quantitative research about students mental health

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings
• My Bibliography
• Collections
• Citation manager

Save citation to file

Email citation, add to collections.

• Create a new collection
• Add to an existing collection

Add to My Bibliography

Your saved search, create a file for external citation management software, your rss feed.

• Search in PubMed
• Search in NLM Catalog
• Add to Search

Examining the mental health of university students: A quantitative and qualitative approach to identifying prevalence, associations, stressors, and interventions

Affiliations.

• 1 Department of Dental Public Health and Behavioural Sciences, University of Missouri-Kansas City School of Dentistry, Kansas City, MO, USA.
• 2 Office of Research and Graduate Programs, University of Missouri-Kansas City School of Dentistry, Kansas City, MO, USA.
• PMID: 35380931
• DOI: 10.1080/07448481.2022.2057192

Objective To identify the prevalence of anxiety, depression, and suicidal ideation that would place university students at risk for mental health disorders. To explore the source of stressors and possible interventions that may benefit student mental health in a university setting.

Participants: University students (n = 483) who had been learning remotely due to the COVID-19 pandemic.

Methods: A mixed-methods cross-sectional survey was administered in 2020.

Results: Students were at an increased rate of depression, anxiety and suicidal ideation as compared to the general population. Female gender, lack of social support, living alone, being a first-generation college student and COVID-19 were significantly associated with mental health disorders. Stressors were identified and categorized into themes and interventions were recognized that may improve student well-being.

Conclusion: Students enrolled in university programs appear to experience significant amounts of anxiety, depression, and suicidal ideation. Additional mental health education, resources, and support is needed.

Keywords: Anxiety; COVID-19; college students; depression; suicidal ideation.

PubMed Disclaimer

Similar articles

• Investigating Mental Health of US College Students During the COVID-19 Pandemic: Cross-Sectional Survey Study. Wang X, Hegde S, Son C, Keller B, Smith A, Sasangohar F. Wang X, et al. J Med Internet Res. 2020 Sep 17;22(9):e22817. doi: 10.2196/22817. J Med Internet Res. 2020. PMID: 32897868 Free PMC article.
• Mental Health Symptoms of University Students 15 Months After the Onset of the COVID-19 Pandemic in France. Wathelet M, Horn M, Creupelandt C, Fovet T, Baubet T, Habran E, Martignène N, Vaiva G, D'Hondt F. Wathelet M, et al. JAMA Netw Open. 2022 Dec 1;5(12):e2249342. doi: 10.1001/jamanetworkopen.2022.49342. JAMA Netw Open. 2022. PMID: 36580328 Free PMC article.
• Prevalence and associated factors of suicidal ideation and attempt among undergraduate medical students of Haramaya University, Ethiopia. A cross sectional study. Asfaw H, Yigzaw N, Yohannis Z, Fekadu G, Alemayehu Y. Asfaw H, et al. PLoS One. 2020 Aug 12;15(8):e0236398. doi: 10.1371/journal.pone.0236398. eCollection 2020. PLoS One. 2020. PMID: 32785295 Free PMC article.
• Prevalence and associated factors of suicidal ideation among college students during the COVID-19 pandemic in China: a 3-wave repeated survey. Liang SW, Liu LL, Peng XD, Chen JB, Huang AD, Wang XY, Zhao JB, Fan F, Liu XC. Liang SW, et al. BMC Psychiatry. 2022 May 15;22(1):336. doi: 10.1186/s12888-022-03968-2. BMC Psychiatry. 2022. PMID: 35570282 Free PMC article.
• The Impact of COVID-Related Restrictions on the Mental Health of Students. Reuter PR. Reuter PR. Adv Exp Med Biol. 2024;1458:35-50. doi: 10.1007/978-3-031-61943-4_3. Adv Exp Med Biol. 2024. PMID: 39102188 Review.
• "You would think she would hug me": Micropractices of Care Between First-Generation College Students and Their Parents During Covid-19. Flores A, Mason KA. Flores A, et al. Cult Med Psychiatry. 2024 Mar;48(1):91-112. doi: 10.1007/s11013-023-09833-5. Epub 2023 Sep 28. Cult Med Psychiatry. 2024. PMID: 37768495
• Search in MeSH

Related information

Linkout - more resources, full text sources.

• Taylor & Francis
• MedlinePlus Health Information

• Citation Manager

NCBI Literature Resources

MeSH PMC Bookshelf Disclaimer

The PubMed wordmark and PubMed logo are registered trademarks of the U.S. Department of Health and Human Services (HHS). Unauthorized use of these marks is strictly prohibited.

This paper is in the following e-collection/theme issue:

Published on 3.9.2020 in Vol 22 , No 9 (2020) : September

Effects of COVID-19 on College Students’ Mental Health in the United States: Interview Survey Study

Authors of this article:

Original Paper

• Changwon Son 1 , BS, MS   ; 
• Sudeep Hegde 1 , BEng, MS, PhD   ; 
• Alec Smith 1 , BS   ; 
• Xiaomei Wang 1 , BS, PhD   ; 
• Farzan Sasangohar 1, 2 , BA, BCS, MASc, SM, PhD  

1 Department of Industrial and Systems Engineering, Texas A&M University, College Station, TX, United States

2 Center for Outcomes Research, Houston Methodist Hospital, Houston, TX, United States

Corresponding Author:

Farzan Sasangohar, BA, BCS, MASc, SM, PhD

Department of Industrial and Systems Engineering

Texas A&M University

College Station, TX, 77843

United States

Phone: 1 979 458 2337

Email: [email protected]

Background: Student mental health in higher education has been an increasing concern. The COVID-19 pandemic situation has brought this vulnerable population into renewed focus.

Objective: Our study aims to conduct a timely assessment of the effects of the COVID-19 pandemic on the mental health of college students.

Methods: We conducted interview surveys with 195 students at a large public university in the United States to understand the effects of the pandemic on their mental health and well-being. The data were analyzed through quantitative and qualitative methods.

Results: Of the 195 students, 138 (71%) indicated increased stress and anxiety due to the COVID-19 outbreak. Multiple stressors were identified that contributed to the increased levels of stress, anxiety, and depressive thoughts among students. These included fear and worry about their own health and of their loved ones (177/195, 91% reported negative impacts of the pandemic), difficulty in concentrating (173/195, 89%), disruptions to sleeping patterns (168/195, 86%), decreased social interactions due to physical distancing (167/195, 86%), and increased concerns on academic performance (159/195, 82%). To cope with stress and anxiety, participants have sought support from others and helped themselves by adopting either negative or positive coping mechanisms.

Conclusions: Due to the long-lasting pandemic situation and onerous measures such as lockdown and stay-at-home orders, the COVID-19 pandemic brings negative impacts on higher education. The findings of our study highlight the urgent need to develop interventions and preventive strategies to address the mental health of college students.

Introduction

Mental health issues are the leading impediment to academic success. Mental illness can affect students’ motivation, concentration, and social interactions—crucial factors for students to succeed in higher education [ 1 ]. The 2019 Annual Report of the Center for Collegiate Mental Health [ 2 ] reported that anxiety continues to be the most common problem (62.7% of 82,685 respondents) among students who completed the Counseling Center Assessment of Psychological Symptoms, with clinicians also reporting that anxiety continues to be the most common diagnosis of the students that seek services at university counseling centers. Consistent with the national trend, Texas A&M University has seen a rise in the number of students seeking services for anxiety disorders over the past 8 years. In 2018, slightly over 50% of students reported anxiety as the main reason for seeking services. Despite the increasing need for mental health care services at postsecondary institutions, alarmingly, only a small portion of students committing suicide contact their institution counseling centers [ 3 ], perhaps due to the stigma associated with mental health. Such negative stigma surrounding mental health diagnosis and care has been found to correlate with a reduction in adherence to treatment and even early termination of treatment [ 4 ].

The COVID-19 pandemic has brought into focus the mental health of various affected populations. It is known that the prevalence of epidemics accentuates or creates new stressors including fear and worry for oneself or loved ones, constraints on physical movement and social activities due to quarantine, and sudden and radical lifestyle changes. A recent review of virus outbreaks and pandemics documented stressors such as infection fears, frustration, boredom, inadequate supplies, inadequate information, financial loss, and stigma [ 5 ]. Much of the current literature on psychological impacts of COVID-19 has emerged from the earliest hot spots in China. Although several studies have assessed mental health issues during epidemics, most have focused on health workers, patients, children, and the general population [ 6 , 7 ]. For example, a recent poll by The Kaiser Family Foundation showed that 47% of those sheltering in place reported negative mental health effects resulting from worry or stress related to COVID-19 [ 8 ]. Nelson et al [ 9 ] have found elevated levels of anxiety and depressive symptoms among general population samples in North America and Europe. However, with the exception of a few studies, notably from China [ 10 - 12 ], there is sparse evidence of the psychological or mental health effects of the current pandemic on college students, who are known to be a vulnerable population [ 13 ]. Although the findings from these studies thus far converge on the uptick of mental health issues among college students, the contributing factors may not necessarily be generalizable to populations in other countries. As highlighted in multiple recent correspondences, there is an urgent need to assess effects of the current pandemic on the mental health and well-being of college students [ 14 - 17 ].

The aim of this study is to identify major stressors associated with the COVID-19 pandemic and to understand their effects on college students’ mental health. This paper documents the findings from online interview surveys conducted in a large university system in Texas.

Study Design

A semistructured interview survey guide was designed with the purpose of assessing the mental health status of college students both quantitatively and qualitatively. In addition, the interview aimed to capture the ways that students have been coping with the stress associated with the pandemic situation. First, our study assesses participants’ general stress levels using the Perceived Stress Scale-10 (PSS) [ 18 ]. PSS is a widely used instrument to measure overall stress in the past month [ 19 ]. Second, participants were asked if their own and peers’ (two separate questions) stress and anxiety increased, decreased, or remained the same because of the COVID-19 pandemic. For those who indicated increased stress and anxiety during the pandemic, we questioned their stress coping strategies and use of available mental health counseling services. We then elicited pandemic-specific stressors and their manifestations across 12 academic-, health-, and lifestyle-related categories of outcomes such as effects on own or loved ones’ health, sleeping habits, eating habits, financial situation, changes to their living environment, academic workload, and social relations. Students were also asked about the impact of COVID-19 on depressive and suicidal thoughts. These constructs were derived from existing literature identifying prominent factors affecting college students’ mental health [ 20 , 21 ]. Feedback on the severity of COVID-19’s impact on these aspects were elicited using a 4-point scale: 0 (none), 1 (mild), 2 (moderate), and 3 (severe). Participants were asked to elaborate on each response. Third, participants were guided to describe stressors, coping strategies, and barriers to mental health treatment during a typical semester without associating with the COVID-19 pandemic. Although multiple analyses of the collected data are currently under progress, PSS results and the COVID-19–related findings are presented in this paper.

Participants

Participants were recruited from the student population of a large university system in Texas, United States. This particular university closed all their campuses on March 23, 2020, and held all its classes virtually in response to the COVID-19 pandemic. In addition, the state of Texas issued a stay-at-home order on April 2, 2020. Most interviews were conducted about 1 month after the stay-at-home order in April 2020. Figure 1 illustrates the trend of cumulative confirmed cases and a timeline of major events that took place in the university and the state of Texas. Participants were recruited by undergraduate student researchers through email, text messaging, and snowball sampling. The only inclusion criteria for participation was that participants should have been enrolled as undergraduate students in the university at the time of the interviews.

The interviews were conducted by 20 undergraduate researchers trained in qualitative methods and the use of the interview survey guide described above. None of the authors conducted the interviews. All interviews were conducted via Zoom [ 22 ] and were audio recorded. The recordings were later transcribed using Otter.ai [ 23 ], an artificial intelligence–based transcription service, and verified for accuracy manually. Prior to the interview, participants were provided an information document about the study approved by the university’s Institutional Review Board (No 2019-1341D). Upon verbal consent, participants were asked to respond to a questionnaire about their demographic information such as age, gender, year of college, and program of study before completing the interview. Participation was voluntary and participants were not compensated.

Data Analysis

First, descriptive statistics were compiled to describe participants’ demographics (eg, age, gender, academic year, and major) and the distribution of the ratings on PSS-10 survey items. A total PSS score per participant was calculated by first reversing the scores of the positive items (4-7, 9, and 10) and then adding all the ten scores. A mean (SD) PSS score was computed to evaluate the overall level of stress and anxiety among the participants during the COVID-19 pandemic. Second, participants’ answers to 12 academic-, health-, and lifestyle-related questions were analyzed to understand relative impacts of the pandemic on various aspects of college students’ mental health. Percentages of participants who indicated negative ratings (ie, mild, moderate, or severe influence) on these questions were calculated and ranked in a descending order. Qualitative answers to the 12 stressors and coping strategies were analyzed using thematic analysis [ 24 , 25 ] similar to the deductive coding step in the grounded theory method [ 26 ]. A single coder (CS), trained in qualitative analysis methods, analyzed the transcripts and identified themes using an open coding process, which does not use a priori codes or codes created prior to the analysis and places an emphasis on information that can be extracted directly from the data. Following the identification of themes, the coder discussed the codes with two other coders (XW and AS) trained in qualitative analysis and mental health research to resolve discrepancies among related themes and discuss saturation. The coders consisted of two Ph.D. students and one postdoctoral fellow at the same university. MAXQDA (VERBI GmbH) [ 27 ] was used as a computer software program to carry out the qualitative analysis.

Of the 266 university students initially recruited by the undergraduate researchers, 17 retreated and 249 participated in this study. There were 3 graduate students and 51 participants who had missing data points and were excluded, and data from 195 participants were used in the analysis. The average age was 20.7 (SD 1.7) years, and there were more female students (111/195, 57%) than male students (84/195, 43%). Approximately 70% of the participants were junior and senior students. About 60% of the participants were majoring in the college of engineering, which was the largest college in the university population ( Table 1 ). The mean PSS score for the 195 participants was 18.8 (SD 4.9), indicating moderate perceived stress in the month prior to the interview ( Table 2 ).

a PSS: Perceived Stress Scale-10.

Challenges to College Students’ Mental Health During COVID-19

Out of 195 participants, 138 (71%) indicated that their stress and anxiety had increased due to the COVID-19 pandemic, whereas 39 (20%) indicated it remained the same and 18 (9%) mentioned that the stress and anxiety had actually decreased. Among those who perceived increased stress and anxiety, only 10 (5%) used mental health counseling services. A vast majority of the participants (n=189, 97%) presumed that other students were experiencing similar stress and anxiety because of COVID-19. As shown in Figure 2 , at least 54% (up to 91% for some categories) of participants indicated negative impacts (either mild, moderate, or severe) of COVID-19 on academic-, health-, and lifestyle-related outcomes. The qualitative analysis yielded two to five themes for each category of outcomes. The chronic health conditions category was excluded from the qualitative analysis due to insufficient qualitative response. Table 3 presents the description and frequency of the themes and select participant quotes.

a Not every participant provided sufficient elaboration to allow for identification of themes, so the frequency of individual themes does not add up to the total number of participants who indicated negative impacts of the COVID-19 outbreak.

b The five-digit alphanumeric value indicates the participant ID.

c TA: teaching assistant.

Concerns for One’s Own Health and the Health of Loved Ones

A vast majority of the participants (177/195, 91%) indicated that COVID-19 increased the level of fear and worry about their own health and the health of their loved ones. Over one-third of those who showed concern (76/177, 43%) were worried about their families and relatives who were more vulnerable, such as older adults, those with existing health problems, and those who are pregnant or gave birth to a child recently. Some of the participants (26/177, 15%) expressed their worry about their family members whose occupation increased their risk of exposure to COVID-19 such as essential and health care workers. Some participants (19/177, 11%) specifically mentioned that they were worried about contracting the virus.

Difficulty With Concentration

A vast majority of participants (173/195, 89%) indicated difficulty in concentrating on academic work due to various sources of distraction. Nearly half of them (79/173, 46%) mentioned that their home is a distractive environment and a more suitable place to relax rather than to study. Participants mentioned that they were more prone to be interrupted by their family members and household chores at home. Other factors affecting students’ concentration were lack of accountability (21/173, 12%) and social media, internet, and video games (19/173, 11%). Some (18/173, 10%) stated that online classes were subject to distraction due to lack of interactions and prolonged attention to a computer screen. Additionally, monotonous life patterns were mentioned by some to negatively affect concentration on academic work (5/173, 3%).

Disruption to Sleep Patterns

A majority of participants (168/195, 86%) reported disruptions to their sleep patterns caused by the COVID-19 pandemic, with over one-third (38%) reporting such disruptions as severe. Half of students who reported some disruption (84/168, 50%) stated that they tended to stay up later or wake up later than they did before the COVID-19 outbreak. Another disruptive impact brought by the pandemic was irregular sleep patterns such as inconsistent time to go to bed and to wake up from day to day (28/168, 17%). Some (12/168, 7%) reported increased hours of sleep, while others (10/168, 6%) had poor sleep quality.

Increased Social Isolation

A majority of participants answered that the pandemic has increased the level of social isolation (167/195, 86%). Over half of these students (91/167, 54%) indicated that their overall interactions with other people such as friends had decreased significantly. In particular, about one-third (52/167, 31%) shared their worries about a lack of in-person interactions such as face-to-face meetings. Others (9/167, 5%) stated that disruptions to their outdoor activities (eg, jogging, hiking) have affected their mental health.

Concerns About Academic Performance

A majority of participants (159/195, 82%) showed concerns about their academic performance being impacted by the pandemic. The biggest perceived challenge was the transition to online classes (61/159, 38%). In particular, participants stated their concerns about sudden changes in the syllabus, the quality of the classes, technical issues with online applications, and the difficulty of learning online. Many participants (36/159, 23%) were worried about progress in research and class projects because of restrictions put in place to keep social distancing and the lack of physical interactions with other students. Some participants (23/159, 14%) mentioned the uncertainty about their grades under the online learning environment to be a major stressor. Others (12/159, 8%) indicated their reduced motivation to learn and tendency to procrastinate.

Disruptions to Eating Patterns

COVID-19 has also negatively impacted a large portion of participants’ dietary patterns (137/195, 70%). Many (35/137, 26%) stated that the amount of eating has increased, including having more snacks since healthy dietary options were reduced, and others (27/137, 20%) addressed that their eating patterns have become inconsistent because of COVID-19, for example, irregular times of eating and skipping meals. Some students (16/137, 12%) reported decreased appetite, whereas others (7/137, 5%) were experiencing emotional eating or a tendency to eat when bored. On the other hand, some students (28/195, 14%) reported that they were having healthier diets, as they were cooking at home and not eating out as much as they used to.

Changes in the Living Environment

A large portion of the participants (130/195, 67%) described that the pandemic has resulted in significant changes in their living conditions. A majority of these students (89/130, 68%) referred to living with family members as being less independent and the environment to be more distractive. For those who stayed in their residence either on- or off-campus (18/130, 14%), a main change in their living environment was reduced personal interactions with roommates. Some (9/130, 7%) mentioned that staying inside longer due to self-quarantine or shelter-in-place orders was a primary change in their living circumstances.

Financial Difficulties

More than half of the participants (115/195, 59%) expressed their concerns about their financial situations being impacted by COVID-19. Many (44/115, 38%) noted that COVID-19 has impacted or is likely to impact their own current and future employment opportunities such as part-time jobs and internships. Some (21/115, 18%) revealed the financial difficulties of their family members, mostly parents, getting laid off or receiving pay cuts in the wake of COVID-19.

Increased Class Workload

The effect of COVID-19 on class workload among the college students was not conclusive. Although slightly over half of participants (106/195, 54%) indicated their academic workload has increased due to COVID-19, the rest stated the workload has remained the same (70/195, 36%) or rather decreased (19/195, 10%). For those who were experiencing increased workloads, nearly half (51/106, 48%) thought they needed to increase their own efforts to catch up with online classes and class projects given the lack of in-person support from instructors or teaching assistants. About one-third of the participants (33/106, 31%) perceived that assignments had increased or became harder to do. Some (6/106, 6%) found that covering the remainder of coursework as the classes resumed after the 2-week break to be challenging.

Depressive Thoughts

When asked about the impact of the COVID-19 pandemic on depressive thoughts, 44% (86/195) mentioned that they were experiencing some depressive thoughts during the COVID-19 pandemic. Major contributors to such depressive thoughts were loneliness (28/86, 33%), insecurity or uncertainty (10/86, 12%), powerlessness or hopelessness (9/86, 10%), concerns about academic performance (7/86, 8%), and overthinking (4/86, 5%).

Suicidal Thoughts

Out of 195 participants, 16 (8%) stated that the pandemic has led to some suicidal thoughts with 5% (10/16) reporting these thoughts as mild and 3% (6/16) as moderate. There were 6 participants (38%) that attributed their suicidal thoughts to the presence of depressive thoughts. Other reasons were related to academic performance (1/16, 6%), problems with family as they returned home (1/16, 6%), and fear from insecurity and uncertainty (1/16, 6%).

Coping Mechanism During COVID-19

To cope with stress and anxiety imposed by COVID-19, college students reported seeking support from others but were mainly using various self-management methods.

Self-Management

The majority of the participants (105/138, 76%) with increased stress due to the outbreak of COVID-19 explained that they were using various means to help themselves cope with stress and anxiety during the pandemic. Some (24/105, 23%) relied on negative coping methods such as ignoring the news about COVID-19 (10/105), sleeping longer (7/105), distracting themselves by doing other tasks (5/105), and drinking or smoking (2/105). Approximately one-third (30/105, 29%) used positive coping methods such as meditation and breathing exercises (18/105), spiritual measures (7/105), keeping routines (4/105), and positive reframing (2/105). A majority of the participants (73/105, 70%) who used self-management mentioned doing relaxing hobbies including physical exercise (31/105), enjoying streaming services and social media (22/105), playing with pets (7/105), journaling (5/105), listening to music (4/105), reading (2/105), and drawing (2/105). Finally, some participants (15/105, 14%) stated that they were planning activities (eg, drafting to-do lists) for academic work and personal matters as a self-distraction method.

Seeking Support From Others

Approximately one-third of the participants (47/138, 34%) mentioned that communicating with their families and friends was a primary way to deal with stress and anxiety during COVID-19. Some explicitly stated that they were using a virtual meeting application such as Zoom frequently to connect to friends and family. Only 1 participant claimed to be receiving support from a professional therapist, and another participant was using Sanvello, a mobile mental health service app provided by the university.

Barriers to Seeking Professional Support During COVID-19

Despite the availability of tele-counseling and widespread promotion of such services by the university, a vast majority of participants who indicated an increase in stress and anxiety (128/138, 93%) claimed that they had not used school counseling services during the pandemic. Reasons for such low use included the condition not being perceived as severe enough to seek the services (4/128, 3%), not comfortable interacting with unfamiliar people (1/128, 0.8%), not comfortable talking about mental health issues over the phone (1/128, 0.8%), and lack of trust in the counseling services (1/128, 0.8%).

Principal Findings

College students comprise a population that is considered particularly vulnerable to mental health concerns. The findings of this study bring into focus the effects of pandemic-related transitions on the mental health and well-being of this specific population. Our findings suggest a considerable negative impact of the COVID-19 pandemic on a variety of academic-, health-, and lifestyle-related outcomes. By conducting online survey interviews in the midst of the pandemic, we found that a majority of the participants were experiencing increased stress and anxiety due to COVID-19. In addition, results of the PSS showed moderate levels of stress among our participants. This is in line with a recent pre–COVID-19 survey conducted in the United Kingdom (mean PSS score 19.79, SD 6.37) [ 28 ]; however, the administration of PSS as interview questions (compared to allowing participants to read and respond to the 10 questions) might have introduced bias and resulted in underreporting.

Among the effects of the pandemic identified, the most prominent was worries about one’s own health and the health of loved ones, followed by difficulty concentrating. These findings are in line with recent studies in China that also found concerns relating to health of oneself and of family members being highly prevalent among the general population during the pandemic. Difficulty in concentrating, frequently expressed by our participants, has previously been shown to adversely affect students’ confidence in themselves [ 29 ], which has known correlations to increased stress and mental health [ 30 ]. In comparison with stress and anxiety in college students’ general life, it appears that countermeasures put in place against COVID-19, such as shelter-in-place orders and social distancing practices, may have underpinned significant changes in students’ lives. For example, a vast majority of the participants noted changes in social relationships, largely due to limited physical interactions with their families and friends. This is similar to recent findings of deteriorated mental health status among Chinese students [ 10 ] and increased internet search queries on negative thoughts in the United States [ 31 ]. The findings on the impact of the pandemic on sleeping and eating habits are also a cause for concern, as these variables have known correlations with depressive symptoms and anxiety [ 20 ].

Although a majority of participants expressed concerns regarding academic performance, interestingly, almost half of the participants reported lower stress levels related to academic pressure and class workload since the pandemic began. This may be due, in part, to decisions taken by professors and the university to ease the students’ sudden transition to distance learning. For instance, this university allowed students to choose a pass/fail option for each course instead of a regular letter grade. Additionally, actions taken by professors, such as reduced course loads, open book examinations, and other allowances on grading requirements, could also have contributed to alleviating or reducing stress. Although participants who returned to their parental home reported concerns about distractions and independence, students might have benefited from family support and reduced social responsibilities. Therefore, the increased stress due to the pandemic may have been offset, at least to some extent.

Alarmingly, 44% (86/195) of the participants reported experiencing an increased level of depressive thoughts, and 8% (16/195) reported having suicidal thoughts associated with the COVID-19 pandemic. Previous research [ 32 ] reported about 3%-7% of the college student population to have suicidal thoughts outside of the pandemic situation. Furthermore, with the exception of high-burnout categories, depression levels among students, reported in several recent studies [ 33 - 35 ], have varied between 29% and 38%, which may suggest an uptick in pandemic-related depressive symptoms among college students similar to recent studies in China [ 10 , 11 ]. Although our participants specifically mentioned several factors such as feelings of loneliness, powerlessness, as well as financial and academic uncertainties, other outcomes that were perceived to be impacted by the COVID-19 pandemic may also act as contributors to depressive thoughts and suicidal ideation. In particular, both difficulty concentrating and changes in sleeping habits are associated with depression [ 20 , 29 , 36 ].

Our study also identifies several coping mechanisms varying between adaptive and maladaptive behaviors. The maladaptive coping behaviors such as denial and disengagement have been shown to be significant predictors of depression among young adults [ 37 ]. In contrast, adaptive coping such as acceptance and proactive behaviors are known to positively impact mental health. Our findings suggest that the majority of our participants exhibited maladaptive coping behaviors. Identifying students’ coping behavior is important to inform the planning and design of support systems. In this regard, participatory models of intervention development can be used, in which researchers’ and psychologists’ engagement with the target population to adapt interventional programs to their specific context has shown promise [ 37 , 38 ]. For instance, Nastasi et al [ 37 ] used a participatory model to develop culture-specific mental health services for high school students in Sri Lanka. Similar approaches can be adopted to engage college students as well to develop a mental health program that leverages their natural positive coping behaviors and addresses their specific challenges.

Participants described several barriers to seeking help, such as lack of trust in counseling services and low comfort levels in sharing mental health issues with others, which may be indicative of stigma. Perceiving social stigma as a barrier to seeking help and availing counseling services and other support is common among students [ 29 ]. One study showed that only a minor fraction of students who screened positive for a mental health problem actually sought help [ 39 ]. Although overcoming the stigma associated with mental health has been discussed at length, practical ways of mitigating this societal challenge remains a gap [ 40 , 41 ]. Our findings suggest that self-management is preferred by students and should be supported in future work. Digital technologies and telehealth applications have shown some promise to enable self-management of mental health issues [ 42 ]. For instance, Youn et al [ 43 ] successfully used social media networks as a means to reach out to college students and screen for depression by administering a standardized scale, the Patient Health Questionnaire-9. Digital web-based platforms have also been proposed to enhance awareness and communication with care providers to reduce stigma related to mental health among children in underserved communities [ 44 ]. For instance, one of the online modules suggested by the authors involves providing information on community-identified barriers to communicating with care providers. Technologies such as mobile apps and smart wearable sensors can also be leveraged to enable self-management and communication with caregivers.

In light of the aforementioned projections of continued COVID-19 cases at the time of this writing [ 45 ] and our findings, there is a need for immediate attention to and support for students and other vulnerable groups who have mental health issues [ 17 ]. As suggested by a recent study [ 46 ] based on the Italian experience of this pandemic, it is essential to assess the population’s stress levels and psychosocial adjustment to plan for necessary support mechanisms, especially during the recovery phase, as well as for similar events in the future. Although the COVID-19 pandemic seems to have resulted in a widespread forced adoption of telehealth services to deliver psychiatric and mental health support, more research is needed to investigate use beyond COVID-19 as well as to improve preparedness for rapid virtualization of psychiatric counseling or tele-psychiatry [ 47 - 49 ].

Limitations and Future Work

To our knowledge, this is the first effort in documenting the psychological impacts of the COVID-19 pandemic on a representative sample of college students in the United States via a virtual interview survey method in the middle of the pandemic. However, several limitations should be noted. First, the sample size for our interview survey was relatively small compared to typical survey-only studies; however, the survey interview approach affords the capture of elaboration and additional clarifying details, and therefore complements the survey-based approaches of prior studies focusing on student mental health during this pandemic [ 10 , 11 , 50 ]. Second, the sample used is from one large university, and findings may not generalize to all college students. However, given the nationwide similarities in universities transitioning to virtual classes and similar stay-at-home orders, we expect reasonable generalizability of these findings. Additionally, a majority of our participants were from engineering majors. Therefore, future work is needed to use a stratified nationwide sample across wider disciplines to verify and amend these findings. Third, although a vast majority of participants answered that they have not used the university counseling service during the pandemic, only a few of them provided reasons. Since finding specific reasons behind the low use is a key to increasing college students’ uptake of available counseling support, future research is warranted to unveil underlying factors that hinder college students’ access to mental health support. Finally, we did not analyze how student mental health problems differ by demographic characteristics (eg, age, gender, academic year, major) or other personal and social contexts (eg, income, religion, use of substances).

Future work could focus on more deeply probing the relationships between various coping mechanisms and stressors. Additionally, further study is needed to determine the effects of the pandemic on students’ mental health and well-being in its later phases beyond the peak period. As seen in the case of health care workers in the aftermath of the severe acute respiratory syndrome outbreak, there is a possibility that the effects of the pandemic on students may linger for a period beyond the peak of the COVID-19 pandemic itself [ 51 ].

Acknowledgments

This research was partly funded by a Texas A&M University President’s Excellence (X-Grant) award.

Conflicts of Interest

None declared.

• Unger K. Handbook on Supported Education: Providing Services for Students With Psychiatric Disabilities. Charleston, SC: BookSurge Publishing; 2007.
• 2019 annual report. Center for Collegiate Mental Health. University Park, PA: Penn State University; 2020.   URL: https://ccmh.memberclicks.net/assets/docs/2019-CCMH-Annual-Report_3.17.20.pdf [accessed 2020-05-11]
• Shuchman M. Falling through the cracks — Virginia Tech and the restructuring of college mental health services. N Engl J Med 2007 Jul 12;357(2):105-110. [ CrossRef ]
• Eisenberg D, Downs MF, Golberstein E, Zivin K. Stigma and help seeking for mental health among college students. Med Care Res Rev 2009 Oct;66(5):522-541. [ CrossRef ] [ Medline ]
• Brooks SK, Webster RK, Smith LE, Woodland L, Wessely S, Greenberg N, et al. The psychological impact of quarantine and how to reduce it: rapid review of the evidence. Lancet 2020 Mar 14;395(10227):912-920 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Lai J, Ma S, Wang Y, Cai Z, Hu J, Wei N, et al. Factors associated with mental health outcomes among health care workers exposed to coronavirus disease 2019. JAMA Netw Open 2020 Mar 02;3(3):e203976 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Xie X, Xue Q, Zhou Y, Zhu K, Liu Q, Zhang J, et al. Mental health status among children in home confinement during the coronavirus disease 2019 outbreak in Hubei Province, China. JAMA Pediatr 2020 Apr 24:e201619 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Kirzinger A, Kearney A, Hamel L, Brodie M. KFF health tracking poll - early April 2020: the impact Of coronavirus on life In America. Kaiser Family Foundation. 2020 Apr 02.   URL: https://www.kff.org/health-reform/report/kff-health-tracking-poll-early-april-2020/ [accessed 2020-05-10]
• Nelson B, Pettitt A, Flannery J, Allen N. Rapid assessment of psychological and epidemiological correlates of COVID-19 concern, financial strain, and health-related behavior change in a large online sample. Int J Methods in Psychiatr Res 2020 Apr 13;21(3):169-184. [ CrossRef ]
• Cao W, Fang Z, Hou G, Han M, Xu X, Dong J, et al. The psychological impact of the COVID-19 epidemic on college students in China. Psychiatry Res 2020 May;287:112934 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Liu X, Liu J, Zhong X. Psychological state of college students during COVID-19 epidemic. SSRN J 2020 Mar 24:A. [ CrossRef ]
• Wang C, Zhao H. The impact of COVID-19 on anxiety in Chinese university students. Front Psychol 2020;11:1168. [ CrossRef ] [ Medline ]
• Bruffaerts R, Mortier P, Kiekens G, Auerbach RP, Cuijpers P, Demyttenaere K, et al. Mental health problems in college freshmen: prevalence and academic functioning. J Affect Disord 2018 Jan 01;225:97-103 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Zhai Y, Du X. Addressing collegiate mental health amid COVID-19 pandemic. Psychiatry Res 2020 Jun;288:113003 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Zhai Y, Du X. Mental health care for international Chinese students affected by the COVID-19 outbreak. Lancet Psychiatry 2020 Apr;7(4):e22 [ FREE Full text ] [ CrossRef ] [ Medline ]
• de Oliveira Araújo FJ, de Lima LSA, Cidade PIM, Nobre CB, Neto MLR. Impact of Sars-Cov-2 and its reverberation in global higher education and mental health. Psychiatry Res 2020 Jun;288:112977 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Holmes EA, O'Connor RC, Perry VH, Tracey I, Wessely S, Arseneault L, et al. Multidisciplinary research priorities for the COVID-19 pandemic: a call for action for mental health science. Lancet Psychiatry 2020 Jun;7(6):547-560 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav 1983 Dec;24(4):385-396. [ Medline ]
• Cohen S, Kessler R, Gordon L. Measuring stress: A Guide for Health and Social Scientists. New York, NY: Oxford University Press; 1997.
• Acharya L, Jin L, Collins W. College life is stressful today - emerging stressors and depressive symptoms in college students. J Am Coll Health 2018 Oct;66(7):655-664. [ CrossRef ] [ Medline ]
• Baghurst T, Kelley BC. An examination of stress in college students over the course of a semester. Health Promot Pract 2014 May;15(3):438-447. [ CrossRef ] [ Medline ]
• Zoom. Zoom Video Communications.   URL: https://zoom.us/
• Otter.ai.   URL: https://otter.ai/login
• Braun V, Clarke V. Using thematic analysis in psychology. Qualitative Res Psychol 2006 Jan;3(2):77-101. [ CrossRef ]
• Guest G, MacQueen K, Namey E. Applied Thematic Analysis. Thousand Oaks, CA: Sage Publications; 2011.
• Corbin J, Strauss A. Basics of Qualitative Research (3rd ed.): Techniques and Procedures for Developing Grounded Theory. Thousand Oaks, CA: Sage Publications; 2014.
• MAXQDA. VERBI Software.   URL: https://www.maxqda.com/
• Denovan A, Dagnall N, Dhingra K, Grogan S. Evaluating the perceived stress scale among UK university students: implications for stress measurement and management. Stud Higher Education 2017 Jun 22;44(1):120-133. [ CrossRef ]
• Martin JM. Stigma and student mental health in higher education. Higher Education Res Dev 2010 Jun;29(3):259-274. [ CrossRef ]
• Zuckerman DM. Stress, self-esteem, and mental health: how does gender make a difference? Sex Roles 1989 Apr;20(7-8):429-444. [ CrossRef ]
• Jacobson NC, Lekkas D, Price G, Heinz MV, Song M, O'Malley AJ, et al. Flattening the mental health curve: COVID-19 stay-at-home orders are associated with alterations in mental health search behavior in the United States. JMIR Ment Health 2020 Jun 01;7(6):e19347 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Zivin K, Eisenberg D, Gollust SE, Golberstein E. Persistence of mental health problems and needs in a college student population. J Affect Disord 2009 Oct;117(3):180-185. [ CrossRef ] [ Medline ]
• Zeng Y, Wang G, Xie C, Hu X, Reinhardt JD. Prevalence and correlates of depression, anxiety and symptoms of stress in vocational college nursing students from Sichuan, China: a cross-sectional study. Psychol Health Med 2019 Aug;24(7):798-811. [ CrossRef ] [ Medline ]
• Nahas ARMF, Elkalmi R, Al-Shami A, Elsayed T. Prevalence of depression among health sciences students: findings from a public university in Malaysia. J Pharm Bioallied Sci 2019;11(2):170-175 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Fitzpatrick O, Biesma R, Conroy RM, McGarvey A. Prevalence and relationship between burnout and depression in our future doctors: a cross-sectional study in a cohort of preclinical and clinical medical students in Ireland. BMJ Open 2019 May 01;9(4):e023297 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Heiligenstein E, Guenther G, Hsu K, Herman K. Depression and academic impairment in college students. J Am Coll Health 1996 Sep;45(2):59-64. [ CrossRef ] [ Medline ]
• Nastasi BK, Sarkar S, Varjas K, Jayasena A. Participatory model of mental health programming: lessons learned from work in a developing country. Sch Psychol Rev 1998;27(2):260-276.
• Cappella E, Jackson DR, Bilal C, Hamre BK, Soulé C. Bridging mental health and education in urban elementary schools: participatory research to inform intervention development. Sch Psychol Rev 2011;40(4):486-508.
• Eisenberg D, Hunt J, Speer N, Zivin K. Mental health service utilization among college students in the United States. J Nerv Ment Dis 2011 May;199(5):301-308. [ CrossRef ] [ Medline ]
• Husain W. Barriers in seeking psychological help: public perception in Pakistan. Community Ment Health J 2020 Jan;56(1):75-78. [ CrossRef ] [ Medline ]
• Fischer EP, McSweeney JC, Wright P, Cheney A, Curran GM, Henderson K, et al. Overcoming barriers to sustained engagement in mental health care: perspectives of rural veterans and providers. J Rural Health 2016 Sep;32(4):429-438. [ CrossRef ] [ Medline ]
• Torous J, Jän Myrick K, Rauseo-Ricupero N, Firth J. Digital mental health and COVID-19: using technology today to accelerate the curve on access and quality tomorrow. JMIR Ment Health 2020 Mar 26;7(3):e18848 [ FREE Full text ] [ CrossRef ] [ Medline ]
• Youn SJ, Trinh N, Shyu I, Chang T, Fava M, Kvedar J, et al. Using online social media, Facebook, in screening for major depressive disorder among college students. Int J Clin Health Psychol 2013 Jan;13(1):74-80. [ CrossRef ]
• Ginossar T, Nelson S. Reducing the health and digital divides: a model for using community-based participatory research approach to e-health interventions in low-income Hispanic communities. J Computer-Mediated Commun 2010;15(4):530-551. [ CrossRef ]
• COVID-19 projections. Institute for Health Metrics and Evaluation. Seattle, WA: University of Washington   URL: https://covid19.healthdata.org/united-states-of-america [accessed 2020-05-20]
• de Girolamo G, Cerveri G, Clerici M, Monzani E, Spinogatti F, Starace F, et al. Mental health in the coronavirus disease 2019 emergency-the Italian response. JAMA Psychiatry 2020 Apr 30:A. [ CrossRef ] [ Medline ]
• Zhou J, Liu L, Xue P, Yang X, Tang X. Mental health response to the COVID-19 outbreak in China. Am J Psychiatry 2020 Jul 01;177(7):574-575. [ CrossRef ] [ Medline ]
• Zhou X, Snoswell CL, Harding LE, Bambling M, Edirippulige S, Bai X, et al. The role of telehealth in reducing the mental health burden from COVID-19. Telemed J E Health 2020 Apr;26(4):377-379. [ CrossRef ] [ Medline ]
• Shore JH, Waugh M, Calderone J, Donahue A, Rodriguez J, Peters D, et al. Evaluation of telepsychiatry-enabled perinatal integrated care. Psychiatr Serv 2020 May 01;71(5):427-432. [ CrossRef ] [ Medline ]
• Huckins JF, daSilva AW, Wang W, Hedlund E, Rogers C, Nepal SK, et al. Mental health and behavior of college students during the early phases of the COVID-19 pandemic: longitudinal smartphone and ecological momentary assessment study. J Med Internet Res 2020 Jun 17;22(6):e20185 [ FREE Full text ] [ CrossRef ] [ Medline ]
• McAlonan GM, Lee AM, Cheung V, Cheung C, Tsang KW, Sham PC, et al. Immediate and sustained psychological impact of an emerging infectious disease outbreak on health care workers. Can J Psychiatry 2007 Apr;52(4):241-247. [ CrossRef ] [ Medline ]

Abbreviations

Edited by G Eysenbach, G Fagherazzi, J Torous; submitted 10.06.20; peer-reviewed by T Liu, V Hagger; comments to author 28.07.20; revised version received 01.08.20; accepted 15.08.20; published 03.09.20

©Changwon Son, Sudeep Hegde, Alec Smith, Xiaomei Wang, Farzan Sasangohar. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 03.09.2020.

This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.

• Open access
• Published: 20 September 2022

Factors that influence mental health of university and college students in the UK: a systematic review

• Fiona Campbell 1 ,
• Lindsay Blank 1 ,
• Anna Cantrell 1 ,
• Susan Baxter 1 ,
• Christopher Blackmore 1 ,
• Jan Dixon 1 &
• Elizabeth Goyder 1  

BMC Public Health volume  22 , Article number:  1778 ( 2022 ) Cite this article

117k Accesses

75 Citations

98 Altmetric

Metrics details

Worsening mental health of students in higher education is a public policy concern and the impact of measures to reduce transmission of COVID-19 has heightened awareness of this issue. Preventing poor mental health and supporting positive mental wellbeing needs to be based on an evidence informed understanding what factors influence the mental health of students.

To identify factors associated with mental health of students in higher education.

We undertook a systematic review of observational studies that measured factors associated with student mental wellbeing and poor mental health. Extensive searches were undertaken across five databases. We included studies undertaken in the UK and published within the last decade (2010–2020). Due to heterogeneity of factors, and diversity of outcomes used to measure wellbeing and poor mental health the findings were analysed and described narratively.

We included 31 studies, most of which were cross sectional in design. Those factors most strongly and consistently associated with increased risk of developing poor mental health included students with experiences of trauma in childhood, those that identify as LGBTQ and students with autism. Factors that promote wellbeing include developing strong and supportive social networks. Students who are prepared and able to adjust to the changes that moving into higher education presents also experience better mental health. Some behaviours that are associated with poor mental health include lack of engagement both with learning and leisure activities and poor mental health literacy.

Improved knowledge of factors associated with poor mental health and also those that increase mental wellbeing can provide a foundation for designing strategies and specific interventions that can prevent poor mental health and ensuring targeted support is available for students at increased risk.

Peer Review reports

Poor mental health of students in further and higher education is an increasing concern for public health and policy [ 1 , 2 , 3 , 4 ]. A 2020 Insight Network survey of students from 10 universities suggests that “1 in 5 students has a current mental health diagnosis” and that “almost half have experienced a serious psychological issue for which they felt they needed professional help”—an increase from 1 in 3 in the same survey conducted in 2018 [ 5 ]. A review of 105 Further Education (FE) colleges in England found that over a three-year period, 85% of colleges reported an increase in mental health difficulties [ 1 ]. Depression and anxiety were both prevalent and widespread in students; all colleges reported students experiencing depression and 99% reported students experiencing severe anxiety [ 5 , 6 ]. A UK cohort study found that levels of psychological distress increase on entering university [ 7 ], and recent evidence suggests that the prevalence of mental health problems among university students, including self-harm and suicide, is rising, [ 3 , 4 ] with increases in demand for services to support student mental health and reports of some universities finding a doubling of the number of students accessing support [ 8 ]. These common mental health difficulties clearly present considerable threat to the mental health and wellbeing of students but their impact also has educational, social and economic consequences such as academic underperformance and increased risk of dropping out of university [ 9 , 10 ].

Policy changes may have had an influence on the student experience, and on the levels of mental health problems seen in the student population; the biggest change has arguably been the move to widen higher education participation and to enable a more diverse demographic to access University education. The trend for widening participation has been continually rising since the late 1960s [ 11 ] but gained impetus in the 2000s through the work of the Higher Education Funding Council for England (HEFCE). Macaskill (2013) [ 12 ] suggests that the increased access to higher education will have resulted in more students attending university from minority groups and less affluent backgrounds, meaning that more students may be vulnerable to mental health problems, and these students may also experience greater challenges in making the transition to higher education.

Another significant change has been the introduction of tuition fees in 1998, which required students to self fund up to £1,000 per academic year. Since then, tuition fees have increased significantly for many students. With the abolition of maintenance grants, around 96% of government support for students now comes in the form of student loans [ 13 ]. It is estimated that in 2017, UK students were graduating with average debts of £50,000, and this figure was even higher for the poorest students [ 13 ]. There is a clear association between a student’s mental health and financial well-being [ 14 ], with “increased financial concern being consistently associated with worse health” [ 15 ].

The extent to which the increase in poor mental health is also being seen amongst non-students of a similar age is not well understood and warrants further study. However, the increase in poor mental health specifically within students in higher education highlights a need to understand what the risk factors are and what might be done within these settings to ensure young people are learning and developing and transitioning into adulthood in environments that promote mental wellbeing.

Commencing higher education represents a key transition point in a young person’s life. It is a stage often accompanied by significant change combined with high expectations of high expectations from students of what university life will be like, and also high expectations from themselves and others around their own academic performance. Relevant factors include moving away from home, learning to live independently, developing new social networks, adjusting to new ways of learning, and now also dealing with the additional greater financial burdens that students now face.

The recent global COVID-19 pandemic has had considerable impact on mental health across society, and there is concern that younger people (ages 18–25) have been particularly affected. Data from Canada [ 16 ] indicate that among survey respondents, “almost two-thirds (64%) of those aged 15 to 24 reported a negative impact on their mental health, while just over one-third (35%) of those aged 65 and older reported a negative impact on their mental health since physical distancing began” (ibid, p.4). This suggests that older adults are more prepared for the kind of social isolation which has been brought about through the response to COVID-19, whereas young adults have found this more difficult to cope with. UK data from the National Union of Students reports that for over half of UK students, their mental health is worse than before the pandemic [ 17 ]. Before COVID-19, students were already reporting increasing levels of mental health problems [ 2 ], but the COVID-19 pandemic has added a layer of “chronic and unpredictable” stress, creating the perfect conditions for a mental health crisis [ 18 ]. An example of this is the referrals (both urgent and routine) of young people with eating disorders for treatment in the NHS which almost doubled in number from 2019 to 2020 [ 19 ]. The travel restrictions enforced during the pandemic have also impacted on student mental health, particularly for international students who may have been unable to commence studies or go home to see friends and family during holidays [ 20 ].

With the increasing awareness and concern in the higher education sector and national bodies regarding student mental health has come increasing focus on how to respond. Various guidelines and best practice have been developed, e.g. ‘Degrees of Disturbance’ [ 21 ], ‘Good Practice Guide on Responding to Student Mental Health Issues: Duty of Care Responsibilities for Student Services in Higher Education’ [ 22 ] and the recent ‘The University Mental Health Charter’ [ 2 ]. Universities UK produced a Good Practice Guide in 2015 called “Student mental wellbeing in higher education” [ 23 ]. An increasing number of initiatives have emerged that are either student-led or jointly developed with students, and which reflect the increasing emphasis students and student bodies place on mental health and well-being and the increased demand for mental health support: Examples include: Nightline— www.nightline.ac.uk , Students Against Depression— www.studentsagainstdepression.org , Student Minds— www.studentminds.org.uk/student-minds-and-mental-wealth.html and The Alliance for Student-Led Wellbeing— www.alliancestudentwellbeing.weebly.com/ .

Although requests for professional support have increased substantially [ 24 ] only a third of students with mental health problems seek support from counselling services in the UK [ 12 ]. Many students encounter barriers to seeking help such as stigma or lack of awareness of services [ 25 ], and without formal support or intervention, there is a risk of deterioration. FE colleges and universities have identified the need to move beyond traditional forms of support and provide alternative, more accessible interventions aimed at improving mental health and well-being. Higher education institutions have a unique opportunity to identify, prevent, and treat mental health problems because they provide support in multiple aspects of students’ lives including academic studies, recreational activities, pastoral and counselling services, and residential accommodation.

In order to develop services that better meet the needs of students and design environments that are supportive of developing mental wellbeing it is necessary to explore and better understand the factors that lead to poor mental health in students.

Research objectives

The overall aim of this review was to identify, appraise and synthesise existing research evidence that explores the aetiology of poor mental health and mental wellbeing amongst students in tertiary level education. We aimed to gain a better understanding of the mechanisms that lead to poor mental health amongst tertiary level students and, in so doing, make evidence-based recommendations for policy, practice and future research priorities. Specific objectives in line with the project brief were to:

To co-produce with stakeholders a conceptual framework for exploring the factors associated with poorer mental health in students in tertiary settings. The factors may be both predictive, identifying students at risk, or causal, explaining why they are at risk. They may also be protective, promoting mental wellbeing.

To conduct a review drawing on qualitative studies, observational studies and surveys to explore the aetiology of poor mental health in students in university and college settings and identify factors which promote mental wellbeing amongst students.

To identify evidence-based recommendations for policy, service provision and future research that focus on prevention and early identification of poor mental health

Methodology

Identification of relevant evidence.

The following inclusion criteria were used to guide the development of the search strategy and the selection of studies.

We included students from a variety of further education settings (16 yrs + or 18 yrs + , including mature students, international students, distance learning students, students at specific transition points).

Universities and colleges in the UK. We were also interested in the context prior to the beginning of tertiary education, including factors during transition from home and secondary education or existing employment to tertiary education.

Any factor shown to be associated with mental health of students in tertiary level education. This included clinical indicators such as diagnosis and treatment and/or referral for depression and anxiety. Self-reported measures of wellbeing, happiness, stress, anxiety and depression were included. We did not include measures of academic achievement or engagement with learning as indicators of mental wellbeing.

Study design

We included cross-sectional and longitudinal studies that looked at factors associated with mental health outcomes in Table 5 .

Data extraction and quality appraisal

We extracted and tabulated key data from the included papers. Data extraction was undertaken by one reviewer, with a 10% sample checked for accuracy and consistency The quality of the included studies were evaluated using the Newcastle-Ottawa Scale [ 26 ] and the findings of the quality appraisal used in weighting the strength of associations and also identifying gaps for future high quality research.

Involvement of stakeholders

We recruited students, ex-students and parents of students to a public involvement group which met on-line three times during the process of the review and following the completion of the review. During a workshop meeting we asked for members of the group to draw on their personal experiences to suggest factors which were not mentioned in the literature.

Methods of synthesis

We undertook a narrative synthesis [ 27 ] due to the heterogeneity in the exposures and outcomes that were measured across the studies. Data showing the direction of effects and the strength of the association (correlation coefficients) were recorded and tabulated to aid comparison between studies.

Search strategy

Searches were conducted in the following electronic databases: Medline, Applied Social Sciences Index and Abstracts (ASSIA), International Bibliography of Social Sciences (IBSS), Science,PsycINFO and Science and Social Sciences Ciatation Indexes. Additional searches of grey literature, and reference lists of included studies were also undertaken.

The search strategy combined a number of terms relating to students and mental health and risk factors. The search terms included both subject (MeSH) and free-text searches. The searches were limited to papers about humans in English, published from 2010 to June 2020. The flow of studies through the review process is summarised in Fig.  1 .

Flow diagram

The full search strategy for Medline is provided in Appendix 1 .

Thirty-one quantitative, observational studies (39 papers) met the inclusion criteria. The total number of students that participated in the quantitative studies was 17,476, with studies ranging in size from 57 to 3706. Eighteen studies recruited student participants from only one university; five studies (10 publications) [ 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 ] included seven or more universities. Six studies (7 publications) [ 35 , 36 , 37 , 38 , 39 , 40 , 41 ] only recruited first year students, while the majority of studies recruited students from a range of year groups. Five studies [ 39 , 42 , 43 , 44 , 45 ] recruited only, or mainly, psychology students which may impact on the generalisability of findings. A number of studies focused on students studying particular subjects including: nursing [ 46 ] medicine [ 47 ], business [ 48 ], sports science [ 49 ]. One study [ 50 ] recruited LGBTQ (lesbian, gay, bisexual, transgender, intersex, queer/questioning) students, and one [ 51 ] recruited students who had attended hospital having self-harmed. In 27 of the studies, there were more female than male participants. The mean age of the participants ranged from 19 to 28 years. Ethnicity was not reported in 19 of the studies. Where ethnicity was reported, the proportion that were ‘white British’ ranged from 71 – 90%. See Table 1 for a summary of the characteristics of the included studies and the participants.

Design and quality appraisal of the included studies

The majority of included studies ( n  = 22) were cross-sectional surveys. Nine studies (10 publications) [ 35 , 36 , 39 , 41 , 43 , 50 , 51 , 52 , 53 , 62 ] were longitudinal in design, recording survey data at different time points to explore changes in the variables being measured. The duration of time that these studies covered ranged from 19 weeks to 12 years. Most of the studies ( n  = 22) only recruited participants from a single university. The use of one university setting and the large number of studies that recruited only psychology students weakens the wider applicability of the included studies.

Quantitative variables

Included studies ( n  = 31) measured a wide range of variables and explored their association with poor mental health and wellbeing. These included individual level factors: age, gender, sexual orientation, ethnicity and a range of psychological variables. They also included factors that related to mental health variables (family history, personal history and mental health literacy), pre-university factors (childhood trauma and parenting behaviour. University level factors including social isolation, adjustment and engagement with learning. Their association was measured against different measures of positive mental health and poor mental health.

Measurement of association and the strength of that association has some limitations in addressing our research question. It cannot prove causality, and nor can it capture fully the complexity of the inter-relationship and compounding aspect of the variables. For example, the stress of adjustment may be manageable, until it is combined with feeling isolated and out of place. Measurement itself may also be misleading, only capturing what is measureable, and may miss variables that are important but not known. We included both qualitative and PPI input to identify missed but important variables.

The wide range of variables and different outcomes, with few studies measuring the same variable and outcomes, prevented meta-analyses of findings which are therefore described narratively.

The variables described were categorised during the analyses into the following categories:

Vulnerabilities – factors that are associated with poor mental health

Individual level factors including; age, ethnicity, gender and a range of psychological variables were all measured against different mental health outcomes including depression, anxiety, paranoia, and suicidal behaviour, self-harm, coping and emotional intelligence.

Six studies [ 40 , 42 , 47 , 50 , 60 , 63 ] examined a student’s ages and association with mental health. There was inconsistency in the study findings, with studies finding that age (21 or older) was associated with fewer depressive symptoms, lower likelihood of suicide ideation and attempt, self-harm, and positively associated with better coping skills and mental wellbeing. This finding was not however consistent across studies and the association was weak. Theoretical models that seek to explain this mechanism have suggested that older age groups may cope better due to emotion-regulation strategies improving with age [ 67 ]. However, those over 30 experienced greater financial stress than those aged 17-19 in another study [ 63 ].

Sexual orientation

Four studies [ 33 , 40 , 64 , 68 ] examined the association between poor mental health and sexual orientation status. In all of the studies LGBTQ students were at significantly greater risk of mental health problems including depression [ 40 ], anxiety [ 40 ], suicidal behaviour [ 33 , 40 , 64 ], self harm [ 33 , 40 , 64 ], use of mental health services [ 33 ] and low levels of wellbeing [ 68 ]. The risk of mental health problems in these students compared with heterosexual students, ranged from OR 1.4 to 4.5. This elevated risk may reflect the greater levels of isolation and discrimination commonly experienced by minority groups.

Nine studies [ 33 , 38 , 39 , 40 , 42 , 47 , 50 , 60 , 63 ] examined whether gender was associated mental health variables. Two studies [ 33 , 47 ] found that being female was statistically significantly associated with use of mental health services, having a current mental health problem, suicide risk, self harm [ 33 ] and depression [ 47 ]. The results were not consistent, with another study [ 60 ] finding the association was not significant. Three studies [ 39 , 40 , 42 ] that considered mediating variables such as adaptability and coping found no difference or very weak associations.

Two studies [ 47 , 60 ] examined the extent to which ethnicity was associated with mental health One study [ 47 ] reported that the risks of depression were significantly greater for those who categorised themselves as non-white (OR 8.36 p = 0.004). Non-white ethnicity was also associated with poorer mental health in another cross-sectional study [ 63 ]. There was no significant difference in the McIntyre et al. (2018) study [ 60 ]. The small number of participants from ethnic minority groups represented across the studies means that this data is very limited.

Family factors

Six studies [ 33 , 40 , 42 , 50 , 60 ] explored the association of a concept that related to a student’s experiences in childhood and before going to university. Three studies [ 40 , 50 , 60 ] explored the impact of ACEs (Adverse Childhood Experiences) assessed using the same scale by Feletti (2009) [ 69 ] and another explored the impact of abuse in childhood [ 46 ]. Two studies examined the impact of attachment anxiety and avoidance [ 42 ], and parental acceptance [ 46 , 59 ]. The studies measured different mental health outcomes including; positive and negative affect, coping, suicide risk, suicide attempt, current mental health problem, use of mental health services, psychological adjustment, depression and anxiety.

The three studies that explored the impact of ACE’s all found a significant and positive relationship with poor mental health amongst university students. O’Neill et al. (2018) [ 50 ] in a longitudinal study ( n  = 739) showed that there was in increased likelihood in self-harm and suicidal behaviours in those with either moderate or high levels of childhood adversities (OR:5.5 to 8.6) [ 50 ]. McIntyre et al. (2018) [ 60 ] ( n  = 1135) also explored other dimensions of adversity including childhood trauma through multiple regression analysis with other predictive variables. They found that childhood trauma was significantly positively correlated with anxiety, depression and paranoia (ß = 0.18, 0.09, 0.18) though the association was not as strong as the correlation seen for loneliness (ß = 0.40) [ 60 ]. McLafferty et al. (2019) [ 40 ] explored the compounding impact of childhood adversity and negative parenting practices (over-control, overprotection and overindulgence) on poor mental health (depression OR 1.8, anxiety OR 2.1 suicidal behaviour OR 2.3, self-harm OR 2.0).

Gaan et al.’s (2019) survey of LGBTQ students ( n  = 1567) found in a multivariate analyses that sexual abuse, other abuse from violence from someone close, and being female had the highest odds ratios for poor mental health and were significantly associated with all poor mental health outcomes [ 33 ].

While childhood trauma and past abuse poses a risk to mental health for all young people it may place additional stresses for students at university. Entry to university represents life stage where there is potential exposure to new and additional stressors, and the possibility that these students may become more isolated and find it more difficult to develop a sense of belonging. Students may be separated for the first time from protective friendships. However, the mechanisms that link childhood adversities and negative psychopathology, self-harm and suicidal behaviour are not clear [ 40 ]. McLafferty et al. (2019) also measured the ability to cope and these are not always impacted by childhood adversities [ 40 ]. They suggest that some children learn to cope and build resilience that may be beneficial.

McLafferty et al. (2019) [ 40 ] also studied parenting practices. Parental over-control and over-indulgence was also related to significantly poorer coping (OR -0.075 p  < 0.05) and this was related to developing poorer coping scores (OR -0.21 p  < 0.001) [ 40 ]. These parenting factors only became risk factors when stress levels were high for students at university. It should be noted that these studies used self-report, and responses regarding views of parenting may be subjective and open to interpretation. Lloyd et al.’s (2014) survey found significant positive correlations between perceived parental acceptance and students’ psychological adjustment, with paternal acceptance being the stronger predictor of adjustment.

Autistic students may display social communication and interaction deficits that can have negative emotional impacts. This may be particularly true during young adulthood, a period of increased social demands and expectations. Two studies [ 56 ] found that those with autism had a low but statistically significant association with poor social problem-solving skills and depression.

Mental health history

Three studies [ 47 , 51 , 68 ] investigated mental health variables and their impact on mental health of students in higher education. These included; a family history of mental illness and a personal history of mental illness.

Students with a family history or a personal history of mental illness appear to have a significantly greater risk of developing problems with mental health at university [ 47 ]. Mahadevan et al. (2010) [ 51 ] found that university students who self-harm have a significantly greater risk (OR 5.33) of having an eating disorder than a comparison group of young adults who self-harm but are not students.

Buffers – factors that are protective of mental wellbeing

Psychological factors.

Twelve studies [ 29 , 39 , 40 , 41 , 42 , 43 , 46 , 49 , 54 , 58 , 64 ] assessed the association of a range of psychological variables and different aspects of mental wellbeing and poor mental health. We categorised these into the following two categories: firstly, psychological variables measuring an individual’s response to change and stressors including adaptability, resilience, grit and emotional regulation [ 39 , 40 , 41 , 42 , 43 , 46 , 49 , 54 , 58 ] and secondly, those that measure self-esteem and body image [ 29 , 64 ].

The evidence from the eight included quantitative studies suggests that students with psychological strengths including; optimism, self-efficacy [ 70 ], resilience, grit [ 58 ], use of positive reappraisal [ 49 ], helpful coping strategies [ 42 ] and emotional intelligence [ 41 , 46 ] are more likely to experience greater mental wellbeing (see Table 2 for a description of the psychological variables measured). The positive association between these psychological strengths and mental well-being had a positive affect with associations ranging from r  = 0.2–0.5 and OR1.27 [ 41 , 43 , 46 , 49 , 54 ] (low to moderate strength of association). The negative associations with depressive symptoms are also statistically significant but with a weaker association ( r  = -0.2—0.3) [ 43 , 49 , 54 ].

Denovan (2017a) [ 43 ] in a longitudinal study found that the association between psychological strengths and positive mental wellbeing was not static and that not all the strengths remained statistically significant over time. The only factors that remained significant during the transition period were self-efficacy and optimism, remaining statistically significant as they started university and 6 months later.

Parental factors

Only one study [ 59 ] explored family factors associated with the development of psychological strengths that would equip young people as they managed the challenges and stressors encountered during the transition to higher education. Lloyd et al. (2014) [ 59 ] found that perceived maternal and paternal acceptance made significant and unique contributions to students’ psychological adjustment. Their research methods are limited by their reliance on retrospective measures and self-report measures of variables, and these results could be influenced by recall bias.

Two studies [ 29 , 64 ] considered the impact of how individuals view themselves on poor mental health. One study considered the impact of self-esteem and the association with non-accidental self-injury (NSSI) and suicide attempt amongst 734 university students. As rates of suicide and NSSI are higher amongst LGBT (lesbian, gay, bisexual, transgender) students, the prevalence of low self-esteem was compared. There was a low but statistically significant association between low self-esteem and NSSI, though not for suicide attempt. A large survey, including participants from seven universities [ 42 ] compared depressive symptoms in students with marked body image concerns, reporting that the risk of depressive symptoms was greater (OR 2.93) than for those with lower levels of body image concerns.

Mental health literacy and help seeking behaviour

Two studies [ 48 , 68 ] investigated attitudes to mental illness, mental health literacy and help seeking for mental health problems.

University students who lack sufficient mental health literacy skills to be able to recognise problems or where there are attitudes that foster shame at admitting to having mental health problems can result in students not recognising problems and/or failing to seek professional help [ 48 , 68 ]. Gorcyznski et al. (2017) [ 68 ] found that women and those who had a history of previous mental health problems exhibited significantly higher levels of mental health literacy. Greater mental health literacy was associated with an increased likelihood that individuals would seek help for mental health problems. They found that many students find it hard to identify symptoms of mental health problems and that 42% of students are unaware of where to access available resources. Of those who expressed an intention to seek help for mental health problems, most expressed a preference for online resources, and seeking help from family and friends, rather than medical professionals such as GPs.

Kotera et al. (2019) [ 48 ] identified self-compassion as an explanatory variable, reducing social comparison, promoting self-acceptance and recognition that discomfort is an inevitable human experience. The study found a strong, significant correlation between self-compassion and mental health symptoms ( r  = -0.6. p  < 0.01).

There again appears to be a cycle of reinforcement, where poor mental health symptoms are felt to be a source of shame and become hidden, help is not sought, and further isolation ensues, leading to further deterioration in mental health. Factors that can interrupt the cycle are self-compassion, leading to more readiness to seek help (see Fig.  2 ).

Poor mental health – cycles of reinforcement

Social networks

Nine studies [ 33 , 38 , 41 , 46 , 51 , 54 , 60 , 64 , 65 ] examined the concepts of loneliness and social support and its association with mental health in university students. One study also included students at other Higher Education Institutions [ 46 ]. Eight of the studies were surveys, and one was a retrospective case control study to examine the differences between university students and age-matched young people (non-university students) who attended hospital following deliberate self-harm [ 51 ].

Included studies demonstrated considerable variation in how they measured the concepts of social isolation, loneliness, social support and a sense of belonging. There were also differences in the types of outcomes measured to assess mental wellbeing and poor mental health. Grouping the studies within a broad category of ‘social factors’ therefore represents a limitation of this review given that different aspects of the phenomena may have been being measured. The tools used to measure these variables also differed. Only one scale (The UCLA loneliness scale) was used across multiple studies [ 41 , 60 , 65 ]. Diverse mental health outcomes were measured across the studies including positive affect, flourishing, self-harm, suicide risk, depression, anxiety and paranoia.

Three studies [ 41 , 60 , 62 ] measuring loneliness, two longitudinally [ 41 , 62 ], found a consistently positive association between loneliness and poor mental health in university students. Greater loneliness was linked to greater anxiety, stress, depression, poor general mental health, paranoia, alcohol abuse and eating disorder problems. The strength of the correlations ranged from 0–3-0.4 and were all statistically significant (see Tables 3 and 4 ). Loneliness was the strongest overall predictor of mental distress, of those measured. A strong identification with university friendship groups was most protective against distress relative to other social identities [ 60 ]. Whether poor mental health is the cause, or the result of loneliness was explored further in the studies. The results suggest that for general mental health, stress, depression and anxiety, loneliness induces or exacerbates symptoms of poor mental health over time [ 60 , 62 ]. The feedback cycle is evident, with loneliness leading to poor mental health which leads to withdrawal from social contacts and further exacerbation of loneliness.

Factors associated with protecting against loneliness by fostering supportive friendships and promoting mental wellbeing were also identified. Beliefs about the value of ‘leisure coping’, and attributes of resilience and emotional intelligence had a moderate, positive and significant association with developing mental wellbeing and were explored in three studies [ 46 , 54 , 66 ].

The transition to and first year at university represent critical times when friendships are developed. Thomas et al. (2020) [ 65 ] explored the factors that predict loneliness in the first year of university. A sense of community and higher levels of ‘social capital’ were significantly associated with lower levels of loneliness. ‘Social capital’ scales measure the development of emotionally supportive friendships and the ability to adjust to the disruption of old friendships as students transition to university. Students able to form close relationships within their first year at university are less likely to experience loneliness (r-0.09, r- 0.36, r- 0.34). One study [ 38 ] investigating the relationship between student experience and being the first in the family to attend university found that these students had lower ratings for peer group interactions.

Young adults at university and in higher education are facing multiple adjustments. Their ability to cope with these is influenced by many factors. Supportive friendships and a sense of belonging are factors that strengthen coping. Nightingale et al. (2012) undertook a longitudinal study to explore what factors were associated with university adjustment in a sample of first year students ( n  = 331) [ 41 ]. They found that higher skills of emotion management and emotional self-efficacy were predictive of stable adjustment. These students also reported the lowest levels of loneliness and depression. This group had the skills to recognise their emotions and cope with stressors and were confident to access support. Students with poor emotion management and low levels of emotional self-efficacy may benefit from intervention to support the development of adaptive coping strategies and seeking support.

The positive and negative feedback loops

The relationship between the variables described appeared to work in positive and negative feedback loops with high levels of social capital easing the formation of a social network which acts as a critical buffer to stressors (see Fig.  3 ). Social networks and support give further strengthening and reinforcement, stimulating positive affect, engagement and flourishing. These, in turn, widen and deepen social networks for support and enhance a sense of wellbeing. Conversely young people who enter the transition to university/higher education with less social capital are less likely to identify with and locate a social network; isolation may follow, along with loneliness, anxiety, further withdrawal from contact with social networks and learning, and depression.

Triggers – factors that may act in combination with other factors to lead to poor mental health

Stress is seen as playing a key role in the development of poor mental health for students in higher education. Theoretical models and empirical studies have suggested that increases in stress are associated with decreases in student mental health [ 12 , 43 ]. Students at university experience the well-recognised stressors associated with academic study such as exams and course work. However, perhaps less well recognised are the processes of transition, requiring adapting to a new social and academic environment (Fisher 1994 cited by Denovan 2017a) [ 43 ]. Por et al. (2011) [ 46 ] in a small ( n  = 130 prospective survey found a statistically significant correlation between higher levels of emotional intelligence and lower levels of perceived stress ( r  = 0.40). Higher perceived stress was also associated with negative affect in two studies [ 43 , 46 ], and strongly negatively associated with positive affect (correlation -0.62) [ 54 ].

University variables

Eleven studies [ 35 , 39 , 47 , 51 , 52 , 54 , 60 , 63 , 65 , 83 , 84 ] explored university variables, and their association with mental health outcomes. The range of factors and their impact on mental health variables is limited, and there is little overlap. Knowledge gaps are shown by factors highlighted by our PPI group as potentially important but not identified in the literature (see Table 5 ). It should be noted that these may reflect the focus of our review, and our exclusion of intervention studies which may evaluate university factors.

High levels of perceived stress caused by exam and course work pressure was positively associated with poor mental health and lack of wellbeing [ 51 , 52 , 54 ]. Other potential stressors including financial anxieties and accommodation factors appeared to be less consistently associated with mental health outcomes [ 35 , 38 , 47 , 51 , 60 , 62 ]. Important mediators and buffers to these stressors are coping strategies and supportive networks (see conceptual model Appendix 2 ). One impact of financial pressures was that students who worked longer hours had less interaction with their peers, limiting the opportunities for these students to benefit from the protective effects of social support.

Red flags – behaviours associated with poor mental health and/or wellbeing

Engagement with learning and leisure activities.

Engagement with learning activities was strongly and positively associated with characteristics of adaptability [ 39 ] and also happiness and wellbeing [ 52 ] (see Fig.  4 ). Boulton et al. (2019) [ 52 ] undertook a longitudinal survey of undergraduate students at a campus-based university. They found that engagement and wellbeing varied during the term but were strongly correlated.

Engagement and wellbeing

Engagement occurred in a wide range of activities and behaviours. The authors suggest that the strong correlation between all forms of engagement with learning has possible instrumental value for the design of systems to monitor student engagement. Monitoring engagement might be used to identify changes in the behaviour of individuals to assist tutors in providing support and pastoral care. Students also were found to benefit from good induction activities provided by the university. Greater induction satisfaction was positively and strongly associated with a sense of community at university and with lower levels of loneliness [ 65 ].

The inte r- related nature of these variables is depicted in Fig.  4 . Greater adaptability is strongly associated with more positive engagement in learning and university life. More engagement is associated with higher mental wellbeing.

Denovan et al. (2017b) [ 54 ] explored leisure coping, its psychosocial functions and its relationship with mental wellbeing. An individual’s beliefs about the benefits of leisure activities to manage stress, facilitate the development of companionship and enhance mood were positively associated with flourishing and were negatively associated with perceived stress. Resilience was also measured. Resilience was strongly and positively associated with leisure coping beliefs and with indicators of mental wellbeing. The authors conclude that resilient individuals are more likely to use constructive means of coping (such as leisure coping) to proactively cultivate positive emotions which counteract the experience of stress and promote wellbeing. Leisure coping is predictive of positive affect which provides a strategy to reduce stress and sustain coping. The belief that friendships acquired through leisure provide social support is an example of leisure coping belief. Strong emotionally attached friendships that develop through participation in shared leisure pursuits are predictive of higher levels of well-being. Friendship bonds formed with fellow students at university are particularly important for maintaining mental health, and opportunities need to be developed and supported to ensure that meaningful social connections are made.

The ‘broaden-and-build theory’ (Fredickson 2004 [ 85 ] cited by [ 54 ]) may offer an explanation for the association seen between resilience, leisure coping and psychological wellbeing. The theory is based upon the role that positive and negative emotions have in shaping human adaptation. Positive emotions broaden thinking, enabling the individual to consider a range of ways of dealing with and adapting to their environment. Conversely, negative emotions narrow thinking and limit options for adapting. The former facilitates flourishing, facilitating future wellbeing. Resilient individuals are more likely to use constructive means of coping which generate positive emotion (Tugade & Fredrickson 2004 [ 86 ], cited by [ 54 ]). Positive emotions therefore lead to growth in coping resources, leading to greater well-being.

Health behaviours at university

Seven studies [ 29 , 31 , 38 , 45 , 51 , 54 , 66 ] examined how lifestyle behaviours might be linked with mental health outcomes. The studies looked at leisure activities [ 63 , 80 ], diet [ 29 ], alcohol use [ 29 , 31 , 38 , 51 ] and sleep [ 45 ].

Depressive symptoms were independently associated with problem drinking and possible alcohol dependence for both genders but were not associated with frequency of drinking and heavy episodic drinking. Students with higher levels of depressive symptoms reported significantly more problem drinking and possible alcohol dependence [ 31 ]. Mahadevan et al. (2010) [ 51 ] compared students and non-students seen in hospital for self-harm and found no difference in harmful use of alcohol and illicit drugs.

Poor sleep quality and increased consumption of unhealthy foods were also positively associated with depressive symptoms and perceived stress [ 29 ]. The correlation with dietary behaviours and poor mental health outcomes was low, but also confirmed by the negative correlation between less perceived stress and depressive symptoms and consumption of a healthier diet.

Physical activity and participation in leisure pursuits were both strongly correlated with mental wellbeing ( r  = 0.4) [ 54 ], and negatively correlated with depressive symptoms and anxiety ( r  = -0.6, -0.7) [ 66 ].

Thirty studies measuring the association between a wide range of factors and poor mental health and mental wellbeing in university and college students were identified and included in this review. Our purpose was to identify the factors that contribute to the growing prevalence of poor mental health amongst students in tertiary level education within the UK. We also aimed to identify factors that promote mental wellbeing and protect against deteriorating poor mental health.

Loneliness and social isolation were strongly associated with poor mental health and a sense of belonging and a strong support network were strongly associated with mental wellbeing and happiness. These associations were strongly positive in the eight studies that explored them and are consistent with other meta-analyses exploring the link between social support and mental health [ 87 ].

Another factor that appeared to be protective was older age when starting university. A wide range of personal traits and characteristics were also explored. Those associated with resilience, ability to adjust and better coping led to improved mental wellbeing. Better engagement appeared as an important mediator to potentially explain the relationship between these two variables. Engagement led to students being able to then tap into those features that are protective and promoting of mental wellbeing.

Other important risk factors for poor mental wellbeing that emerged were those students with existing or previous mental illness. Students on the autism spectrum and those with poor social problem-solving also were more likely to suffer from poor mental health. Negative self-image was also associated with poor mental health at university. Eating disorders were strongly associated with poor mental wellbeing and were found to be far more of a risk in students at university than in a comparative group of young people not in higher education. Other studies of university students also found that pre-existing poor mental health was a strong predictor of poor mental health in university students [ 88 ].

At a family level, the experience of childhood trauma and adverse experiences including, for example, neglect, household dysfunction or abuse, were strongly associated with poor mental health in young people at university. Students with a greater number of ‘adverse childhood experiences’ were at significantly greater risk of poor mental health than those students without experience of childhood trauma. This was also identified in a review of factors associated with depression and suicide related outcomes amongst university undergraduate students [ 88 ].

Our findings, in contrast to findings from other studies of university students, did not find that female gender associated with poor mental health and wellbeing, and it also found that being a mature student was protective of mental wellbeing.

Exam and course work pressure was associated with perceived stress and poor mental health. A lack of engagement with learning activities was also associated with poor mental health. A number of variables were not consistently shown to be associated with poor mental health including financial concerns and accommodation factors. Very little evidence related to university organisation or support structures was assessed in the evidence. One study found that a good induction programme had benefits for student mental wellbeing and may be a factor that enables students to become a part of a social network positive reinforcement cycle. Involvement in leisure activities was also found to be associated with improved coping strategies and better mental wellbeing. Students with poorer mental health tended to also eat in a less healthy manner, consume more harmful levels of alcohol, and experience poorer sleep.

This evidence review of the factors that influence mental health and wellbeing indicate areas where universities and higher education settings could develop and evaluate innovations in practice. These include:

Interventions before university to improve preparation of young people and their families for the transition to university.

Exploratory work to identify the acceptability and feasibility of identifying students at risk or who many be exhibiting indications of deteriorating mental health

Interventions that set out to foster a sense of belonging and identify

Creating environments that are helpful for building social networks

Improving mental health literacy and access to high quality support services

This review has a number of limitations. Most of the included studies were cross-sectional in design, with a small number being longitudinal ( n  = 7), following students over a period of time to observe changes in the outcomes being measured. Two limitations of these sources of data is that they help to understand associations but do not reveal causality; secondly, we can only report the findings for those variables that were measured, and we therefore have to support causation in assuming these are the only factors that are related to mental health.

Furthermore, our approach has segregated and categorised variables in order to better understand the extent to which they impact mental health. This approach does not sufficiently explore or reveal the extent to which variables may compound one another, for example, feeling the stress of new ways of learning may not be a factor that influences mental health until it is combined with a sense of loneliness, anxiety about financial debt and a lack of parental support. We have used our PPI group and the development of vignettes of their experiences to seek to illustrate the compounding nature of the variables identified.

We limited our inclusion criteria to studies undertaken in the UK and published within the last decade (2009–2020), again meaning we may have limited our inclusion of relevant data. We also undertook single data extraction of data which may increase the risk of error in our data.

Understanding factors that influence students’ mental health and wellbeing offers the potential to find ways to identify strategies that enhance the students’ abilities to cope with the challenges of higher education. This review revealed a wide range of variables and the mechanisms that may explain how they impact upon mental wellbeing and increase the risk of poor mental health amongst students. It also identified a need for interventions that are implemented before young people make the transition to higher education. We both identified young people who are particularly vulnerable and the factors that arise that exacerbate poor mental health. We highlight that a sense of belonging and supportive networks are important buffers and that there are indicators including lack of engagement that may enable early intervention to provide targeted and appropriate support.

Availability of data and materials

Further details of the study and the findings can be provided on request to the lead author ([email protected]).

Association of Colleges. Association of Colleges’ survey on students with mental health conditions in further education. London: 2017.

Google Scholar  

Hughes G, Spanner L. The University Mental Health Charter. Leeds: Student Minds; 2019.

Sivertsen B, Hysing M, Knapstad M, Harvey AG, Reneflot A, Lønning KJ, et al. Suicide attempts and non-suicidal self-harm among university students: prevalence study. BJPsych Open. 2019;5(2):e26.

Article   PubMed   PubMed Central   Google Scholar  

Storrie K, Ahern K, Tuckett A. A systematic review: students with mental health problems—a growing problem. Int J Nurs Pract. 2010;16(1):1–6.

Article   PubMed   Google Scholar  

Pereira S, Reay K, Bottell J, Walker L, Dzikiti C, Platt C, Goodrham C. Student Mental Health Survey 2018: A large scale study into the prevalence of student mental illness within UK universities. 2019.

Bayram N, Bilgel N. The prevalence and socio-demographic correlations of depression, anxiety and stress among a group of university students. Soc Psychiatry Psychiatr Epidemiol. 2008;43(8):667–72.

Bewick B, Koutsopoulou G, Miles J, Slaa E, Barkham M. Changes in undergraduate students’ psychological well-being as they progress through university. Stud High Educ. 2010;35(6):633–45.

Article   Google Scholar  

Thorley C. Not By Degrees: Not by degrees: Improving student mental health in the UK’s universities. London: IPPR; 2017.

Eisenberg D, Golberstein E, Hunt JB. Mental health and academic success in college. BE J Econ Anal Pol. 2009;9(1):1–37.

Hysenbegasi A, Hass SL, Rowland CR. The impact of depression on the academic productivity of university students. J Ment Health Policy Econ. 2005;8(3):145.

PubMed   Google Scholar  

Chowdry H, Crawford C, Dearden L, Goodman A, Vignoles A. Widening participation in higher education: analysis using linked administrative data. J R Stat Soc A Stat Soc. 2013;176(2):431–57.

Macaskill A. The mental health of university students in the United Kingdom. Br J Guid Couns. 2013;41(4):426–41.

Belfield C, Britton J, van der Erve L. Higher Education finance reform: Raising the repayment threshold to£ 25,000 and freezing the fee cap at £ 9,250: Institute for Fiscal Studies Briefing note. London: Institute for Fiscal Studies; 2017. Available from https://ifs.org.uk/publications/9964 .

Benson-Egglenton J. The financial circumstances associated with high and low wellbeing in undergraduate students: a case study of an English Russell Group institution. J Furth High Educ. 2019;43(7):901–13.

Jessop DC, Herberts C, Solomon L. The impact of financial circumstances on student health. Br J Health Psychol. 2005;10(3):421–39.

(2020) SCSC. Canadians’ mental health during the COVID-19 pandemic. 2020.

(NUS) NUoS. Coronavirus Student Survey phase III November 2020. 2020.

Hellemans K, Abizaid A, Gabrys R, McQuaid R, Patterson Z. For university students, COVID-19 stress creates perfect conditions for mental health crises. The Conversation. 2020. Available from: https://theconversation.com/for-university-students-covid-19-stress-creates-perfect-conditions-for-mental-health-crises-149127 .

England N. Children and Young People with an Eating Disorder Waiting Times: NHS England; 2021 [Available from: https://www.england.nhs.uk/statistics/statistical-work-areas/cyped-waiting-times/

King JA, Cabarkapa S, Leow FH, Ng CH. Addressing international student mental health during COVID-19: an imperative overdue. Australas Psychiatry. 2020;28(4):469.

Rana R, Smith E, Walking J. Degrees of disturbance: the new agenda; the Impact of Increasing Levels of Psychological Disturbance Amongst Students in Higher Education. England: Association for University and College Counselling Rugby; 1999.

AMOSSHE. Responding to student mental health issues: 'Duty of Care' responsibilities for student services in higher education. https://www.amosshe.org.uk/resources/Documents/AMOSSHE_Duty_of_Care_2001.pdf [accessed 24.12.2020]. (2001).

Universities UK. Student mental wellbeing in higher education. Good practice guide. London: Universities UK; 2015.

Williams M, Coare P, Marvell R, Pollard E, Houghton A-M, Anderson J. 2015. Understanding provision for students with mental health problems and intensive support needs: Report to HEFCE by the Institute for Employment Studies (IES) and Researching Equity, Access and Partnership (REAP). Institute for Employment Studies.

Hunt J, Eisenberg D. Mental health problems and help-seeking behavior among college students. J Adolesc Health. 2010;46(1):3–10.

Wells GA, Shea B, O’Connell D, Peterson J, Welch V, Losos M, Tugwell P. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. In.: Oxford; 2000.

Campbell M, McKenzie JE, Sowden A, Katikireddi SV, Brennan SE, Ellis S, et al. Synthesis without meta-analysis (SWiM) in systematic reviews: reporting guideline. BMJ. 2020;368:l6890.

El Ansari W, Adetunji H, Oskrochi R. Food and mental health: relationship between food and perceived stress and depressive symptoms among university students in the United Kingdom. Cent Eur J Public Health. 2014a;22(2):90–7.

El Ansari W, Dibba E, Stock C. Body image concerns: levels, correlates and gender differences among students in the United Kingdom. Cent Eur J Public Health. 2014b;22(2):106–17.

Ansari EL, W, Oskrochi R, Stock C. Symptoms and health complaints and their association with perceived stress: Students from seven universities in England, Wales and Northern Ireland. J Public Health. 2013;21(5):413–25.

El Ansari W, Sebena R, Stock C. Do importance of religious faith and healthy lifestyle modify the relationships between depressive symptoms and four indicators of alcohol consumption? A survey of students across seven universities in England, Wales, and Northern Ireland. Subst Use Misuse. 2014c;49(3):211–20.

El Ansari W, Stock C. Is the health and wellbeing of university students associated with their academic performance? Cross sectional findings from the United Kingdom. International Journal of Environmental Research & Public Health [Electronic Resource]. 2010;7(2):509–27.

Gnan GH, Rahman Q, Ussher G, Baker D, West E, Rimes KA. General and LGBTQ-specific factors associated with mental health and suicide risk among LGBTQ students. J Youth Stud. 2019;22(10):1393–408.

Jackson SL, Dritschel B. Modeling the impact of social problem-solving deficits on depressive vulnerability in the broader autism phenotype. Res Aut Spectr Disord. 2016;21:128–38.

Richardson T, Elliott P, Roberts R. The impact of tuition fees amount on mental health over time in British students. J Public Health. 2015;37(3):412–8.

Article   CAS   Google Scholar  

Richardson T, Mma Y, Jansen M, Elliott P, Roberts R. Financial difficulties and psychosis risk in British undergraduate students: a longitudinal analysis. J Public Ment Health. 2018;17(2):61–8.

Thomas L, Briggs P, Hart A, Kerrigan F. Understanding social media and identity work in young people transitioning to university. Comput Hum Behav. 2017;76:541–53.

Hixenbaugh P, Dewart H, Towell T. What enables students to succeed? An investigation of socio-demographic, health and student experience variables. Psychodyn Pract. 2012;18(3):285–301.

Holliman A, Martin A, Collie R. Adaptability, engagement, and degree completion: a longitudinal investigation of university students. Educ Psychol. 2018;38(6):785–99.

McLafferty M, Armour C, Bunting B, Ennis E, Lapsley C, Murray E, et al. Coping, stress, and negative childhood experiences: the link to psychopathology, self-harm, and suicidal behavior. Psychic J. 2019;8(3):293–306.

Nightingale S, Roberts S, Tariq V, Appleby Y, Barnes L, Harris R, et al. Trajectories of university adjustment in the United Kingdom: EMOTION management and emotional self-efficacy protect against initial poor adjustment. Learn Individ Differ. 2013;27:174–81.

Berry K, Kingswell S. An investigation of adult attachment and coping with exam-related stress. Br J Guid Couns. 2012;40(4):315.

Denovan A, Macaskill A. Stress and subjective well-being among first year UK undergraduate students. J Happiness Stud. 2017a;18(2):505–25.

Hassel S, Ridout N. An investigation of first-year students’ and lecturers’ expectations of university education. Front Psychol. 2018;8:2218.

Norbury R, Evans S. Time to think: subjective sleep quality, trait anxiety and university start time. Psychiatry Res. 2019;271:214–9.

Por J, Barriball L, Fitzpatrick J, Roberts J. Emotional intelligence: its relationship to stress, coping, well-being and professional performance in nursing students. Nurse Educ Today. 2011;31(8):855.

Honney K, Buszewicz M, Coppola W, Griffin M. Comparison of levels of depression in medical and non-medical students. Clin Teach. 2010;7(3):180–4.

Kotera Y, Conway E, Van Gordon W. Mental health of UK university business students: Relationship with shame, motivation and self-compassion. Journal of Education for Business. 2019;94(1):11–20.

Oliver EJ, Markland D, Hardy J. Interpretation of self-talk and post-lecture affective states of higher education students: a self-determination theory perspective. Br J Educ Psychol. 2010;80(Pt 2):307–23.

O’Neill S, McLafferty M, Ennis E, Lapsley C, Bjourson T, Armour C, et al. Socio-demographic, mental health and childhood adversity risk factors for self-harm and suicidal behaviour in College students in Northern Ireland. J Affect Disord. 2018;239:58–65.

Mahadevan S, Hawton K, Casey D. Deliberate self-harm in Oxford University students, 1993–2005: a descriptive and case-control study. Soc Psychiatry Psychiatr Epidemiol. 2010;45(2):211–9.

Boulton CA, Hughes E, Kent C, Smith JR, Williams HTP. Student engagement and wellbeing over time at a higher education institution. PLoS One [Electronic Resource]. 2019;14(11): e0225770.

Davies EL, Paltoglou AE. Public self-consciousness, pre-loading and drinking harms among university students. Subst Use Misuse. 2019;54(5):747–57.

Denovan A, Macaskill A. Stress, resilience and leisure coping among university students: Applying the broaden-and-build theory. Leisure Studies. 2017b;36(6):852–65.

El Ansari W, Vallentin-Holbech L, Stock C. Predictors of illicit drug/s use among university students in Northern Ireland, Wales and England. Glob J Health Sci. 2015;7(4):18–29.

Freeth M, Bullock T, Milne E. The distribution of and relationship between autistic traits and social anxiety in a UK student population. Autism. 2013;17(5):571–81.

Jessop DC, Reid M, Solomon L. Financial concern predicts deteriorations in mental and physical health among university students. Psychology Health. 2020;35(2):196–209.

Kannangara CS, Allen RE, Waugh G, Nahar N, Khan SZN, Rogerson S, Carson J. All that glitters is not grit: Three studies of grit in university students. Front Psychol. 2018;9:1539.

Lloyd J, Ward T, Young J. Do parental interpersonal power and prestige moderate the relationship between parental acceptance and psychological adjustment in U.K. Students? Cross-Cultural Research. The Journal of Comparative Social Science. 2014;48(3):326–35.

McIntyre JC, Worsley J, Corcoran R, Harrison Woods P, Bentall RP. Academic and non-academic predictors of student psychological distress: the role of social identity and loneliness. J Ment Health. 2018;27(3):230–9.

Ribchester C, Ross K, Rees EL. Examining the impact of pre-induction social networking on the student transition into higher education. Innov Educ Teach Int. 2014;51(4):355–65.

Richardson T, Elliott P, Roberts R. Relationship between loneliness and mental health in students. J Public Ment Health. 2017a;16(2):48–54.

Richardson T, Elliott P, Roberts R, Jansen M. A Longitudinal Study of Financial Difficulties and Mental Health in a National Sample of British Undergraduate Students. Community Ment Health J. 2017;53(3):344–52.

Taylor PJ, Dhingra K, Dickson JM, McDermott E. Psychological Correlates of Self-Harm within Gay, Lesbian and Bisexual UK University Students. Arch Suicide Res. 2020;24(sup1):41–56.

Thomas L, Orme E, Kerrigan F. Student loneliness: The role of social media through life transitions. Comput Educ. 2020;146:103754.

Tyson P, Wilson K, Crone D, Brailsford R, Laws K. Physical activity and mental health in a student population. J Ment Health. 2010;19(6):492–9.

Folkman S. The Oxford handbook of stress, health, and coping. Oxford: Oxford University Press; 2011.

Gorczynski P, Sims-schouten W, Hill D, Wilson JC. Examining mental health literacy, help seeking behaviours, and mental health outcomes in UK university students. J Ment Health Train Educ Pract. 2017;12(2):111–20.

Felitti VJ. Adverse childhood experiences and adult health. Acad Pediatr. 2009;9(3):131–2.

Denovan A, Macaskill A. An interpretative phenomenological analysis of stress and coping in first year undergraduates. Br Educ Res J. 2013;39(6):1002–24.

Bandura A. Self-efficacy: The foundation of agency. Control of human behavior, mental processes, and consciousness: Essays in honor of the 60th birthday of August Flammer. 2000;16.

Martin AJ, Nejad H, Colmar S, Liem GAD. Adaptability: Conceptual and empirical perspectives on responses to change, novelty and uncertainty. J Psychol Couns Sch. 2012;22(1):58–81.

Lazarus RS, Folkman S. Stress, appraisal, and coping: Springer publishing company; 1984.

Gross JJ. Emotion regulation: Past, present, future. Cogn Emot. 1999;13(5):551–73.

Mayer JD, Salovey P, Caruso DR. TARGET ARTICLES:" Emotional Intelligence: Theory, Findings, and Implications". Psychol Inq. 2004;15(3):197–215.

Duckworth AL, Peterson C, Matthews MD, Kelly DR. Grit: perseverance and passion for long-term goals. J Pers Soc Psychol. 2007;92(6):1087.

Snyder CR, Ilardi SS, Cheavens J, Michael ST, Yamhure L, Sympson S. The role of hope in cognitive-behavior therapies. Cognit Ther Res. 2000;24(6):747–62.

Scheier MF, Carver CS, Bridges MW. Optimism, pessimism, and psychological well-being. 2001.

Seligman ME. Positive psychology in practice: Wiley; 2012.

Masten AS. Ordinary magic: Lessons from research on resilience in human development. Education Canada. 2009;49(3):28–32.

Rosenberg M, Schooler C, Schoenbach C, Rosenberg F. Global self-esteem and specific self-esteem: Different concepts, different outcomes. Am Sociol Rev. 1995:141–56.

Oliver EJ, Markland D, Hardy J. Interpretation of self-talk and post-lecture affective states of higher education students: A self-determination theory perspective. Br J Educ Psychol. 2010;80(2):307–23.

Hofmann W, Friese M, Strack F. Impulse and self-control from a dual-systems perspective. Perspect Psychol Sci. 2009;4(2):162–76.

Aceijas C, Waldhausl S, Lambert N, Cassar S, Bello-Corassa R. Determinants of health-related lifestyles among university students. Perspect Public Health. 2017;137(4):227–36.

Fredrickson BL. The broaden–and–build theory of positive emotions. Philos Trans R Soc Lond B Biol Sci. 2004;359(1449):1367–77.

Tugade MM, Fredrickson BL, Feldman Barrett L. Psychological resilience and positive emotional granularity: Examining the benefits of positive emotions on coping and health. J Pers. 2004;72(6):1161–90.

Harandi TF, Taghinasab MM, Nayeri TD. The correlation of social support with mental health: A meta-analysis. Electron physician. 2017;9(9):5212.

Sheldon E, Simmonds-Buckley M, Bone C, Mascarenhas T, Chan N, Wincott M, Gleeson H, Sow K, Hind D, Barkham M. Prevalence and risk factors for mental health problems in university undergraduate students: A systematic review with meta-analysis. J Affect Disord. 2021;287:282–92.

Download references

Acknowledgements

We acknowledge the input from our public advisory group which included current and former students, and family members of students who have struggled with their mental health. The group gave us their extremely valuable insights to assist our understanding of the evidence.

This project was supported by funding from the National Institute for Health Research as part of the NIHR Public Health Research  Programme (fuding reference 127659 Public Health Review Team). The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.

Author information

Authors and affiliations.

University of Sheffield, Sheffield, UK

Fiona Campbell, Lindsay Blank, Anna Cantrell, Susan Baxter, Christopher Blackmore, Jan Dixon & Elizabeth Goyder

You can also search for this author in PubMed   Google Scholar

Contributions

All of the included authors designed the project methods and prepared a protocol. A.C. designed the search strategy. F.C, L.B and C.B screened the identified citations and undertook data extraction. S.B. led the PPI involvement. JD participated as a member of the PPI group. F.C and L.B undertook the analysis. F.C. and L.B wrote the main manuscript text. All authors reviewed the manuscript. F.C designed Figs. 2 , 3 and 4 . The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Fiona Campbell .

Ethics declarations

Ethics approval and consent to participate.

Not applicable as this was secondary research.

Consent for publication

Not applicable.

Competing interests

None of the authors have competing interests or other interests that might be perceived to influence the results and/or discussion reported in this paper.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1., additional file 2., rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Campbell, F., Blank, L., Cantrell, A. et al. Factors that influence mental health of university and college students in the UK: a systematic review. BMC Public Health 22 , 1778 (2022). https://doi.org/10.1186/s12889-022-13943-x

Download citation

Received : 03 February 2022

Accepted : 25 July 2022

Published : 20 September 2022

DOI : https://doi.org/10.1186/s12889-022-13943-x

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Student mental health
• Mental wellbeing
• Risk factors
• Rapid review

BMC Public Health

ISSN: 1471-2458

• General enquiries: [email protected]

Student mental health is in crisis. Campuses are rethinking their approach

Amid massive increases in demand for care, psychologists are helping colleges and universities embrace a broader culture of well-being and better equipping faculty to support students in need

Vol. 53 No. 7 Print version: page 60

• Mental Health

By nearly every metric, student mental health is worsening. During the 2020–2021 school year, more than 60% of college students met the criteria for at least one mental health problem, according to the Healthy Minds Study, which collects data from 373 campuses nationwide ( Lipson, S. K., et al., Journal of Affective Disorders , Vol. 306, 2022 ). In another national survey, almost three quarters of students reported moderate or severe psychological distress ( National College Health Assessment , American College Health Association, 2021).

Even before the pandemic, schools were facing a surge in demand for care that far outpaced capacity, and it has become increasingly clear that the traditional counseling center model is ill-equipped to solve the problem.

“Counseling centers have seen extraordinary increases in demand over the past decade,” said Michael Gerard Mason, PhD, associate dean of African American Affairs at the University of Virginia (UVA) and a longtime college counselor. “[At UVA], our counseling staff has almost tripled in size, but even if we continue hiring, I don’t think we could ever staff our way out of this challenge.”

Some of the reasons for that increase are positive. Compared with past generations, more students on campus today have accessed mental health treatment before college, suggesting that higher education is now an option for a larger segment of society, said Micky Sharma, PsyD, who directs student life’s counseling and consultation service at The Ohio State University (OSU). Stigma around mental health issues also continues to drop, leading more people to seek help instead of suffering in silence.

But college students today are also juggling a dizzying array of challenges, from coursework, relationships, and adjustment to campus life to economic strain, social injustice, mass violence, and various forms of loss related to Covid -19.

As a result, school leaders are starting to think outside the box about how to help. Institutions across the country are embracing approaches such as group therapy, peer counseling, and telehealth. They’re also better equipping faculty and staff to spot—and support—students in distress, and rethinking how to respond when a crisis occurs. And many schools are finding ways to incorporate a broader culture of wellness into their policies, systems, and day-to-day campus life.

“This increase in demand has challenged institutions to think holistically and take a multifaceted approach to supporting students,” said Kevin Shollenberger, the vice provost for student health and well-being at Johns Hopkins University. “It really has to be everyone’s responsibility at the university to create a culture of well-being.”

Higher caseloads, creative solutions

The number of students seeking help at campus counseling centers increased almost 40% between 2009 and 2015 and continued to rise until the pandemic began, according to data from Penn State University’s Center for Collegiate Mental Health (CCMH), a research-practice network of more than 700 college and university counseling centers ( CCMH Annual Report , 2015 ).

That rising demand hasn’t been matched by a corresponding rise in funding, which has led to higher caseloads. Nationwide, the average annual caseload for a typical full-time college counselor is about 120 students, with some centers averaging more than 300 students per counselor ( CCMH Annual Report , 2021 ).

“We find that high-caseload centers tend to provide less care to students experiencing a wide range of problems, including those with safety concerns and critical issues—such as suicidality and trauma—that are often prioritized by institutions,” said psychologist Brett Scofield, PhD, executive director of CCMH.

To minimize students slipping through the cracks, schools are dedicating more resources to rapid access and assessment, where students can walk in for a same-day intake or single counseling session, rather than languishing on a waitlist for weeks or months. Following an evaluation, many schools employ a stepped-care model, where the students who are most in need receive the most intensive care.

Given the wide range of concerns students are facing, experts say this approach makes more sense than offering traditional therapy to everyone.

“Early on, it was just about more, more, more clinicians,” said counseling psychologist Carla McCowan, PhD, director of the counseling center at the University of Illinois at Urbana-Champaign. “In the past few years, more centers are thinking creatively about how to meet the demand. Not every student needs individual therapy, but many need opportunities to increase their resilience, build new skills, and connect with one another.”

Students who are struggling with academic demands, for instance, may benefit from workshops on stress, sleep, time management, and goal-setting. Those who are mourning the loss of a typical college experience because of the pandemic—or facing adjustment issues such as loneliness, low self-esteem, or interpersonal conflict—are good candidates for peer counseling. Meanwhile, students with more acute concerns, including disordered eating, trauma following a sexual assault, or depression, can still access one-on-one sessions with professional counselors.

As they move away from a sole reliance on individual therapy, schools are also working to shift the narrative about what mental health care on campus looks like. Scofield said it’s crucial to manage expectations among students and their families, ideally shortly after (or even before) enrollment. For example, most counseling centers won’t be able to offer unlimited weekly sessions throughout a student’s college career—and those who require that level of support will likely be better served with a referral to a community provider.

“We really want to encourage institutions to be transparent about the services they can realistically provide based on the current staffing levels at a counseling center,” Scofield said.

The first line of defense

Faculty may be hired to teach, but schools are also starting to rely on them as “first responders” who can help identify students in distress, said psychologist Hideko Sera, PsyD, director of the Office of Equity, Inclusion, and Belonging at Morehouse College, a historically Black men’s college in Atlanta. During the pandemic, that trend accelerated.

“Throughout the remote learning phase of the pandemic, faculty really became students’ main points of contact with the university,” said Bridgette Hard, PhD, an associate professor and director of undergraduate studies in psychology and neuroscience at Duke University. “It became more important than ever for faculty to be able to detect when a student might be struggling.”

Many felt ill-equipped to do so, though, with some wondering if it was even in their scope of practice to approach students about their mental health without specialized training, Mason said.

Schools are using several approaches to clarify expectations of faculty and give them tools to help. About 900 faculty and staff at the University of North Carolina have received training in Mental Health First Aid , which provides basic skills for supporting people with mental health and substance use issues. Other institutions are offering workshops and materials that teach faculty to “recognize, respond, and refer,” including Penn State’s Red Folder campaign .

Faculty are taught that a sudden change in behavior—including a drop in attendance, failure to submit assignments, or a disheveled appearance—may indicate that a student is struggling. Staff across campus, including athletic coaches and academic advisers, can also monitor students for signs of distress. (At Penn State, eating disorder referrals can even come from staff working in food service, said counseling psychologist Natalie Hernandez DePalma, PhD, senior director of the school’s counseling and psychological services.) Responding can be as simple as reaching out and asking if everything is going OK.

Referral options vary but may include directing a student to a wellness seminar or calling the counseling center to make an appointment, which can help students access services that they may be less likely to seek on their own, Hernandez DePalma said. Many schools also offer reporting systems, such as DukeReach at Duke University , that allow anyone on campus to express concern about a student if they are unsure how to respond. Trained care providers can then follow up with a welfare check or offer other forms of support.

“Faculty aren’t expected to be counselors, just to show a sense of care that they notice something might be going on, and to know where to refer students,” Shollenberger said.

At Johns Hopkins, he and his team have also worked with faculty on ways to discuss difficult world events during class after hearing from students that it felt jarring when major incidents such as George Floyd’s murder or the war in Ukraine went unacknowledged during class.

Many schools also support faculty by embedding counselors within academic units, where they are more visible to students and can develop cultural expertise (the needs of students studying engineering may differ somewhat from those in fine arts, for instance).

When it comes to course policy, even small changes can make a big difference for students, said Diana Brecher, PhD, a clinical psychologist and scholar-in-residence for positive psychology at Toronto Metropolitan University (TMU), formerly Ryerson University. For example, instructors might allow students a 7-day window to submit assignments, giving them agency to coordinate with other coursework and obligations. Setting deadlines in the late afternoon or early evening, as opposed to at midnight, can also help promote student wellness.

At Moraine Valley Community College (MVCC) near Chicago, Shelita Shaw, an assistant professor of communications, devised new class policies and assignments when she noticed students struggling with mental health and motivation. Those included mental health days, mindful journaling, and a trip with family and friends to a Chicago landmark, such as Millennium Park or Navy Pier—where many MVCC students had never been.

Faculty in the psychology department may have a unique opportunity to leverage insights from their own discipline to improve student well-being. Hard, who teaches introductory psychology at Duke, weaves in messages about how students can apply research insights on emotion regulation, learning and memory, and a positive “stress mindset” to their lives ( Crum, A. J., et al., Anxiety, Stress, & Coping , Vol. 30, No. 4, 2017 ).

Along with her colleague Deena Kara Shaffer, PhD, Brecher cocreated TMU’s Thriving in Action curriculum, which is delivered through a 10-week in-person workshop series and via a for-credit elective course. The material is also freely available for students to explore online . The for-credit course includes lectures on gratitude, attention, healthy habits, and other topics informed by psychological research that are intended to set students up for success in studying, relationships, and campus life.

“We try to embed a healthy approach to studying in the way we teach the class,” Brecher said. “For example, we shift activities every 20 minutes or so to help students sustain attention and stamina throughout the lesson.”

Creative approaches to support

Given the crucial role of social connection in maintaining and restoring mental health, many schools have invested in group therapy. Groups can help students work through challenges such as social anxiety, eating disorders, sexual assault, racial trauma, grief and loss, chronic illness, and more—with the support of professional counselors and peers. Some cater to specific populations, including those who tend to engage less with traditional counseling services. At Florida Gulf Coast University (FGCU), for example, the “Bold Eagles” support group welcomes men who are exploring their emotions and gender roles.

The widespread popularity of group therapy highlights the decrease in stigma around mental health services on college campuses, said Jon Brunner, PhD, the senior director of counseling and wellness services at FGCU. At smaller schools, creating peer support groups that feel anonymous may be more challenging, but providing clear guidelines about group participation, including confidentiality, can help put students at ease, Brunner said.

Less formal groups, sometimes called “counselor chats,” meet in public spaces around campus and can be especially helpful for reaching underserved groups—such as international students, first-generation college students, and students of color—who may be less likely to seek services at a counseling center. At Johns Hopkins, a thriving international student support group holds weekly meetings in a café next to the library. Counselors typically facilitate such meetings, often through partnerships with campus centers or groups that support specific populations, such as LGBTQ students or student athletes.

“It’s important for students to see counselors out and about, engaging with the campus community,” McCowan said. “Otherwise, you’re only seeing the students who are comfortable coming in the door.”

Peer counseling is another means of leveraging social connectedness to help students stay well. At UVA, Mason and his colleagues found that about 75% of students reached out to a peer first when they were in distress, while only about 11% contacted faculty, staff, or administrators.

“What we started to understand was that in many ways, the people who had the least capacity to provide a professional level of help were the ones most likely to provide it,” he said.

Project Rise , a peer counseling service created by and for Black students at UVA, was one antidote to this. Mason also helped launch a two-part course, “Hoos Helping Hoos,” (a nod to UVA’s unofficial nickname, the Wahoos) to train students across the university on empathy, mentoring, and active listening skills.

At Washington University in St. Louis, Uncle Joe’s Peer Counseling and Resource Center offers confidential one-on-one sessions, in person and over the phone, to help fellow students manage anxiety, depression, academic stress, and other campus-life issues. Their peer counselors each receive more than 100 hours of training, including everything from basic counseling skills to handling suicidality.

Uncle Joe’s codirectors, Colleen Avila and Ruchika Kamojjala, say the service is popular because it’s run by students and doesn’t require a long-term investment the way traditional psychotherapy does.

“We can form a connection, but it doesn’t have to feel like a commitment,” said Avila, a senior studying studio art and philosophy-neuroscience-psychology. “It’s completely anonymous, one time per issue, and it’s there whenever you feel like you need it.”

As part of the shift toward rapid access, many schools also offer “Let’s Talk” programs , which allow students to drop in for an informal one-on-one session with a counselor. Some also contract with telehealth platforms, such as WellTrack and SilverCloud, to ensure that services are available whenever students need them. A range of additional resources—including sleep seminars, stress management workshops, wellness coaching, and free subscriptions to Calm, Headspace, and other apps—are also becoming increasingly available to students.

Those approaches can address many student concerns, but institutions also need to be prepared to aid students during a mental health crisis, and some are rethinking how best to do so. Penn State offers a crisis line, available anytime, staffed with counselors ready to talk or deploy on an active rescue. Johns Hopkins is piloting a behavioral health crisis support program, similar to one used by the New York City Police Department, that dispatches trained crisis clinicians alongside public safety officers to conduct wellness checks.

A culture of wellness

With mental health resources no longer confined to the counseling center, schools need a way to connect students to a range of available services. At OSU, Sharma was part of a group of students, staff, and administrators who visited Apple Park in Cupertino, California, to develop the Ohio State: Wellness App .

Students can use the app to create their own “wellness plan” and access timely content, such as advice for managing stress during final exams. They can also connect with friends to share articles and set goals—for instance, challenging a friend to attend two yoga classes every week for a month. OSU’s apps had more than 240,000 users last year.

At Johns Hopkins, administrators are exploring how to adapt school policies and procedures to better support student wellness, Shollenberger said. For example, they adapted their leave policy—including how refunds, grades, and health insurance are handled—so that students can take time off with fewer barriers. The university also launched an educational campaign this fall to help international students navigate student health insurance plans after noticing below average use by that group.

Students are a key part of the effort to improve mental health care, including at the systemic level. At Morehouse College, Sera serves as the adviser for Chill , a student-led advocacy and allyship organization that includes members from Spelman College and Clark Atlanta University, two other HBCUs in the area. The group, which received training on federal advocacy from APA’s Advocacy Office earlier this year, aims to lobby public officials—including U.S. Senator Raphael Warnock, a Morehouse College alumnus—to increase mental health resources for students of color.

“This work is very aligned with the spirit of HBCUs, which are often the ones raising voices at the national level to advocate for the betterment of Black and Brown communities,” Sera said.

Despite the creative approaches that students, faculty, staff, and administrators are employing, students continue to struggle, and most of those doing this work agree that more support is still urgently needed.

“The work we do is important, but it can also be exhausting,” said Kamojjala, of Uncle Joe’s peer counseling, which operates on a volunteer basis. “Students just need more support, and this work won’t be sustainable in the long run if that doesn’t arrive.”

Further reading

Overwhelmed: The real campus mental-health crisis and new models for well-being The Chronicle of Higher Education, 2022

Mental health in college populations: A multidisciplinary review of what works, evidence gaps, and paths forward Abelson, S., et al., Higher Education: Handbook of Theory and Research, 2022

Student mental health status report: Struggles, stressors, supports Ezarik, M., Inside Higher Ed, 2022

Before heading to college, make a mental health checklist Caron, C., The New York Times, 2022

Related topics

• Mental health
• Stress effects on the body

Contact APA

You may also like.

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

Psychological impacts from COVID-19 among university students: Risk factors across seven states in the United States

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliation Virtual Reality & Nature Lab, Department of Parks, Recreation and Tourism Management, Clemson University, Clemson, SC, United States of America

Roles Conceptualization, Data curation, Investigation, Methodology, Visualization, Writing – review & editing

Affiliation Department of Parks, Recreation and Tourism Management, North Carolina State University, Raleigh, NC, United States of America

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Software, Writing – review & editing

Affiliation Department of Parks, Recreation and Tourism Management, Clemson University, Clemson, SC, United States of America

Roles Conceptualization, Data curation, Methodology, Writing – review & editing

Affiliation Department of City and Metropolitan Planning, The University of Utah, Salt Lake City, UT, United States of America

Roles Data curation, Writing – review & editing

Affiliation Sustainability and Happiness Research Lab, School of Sustainability, Arizona State University, Tempe, AZ, United States of America

Roles Formal analysis, Methodology, Visualization, Writing – review & editing

Affiliation Advanced Computing & Data Science, Clemson Computing & Information Technology, Clemson University, Clemson, SC, United States of America

Affiliation Department of Society and Conservation, W.A. Franke College of Forestry and Conservation, University of Montana, Missoula, MT, United States of America

Roles Data curation, Methodology, Visualization, Writing – review & editing

Affiliation Recreation, Park, and Tourism Management Department, College of Health and Human Development, Pennsylvania State University, PA, United States of America

Affiliation Forest Ecosystems and Society, College of Forestry, Oregon State University, Corvallis, OR, United States of America

Roles Methodology, Software, Writing – review & editing

Affiliation Department of Human Geography and Spatial Planning, Faculty of Geosciences, Utrecht University, Utrecht, The Netherlands

Roles Methodology, Writing – review & editing

Affiliation Environment & Well-Being Lab, School of Environmental and Forest Sciences, University of Washington, Seattle, WA, United States of America

Affiliation School of Nursing, Oregon Health & Science University, Portland, OR, United States of America

• Matthew H. E. M. Browning, 
• Lincoln R. Larson, 
• Iryna Sharaievska, 
• Alessandro Rigolon, 
• Olivia McAnirlin, 
• Lauren Mullenbach, 
• Scott Cloutier, 
• Tue M. Vu, 
• Jennifer Thomsen, 

• Published: January 7, 2021
• https://doi.org/10.1371/journal.pone.0245327
• Peer Review
• Reader Comments

26 Aug 2022: Browning MHEM, Larson LR, Sharaievska I, Rigolon A, McAnirlin O, et al. (2022) Correction: Psychological impacts from COVID-19 among university students: Risk factors across seven states in the United States. PLOS ONE 17(8): e0273938. https://doi.org/10.1371/journal.pone.0273938 View correction

University students are increasingly recognized as a vulnerable population, suffering from higher levels of anxiety, depression, substance abuse, and disordered eating compared to the general population. Therefore, when the nature of their educational experience radically changes—such as sheltering in place during the COVID-19 pandemic—the burden on the mental health of this vulnerable population is amplified. The objectives of this study are to 1) identify the array of psychological impacts COVID-19 has on students, 2) develop profiles to characterize students' anticipated levels of psychological impact during the pandemic, and 3) evaluate potential sociodemographic, lifestyle-related, and awareness of people infected with COVID-19 risk factors that could make students more likely to experience these impacts.

Cross-sectional data were collected through web-based questionnaires from seven U.S. universities. Representative and convenience sampling was used to invite students to complete the questionnaires in mid-March to early-May 2020, when most coronavirus-related sheltering in place orders were in effect. We received 2,534 completed responses, of which 61% were from women, 79% from non-Hispanic Whites, and 20% from graduate students.

Exploratory factor analysis on close-ended responses resulted in two latent constructs, which we used to identify profiles of students with latent profile analysis, including high (45% of sample), moderate (40%), and low (14%) levels of psychological impact. Bivariate associations showed students who were women, were non-Hispanic Asian, in fair/poor health, of below-average relative family income, or who knew someone infected with COVID-19 experienced higher levels of psychological impact. Students who were non-Hispanic White, above-average social class, spent at least two hours outside, or less than eight hours on electronic screens were likely to experience lower levels of psychological impact. Multivariate modeling (mixed-effects logistic regression) showed that being a woman, having fair/poor general health status, being 18 to 24 years old, spending 8 or more hours on screens daily, and knowing someone infected predicted higher levels of psychological impact when risk factors were considered simultaneously.

Inadequate efforts to recognize and address college students’ mental health challenges, especially during a pandemic, could have long-term consequences on their health and education.

Citation: Browning MHEM, Larson LR, Sharaievska I, Rigolon A, McAnirlin O, Mullenbach L, et al. (2021) Psychological impacts from COVID-19 among university students: Risk factors across seven states in the United States. PLoS ONE 16(1): e0245327. https://doi.org/10.1371/journal.pone.0245327

Editor: Chung-Ying Lin, Hong Kong Polytechnic University, HONG KONG

Received: August 4, 2020; Accepted: December 28, 2020; Published: January 7, 2021

Copyright: © 2021 Browning et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the manuscript and its Supporting Information files.

Funding: The author(s) received no specific funding for this work.

Competing interests: The authors have declared that no competing interests exist.

1 Introduction

A large number of studies support that the conclusion that the novel coronavirus (SARS-CoV-2) and its corresponding disease (COVID-19) have dramatically impacted people's mental health and behavior [ 1 – 5 ], with very few studies suggesting otherwise [ 6 ]. Mental health hotlines in the United States experienced 1,000% increases during the month of April, when most people were under lockdown because of the pandemic [ 7 ]. Some medical facilities have seen more deaths from suicide, presumably because of exceedingly poor mental health, than from COVID-19 infections [ 8 ]. Substance disorders in many people who were previously abstinent are expected to relapse during COVID-19, which will cause long-term economic and health impacts [ 9 ].

Although impacts are felt across populations—and especially in socially-disadvantaged communities and individuals employed as essential workers—college students are among the most strongly affected by COVID-19 because of uncertainty regarding academic success, future careers, and social life during college, amongst other concerns [ 10 ]. Even before the pandemic, students across the globe experienced increasing levels of anxiety, depressive moods, lack of self-esteem, psychosomatic problems, substance abuse, and suicidality [ 11 ]. Therefore, students may need additional resources and services to deal with the physical and mental health repercussions of the disease.

University administrators could best serve students if they better understood the impacts of COVID-19 and the risk factors of its psychological impacts. These impacts are of critical importance to warrant immediate mental health interventions focused on prevention and treatment [ 12 ]. Psychiatric and counseling services have historically been underutilized by college students [ 13 , 14 ]. Understanding what subpopulations may suffer from unique combinations of psychological impacts may facilitate targeted interventions and successful treatment and coping strategies for individuals at greatest risk.

A recent review highlights some of the documented psychological impacts of COVID-19 on college students [ 15 ]. Many feel increased stress levels and anxiety and depressive symptoms as a result of changed delivery and uncertainty of university education, technological concerns of online courses, being far from home, social isolation, decreased family income, and future employment. These impacts have been observed in universities across the world [ 10 ].

Studies of the psychological impacts of COVID-19 on college studies in the United States, however, have been limited in their generalizability [ 10 ] due to examination of single institutions only [ 10 , 16 , 17 ]. We are aware of no studies that have been conducted with college students at multiple institutions across the United States during the pandemic. These schools collectively represent a somewhat unique context within higher education. The United States educates large numbers of students from around the world [ 18 , 19 ]. Diverse student bodies may show different risk factors from more culturally-homogenous student bodies because of the diversity of value orientations [ 20 ] and sources of media consumption [ 16 , 21 – 23 ]. Further, colleges in the United States cost more than higher education institutions nearly anywhere else in the world [ 24 ]; therefore, financial concerns may be particularly apparent in the United States. The United States also experienced the lowest global recovery rate from infection–in other words, the highest mortality rate post-infection–in the weeks leading up to the timing of the current study (April and May, 2020) [ 25 ]. This country continues to witness the highest incidence and mortality rates among Global North countries [ 26 ]. Such high rates aggravate the psychological impacts of the disease on infected and non-infected individuals [ 1 ].

In the current study, we investigate the psychological impacts of COVID-19 and associated risk factors on college students at seven universities across the United States. Our objectives are three-fold: 1) identify the array of psychological impacts COVID-19 has on students, 2) develop profiles to characterize students' anticipated levels of psychological impact during the pandemic, and 3) evaluate potential sociodemographic, lifestyle-related, and awareness of people infected with COVID-19 risk factors that could make students more likely to experience these impacts.

2.1 Study population

In spring 2020, 14,174 participants were recruited cross-sectionally from representative and targeted samples at seven large, state universities, which in sum enroll more than 238,000 students. Universities included Arizona State University in Tempe, AZ (approximately 52,000 undergraduate/graduate students enrolled in 2019); Clemson University in Clemson, SC (approx. 25,000); North Carolina State University in Raleigh, NC (approx. 34,000); Oregon State University in Corvallis, OR (approx. 29,000); Pennsylvania State University in State College, PA (approx. 54,000); University of Montana in Missoula, MT (approx. 11,000); and The University of Utah in Salt Lake City, UT (approx. 33,000). One institution (North Carolina State University) was able to utilize a university-wide representative sample. Other institutions used targeted samples in the home college(s) or department(s) of the corresponding author. Selection of sampling scheme (i.e., representative or targeted) was determined by human subject review board allowances and listserv availability. (Recruitment occurred over email listservs and course website announcements.)

This research was deemed exempt from the Clemson University Institutional Review Board. Also, all subjects provided written consent when they completed the online survey.

Recruitment started as soon as human subject approval was awarded and occurred over a two-to-three-week window at each institution. Because approval took longer at some institutions, nationwide recruitment was staggered. No compensation for participation was provided. Sampling frames and recruitment windows are detailed in Table 1 .

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

https://doi.org/10.1371/journal.pone.0245327.t001

Of the 14,174 students invited to participate in the survey, we received 2,534 responses with data on most of the relevant variables; thus, this sample size was available for most of the descriptive statistics and bivariate associations. Missing/not reported data on race/ethnicity and gender occurred in approximately 11% of respondents. Therefore, complete data for multivariate analyses with all risk factors entered simultaneously—including race/ethnicity and gender—were available for 2,140 students. Table 2 provides the sample characteristics.

https://doi.org/10.1371/journal.pone.0245327.t002

2.2 Measures

2.2.1 psychological impacts..

2 . 2 . 1 . 1 Qualitative assessment . We expected that it would be difficult to parsimoniously and comprehensively capture the broad array of impacts from COVID-19 on students with quantitative measures. Therefore, we utilized an open-ended questionnaire item that asked respondents, "We are interested in the ways that the coronavirus (COVID-19) pandemic has changed how you feel and behave. What are the first three ways that come to mind?" Three responses were required, and a fourth response was optional. This question was placed at the beginning of the questionnaire to avoid priming and order effects [ 27 , 28 ].

2 . 2 . 1 . 2 Quantitative assessment . Regarding our selection of quantitative impacts to measure, we chose nine survey items based on information gathered from a review of previous research and new interview data. These nine survey items measured the following concepts: negative emotion states, preoccupation with COVID-19, feeling stressed, worry, and time demands.

Regarding the review of previous research, we examined studies of other large-scale disasters (i.e., the World Trade Center terrorist attacks on September 11, 2001; previous epidemics requiring quarantine), which are almost always associated with psychological impacts on the general population [ 29 ]. These studies provided some guidance on what impacts to measure for the impacts of COVID-19 on college students.

Regarding new interview data, the corresponding author of the current study conducted unstructured interviews with adults on their experiences in the early stages of the COVID-19 pandemic. These interviews consisted of recruiting ten participants aged 18 years or older in February 2020. Recruitment occurred in both low-risk and high-risk regions of the United States, including urban areas in Washington and rural areas in Tennessee, Iowa, and South Carolina. The interviews captured the feelings that interviewees experienced during the pandemic.

Negative emotion states comprised four of the survey items. Each item explained one of the basic negative emotions (i.e., being afraid, irritable, guilty, and sad) identified during the development of the positive and negative affect schedule (PANAS) [ 30 ]. Items were measured using the visual analogue scale (VAS) to provide data across a wide range of responses (1–100) with minimal participant burden [ 31 ]. Prompts asked respondents to indicate the extent to which they felt these things when they thought about the pandemic.

Preoccupation and feeling stressed comprised two more survey items. These were also measured with the VAS. Prompts once again asked respondents to indicate the extent to which they felt these things when they thought about the pandemic.

One more survey item measured worry—specifically anxious arousal. It was measured with a single item ("I worry a lot") from the Penn State Worry Questionnaire (PSWQ) that is strongly associated with the entire 16-item PSWQ, r = 0.80 [ 32 ]. Therefore, this single item succinctly captures the concept of worry/anxious arousal. A 5-point Likert-type agree-disagree response scale was used.

Two more survey items measured time demands. These were developed from survey prompts in the eating disorder literature [ 33 ]. Specifically, we asked to what extent respondents believed they spent a lot of time/thought on the pandemic, and to what extent they believed they spent too much time/thought on the pandemic. Once again, a 5-point Likert-type agree-disagree response scale was used.

The prompts for all nine of these survey items were delivered as reactions to the coronavirus rather than measures of general psychological states. Example include: "how stressed do you feel when you think about coronavirus," and "to what extent do you agree/disagree with the following: I worry about coronavirus all of the time."

2.2.2 Risk factors.

Sociodemographic factors were self-reported and allowed identification of potential differences in impact levels by gender, age, race/ethnicity, socioeconomic status (SES), and academic status (undergraduate vs. graduate-seeking). SES was measured with perceived social class, which has been shown to accurately represent SES in student populations, using a battery of seven questions on class, parental education, and relative family income [ 34 , 35 ]. To measure academic status, we asked respondents whether they were in pursuit of an undergraduate or graduate degree.

To account for possible lifestyle-related risk factors, we first considered general health factors such as general health status and body mass index (BMI). Health status was measured with a single item on respondents' "health in general" and a 5-point response scale (poor to excellent) [ 36 ]. BMI was calculated from self-reported height and weight. BMI has been implicated as a risk factor or confounder of the psychological impacts of COVID-19 [ 37 , 38 ].

Another set of plausible lifestyle-related risk factors was time use. We utilized a recent recall question structure from the American Time Use Survey that strongly predicts objective time use and activity measures [ 39 ]. Three items were used to ask respondents to indicate how many hours they spent outdoors (at a park, on a greenway/trail, in a neighborhood/yard, etc.), in front of a screen (on a smartphone/computer, watching television, online gaming, etc.), and engaged in moderate or vigorous physical activity that caused an increase in breathing or heart rate (fast walking, running, etc.) in the past 24 hours [ 40 , 41 ].

Regarding awareness of COVID-19 victims as a potential risk factor, we included two measures of knowing people who were diagnosed with the virus: someone in their family and someone in their community [ 42 ].

2.3 Analyses

To accomplish Objective 1, qualitative data from the open-ended responses were analyzed using content analysis with an inductive approach [ 43 , 44 ]. Two independent researchers examined the data systematically to identify patterns and codes [ 43 ]. Each response was coded separately and reviewed for agreement [ 45 , 46 ]. Interrater/intercoder agreement (kappa) score was 94.94% [ 47 ].

Objective 2 was accomplished in three steps. These included data imputations, data reduction, and profile identification.

We imputed missing values by bag imputation, which fits a machine learning regression tree model for each predictor as a function of all others [ 48 ]. In our dataset, 5.2% of the quantitative data were missing and imputed.

Next, we reduced the survey items related to levels of psychological impact into latent constructs using exploratory factor analysis (EFA) with oblimin rotation [ 49 ]. Scree plots and Very Simple Structure (VSS) criterion were used to identify the number of factors [ 50 ]. The VSS criterion evaluates the magnitude of the changes in goodness of fit with each increase in the number of extracted factors.

Last, using the composite scores from the EFA, we used the identified latent constructs from the psychological impact survey items as input variables in a latent profile analysis (LPA) [ 51 ]. Criteria for determining the number of profiles in the LPA included statistical adequacy of the solution and interpretability of each profile [ 52 ]. Indices used to determine statistical adequacy included the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) and sample-size adjusted Bayesian Information Criterion [ 53 ]. For each of these indices, lower values represented better model fit. Also, the entropy criterion was calculated as a measure of classification precision [ 54 ]. We favored a parsimonious solution with fewer profiles over a more complex solution if this improved the interpretability of the LPA [ 53 ]. Z-scores of the input variables were used to interpret the profiles. The criteria to assign low and high values is not established and so we adopted previous studies' thresholds [ 53 ]. These included standardized scores between +0.5 and -0.5 being labelled as moderate, scores above 0.5 being labelled as high, and scores below -0.5 being labelled as low levels of psychological impact from COVID-19.

Objective 3 was achieved by modeling unadjusted (bivariate) and adjusted (multivariate) relationships between risk factors and profiles from the LPA. Unadjusted results are presented because multivariate models used a dichotomous outcome variable to distinguish students in the highest profile of psychological impact from those in the moderate or low profiles of impact (see Results for profile development and sample sizes within each profile). Determining risk factors for being in the high impact profile was deemed more important and actionable by university administrators than determining risk factors for each of the lower impact profiles, as would have been accomplished with a multinomial model. Thus, this modeling approach served a practical function; results could inform university administrators with tight budgets on how to prioritize funding for mental health interventions amongst students at greatest risk of high levels of psychological impact. Unadjusted results remained relevant, however, since they served the function of comparing risk factors between each level of impact profile in a simpler format than the output of a multinomial regression model.

For the unadjusted results, risk factors were evaluated with chi-squared contingency tables. Residuals from observed versus expected count comparisons determined the direction of the effect of the risk factors (i.e., more or less likely that a group was classified to a higher impact profile than another profile). Statistical significance of risk factors was calculated with Bonferroni adjustments to reduce Type I Error [ 55 ]. Continuous measures were reduced to dichotomous or categorical factors based on clinically meaningful levels, past research, and data distributions. BMI was classified into four categories (less than 18 = underweight; 18 to 24.99 = normal; 25 to 29.99 = overweight; 30.0 and over = obese) [ 56 ]. General health was separated into two groups: poor/fair health and good/very good/excellent health [ 53 ]. Screen time was separated into less than eight hours on a device and eight or more hours on a device [ 57 ]. Time outdoors was split into three groups: Less than 1.00 hour, 1.00 to 1.99 hours, and 2.00 hours or more [ 58 , 59 ]. Time spent exercising was also split into three groups: 30.00 minutes or less, 30.01 to 59.99 minutes, and 1.00 hour or more [ 60 ]. In addition, social class and relative income were split into three levels: below average, average, or above average. Levels of education were split into two levels: less than a 4-year college degree and a 4-year college degree or more [ 61 ].

For the adjusted results, we conducted generalized mixed-effects logistic regression to examine risk factors simultaneously and control for random (grouping) effects by institutional affiliation. To avoid collinearity in SES measures, whichever item correlated most strongly with psychological impacts was entered in the model. We used Variance Inflation Factor (VIF) values to test for multicollinearity. The proportion of variance explained was measured with conditional and marginal R 2 coefficients of determination [ 62 – 64 ]. Marginal R 2 represents the contribution of the predictors, which are modelled as fixed effects, whereas conditional R 2 accounts for the additional contribution of institutional affiliation (random effect) in addition to the fixed effects.

As a sensitivity analysis, we ran a logistic regression model with a subsample of respondents from the university that obtained a representative sample (North Carolina State University). This allowed us to evaluate the robustness of our nationwide sample, which otherwise utilized a convenience sampling approach.

Analyses were conducted in Excel for Mac Version 16.38 and R Version 3.6.2.

3.1 Array of impacts

Qualitative data from the open-ended responses demonstrated a broad array of impacts from COVID-19 on college students’ feelings ( Table 3 ) and behaviors ( Table 4 ). The most common changes in how students felt compared to before the pandemic were increased lack of motivation, anxiety, stress, and isolation. For example, one of the students reflected, “I'm normally extremely motivated, and I've never struggled with depression, but have recently felt very sluggish and melancholy.” Another student described their feelings related to isolation as “I feel trapped. I don't have anywhere I need to go since I can't socialize, and I have schoolwork. But yet I still feel trapped due to actual restrictions and suggestions.” The most frequent changes in student behavior compared to before the pandemic included more social distancing, more education changes, and less going out. Other concerning changes ranged from entrapment, boredom, fatigue, hopelessness, guilt, and inconvenience to hygiene, sleep, housing, employment, personal finances, and caretaking. For example, some students expressed their frustration with the financial situation, including one statement indicating: “I am BROKE. I lost my job because of this pandemic and now I can’t pay for groceries.” Other students were concerned about online learning. For example, one student commented: “I am constantly on edge about coursework: Did the computer register I submitted my exam? Did I see everything my teacher posted in Moodle? What happens if my internet goes out and I miss an assignment?”

https://doi.org/10.1371/journal.pone.0245327.t003

https://doi.org/10.1371/journal.pone.0245327.t004

Smaller numbers of students reported positive changes from the COVID-19 pandemic as well. These included optimism, productivity, adaptation, and empathy, as highlighted in the following quotes: “I've affirmed that people are capable of adapting in any circumstances” and “[I felt a] higher degree of empathy toward my community”.

3.2 Psychological impact profiles

Mean values of the psychological impact survey items are shown in Table 5 . Eight of these were included in the EFA. (Feeling guilty demonstrated low communality ( h 2 = .21) and was removed from further analyses.) All eight items displayed relatively normal distributions ( S1 Fig ). Criteria of the resulting model were acceptable: Tucker Lewis Index = 0.95; Kaiser-Meyer-Olkin (KMO) factor adequacy measure of sampling adequacy (MSA) = .89 [ 65 ]; significant Bartlett's test of sphericity, χ 2 (28) = 10503, p < .001. The VSS Criterion [ 50 ] achieved a maximum of .93 with a two-factor solution, compared to .89 for a one-factor solution or .94 for a three-factor solution ( S1 Table , S2 and S3 Figs). We labelled the first factor as "Emotional Distress" since it was composed largely of negative affect items (afraid, irritable, sad, preoccupied and stressed). The second factor was composed of three items dealing with how time was spent presumably in worry during the pandemic (worry, too much time and a lot of time), and so we labelled it "Worry Time." This is a term from clinical psychology that describes time spent reflecting on all the possible impacts of a health concern, including those worries that an individual cannot do anything about [ 66 ]. The internal reliability of the factors was high, Cronbach's ⍺ = .87 for Worry Time and .83 for Emotional Distress.

https://doi.org/10.1371/journal.pone.0245327.t005

A three-profile solution fit the data best for the LPA. Information criteria decreased with additional profiles up to a five-profile solution, indicating a better model fit ( S2 Table , S4 Fig ). The elbow plot suggested minor improvements in model fit after a three-profile solution. Adding a fourth or fifth profile provided less interpretable results. Based on the combined information from the statistical criteria and interpretability, we retained a three-profile solution as our final model.

The three levels of psychological impact from COVID-19 resulting from the LPA are depicted in Fig 1 . Positive z-scores indicate higher levels of impact and negative z-scores indicate lower levels of impact, compared to the average. Profile 1 ("high") represented students with higher than average levels of the two factors measuring psychological impacts (Emotional Distress, Worry Time) stemming from COVID-19. Profile 2 ("moderate") represented students with moderate levels of the two factors, and profile 3 ("low") represented students with low levels of the two factors. Regarding profile membership, 45.2% of students (n = 1,146) were within the high impact profile, whereas 40.4% (n = 1,025) were in the moderate profile and 14.3% (n = 363) were in the low profile.

Means and standard errors shown.

https://doi.org/10.1371/journal.pone.0245327.g001

3.3 Risk factors

A summary of the risk factors with significant differences between impact profiles based on bivariate Chi-square tests is depicted in Fig 2 . With respect to sociodemographic factors, women were more likely to be at risk than men ( χ 2 (2) = 88, p < .001). Specifically, women were more likely to be in the high profile (residuals (RES) = 8.02, p < .001) and less likely to be in the moderate (RES = -2.75, p = .036) or low (RES = -7.54, p < .001) profile. Men demonstrated the opposite pattern. We did not observe differences by academic status ( χ 2 (2) = .3, p = .9), although we did observe differences by age ( χ 2 (4) = 15, p = .005). Students who were 18 to 24 years old were more likely to be in the moderate profile (RES = 3.81, p = .0013), and students who were 25 to 32 years old were less likely to be in the moderate profile (RES = -3.03, p = .022) than other profiles. No other significant differences between age groups by profile were found, p > .05.

Sociodemographic (a), lifestyle (b), and COVID-19 victim awareness (c) risk factors associated with high, moderate, and low psychological impact profiles for students across the United States. Residuals from Pearson's chi-squared tests depict likelihood of profile membership based on risk factor. Only significant factors ( p < .05) are reported. Reference groups include men; over 32 age; other race/ethnicity; average/above average SES (social class and relative family income); good/very good/excellent general health; less than 2 hours of time outdoors; less than 8 hours of screen time; and not knowing someone infected (COVID-19).

https://doi.org/10.1371/journal.pone.0245327.g002

We also observed racial/ethnic and SES differences in psychological impact levels. Specifically, we found differences by race/ethnicity ( χ 2 (6) = 18, p = .007) with non-Hispanic Whites being more likely to be in the low profile (RES = 2.98, p = .035) and Non-Hispanic Asians being less likely to be in the low impact profile (RES = -3.42, p = .0076). No differences in impact profiles were observed for non-Hispanic Black students or Hispanic students, although sample sizes were small ( n = 95 and 98, respectively). Parental educational achievement measures further showed no differences in profiles, p = .5 for maternal and .9 for paternal. No differences were observed for parental social class either, p = .1 for maternal and .2 for paternal. In contrast, student social class ( χ 2 (4) = 14, p = .008), and relative family income ( χ 2 (4) = 14, p = .008) differed by impact profile. Students who reported above-average social class were more likely to be in the low profile (RES = 3.07, p = .019), and students who reported below-average relative family income were more likely to be in the high profile (RES = 3.38, p = .0065). No other significant differences between ethnoracial groups or SES measures by profile were found, p > .05.

Lifestyle-related factors predicted differences in impact profiles. For instance, general health predicted assignment to different impact profiles ( χ 2 (2) = 41, p < .001). Students with fair/poor health were more likely to be in the high profile (RES = 5.90, p < .001) and less likely to be in the moderate (RES = -2.67, p = .045) or low profile (RES = -4.58, p < .001). Students with good/very good/excellent health displayed the opposite pattern. No difference in impact profiles was observed for BMI ( χ 2 (6) = 9, p = .2). We observed differences in impact profiles by time outdoors ( χ 2 (4) = 13, p = .01) and screen time ( χ 2 (2) = 14, p = .001) but not by exercise time ( χ 2 (4) = 6, p = .2). Students who reported spending two or more hours outdoors were less likely to be in the high profile (RES = -3.17, p = .014), and students who reported spending more than eight hours on a device were more likely to be in the high profile (RES = 3.06, p = .013) and less likely to be in the moderate profile (RES = -3.67, p = .0014). Students spending less than eight hours on a device displayed the exact opposite trend. No other pair-wise comparisons in lifestyle-related factors were significant, p > .05.

Lastly, knowing someone who was infected with COVID-19 increased the likelihood of being at risk of psychological impacts ( χ 2 (2) = 14, p < .001). Students who knew someone in their family or community who was infected were more likely to be in the high profile (RES = 3.06, p = .013) and less likely to be in the moderate profile (RES = -3.67, p = .0014). Students who did not know an infected person displayed the opposite pattern.

Five variables remained significant predictors of impact profiles in models adjusting for all risk factors simultaneously while controlling for institutional affiliation ( Table 6 ). The SES measure entered in these models was social class of student, because it correlated more highly with psychological impact levels than other measures ( S5 Fig ). Students who were women, fair/poor general health, 18 to 24 years old, reporting 8 or more hours of screen time, and who knew someone infected with COVID-19 were more likely to be in the high profile. Non-Hispanic Asian students were marginally more likely to be in the high impact profile, p = .091. Effect sizes varied; women were approximately twice as likely to be assigned to the high impact profile as the moderate/low profile. Other predictors increased (or decreased) the likelihood of being in the high impact profile by approximately 20% to 40%. No institutions emerged as significant random effects ( S6 Fig ). VIF values < 2.0 indicated no multicollinearity. Approximately 7% of the variance was explained by the predictors and institutional affiliations.

https://doi.org/10.1371/journal.pone.0245327.t006

Sensitivity analyses with a subsample of respondents from the representative sample at North Carolina State University identified a similar set of predictors of psychological impact levels ( S3 Table ). Gender, age, general health, and knowing someone infected remained significant predictors. In contrast, screen time was no longer significant. Being Non-Hispanic Asian as marginally significant, p = .070, and social class was significant, p = .0038. Students of above average social class were 23.0% less likely to be assigned to the high impact profile.

4 Discussion

4.1 key findings and interpretation of results.

To evaluate the psychological impacts of COVID-19 on students in the United States, we collected over 2,500 survey responses from students at seven universities in late-February to mid-May 2020. Qualitative data from open-ended responses showed students experienced largely negative impacts of COVID-19 on psychological health and lifestyle behaviors. Among the most commonly reported changes were lack of motivation, anxiety, stress, and isolation, as well as social distancing, education changes, and going out less. Similar findings were reported by another study exploring the impact of COVID-19 on students at a single college in the United States, revealing increases in sedentary lifestyle, anxiety, and depressive symptoms [ 16 ]. A global study examining experiences of students in 62 countries, including one university in the United States, found that students’ expressed concerns about their academic and professional careers, as well as feelings of boredom, anxiety and frustration [ 10 ]. Increased anger, sadness, anxiety and fear were also reported by students in China [ 67 ]. Students in Switzerland reported a decrease in social interaction and higher levels of stress, anxiety, and loneliness [ 68 ]. More generally, adults have reported decreases in physical activity and food consumption increases during the COVID-19 pandemic quarantine compared to beforehand, as well as increases in binge drinking on average [ 69 ], which was identified in a small portion of our student respondents as well. Slight differences between our studies' results and results from studies conducted elsewhere may be due to the differences in student experience by geographical location. The United States is providing relatively little financial relief to college students during the pandemic compared to other Global North countries [ 70 ].

Quantitative survey measures captured the majority of the content that students entered in the open-ended responses (i.e., worry, stress, and fear) and informed the development of impact profiles. Students were assigned to one of three profiles—low (14% of the sample), moderate (40%) and high (45%)—based on the psychological impacts they reported experiencing in response to COVID-19.

In unadjusted models, students who were women, non-Hispanic Asian, in fair/poor health, of below-average relative family income, or someone who knew a family/community member infected with COVID-19 were at risk of higher levels of psychological impact. Students who were non-Hispanic White, above-average social class, spent at least two hours outside in the past day, or spent less than eight hours on screens in the past day were at less risk.

In multivariate models controlling, being a woman, being younger (18 to 24 years old), having poor/fair general health, reporting more screen time, and knowing someone infected were statistically significant risk factors. SES and identifying as non-Hispanic Asian were additional significant risk factors in the subsample of respondents obtained from representative sampling, whereas screen time was not significant in this sensitivity analysis.

These risk factor findings generally match those found in other studies that employed a case study approach within single United States universities/colleges. One longitudinal study of students at a public university in Nevada (n = 205) found that anxiety and depressive symptoms were greater in April 2020 than in prior months [ 17 ]. Women reported greater disruption to daily activities, mental and physical health, and personal finances than men. Contrary to our unadjusted findings, Asian or Asian-American students in the Nevada study reported lower levels of anxiety and depression than other races. A second longitudinal study with undergraduate students (n = 217) at a small liberal arts college in New Hampshire also found increases in anxiety, depression, and sedentary time during April 2020 relative to prior months [ 16 ]. COVID-19 risk factors for college students at other countries have been strikingly similar, as explained below.

Over ten studies, including several with college student populations, identify women as being at greater risk of psychological distress during the COVID-19 pandemic [ 1 , 10 , 21 , 71 – 77 ]. Women are generally prone to depression and anxiety disorders [ 14 ], and although initial evidence indicated men were more susceptible to infection [ 77 ], our study supports the assertion that women appear to be more strongly impacted by the long-term psychological impacts of the pandemic. This observation may be attributable to higher levels of pre-existing psychopathology in women as well as gender differences in fear processing, which could translate to exacerbations of symptoms [ 78 ]. Also, male students tend to have higher confidence in the computer skills necessary for the transition to online course delivery [ 10 ]. Meanwhile, women are more concerned about impacts on their professional career and ability to study than men, on average [ 10 ]. One study attributed these gender differences to greater emotional expression, less tolerance for uncertainty, and less-effective coping strategies amongst students who are women [ 75 ]. Women have also reported being more susceptible to "emotional hunger" and subsequent increased food intake than men during COVID-19 quarantine; these behaviors can lead to weight gain and poor mental health [ 73 ].

Our findings that fair/poor general health is a risk factor has been documented in numerous other populations during COVID-19 [ 79 , 80 ]. In addition to comorbidity between mental and physical health status, people with pre-existing health problems and those with poor mental health show lower preparedness for disasters and suffer disproportionately more from disaster-related outcomes [ 81 ].

Several reasons explain our findings that younger students may be at greater risk than older students. Younger students (i.e., 18 to 24 years old, regardless of academic status) tend to be more worried about their future education and ability to pay for college education than older students [ 10 ]. Younger people also engage in social media more than older people during the pandemic [ 12 , 82 ]. Given the dominance of the COVID-19 pandemic in the news, younger "always-on" students may be exposed to greater amounts of risk-elevating messages, which can lead to anxiety and poor mental health [ 16 , 75 ].

Regarding our findings that non-Hispanic Asian students may be at greater risk than other races/ethnicities, several studies show higher psychological distress from COVID-19 in this population [ 10 ]. Asians and Asian Americans have reported being discriminated against by other students on social media during the pandemic [ 83 ]. Further, this population has experienced long-standing barriers to receiving and participating in mental health services [ 84 ].

The current study provides some support toward the mounting evidence that excessive screen time, including during the pandemic, may negatively impact mental health [ 85 ]. People who manage COVID-19 anxiety with excessive use of smartphones and other screen-based technology inadvertently learn more about the virus from the news, which fuels anxiety and ongoing coping through screens, thus causing a downward spiral [ 82 ]. Excessive use of digital media also detracts from time that could be spent on other health-promoting activities such as outdoor recreation [ 86 ]. Our study supports these relationships, suggesting negative impacts of screen time and positive impacts of "green time" on students' psychological health. The unadjusted analyses suggested that outdoor time predicted psychological impacts of COVID-19, although this variable was not significant in multivariate models. Other studies justify its consideration as a risk factor by university administrators. Both outdoor recreation [ 87 ] and nature exposure [ 88 , 89 ] can improve psychosocial and eudaimonic well-being [ 90 , 91 ]. Recent studies of people across the world show protective psychological effects of park and green space access during the pandemic [ 92 ] as well as lower rates of infection and mortality [ 93 ].

The finding that knowing someone infected is a risk factor for psychological impacts of COVID-19 is intuitive. Familiarity can increase the salience and perceived risk of becoming infected and dealing with subsequent health concerns, like COVID-19-related death [ 79 ]. Also, the threat of death from COVID has been associated with students' mental health and explainable by unhealthy levels of smartphone use [ 82 ].

As suggested in our unadjusted analyses and the multivariate model with the representative sample, SES may influence students' mental health during the pandemic. This might be a result of financial concerns affecting college students and their families [ 10 ]. SES has been documented as a predictor of COVID-19 fear and mental health concerns in other populations [ 10 , 74 , 79 , 94 – 98 ]. Students coming from low-SES families may be more concerned about basic needs, like food and shelter, caused by loss of their or their parent's income [ 99 ]. Furthermore, since low-SES families are more susceptible to COVID-19 infection [ 98 ], students may be more concerned for their own and their families' safety.

4.2 Recommendations for universities

Given the large percentage of students assigned to the high psychological impact profile, universities would be well-served to address the mental health needs of their entire student body. Select programs that have promoted mental health—such as those at the University of Connecticut, University of Kentucky, and Northeastern—include virtual group exercise and meditation/mindfulness sessions, accountability buddies and exercise challenges and tele-medicine/counseling visits [ 99 ]. These group meetings may be helpful not only in lowering anxiety but also in decreasing the sense of isolation reported by the students in this study. Digital interventions for students with clinical levels of anxiety or depression as well as potential for self-harm or suicide can involve automated and blended therapeutic interventions (such as apps and online programs), calls/text messages to reach those with less digital resources, suicide risk assessments, chatlines and forums, and other technologies to monitor risk either passively or actively [ 80 ]. Recently, Chen et al. [ 100 ] recommended a six-step intervention for the reduction in psychological impact risk amongst Chinese college students. These steps included the delivery of positive pandemic-related information, reduction in negative behavior, learning about stress management techniques, improvements in family relationships, increases in positive behavior, and adjustments in academic expectations.

Given the likelihood of ongoing psychological distress from COVID-19, universities may also consider helping students maintain healthy mindsets rather than avoiding stress [ 101 ]. In support of this proposition are recent findings that cognitive and behavioral avoidance (i.e., avoiding situations where exposure is possible and difficult thoughts about the pandemic) was the most consistent predictor of increased anxiety and depressive symptoms during the pandemic [ 17 ]. Cognitive reappraisal of stressful situations can alter their negative impacts [ 102 , 103 ]. Training students to shift their educational experience mind-set to one that focuses on the "silver linings" and emerging opportunities may lead to "stress-related growth" and "toughening" [ 104 , 105 ]. Adaptive mindsets can also help reorganize priorities to develop deeper relationships and greater appreciation of life [ 106 ], as well as help students to adjust to new ways of learning. Since a portion of the students in this study reported feeling less motivated, productive, and able to focus, switching to an adaptive mindset may help students persevere in their education and later in life. Finally, mindset reappraisals can increase well-being, decrease negative health symptoms, and boost physiological functioning under acute stress when a family member becomes infected or the pandemic creates rapid shifts in policies and procedures that affect students [ 107 , 108 ].

Universities can further develop platforms that facilitate safe student social interaction. Many students seek out social interaction during their university experience [ 109 – 111 ]. However, as the findings of this study revealed, students’ opportunities for socializing significantly decreased in the early stages of COVID-19. Missing "going out" and important milestone events (e.g., graduation, last sporting event) was a frequent response from our student participants. Other studies found that in order to maintain students' mental health during the first wave of the COVID-19 pandemic, they communicated online with close family members or roommates at least daily [ 10 ]. With college students, physical distancing does not and should not require "social distancing" [ 101 ]. Both synchronous (i.e., Zoom) and asynchronous (i.e., Facebook group) online interactions can foster bonding and bridging social connection [ 112 – 115 ], which can extend beyond social media posts and email listservs. Normal venues where people congregate such as places of worship, gyms, cafeterias, yoga studios and classes can be replicated online or even held outdoors in temperate weather on a schedule similar to what was in place prior to the pandemic [ 116 ]. Other recently-successful interventions include the facilitated online sharing of recipes, books, and podcasts as well as virtual movie, game, trivia, or happy hour nights [ 99 ]. Providing support to student organizations to coordinate these virtual social activities could accelerate the availability of these resources.

Colleges and universities also have a moral obligation to boost their outreach to particularly vulnerable groups–that is, populations at risk of high levels of psychological impact from COVID-19 [ 14 ]. As documented in the impact profiles of our study, people at increased risk include women, younger students, students with pre-existing health concerns, students spending at least one-third of their day (including time spent sleeping) on screens, and students with family or community members who are infected with COVID-19. Monitoring and reporting rates of anxiety, depression, self-harm, suicide and other mental health issues within these groups is necessary to allocate counseling services and intervene pre-emptively and at times of acute symptomology [ 80 ]. Further, universities can provide accommodations for assignments and exams using a more personalized approach to learning and create enhanced opportunities for virtual social interactions with peers. These efforts may help at-risk groups succeed academically, build stronger relationships, and enhance their sense of belonging during distant learning [ 117 ].

Students in this study also expressed stress and anxiety associated with changes in education mode during the pandemics. As previous research has found, academic success may be supported with virtual town halls, regular email check-ins, virtual office hours, and peer mentoring [ 104 ]. Globally, students' satisfaction with university response to COVID-19 is predicted by students' satisfaction with pre-recorded videos during online course delivery, sufficient information on exams, satisfaction with teaching staff, satisfaction with websites and social media information with regular updates from the university, hopefulness, (lack of) boredom, (lack of) study issues, being on scholarship, being able to pay for school, and study discipline (social sciences tend to be less impacted than hard sciences or engineering) [ 10 ]. Universities may be encouraged by findings from another study on the switch to online courses; this study found many students were not challenged by the transition because of their aptitude toward digital learning and new technologies [ 118 ]. However, another study found new software platforms can be a challenge for some students [ 10 ].

4.3 Strengths and limitations

The primary strength of this study is the development of psychological impact profiles using data from universities across the United States. This sampling approach is also a limitation, however. Whereas all the included universities were teaching exclusively online during the study, their respective states and localities may have experienced differing levels of social distancing policy and enforcement. Another limitation related to the sample is the high percentage of non-Hispanic Whites. This occurrence was likely the result of the demographic composition of the colleges and departments targeted for recruitment [ 119 ]. Selection bias related to which students participated in the study questionnaire based on interest and access/availability is also possible [ 3 ].

Another limitation is the quantitative assessment of the psychological impacts of COVID-19, which could have limited the utility of our impact profiles. We did not measure substance abuse, which is expected to be a ramification of the virus [ 116 ] and which anxious individuals are prone to under-report [ 120 ]. Such counterproductive coping behaviors could be particularly problematic for college students [ 121 ]. Further, because our predictors explained a small amount of variance of the profiles, other unmeasured (or better measured) factors might predict students’ psychological risk. For example, our single-item measures of leisure time activities could be improved with a more comprehensive assessment of time budgets such as those employed in episodic time use surveys [ 122 ].

We were primarily interested in reactions to the pandemic rather than how people were feeling/behaving during the pandemic. Therefore, we did not employ standardized measures of stress, anxiety, depression, or well-being. This limits our findings from being directly compared to other studies and pooled in meta-analyses.

Lastly, our measures were retrospective rather than longitudinal, which decreases our ability to say with confidence that the reported impacts were caused by COVID-19. However, we are fairly confident that the findings are attributable to the pandemic given our survey prompts. They specified students' responses to COVID-19 rather than asked generalized psychological states, and the findings strongly aligned with those of longitudinal studies of college students during the pandemic [ 16 , 17 , 37 , 123 – 125 ].

5 Conclusion

Our cross-sectional study found that being a woman, being of younger age, experiencing poor/fair general health, spending extensive time on screens, and knowing someone infected with COVID-19 were risk factors for higher levels of psychological impact during the pandemic among college students in the United States. Unadjusted analyses also suggested that students who were non-Hispanic White, were not non-Hispanic Asian, were of higher-SES, or spent at least two hours outside experienced lower levels of psychological impact. That said, all students surveyed reported being negatively affected by the pandemic in some way, and 59% of respondents experienced high levels of psychological impact.

At the time that these data were collected, the education of over 1.5 billion students across the world were affected by COVID-19 [ 126 ]. Rates of student psychological distress were as high as 90% [ 17 , 127 ]. Students must "Maslow before they can Bloom; " in other words, their basic physiological, psychological, and safety needs must be met prior to them focusing on–much less excelling–in academic life [ 99 ]. We recommend that university administrators take aggressive, proactive steps to support the mental health and educational success of their students at all times, but particularly during times of uncertainty and crisis–notably, the COVID-19 pandemic.

Supporting information

S1 fig. distributions and relationships between covid-19 psychological impact survey items, including histograms, pearson correlation coefficients, and scatter plots..

https://doi.org/10.1371/journal.pone.0245327.s001

S2 Fig. Diagram of EFA on COVID-19 psychological impact survey items.

https://doi.org/10.1371/journal.pone.0245327.s002

S3 Fig. Scree plot of EFA on COVID-19 psychological impact survey items.

https://doi.org/10.1371/journal.pone.0245327.s003

S4 Fig. Elbow plot of the information criteria for the latent profile analysis.

https://doi.org/10.1371/journal.pone.0245327.s004

S5 Fig. Correlations between socio-economic measures and the two psychological impact profiles.

https://doi.org/10.1371/journal.pone.0245327.s005

S6 Fig. Conditional mean values ("condval") and standard deviations of institutional affiliation (university) random effects from mixed-effects logistic regression predicting high versus medium/low psychological impact profile from COVID-19.

https://doi.org/10.1371/journal.pone.0245327.s006

S1 Table. Item loadings and fit statistics of EFA on COVID-19 psychological impact survey items.

https://doi.org/10.1371/journal.pone.0245327.s007

S2 Table. Fit indices, entropy and model comparisons for estimated latent profile analyses models.

https://doi.org/10.1371/journal.pone.0245327.s008

S3 Table. Results of binomial logistic regression modelling likelihood of risk factors predicting high versus low/moderate levels of COVID-19 psychological impact for students at North Carolina State University, where a representative sample was collected ( N = 1,312).

https://doi.org/10.1371/journal.pone.0245327.s009

https://doi.org/10.1371/journal.pone.0245327.s010

• View Article
• PubMed/NCBI
• Google Scholar
• 2. Vindegaard N, Benros ME. COVID-19 pandemic and mental health consequences: Systematic review of the current evidence. Brain, Behavior, and Immunity. Academic Press; 2020. https://doi.org/10.1016/j.bbi.2020.05.048 pmid:32485289
• 4. Loades ME, Chatburn E, Higson-Sweeney N, Reynolds S, Shafran R, Brigden A, et al. (2020). Rapid systematic review: The impact of social isolation and loneliness on the mental health of children and adolescents in the context of COVID-19. J Am Acad Child Adolesc Psychiatry. American Academy of Child & Adolescent Psychiatry; 2020; 1–46. https://doi.org/10.1016/j.jaac.2020.05.009 pmid:32504808
• 5. Luo M, Guo L, Yu M, Jiang W, Wang H. The psychological and mental impact of coronavirus disease 2019 (COVID-19) on medical staff and general public–A systematic review and meta-analysis. Psychiatry Research. Elsevier Ireland Ltd; 2020;291: 113190. https://doi.org/10.1016/j.psychres.2020.113190 pmid:32563745
• 7. Cunningham PW, Firozi P. The Health 202: Texts to federal government mental health hotline up roughly 1,000 percent. In: washingtonpost.com [Internet]. 4 May 2020 [cited 23 Jun 2020] pp. 1–15. Available: https://www.washingtonpost.com/news/powerpost/paloma/the-health-202/2020/05/04/the-health-202-texts-to-federal-government-mental-health-hotline-up-roughly-1-000-percent/5eaae16c602ff15fb0021568/
• 8. Hollyfield A. Coronavirus impact: Suicides on the rise amid shelter-in-place order, Bay Area medical professionals say. In: abcnews.com [Internet]. 21 May 2020 [cited 23 Jun 2020] pp. 1–4. Available: https://abc7news.com/suicide-covid-19-coronavirus-rates-during-pandemic-death-by/6201962/
• 10. Aristovnik A, Keržič D, Ravšelj D, Tomaževič N, Umek L. Impacts of the COVID-19 pandemic on life of higher education students: A global perspective. preprints.org . 2020. https://doi.org/10.20944/preprints202008.0246.v2
• 12. Stieger S, Lewetz D, Swami V. Psychological well-being under conditions of lockdown: An experience sampling study in Austria during the COVID-19 pandemic. psyarxiv.com . 2020. https://doi.org/10.31234/osf.io/qjhfp
• 13. Gruttadaro D, Crudo D. College students speak: A survey on mental health [Internet]. National Alliance on Mental Health. Fairfax, VA; 2012 Nov pp. 1–24. Available: https://www.nami.org/Support-Education/Publications-Reports/Survey-Reports/College-Students-Speak_A-Survey-Report-on-Mental-H
• 15. Sankhi S, Marasine NR. Impact of COVID-19 Impact of COVID-19 pandemic on mental health of the general population, students, and health care workers: A review. preprints.org . 2020. https://doi.org/10.20944/preprints202007.0616.v1
• 17. Zimmermann M, Bledsoe C, Papa A. The impact of the COVID-19 pandemic on college student mental health: A longitudinal examination of risk and protective factors. psyarxiv.com . 2020. https://doi.org/10.31234/osf.io/2y7hu
• 24. Marmolejo F. America spends more on college than virtually any other country. In: Forbes.com [Internet]. 10 Sep 2020 [cited 10 Sep 2020] pp. 1–9. Available: https://www.forbes.com/sites/prestoncooper2/2019/09/22/america-spends-more-on-college-than-virtually-any-other-country/#2fe46f483348
• 25. Bhagavathula AS, Rahmani J, Aldhaleei WA, Kumar P, Rovetta A. Global, regional and national incidence and case-fatality rates of novel coronavirus (COVID-19) across 154 countries and territories: A systematic assessment of cases reported from January to March 16, 2020. medRxiv.org . 2020. https://doi.org/10.1101/2020.03.26.20044743
• 30. Watson D, Clark LA. The PANAS-X: Manual for the positive and negative affect schedule-expanded form. University of Iowa; 1994.
• 36. Centers for Disease Control and Prevention. Healthy Days Core Module: HRQOL-14 Measure. In: cdc.gov . 31 Oct 2018 pp. 1–10.
• 37. Xiao J, Jiang Y, Zhang Y, Gu X, Ma W, Zhuang B. The impact of psychology interventions on changing mental health status and sleep quality in university students during the COVID-19 pandemic. medRxiv.org . 2020. https://doi.org/10.1101/2020.09.01.20186411
• 43. Lune H, Berg BL. Qualitative Research Methods for the Social Sciences. Pearson; 2016.
• 44. Maxwell JA. Qualitative Research Design: An Interactive Approach. SAGE; 2013.
• 48. Kuhn M. caret: Classification and Regression Training. In: cran.r-project.org [Internet]. 20 Mar 2020 [cited 24 Jun 2020] pp. 1–2. Available: https://cran.r-project.org/web/packages/caret/index.html
• 51. Hennig C, Meila M, Murtagh F, Rocci R. Handbook of Cluster Analysis. Chapman and Hall/CRC; 2015.
• 65. Tabachnick BG, Fidell LS. Using multivariate statistics. 5 ed. Boston, MA: Allyn & Bacon; 2007.
• 69. Ammar A, Brach M, Trabelsi K, Chtourou H, Boukhris O, Masmoudi L, et al. Effects of COVID-19 home confinement on physical activity and eating behaviour Preliminary results of the ECLB-COVID19 international online-survey. medRxiv.org . 2020. pp. 1–24. https://doi.org/10.1101/2020.05.04.20072447
• 70. Holzer HJ. The COVID-19 crisis: How do U.S. employment and health outcomes compare to other OECD countries? In: Brookings Institute [Internet]. Jun 2020 [cited 17 Sep 2020] pp. 1–12. Available: https://www.brookings.edu/research/the-covid-19-crisis-how-do-u-s-economic-and-health-outcomes-compare-to-other-oecd-countries/
• 83. Duong V, Pham P, Yang T, Wang Y, Luo J. The Ivory Tower lost: How college students respond differently than the general public to the COVID-19 pandemic. arxiv.org . 2020.
• 93. Klompmaker JO, Hart JE, Holland I, Sabath MB, Wu X, Laden F, et al. County-level exposures to greenness and associations with COVID-19 incidence and mortality in the United States. medRxiv.org . 2020. pp. 1–14. https://doi.org/10.1101/2020.08.26.20181644 pmid:32908990
• 126. UNESCO. School closures caused by Coronavirus (Covid-19). In: en.unesco.org . 11 Sep 2020 pp. 1–4.
• 127. Krakauer Hubner von C, Bruscatto ML, Lima RD. Distress among Brazilian university students due to the Covid-19 pandemic: survey results and reflections. medRxiv.org . 2020. https://doi.org/10.1101/2020.06.19.20135251

Advertisement

A quantitative approach to the intersectional study of mental health inequalities during the COVID-19 pandemic in UK young adults

• Open access
• Published: 24 January 2023
• Volume 59 , pages 417–429, ( 2024 )

Cite this article

You have full access to this open access article

• Darío Moreno-Agostino 1 , 2 ,
• Charlotte Woodhead 2 , 3 ,
• George B. Ploubidis 1 , 2   na1 &
• Jayati Das-Munshi 2 , 3 , 4   na1  

5434 Accesses

11 Citations

10 Altmetric

Explore all metrics

Mental health inequalities across social identities/positions during the COVID-19 pandemic have been mostly reported independently from each other or in a limited way (e.g., at the intersection between age and sex or gender). We aim to provide an inclusive socio-demographic mapping of different mental health measures in the population using quantitative methods that are consistent with an intersectional perspective.

Data included 8,588 participants from two British cohorts (born in 1990 and 2000–2002, respectively), collected in February/March 2021 (during the third UK nationwide lockdown). Measures of anxiety and depressive symptomatology, loneliness, and life satisfaction were analysed using Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy (MAIHDA) models.

We found evidence of large mental health inequalities across intersectional strata. Large proportions of those inequalities were accounted for by the additive effects of the variables used to define the intersections, with some of the largest gaps associated with sexual orientation (with sexual minority groups showing substantially worse outcomes). Additional inequalities were found by cohort/generation, birth sex, racial/ethnic groups, and socioeconomic position. Intersectional effects were observed mostly in intersections defined by combinations of privileged and marginalised social identities/positions (e.g., lower-than-expected life satisfaction in South Asian men in their thirties from a sexual minority and a disadvantaged childhood social class).

We found substantial inequalities largely cutting across intersectional strata defined by multiple co-constituting social identities/positions. The large gaps found by sexual orientation extend the existing evidence that sexual minority groups were disproportionately affected by the pandemic. Study implications and limitations are discussed.

Similar content being viewed by others

Changes in young adults' mental well-being before and during the early stage of the COVID-19 pandemic: disparities between ethnic groups in Germany

COVID-19-Induced Inequalities and Mental Health: Testing the Moderating Roles of Self-rated Health and Race/Ethnicity

A repeated cross-sectional and longitudinal study of mental health and wellbeing during COVID-19 lockdowns in Victoria, Australia

Avoid common mistakes on your manuscript.

Introduction

The quantitative study of health inequalities has often been inadequately underpinned by social theory [ 1 ]. Quantitative studies have frequently focused on examining inequalities in relation to broad social categories such as gender, race/ethnicity, and socioeconomic position (SEP), with the social forces driving these inequalities [ 2 , 3 , 4 , 5 ] often being under-acknowledged. This can contribute to the perpetuation of deficit-based or damage-centred perspectives which locate the “problem” of inequality within the group(s) being examined rather than the underlying structures and processes [ 6 , 7 ], which serve as the up-stream, fundamental causes of such inequalities [ 5 ]. Similarly, the complexity of personal experience, in that people occupy more than one social identity/position which can include a mix of advantaged and disadvantaged identities/positions that are dynamic and context-dependent [ 8 , 9 , 10 ], gets frequently under-recognised.

Intersectionality theory [ 11 ] supports a move away from some of these issues by highlighting that social identities and positions are “interdependent and mutually constitutive rather than independent and uni-dimensional” [ 12 ]. It acknowledges that, due to interlocking systems of oppression, the experiences of a person living at a particular intersection (e.g., Black woman) cannot be understood by independently looking at the experiences associated to each of the identities and positions that define it (in the same example, the experiences associated with being Black and a woman).

Although intersectional research poses challenges for both qualitative and quantitative methodological approaches [ 12 ], it has relied mostly on qualitative methods. Quantitative approaches to intersectionality have been criticised for their potential to unintentionally reinforce the idea that the observed inequalities may be natural or intractable [ 13 , 14 ] and “blunt [the] critical edge and transformative aims” of intersectionality [ 3 ] by simply describing those inequalities. Intercategorical approaches to intersectional complexity [ 15 ], where analytical categories (e.g., based on gender) are used to explore inequalities, and the focus on identifying significant differences across such categories (a focus that has been named “intersectionality as a testable hypothesis”) [ 11 ], have also been criticised. By focusing on the differences between groups, these approaches may dismiss the differences within those groups, unintentionally reinforcing the idea that they are homogeneous [ 11 ]. Furthermore, the use of the most privileged categories (e.g., White, male) as the reference can implicitly maintain the idea of dominant groups being the standard to which the rest of categories should be compared [ 16 ]. This can also result in a lack of evidence on intersections defined by combinations of privileged and marginalised identities and positions, which is essential to understand and address health inequalities [ 17 ].

Nonetheless, quantitative approaches provide unique opportunities to accurately document population health inequalities [ 14 ]. First, many of the above-mentioned critiques are not inherent to quantitative methods [ 12 , 18 ]. Categories can be provisionally adopted to explore inequalities across intersections [ 15 ] and acknowledged as proxies for the interlocking systems of oppression [ 14 , 17 ]. Furthermore, aspects such as SEP reflect material conditions rather than social constructions. In addition, novel quantitative approaches [ 18 , 19 , 20 , 21 , 22 , 23 , 24 ] can help overcome some of the critiques. Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy (MAIHDA) models [ 17 , 23 ] constitute a paradigmatic example. Unlike more traditional quantitative intercategorical approaches (e.g., fixed-effects regression models with interaction terms), MAIHDA models open the way to provide evidence at intersections that would otherwise be overlooked [ 18 , 21 ]. Moreover, they provide estimates of the variability/heterogeneity within those intersections and the proportion of variability that is attributable to differences between them [known as Variance Partition Coefficient (VPC) or Intra-Class Correlation (ICC)]. Such estimates can be interpreted as a measure of the “discriminatory accuracy” of the categories provisionally adopted to define the intersections, and can be relevant to inform public policy, because targeting interventions at specific intersections when very little of the variability is attributable to differences between intersections (i.e., when discriminatory accuracy is low) may lead to ineffective interventions [ 23 ].

MAIHDA models focus on the difference between the expected levels at particular intersections, operationalised as the sum of the effects of each of the categories that define them (i.e., the “sum of the parts” or the additive effects), and the observed levels at those intersections. Such “excess” or residual effect represents what is above and beyond the “sum of the parts”, what is unique to that particular intersection: the “intersectional effect”. Intersectional effects represent the impact of experiences of marginalisation and/or privilege due to interlocking systems of oppression in the outcomes under study [ 25 ]. The distinction between intersectional “experiences” and “effects” is crucial: failure to find significant intersectional effects does not preclude the existence of different experiences lived by different intersections [ 21 , 25 ]. Hence, MAIHDA models provide the opportunity to study intersectional complexity from one angle, which can then be complemented by qualitative, experiential, and other quantitative approaches for a more complete understanding [ 18 , 26 ]. This angle is descriptive in the sense that it does not engage in the statistical analysis of causal processes driving the inequalities described [ 13 ]. However, by explicitly engaging with social theory, they can situate those inequalities in the context of the underpinning social processes causing them, thus “maintain(ing) the critical and transformative edge of intersectionality” [ 1 ].

An applied example: mental health inequalities during the COVID-19 pandemic in the UK

The onset of the COVID-19 pandemic has had unequal implications for different groups within the population [ 27 , 28 ]. Evidence suggests disproportional mental health effects among disadvantaged population groups including adolescents and young adults, women, racialised and ethnically minoritised groups, sexual and gender minority groups, and those in more disadvantaged SEP [ 29 ]. UK-based research replicates these findings in outcomes such as anxiety and depressive symptomatology, psychological distress, loneliness, and life satisfaction [ 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 ]. In most cases, however, mental health inequalities by different social identities and positions have been reported independently from each other. Hence, the mutual co-constitution of those broader social categories has been left unacknowledged (or has been acknowledged in a very limited way, such as at the intersection between age and sex or gender) [ 12 , 18 ].

Using MAIHDA models, this study aims to provide evidence within the UK on mental health across multiple intersectional positions defined by categories closely tied to social power such as age, sex, race/ethnicity, sexual orientation, and SEP. This will first provide a “socio-demographic mapping” of the levels of different mental health measures within the population [ 14 ], which in turn will support the development of hypotheses for further research and suggest avenues for public health resource allocation.

This study focused on the most recent assessment of two British cohorts: Next Steps (NS) [ 44 ] and Millennium Cohort Study (MCS) [ 45 ], with participants born in 1990 and 2000–2002, respectively. This assessment took place in February/March 2021, during the third nationwide lockdown [ 46 ], as part of the third wave of the ‘COVID-19 Survey’ [ 47 ]. Both cohorts implemented oversampling methods to ensure representation from marginalised populations [ 44 , 45 ]. We focused on participants who were alive and still residing in the UK during the third wave of the COVID-19 Survey (February/March 2021). Due to the use of web and telephone interviews, the largest response rates within the target population were achieved in this wave of the COVID-19 Survey: 26.4% (NS) and 23.0% (MCS). Overall, 8588 participants (4167 from NS, 4421 from MCS) were included. All participants provided informed consent. Further details on the sample and procedure are available elsewhere [ 47 ].

Measures of anxiety symptomatology, depressive symptomatology, loneliness, and life satisfaction were collected using the same assessment tools across the two cohorts. Anxiety and depressive symptomatology were measured using the 2-item versions of the Generalised Anxiety Disorder (GAD-2) [ 48 ] and Patient Health Questionnaire (PHQ-2) [ 49 ]. These questionnaires enquire about how frequently the respondent has been bothered by core experiences of anxiety or depression, respectively, with scores ranging from 0 (lowest anxiety/depression) to 6 (highest anxiety/depression). Loneliness was measured with the University of California Los Angeles 3-item loneliness scale (UCLA-3) [ 50 ], which enquires about the extent to which the respondent has felt lack of companionship, left out, or isolated from others, and with scores ranging from 3 (lowest loneliness) to 9 (highest loneliness). Life satisfaction was measured with the Office for National Statistics (ONS) single question [ 51 ], with scores ranging from 0 (lowest life satisfaction) to 10 (highest life satisfaction).

Indicators/proxies for social identities/positions

Cohort/generation was assigned from the cohort of provenance. NS participants were in their early 30s at the time of the interview, whereas MCS participants were in their late teens/early 20s.

Information on birth sex as a binary variable (female or male) was obtained from the parents in the earlier waves.

Information on race/ethnicity corresponded to the most recent self-designated racial/ethnic group, complemented by the parents’ report wherever the former was not available. Responses were obtained following the ONS criteria [ 52 ] and, due to the small number of participants in some of the individual groups, grouped into White (including all White groups), Mixed (including all Mixed groups), South Asian (including Indian, Pakistani, and Bangladeshi groups), Black (including Black African, Black Caribbean, and Black British groups), and Other (including all ethnicities not included in the previous groups).

Self-reported information on sexual orientation was obtained from participants. Due to the small number of cases in some of the minority categories, we grouped participants into heterosexual versus sexual minority (including bisexual, gay/lesbian, and other) for analyses.

The residential Index of Multiple Deprivation (IMD) was used as an indicator of the current household SEP. A binary variable indicating whether the person lived in an area above (less deprived) or below (more deprived) the within-country median IMD rank was derived (the methodology used to generate IMDs varies across UK countries [ 53 ]). Self-reported information on housing tenure, collected during the COVID-19 Survey and grouped into Owners (including part owners) and Not owners, was used as an alternative indicator of the current household SEP. Finally, harmonised data on parental social class at age 11/14 years were used as an indicator of the household SEP during childhood [ 54 ], grouped into Non-manual/advantaged (including Professional, Managerial and Technical, and Skilled non-manual groups) and Manual/disadvantaged (including Skilled manual, Partly skilled, and Unskilled). Residential IMD was prioritised as SEP indicator due to the smaller number of missing data.

Intersectional strata were first generated not including socioeconomic indicators, resulting in 2 (cohorts/generations) * 2 (birth sex) * 5 (ethnicity groups) * 2 (sexual orientation) = 40 intersectional strata (stratification 40). Strata including indicators of SEP were then generated using either residential IMD rank (stratification 80a), current housing tenure (stratification 80b), or harmonised childhood social class (stratification 80c), resulting in up to 80 strata reflecting the intersection with different aspects of the SEP.

Statistical analysis

We used MAIHDA models [ 17 , 23 ] to obtain estimates of the residual/intersectional effects (i.e., what is beyond what would be expected based on the fixed/main/additive effects, conceptually similar to interaction effects) and predicted effects (including both the expected and residual/intersectional effects) at the different intersectional strata in each outcome. We first estimated intercepts-only (or “null” [ 17 , 25 ]) models with no predictors to obtain estimates of the degree of clustering or correlation within the strata (or, analogously, the proportion of the variance explained by differences across strata) ( VPC intercepts-only ). Then, main models were estimated including the variables adopted to define the intersectional strata as predictors. The fixed effects of each of those predictors (cohort/generation, birth sex, racial/ethnic group, sexual orientation, and the appropriate SEP indicator depending on the stratification used) represent the main/additive effects of the specific category across all intersections (non-intersectional effects). The VPC from the main models (VPC main ) returns information on the degree of clustering or correlation within intersectional strata (or, analogously, the proportion of the variance explained by differences across strata) after accounting for the fixed (or main, or additive) effects of each of the variables used to define these (the “sum of the parts”) [ 17 ]. The percentage of between-strata variance accounted for by the inclusion of those main/additive effects, or Proportional Change in Variance (PCV), was obtained as

Models were estimated using the four above-mentioned stratifications (40, 80a, 80b, and 80c). Following the procedure and code laid out by Dr Claire Evans [ 21 ], models were first estimated using Bayesian Markov Chain Monte Carlo (MCMC) procedures [ 55 ] with diffuse priors, initialisation values obtained from analogous models estimated with quasi-likelihood methods, and 50,000 iterations with a burn-in period of 5000 iterations and thinning every 50 iterations. Stratum-specific residual values (the intersectional effects [ 25 ]) and predicted values (including both the stratum-specific residuals and the fixed effects of each of the social identities/positions defining the stratum) were obtained from the main models, and 95% credible intervals (CI) were constructed using the 2.5 and 97.5 percentiles of those values across the MCMC iterations.

Initial checks (Supplementary Appendix S1) suggested that survey non-response was introducing bias. Based on these results, Bayesian MCMC MAIHDA models may be adequate to provide a socio-demographic mapping of the mental health levels among the survey respondents. Weighted analyses to account for the survey design and non-response are not yet implemented in Bayesian MCMC MAIHDA models. We estimated an identical set of models with maximum-likelihood (ML) estimation using weights to account for survey design and non-response, thus increasing the generalisability of the results beyond the survey respondents to each survey’s target population. One caveat is that ML estimation does not provide confidence intervals for the stratum-specific residuals (the intersectional effects).

Fixed-effects multiple regression models including the interaction across all the variables adopted to define the intersections were estimated for comparison purposes. Details on the rationale for these additional analyses are available in Supplementary Appendix S2.

MCMC MAIHDA models were estimated in MLwiN version 3.01 [ 56 ], using the runmlwin function [ 57 ] in Stata/MP 17.0 [ 58 ]. ML MAIHDA models and multivariable regression models were estimated in Stata/MP 17.0.

Most participants across both cohorts were female, White, and heterosexual (Supplementary Table S1). Sample sizes varied across models due to different missingness in the outcomes and SEP indicators. When accounting for the SEP indicators, some strata corresponding to intersections with racial/ethnic and sexual minority groups had no observations (Supplementary Table S2). There was a large variability in the number of observations by stratum, ranging from 1 to 1669, and the percentage of strata with 20 or more observations ranged from 45.0% to 62.5% (Supplementary Table S3).

As shown in Table 1 , the degree of clustering into the intersectional strata (or, analogously, the proportion of variance explained by differences across strata) before including the fixed effects of the variables used to define them (the VPC intercepts-only ) was generally larger for anxiety and depressive symptomatology than for loneliness and life satisfaction. This suggests that the discriminatory accuracy of the variables defining the intersections was generally larger for anxiety and depressive symptomatology. The discriminatory accuracy varied across outcomes when using different SEP indicators, being largest for anxiety and depressive symptomatology when using IMD rank, housing tenure for loneliness, and childhood social class for life satisfaction. PCVs under MCMC (unweighted) were large in all cases (> 90.0%), indicating that the main/additive effects accounted for most of the variability between clusters. PCVs were generally smaller under ML (weighted) due to larger proportions of residual variance between strata (VPC main ), suggesting larger intersectional effects.

Results from the MCMC (unweighted) models using 40 intersectional strata evidenced large inequalities across strata in the predicted values of all outcomes (Supplementary Figure S1). Although most differences were accounted for by the main/additive effects of the variables defining the strata, and all intersectional effects overlapped with zero (no effect), some strata had higher- or lower-than-expected levels (Supplementary Figure S2). Results from the ML (weighted) models (Supplementary Figures S3-S4) were similar, with most of the differences across strata being accounted for by the main effects as indicated by the high PCVs (Table 1 ).

Figure  1 and Fig.  2 provide a ‘socio-demographic mapping’ of the predicted levels in the different mental health outcomes using residential IMD rank as SEP indicator according to the MCMC (unweighted) estimation, evidencing large inequalities across intersectional strata. Fixed (main/additive) and random effects from these MCMC models are included in Supplementary Table S4. Most of the privileged categories (male, heterosexual, socioeconomically advantaged) showed better outcome levels, with large and consistent gaps across sexes, sexual orientations, and cohorts/generations (participants in their 30s showed better results than those a decade younger across all outcomes). Inequalities by IMD rank were comparatively smaller. Black participants generally showed the lowest levels of anxiety and depressive symptomatology and loneliness. This was not the case for life satisfaction, where Black and White participants showed fairly similar results across intersections with other variables, and the lowest levels were observed among those in the “Other” ethnicity group, which were also among the intersections showing the worst mental health outcomes. Using different SEP indicators (Supplementary Figures S5–S6) led to very similar results, although gaps by SEP were typically larger when using housing tenure as indicator. The divide by sexual orientation was consistent across all outcomes, accounting for some of the largest gaps in all outcomes.

Anxiety and depressive symptomatology predicted values of each intersectional stratum using residential Index of Multiple Deprivation (IMD) rank as the indicator of socioeconomic position. Markov Chain Monte Carlo (MCMC) estimation, unweighted results. M male, F female. White includes all White groups; South Asian includes Bangladeshi, Indian, and Pakistani groups; Black includes Black African, Black Caribbean, and Black British groups; Other includes all other ethnic group not included in the other categories

Loneliness and life satisfaction predicted values of each intersectional stratum using residential Index of Multiple Deprivation (IMD) rank as the indicator of socioeconomic position. Markov Chain Monte Carlo (MCMC) estimation, unweighted results. M male, F female. White includes all White groups; South Asian includes Bangladeshi, Indian, and Pakistani groups; Black includes Black African, Black Caribbean, and Black British groups; Other includes all other ethnic group not included in the other categories

The ‘socio-demographic mapping’ of the predicted values at each intersection was more heterogeneous when accounting for survey design and non-response (Supplementary Figures S7-S8). The fixed/main/additive effects from these models (Supplementary Table S5) were, however, largely similar to those from the unweighted models, and sexual orientation was again associated with most of the largest gaps across all stratifications and outcomes. Most differences in fixed effects between weighted and unweighted approaches were found by racial/ethnic group. Being in the “Other” ethnicity group was associated with worse levels in anxiety, whereas those in the Mixed ethnicity group showed worse loneliness and life satisfaction outcomes.

Figure  3 and Fig.  4 show the residual values (intersectional effects) of each intersectional stratum using residential IMD as SEP indicator according to the MCMC (unweighted) estimation (similar plots using the alternative SEP indicators are available in Supplementary Figures S9–S12). All intersectional effects’ CIs overlapped with or were very close to zero (no effect). The only significant intersectional effect corresponded to the loneliness levels of the stratum including White heterosexual males in their 30s owning/part owning a house, which were M residual  = − 0.19 (95% CI − 0.39, − 0.005) lower-than-expected. Some strata at the intersection between privileged and marginalised social identities/positions tended to have worse-than-expected (e.g., South Asian heterosexual males in their 30s living in less deprived areas) or better-than-expected (e.g., South Asian heterosexual males in their teens/20s living in more deprived areas) outcomes.

Anxiety and depressive symptomatology residual values (intersectional effects) and 95% credible intervals of each intersectional stratum using residential Index of Multiple Deprivation (IMD) rank as the indicator of socioeconomic position. Markov Chain Monte Carlo (MCMC) estimation, unweighted results. Strata defined by generation/cohort (first digit: 1 Next Steps/1990, 2 Millennium Cohort Study/2000–2002), birth sex (second digit: 0 Male, 1 Female), ethnicity (third digit: 1 White, 2 Mixed, 3 South Asian, 4 Black, 5 Other), sexual orientation (fourth digit: 0 Heterosexual, 1 Sexual minority), residential IMD rank (fifth digit: 0 More deprived, 1 Less deprived)

Loneliness and life satisfaction residual values (intersectional effects) and 95% credible intervals of each intersectional stratum using residential Index of Multiple Deprivation (IMD) rank as the indicator of socioeconomic position. Markov Chain Monte Carlo (MCMC) estimation, unweighted results. Strata defined by generation/cohort (first digit: 1 Next Steps/1990, 2 Millennium Cohort Study/2000–2002), birth sex (second digit: 0 Male, 1 Female), ethnicity (third digit: 1 White, 2 Mixed, 3 South Asian, 4 Black, 5 Other), sexual orientation (fourth digit: 0 Heterosexual, 1 Sexual minority), residential IMD rank (fifth digit: 0 More deprived, 1 Less deprived)

Larger residual values (intersectional effects) at some intersections were found in the weighted analyses (ML estimation, Supplementary Figures S13–18). Most of the largest intersectional effects corresponded to strata at the intersection of privileged and marginalised social identities/positions. For instance, the largest worse-than-expected levels were found for anxiety among South Asian heterosexual men in their thirties living in less deprived areas ( M residual  = 0.52); for depression among heterosexual men in their 30s from the “Other” ethnicity group living in more deprived areas ( M residual  = 1.10) (Supplementary Figure S13); for loneliness among South Asian heterosexual women in their teens/20s owning a house ( M residual  = 1.05) (Supplementary Figure S16); and for life satisfaction among South Asian men in their thirties from a sexual minority and a disadvantaged social class at childhood ( M residual  = − 1.52) (Supplementary Figure S18).

Comparison with fixed-effects multiple regression approach

Results from the fixed-effects multiple regression approach are included in Supplementary Tables S6-S9. Several interaction terms were statistically significant. In line with the differences across the MCMC unweighted and ML weighted MAIHDA models, the unweighted and weighted regression models’ results varied in some cases, with some interaction terms becoming statistically significant after accounting for the survey and non-response weights, often involving sexual and ethnic minorities. Many of the significant interaction terms found under both approaches were based on very few (down to two) observations and using a specific intersection (White heterosexual males in their 30s in a disadvantaged SEP) as reference.

We aimed to provide a “socio-demographic mapping” [ 14 ] of the mental health inequalities within the UK population during the COVID-19 pandemic from an intersectional perspective, using MAIHDA models. We documented levels of anxiety, depression, loneliness, and life satisfaction across multiple intersecting social identities/positions tied to social power and explored whether there were intersectional effects observable above and beyond the effects associated with any identity/position in isolation. In our first approach, similar to previous MAIHDA applications [ 21 , 25 , 59 , 60 ], we found that, among the study participants, most of the differences across intersectional strata were accounted for by the additive effects of the social identities/positions used to define those intersections. Our second approach aimed to account for the biasing effect of differential non-response to make the results generalisable beyond the study participants. Using this approach, we found even larger inequalities across strata and different-than-expected outcome levels in some intersectional strata, defined in most cases by combinations of privileged and marginalised social identities/positions. Both approaches evidenced the existence of large inequalities in all outcomes. Some of the largest inequalities were observed by sexual orientation, followed by birth sex and cohort/generation, with sexual minorities, females, and younger people (in their teens/20s) showing worse levels. These findings exemplify the multifaceted way in which mental (ill) health inequalities are socially patterned [ 5 ].

From a methodological standpoint, our study showcases some of the desirable features of MAIHDA models to the quantitative analysis of inequalities from an intersectional perspective. All intersections (multiply advantaged and disadvantaged, as well as all combinations in between) were included and voiced in the “socio-demographic mapping” [ 14 ], which prevented reinforcing the idea of reference categories as the “standard” [ 16 , 17 ]. Combinations of privileged and marginalised identities were among those with the largest positive and negative intersectional effects in the two MAIHDA modelling strategies used, highlighting how inequalities are not limited to groups with multiply advantaged or disadvantaged positions, and that they may also be contextually contingent [ 8 , 9 ]. Importantly, the lack of evidence of significant or large intersectional effects, regardless of the quantitative approach used, does not rule out the existence of different intersectional lived experiences [ 25 ], as they may not necessarily reflect upon differences in the outcomes under study. Using MAIHDA models also helped us to further embrace intersectional complexity by acknowledging the existence of heterogeneity not only between but also within intersections [ 11 ]. Discriminatory accuracy levels were similar or larger than those found in most applications of MAIHDA (where VPC intercepts-only or ICC tend to be < 0.05 [ 22 ]), and generally larger for anxiety and depressive symptomatology than for loneliness and life satisfaction. These varied across stratifications using different SEP indicators, suggesting that the experiences attached to these SEP indicators may have different impacts across different outcomes.

From a substantive standpoint, our study covers a gap in the knowledge about population mental health inequalities during the pandemic from an intersectional perspective [ 29 ], and particularly among young adults who, according to previous evidence [ 35 , 37 , 38 , 41 , 42 ], have been most adversely affected by the pandemic. Women, young adults, and those in more disadvantaged socioeconomic positions had worse mental health at the time of data collection (February/March 2021, during the third UK nationwide lockdown). These results exemplify the structural, up-stream, fundamental causes (e.g., sexism, classism, heteronormativity) of inequality, leading to differential exposures to experiences such as discrimination and stigma [ 2 , 5 ]. The mental health inequalities by sexual orientation observed in our study are a grim example of this, extending recent evidence from earlier data collection time points in MCS [ 61 , 62 , 63 ] and showing that these inequalities are large and, in most cases, cut across different mental health outcomes, cohorts/generations, sexes, racial/ethnic groups, and socioeconomic levels. Inequalities by sexual orientation may be explained by the differential exposure to experiences such as reduced peer support availability and increased exposure to discrimination or familial rejection (e.g., increased time spent in family contexts that may have been unsupportive), as well as poorer pre-pandemic health and mental health [ 64 , 65 , 66 , 67 ]. Although disproportionate COVID-19 infection and mortality rates in minoritised racial/ethnic groups have been documented [ 68 ], we did not find consistent evidence of mental health inequalities by racial/ethnic groups. The weighted results suggested that some racial/ethnic groups (particularly the Mixed and “Other” ethnicity groups) had worse levels in multiple outcomes. This goes in line with previous research suggesting larger distress levels during the pandemic in the UK general adult population using similar groups [ 42 ], and adds to the mixed evidence on loneliness, where coarser ethnicity/racial groups (White vs non-White) have been used [ 30 , 69 ]. Estimates of the additive/main effects associated with different racial/ethnic groups were the most variable across the two MAIHDA modelling approaches used (unweighted vs weighted), suggesting a larger bias of non-response in these estimates.

Limitations and future directions

This is, to our knowledge, the first study to document population inequalities in different mental health outcomes during the pandemic using MAIHDA models, with the already mentioned advantages of doing so relative to other more traditional approaches. These results must be interpreted considering several limitations. Despite the diversity in the cohorts, the number of participants from racially/ethnically minoritised and sexual minority groups was small. This had multiple implications for our study. We had to group some of the least frequent categories (e.g., sexual minorities and ethnic groups), lumping together people with different experiences, perspectives, histories, cultures, and complexity in relation to experiences of marginalisation and oppression, thus obscuring (and increasing) the sources of heterogeneity within intersections. Even after grouping those categories, some of the intersections had none or very few observations, which prevented us from mapping the missing intersections and likely limited our ability to detect intersectional effects at some of these intersections which may be at risk. The small sample size at some intersections may also explain some of the differences across the MAIHDA models and the multiple regression fixed-effects models: MAIHDA models introduce a correction (shrinkage) to adjust the estimates of intersections by their precision (based on their size) [ 22 ]. This has been documented to result in smaller number of statistically significant intersectional effects compared to fixed-effects approaches [ 21 , 25 ], which do not include this correction thus potentially resulting in significant interaction effects based on very few observations. Surveys designed to ensure sufficient sizes at all intersections to be studied are needed to overcome these limitations [ 12 , 18 ].

Second, the small number of indicators in our outcome measures contributed towards measurement error, thus artificially increasing their heterogeneity. This also prevented us from exploring the equivalence of the measures across the intersections under study. Future research using longer versions of these and other instruments may result in more reliable/accurate outcome measurements, while also enabling testing measurement equivalence using suitable methods [ 70 ].

Third, due to differential non-response across groups [ 47 ], the results from the MCMC analyses, which permit assessing the statistical significance of the intersectional effects, may only be generalisable to the study participants. Since weighted analyses have not yet been implemented for MCMC MAIHDA models, we tried to overcome this limitation by re-estimating the MAIHDA models with ML using survey and non-response weights, at the cost of not obtaining confidence intervals for the intersectional effects. Both approaches resulted in remarkably similar main/additive effects, but both discriminatory accuracy and intersectional effects were generally larger in the weighted results. Aside the obvious need for implementation of weighted analysis in standard MAIHDA models, boostrapping conditioned on clusters defined by intersections may be a potential solution to obtain confidence intervals for the intersectional effects when using weighted ML, but methodological work beyond the scope of this paper, including formal simulations, is needed to test this approach.

Fourth, the “socio-demographic mapping” provided is only applicable to the social identities/positions under study: we were, for instance, unable to examine mental health of transgender and gender diverse groups despite evidence suggesting they were also disproportionately adversely affected by the pandemic [ 71 , 72 ].

Finally, the cross-sectional design provides a snapshot of the inequalities at one time-point, coinciding with a lockdown period. This may not be generalisable to other pandemic periods, as longitudinal UK-based evidence shows that levels of different mental health measures changed over the pandemic course [ 30 , 31 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 ]. Future studies may cover this gap by extending the MAIHDA modelling approach to longitudinal designs.

Conclusions

We have illustrated how quantitative methods can be used to study population intersectional mental health inequalities. Our study evidences large mental health inequalities across (and within) intersectional strata in the population. Large proportions of these inequalities can be accounted for by the main/additive effects of the variables used to define those intersections (cohort/generation, birth sex, racial/ethnic group, sexual orientation, and SEP), with particularly large inequalities by sexual orientation across all studied outcomes. Our analyses also suggest that some of those inequalities were not strictly equivalent across all intersections and support the notion (and the importance of acknowledging) that inequalities are not limited to groups with multiply advantaged or disadvantaged identities/positions. The large gaps found by sexual orientation support and extend existing evidence that sexual minority groups were disproportionately affected by the pandemic. Interventions to provide support, along with further research aimed at understanding intersectional experiences of discrimination across different racial/ethnic groups and socioeconomic levels, are crucial.

Data availability

Deidentified data and documentation on Next Steps [SN 2000030] and Millennium Cohort Study [SN 2000031] are available from the UK Data Service: https://ukdataservice.ac.uk/ .

Evans CR (2019) Modeling the intersectionality of processes in the social production of health inequalities. Soc Sci Med 226:249–253. https://doi.org/10.1016/j.socscimed.2019.01.017

Article   PubMed   Google Scholar  

Hatzenbuehler ML, Phelan JC, Link BG (2013) Stigma as a fundamental cause of population health inequalities. Am J Public Health 103(5):813–821. https://doi.org/10.2105/AJPH.2012.301069

Article   PubMed   PubMed Central   Google Scholar  

May VM (2015) Pursuing intersectionality unsettling dominant imaginaries. Contemporary sociological perspectives. Routledge, Taylor & Francis Group, New York

Google Scholar  

Link BG, Garcia SJ (2021) Diversions: how the underrepresentation of research on advantaged groups leaves explanations for health inequalities incomplete. J Health Soc Behav 62(3):334–349. https://doi.org/10.1177/00221465211028152

Link BG, Phelan J (1995) Social conditions as fundamental causes of disease. J Health Soc Behav Spec No:80–94

Article   CAS   PubMed   Google Scholar  

Tuck E (2009) Suspending damage: a letter to communities. Harv Educ Rev 79(3):409–428. https://doi.org/10.17763/haer.79.3.n0016675661t3n15

Article   Google Scholar  

Butler J (2016) Rethinking vulnerability and resistance. In: Butler J, Gambetti Z, Sabsay L (eds) Vulnerability in resistance. Duke University Press, Durham, pp 12–27

Chapter   Google Scholar  

McLeod JD (2015) Why and how inequality matters. J Health Soc Behav 56(2):149–165. https://doi.org/10.1177/0022146515581619

Rhead RD, Woodhead C, Ahmad G, Das-Munshi J, McManus S, Hatch SL (2022) A comparison of single and intersectional social identities associated with discrimination and mental health service use: data from the 2014 adult psychiatric morbidity survey in England. Soc Psychiatr Psychiatr Epidemiol. https://doi.org/10.1007/s00127-022-02259-1

McLeod JD, Lively KJ (2007) Social psychology and stress research. In: Avison WR, McLeod JD, Pescosolido BA (eds) Mental health, social mirror. Springer, Boston, pp 275–303. https://doi.org/10.1007/978-0-387-36320-2_12

Hancock A-M (2013) Empirical intersectionality: a tale of two approaches. UC Irvine Law Review 3:259–296

Bowleg L (2008) When Black + Lesbian + Woman ≠ Black Lesbian Woman: the methodological challenges of qualitative and quantitative intersectionality research. Sex Roles 59(5):312–325. https://doi.org/10.1007/s11199-008-9400-z

Bauer GR, Scheim AI (2019) Methods for analytic intercategorical intersectionality in quantitative research: discrimination as a mediator of health inequalities. Soc Sci Med 226:236–245. https://doi.org/10.1016/j.socscimed.2018.12.015

Bauer GR (2014) Incorporating intersectionality theory into population health research methodology: challenges and the potential to advance health equity. Soc Sci Med 110:10–17. https://doi.org/10.1016/j.socscimed.2014.03.022

McCall L (2005) The complexity of intersectionality. Signs 30(3):1771–1800

Choo HY, Ferree MM (2010) Practicing intersectionality in sociological research: a critical analysis of inclusions, interactions, and institutions in the study of inequalities. Sociol Theory 28(2):129–149. https://doi.org/10.1111/j.1467-9558.2010.01370.x

Evans CR, Williams DR, Onnela JP, Subramanian SV (2018) A multilevel approach to modeling health inequalities at the intersection of multiple social identities. Soc Sci Med 203:64–73. https://doi.org/10.1016/j.socscimed.2017.11.011

Bowleg L, Bauer G (2016) Invited reflection: quantifying intersectionality. Psychol Women Q 40(3):337–341. https://doi.org/10.1177/0361684316654282

Bauer GR, Churchill SM, Mahendran M, Walwyn C, Lizotte D, Villa-Rueda AA (2021) Intersectionality in quantitative research: a systematic review of its emergence and applications of theory and methods. SSM Popul Health 14:100798. https://doi.org/10.1016/j.ssmph.2021.100798

Bauer GR, Scheim AI (2019) Advancing quantitative intersectionality research methods: Intracategorical and intercategorical approaches to shared and differential constructs. Soc Sci Med 226:260–262. https://doi.org/10.1016/j.socscimed.2019.03.018

Evans CR (2019) Adding interactions to models of intersectional health inequalities: comparing multilevel and conventional methods. Soc Sci Med 221:95–105. https://doi.org/10.1016/j.socscimed.2018.11.036

Evans CR, Leckie G, Merlo J (2020) Multilevel versus single-level regression for the analysis of multilevel information: the case of quantitative intersectional analysis. Soc Sci Med 245:112499. https://doi.org/10.1016/j.socscimed.2019.112499

Merlo J (2018) Multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) within an intersectional framework. Soc Sci Med 203:74–80. https://doi.org/10.1016/j.socscimed.2017.12.026

Bauer GR, Mahendran M, Walwyn C, Shokoohi M (2022) Latent variable and clustering methods in intersectionality research: systematic review of methods applications. Soc Psychiatr Psychiatr Epidemiol 57(2):221–237. https://doi.org/10.1007/s00127-021-02195-6

Evans CR, Erickson N (2019) Intersectionality and depression in adolescence and early adulthood: a MAIHDA analysis of the national longitudinal study of adolescent to adult health, 1995–2008. Soc Sci Med 220:1–11. https://doi.org/10.1016/j.socscimed.2018.10.019

Glymour MM, Rudolph KE (2016) Causal inference challenges in social epidemiology: Bias, specificity, and imagination. Soc Sci Med 166:258–265. https://doi.org/10.1016/j.socscimed.2016.07.045

Marmot M, Allen J, Goldblatt P, Herd E, Morrison J (2020) Build back fairer: the COVID-19 Marmot review the pandemic, socioeconomic and health inequalities in England. Institute of Health Equity, London

Bambra C, Riordan R, Ford J, Matthews F (2020) The COVID-19 pandemic and health inequalities. J Epidemiol Commun Health 74(11):964–968. https://doi.org/10.1136/jech-2020-214401

Gibson B, Schneider J, Talamonti D, Forshaw M (2021) The Impact of inequality on mental health outcomes during the COVID-19 pandemic: a systematic review. Can Psychol-Psychologie Canadienne 62(1):101–126. https://doi.org/10.1037/cap0000272

Bu F, Steptoe A, Fancourt D (2020) Who is lonely in lockdown? cross-cohort analyses of predictors of loneliness before and during the COVID-19 pandemic. Public Health 186:31–34. https://doi.org/10.1016/j.puhe.2020.06.036

Creese B, Khan Z, Henley W, O’Dwyer S, Corbett A, Vasconcelos Da Silva M, Mills K, Wright N, Testad I, Aarsland D, Ballard C (2021) Loneliness, physical activity, and mental health during COVID-19: a longitudinal analysis of depression and anxiety in adults over the age of 50 between 2015 and 2020. Int Psychogeriatr 33(5):505–514. https://doi.org/10.1017/S1041610220004135

Creswell C, Shum A, Pearcey S, Skripkauskaite S, Patalay P, Waite P (2021) Young people’s mental health during the COVID-19 pandemic. Lancet Child Adolesc Health 5(8):535–537. https://doi.org/10.1016/S2352-4642(21)00177-2

Article   CAS   PubMed   PubMed Central   Google Scholar  

Ellwardt L, Prag P (2021) Heterogeneous mental health development during the COVID-19 pandemic in the United Kingdom. Sci Rep 11(1):15958. https://doi.org/10.1038/s41598-021-95490-w

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Fancourt D, Steptoe A, Bu F (2021) Trajectories of anxiety and depressive symptoms during enforced isolation due to COVID-19 in England: a longitudinal observational study. Lancet Psychiatr 8(2):141–149. https://doi.org/10.1016/S2215-0366(20)30482-X

Kwong ASF, Pearson RM, Adams MJ, Northstone K, Tilling K, Smith D, Fawns-Ritchie C, Bould H, Warne N, Zammit S, Gunnell DJ, Moran PA, Micali N, Reichenberg A, Hickman M, Rai D, Haworth S, Campbell A, Altschul D, Flaig R, McIntosh AM, Lawlor DA, Porteous D, Timpson NJ (2021) Mental health before and during the COVID-19 pandemic in two longitudinal UK population cohorts. Br J Psychiatr 218(6):334–343. https://doi.org/10.1192/bjp.2020.242

Article   CAS   Google Scholar  

Moreno-Agostino D, Fisher HL, Goodman A, Hatch SL, Morgan C, Richards M, Das-Munshi J, Ploubidis GB. Long-term psychological distress trajectories and the COVID-19 pandemic in three British birth cohorts: a multi-cohort study. PLOS Medicine. https://doi.org/10.1371/journal.pmed.1004145 (in press)

Moreno-Agostino D, Fisher HL, Hatch SL, Morgan C, Ploubidis GB, Das-Munshi J (2022) Generational, sex, and socioeconomic inequalities in mental and social wellbeing during the COVID-19 pandemic: prospective longitudinal observational study of five UK cohorts. Psychological Medicine. https://doi.org/10.1017/S0033291722003348

Niedzwiedz CL, Green MJ, Benzeval M, Campbell D, Craig P, Demou E, Leyland A, Pearce A, Thomson R, Whitley E, Katikireddi SV (2021) Mental health and health behaviours before and during the initial phase of the COVID-19 lockdown: longitudinal analyses of the UK household longitudinal study. J Epidemiol Commun Health 75(3):224–231. https://doi.org/10.1136/jech-2020-215060

O’Connor RC, Wetherall K, Cleare S, McClelland H, Melson AJ, Niedzwiedz CL, O’Carroll RE, O’Connor DB, Platt S, Scowcroft E, Watson B, Zortea T, Ferguson E, Robb KA (2021) Mental health and well-being during the COVID-19 pandemic: longitudinal analyses of adults in the UK COVID-19 mental health & wellbeing study. Br J Psychiatr 218(6):326–333. https://doi.org/10.1192/bjp.2020.212

Patel K, Robertson E, Kwong ASF, Griffith GJ, Willan K, Green MJ, Di Gessa G, Huggins CF, McElroy E, Thompson EJ, Maddock J, Niedzwiedz CL, Henderson M, Richards M, Steptoe A, Ploubidis GB, Moltrecht B, Booth C, Fitzsimons E, Silverwood R, Patalay P, Porteous D, Katikireddi SV (2022) Psychological distress before and during the COVID-19 pandemic among adults in the United Kingdom based on coordinated analyses of 11 longitudinal studies. JAMA Netw Open 5(4):e227629. https://doi.org/10.1001/jamanetworkopen.2022.7629

Pierce M, Hope H, Ford T, Hatch S, Hotopf M, John A, Kontopantelis E, Webb R, Wessely S, McManus S, Abel KM (2020) Mental health before and during the COVID-19 pandemic: a longitudinal probability sample survey of the UK population. Lancet Psychiatr 7(10):883–892. https://doi.org/10.1016/S2215-0366(20)30308-4

Pierce M, McManus S, Hope H, Hotopf M, Ford T, Hatch SL, John A, Kontopantelis E, Webb RT, Wessely S, Abel KM (2021) Mental health responses to the COVID-19 pandemic: a latent class trajectory analysis using longitudinal UK data. Lancet Psychiatr 8(7):610–619. https://doi.org/10.1016/S2215-0366(21)00151-6

Zaninotto P, Iob E, Demakakos P, Steptoe A (2021) Immediate and longer-term changes in the mental health and well-being of older adults in England during the COVID-19 pandemic. JAMA Psychiat. https://doi.org/10.1001/jamapsychiatry.2021.3749

Calderwood L, Sánchez C (2016) Next steps (formerly known as the longitudinal study of young people in England). Open Health Data. https://doi.org/10.5334/ohd.16

Connelly R, Platt L (2014) Cohort profile: UK millennium cohort study (MCS). Int J Epidemiol 43(6):1719–1725. https://doi.org/10.1093/ije/dyu001

Institute for Government (2021) Timeline of UK government coronavirus lockdowns. https://www.instituteforgovernment.org.uk/charts/uk-government-coronavirus-lockdowns . Accessed 4 Nov 2021

Brown M, Goodman A, Peters A, Ploubidis GB, Sanchez A, Silverwood R, Smith K (2021) COVID-19 survey in five national longitudinal studies: Waves 1, 2 and 3 User Guide (Version 3). UCL Centre for Longitudinal Studies and MRC Unit for Lifelong Health and Ageing, London

Kroenke K, Spitzer RL, Williams JB, Monahan PO, Lowe B (2007) Anxiety disorders in primary care: prevalence, impairment, comorbidity, and detection. Ann Intern Med 146(5):317–325. https://doi.org/10.7326/0003-4819-146-5-200703060-00004

Kroenke K, Spitzer RL, Williams JB (2003) The patient health questionnaire-2: validity of a two-item depression screener. Med Care 41(11):1284–1292. https://doi.org/10.1097/01.MLR.0000093487.78664.3C

Hughes ME, Waite LJ, Hawkley LC, Cacioppo JT (2004) A short scale for measuring loneliness in large surveys: results from two population-based studies. Res Aging 26(6):655–672. https://doi.org/10.1177/0164027504268574

Office for National Statistics (2018) Personal well-being user guidance. https://www.ons.gov.uk/peoplepopulationandcommunity/wellbeing/methodologies/personalwellbeingsurveyuserguide . Accessed 28 Sep 2021

Office for National Statistics (2022) Ethnic group, national identity and religion. https://www.ons.gov.uk/methodology/classificationsandstandards/measuringequality/ethnicgroupnationalidentityandreligion . Accessed 02 Feb 2022

Noble S, McLennan D, Noble M, Plunkett E, Gutacker N, Silk M, Wright G (2019) The English indices of deprivation 2019. Research report. Ministry of Housing, Communities and Local Government, London

Dodgeon B, Morris T, Crawford C, Parsons S, Vignoles A, Oldfield Z, O’Neill D (2011) CLOSER work package 2: Harmonised socio-economic measures user guide (revised). CLOSER, London

Browne WJ (2012) MCMC estimation in MLwiN, v2.26. Centre for multilevel modelling. University of Bristol, Bristol

Rasbash J, Charlton C, Browne WJ, Healy M, Cameron B (2009) MLwiN Version 2.1. Centre for multilevel modelling. University of Bristol, Bristol

Leckie G, Charlton C (2013) runmlwin: a program to run the MLwiN multilevel modeling software from within stata. J Stat Softw 52:1–40

StataCorp (2021) Stata statistical software: release 17. StataCorp LLC, College Station

Axelsson Fisk S, Mulinari S, Wemrell M, Leckie G, Perez Vicente R, Merlo J (2018) Chronic obstructive pulmonary disease in Sweden: an intersectional multilevel analysis of individual heterogeneity and discriminatory accuracy. SSM Popul Health 4:334–346. https://doi.org/10.1016/j.ssmph.2018.03.005

Persmark A, Wemrell M, Zettermark S, Leckie G, Subramanian SV, Merlo J (2019) Precision public health: Mapping socioeconomic disparities in opioid dispensations at Swedish pharmacies by multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA). PLoS ONE 14(8):e0220322. https://doi.org/10.1371/journal.pone.0220322

Becares L, Kneale D (2022) Inequalities in mental health, self-rated health, and social support among sexual minority young adults during the COVID-19 pandemic: analyses from the UK millennium cohort study. Soc Psychiatr Psychiatr Epidemiol. https://doi.org/10.1007/s00127-022-02291-1

Amos R, Manalastas EJ, White R, Bos H, Patalay P (2020) Mental health, social adversity, and health-related outcomes in sexual minority adolescents: a contemporary national cohort study. Lancet Child Adolesc Health 4(1):36–45. https://doi.org/10.1016/S2352-4642(19)30339-6

Patalay P, Fitzsimons E (2021) Psychological distress, self-harm and attempted suicide in UK 17-year olds: prevalence and sociodemographic inequalities. Br J Psychiatry 219(2):437–439. https://doi.org/10.1192/bjp.2020.258

Ormiston CK, Williams F (2022) LGBTQ youth mental health during COVID-19: unmet needs in public health and policy. Lancet 399(10324):501–503. https://doi.org/10.1016/S0140-6736(21)02872-5

Salerno JP, Devadas J, Pease M, Nketia B, Fish JN (2020) Sexual and gender minority stress amid the COVID-19 pandemic: implications for LGBTQ young persons’ mental health and well-being. Public Health Rep 135(6):721–727. https://doi.org/10.1177/0033354920954511

Salerno JP, Doan L, Sayer LC, Drotning KJ, Rinderknecht RG, Fish JN (2021) Changes in mental health and well-being are associated with living arrangements with parents during COVID-19 among sexual minority young persons in the US. Psychol Sex Orientat Gender Divers. https://doi.org/10.1037/sgd0000520

Fish JN, McInroy LB, Paceley MS, Williams ND, Henderson S, Levine DS, Edsall RN (2020) “I’m kinda stuck at home with unsupportive parents right now”: LGBTQ youths’ experiences with COVID-19 and the importance of online support. J Adolesc Health 67(3):450–452. https://doi.org/10.1016/j.jadohealth.2020.06.002

Katikireddi SV, Hainey KJ, Beale S (2021) The impact of COVID-19 on different population subgroups: ethnic, gender and age-related disadvantage. J R Coll Physicians Edinb 51(S1):S40–S46. https://doi.org/10.4997/JRCPE.2021.240

Bu F, Steptoe A, Fancourt D (2020) Loneliness during a strict lockdown: Trajectories and predictors during the COVID-19 pandemic in 38,217 United Kingdom adults. Soc Sci Med 265:113521. https://doi.org/10.1016/j.socscimed.2020.113521

Muthen B, Asparouhov T (2018) Recent methods for the study of measurement invariance with many groups: alignment and random effects. Sociol Methods Res 47(4):637–664. https://doi.org/10.1177/0049124117701488

Article   MathSciNet   Google Scholar  

Hawke LD, Hayes E, Darnay K, Henderson J (2021) Mental health among transgender and gender diverse youth: an exploration of effects during the COVID-19 pandemic. Psychol Sex Orientat Gend Divers 8(2):180–187. https://doi.org/10.1037/sgd0000467

Wang Y, Pan B, Liu Y, Wilson A, Ou J, Chen R (2020) Health care and mental health challenges for transgender individuals during the COVID-19 pandemic. Lancet Diabetes Endocrinol 8(7):564–565. https://doi.org/10.1016/S2213-8587(20)30182-0

Download references

Acknowledgements

We would like to thank all individuals who participated in the two cohort studies for so generously giving up their time over so many years, and all the study team members for their tremendous efforts in collecting and managing the data. We would also like to thank Dr Clare R. Evans for her input, for providing very helpful code to run the analyses used in this study, and for her inspiring work; and Dr Annie Irvine, Dr Rochelle A. Burgess, Dr Dörte Bemme, Dr Dominique Behague, and the anonymous reviewers for their feedback and recommendations to improve the manuscript.

This paper represents independent research part supported by the ESRC Centre for Society and Mental Health at King’s College London [ES/S012567/1]. DM, CW, GBP, and JD are part supported by the ESRC Centre for Society and Mental Health at King's College London [ES/S012567/1]. JD is also supported by the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London and the NIHR Applied Research Collaboration South London (NIHR ARC South London) at King's College Hospital NHS Foundation Trust. The views expressed are those of the authors and not necessarily those of the ESRC, NIHR, the Department of Health and Social Care, or King’s College London. Next Steps and the Millennium Cohort Study are supported by the Centre for Longitudinal Studies, Resource Centre 2015-20 grant [ES/M001660/1] and a host of other co-funders. The COVID-19 data collections were funded by the UKRI grant Understanding the economic, social and health impacts of COVID-19 using lifetime data: evidence from 5 nationally representative UK cohorts [ES/V012789/1].

Author information

George B. Ploubidis and Jayati Das-Munshi are joint senior authors.

Authors and Affiliations

Centre for Longitudinal Studies, UCL Social Research Institute, University College London, 55-59 Gordon Square, London, WC1H 0NU, UK

Darío Moreno-Agostino & George B. Ploubidis

ESRC Centre for Society and Mental Health, King’s College London, Melbourne House, 44-46 Aldwych, London, WC2B 4LL, UK

Darío Moreno-Agostino, Charlotte Woodhead, George B. Ploubidis & Jayati Das-Munshi

Department of Psychological Medicine, King’s College London, Institute of Psychiatry, Psychology & Neuroscience, 16 De Crespigny Park, London, SE5 8AF, UK

Charlotte Woodhead & Jayati Das-Munshi

South London and Maudsley NHS Trust, London, UK

Jayati Das-Munshi

You can also search for this author in PubMed   Google Scholar

Contributions

D.M. conceived the study and carried out the analyses. D.M. and C.W. prepared the first draft. G.B.P. and J.D. supervised the project and provided critical feedback. All authors reviewed and contributed to the final manuscript.

Corresponding author

Correspondence to Darío Moreno-Agostino .

Ethics declarations

Conflict of interest.

None to declare.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (PDF 2046 KB)

Rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Moreno-Agostino, D., Woodhead, C., Ploubidis, G.B. et al. A quantitative approach to the intersectional study of mental health inequalities during the COVID-19 pandemic in UK young adults. Soc Psychiatry Psychiatr Epidemiol 59 , 417–429 (2024). https://doi.org/10.1007/s00127-023-02424-0

Download citation

Received : 13 September 2022

Accepted : 12 January 2023

Published : 24 January 2023

Issue Date : March 2024

DOI : https://doi.org/10.1007/s00127-023-02424-0

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Multilevel modelling
• Intersectionality
• Multilevel analysis of individual heterogeneity and discriminatory accuracy
• Intercategorical complexity
• Find a journal
• Publish with us
• Track your research

• Open access
• Published: 22 June 2023

Social media and mental health in students: a cross-sectional study during the Covid-19 pandemic

• Abouzar Nazari   ORCID: orcid.org/0000-0003-2155-5438 1 ,
• Maede Hosseinnia   ORCID: orcid.org/0000-0002-2248-7011 2 ,
• Samaneh Torkian 3 &
• Gholamreza Garmaroudi   ORCID: orcid.org/0000-0001-7449-227X 4  

BMC Psychiatry volume  23 , Article number:  458 ( 2023 ) Cite this article

69k Accesses

5 Citations

17 Altmetric

Metrics details

Social media causes increased use and problems due to their attractions. Hence, it can affect mental health, especially in students. The present study was conducted with the aim of determining the relationship between the use of social media and the mental health of students.

Materials and methods

The current cross-sectional study was conducted in 2021 on 781 university students in Lorestan province, who were selected by the Convenience Sampling method. The data was collected using a questionnaire on demographic characteristics, social media, problematic use of social media, and mental health (DASS-21). Data were analyzed in SPSS-26 software.

Shows that marital status, major, and household income are significantly associated with lower DASS21 scores (a lower DASS21 score means better mental health status). Also, problematic use of social media (β = 3.54, 95% CI: (3.23, 3.85)) was significantly associated with higher mental health scores (a higher DASS21 score means worse mental health status). Income and social media use (β = 1.02, 95% CI: 0.78, 1.25) were significantly associated with higher DASS21 scores (a higher DASS21 score means worse mental health status). Major was significantly associated with lower DASS21 scores (a lower DASS21 score means better mental health status).

This study indicated that social media had a direct relationship with mental health. Despite the large amount of evidence suggesting that social media harms mental health, more research is still necessary to determine the cause and how social media can be used without harmful effects.

Peer Review reports

• Social media

Social media is one of the newest and most popular internet services, which has caused significant progress in the social systems of different countries in recent years [ 1 , 2 ]. The use of the Internet has become popular among people in such a way that its use has become inevitable and has made life difficult for those who use it excessively [ 3 ]. Social media has attracted the attention of millions of users around the world owing to the possibility of fast communication, access to a large amount of information, and its widespread dissemination [ 4 ]. Facebook, WhatsApp, Instagram, and Twitter are the most popular media that have attractive and diverse spaces for online communication among users, especially the young generation [ 5 , 6 ].

According to studies, at least 55% of the world’s population used social media in 2022 [ 7 ]. Iranian statistics also indicate that 78.5% of people use at least one social media. WhatsApp, with 71.1% of users, Instagram, with 49.4%, and Telegram, with 31.6% are the most popular social media among Iranians [ 8 , 9 ].

The use of social media has increased significantly in all age groups due to the origin of the COVID-19 pandemic [ 10 ] .It affected younger people, especially students, due to educational and other purposes [ 11 , 12 ]. Because of the sudden onset of the COVID-19 pandemic, educational institutions and learners had to accept e-learning as the only sustainable education option [ 13 ]. The rapid migration to E-learning has brought several challenges that can have both positive and negative consequences [ 14 ].

Unlike traditional media, where users are passive, social media enables people to create and share content; hence, they have become popular tools for social interaction [ 15 ].The freedom to choose to participate in the company of friends, anonymity, moderation, encouragement, the free exchange of feelings, and network interactions without physical presence and the constraints of the real world are some of the most significant factors that influence users’ continued activity in social media [ 16 ]. In social media, people can interact, maintain relationships, make new friends, and find out more about the people they know offline [ 17 ]. However, this popularity has resulted in significant lifestyle changes, as well as intentional or unintentional changes in various aspects of human social life [ 18 ]. Despite many advantages, the high use of social media brings negative physical, psychological, and social problems and consequences [ 19 ], but despite the use and access of more people to the Internet, its consequences and crises have been ignored [ 20 ].

Use of social media and mental health

Spending too much time on social media can easily become problematic [ 21 ]. Excessive use of social media, called problematic use, has symptoms similar to addiction [ 22 , 23 ]. Problematic use of social media represents a non-drug-related disorder in which harmful effects emerge due to preoccupation and compulsion to over-participate in social media platforms despite its highly negative consequences [ 24 , 25 , 26 ], which leads to adverse consequences of mental health, including anxiety, depression, lower well-being, and lower self-esteem [ 27 , 28 , 29 ].

Mental health & use of social media

Mental health is the main pillar of healthy human societies, which plays a vital role in ensuring the dynamism and efficiency of any society in such a way that other parts of health cannot be achieved without mental health [ 30 ]. According to World Health Organization’s (WHO) definition, mental health refers to a person’s ability to communicate with others [ 31 ]. Some researchers believe that social relationships can significantly affect mental health and improve quality of life by creating a sense of belonging and social identity [ 32 ]. It is also reported that people with higher social interactions have higher physical and mental health [ 33 ].

Scientific evidence also shows that social media affect people’s mental health [ 34 ]. Social studies and critiques often emphasize the investigation of the negative effects of Internet use [ 35 ]. For example, Kim et al. studied 1573 participants aged 18–64 years and reported that Internet addiction and social media use were associated with higher levels of depression and suicidal thoughts [ 36 ]. Zadar also studied adults and reported that excessive use of social media and the Internet was correlated with stress, sleep disturbances, and personality disorders [ 37 ]. Richards et al. reported the negative effects of the Internet and social media on the health and quality of life of adolescents [ 38 ]. There have been numerous studies that examine Internet addiction and its associated problems in young people [ 39 , 40 ], as well as reports of the effects of social media use on young people’s mental health [ 41 , 42 ].

A study on Iranian students showed that social media leads to depression, anxiety, and mental health decline [ 25 ]. A study on Iranian students showed that social media leads to depression, anxiety, and mental health decline [ 25 ]. But no study has investigated the effects of social media on the mental health of students from a more traditional province with lower individualism and higher levels of social support (where they were thought to have lower social media use and better mental health) during the COVID-19 pandemic. As social media became more and more vital to university students’ social lives during the lockdowns, students were likely at increased risk of social media addiction, which could harm their mental health. University students depended more on social media due to the limitations of face-to-face interactions. In addition, previous studies were conducted exclusively on students in specific fields. However, in our study, all fields, including medical and non-medical science fields were investigated.

The present study was conducted to determine the relationship between the use of social media and mental health in students in Lorestan Province during the COVID-19 pandemic.

Study design and participants

The current study was descriptive-analytical, cross-sectional, and conducted from February to March 2022 with a statistical population made up of students in all academic grades at universities in Lorestan Province (19 scientific and academic centers, including centers under the supervision of the Ministry of Health and the Ministry of Science).

Sample size

According to the convenience sampling method, 781 people were chosen as participants in the present study. During the sampling, a questionnaire was created and uploaded virtually on Porsline’s website, and then the questionnaire link was shared in educational and academic groups on social media for students to complete the questionnaire under inclusion criteria (being a student at the University of Lorestan and consenting to participate in the study).

The research tools included the demographic information questionnaire, the standard social media use questionnaire, and the mental health questionnaire.

Demographic information

The demographic information age, gender, ethnicity, province of residence, urban or rural, place of residence, semester, and the field of study, marital status, household income, education level, and employment status were recorded.

Psychological assessment

The students were subjected to the Persian version of the Depression Anxiety Stress Scale (DASS21). It consists of three self-report scales designed to measure different emotional states. DASS21 questions were adjusted according to their importance and the culture of Iranian students. The DASS21 scale was scored on a four-point scale to assess the extent to which participants experienced each condition over the past few weeks. The scoring method was such that each question was scored from 0 (never) to 3 (very high). Samani (2008) found that the questionnaire has a validity of 0.77 and a Cronbach’s alpha of 0.82 [ 43 ].

Use of social media questionnaire

Among the 13 questions on social media use in the questionnaire, seven were asked on a Likert scale (never, sometimes, often, almost, and always) that examined the problematic use of social media, and six were asked about how much time users spend on social media. Because some items were related to the type of social media platform, which is not available today, and users now use newer social media platforms such as WhatsApp and Instagram, the questionnaires were modified by experts and fundamentally changed, and a 22-item questionnaire was obtained that covered the frequency of using social media. Cronbach’s alpha was equal to 0.705 for the first part, 0.794 for the second part, and 0.830 for all questions [ 44 ]. Considering the importance of the problematic use of the social media, six questions about the problematic use were measured separately.

To confirm the validity of the questionnaire, a panel of experts with CVR 0.49 and CVI 0.70 was used. Its reliability was also obtained (0.784) using Cronbach’s alpha coefficient. Finally, the questionnaire was tested in a class with 30 students to check the level of difficulty and comprehension of the questionnaire. Finally, a 22-item questionnaire was obtained, of which six items were about the problematic use of social media and the remaining 16 questions were about the rate and frequency of using social media. Cronbach’s alpha was 0.705 for the first part, including questions about the problematic use of the social media, and 0.794 for the second part, including questions about the rate and frequency of using the social media. The total Cronbach’s alpha for all questions was 0.830. Six questions about the problematic use of social media were measured separately due to the importance of the problematic use of social media. Also, a separate score was considered for each question. The scores of these six questions on the problematic use of the social media were summed, and a single score was obtained for analysis.

Statistical analysis

Data were analyzed using the Statistical Package for Social Sciences (SPSS) version 26.0 (SPSS Inc., Chicago, IL, USA). The normal distribution of continuous variables was analyzed using the Kolmogorov-Smirnov test, histogram, and P-P diagram, which showed that they are not normally distributed. Descriptive statistics were calculated for all variables. Comparison between groups was done using Mann-Whitney and Kruskal-Wallis non-parametric tests. Multiple linear regression analysis was used to investigate the relationship between mental health, problematic use of social media, and social media use (The result of merging the Frequency of using social media and Time to use social media). Generalized Linear Models (GLM) were used to assess the association between mental health with the use of social media and problematic use of social media. Due to the high correlation (r = 0.585, p = < 0.001) between the use of social media and problematic use of social media, collinearity, we run two separate GLM models. Regression coefficients (β) and adjusted β (β*) with 95% CI and P-value were reported.

A total of 781 participants completed the questionnaires, of which 64.4% were women and 71.3% were single. The minimum age of the participants was 17 years, the maximum age was 45 years, and about half of them (48.9%) were between 21 and 25 years old. A total of 53.4% of the participants had bachelor’s degrees. The income level of 23.2% of participants was less than five million Tomans (the currency of Iran), and 69.7% of the participants were unemployed. 88.1% were living with their families and 70.8% were studying in non-medical fields. 86% of the participants lived in the city, and 58.9% were in their fourth semester or higher. Considering that the research was conducted in a Lorish Province, 43.8% of participants were from the Lorish ethnicity.

The mean total score of mental health was 12.30 with a standard deviation of 30.38, and the mean total score of social media was 14.5557 with a standard deviation of 7.74140.

Table  1 presents a comparison of the mean problematic use of social media and mental health with demographic variables. Considering the non-normality of the hypothesis H0, to compare the means of the independent variables, Mann-Whitney non-parametric tests (for the variables of gender, the field of study, academic semester, employment status, province of residence, and whether it is rural or urban) and Kruskal Wallis (for the variables age, ethnicity, level of education, household income and marital status). According to the obtained results, it was found that the score of problematic use of social media is significantly higher in women, the age group less than 20 years, unemployed, non-native students, dormitory students, and students living with friends or alone, Fars students, students with a household income level of fewer than 7 million Tomans(Iranian currency), and single, divorced, and widowed students were higher than the other groups(P < 0.05).

By comparing the mean score of mental health with demographic variables using non-parametric Mann-Whitney and Kruskal Wallis tests, it was found that there is a significant difference between the variable of poor mental health and all demographic variables (except for the semester variable), residence status (rural or urban) and education level. (There was a significant relationship (P < 0.05). In such a way that the mental health condition was worse in women, age group less than 20 years old, non-medical science, unemployed, non-native, and dormitory students. Also, Fars students, divorced, widowed, and students with a household income of fewer than 5 million Tomans (Iranian currency) showed poorer mental health status. (Table  1 ).

The final model shows that marital status, field, and household income were significantly associated with lower DASS21 scores (a lower DASS21 score means better mental health status). Being single (β* = -23.03, 95% CI: (-33.10, -12.96), being married (β* = -38.78, 95% CI: -51.23, -26.33), was in Medical sciences fields (β* = -8.15, 95% CI: -11.37, -4.94), and have income 7–10 million (β* = -5.66, 95% CI: -9.62, -1.71) were significantly associated with lower DASS21 scores (a lower DASS21 score means better mental health status). Problematic use of social media (β* = 3.54, 95% CI: (3.23, 3.85) was significantly associated with higher mental health scores (a higher DASS21 score means worse mental health status). (Table  2 )

Age, income, and use of social media (β* = 1.02, 95% CI: 0.78, 1.25) were significantly associated with higher DASS21 scores (a higher DASS21 score means worse mental health status). Marital status and field were significantly associated with lower DASS21 scores (a lower DASS21 score means better mental health status). Age groups < 20 years (β* = 6.36, 95% CI: 0.78, 11.95) and income group < 5 million (β* = 6.58, 95% CI: 1.47, 11.70) increased mental health scores. Being single (β* = -34.72, 95% CI: -47.06, -38.78), being married (β* = -38.78, 95% CI: -51.23, -26.33) and in medical sciences fields (β* = -8.17, 95% CI: -12.09, -4.24) decreased DASS21 scores. (Table  3 )

The main purpose of this study was to determine the relationship between social media use and mental health among students during the COVID-19 pandemic.

University students are more reliant on social media because of the limitations of in-person interactions [ 45 ]. Since social media has become more and more vital to the social lives of university students during the pandemic, students may be at increased risk of social media addiction, which may be harmful to their mental health [ 14 ].

During non-adulthood, peer relations and approval are critical and social media seems to meet these needs. For example, connection and communication with friends make them feel better and happier, especially during the COVID-19 pandemic and national lockdowns where face-to-face communication was restricted [ 46 ]. Kele’s study showed that the COVID-19 pandemic has increased the time spent on social media, and the frequency of online activities [ 47 ].

Because of the COVID-19 pandemic, e-learning became the only sustainable option for students [ 13 ]. This abrupt transition can lead to depression, stress, or anxiety for some students due to insufficient time to adjust to the new learning environment. The role of social media is also important to some university students [ 48 ].

Staying at home, having nothing else to do, and being unable to go out and meet with friends due to the lockdown measures increased the time spent on social media and the frequency of online activities, which influenced their mental health negatively [ 49 ]. These reasons may explain the findings of previous studies that found an increase in depression and anxiety among adolescents who were healthy before the COVID-19 pandemic [ 50 ].

According to the results, there was a statistically significant relationship between social media use and mental health in students, in such a way that one Unit increase in the score of social media use enhanced the score of mental health. These two variables were directly correlated. Consistent with the current study, many studies have shown a significant relationship between higher use of social media and lower mental health in students [ 45 , 51 , 52 , 53 , 54 ].

Inconsistent with the findings of the present study, some previous studies reported the positive effect of social media use on mental health [ 55 , 56 , 57 ]. The differences in findings could be attributed to the time and location of the studies. Anderson’s study in France in 2018 found no significant relationship between social media use and mental health. This may be because of the differences between the tools for measuring the ability to detect fake news and health literacy and the scales of the research [ 4 ].

The present study showed that the impact of using social media on the mental health of students was higher than Lebni’s study, which was conducted in 2020 [ 25 ]. Also, in Dost Mohammad’s study in 2018, the effect of using social media on the mental health of students was reported to be lower than in the present study [ 58 ]. Entezari’s study in 2021, was also lower than the present study [ 59 ]. It seems that the excessive use of social media during the COVID-19 pandemic was the reason for the greater effects of social media on students’ mental health.

The use of social media has positive and negative characteristics. Social media is most useful for rapidly disseminating timely information via widely accessible platforms [ 4 ]. Among the types of studies, at least one shows an inverse relationship between the use of social media and mental health [ 53 ]. While social media can serve as a tool for fostering connection during periods of physical isolation, the mental health implications of social media being used as a news source are tenuous [ 45 ].

The results of the GLM analysis indicated that there was a statistically significant relationship between the problematic use of social media and mental health in students in such a way that one-unit increase in the score of problematic use of social media enhanced the mental health score, and it was found that the two variables had a direct relationship. Consistent with our study, Boer’s study showed that problematic use of social media may highlight the potential risk to adolescent mental health [ 60 ]. Malaeb also reported that the problematic use of social media had a positive relationship with mental health [ 61 ], but that study was conducted on adults and had a smaller sample size before the COVID-19 pandemic.

Saputri’s study found that excessive social media use likely harms the mental health of university students since students with higher social media addiction scores had a greater risk of experiencing mild depression [ 62 ]. A systematic literature review before the COVID-19 pandemic (2019) found that the time spent by adolescents on social media was associated with depression, anxiety, and psychological distress [ 63 ]. Marino’s study (2018) reported a significant correlation between the problematic use of social media by students and psychological distress [ 64 ].

Social media has become more vital for students’ social lives owing to online education during the COVID-19 pandemic. Therefore, this group is more at risk of addiction to social media and may experience more mental health problems than other groups. Lebni also indicated that students’ higher use of the Internet led to anxiety, depression, and adverse mental health, but the main purpose of the study was to investigate the effects of such factors on student’s academic performance [ 25 ]. Previous studies indicated that individuals who spent more time on social media had lower self-esteem and higher levels of anxiety and depression [ 65 , 66 ]. In the present study, students with higher social media addiction scores were at higher mental health risk. Such a finding was consistent with research by Gao et al., who found that the excessive use of social media during the pandemic had adverse effects on social health [ 14 ]. Cheng et al. indicated that using the Internet, especially for communication with people, can harm mental health by changing the quality of social relationships, face-to-face communication, and changes in social support [ 24 ].

A reason for the significant relationship between social media use and mental health in students during the COVID-19 pandemic in the present study was probably the students’ intentional or unintentional use of online communication. Unfortunately, social media published information, which might be incorrect, in this pandemic that caused public fear and threatened mental health.

During the pandemic, social media played essential roles in learning and leisure activities. Due to electronic education, staying at home, and long leisure time, students had more time, frequency, and opportunities to use social media in this pandemic. Such a high reliance on social media may threaten student’s mental health. Lee et al. conducted a study during the COVID-19 pandemic and confirmed that young people who used social media had higher symptoms of depression and loneliness than before the COVID-19 pandemic [ 67 ].

The present study showed that there was a significant positive relationship between problematic use of social media and gender, so that women were more willing to use social media, probably because they had more opportunities to use social media as they stayed at home more than men; hence, they were more exposed to problematic use of social media. Consistent with our study, Andreassen reported that being a woman was an important factor in social media addiction [ 68 ]. In contrast to our study, Azizi’s study in Iran showed that male students use social media significantly more than female students, possibly due to differences in demographic variables in each population [ 69 ].

Moreover, there was a significant relationship between age and problematic use of social media in that people younger than 20 were more willing to use social media in a problematic way. Consistent with the present study, Perrin also indicated that younger people further used social media [ 70 ].

According to the findings, unemployed students used social media more than employed ones, probably because they had more time to spend in virtual space, leading to higher use and the possibility of problematic use of social media [ 71 ].

Moreover, non-native students were more willing to use the social media probably because students who lived far away from their families used social media problematically due to the lack of family control over hours of use and higher opportunities [ 72 ] .

The results showed that rural students have a greater tendency to use social Medias than urban students. Inconsistent with this finding, Perrin reported that urban people were more willing to use the social media. The difference was probably due to different research times and places or different target groups [ 70 ].

According to the current study, people with low household income were more likely to use social media, most likely because low-income people seek free information and services due to a lack of access to facilities and equipment in the real world or because they seek assimilation with people around them. Inconsistent with our findings, Hruska et al. reported that people with high household income levels made much use of social media [ 73 ], probably because of cultural, economic, and social differences or different information measurement tools.

Furthermore, single, divorced, and widowed students used social media more than married students. This is because they spend more time on social media due to the need for more emotional attention, the search for a life partner, or a feeling of loneliness. This also led to the problematic use of social media [ 74 ].

According to the results, Fars people used social media more than other ethnic groups, but this difference was insignificant. This finding was consistent with Perrin’s study, but the population consisted of people aged 18 to 65 [ 70 ].

In the current study, there was a significant relationship between gender and mental health, so that women had lower mental health than men. The difference was in health sociology. Consistent with the present study, Ghasemi et al. indicated that it appeared necessary to pay more attention to women’s health and create an opportunity for them to use health services [ 75 ].

The findings revealed that unemployed students had lower mental health than employed students, most likely because unemployed individuals have lower mental health due to not having a job and being economically dependent on others, as well as feeling incompetent at times. Consistent with the present study, Bialowolski reported that unemployment and low income caused mental disorders and threatened mental health [ 76 ].

According to this study, non-native students have lower mental health than native students because they live far from their families. The family plays an imperative role in improving the mental health of their children, and mental health requires their support. Also, the economic, social, and support problems caused by being away from the family have endangered their mental health [ 77 ].

Another important factor of the current study was that married people had higher mental health than single people. In addition, divorced and widowed students had lower mental health [ 78 ]. Possibly due to the social pressure they suffer in Iranian society. Furthermore, they received lower emotional support than married people. Therefore, their lower mental health seemed logical [ 79 , 80 , 81 ]. A large study in a European population also reported differences in the likelihood of mood, anxiety, and personality disorders between separated/divorced and married mothers [ 82 ].

A key point confirmed in other studies is the relationship between low incomes with mental health. A meta-analysis by Lorant indicated that economic and social inequalities caused mental disorders [ 83 ]. Safran also reported that the probability of developing mental disorders in people with low socioeconomic status is up to three times higher than that of people with the highest socioeconomic status [ 84 ]. Bialowolski’s study was consistent with the current study but Bialowolski’s study examined employees [ 76 ].

The present study was conducted during the COVID-19 pandemic and therefore had limitations in accessing students. Another limitation was the use of self-reporting tools. Participants may show positive self-presentation by over- or under-reporting their social media-related behaviors and some mental health-related items, which may directly or indirectly lead to social desirability bias, information bias, and reporting bias. Small sample sizes and convenience sampling limit student population representativeness and generalizability. This study was based on cross-sectional data. Therefore, the estimation results should be seen as associative rather than causative. Future studies would need to investigate causal effects using a longitudinal or cohort design, or another causal effect research design.

The findings of this study indicated that the high use of social media affected students’ mental health. Furthermore, the problematic use of the social media had a direct relationship with mental health. Variables such as age, gender, income level, marital status, and unemployment of non-native students had significant relationships with social media use and mental health. Despite the large amount of evidence suggesting that social media harms mental health, more research is still necessary to determine the cause and how social media can be used without harmful effects. It is imperative to better understand the relationship between social media use and mental health symptoms among young people to prevent such a negative outcome.

Data Availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Austin LL, Jin Y. Social media and crisis communication. Routledge New York; 2018.

Carr CT, Hayes RAJAjoc. Social media: defining, developing, and divining. 2015;23(1):46–65.

Fathi A, Sadeghi S, Maleki Rad AA, Sharifi Rahnmo S, Rostami H, Abdolmohammadi KJIJoP et al. The role of Cyberspace Use on Lifestyle promoting health and coronary anxiety in Young Peopl. 2020;26(3):332–47.

Anderson M, Jiang JJPRC. Teens, social media & technology 2018. 2018;31(2018):1673-89.

Nakaya ACJW. Internet and social media addiction. 2015;12(2):1–3.

Hemayatkhah M. The Impact Of Virtual Social Networks On Women’s Social Deviations. 2021.

Chaffey DJSISMM. Global social media research summary 2016. 2016.

Naderi N, Pourjamshidi HJJoBAR. The role of social networks on entrepreneurial orientation of students: a case study of Razi University. 2022;13(26):291–309.

Narmenji SM, Nowkarizi M, Riahinia N, Zerehsaz MJS. Investigating the behavior of sharing information of students in the virtual Social Networks: Instagram, Telegram and WhatsApp. Manage ToI. 2020;6(4):49–78.

Google Scholar  

Zarocostas JJTl. How to fight an infodemic. 2020;395(10225):676.

de Ávila GB, Dos Santos ÉN, Jansen K. Barros FCJTip, psychotherapy. Internet addiction in students from an educational institution in Southern Brazil: prevalence and associated factors. 2020;42:302–10.

Cain JJAjope. It’s time to confront student mental health issues associated with smartphones and social media. 2018;82(7).

Cavus N, Sani AS, Haruna Y, Lawan AAJS. Efficacy of Social networking Sites for Sustainable Education in the era of COVID-19: a systematic review. 2021;13(2):808.

Gao J, Zheng P, Jia Y, Chen H, Mao Y, Chen S et al. Mental health problems and social media exposure during COVID-19 outbreak. 2020;15(4):e0231924.

Moreno MA, Kolb JJPC. Social networking sites and adolescent health. 2012;59(3):601–12.

Claval PJLEg. Marxism and space. 1993;1(1):73–96.

Thomas L, Orme E, Kerrigan FJC. Student loneliness: the role of social media through life transitions. Education. 2020;146:103754.

Dixon SJCSOhwscsg-s-n-r-b-n-o-u. Global social networks ranked by number of users 2022. 2022.

Sindermann C, Elhai JD, Montag CJPR. Predicting tendencies towards the disordered use of Facebook’s social media platforms: On the role of personality, impulsivity, and social anxiety. 2020;285:112793.

Zheng B, Bi G, Liu H, Lowry PBJH. Corporate crisis management on social media: A morality violations perspective. 2020;6(7):e04435.

Schou Andreassen C, Pallesen SJCpd. Social network site addiction-an overview. 2014;20(25):4053–61.

Hussain Z, Griffiths MDJIJoMH. Addiction. The associations between problematic social networking site use and sleep quality, attention-deficit hyperactivity disorder, depression. anxiety and stress. 2021;19:686–700.

Hussain Z, Starcevic, VJCoip. Problematic social networking site use: a brief review of recent research methods and the way forward. 2020;36:89–95.

Cheng C, Lau Y-c, Chan L, Luk JWJAB. Prevalence of social media addiction across 32 nations: Meta-analysis with subgroup analysis of classification schemes and cultural values. 2021;117:106845.

Lebni JY, Toghroli R, Abbas J, NeJhaddadgar N, Salahshoor MR, Mansourian M, et al. A study of internet addiction and its effects on mental health. A study based on Iranian University Students; 2020. p. 9.

Griffiths MD, Kuss DJ, Demetrovics ZJBa. Social networking addiction: An overview of preliminary findings. 2014:119 – 41.

O’reilly M, Dogra N, Whiteman N, Hughes J, Eruyar S, Reilly PJCcp et al. Is social media bad for mental health and wellbeing? Exploring the perspectives of adolescents. 2018;23(4):601–13.

Huang CJC, Behavior, Networking S. Time spent on social network sites and psychological well-being: a meta-analysis. 2017;20(6):346–54.

MPh MMJJMAT. Facebook addiction and its relationship with mental health among thai high school students. 2015;98(3):S81–S90.

Galderisi S, Heinz A, Kastrup M, Beezhold J. Sartorius NJWp. Toward a new definition of mental health. 2015;14(2):231.

Health WHODoM, Abuse S, Organization WH, Health WHODoM, Health SAM, Evidence WHOMH, et al. Mental health atlas 2005. World Health Organization; 2005.

Massari LJISF. Analysis of MySpace user profiles. 2010;12:361-7.

Chang P-J, Wray L, Lin YJHP. Social relationships, leisure activity, and health in older adults. 2014;33(6):516.

Cunningham S, Hudson C, Harkness K. Social media and depression symptoms: a meta-analysis. Res Child Adolesc Psychopathol. 2021;49(2):241–53.

Article   PubMed   Google Scholar  

Haand R, Shuwang ZJIJoA. Youth. The relationship between social media addiction and depression: a quantitative study among university students in Khost. Afghanistan. 2020;25(1):780–6.

Kim B-S, Chang SM, Park JE, Seong SJ, Won SH. Cho MJJPr. Prevalence, correlates, psychiatric comorbidities, and suicidality in a community population with problematic Internet use. 2016;244:249 – 56.

Zadra S, Bischof G, Besser B, Bischof A, Meyer C, John U et al. The association between internet addiction and personality disorders in a general population-based sample. 2016;5(4):691–9.

Richards D, Caldwell PH. Go HJJop, health c. Impact of social media on the health of children and young people. 2015;51(12):1152–7.

Al-Menayes JJJP, Sciences B. Dimensions of social media addiction among university students in Kuwait. 2015;4(1):23–8.

Hou Y, Xiong D, Jiang T, Song L, Wang QJCJoproc. Social media addiction: Its impact, mediation, and intervention. 2019;13(1).

Chen I-H, Pakpour AH, Leung H, Potenza MN, Su J-A, Lin C-Y et al. Comparing generalized and specific problematic smartphone/internet use: longitudinal relationships between smartphone application-based addiction and social media addiction and psychological distress. 2020;9(2):410–9.

Tang CS-k. Koh YYWJAjop. Online social networking addiction among college students in Singapore: Comorbidity with behavioral addiction and affective disorder. 2017;25:175–8.

Samani S, Jokar B. Validity and reliability short-form version of the Depression, Anxiety and Stress. 2001.

Taghavifard MT, Ghafurian Shagerdi A, Behboodi OJJoBAR. The Effect of Market Orientation on Business Performance. 2015;7(13):205–27.

Haddad JM, Macenski C, Mosier-Mills A, Hibara A, Kester K, Schneider M et al. The impact of social media on college mental health during the COVID-19 pandemic: a multinational review of the existing literature. 2021;23:1–12.

Fiacco A. Adolescent perspectives on media use: a qualitative study. Antioch University; 2020.

Keles B, Grealish A, Leamy MJCP. The beauty and the beast of social media: an interpretative phenomenological analysis of the impact of adolescents’ social media experiences on their mental health during the Covid-19 pandemic. 2023:1–17.

Lim LTS, Regencia ZJG, Dela Cruz JRC, Ho FDV, Rodolfo MS, Ly-Uson J et al. Assessing the effect of the COVID-19 pandemic, shift to online learning, and social media use on the mental health of college students in the Philippines: a mixed-method study protocol. 2022;17(5):e0267555.

Cohen ZP, Cosgrove KT, DeVille DC, Akeman E, Singh MK, White E et al. The impact of COVID-19 on adolescent mental health: preliminary findings from a longitudinal sample of healthy and at-risk adolescents. 2021:440.

Levita L, Miller JG, Hartman TK, Murphy J, Shevlin M, McBride O et al. Report1: Impact of Covid-19 on young people aged 13–24 in the UK-preliminary findings. 2020.

Conrad RC, Koire A, Pinder-Amaker S, Liu CHJJoPR. College student mental health risks during the COVID-19 pandemic: Implications of campus relocation. 2021;136:117 – 26.

Li Y, Zhao J, Ma Z, McReynolds LS, Lin D, Chen Z et al. Mental health among college students during the COVID-19 pandemic in China: a 2-wave longitudinal survey. 2021;281:597–604.

Zhao N, Zhou GJAPH, Well-Being. Social media use and mental health during the COVID‐19 pandemic: moderator role of disaster stressor and mediator role of negative affect. 2020;12(4):1019–38.

Saputri RAM, Yumarni, TJIJoMH. Addiction. Social media addiction and mental health among university students during the COVID-19 pandemic in Indonesia. 2021:1–15.

Glaser P, Liu JH, Hakim MA, Vilar R, Zhang RJNZJoP. Is social media use for networking positive or negative? Offline social capital and internet addiction as mediators for the relationship between social media use and mental health. 2018;47(3):12–8.

Nabi RL, Prestin A, So JJC. Behavior, networking s. Facebook friends with (health) benefits? Exploring social network site use and perceptions of social support, stress, and well-being. 2013;16(10):721–7.

Campisi J, Folan D, Diehl G, Kable T, Rademeyer CJPR. Social media users have different experiences, motivations, and quality of life. 2015;228(3):774–80.

Dost Mohammadi M, Khojaste S, Doukaneifard FJJoQRiHS. Investigating the relationship between the use of social networks with self-confidence and mental health of faculty members and students of Payam Noor University. Kerman. 2020;8(4):16–27.

Franco JA. Carrier LMJHb, technologies e. Social media use and depression, anxiety, and stress in Latinos: A correlational study. 2020;2(3):227 – 41.

Boer M, Stevens GW, Finkenauer C, de Looze ME, Eijnden RJJCiHB. Social media use intensity, social media use problems, and mental health among adolescents: Investigating directionality and mediating processes. 2021;116:106645.

Malaeb D, Salameh P, Barbar S, Awad E, Haddad C, Hallit R et al. Problematic social media use and mental health (depression, anxiety, and insomnia) among Lebanese adults: Any mediating effect of stress? 2021;57(2):539 – 49.

Saputri RAM, Yumarni TJIJoMH. Addiction. Social media addiction and mental health among university students during the COVID-19 pandemic in Indonesia. 2023;21(1):96–110.

Keles B, McCrae N, Grealish AJIJoA. Youth. A systematic review: the influence of social media on depression, anxiety and psychological distress in adolescents. 2020;25(1):79–93.

Marino C, Gini G, Vieno A, Spada MMJJoAD. The associations between problematic Facebook use, psychological distress and well-being among adolescents and young adults: a systematic review and meta-analysis. 2018;226:274–81.

Vannucci A, Flannery KM. Ohannessian CMJJoad. Social media use and anxiety in emerging adults. 2017;207:163-6.

Huang CJC. behavior, networking s. Internet use and psychological well-being: A meta-analysis. 2010;13(3):241-9.

Lee Y, Yang BX, Liu Q, Luo D, Kang L, Yang F et al. Synergistic effect of social media use and psychological distress on depression in China during the COVID-19 epidemic. 2020;74(10):552.

Andreassen CS, Pallesen S, Griffiths MDJAb. The relationship between addictive use of social media, narcissism, and self-esteem: findings from a large national survey. 2017;64:287–93.

Azizi SM, Soroush A, Khatony AJBp. The relationship between social networking addiction and academic performance in iranian students of medical sciences: a cross-sectional study. 2019;7(1):1–8.

Perrin AJPrc. Social media usage. 2015;125:52–68.

Kemp. SJD-RAhdcrm-t-h-t-w-u-s-m. More than half of the people on earth now use social media. 2020.

Vejmelka L, Matković R, Borković DKJIJoC, Youth, Studies F. Online at risk! Online activities of children in dormitories: experiences in a croatian county. 2020;11(4):54–79.

Hruska J, Maresova PJS. Use of social media platforms among adults in the United States—behavior on social media. 2020;10(1):27.

Huo J, Desai R, Hong Y-R, Turner K, Mainous AG III, Bian JJCC. Use of social media in health communication: findings from the health information national trends survey 2013, 2014, and 2017. 2019;26(1):1073274819841442.

Ghasemi S, Delavaran A, Karimi ZJQEM. A comparative study of the total index of mental health in males and females through meta-analysis method. 2012;3(10):159–75.

Bialowolski P, Weziak-Bialowolska D, Lee MT, Chen Y, VanderWeele TJ, McNeely EJSS et al. The role of financial conditions for physical and mental health. Evidence from a longitudinal survey and insurance claims data. 2021;281:114041.

Moradi GJSWQ. Correlation between social capital and mental health among hostel students of tehran and shiraz universities. 2008;8(30):143–70.

Mohebbi Z, Setoodeh G, Torabizadeh C, Rambod MJIyeee. State of mental health and associated factors in nursing students from Southeastern Iran. 2019;37(3).

Grundström J, Konttinen H, Berg N, Kiviruusu OJS-ph. Associations between relationship status and mental well-being in different life phases from young to middle adulthood. 2021;14:100774.

Vaingankar JA, Abdin E, Chong SA, Shafie S, Sambasivam R, Zhang YJ et al. The association of mental disorders with perceived social support, and the role of marital status: results from a national cross-sectional survey. 2020;78(1):1–11.

McLuckie A, Matheson KM, Landers AL, Landine J, Novick J, Barrett T et al. The relationship between psychological distress and perception of emotional support in medical students and residents and implications for educational institutions. 2018;42:41–7.

Afifi TO, Cox BJ. Enns MWJSp, epidemiology p. Mental health profiles among married, never-married, and separated/divorced mothers in a nationally representative sample. 2006;41:122–9.

Lorant V, Deliège D, Eaton W, Robert A, Philippot P. Ansseau MJAjoe. Socioeconomic inequalities in depression: a meta-analysis. 2003;157(2):98–112.

Safran MA, Mays RA Jr, Huang LN, McCuan R, Pham PK, Fisher SK, et al. Mental health disparities. 2009;99(11):1962–6.

Download references

Acknowledgements

The authors would like to express their gratitude to all academic officials of Lorestan universities and Mr. Mohsen Amani for their cooperation in data collection.

Not applicable.

Author information

Authors and affiliations.

Department of Health Education and Promotion, Faculty of Health, Tehran University of Medical Sciences, Tehran, 1417613151, Iran

Abouzar Nazari

Department of Health Education and Promotion, Faculty of Health, Isfahan University of Medical Sciences, Isfahan, Iran

Maede Hosseinnia

Department of Epidemiology and Biostatistics, Faculty of Health, Iran University of Medical Sciences, Tehran, 1417613151, Iran

Samaneh Torkian

Department of Health Education and Promotion, School of Health, Tehran University of Medical Sciences, Tehran, 1417613151, Iran

Gholamreza Garmaroudi

You can also search for this author in PubMed   Google Scholar

Contributions

Abouzar Nazari and Maedeh Hossennia designed the study, collected the data and drafted the manuscript. Samaneh Torkian performed the statistical analysis and prepared the tables. Gholamreza Garmaroudi, as the responsible author, supervised the entire study. All authors reviewed and edited the draft manuscript and approved the final version.

Corresponding author

Correspondence to Gholamreza Garmaroudi .

Ethics declarations

Ethics approval and consent to participate.

Permission was obtained from the Ethics Committee of the Tehran University of Medical Sciences (IR.TUMS.SPH.REC.1400.258) before starting the study and follows the principles outlined in the 1964 Helsinki Declaration and its subsequent amendments. Participants were informed about the purpose and benefits of the study. Sending the completed questionnaire was considered as informed consent to participate in the research. The respondents’ participation was completely consensual, anonymous, and voluntary. (The present data were collected before social media filtering in Iran).

Consent for publication

Competing interests.

The authors declare that they have no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Nazari, A., Hosseinnia, M., Torkian, S. et al. Social media and mental health in students: a cross-sectional study during the Covid-19 pandemic. BMC Psychiatry 23 , 458 (2023). https://doi.org/10.1186/s12888-023-04859-w

Download citation

Received : 31 January 2023

Accepted : 10 May 2023

Published : 22 June 2023

DOI : https://doi.org/10.1186/s12888-023-04859-w

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Mental Health

BMC Psychiatry

ISSN: 1471-244X

• General enquiries: [email protected]

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 04 September 2024

Predicting anxiety treatment outcome in community mental health services using linked health administrative data

• Kevin E. K. Chai 1 ,
• Kyran Graham-Schmidt 2 ,
• Crystal M. Y. Lee 1 ,
• Daniel Rock 3 , 4 , 5 ,
• Mathew Coleman 6 ,
• Kim S. Betts 1 ,
• Suzanne Robinson 1 , 7 &
• Peter M. McEvoy 1 , 8  

Scientific Reports volume  14 , Article number:  20559 ( 2024 ) Cite this article

Metrics details

• Health care

Machine learning

• Psychiatric disorders

Anxiety disorders is ranked as the most common class of mental illness disorders globally, affecting hundreds of millions of people and significantly impacting daily life. Developing reliable predictive models for anxiety treatment outcomes holds immense potential to help guide the development of personalised care, optimise resource allocation and improve patient outcomes. This research investigates whether community mental health treatment for anxiety disorder is associated with reliable changes in Kessler psychological distress scale (K10) scores and whether pre-treatment K10 scores and past health service interactions can accurately predict reliable change (improvement). The K10 assessment was administered to 46,938 public patients in a community setting within the Western Australia dataset in 2005–2022; of whom 3794 in 4067 episodes of care were reassessed at least twice for anxiety disorders, obsessive–compulsive disorder, or reaction to severe stress and adjustment disorders (ICD-10 codes F40–F43). Reliable change on the K10 was calculated and used with the post-treatment score as the outcome variables. Machine learning models were developed using features from a large health service administrative linked dataset that includes the pre-treatment K10 assessment as well as community mental health episodes of care, emergency department presentations, and inpatient admissions for prediction. The classification model achieved an area under the receiver operating characteristic curve of 0.76 as well as an F1 score, precision and recall of 0.69, and the regression model achieved an R 2 of 0.37 with mean absolute error of 5.58 on the test dataset. While the prediction models achieved moderate performance, they also underscore the necessity for regular patient monitoring and the collection of more clinically relevant and contextual patient data to further improve prediction of treatment outcomes.

Similar content being viewed by others

Predictive modeling of depression and anxiety using electronic health records and a novel machine learning approach with artificial intelligence

Trends in incident diagnoses and drug prescriptions for anxiety and depression during the COVID-19 pandemic: an 18-month follow-up study based on the UK Biobank

A nationwide cross-sectional study on the association of patient-level factors with financial anxiety in the context of chronic medical conditions

Introduction.

Anxiety disorders are the most common class of mental illness in Australia, affecting 3.4 million adults aged 16 years and older or 17.2% of the population in 2020–2022 1 . Similarly in the United States, anxiety disorders are also the most common estimated to affect 30.6% of the population aged 18 years and older in 2020–2022 2 . These disorders are characterized by excessive worry, fear, and nervousness that can interfere with daily life. There are several different types of anxiety disorders, including generalized anxiety disorder, panic disorder, social anxiety disorder, and specific phobias. Historically, obsessive compulsive disorder and fear and stressor-related disorders (e.g., posttraumatic stress disorder) were considered anxiety disorders in the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV, American Psychiatric Association, APA, 1994) although more recent nosologies consider them separate but related classes of disorders (DSM-5, APA, 2013). Within the International Classification of Diseases (ICD version 10, 2019; ICD version 11, 2023), these disorders are three categories within the mental, behavioural or neurodevelopmental disorders.

Primary care is the main source of treatment for anxiety disorders and, where required, providers more commonly refer patients to private specialist services than to public services 3 . Nonetheless, community mental health services remain important for patients who cannot afford or access private providers 4 . Public services refer to government funded and operated specialised mental health care provided by community and hospital based ambulatory care services, such as outpatient and day clinics 5 and offer a variety of ongoing treatment options including psychotherapy, medication, and support groups. A continuing challenge for clinicians and services in all settings is to predict how well an individual will respond to treatment. There are many factors that can influence outcomes, such as the severity of the disorder, the patient's readiness for change, the quality of the treatment they receive, and external factors that reflect the overall complexity of human lives (e.g., relationship breakdown, financial hardship, workplace redundancy, bereavement) 5 , 6 , 7 .

Being able to accurately predict patient outcomes would be beneficial 7 , 8 , 9 , 10 . First, it would allow clinicians to tailor treatment plans to the individual needs of each patient, for example, by targeting known risk factors for disengagement or poor clinical outcomes. This could improve patient outcomes and reduce the need for patients to try multiple standardised treatments before finding one that works. Second, it would allow clinical planners in mental health services to allocate resources more effectively. For example, services could focus on providing more intensive treatment to patients who are at high risk of deterioration. Third, it could help identify patients who are unlikely to respond to treatment and may need additional support.

Promising methods for predicting patient outcomes for anxiety disorders and other mental illnesses include clinical prediction tools, patient-reported outcome measures, and machine learning 9 , 10 , 11 . These methods are commonly based on predictors such as patient demographics, clinical symptoms, treatment history, from different modes of data such as electronic health records, biometrics, and radiology and machine learning techniques such as logistic regression, random forests, support vector machines, gradient boosting and neural networks on datasets comprising of 4184 undergraduate students 9 and 1249 participants from a mental healthcare provider 11 .

Research on the prediction of treatment outcomes in mental health show that it is difficult, either because treatment outcomes genuinely do not vary based on individual differences or due to a range of methodological limitations, such as investigations of variables based on convenience rather than strong theory; the lack of consideration of the complex interplay between relational and content components of psychotherapy; low statistical power due to studies being designed to evaluate main effects of treatments rather than moderators of symptom change; overly homogenous samples due to exclusion criteria in randomised trials; over-reliance on significance testing without due consideration to effect sizes; failure to probe interactions to understand patterns of effects; and neglecting non-linear relationships within the context of complex relationships for humans in the real world 8 , 12 .

The alternative of relying on clinician intuition is also fraught. The biases clinicians bring to predicting psychotherapy outcomes have been long known 13 , 14 , 15 . Researchers have recently suggested that machine learning approaches that use large databases, theory-informed parameters and include complex relationships with multiple predictors of responder status, could address many of these issues 8 , 16 , 17 . Models that explain patterns in historical data and predict future outcomes, would hold promise for informing and improving the quality of care for people with anxiety disorders.

The aims of this study were to (a) investigate associations between demographic, treatment, and clinical variables and changes in psychological distress while patients were engaged with community mental health services and (b) develop machine learning models to predict reliable change in Kessler (K10) psychological distress scores using a patient’s pre-treatment (K10) scores within a community mental health setting and their past health service interactions for anxiety disorders. No previous research has used a large sample of demographic, clinical, and treatment service data administratively collected within community mental health services over a 17-year period to predict changes in psychological distress using machine learning models.

Study population

This study was approved by the Department of Health Western Australia Human Research Ethics Committee (approval number: RGS0000004782) and the Curtin University Human Research Ethics Committee (approval number: HRE2022-0001) with a waiver of informed consent obtained from the Department of Health Western Australia Human Research Ethics Committee. All methods in this study were performed in accordance with the relevant guidelines and regulations.

The study cohort was collated from a linked mental health dataset provided by the Department of Health Western Australia which is described elsewhere 18 . The linked dataset is comprised of records related to mental health assessments, community mental health service usage, emergency department presentations and inpatient admissions from 2005–2022.

For this study, we restricted the dataset to records from community mental health services where an anxiety disorder (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD): F40–F43) 19 was recorded at any time in the episode of care and to episodes of care with at least two assessments (pre and post treatment ≥ 2 weeks and ≤ 4 months apart) for determining the outcome of the treatment. Based on community mental health dataset collection rules, assessments are not to be reported for brief community interventions (< 2 weeks) and that assessments should be completed at least every three months (we adjusted to 4 months to allow delays and scheduling issues). Data is included from eligible patient episodes of care, with the first pre/post assessment used for each individual episode. Allowing multiple care episodes per patient better represents real-world conditions, providing a more accurate evaluation of the predictive model’s performance on each patient encounter. We conducted a sensitivity analysis comparing the use of single and multiple episodes of care in Supplementary Discussion 1 . ICD-10 was used as 99% of records in the community mental health data collection period within the study population used this classification.

The dataset preparation steps for defining the study population (Table 1 ) and the number of records from each anxiety disorder ICD-10 code (Table 2 ) are presented below.

Primary outcome measure

The K10 assessment is a self-reported measure of anxiety and depression symptoms characteristic of the broad construct of psychological distress 20 . It comprises of 10 questions about emotional states assessed on a five-level response scale (1 = none of the time, 2 = a little of the time, 3 = some of the time, 4 = most of the time, 5 = all of the time). The responses from the 10 questions can be summed to a total ranging from 10 to 50, where lower scores represent lower levels of distress. The K10 has high internal reliability (Cronbach's alpha = 0.93) 21 , distinguishes people with and without anxiety disorders 22 , and has been shown to be highly sensitive to change during psychotherapy 23 . We calculated Cronbach’s alpha for each ICD-10 code in our dataset using the Pingouin Python statistical package 24 .

Data analysis plan

Treatment outcome.

The treatment outcome and its effectiveness were determined by subtracting the post-treatment score from the pre-treatment score. Given that changes in scores reflect true change plus measurement error, Jacobson and Traux proposed the Reliable Change Index (RCI) to evaluate the effectiveness of therapies and interventions based on pre/post treatment scores 25 . The RCI estimates the magnitude of change in a measure’s observed score required before assuming that true change has occurred (i.e., not attributable to measurement error). The RCI is calculated by dividing the difference between the two scores by the standard error of the difference. RCI values ≥ 1.96 represent reliable improvement, RCI values ≤ 1.96 represent reliable deterioration and RCI values between − 1.96 and 1.96 represent no reliable change. The K10 was used as both a continuous outcome variable (post-treatment score) and to classify individuals with respect to whether they reliably improved, deteriorated, or remained unchanged between pre-treatment and post-treatment. The calculation of the RCI and subsequent analysis were conducted using Python 3.9.

The dataset of the study population was prepared with the prediction model features restricted to data from the K10 pre-assessment and previous community mental health episodes of care, in addition to emergency department and inpatient mental health service events (Fig.  1 ).

The data sources and features that are available for the prediction model at pre-assessment are depicted to the left of the dashed line. The first pre/post assessment is used for each episode of care and patients may have multiple eligible episodes of care in the dataset. ED emergency department.

The features extracted and created from these data sources are presented in Table 3 with definitions provided in Supplementary Table 1 . The dataset is split into a 70%/30% training and test set using fivefold random subsampling stratified cross validation in machine learning experiments.

Classification and regression models are used to predict the reliable change category (deterioration/no reliable change vs. reliable improvement) and post treatment score as a continuous variable, respectively. Models were trained using the Python scikit-learn library 26 . Training (70%) and testing (30%) datasets were created using a stratified fivefold repeated random sub-sampling cross-validation method.

Model selection

PyCaret 27 , an automated machine learning (AutoML) software library, was used to initially experiment with several machine learning algorithms by splitting only the training dataset into 70/30% using fivefold random sampling cross validation. These initial results will be used to select the most suitable classification and regression methods for subsequent experiments.

Model evaluation

The classification models are evaluated using the Receiver Operating Characteristic (ROC) Area Under the Curve (AUC), precision, recall, F1 score (harmonic mean of precision and recall) and a confusion matrix to identify how often a model gets predictions right (true positives/negatives) and wrong (false positives/negatives) for each reliable change category. An AUC of 1 is considered to have perfect predictive power while an AUC 0.5 suggests no predictive power beyond random chance 28 . The regression models are evaluated using predicted R squared (R 2 ) and the mean absolute error 29 . The predicted post-treatment scores from the regression model were also used to classify episodes of care into the reliable change categories for evaluation.

Feature importance and selection

Shapley Additive Explanations (SHAP) is a game theory inspired technique commonly used to explain the importance and contribution of features in prediction modelling 30 , 31 . It is a model agnostic approach applied to both classification and regression models in our experiments using the SHAP Python library 31 . Furthermore, a greedy forward feature selection method 32 was applied, which involved sequentially adding the feature that provides the largest contribution to the model until a pre-defined stopping criterion was met. The stopping criteria used in experiments for classification were F1 improvement > 0.01 and mean absolute error (MAE) improvement < 0.001 for regression.

The distribution of score changes between pre/post-treatment is shown in Fig.  2 . 2882 (71%) episodes of care showed a reduction in K10 score after treatment, 872 (21%) exhibited an increase in K10 after treatment and 313 (8%) remained unchanged.

The difference (score change) between pre/post treatment Kessler psychological distress scale (K10) total scores.

The RCI method was applied on the dataset, where K10 reliability coefficients (Cronbach's alphas) of 0.92–0.94 were calculated for each of the ICD codes. The pattern of reliable change for F43 (Reaction to severe stress, and adjustment disorders) is illustrated in Fig.  3 . These boundaries vary for other ICD codes (F40, F41, F42) as the reliable change index was calculated and applied separately for each diagnosis (Supplementary Fig.  1 ).

Pre/post treatment scores for F43: Reaction to severe stress, and adjustment disorders. The dashed green lines represent the boundaries of the reliable change index, with the area to the left representing reliable deterioration and the area to the right representing reliable improvement. The area between the green lines represents no reliable change.

Descriptive statistics for the dataset are reported in Table 4 . Altogether, 4067 episodes of care were available for analysis that comprised predominately of females (67%) and a mean (SD) age of 40.2 (17.9) years. The deteriorated reliable change category had low representation (212 records or 5%) and was merged with the no reliable change category (total of 2446 records or 60%) for machine learning experiments.

The machine learning results are presented in two sections (a) classification for predicting the reliable change category and (b) regression for predicting post-assessment scores.

Classification

PyCaret (AutoML) was used to initially experiment with several classification models on the training dataset using cross-validation as presented in Table 5 . Gradient boosting achieved the highest AUC (0.72) and F1 score (0.57). All the models outperform the baseline classifier (AUC = 0.5) that predicts all records as the majority class (deteriorated/no reliable change). Based on these results, gradient boosting was selected for subsequent experiments.

The gradient boosting model was run on both the train and test datasets achieving an average F1 score of 0.66 (0.66–0.69) over fivefold cross validation, with the best model achieving an AUC of 0.77 and F1 of 0.69 (Table 6 ).

The confusion matrix and ROC of the best model is presented in Fig.  4 . The confusion matrix highlighted that the model performed better in classifying episodes of care with deterioration/no reliable change (551 out of 734 (75%) correctly classified) than those that demonstrated reliable improvement (306 out of 487 (63%) correct).

( A ) Classification confusion matrix shows how often the model correctly predicted each class (true positives/negatives) and how often it made mistakes (false positives/negatives). ( B ) The receiver operating characteristic curve on the test dataset shows the sensitivity and specificity at different thresholds for prediction.

The top 20 features based on the SHAP values and feature selection results are shown in Supplementary Table 2 and Supplementary Fig.  2 . The top 2 features from both methods were the pre-assessment score and the collection stage (review). Only using the pre-assessment score achieved a 0.62 F1 score with the admission collection stage increasing the prediction performance to 0.66 and years since the previous emergency contact to 0.69. The additional 4 selected features only improve the model performance to 0.70 (+ 0.1 F1 score).

AutoML was applied to experiment with several regression models on the training dataset using cross-validation as presented in Table 7 . Gradient boosting achieved the top performance with a 0.33 R 2 and 5.82 MAE. All models, except for decision tree, outperformed the baseline regressor that predicts the mean post-treatment score for all records. The gradient boosting model was selected for subsequent experiments.

The gradient boosting model achieved an average MAE of 5.73 (5.58–5.83) over fivefold cross validation with the best model achieving an R 2 of 0.39, 0.37 and MAE values of 5.65, 5.58 on the train and test dataset, respectively (Table 8 ).

The top 20 features based on the absolute SHAP values and feature selection results are shown in Supplementary Table 3 and Supplementary Fig.  3 . Feature selection identified the pre-assessment score and the collection stage (admission) as the top features achieving a 5.75 and 5.59 MAE. The other 5 selected features only reduced the MAE to 5.52 (− 0.07).

Regression applied classification

The regression model predicted the post-assessment score and was used to classify episodes of care into reliable change. The regression applied classification results (Table 9 ) showed a decline when compared to the classification model with an F1 score of 0.69 vs. 0.67 on the test set. The AUC cannot be computed for comparison as the regression model does not generate classification probabilities.

The confusion matrix of the regression applied classification is shown in Fig.  5 . These results when compared to the classification model showed that the regression model performed poorer in predicting improved reliable change (306 vs. 304), and deterioration/no reliable change (551 vs. 533).

Regression applied classification confusion matrix shows how often the model correctly predicted each class (true positives/negatives) and how often it made mistakes (false positives/negatives).

This study aimed to investigate whether community mental health treatment is related to improvements in psychological distress and develop machine learning models for predicting reliable change and post-treatment scores in anxiety disorder treatments. The discussion will now assess whether the results and findings adequately achieved these aims.

Prediction performance

The classification model achieved an AUC of 0.76 on the test dataset of 1193 patients and an AUC between 0.75 and 0.90 indicates a moderate score in psychology and human behavioural research 33 , 34 . Our results are similar to a study that achieved an AUC of 0.73 on a test dataset of 1255 undergraduate students 9 and outperformed another study that achieved an AUC of 0.60 on 279 patients in their test dataset 11 . The regression model achieved a R 2 on the test dataset and a R 2 between 0.3 and 0.5 is generally considered a weak effect 35 but can be considered as moderate in the context of human behavioural and psychology research 36 . Furthermore, A MAE of 5.58 for the regression model could be interpreted as a relatively large error for downstream tasks such as using the predicted post-treatment scores to classify reliable change. The classification and regression applied classification model achieved similar performance and both outperformed the baseline models. The moderate performance indicates that the models could be further improved with more data and/or better discriminating features. However, there is likely to be an upper limit on prediction performance given the inherent complexity of human lives in predicting the outcome of patient treatments (i.e. Bayes error) 29 .

Classification and regression

The classification model generated probabilities for each class, which helped identify appropriate classification thresholds using the ROC and AUC evaluation metrics. However, a strength of the regression model is that it predicted the post-treatment score, which allowed for the use of classification systems such as reliable change and could potentially be used for other metrics of recovery. Furthermore, the SHAP values of the regression model were easier to interpret as a higher SHAP value indicated a higher predicted post-assessment score (poorer outcomes) compared to classification where a higher SHAP value represents as a lower post-assessment score (improved reliable change). For example, a high pre-assessment score (poor outcome) for classification resulted in the model predicting towards reliable improvement, possibly due to higher pre-assessment scores having more potential to change by post-assessment (i.e. lower scores experiencing a floor effect). However, for regression, a high pre-assessment score (poor outcome) would predict towards high post-assessment scores (poor outcomes).

Model features

The SHAP analysis and feature selection experiments showed that the pre-assessment score was the most important feature, with the assessment collection stage (admission, review) improving prediction with the remaining features providing only a minor contribution to the overall performance. However, a strength of having fewer contributing features is that the model is simpler to implement and translate into clinical software. These top features were, however, not particularly helpful for future treatment-matching, although the challenge of discovering robust predictors of mental health treatment outcomes is well known 8 , 12 . A shift from capturing predominantly health service activity data to capturing more clinically relevant data (e.g., therapeutic process, treatments delivered) along with contextual factors (i.e., non-therapy factors such as life stressors), and implementing more regular patient outcome monitoring 37 to more readily identify when a clinical intervention is not working and could be adapted or stopped, may be required to improve prediction. A cardiologist would not contemplate diagnosing and evaluating interventions for heart disease from single datapoints three months apart, and yet mental health services are expected to do so.

Clinically relevant data

While the study dataset can be seen as a strength (i.e. linked population dataset collected over a 17-year period for training and evaluating prediction models) it is still limited and can be further enhanced. The collection of administrative patient data is often driven by compliance and reporting requirements rather than a clear understanding of its clinical utility. This can lead to the accumulation of vast amounts of data that are difficult to analyse and interpret, providing limited insights into patient care and outcomes. Moreover, the focus on compliance can divert resources away from efforts to collect and curate data that is directly relevant to clinical decision-making while burdening clinicians with onerous data entry administrative tasks. For instance, measures of key individual differences theorised to play a critical role in the aetiology and maintenance of anxiety disorders, such as anxiety sensitivity 38 , intolerance of uncertainty 39 , and experiential avoidance 40 , may help with case formulation, treatment planning, and outcome monitoring. The degree to which interventions successfully modify these factors would be expected to determine downstream impacts on symptom change across the anxiety disorders. Patients’ satisfaction and engagement with the service (e.g., attendance frequency and duration), relational factors between the clinician and patient (e.g., working alliance 41 ), and social determinants (e.g., interpersonal supports and stressors, financial stressors, adverse childhood experiences 42 , 43 ) may also help focus clinicians’ and consumers’ attention on factors likely to have the largest impact on mental health and wellbeing and thereby improve outcomes and their prediction. Outcomes beyond symptom change that capture broader intervention impacts (e.g., quality of life), or monitoring progress on idiographic presenting problems (those specific and of highest priority to the individual), may be particularly valued by consumers 44 , although there is evidence that improvements in quality of life are largely mediated by symptom change 45 . Routine monitoring of known predictors of mental health and wellbeing would facilitate outcome evaluation and benchmarking, whereby novel interventions and service models can be compared over time to previous benchmarks. Without these data, services have no way of knowing if outcomes are worsening, maintaining, or improving over time, which would help with treatment planning. There is evidence that regular and routine outcome monitoring (e.g., session-by-session) that is used collaboratively by consumers and clinicians can improve outcomes, decrease negative outcomes for consumers at risk of not benefiting from treatment, and increase cost-effectiveness of interventions 46 . Future research incorporating and documenting these measures and processes would likely produce more robust and informative predictive models.

The inability of the prediction models to produce higher or more robust performance might suggest that the health administrative data being collected and made available for research lacks clinical relevance, which makes its collection and use difficult to justify. The resources invested in collecting and storing this data could be better utilised towards initiatives that directly improve patient care. Moreover, relying on data that fails to provide meaningful insights could lead to misguided policy decisions and interventions that may not produce the desired outcomes. Administrative data collected solely for service utilisation and planning metrics are insufficient for evaluating quality of care, identifying impacts of service innovations, and ensuring consumer outcomes improve over time. If the priority is maximising patient recovery, then infrastructure (e.g., digital platforms) and measures that routinely, regularly and effectively capture consumer-driven priorities are required to ensure interventions are on track for positive outcomes, or, if not, can be, collaboratively and rapidly responded to by the consumer and healthcare worker to process back on track.

Clinical assessment

A limitation of the model and experiments are features provided by clinicians in their assessments of the patients such as unstructured clinical notes. While these features could aid in prediction, it is noteworthy to highlight that it is also difficult for clinicians to predict, based only from the initial pre-assessment, whether a patient will drop out, be treatment resistant or improve. If this cannot be predicted accurately and reliably by clinical experts 13 , 14 , 15 , then it may be no different when developing and using predictive models. Future research including a combination of clinician, consumer, and administrative data may improve predictive models.

Predicting patient outcomes in mental health is a complex and difficult task but is essential for improving the quality of care for people with anxiety disorders. Research on the prediction of patient outcomes is ongoing and the preliminary findings to date are promising. This study developed classification and regression models that showed moderate prediction performance with features that would be relatively easy to collect and implement in health services organisations and clinics on a linked health administrative dataset collected over a 17-year period. Future research using regular patient outcome monitoring, clinical assessment, consumer and administrative data, may yield more accurate and reliable models for predicting patient outcomes. This will have a significant impact on the lives of people with anxiety disorders and will inform healthcare policy planning.

Data availability

The data that support the findings of this study are available from Government of Western Australia Department of Health ( https://www.datalinkage-wa.org.au/ ) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. The corresponding author can provide clarification of the dataset used for the study but for access to the data, contact the Western Australia Department of Health at [email protected].

National Study of Mental Health and Wellbeing. https://www.abs.gov.au/statistics/health/mental-health/national-study-mental-health-and-wellbeing/latest-release (2023).

Villaume, S. C., Chen, S. & Adam, E. K. Age disparities in prevalence of anxiety and depression among US adults during the COVID-19 pandemic. JAMA Netw. Open 6 (11), e2345073 (2023).

Article   Google Scholar  

Australian Institute of Health and Welfare. Medicare-subsidised mental health-specific services . https://www.aihw.gov.au/mental-health/topic-areas/medicare-subsidised-services (2023).

Castillo, E. G. et al. Community interventions to promote mental health and social equity. Curr. Psychiatry Rep. 21 , 1–14. https://doi.org/10.1007/11920-019-1017-0 (2019).

Australian institute of Health and Welfare. Community Services—Mental health AIHW . https://www.aihw.gov.au/mental-health/topic-areas/community-services (2023).

McMahon, F. J. Prediction of treatment outcomes in psychiatry—Where do we stand?. Dialogues Clin. Neurosci. 16 (4), 455–464 (2014).

Article   PubMed   PubMed Central   Google Scholar  

Chekroud, A. M. et al. The promise of machine learning in predicting treatment outcomes in psychiatry. World Psychiatry 20 (2), 154–170 (2021).

Eilertsen, S. E. H. & Eilertsen, T. H. Why is it so hard to identify (consistent) predictors of treatment outcome in psychotherapy? Clinical and research perspectives. BMC Psychol. 11 (1), 198 (2023).

Nemesure, M. D., Heinz, M. V., Huang, R. & Jacobson, N. C. Predictive modeling of depression and anxiety using electronic health records and a novel machine learning approach with artificial intelligence. Sci. Rep. 11 (1), 1980 (2021).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Stanojevic, M., Norris, L. A., Kendall, P. C. & Obradovic, Z. Predicting anxiety treatment outcomes with machine learning. In 2022 21st IEEE International Conference on Machine Learning and Applications ( ICMLA ) 957–962 (IEEE, 2022).

Hornstein, S., Forman-Hoffman, V., Nazander, A., Ranta, K. & Hilbert, K. Predicting therapy outcome in a digital mental health intervention for depression and anxiety: A machine learning approach. Digit. Health 7 , 20552076211060659 (2021).

PubMed   PubMed Central   Google Scholar  

Erceg-Hurn, D. M., Campbell, B. N. & McEvoy, P. M. What explains the failure to identify replicable moderators of symptom change in social anxiety disorder?. J. Anxiety Disord. 94 , 102676 (2023).

Article   PubMed   Google Scholar  

Meehl, P. E. Clinical versus statistical prediction: A theoretical analysis and a review of the evidence (1954).

Dawes, R. M., Faust, D. & Meehl, P. E. Clinical versus actuarial judgment. Science 243 (4899), 1668–1674 (1989).

Article   ADS   CAS   PubMed   Google Scholar  

Lilienfeld, S. O., Ritschel, L. A., Lynn, S. J., Cautin, R. L. & Latzman, R. D. Why ineffective psychotherapies appear to work: A taxonomy of causes of spurious therapeutic effectiveness. Perspect. Psychol. Sci. 9 (4), 355–387 (2014).

Mululo, S. C. C., Menezes, G. B. D., Vigne, P. & Fontenelle, L. F. A review on predictors of treatment outcome in social anxiety disorder. Braz. J. Psychiatry 34 , 92–100 (2012).

Ang, Y. S. & Pizzagalli, D. A. Predictors of treatment outcome in adolescent depression. Curr. Treat. Options Psychiatry 8 , 18–28 (2021).

Lee, C. M. Y. et al. Patterns of mental service utilisation: A population-based linkage of over 17 years of health administrative records. Community Ment. Health J. https://doi.org/10.1007/s10597-024-01300-8 (2024).

National Centre for Classification in Health. The International Statistical Classification of Diseases and Related Health Problems, Australian Modification (ICD-10-AM) 10th edn. (Independent Hospital Pricing Authority, 2017).

Google Scholar  

Kessler, R. C. M. D. & Mroczek, D. An Update of the Development of Mental Health Screening Scales for the US National Health Interview Stud y (University of Michigan, Survey Research Center of the Institute for Social Research, 1992).

Kessler, R. C. et al. Screening for serious mental illness in the general population. Arch. Gen. Psychiatry 60 (2), 184–189 (2003).

Andrews, G. & Slade, T. Interpreting scores on the Kessler Psychological Distress Scale (K10). Aust. N. Z. J. Public Health 25 , 494–497 (2001).

Article   CAS   PubMed   Google Scholar  

McEvoy, P. M. et al. Group metacognitive therapy for repetitive negative thinking in primary and non-primary generalized anxiety disorder: An effectiveness trial. J. Affect. Disord. 175 , 124–132 (2015).

Vallat, R. Pingouin: Statistics in Python. J. Open Source Softw. 3 (31), 1026. https://doi.org/10.21105/joss.01026 (2018).

Article   ADS   Google Scholar  

Jacobson, N. S. & Truax, P. Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. J. Consult. Clin. Psychol. 59 , 12–19 (1992).

Pedregosa, F. et al. Scikit-learn: Machine learning in Python. J. Mach. Learn. Res. 12 , 2825–2830 (2011).

MathSciNet   Google Scholar  

Ali, M. PyCaret: An open source, low-code machine learning library in Python. https://www.pycaret.org (2020).

Mandrekar, J. N. Receiver operating characteristic curve in diagnostic test assessment. J. Thorac. Oncol. 5 (9), 1315–1316 (2010).

Bishop, C. M. & Nasrabadi, N. M. Pattern Recognition and Machine Learning Vol. 4, 738 (Springer, 2006).

Strumbelj, E. & Kononenko, I. Explaining prediction models and individual predictions with feature contributions. Knowl. Inf. Syst. J. https://doi.org/10.1007/s10115-013-0679-x (2014).

Lundberg, S. M., & Lee, S. I. A unified approach to interpreting model predictions. In Advances in Neural Information Processing Systems ( NeurIPS ) (2017).

Guyon, I. & Elisseeff, A. An introduction to variable and feature selection. J. Mach. Learn. Res. 3 (2), 1157–1182 (2003).

Swets, J. A. Measuring the accuracy of diagnostic systems. Science 240 (4857), 1285–1293 (1988).

Article   ADS   MathSciNet   CAS   PubMed   Google Scholar  

Streiner, D. L. & Cairney, J. What’s under the ROC? An introduction to receiver operating characteristics curves. Can. J. Psychiatry 52 (2), 121–128 (2007).

Moore, D. S., Notz, W. & Fligner, M. A. The Basic Practice of Statistics (W.H. Freeman and Company, 2013).

Cohen, J. Statistical Power Analysis for the Behavioral Sciences 2nd edn. (Lawrence Erlbaum Associates, 1988).

Lambert, M. J. & Harmon, K. L. The merits of implementing routine outcome monitoring in clinical practice. Clin. Psychol. Sci. Pract. 25 (4), e12268 (2018).

Naragon-Gainey, K. Meta-analysis of the relations of anxiety sensitivity to the depressive and anxiety disorders. Psychol. Bull. 136 (1), 128 (2010).

McEvoy, P. M., Hyett, M. P., Shihata, S., Price, J. E. & Strachan, L. The impact of methodological and measurement factors on transdiagnostic associations with intolerance of uncertainty: A meta-analysis. Clin. Psychol. Rev. 73 , 101778 (2019).

Akbari, M., Seydavi, M., Hosseini, Z. S., Krafft, J. & Levin, M. E. Experiential avoidance in depression, anxiety, obsessive-compulsive related, and posttraumatic stress disorders: A comprehensive systematic review and meta-analysis. J. Context. Behav. Sci. 24 , 65–78 (2022).

Vaz, A. M., Ferreira, L. I., Gelso, C. & Janeiro, L. The sister concepts of working alliance and real relationship: A meta-analysis. Counsel. Psychol. Q. 37 (2), 247–268 (2024).

de Graaf, R., ten Have, M., Tuithof, M. & van Dorsselaer, S. First-incidence of DSM-IV mood, anxiety and substance use disorders and its determinants: Results from the Netherlands Mental Health Survey and Incidence Study-2. J. Affect. Disord. 149 (1–3), 100–107 (2013).

Sharma, S., Powers, A., Bradley, B. & Ressler, K. J. Gene × environment determinants of stress-and anxiety-related disorders. Annu. Rev. Psychol. 67 (1), 239–261 (2016).

Cuijpers, P. Targets and outcomes of psychotherapies for mental disorders: An overview. World Psychiatry 18 (3), 276–285 (2019).

Lundqvist, L. O. et al. Influence of mental health service provision on the perceived quality of life among psychiatric outpatients: Associations and mediating factors. Front. Psychiatry 14 , 1282466 (2024).

McAleavey, A. A., de Jong, K., Nissen-Lie, H. A., Boswell, J. F., Moltu, C. & Lutz, W. (2024). Routine outcome monitoring and clinical feedback in psychotherapy: Recent advances and future directions. Administration and Policy in Mental Health and Mental Health Services Research , 1–15.

Download references

Acknowledgements

This work was supported by the Digital Health Cooperative Research Centre (DHCRC) [DHCRC-0076]. DHCRC is funded under the Australian Commonwealth’s Cooperative Research Centres (CRC) Program. The funder had no role in the study design, data collection and analysis, decision to publish, or preparation for the manuscript. The authors wish to thank Justin Manuel from Western Australia Country Health Service for his ongoing contribution to the overall project and to the staff from the Department of Health WA’s Data Linkage Services and the Hospital Morbidity Data Collection, Emergency Department Data Collection, and Mental Health Data Collection.

Author information

Authors and affiliations.

School of Population Health, Curtin University, Perth, WA, Australia

Kevin E. K. Chai, Crystal M. Y. Lee, Kim S. Betts, Suzanne Robinson & Peter M. McEvoy

Department of Health, Perth, WA, Australia

Kyran Graham-Schmidt

Western Australia Primary Health Alliance, Perth, WA, Australia

Daniel Rock

Discipline of Psychiatry, Medical School, University of Western Australia, Perth, WA, Australia

Faculty of Health, Health Research Institute, University of Canberra, Canberra, ACT, Australia

Western Australia Country Health Service, Albany, WA, Australia

Mathew Coleman

Deakin Health Economics, Deakin University, Melbourne, VIC, Australia

Suzanne Robinson

Centre for Clinical Interventions, North Metropolitan Health Service, Perth, WA, Australia

Peter M. McEvoy

You can also search for this author in PubMed   Google Scholar

Contributions

K.E.K.C analysed the data, conducted the experiments and drafted the manuscript. K.E.K.C, K.G.S C.M.Y.L, P.M.M., D.R, M.C conceived the design and P.M.M, K.G.S, D.R, M.C provided clinical advice for the project. K.S.B, P.M, D.R, S.R secured funding for the project. All authors contributed to the critical revision of the manuscript and approved the final version of the article to be published.

Corresponding author

Correspondence to Kevin E. K. Chai .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Ethics approval

Additional information, publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary information., rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ .

Reprints and permissions

About this article

Cite this article.

Chai, K.E.K., Graham-Schmidt, K., Lee, C.M.Y. et al. Predicting anxiety treatment outcome in community mental health services using linked health administrative data. Sci Rep 14 , 20559 (2024). https://doi.org/10.1038/s41598-024-71557-2

Download citation

Received : 17 May 2024

Accepted : 29 August 2024

Published : 04 September 2024

DOI : https://doi.org/10.1038/s41598-024-71557-2

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

NASPA Releases Research on Top Issues in Student Affairs

August 28, 2024 naspa.

WASHINGTON, DC - August 28, 2024— Findings from NASPA’s inaugural survey, 2024 NASPA Top Issues in Student Affairs, highlight the critical issues shaping how student affairs professionals support students and where they are focusing efforts in an ever-changing higher education landscape. The survey gathered perspectives from vice presidents and assistant vice presidents of student affairs as they ranked the level of importance of 50 issues across five key areas by level of importance to their institution in 2024. The key areas that issues were sorted within include Administration & Governance, Assessment & Evaluation; Health, Safety & Well-Being; Justice, Equity, Diversity & Inclusion; and Technology & Innovation. 

Student affairs leaders ranked issues related to Health, Safety, & Well-Being as very important to their institutions in 2024, with the top three being:

Providing health, safety, and well-being education and training for students (83%)

Increasing access to mental health services (82%)

Enhancing campus-wide collaboration on health and safety (80%)

The survey reflects the input of senior leaders from a diverse range of institutions and is informed by key trends in the field. Moving forward, NASPA aims to refine this research with input from a broader panel of student affairs professionals, ensuring the survey remains relevant and comprehensive over time.

NASPA President Dr. Amelia Parnell emphasized, “This inaugural examination of top issues provides valuable insights regarding the areas that senior leaders are prioritizing. As we continue to conduct this annual survey, we will also explore distinctions across various institution types, which will help us better understand the nuances of each topic.” The full report is now available on the NASPA website: Read the Report

For more information, contact [email protected]

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Afr Health Sci
• v.19(2); 2019 Jun

A quantitative assessment of the views of mental health professionals on exercise for people with mental illness: perspectives from a low-resource setting

Davy vancampfort.

1 KU Leuven Department of Rehabilitation Sciences, Leuven, Belgium

2 KU Leuven, University Psychiatric Center KU Leuven, Leuven-Kortenberg, Belgium

Robert Stanton

3 Central Queensland University, School of Health. Medical and Applied Sciences, North Rockhampton, Australia

Michel Probst

Marc de hert, ruud van winkel.

4 KU Leuven, Centre for Contexual Psychiatry, Leuven, Belgium

Inez Myin-Germeys

Eugene kinyanda.

5 MRC/UVRI, Uganda Research Unit on AIDS, Entebbe, Uganda

6 Department of Psychiatry, Makerere College of Health Sciences, Kampala, Uganda

7 Senior Wellcome Trust Fellowship, London

James Mugisha

8 Butabika National Referral Mental Hospital, Kampala, Uganda

9 Kyambogo University, Kampala, Uganda

Exercise is nowadays considered as an evidence-based treatment modality in people with mental illness. Nurses and occupational therapists working in low-resourced mental health settings are well-placed to provide exercise advice for people with mental illness.

We examined the current exercise prescription practices employed by Ugandan health care professionals when working with people with mental illness, and identified perceived barriers to exercise prescription and exercise participation for people with mental illness.

In this study, 31 Ugandan health care professionals 20 men; 31.2 ± 7.1 years completed the Exercise in Mental Illness Questionnaire- Health Professionals Version EMIQ-HP.

The vast majority of the respondents 29/31, 94% reported they prescribed exercise at least “occasionally” to people with mental illness. Exercise-prescription parameters used were consistent with those recommended for people with mental illness. Regarding barriers to exercise participation, coping with side effects of psychotropic medication at the individual level and reducing stigma at community level should be prioritized.

A health care reform to enable collaboration with exercise professionals, such as exercise physiologists or physiotherapists, might increase exercise uptake for people with mental illness, thereby improving health outcomes for this vulnerable population.

Introduction

Mental illness is the leading cause of years lived with disability YLD in sub-Saharan Africa SSA, accounting for about one fifth of all disability-associated burden YLD 1 . It is estimated that the burden will more than double by 2050 Institute for Health Metrics and Evaluation, 2013. The consequences of the rising and devastating burden of mental illness is not only having an impact on the individual but also on the family and community as a whole. The quality of life of those affected is severely reduced and economic costs are significant 2 . Moreover, physical co-morbidities 3 , 4 , chronic pain 5 , 6 and HIV/AIDS 7 , 8 are more common in people with mental illness and add to the disability and burden. Despite this tremendous burden, most SSA countries invest less than 1% of the total health budget on mental health World Health Organization, 2011. As a result, mental health services are poorly resourced and considered inaccessible 9 . Therefore, it is not a surprise that treatment rates for people with mental disorders remain low, with less than 10% receiving mental health care 10 .

Despite significant efforts by the Ugandan Ministry of Health to improve access to mental health services 11 , treatment gaps remain, in part, due to the cultural beliefs and help-seeking behaviors of the Ugandan population, who often seek traditional medicine as first-line intervention, as opposed to Westernized care 9 . Community-based rehabilitation, psychoeducation and social support are recommended for low resource settings such as Uganda, with assertive community care and cognitive behavioral therapy recommended as additions in higher resourced settings with stronger service-delivery platforms 12 .

In recent years there has been an increasing interest in exercise as a stand-alone or complementary treatment modality for people with mental illnesses such as depression 13 , schizophrenia 14 , bipolar disorders 15 , alcohol use disorders 16 , post-traumatic stress disorder 17 and anxiety disorders 18 . Exercise supports patients in managing their psychiatric symptoms, andit improves the physical health and quality of life 19 . Since exercise may be implemented at low cost and often requires no or minimal resources and can be easily tailored to accommodate co-morbidities or injuries, it may be attractive in low resource settings. The potential role of exercise interventions however seems to be given low priority and to be neglected in these low resource settings 20 . This is not surprising since the emphasis in health service delivery in SSA is based on the biomedical model as opposed to the biopsychosocial model with an important focus on pharmacotherapy in the management of mental disorders Mugisha, 2016.

Although physio-therapists and exercise physiologists are ideally placed to deliver exercise interventions 21 – 24 , these clinical roles are currently not available in many low resource settings in SSA countries 20 . In clinical practice, the existing staff including mental health nurses, occupational therapists, psychologists, doctors and psychiatrists, are currently better placed to deliver exercise counseling for people with mental illness.Qualitative 25 and quantitative Stanton et al., 2015 a studies in high income countries suggest that nurses working in mental health settings acknowledge the value of exercise for people with mental illness and believe providing exercise advice is part of their role. However, such data is lacking in low resource settings, but are urgently needed in order to influence policy and practice and maximize access to the therapeutic potential of exercise at all levels of care. Such data may also help address personal factors including low confidence and limited training in exercise prescription 26 , and systemic barriers such as competing work priorities 27 , 28 that limit the provision of exercise programs for people with mental illness in low resource settings.

Therefore, in order to better inform the development of exercise interventions that can be implemented in low resource settings at all levels of care in and to define specific training needs, a comprehensive assessment of the current knowledge, attitudes, beliefs and behaviors of health practitioners working in mental health settings regarding the prescription of exercise for people with a mental illness is required. The aims of the present study are twofold. Firstly, to examine the current exercise prescription practices employed by Ugandan health care professionals when working with people with mental illness. Secondly, to identify perceived barriers to exercise prescription and perceived barriers to exercise participation for people with mental illness.

Study design

This was a cross-sectional study.

Study setting and procedure

This study was a cross-sectional investigation undertaken at Uganda's only psychiatric hospital, the Butabika National Referral Mental Hospital. The nurses and occupation therapists working in two adult long-term care units were invited to participate. Combined, these two units could accommodate 110 in-patients and employs 32 nurses and 2 occupational therapists. First, all the nurses and occupational therapists were provided with an information sheet outlining the purpose of the study and with the questionnaire. The information sheet and questionnaire were provided by a research nurse who was not working in the two adult long-term care units. There were no exclusion criteria. The information sheet stated that the research nurse was available upon request to assist in the completion of the questionnaire. After one month, the research nurse gave a one-time reminder to the staff members who volunteered to participate, to complete the questionnaire within the following month. A self-administered questionnaire were used to collect data since the participants were fluent and competent in English. Content validity, conceptual equivalence and cultural sensitivity were also not an issue. No incentive for completion of the survey was offered. Participation was anonymous with questionnaires placed in a sealed box not observable to other staff members. Informed consent was assumed on completion and return of the survey. Data were collected during November and December, 2017. Ethical clearance for the study was received from the local Butabika Hospital Research Committee.

Study instrument

Participants completed the Exercise in Mental Illness Questionnaire- Health Professionals Version EMIQ-HP for which content validity and test-retest reliability have previously been established 29 . The instrument comprises six domains of exercise knowledge, exercise beliefs, exercise prescription behaviors, barriers to exercise, personal exercise habits and demographics. Time to complete the paper-based survey was approximately 20 minutes. Exercise-prescription practices were determined using the question; “Do you prescribe exercise to people with a mental illness?” with four response options of: “Never”, “Occasionally”, “Most of the time” and “Always”. Self-rated knowledge and confidence to prescribe exercise for people with mental illness were assessed using Likert-response questions, 1 = “Very poor”, and 5 = “Excellent”. To examine the views of other well-established treatment strategies for mental illness, respondents were asked to rate how valuable they believed each treatment was, compared to exercise, using a five-point Likert scale where 1 = “Significantly less than exercise”, and 5 = “Significantly better than exercise”. Electroconvulsive therapy and bright light therapy were removed from the list of well-established treatments, as they are not practiced in the setting we investigated. Respondents then answered questions regarding strategies used to prescribe exercise including the frequency, intensity, duration, and type of exercise duration, frequency, using fixed response options. Level of agreement questions using a five-point Likertscale with anchors from 1 = “Strongly disagree” to5 = “Strongly agree” were used to examine respondents' views regarding barriers to exercise prescription for people with mental illness, and exercise participation by people with mental illness. Future training needs were examined with respect to level and topics of interest for professional development. Responses to statements for each subsection were then summed, thus a higher score indicates a higher level of agreement. Finally, the following demographic data were captured as part of the EMIQ-HP: gender male / female, age years, current marital status married or not married, years in profession, and full time employment yes or no.

Statistical analysis

Participant demographics, exercise prescription practices and responses to statements regarding barriers to exercise prescription for; and barriers to exercise participation by people with mental illness are reported using descriptive statistics mean ± SD, frequencies. In accordance with previous studies 29 , 30 , responses to statements were collapsed to three categories; “Agree”, “Neutral”, and “Disagree”. Since the Likert scale responses are not assumed to be on an equal interval scale, and frequency of responses to “Strongly agree” and “Agree” are low, these responses were collapsed to “Agree”. Based on rating scale optimization, collapsing the positive responses “Strongly agree” and “Agree” into one category is logical and does not create an artificial new category. Similarly, combining negative responses “Strongly disagree” and “Disagree” demonstrates the strength of these responses, compared to neutral and positive responses 31 .

Participants

Thirty-one health care professionals, representing 91% of potential respondents completed the EMIQ-HP. Respondents included 10 nurses, 19 nurses with specialist mental health nursing qualification and 2 occupational therapists. The characteristics of respondents are shown in Table 1 .

Demographic characteristics of respondents n=31

Frequency of exercise prescription

Three respondents 10% reported ‘Always’ prescribing exercise, 3 (10%) reported prescribing exercise ‘Most of the time’, 23 (74%) reported prescribing exercise ‘Occasionally’ and two 6% reported ‘Never’ prescribing exercise.

Knowledge about and confidence regarding exercise prescription

Sixteen respondents (52%) indicated that they had a formal training in exercise prescription. The mean ±SD response for self-reported knowledge and confidence scores was3.6± 0.6 and 3.6±0.5, respectively. Nine respondents 29% reported a “Good” or “Excellent” knowledge of exercise prescription for mental illness. Similarly, 10 respondents (32%) reported that they are confident at prescribing exercise for people with mental illness to be ''Good” or “Excellent”.

Views of health care professionals comparing established treatments to exercise for the treatment of mental illness

Overall, between 74% and 90% of respondents believed other treatment modalities to be of equal or greater value compared to exercise. The majority of respondents n=24, 77% believed medication is ‘Somewhat’, or ‘Significantly’ more valuable than exercise. Between 45% and 65% of respondents believed other treatment modalities are ‘Somewhat’, or ‘Significantly’ more valuable than exercise. Slightly more than one-third n=11, 35% of respondents believed social skills training is of equal value to exercise. A summary of the findings regarding the value to treatments compared to exercise is shown in Table 2 .

Comparison of established treatments to exercise for the treatment of mental illness

Exercise prescription strategies

When considering the strategies used to prescribe exercise to people with mental illness, personal discussion, including the development of an individualized program was the most frequently used strategy n=19/29, 65%. Only one respondent indicated referral to an exercise physiologist / physiotherapist for exercise prescription. The most commonly reported recommendation for exercise frequency was to exercise “As often as they can” n=12/29, 41% followed by on “Most days of the week” n=10/29, 34%. The most frequently recommended exercise intensity for people with mental illness was “At a level that makes them feel good” n=9/29, 31%, followed by “Moderate” n=7/29, 24%. “30 minutes per day'' n=11/29, 38% was the most frequently prescribed exercise duration followed by “Exercising as long as they can” n=7/29, 24%. Relaxation exercises such as yoga or Tai Chi n=16/29, 55% were the most commonly prescribed mode of exercise followed by aerobic exercise n=10/29, 34%.

Barriers to exercise prescription

Responses to statements regarding the barriers to exercise prescription for people with mental illness are shown in Table 3 . When collapsed to categories of ‘Agree’, ‘Neutral’, and ‘Disagree’, just over half n=18, 58% agreed that patient's mental health makes it impossible for them to participate in exercise. Almost half n=13, 45% agreed that getting injured during exercise is a concern. Overwhelmingly however, 87% of respondents n=27 agreed that exercise will be beneficial, and were interested in exercise prescription for this population. Only 13% n=4 agreed that exercise prescription is not part of their job, but 16% agreed that they did not know how to prescribe exercise for people with mental illness. Importantly, 71% n=22 agreed that exercise prescription for people with mental illness is best delivered by an exercise professional.

Level of agreement [n %] with statements regarding barriers to exercise prescription for people with mental illness

Barriers to participation

The agreement with statements expressed by people with mental illness regarding exercise participation is shown in Table 4. In a manner similar to the responses to statements regarding barriers to exercise prescription, scale optimization was performed to result in three categories. When collapsed to categories of “Agree”, “Neutral”, and “Disagree”, almost three-quarters of respondents n=23, 74% agreed with the consumer view that “There is too much stigma attached to having a mental illness.” while more than half n=18, 58% agreed with the statement “There are too many side effects from the medications.”

Training needs for health care professionals

Participants were cognizant of the need for ongoing professional development in the field. More than two-thirds of respondents 23/31 indicated they would “Definitely” attend further training for exercise prescription for people with mental illness, with the most commonly reported topics of interest being “How to assess the patients' suitability for physical activity?” n=22, 71% and “How to get and maintain motivation in people with mental illness?” n=18, 58%.

General findings

The present study is the first to provide new insight from the perspectives of health professionals working in a long-term adult inpatient mental health facility in a low resource country, with regard to the prescription of exercise to people with mental illness. The 31 respondents in the present survey represent approximately 90% of the health care professionals working in the mental health setting explored.

The vast majority 29/31, 94% reported that they prescribed at least “occasionally” exercise to their patients. The positive attitude of nurses and occupational therapists towards exercise is in line with previous research in other parts of the world. For example, a British study 32 reported that 77% of mental health nurses felt that providing exercise advice and referring to a community facility was part of their role while in an Australian study Stanton et al. 33 , 2015b 72% of the nurses reported prescribing exercise to mental health consumers.

Participants self-reported a high level of knowledge and confidence in prescribing exercise for people with mental illness. This high level of knowledge is also reflected in the exercise-prescription parameters for exercise frequency, intensity, duration, and type recommended by respondents. These are consistent with those recommended for people with mental illness 33 , 34 . International guidelines call for aerobic exercise to be performed 3 to 5 days per week for 30 min at low-to-moderate or self-selected intensity 33 , 34 . The popular view regarding exercising at a level that makes them feel good, and for as long as they like, is consistent with the use of autonomous regulation in exercise prescription for people with mental illness 35 – 37 and consistent with approaches used in other health professional groups 38 , 39 .

The high level of knowledge and confidence in prescribing exercise for people with mental illness is perhaps unsurprising given that more than half of the existing work staff indicated that they are trained in exercise prescription and implementing lifestyle interventions for people with chronic or complex health conditions, a rate which is for example much higher than in Australia where only 11% of the nurses reported having any formal training in exercise prescription 38 . Since Butabika Hospital is a national referral hospital, many of its staff are likely also more exposed to information related to exercise compared to work staff in more rural areas, owing to the fact that the hospital runs a bigger budget, from both local resources and donors for continued medical education 27 , 28 . On the other hand, almost 75% of the respondents indicated that they would “definitely” attend further training for exercise prescription for people with mentalillness, in particular related to how to assess patients and how to motivate them towards an active lifestyle. More than seventy percent of the participants also reported that exercise to people with mental illness is actually best delivered by an exercise professional, although only one respondent referred patients to such an exercise professional. A potential reason for the very low referral rate is the lack of exercise specialists working in mental health care settings in low income countries 20 . It is likely that due to the strong biomedical focus on pharmacotherapy 27 , policy makers are yet to be fully aware of the benefits of including exercise specialists in the Ugandan mental health care system. Hence, a need to re-orient the current health care system including policy makers to embrace these professions in the management of mental health problems is needed. Internationally, exercise physiologists 24 and physiotherapists 21 are the health professional groups with expertise in exercise prescription for people with mentalillness. Both health professional groups are trained in exercise prescription for people with chronic and complex health conditions including for those with mental illness. Thus, exercise professionals are able to develop and deliver cost- and resource-efficient exercise interventions. To date, however, few people in Uganda, and Sub-Saharan African as a whole with mental illness are referred to exercise specialists in primary health care settings 27 , 28 . One of reasons might be the lack of mental health training for these exercise professionals in this part of the world 20 .

Despite the fact that the respondents reported a high level of knowledge and confidence in prescribing exercise for people with mental illness, the potential of exercise within the multidisciplinary treatment seems not yet to be fully endorsed in low resource countries. “Standard treatments” for mental illness were generally perceived as of greater therapeutic value to exercise. One reason might be the previously reported strong biomedical focus, while clinicians tend to favor interventions related to their own discipline, for example occupational therapists favor vocational rehabilitation and social skills training while nurses favor family support. Another issue might be the socio-cultural views of mental illness whereby potential patients do not routinely seek treatment due to the high levels of stigma, and where treatment is provided traditionally through non-Westernized approaches. Thus, exercise as part of any treatment strategy is largely underutilized.

In the current study, we also explored barriers to exercise prescription for health care professionals and participation by mental health consumers. A previous study in physical therapists demonstrated that a-motivation by mental health consumers is the most significant barrier to exercise participation 40 , while barriers to exercise prescription by nurses working in mental health in Australia extend to the systemic level 41 . For example, previous research highlighted how the fragmentation of roles, prioritization of other tasks, lack of time, and limited resources impact on the prescription of exercise by nurses working in mental health in Australia 41 . Surprisingly, in our study these barriers were not endorsed by more than half of our respondents. In our study, respondents agreed with a number of statements regarding barriers to exercise participation proposed by people with mental illness. This was especially the case for statements located more at the individual level such as the side effects from pharmacotherapy, and at the community level where stigma and negative attitudes surrounding mental illness were considered a major barrier for people with mental illness to engage in exercise. Therefore, in order to facilitate exercise uptake, deliberate efforts need to be undertaken within the hospital to assist patients in coping with the side effects of their pharmacotherapy while at the community level public health campaigns are needed to reduce the stigma associated with mental illness. These changes should be augmented by professional development opportunities suggested by respondents including the assessment, initiation, and motivation for continued exercise participation by people with mental illness.

Limitations

The present study should be considered in the light of some limitations. First, we were not able to obtain completed surveys from all health care professionals working in mental health setting where the study was undertaken. This could be due to the time commitment required, personal concerns regarding the knowledge related to exercise for people with mental illness and the lack of incentive for participation. Uganda also has a small mental health workforce with around 28 psychiatrists and 230 mental health nurses, most of whom work at Butabika 42 , thus competing priorities may have affected the survey response rate. However, considerable effort was directed towards recruitment and the proportion of professionals who completed surveys represents approximately 90% of the eligible staff. Second, the present survey was also limited to only one hospital. Butabika is however the only national mental health referral center in Uganda. Together with a small cohort, the generalizability of our findings remains to be confirmed while interdisciplinary comparisons were not possible. Third, although the EMIQ-HP has been validated before in Australia 29 , the validity for the mental health workforce in low income countries is unknown.

The present findings suggest nurses and occupational therapists who participated in this study are supportive of exercise, and those who prescribe exercise do so in accordance with accepted protocols. Moreover, respondents disagree with many of the commonly cited barriers to exercise prescription and participation in the current literature. Regarding barriers to exercise participation, reducing stigma at community level should be prioritized. Collaboration with exercise professionals such as exercise physiologists and physiotherapists as part of a multidisciplinary approach to mental health care could increase exercise uptake and consequently improve health outcomes for mental health consumers. Further examination in larger cohorts including all relevant healthcare disciplines will progress our understanding of the delivery of exercise for people with mental illness in low resourced settings.

Acknowledgements

The authors would like to thank the nurses of the Butabika National Referral and Mental Health Hospital who completed the questionnaires for the purpose of this study.

Conflict of interest

None to declare from either author.

This research was funded by Geestkracht VZW.

Role of funding source

The funding organization had no role in the research at any stage, nor influenced the decision to publish the article.