexplored in university student populations.2 The existing childhood adversity literature focusing in the general population has reported an association between childhood adversities such as abuse, neglect, and early loss of attachment figures6,7 and problems with emotional responses, cognition, and self-concept, with persistent vulnerability to poor mental health into adulthood.8–10 Compared to children who have not experienced such adversities, those who report higher levels of exposure are four times more likely to develop a mental illness in adulthood.11 Specifically, people with childhood exposure to abuse, neglect, or early loss have a higher risk of depression, anxiety, substance use disorders, and suicidality.11,12
Some studies have reported an association between childhood adversities and poor mental health among university students. However, these studies have some methodological limitations, such as their reliance on cross-sectional designs, the use of composite scores that merely count the number of adversities, and their focus on a limited set of mental health outcomes.2,13–15 In addition, entry to university is associated with several new stressful challenges (i.e., more autonomy, financial concerns) and other specific risk factors, including maladaptive coping (such as smoking, substance use), separation from family and other support systems, changes to one’s lifestyle (sleep disruption, lack of routine), and increased academic load—all potentially predisposing students to poor mental health.1,5 Also, the application of the life-course perspective suggests that childhood adversities may influence mental health trajectories throughout life16,17 through mechanisms such as perceived stress, health risk behaviors, and lifestyle factors.2,13,18 We hypothesized that such psychosocial and behavioral factors might mediate the association between childhood adversities and poor mental health. Identification of the mediating pathways might inform early intervention efforts aimed at improving student mental health. The conceptualized causal mechanism can be illustrated using a novel epidemiological tool, that is, Directed Acyclic Graphs (DAGs).19
Hence, the objective of this analysis was to assess the association between childhood adversity and common mental health concerns and assess the mediating effects of psychosocial and behavioral factors among university students.
The study analyzed data from the U-Flourish Student Well-Being and Academic Success Survey as part of an ongoing longitudinal study of successive cohorts of first-year undergraduate students recruited from Queen’s University in Ontario, Canada, starting in Fall 2018. The survey intended to enroll all first-year undergraduate students who were invited to complete an online survey following a student-led awareness and engagement campaign at the entry to university (Fall: Time 1). Consenting students completed the survey, which collected sociodemographic data, psychological and lifestyle information, family and personal mental health history, history of childhood adversity, and current mental health status. Those students further completed a follow-up survey at the end of the academic year prior to final exams (Spring: Time 2), followed similarly by repeat surveys at the subsequent start of Fall Term (Time 3) and end of next Winter Term (Time 4). Full details of the U-Flourish survey study methodology have been published elsewhere.20 This analysis included data from the two cohorts of first-year students entering Queen’s University in 2018 and 2019. Data from Fall 2018 to Spring 2020 for Cohort 1 (four time points) and Fall 2019 to Spring 2020 for Cohort 2 (two time points) were analyzed.
Out of 5,242 eligible students in Fall 2018 (Cohort 1), 2,994 (57.1%) completed the baseline survey, 1914 completed at Time 2, 914 completed at Time 3, and 619 completed at Time 4. Similarly, out of 5,123 students in Fall 2019 (Cohort 2), 2,949 (57.6%) completed the baseline survey, and 1,119 completed at Time 2. There were less than 10% missing data in each of the exposure and outcome variables at baseline.
Childhood Adversity Exposure. Students were asked about traumatic experiences from childhood and adolescence in the baseline survey, using questions derived from the Childhood Experience of Care and Abuse (CECA) questionnaire.21 The adversity variables were defined based on a “yes” response to the following binary response questions:
Mental Health Outcomes. Mental health outcomes were assessed in the baseline and all follow-up surveys. Depressive and anxiety symptoms were measured using the 9-item Patient Health Questionnaire (PHQ-9)22 and the 7-item Generalized Anxiety Disorder (GAD-7) scale, respectively.23 Suicidal ideation, suicidal attempts, and nonsuicidal self-harm were measured using questions derived from the Columbia Suicide Severity Rating Scale.24 The specific outcome variables were defined as follows:
Functional impairment related to symptoms of depression and anxiety was considered present if the symptoms in PHQ-9 and GADS-7 scales were rated as making it “very difficult” or “extremely difficult” to work, take care of things at home or get along with other people.25 The reference period for suicidality and non-suicidal self-harm was lifetime at baseline and past 6 months for the follow-up surveys.
Possible Modifying and/or Confounding Variables. Age in years, gender (female vs. male), ethnicity (White vs. non-White), history of reported mental illness in first-degree family members (such as mood, anxiety, psychotic, and substance misuse disorder), and maternal and paternal education (high school or less vs. postsecondary or university) have been reported to be associated with childhood adversity26,27 and adult mental health.8,16,18,28 These variables may not fall on the causal pathway and thus were examined as confounders (i.e., the association may be distorted due to intermixing of the effects of exposure with these variables) or effect modifiers (i.e., the magnitude of the effect may differ depending on the level of these variables)29 in the associations under study.
Possible Mediating Variables. Various psychosocial and behavioral variables, which have been reported to be associated with both childhood adversities and poor mental health and may fall on the causal pathway, were examined as potential mediators.2,13,18 The variables were defined as follows:
All these variables (except BMI) were measured repeatedly across surveys, and therefore multiple time points were used in the analyses. Refer to Table 1 for information about reference categories for all the variables.
The distribution of sample characteristics and prevalence of mental health outcomes were similar across Cohort 1 and Cohort 2 (Supplementary Tables 1 and 2) at the baseline survey. Therefore, for subsequent analyses, data from the two cohorts were combined. Cohort 2 had “unsure” as one of the response options on the sexual abuse item, and the unsure category was significantly associated with mental health outcomes (i.e., appeared a similar response to the “yes” category). Hence, in the combined sample, the unsure category for sexual abuse was collapsed with the “yes” category. At baseline (Supplementary Table 5) and consistently across time points (results not shown), the estimates for the association between physical abuse, sexual abuse, and bullying with all the mental health outcomes were similar, while parental separation/divorce had a smaller strength of association.
Based on consideration of homogeneity and improvement of precision, we created composite indicators of adversity exposure and mental health outcomes. We analyzed two separate indicators of childhood adversity in further primary analysis: any childhood abuse and parental separation or divorce. “Any abuse” was defined as reported exposure to physical abuse, sexual abuse, or bullying. A composite dichotomous variable, “any mental health outcome,” was created, representing screening positive for at least one of the following mental health outcomes: anxiety and depression symptoms over the screening threshold, self-reported suicidal thoughts/attempts, and non-suicidal self-harm. The pattern of association between specific childhood adversities and mental health outcomes was similar across all the time points. Therefore, repeated measures logistic regression, using Generalized Estimating Equations (GEE), was used to produce population-averaged estimates of effect, accounting for within-subject correlations.35 Exchangeable working correlation and robust standard errors were used in the analyses.
The regression models were adjusted for probable confounders determined a priori: age, gender, ethnicity, history of reported mental illness in first-degree family members, and maternal and paternal education. Due to a lack of power, we did not include the “other” gender group (0.7% of the sample) in the analyses. A conceptual model of the relationship between exposure, outcome, and potential mediators and confounders was formulated using a DAG (Supplementary Figure 1).19
Before adjustment, effect modification by the variables mentioned above was assessed using their interaction terms with childhood adversities in the models (P-value <0.05 as evidence of modification). If effect modification was ruled out, these covariates were adjusted for in the subsequent models. Then, potential mediators were adjusted, one at a time, to examine whether those adjusted estimates for childhood adversity and mental health weakened (or lost statistical significance), which would suggest mediation of the association. Instead of creating a cumulative exposure variable, these covariates at each time point were allowed to be described independently in the GEE models.36 Descriptive results are presented as percentages, means, and medians along with corresponding 95% confidence intervals (CIs), SDs, and interquartile range (IQR), respectively. Results from the regression models are presented as odds ratios (ORs) and 95% CIs. All analyses were conducted using STATA software version 16.0.37 The study was reviewed for ethical compliance and approved by the Queen’s University and Affiliated Teaching Hospitals Research Ethics Board (HSREB PSIY-609-18) and the University of Calgary Health Research Ethics Board (REB20-0436).
Table 1 presents the sample characteristics in the combined sample (n = 5,943). The mean age of the sample at baseline was 18.1 years (SD 1.6), and the majority were females (67.1%). Around 13% of the sample reported past-month tobacco or vape use, 17% reported weekly or more frequent cannabis use, and 24% reported poor sleep quality. Approximately one-third of students reported exposure to at least one form of childhood abuse. The most common childhood adversity was peer bullying (20.9%) and the least reported was physical abuse (7.4%).
At baseline, 53% (95% CI, 51.8 to 54.4) of the respondents screened positive for at least one of the mental health outcomes, which remained roughly similar throughout the follow-up, even though suicidality at baseline was reported for a lifetime and subsequently for over the past 6 months (since the last survey). The prevalence of mental health outcomes was consistently higher among those exposed to any childhood adversity at each time point (Supplementary Table 4). The prevalence of each mental health outcome at all time points in the combined sample is reported in the Supplementary Table 3. At baseline, 42.8% (95% CI, 40.4 to 45.3) and 44.0% (95% CI, 41.7 to 46.4) of the respondents who had clinically significant levels of depression and anxiety symptoms reported functional impairment.
As decided a priori, the associations were adjusted for age, gender, ethnicity, family history of mental illness, and parental education. There was no evidence of meaningful effect modification by these variables, as the interaction term P-values were >0.05. Exposure to at least one form of childhood abuse and parental separation or divorce were significantly associated with the mental health outcomes, adjusted ORs being 2.89 (95%CI,2.58to3.23)and1.29(95%CI,1.12to1.50),respectively (Table 2). Both categories of adversity were independent predictors of the mental health outcomes, with a stronger association observed for childhood abuse (OR 2.87: 95% CI, 2.56 to 3.21) than parental separation or divorce prior to the age of 16 years (OR 1.19: 95% CI, 1.03 to 1.38), each adjusted for the other adversity and the confounders.
Table 3 presents the association between each potential mediator and the composite mental health outcome. Most of these psychosocial and behavioral variables were significantly associated with the mental health outcome.
Table 4 presents the association between childhood adversities and mental health outcomes, adjusting for each potential mediator separately. The association between any abuse and the composite indicator of mental health outcome remained stable when adjusted for most covariates (such as cannabis use, binge drinking, and social support). However, it weakened with adjustment for perceived stress, self-esteem, and clinically significant insomnia separately (Table 5). When these three variables were adjusted simultaneously in the full mediation model, the adjusted OR became 2.05 (95% CI, 1.80 to 2.34). Similarly, only adjustment for self-esteem weakened the association between parental divorce and the mental health outcomes (adjusted OR 1.17 (95% CI, 1.00 to 1.36)).
This analysis identified that childhood abuse and parental separation or divorce are associated with common mental health concerns in undergraduate students. Potentially modifiable psychological factors, namely perceived stress, selfesteem, and insomnia appeared to mediate the association between childhood abuse and adverse mental health. Self-esteem also mediated the association of parental divorce with poor mental health.
In this study, first-year university students reported high levels of childhood adversity in keeping with epidemiological data.2,15,38,39 Further, the high prevalence of mental health concerns among university students found in the study has also been reported consistently in the literature.5,40,41 The prevalence estimates reported for specific mental health outcomes in this study were slightly higher than those reported in a study of Canadian postgraduate students,5 potentially due to the difference in the sample included (household residents only vs. living on- or off-campus) and the measurement instrument used (diagnostic vs. screening tool). Also, there was an overrepresentation of females in our study, and females report a higher prevalence of mental health problems than males.5
The finding that childhood adversity is a strong predictor of psychopathology in university students aligns with the extant literature.38,42 For instance, a study of South African university students reported that at least one adverse exposure (such as bullying, physical and sexual abuse, and parental psychopathology) was significantly associated with depression.15 A recent systematic review reported a significant association between physical and sexual abuse and suicidal attempts among youth.43 The importance of parental divorce in childhood as a risk factor for adverse mental health, including depression, anxiety, suicidal thoughts, and attempts, has also been reported among college students.44,45 These findings suggest that early recognition of risk factors, including childhood adversity in help-seeking students, might help map intervention to specific student mental health needs. However, further intervention studies are required to generate evidence-based interventions.
The specific adversities were independent predictors of adverse mental health in adulthood, independent of the other adversity and known confounders. However, our analysis indicated two different dimensions of adversity; the ORs were statistically significant but consistently lower for parental separation/divorce compared to childhood abuse. This finding supports the emerging argument that childhood adversities have various dimensions, such as threat-based (e.g., physical abuse, emotional abuse) and deprivation-based (e.g., parental divorce and substance abuse),46 and may differentially contribute to reduced health.47
Another important finding was the potential mediating effect of perceived stress, self-esteem, and clinically significant insomnia in the relationship between childhood adversities and mental health outcomes. These findings are consistent with the existing literature that reported childhood abuse is significantly associated with increased perceived stress and maladaptive coping strategies which may be due to the adverse effect of abuse and deprivation on the maturation of self-regulatory systems and abnormally heightened stress response.48 Stress has been reported to be a significant risk factor for adverse mental health,49 and persons with high stress are also reported to use mental health services frequently.50 Also, it has been reported earlier that adolescents with childhood adversity are more likely to experience insomnia,51 which may be a precipitating factor for mental illness.52 Similarly, students with childhood abuse and parental divorce or separation may have low self-esteem, leading to maladjustment and social withdrawal, which may increase vulnerability to mood-related problems.53–55 This finding suggests that healthy stress coping interventions,48,56–58 promotion of a healthy lifestyle (especially targeting insomnia symptoms),51 and strategies for recovery of self-esteem (e.g., encouraging positive self-evaluations)1,54 need to be examined as potential approaches in prevention and management of mental health concerns in university students. The covariates were measured longitudinally from the start of university, and adversities reportedly occurred in childhood, suggesting mediation; however, due to the nature of data collection (data on adversities and covariates collected simultaneously), the mediation hypothesis could not be established deterministically.
The study has several strengths and limitations to consider. The longitudinal nature of the study, large sample size, and the use of validated measures to assess the outcomes are the strengths. Females were overrepresented throughout the survey, and the estimates were adjusted for gender to make them comparable across gender categories. Although 57% is a reasonable response rate for this type of survey, there is still a potential for selection bias due to nonparticipation and attrition during follow-up. The magnitude of any such bias cannot be quantified using existing data. Nonetheless, the samples were broadly representative of the eligible first-year student population according to the program of study, international student status, and ethnicity.25 The retrospective reports of childhood adversities may have introduced recall bias and possibly strengthened the associations if the misclassification may have been differential (i.e., students with poor mental health may be more likely to recall adverse experiences than their counterparts).59 The analysis was not limited to only those students indicating functional impairment associated with symptoms of depression and anxiety. Future studies should seek to quantify the impact on functional impairment, given its importance for informing targeted intervention strategies. Finally, the generalization of findings to other university student populations requires caution, especially when personal and attitudinal variables are analyzed.60
The findings suggest that childhood adversities are common and associated with adverse mental health among undergraduate university students. Characterization of this association provides supportive evidence for incorporating risk factors such as early adversity into a more individualized risk profile, which may help with clinical triage and mapping to the appropriate supports and level of care for students seeking mental health support.1,61 Psychosocial factors (i.e., perceived stress, low self-esteem, and low sleep quality) identified as mediators in the study should be further explored through intervention studies as targets to mitigate the harms to mental health associated with childhood adversities. Student-tailored mental health literacy may support the development of healthy socioemotional coping resources and facilitate timely help-seeking through reduced stigma62 and increased emotional self-awareness.63
All co-authors meet the authorship criteria. A.D. is the nominated principal investigator of the U-Flourish Student Well-Being and Academic Success study. A.B. and S.B.P. conceived the study, guided also by discussions with A.D., A.G.M.B., G.D., and K.A. A.B. analyzed the data and prepared the first draft. A.D., S.B.P., K.A., and N.K. were involved in the discussion of data analysis and interpretation of the results. G.D., D.D, M.L., J.B., D.R., and S.C. also provided intellectual content to the study at all stages. All authors assisted with the preparation of this manuscript and read and approved the final draft.
Access to the data used in the study may be provided upon request to the principal investigator of the U-Flourish Student Well-Being and Academic Success study.
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article. This work was supported by the Alberta Innovates Health Solutions (AIHS) (grant number 201810552) Graduate Studentship, Canadian Institute of Health Research (CIHR), and Cuthbertson and Fischer Chair in Pediatric Mental Health, Rossy Family Foundation.
Asmita Bhattarai https://orcid.org/0000-0001-6689-7218
Nathan King https://orcid.org/0000-0003-4664-5456
Kamala Adhikari https://orcid.org/0000-0003-2872-9496
Andrew G.M. Bulloch https://orcid.org/0000-0003-3305-7874
Scott B. Patten https://orcid.org/0000-0001-9871-4041
Anne Duffy https://orcid.org/0000-0002-5895-075X
Supplemental material for this article is available online.
1 Departments of Community Health Sciences & Psychiatry, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada
2 Department of Public Health Sciences, Queen’s University, Kingston, Ontario, Canada
3 Provincial Population and Public Health, Alberta Health Services, Calgary, Alberta, Canada
4 Faculty of Social Work, University of Calgary, Calgary, Alberta, Canada
5 Department of Pharmacology and Toxicology, University of Toronto, Toronto, Ontario, Canada
6 Department of Biomedical and Molecular Sciences, Queen’s University, Kingston, Ontario, Canada
7 Department of Psychiatry, Queen’s University, Kingston, Ontario, Canada
8 Department of Psychiatry, University of Oxford, Oxford, UK
Corresponding Author:Asmita Bhattarai, Department of Community Health Sciences & Psychiatry, Cumming School of Medicine, University of Calgary, 3280 Hospital Drive NW, Calgary, Alberta, Canada T2N4Z6.Email: asmita.bhattarai1@ucalgary.ca