The Impact of Resilience Interventions on University Students’ Mental Health and Well-Being: A Systematic Review

: Attending university is a time of considerable change, and there are rising concerns about the mental health and well-being of university students, leading to calls for a whole-university approach to student support. Resilience interventions offer an opportunity to improve mental health and well-being, whilst also developing a much sought-after graduate attribute. We conducted a systematic review of interventions designed to increase university students’ levels of resilience and examined the impact of these on students’ mental health and well-being. Five databases identified 1377 unique records, 47 of which were eligible for inclusion. Eligible studies were drawn from a range of countries and adopted different designs, with the most common being the randomised controlled trial (RCT). Interventions were classified into mindfulness, skills-based, psychoeducation, and coaching, with delivery both online and face-to-face. The most common outcomes were depression, anxiety, stress, and well-being. The ‘Quality Assessment Tool for Before-After (Pre-/Post) Studies’ was used to assess the risk of bias with most studies rated as fair. Overall, the studies indicated that there is little evidence of a positive effect on depression, but stress and anxiety may be reduced following interventions. Well-being data were inconclusive. Interestingly, most interventions did not impact measures of resilience or mindfulness, despite the training targeting these constructs. The available research is currently limited and there is a need for more high-quality designs providing descriptions of interventions.


Introduction
Attending university is often considered an exciting rite of passage that typically coincides with the transition to adulthood, marking a period of psychological, sociological, and biological development [1].During university study, students develop specialist knowledge in their chosen disciplines as well as so-called graduate attributes [2].Whilst the attributes identified vary across institutions [3], mapping studies have indicated some shared attributes, including time management, organization, conflict management, good communication, and resilience [4].Keeping up with academic study as well as developing these attributes in higher education (HE) can create a high workload, which contributes to stress in student populations [5].Estimates of the prevalence of stress vary, with some studies reporting as much as 84% of students experiencing stress [6].Such stress can negatively impact academic performance [7,8].Additionally, the impact of stress on mental health is well-established with chronic stress known to be a risk factor for both internalizing and externalizing difficulties [9].In line with reports of high stress levels in students, there has been increasing concern about students' mental health globally with reports of a mental health crisis in this population [10][11][12].These reports are not unfounded, with a study of over 14,000 students across eight countries in 19 universities, finding that over 30% met the diagnostic criteria for at least one mental health condition [13].The most common condition Educ.Sci.2024, 14, 510 2 of 25 reported was depression, followed by anxiety, with various demographic characteristics (e.g., gender, age, and religiosity), correlating with mental health difficulties [13].
Several different approaches have been taken to support students in managing stress and mental health conditions.This includes universal interventions aimed at improving well-being, which may be perceived as less stigmatizing and appeal to students who might not otherwise seek help [14], peer support programmes [15], relaxation, and a range of training programmes, including social skills, attention, and mindfulness [16].Previous reviews have indicated that training programmes for stress management can be beneficial for university students [17], whilst also noting the importance of tailoring programmes to this specific cohort [16].One area of particular interest is resilience training, which has been shown to be beneficial as a universal intervention for university students [18] and aligns with the development of graduate attributes [4], offering an excellent opportunity to embed student support directly into the curriculum as part of a settings-based approach to students' mental health.This approach has been widely encouraged and is a consideration in the UK University Mental Health Charter scheme [19].
The American Psychological Association defines resilience as "the process and outcome of successfully adapting to difficult or challenging life experiences, especially through mental, emotional, and behavioural flexibility and adjustment to external and internal demands" [20].This complex definition reflects the fact that resilience can be defined as a trait, process, or outcome [21,22], which has created difficulties for researchers in adopting a universal definition [23].Despite these difficulties, current research allows resilience to be depicted as a set of competencies [24].When viewed as a set of competencies, it is conceivable that resilience can be taught or trained, something supported by prior research [25,26].Furthermore, a competency approach to resilience creates a framework that supports the development of interventions, which can target external factors, such as building relationships and communities within specific contexts (e.g., support systems), and individual factors (e.g., coping skills, problem-solving, and cognitive flexibility) [18], the latter of which also aligns with the sought-after skills within HE graduate attributes [4].It is perhaps unsurprising then that research on HE has identified that resilience is beneficial in students, helping them to adjust to university life and overcome adversity, as well as improve their well-being and academic outcomes [27][28][29][30][31].
As interest in resilience training within HE has grown, several reviews have been conducted in this area, with a focus specifically on students.A systematic review published in 2016 focused on undergraduate indigenous students only and identified limited evidence as an issue [32].Two scoping reviews followed, one of which focused on health profession students only, and in both cases, the researchers also identified a lack of methodological constraints within the studied review, making conclusions hard to establish [33,34].More recently, a more comprehensive meta-analytic systematic review was conducted, which focused on students across disciplines and explored the effects of resilience training as well as the features of the training programmes [18].This review exclusively focused on randomised controlled trials (RCTs) and examined primary outcome measures of resilience, depressive symptoms, and stress, alongside several secondary outcomes.The review included 29 studies from 2008 to 2020 and reported that resilience interventions within RCTs have a small beneficial effect on symptoms of depression and stress and a small positive effect on resilience.Whilst the focus on RCTs does result in the use of a robust study design, it also risks missing considerable research within an education context.For example, a recent systematic review examining well-being interventions within education found that only 17.4% of studies used an RCT design [35] and the Education Endowment Foundation identified several challenges to RCTs in educational contexts [36].Furthermore, whilst randomisation is suitable for health-related intervention assessment, which is considered a strength of RCTs, it is unlikely to account for a considerable amount of the sources of error within an educational context [37].As such, whilst the previous review included high-quality evidence, it is unlikely that the exclusion of other designs had created the robustness reported for other disciplines, and it is likely that considerable research was omitted.The current review aims to build upon previous reviews in this area by including students of all disciplines rather than just health profession students, and incorporating the most common experimental designs employed in education research to address the research question "What is the effect of resilience interventions on mental health and well-being in university students?".

Protocol Registration
This review was conducted in accordance with PRISMA guidelines [38].The systematic review protocol was registered in the prospective register of systematic reviews, PROSPERO (registration number CRD42022315583 [39]).

Eligibility Criteria
Studies had to fulfil several eligibility criteria to be included in this review.Firstly, studies had to be published in English and be peer-reviewed primary research articles.Secondly, they had to focus on students in post-compulsory education (16 years old or older), including full-time undergraduate or postgraduate students registered for an oncampus program.We opted to exclude those studying for a distance learning programme, given that the majority of HE qualifications are campus-based, even since COVID-19.Additionally, previous research has indicated that the challenges faced by distance students are distinct from those on campus and that this means different adjustments are required to be resilient [40].Thirdly, the interventions studied had to be resilience-based, i.e., targeting individual factors, which can support greater resilience such as attention, behaviour and mood control, coping skills, self-esteem, cognitive flexibility, optimism, and problemsolving [24].This could be in the form of psychoeducation, coaching, skills-based training, and mindfulness-based interventions.Interventions targeting non-student populations, such as instructors and staff, were excluded.We deliberately selected a wide range of intervention approaches to reflect the diverse mediating processes and mechanisms that can impact resilience according to a recent conceptual framework [18].Fourthly, eligible studies focused on quantitative research, with observational, randomised controlled trials, non-randomised controlled trials (otherwise known as quasi-experimental), and crosssectional studies.Qualitative studies were excluded from this review, in line with previous reviews [18,33,34].Comparison groups for these designs were either the non-intervention control group, where available, or the pre-scores in pre/post-study designs.Finally, to be eligible, the study needed to include at least one measure of well-being (e.g., mental well-being, quality of life, happiness, self-efficacy, life satisfaction, positive affect, and coping), or mental health (e.g., depression, anxiety, stress, and negative affect).

Search Strategy
Literature searches were performed on five databases (PsycINFO, Medline, Embase, Global Health, and Web of Science), which included a considerable amount of educationbased literature from the earliest possible dates to February 2024, when the last searches were conducted.Searching was completed in two phases, with initial searching conducted by April 2022 and an updated search run in February 2024.Search terms were related to the population (undergraduate OR post-graduate or graduate or adult learner or "university student"), intervention (psychoeducation or resilience interventions), and outcomes (resilien* OR optimis* or coping or mental health or depress* or anxiety or well-being), with all search categories combined with AND (see published protocol).

Selection Process
Preliminary searches were conducted by two authors (Authors 1 and 2) to check the viability of search terms before Author 1 ran the first search phase in April 2022.Studies were exported into Endnote X9 software and duplicates were removed before the list was exported into Rayyan software [41], where a further duplication check was conducted.Author 1 then completed the title and abstract screening and full-text screening in collaboration with Author 2. Any disagreements were planned to be discussed and resolved by Authors 3 and 4, but no disagreements were found.An updated search was run in February 2024 by Author 1, which identified a further 303 papers after the removal of duplicates.Author 1 and Author 4 conducted title and abstract screening independently, with no disagreements reported.Author 1 then completed the full-text screening in collaboration with Author 4. Full details of the selection process are shown in Figure 1.

Selection Process
Preliminary searches were conducted by two authors (Authors 1 and 2) to check the viability of search terms before Author 1 ran the first search phase in April 2022.Studies were exported into Endnote X9 software and duplicates were removed before the list was exported into Rayyan software [41], where a further duplication check was conducted.Author 1 then completed the title and abstract screening and full-text screening in collaboration with Author 2. Any disagreements were planned to be discussed and resolved by Authors 3 and 4, but no disagreements were found.An updated search was run in February 2024 by Author 1, which identified a further 303 papers after the removal of duplicates.Author 1 and Author 4 conducted title and abstract screening independently, with no disagreements reported.Author 1 then completed the full-text screening in collaboration with Author 4. Full details of the selection process are shown in Figure 1.

Data Collection and Extraction
A data extraction form was developed in Excel by all authors, and studies included in the review underwent data extraction, whereby the following information was extracted for each paper: (i) Study characteristics: study design, the country where the study was conducted, exclusion/inclusion criteria, and publication year; (ii) intervention characteristics: types of resilience interventions used in the study with detailed descriptions, delivery modes, intervention facilitator roles, number of sessions in the intervention, and

Data Collection and Extraction
A data extraction form was developed in Excel by all authors, and studies included in the review underwent data extraction, whereby the following information was extracted for each paper: (i) Study characteristics: study design, the country where the study was conducted, exclusion/inclusion criteria, and publication year; (ii) intervention characteristics: types of resilience interventions used in the study with detailed descriptions, delivery modes, intervention facilitator roles, number of sessions in the intervention, and duration of intervention; (iii) Study population characteristics: graduate or undergraduate, sample size, cohort type, participants' mean age, standard deviation, participant's identified gender, and ethnicity; (iv) Outcome measures: measures used for mental health and well-being, as well as any measures used to assess the impact of treatment on the targeted construct Educ.Sci.2024, 14, 510 5 of 25 (e.g., resilience or mindfulness); (v) Findings: for each identified outcome, we extracted the number of participants in the intervention and control group, mean and standard deviations at both T1 (pre-intervention) and T2 (immediately post-intervention), confidence intervals, and effect sizes when possible.Follow-up points beyond T2 were not extracted.Data extraction was completed by Author 1 and any ambiguities were discussed with all authors.

Quality Assessment
Risk of bias (RoB) assessment was conducted by Author 1 in collaboration with Author 2 using the National Heart Lung and Blood Institute tool [42], previously adapted for education-based studies [35].Risk of bias rating was calculated by the percentage of RoB criteria rated 'yes': rating categories included very poor (0-24%), poor (25-49%), fair (50-74%), and good (75-100%), as previously adopted in reviews focused on education and well-being [35].

Data Analysis
For each study in this systematic review, mean differences (pre/post-intervention) for each outcome were calculated using the publications' reported raw scores.'Cannot Determine' [CD] notation was used if the mean differences could not be calculated due to missing raw data.Outcome data (for pre-post intervention) were extracted; 95% confidence intervals along with Cohen's d were calculated for each study with a control group using STATA15 (StataCorp, 2017).Raw data used in Cohen's d calculation included the sample size, mean, and standard deviation (pre and post-intervention).In the case of studies containing a control group, it was stated whether the results significantly favoured the intervention or control or if there was no significant difference between conditions (NS).Where studies had no control group, Cohen's d values and 95% confidence intervals were reported.The direction of effect was reported if effect sizes could not be calculated due to missing raw data, according to the results section of individual studies.
Populations and interventions were very heterogeneous, and the number of studies was relatively small, making a meta-analysis inappropriate.Instead, we employed the vote-counting method as defined by SWiM guidelines [43].Vote counting was conducted by intervention type to judge whether the different types of resilience intervention improved measures of mental health and well-being, as well as measures of resilience and mindfulness, where collected.For each type of intervention, the percentage of studies showing an effect on the outcomes was reported along with the binomial test, indicating the probability of the results if the intervention was ineffective (i.e., equal to 0.5), and the 95% confidence intervals for the percentage of effects favouring the intervention [44].The syntax 'bitesti X Y 0.5' was used to calculate the binomial test calculated through STATA 15 (StataCorp, 2017).Intervals were calculated following the syntax 'cii proportions X Y, level (95)'; X means the number of effects and Y means the number of intervention-favouring effects.Additionally, given the rise in online delivery, we examined the association between the likelihood of finding a positive effect and the mode of delivery across interventions using chi-square analysis for the main outcomes reported.

Overview of Included Studies
Details of all the study characteristics are shown in Table 1 and summarised in the following sections.

Intervention Categories
The theoretical foundations of the interventions were deduced from the descriptions by the study authors and the mediating processes that were targeted according to the framework developed previously [18], as well as previous systematic reviews in this area [92] and allowed categorisation into distinct types of intervention.

Results of Individual Studies
The effect estimates of the outcomes in each individual study are provided in Table 2 and are summarised by intervention type below.

Coaching Interventions
There were four coaching studies [58,61,73,83].Three studies measured the effects on depression, anxiety, and resilience [58,61,83].The three studies were consistent in reporting no effects on resilience.Two of the three studies also reported no effect on depression [58,83], whilst one reported a positive effect of the intervention on depression [61] (33% (95% CI 0.8-90.6%),p = 1.00).For anxiety, two of the three studies measuring this reported a positive effect of the coaching intervention on anxiety [61,83] (67% (95% CI 9.4-99.2%),p = 1.00).The final study measured only positive affect but reported that the intervention had a positive impact on this [74], (100% (95% CI 2.50-100%), p = 1.00).Collectively, the results of these studies suggest no impact on resilience or depression but possibly some beneficial effects on anxiety.
Well-being measures were less commonly collected in this type of intervention.Three studies found that the intervention had a positive effect on well-being [69,72,77] (100% (95% CI 29.2-100%), p = 0.25), one study reported no effect on happiness [56], another noted a positive effect on a positive mental health measure [88] (100% (95% CI 2.5-100%), p = 1.00), and a final study showed that life satisfaction was helped by two of the three interventions tested, most notably the mindfulness-based stress reduction programme [81] (100% (95% CI 2.5-100%), p = 1.00).Despite the limited number of studies available, most reported a positive effect on well-being.
Measures relating to well-being were less frequently assessed in these studies.Wellbeing was measured in two studies, with both reporting no effect [66,80].Self-efficacy [50] (100% (95% CI 2.5-100%), p = 1.00) and positive affect [59] (100% (95% CI 2.5-00%), p = 1.00) were found to be improved after the intervention, whilst quality of life was unaffected [76].The few studies examining well-being and the range of measures mean that it is premature to make any conclusions on the impact of skills-based training interventions on well-being.
Resilience was measured in eight studies.Seven reported no effects [51,52,63,66,68,76,83] and only one reported a beneficial effect of the intervention [74] (12.5% (95% CI 0.32-52.6%),p = 0.07).Mindfulness was measured in one study but found not to be affected [50].These studies indicate that the skills-based training interventions are not impacting resilience as measured in these studies.

Mode of Delivery
Due to the relatively small number of studies overall, it was not possible to examine whether the likelihood of a positive outcome was associated with a specific mode of delivery within each intervention category.However, across intervention types, we examined whether there was an association between a positive effect being found and mode of delivery, considering only face-to-face or online delivery as only two studies took a hybrid approach.There was no significant association between the likelihood of a positive effect and delivery mode for measures of depression (χ 2 (1) = 0.22, p = 0.485), anxiety (χ 2 (1) = 0.20 p = 0.500), stress (χ 2 (1) = 0.01, p = 0.633), well-being (χ 2 (1) = 3.73, p = 0.143), resilience (χ 2 (1) = 0.27, p = 0.554), or mindfulness (χ 2 (1) = 0.00, p = 0.774).

Discussion
The aim of this review was to answer the research question "What is the effect of resilience interventions on mental health and well-being in university students?".We identified 47 publications for inclusion, all of which had assessed at least one measure of mental health or well-being.As expected in an educational context [36], whilst the single biggest design category was the RCT, NRCT and pre-post designs made up a significant proportion of the studies included.Most also focused on undergraduate students without recruiting specific disciplines of study (Table 1).However, the characteristics of participants were not generally described in detail with studies not including details pertaining to ethnicity, for example.We categorised the interventions into coaching, psychoeducation, mindfulness, and skills-based training, with the latter two being the most prominent.Mental health measures varied but the most frequently assessed constructs were depression, anxiety, and stress, in line with the only previous systematic review in this area, which identified measures of depression and stress [18].Well-being was also measured with a range of scales, including those focused on mental well-being, self-efficacy, happiness, and measures of quality of life or life satisfaction (Table 2).A total of 38% of the studies also measured resilience and 23% measured mindfulness.
Analysis of the results of the different types of interventions revealed that coaching interventions had no clear effects on resilience or depression but may be beneficial to anxiety.Most of these studies were rated as fair in terms of quality, although the small number of studies overall limits any conclusions that can be drawn.A previous systematic review found beneficial effects on depression and resilience overall but this included only one coaching study [18].For psychoeducation intervention, a more diverse selection of outcomes was recorded but results were inconsistent for measures pertaining to mental health and well-being with little evidence of impact.This inconsistency was found despite most studies using this approach being rated as fair or good in terms of quality.Only two studies examined resilience in response to these interventions and found no impact.Most studies using mindfulness interventions were also deemed fair or good in the quality assessment and these interventions demonstrated more consistent results for stress and anxiety, both of which appeared to benefit from the intervention.The latter is in line with a previous review, which reported the beneficial effects on stress but did not examine anxiety [18].There were also some indications about the beneficial effects on depression and distress but there were too few studies to make firm conclusions.Well-being measures were more limited for mindfulness interventions, but they did reveal positive effects.Despite the positive impacts of these interventions on anxiety and stress, measures of mindfulness did not improve post-intervention, and there was no consistent impact on resilience.This could indicate that whilst the interventions were beneficial, the mechanism may not be as intended.Skills-based training interventions resulted in inconsistent findings for depression and anxiety but did appear helpful for test anxiety and stress.There were fewer studies on well-being, and they reported mixed results, making it inappropriate to draw any firm conclusions.As with mindfulness interventions, although some improvements were seen in mental health measures, these were not generally accompanied by improvements in resilience or mindfulness, which could suggest this type of training is tapping into another important construct.Furthermore, even where beneficial effects were found, it is noteworthy that almost half of the studies in this category were recorded as having poor quality ratings, with the remainder rated as fair, meaning no good quality studies were included.The lack of impact of resilience-based interventions on resilience itself is in line with previous findings that resilience interventions had limited effects on the resilience of university students [92].Additionally, even where an effect was found, it was deemed to be of a small effect size [18].This is arguably concerning given resilience is a key graduate attribute [4] and, thus, its development is important for HE outcomes in general as well as mental health and well-being.
It should be acknowledged that there was a huge variety in the types of interventions employed in the reviewed studies (Table 1).Apart from the four categories of intervention, they also differed in terms of the number of sessions, facilitator type, and number of students involved.Because of the small number of studies overall and the diversity of approaches, it was not possible to assess statistically whether characteristics such as the facilitator type impacted the likelihood of a study reporting a positive effect.However, for the mode of delivery, there appeared to be no impact on whether there was a positive effect recorded for the most common outcomes, suggesting that any mode could be effective.
Despite this review aligning with (and extending) previous research, there are several limitations to the study.Firstly, due to the limited number of studies and missing data, we were not able to conduct a meta-analysis, although, in line with SWiM guidelines, we utilised vote-counting [43].Secondly, we focused only on quantitative studies, which may have resulted in important findings being missed in qualitative research.Similarly, we utilised only peer-reviewed journal articles.It is possible that grey literature would have provided different sources for inclusion and future reviews should consider this.Finally, we focused on students attending campus-based degree programs, even if the intervention was delivered online.This was done because the majority of degree programs offered are campus-based rather than distance learning, and distance learning appears to impact resilience differently [40].However, future research should consider this cohort as well.

Conclusions
Although there are limitations to the current review, this is, to our knowledge, the first systematic review of the effect of resilience interventions on mental health and well-being, which includes a range of study designs and centres on higher education.Overall, this review suggests that there may be some benefits to such interventions for stress and anxiety but they are not necessarily accompanied by changes in measures of resilience or mindfulness, which may indicate different underlying mechanisms.Furthermore, this review demonstrates that more high-quality studies are required, as has been previously noted in reviews in this area [32][33][34]108].This is particularly noticeable for skills-based interventions.A greater number of studies would also allow for a comparison of intervention features, which was not possible here beyond the mode of delivery.
, ANX = Anxiety, RES = Resilience, MIND = Mindfulness CG = Control Group; IG = Intervention Group; N/A = Not Applicable; NR = Not Reported; NS = Not Significant; Sig.= Significant; * = p < 0.05.CD = Cannot Determine.DASS = Depression Anxiety and Stress Scale, MAAS = Mindful Attention Awareness Scale, SAS = Stressor Appraisals Scale, GAD-7 = Generalized Anxiety Disorder Assessment, Q-LES-Q-SF = The Quality-of-Life Enjoyment and Satisfaction Questionnaire Short Form, PHQ-9 = Patient Health Questionnaire, PSS = Perceived Stress Scale, BDI = Beck Depression Inventory, BDI-II = Second Version of Beck Depression Inventory, STAI = State Trait Anxiety Inventory, FFMQ = Five-Facet Mindfulness Questionnaire, FFMQ-SF = Five-Facet Mindfulness Questionnaire Short Form, CORE 10 = Clinical outcomes in Routine Evaluation 10, SES = Self Efficacy Scale, CD-RISC = Connor Davidson Resilience Scale, HADS = Hospital Anxiety and Depression Scale, LSS = Life Satisfaction Scale, PANAS = Positive and Negative Affect Schedule, BAI = Beck Anxiety Inventory, BRS = Brief Resilience Scale, SWEMWBS = Short Warwick Edinburgh Mental Well-being Scale, CES-D = Center for Epidemiological Studies Depression Scale, WEMWBS = Warwick Edinburgh Mental Well-being Scale, WHOQOL-BRIEF = World Health Organization Quality of Life Scale, GHQ-12 = General Health Questionnaire, WTAS = Westside Test Anxiety Scale, MASQ = Mood and Anxiety Symptoms Questionnaire, SWLS = Satisfaction with Life Scale.Final column: Sig.improvement used for designs without a control.NB.Information not reported within the table was not reported in the reviewed studies.β : CWYC = Control What You Can intervention groups (IG1) encouraging participants to focus on things they could control.CDC = Centers for Disease Control and Prevention recommended stress management (IG2), MV = Mastery Version AV = Attention Version.Sky = SKY Campus Happiness, MBSR = Mindfulness-based stress reduction, EI = Foundations of emotional intelligence.REBT = Rational Emotive Behaviour Therapy PE = Psychoeducation.

Table 1 .
Summary of study characteristics.

Table 2 .
Summary of effects reported for studies for measures of mental health and well-being.Where studies also collected data on resilience or related constructs these were included.