Authors: Dalal Hammoudi Halat, Waqas Sami, Abderrezzaq Soltani, Ahmed Malki
Categories: Systematic Review, Academics, Faculty, Intervention, Mental health, Systematic review and meta-analysis, University
Source: BMC Public Health
Authors: Dalal Hammoudi Halat, Waqas Sami, Abderrezzaq Soltani, Ahmed Malki
While there is growing evidence highlighting the prevalence of mental health concerns among university faculty, few studies have examined mental health interventions in this population. The objective of this systematic review and meta-analysis was to collect and critically appraise the available evidence about the effectiveness of interventions designed to improve the mental health of faculty.
A systematic search was conducted by searching PubMed, EMBASE, Scopus, Web of Science, and CINAHL to identify relevant studies published in English language from January 1st, 2000 until October 1st, 2023. The search focused on studies done on academic faculty to describe interventions or support programs aimed at improving mental health outcomes, with comparison of mental health data before and after the intervention and an improvement in mental health as study outcome. A random effect meta-analysis method was used to estimate the effectiveness of interventions on faculty mental health.
Ten publications with 891 participants from 2,217 retrieved records were included. The random effect model showed substantial heterogeneity (I^2^ = 84.8%, 95% CI: 73.8 − 91.2%, p < 0.001). The pooled SMD was − 1.41 (95% CI: -2.81–0.004) showing a large effect, and it significantly favors the use of intervention for reducing mental health issues among faculty members. The effect size estimates for all included studies ranged from small to large, showing the positive effect of intervention on faculty mental health. Multimodal inference analysis showed that, of the many studied factors for faculty mental health, the region was the most important predictor of intervention effectiveness. However, when the significance of quantitative moderators was tested using meta-regression, age (p = 0.9491) and duration of intervention (p = 0.1284) were not statistically significant.
Interventions aimed at enhancing the mental health of university faculty were overall significant; however, individual studies showed heterogeneous results. Making efforts to enhance the mental health of faculty is crucial and has been proven effective; nevertheless, the existing evidence necessitates further research in this area. For interventions to be effective, it is imperative to tailor them to the specific environment and to the unique characteristics of faculty members.
CRD42023490388
The online version contains supplementary material available at 10.1186/s12889-024-20402-2.
The latest report from the World Health Organization (WHO) on transforming mental health for all reveals that over one billion people globally, are living with a mental health disorder [1, 2]. Although anxiety and depression account for the largest burden of mental health issues with prevalence of almost 31% and 29% respectively, other mental health issues, such as bipolar disorder, eating disorders, post-traumatic stress disorder and others, affect millions of people worldwide [3]. This is often aggravated by the fact that most societies and the majority of health and social care systems neglect mental health or offer it weak and deficient provision, hence, do not adequately provide nor prioritize the care and support people need [4]. In the context of funding for mental health service, it is reported that, on average, about 2% only of national health budgets is allocated to this sector [5, 6].
Given that workers often spend the majority of their waking hours at work, the importance of occupational mental health has long been recognized [7]. Work settings can either exert positive influences on mental well-being, or conversely, lead to mental health challenges precipitated by job stress, conflicts, organizational cultures, and excessive work demands [8]. The ramifications of poor mental health in the workplace extend to various stakeholders, resulting in diminished productivity, strained workplace relations, and behavioral issues [9].
Evidence shows that mental health disorders are prevalent among academic faculty working at universities [10–14]. Recently, over 70% of faculty from 16 countries had moderate to high levels of psychological distress [15]; moreover, a minimum of 30%, 63%, and 26% of the faculty in another study showed at least moderate levels of depression, anxiety, and stress respectively [16], while up to 55% of academics reported depression, anxiety, and burnout in a study from 10 big universities in the US [17]. A job in academia can be rewarding; however, faculty are expected to accomplish their teaching load, collaborate with peers, publish high quality research, and provide services to the university and the wider community. With such array of tasks, and the pressure to deliver expectations within short timeframes, mental health and well-being among faculty may take a backseat [18]. Several factors affect well-being in academia like the busy nature of academic life, with different responsibilities alongside personal and family commitments [19]. Other factors are dependent on the profound changes in higher education structures, with more marketization, globalization, and increased generation of pedagogical, managerial, and scholarly tasks [20–22]. Moreover, the interaction of faculty with students may have its toll on the well-being of faculty, especially with increased class sizes, academic entitlement, and engagement of students in faculty evaluation as key stakeholders [23–27]. With COVID-19 pandemic, the levels of stress among university faculty increased [28], as a result of change in the academic norms, social disruption brought about by lockdown, and the need to suddenly adapt to remote education, while work-life balance was adversely affected [29–31]. Such rising frequency of mental issues and burnout as occupational phenomena for academics, along with associated influence on the academic community, stand as warning signs that actions in this field are much needed [32].
Generally, the workplace is assumed to be an ideal setting for mental health interventions and their uptake by the working population [33, 34]. Targeting the workplace by system-level interventions, including organizational, cultural, social, and physical facets was shown to be capable of creating efficient and sustainable health and wellbeing [35]. Several studies have investigated the effect of interventions in the workplace including physical activity [36, 37], web-based programs [38], cognitive behavioral therapy [39, 40] and others [41–43], on the mental health of employees. However, in higher education, the evidence of the effectiveness of mental health intervention programs is still inconclusive and primarily dedicated to students [44]. Improving mental health of faculty requires a proactive, systematic approach that prioritizes the well-being of academics, and is rooted in organizational planning, leadership support, and efforts by faculty themselves [18]. For instance, interventional strategies including counseling, leadership training, reward mechanisms, and improved communication across university campuses have been described [45]. Likewise, an ordered interventional format based on proper organization and planning, building strong networks, self-assessment of mental health, and utilizing resources for mental health support, is also another structure previously described for an efficient faculty mental health improvement program [19].
Although published research indicates the importance of wellness cultures in improving mental health of university faculty [17], and calls for active interventions in this regard [18, 19, 46], there is a knowledge gap that probably obstructs the creation of precise and impactful interventions. To the best of our knowledge, at present, systematic synthesis of interventions designed to improve mental health and well-being of faculty are scarce in literature. Hence, the aim of this systematic review and meta-analysis is to synthesize the available evidence on interventions or support programs targeting faculty mental health, to estimate their effectiveness, by comparing the mental health of faculty before and after such interventions.
The protocol for this study was registered on PROSPERO (CRD42023490388) and was conducted in line with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement [47]. The PRISMA checklists for this article are available in Supplementary files 1 and 2.
A systematic search of literature was conducted on five electronic databases that index articles from various disciplines including health and education. These databases were PubMed, Embase, Scopus, Web of Science, and CINAHL. All databases were searched from January 1st, 2000, until October 1st, 2023. A search strategy was used to identify eligible studies for inclusion in the systematic review. We included mental health, faculty, university or higher education institution, and intervention as our keywords. We also manually identified some studies by hand search on studies’ references. In two-weeks period after the initial search, a repetition of the search was done to verify and update search results and include any missing articles. During the search, no restrictions were placed on publication status as open access full text or restricted access; however, we limited the search to studies in peer-reviewed journals and in English language. The PICO model, that covers the targeted population, type of intervention, a comparative group for intervention, and study outcomes, was used. The PICO items and the search criteria are shown in Table 1.
In order to eliminate any duplicate entries, all records were imported into EndNote, a reference management software. Once duplicates were removed, two authors (DHH and WS) screened the remaining non-duplicate titles and abstracts based on pre-established eligibility criteria. Any discrepancies that emerged upon screening were resolved through a consensus between the two authors.
Table 1PICO strategy description and search criteria used in this studyPICO items P (Patient or population) Faculty members within higher education institutions I (Intervention) Interventions or support programs aimed at improving mental health outcomes, including reduced stress, burnout, anxiety, depression, increased job satisfaction, or overall well-being C (Control or comparison) Comparison to mental health prior to the intervention O (Outcome) Improvement in mental health Search criteria
Database
Search query PubMed(((mental health[Title/Abstract]) AND (faculty[Title/Abstract])) AND (university[Title/Abstract])) OR (higher education institution[Title/Abstract]) AND (intervention[Title/Abstract])))Embase(‘mental health’:ab, ti AND ab, ti AND ab, ti OR ‘higher education institution’:ab, ti) AND ab, ti NOT ab, tiScopus( TITLE-ABS-KEY ( mental AND health ) AND TITLE-ABS-KEY ( faculty ) OR TITLE-ABS-KEY ( higher AND education AND institution ) AND TITLE-ABS-KEY ( intervention ) ) AND PUBYEAR > 1999 AND ( LIMIT-TO ( LANGUAGE , “English” ) )Web of Science((((((TS=(Mental health )) AND TS=(faculty)) AND TS=(university)) OR TS=(higher education institution)) AND TS=(Intervention)) NOT TS=(student)) NOT TS=(patient)CINAHLmental health AND faculty AND university OR higher education institutions AND intervention NOT students NOT ( patients or clients )
Among all initial identified records, and for a study to be included in the systematic review, the inclusion criteria were as (i) The study examined a population of faculty who work in a higher education institution as their primary job; (ii) The study applied an interventional approach, in which participating faculty received a mental health intervention, and a comparison was realized before and after the intervention to measure mental health outcomes; (iii) The main study findings were reported in terms of a tangible outcome measure in the form of quantitative description of mental health of faculty. The included studies can investigate a positive mental health outcome (for instance, well-being) or a negative one (for instance, depression, anxiety, or stress). The presence of a control group in the study, or a randomized controlled trial study design were possible for inclusion if other inclusion criteria were met, but were not necessary.
Studies which were simply descriptive of faculty mental health such as prevalence studies, or those which related faculty mental health to other factors (like physical health or demographics) without an intervention being done, were excluded. Studies on mental health of students or school teachers outside tertiary education setting were excluded. Case reports, book chapters, pre-prints, study protocols with unpublished results, reviews, qualitative investigations, and conference papers were excluded as well.
Initial screening of titles and abstracts from the original search was first conducted independently by two reviewers with experience in mental health research and in conducting systematic review and meta-analysis (DHH and WS). Next, full texts of relevant studies were reviewed, and only those satisfying the inclusion criteria were retained. Potentially eligible full texts were reviewed in depth, and the process was repeated for consistency purposes, with documentation of eligible studies. Discrepancies were resolved by discussion and thorough reading of the full texts. Reasons for exclusion were discussed and confirmed to fall within the exclusion criteria. When accessible, reference lists and grey literature were searched and discussed as well. The panel of four authors participated in different discussions and revisions of the systematic review and meta-analysis steps. Authors had experience in mental health in academia, conducting systematic review and meta-analysis, and were senior academics with oversight on faculty, programs, and courses.
Extraction of data from eligible studies was done by two authors (DHH and WS). Any inconsistency in the information extracted was resolved through discussion among them. The extracted information included the following first author, year, title of the study, country, faculty age, gender, and specialty, sample size, type and description of the intervention used, duration of the intervention, specific mental health measure or scoring instrument used with minimum scores, maximum scores, and number of items, pre-intervention and post-intervention mental health scores with standard deviations, significance of the intervention, and the mean of pre- and post-intervention differences observed in mental health measurements. Based on the study design the mean differences between pre- and post-values were taken from studies where there was no control group. One author (WS) piloted the extraction tables and coded the extracted data. Another author (DHH) realized full data extraction, and both authors reviewed all retrieved records for accuracy.
The data was analyzed using R Core Team (R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, 2023. URL http://www.R-project.org/). We used the packages “meta”, “metafor” and “dmetar” for conducting the analysis, whereas the package “esc” was used to calculate the effect size and standard error (SE). The Standardized Mean Difference (SMD) was calculated by taking the pre-post mean differences and dividing it by pre-SD. The SMD is a statistical measure used in meta-analysis to combine the results of studies that use different measurement scales or units. This allows for a more consistent comparison of effects across studies. In addition, the Hedges’ g correction factor was applied on SMDs to adjust for studies with small sample sizes. The effect sizes were interpreted as “small”, “medium” and “large” based on 0.2, 0.5 and 0.8 criteria. The Cochran’s Q and I^2^ tests were applied to quantify heterogeneity [48] using the suggested criteria of 25% (low), 50% (moderate), and 75% (high). The between study variation tau-squared (τ^2^) was estimated using restricted maximum likelihood (REML) method. Hartung-Knapp (hakn) adjustment was also made for the random effects model [49]. To identify the source of heterogeneity, a sub-group analysis was performed, and the significance of various moderators was assessed using forest plot with statistical tests. Funnel plot, Egger’s test and Trim & Fill methods [50, 51] were used to assess the publication bias. A multi-model inference analysis was also carried out to identify predictors based on their importance. Meta regression was applied to observe the significance of quantitative moderators (age and duration of intervention). A p-value less than 0.05 was considered statistically significant.
Upon the initial database search, 2,217 records were obtained. After duplicates removal, the title and abstract of 2,021 records were screened independently by two authors (DHH and WS), based on the eligibility criteria for this study. After title and abstract screening, 2001 articles were excluded and 20 articles were retained for full text review. Of these, 10 were selected as fitting the inclusion criteria while 10 were excluded as they do not document pre-and post-intervention mental health measurements, or estimate knowledge or literacy rather than mental health scores, or do not fit the inclusion criteria. The 10 retained articles were included in both the systematic review and the meta-analysis, and encompassed a total of 891 participants. The PRISMA flowchart for this study is shown in Fig. 1. The main characteristics of the included studies and the essential findings are described in Tables 2 and 3 respectively.
Fig. 1PRISMA flowchart for this study
Table 2Characteristics of included studiesStudy numberFirst author and date of publicationCountryAverage age of facultyFaculty group by specialtyFaculty genderStudy designDescription of the interventionType of interventionIntervention duration in weeks1Brewer et al., 2019 [52]AustraliaNot specified46% in health sciences87% femalesPre- and post-intervention surveysFace-to-face course in leadership development, aimed at enhancing the understanding of leadership in higher education, workplace capacity building, and resilience enhancement strategies.Mental12Garcia et al., 2023 [53]USA31Not specified70.6% femalesTreatment and control groups with pre- and post-intervention assessmentsFibit program, whereby participants were supplied by a Fitbit device linked to their mobiles to log their activity, and were instructed on breathing meditation and mindfulness exercises.Mixed23Guerra et al., 2022 [54]Italy36.8Dental Faculty41.2% femalesRandomized controlled clinical trial with pre- and post-intervention mental score measurementsA range of exercises including yoga, meditation, martial arts, and training in empathic communication.Mixed44Ikiugu et al., 2022 [55]USA26.73Not specified90.48% femalesPre- and post-intervention surveyUse of meaningful and psychologically rewarding occupations as therapeutic media, including physical exercise, arts, and crafts.Mixed65Lim et al., 2023 [56]Malaysia24Not specified68.9% femalesRandomized controlled study with measurements as baseline and after interventionDigital mental health intervention, called ThoughtFullChat, which consists of one-on-one, asynchronous, text-based coaching with certified mental health professionals, and self-guided tools, such as daily mood loggings, professionally curated guided lessons, and personal emotional health assessment.Mental126Ogbuanya et al., 2017 [57]Nigeria27.85Electronics workshop instructors42.9% malesPretest-posttest questionnaire with a waitlist control groupRational emotive behavior coaching derived from evidence-based Rational Emotive Behavioral Therapy (REBT), and including sensory awareness training, Socratic questioning technique, Future Self technique, value identification, Wheel of Life technique, Big “A” Agenda technique, relaxation and rational emotive imagery techniques.Mental147Rodrigues-Jimenes et al., 2022 [58]Spain45Not specified72.7% femalesControlled, randomized, pre-post experimental designHatha Yoga, consisting of postures and counter postures aimed at increasing flexibility and strength, with a focus on body alignment and lateralization.Mixed88Sood et al., 2014 [59]USA47.4Radiology45% femalesRandomized intervention and control groups with pre- and post-intervention questionnaireSMART program of attention and interpretation, by focus on the external world, deferring unrefined judgments, cultivating and guiding interpretations by five higher-order gratitude, compassion, acceptance, meaning, and forgiveness. Also, deep diaphragmatic breathing exercises for relaxation were practiced.Mixed129Stefansdottir and Sutherland 2005 [60]IcelandNot specifiedNot specifiedNot specifiedPre- and post-intervention measurement of a stress indicatorPreference awareness education workshops and homeworks, explaining transactional approach to stress, communication styles, planning self-development, and planning how to handle work stressors.Mental710Ugwoke et al., 2017 [61]Nigeria38.65Technical college teachers50.54% femalesPretest-posttest questionnaire with a waitlist control groupRational emotive health education intervention, including relaxation, stretching, and cognitive restructuring exercises.Mental10Mental: refers to a mental intervention type only; Mixed: refers to an intervention where both mental and physical (like exercise) methods were used
Table 3Main findings of included studiesStudy numberFirst author and date of publicationSample size pre-interventionSample size post-interventionMental health instrument usedMean mental health score pre-interventionMean mental health score post-interventionSMDStudy findings on the effect of intervention on mental health of faculty1Brewer et al., 2019 [52]3817K-1018.5816.06-0.35Distress levels pre-intervention scored an average of 18.5 ± 7.21 on the K-10 scale, while the post-intervention levels scored an average of 16.05 ± 5.11, indicating that the program was positively received by participants.2Garcia et al., 2023 [53]1817HADS10.418.06-0.49The levels of anxiety in participants decreased post-intervention compared to baseline, CI [0.48, 4.22].3Guerra et al., 2022 [52]1817MCS45.952.80.21The mental score levels significantly improved from pre– (49.5 ± 44.45) to post-intervention (52.8 ± 50.05; p = 0.002).4Ikiugu et al., 2022 [55]33RAND SF-36 Energy57.8665.710.63A statistically significant effect on well-being was observed between pre-intervention measurement (57.86 ± 12.51) and post-intervention measurement (65.71 ± 15.54; p = 0.03).5Lim et al., 2023 [56]197167DASS-214.554.17-0.11A significant decrease was observed in depression (P = 0.02) and anxiety (P = 0.002) scores compared before and after the intervention.6Ogbuanya et al., 2017 [57]5555EWIOSS44.1815.16-5.41A significant reduction of stress was observed between pre-intervention levels (44.18 ± 5.36) and post-intervention levels (15.66 ± 5.07; p = 0.00).7Rodrigues-Jimenes et al., 2022 [58]1111PSS17.8812.13-0.88A significant reduction of stress was observed between pre-intervention levels (17.88 ± 6.56) and post-intervention levels (12.13 ± 6.96; p = 0.012).8Sood et al., 2014 [59]1311PSS2519.6-0.73 A significant reduction of stress was observed between pre-intervention levels (25.0 ± 7.4) and post-intervention levels (19.6 ± 5.6; p = 0.020).9Stefansdottir and Sutherland, 2005 [60]2727OSI2.9882.856-0.16The mean level of perceived stress decreased from 2.998 ± 0.826 to 2.856 ± 0.737 after the intervention, which was not statistically significant change (p = 0.244). Upon subgroup analysis, extravert-thinking faculty and those suffering high pre-intervention stress reported decreased stress levels post-intervention.10Ugwoke et al., 2017 [61]9393TSQ58.6732.15-2.95Significant decrease in stress levels (p < 0.001) from pre-intervention at 59.67 ± 9.00 to post-intervention at 32.15 ± 3.17 was observed.SMD: Standardized Mean Difference. K-10: Kessler Psychological Distress Scale. HADS: Hospital Anxiety and Depression Scale. MCS: Mental Composite Score. RAND SF-36 Energy: 36-Item Short Form Health Survey developed by RAND Corporation, USA. DASS-21: Depression, Anxiety, and Stress Scale-21 items. EWIOSS: Electronic Workshop Instructors’ Occupational Stress Scale. PSS: Perceived Stress Scale. OSI: Occupational Stress Indicator. TSQ: Teachers’ Stress Questionnaire
For studies included in the systematic review and meta-analysis, the Cochrane Risk of Bias 2.0 (RoB 2.0) tool [62] was used to evaluate the risk of bias in randomized controlled trials (RCTs) [54, 56, 58, 59], whereas the NIH Quality Assessment tool was used for before-after (pre-post) studies [63]. The different domains under which the two tools were used are presented through Fig. 2; Table 4 respectively.
Fig. 2Risk of bias assessment across different domains using the RoB 2.0 tool [42]
Table 4Risk of bias assessment across different domains using the NIH tool [63]StudyQ1Q2Q3Q4Q5Q6Q7Q8Q9Q10Q11Q12Total ScoreQuality RatingBrewer et al., 2019 [52]YYYYNRYYNNRNNY7/12 (58.33%)FairGarcia et al., 2023 [53]YYYYNYYNRYNYY9/12 (75.00%)GoodIkiugu et al., 2022 [55]YYYYYYYNYNNY9/12 (75.00%)GoodOgbuanya et al., 2017 [57]YYYYNRYYNNRNNY7/12 (58.33%)FairStefansdottir et al. and Sutherland, 2005 [60]YYYYNRYYNNRNNY7/12 (58.33%)FairUgwoke et al., 2017 [61]YYYYNYYNRYNYY9/12 (75.00%)GoodQ1: Was the study question or objective clearly stated?, Q2: Were eligibility/selection criteria for the study population prespecified and clearly described?, Q3: Were the participants in the study representative of those who would be eligible for the test/service/intervention in the general or clinical population of interest?, Q4: Were all eligible participants that met the prespecified entry criteria enrolled?, Q5: Was the sample size sufficiently large to provide confidence in the findings?, Q6: Was the test/service/intervention clearly described and delivered consistently across the study population?, Q7: Were the outcome measures prespecified, clearly defined, valid, reliable, and assessed consistently across all study participants?, Q8: Were the people assessing the outcomes blinded to the participants’ exposures/interventions?, Q9: Was the loss to follow-up after baseline 20% or less? Were those lost to follow-up accounted for in the analysis?, Q10: Did the statistical methods examine changes in outcome measures from before to after the intervention? Were statistical tests done that provided p values for the pre-to-post changes?, Q11: Were outcome measures of interest taken multiple times before the intervention and multiple times after the intervention (i.e., did they use an interrupted time-series design)?, Q12: If the intervention was conducted at a group level (e.g., a whole hospital, a community, etc.) did the statistical analysis take into account the use of individual-level data to determine effects at the group level?; Total Score: Number of yes; NR: not reported; N: no; Y: yes. Quality Rating: Poor < 50%, Fair 50–75%, Good ≥ 75%
The Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach [64] was used to assess the overall certainty of the evidence from the studies included in the systematic review. This approach considered the risk of bias assessment, inconsistency of results, indirectness of evidence, imprecision of effect estimates, and potential publication bias, as shown in Table 5.
Table 5GRADE assessment results of the included studiesStudy designNumber of studiesObserved common findingRisk of BiasInconsistencyIndirectnessImprecisionOthersCertaintyRandomized Controlled Trials (RCTs)4Overall significant improvement in mental health scores post-intervention.Low to moderateLowLowModerate (due to small sample sizes in some studies)NoneModeratePre- and Post- Intervention studies6General improvement in mental health outcomes, though with some variability between studies.Low to moderateModerateLowLow to moderateNoneLow to Moderate
The meta-analysis shown in Fig. 3 (forest plot) synthesized data from a total of 10 studies included in this review to evaluate the impact of various interventions on the mental health of faculty members. Employing a random effects (RE) model, the analysis produced a pooled SMD of -1.41, indicating that, on average, intervention participants experienced a positive effect in mental health issues compared to controls. The 95% confidence interval for SMDs ranged from − 2.81 to 0.004, which implies that the effect of the intervention was significant and the likelihood that this is a chance finding is very small. However, despite the overall positive outcome, the individual studies showed a wide range of effect sizes. For instance, Brewer (2019) [52] and Garcia (2023) [53] reported small, non-significant effect sizes with SMDs of -0.35 and − 0.49, respectively. On the other hand, Ogbuanya (2017) [57] presented a very large and significant negative effect size of -5.41, suggesting a substantial improvement in mental health following the intervention. In contrast to these findings, two studies, Guerra (2022) [54] and Ikiugu (2022) [55] actually reported positive SMDs of 0.21 and 0.63, although with very wide confidence intervals, indicating a high degree of uncertainty about these estimates and suggesting that in these particular studies, the interventions might not have had the intended beneficial effects. The heterogeneity across these studies was quantified by an I^2^ statistic of 84.8%, which falls within a 95% confidence interval of 73.8–91.2% and p < 0.001.
Fig. 3Forest plot displaying the SMD, SE, 95%CI, I^2^, τ^2^ and weight for each study
The forest plot elucidated in Fig. 4 shows the sub-group analysis conducted to explore the sources of the high heterogeneity observed in the overall results that were represented through Fig. 3. The heterogeneity, which was initially quantified by an I² statistic of 84.8%, indicated substantial variability in the effect sizes of the interventions across different studies. To better understand this variation, we analyzed potential moderators including age, duration of intervention, type of intervention, region, and faculty group, each investigated separately using forest plots. The analysis revealed a significant finding when the moderator “region” was examined. Specifically, studies originating from Africa were found to account for a considerable portion of the heterogeneity. The I² for studies from Africa dropped to 77%, with a p-value of 0.04, suggesting that the variation in effect sizes from this region was less random and more systematic than previously understood. The pooled SMD for studies from Africa was quite large at -4.25, though it had a wide 95% confidence interval ranging from − 19.90 to 11.40, indicating significant uncertainty around the estimate. The regions were coded numerically, with region 1 representing Africa, region 2 representing Asia, region 3 representing Australia, region 4 representing Europe, and region 5 representing North America. The forest plot for region 1 (Africa) shows the individual and pooled SMDs, with Ogbuanya (2017) [57] and Ugwoke (2017) [61] contributing large and significant negative effect sizes of -5.41 and − 2.95, respectively. This suggests that the interventions in these studies were particularly effective at reducing mental health issues among faculty members. In comparison, the forest plots for the other regions indicate more modest effect sizes with narrower confidence intervals and, notably, lower heterogeneity. For instance, the studies from Europe (region 4) showed an I² of 0%, implying no observed heterogeneity, and a pooled SMD of -0.16, which is much smaller than the effect seen in studies from Africa. Similarly, studies from North America (region 5) presented a small pooled SMD of -0.53. The sub-group analysis also included a test for subgroup differences, which produced a chi-square statistic of 17.82 with 4 degrees of freedom, resulting in a highly significant p-value of less than 0.01. This outcome confirms that the differences between regions are statistically significant, and that the region is an important factor in explaining the heterogeneity observed in the analysis.
Fig. 4Sub-group analysis by region; region 1 = Africa, region 2 = Asia, region 3 = Australia, region 4 = Europe, region 5 = North America
Figure 5 displays the forest plot that was made to comprehend the sources of heterogeneity in the effectiveness of various interventions. The moderator “type of intervention” was scrutinized to determine its impact on the observed variance across studies. The interventions were categorized as “mental interventions,” which exclusively addressed psychological or mental health outcomes, and “mixed interventions,” which likely included a combination of mental, physical, or other types of interventions. The results indicated that “mental interventions” alone significantly contributed to the heterogeneity among the studies, as reflected by a high I² value of 93% and a significant p-value of less than 0.001. The pooled SMD for mental interventions was − 1.76, with a 95% confidence interval ranging from − 4.68 to 1.17. Although the pooled effect suggests a potential benefit of mental interventions, the wide confidence interval indicates considerable uncertainty around this estimate. Notable studies such as Brewer (2019) [52], Lim (2023) [56], Ogbuanya (2017) [57], and Ugwoke (2017) [61] highlights the variation in SMDs, with Ogbuanya (2017) [57] reporting a notably large negative effect size of -5.41, suggesting a substantial benefit from the mental intervention used in that study. This stark difference in heterogeneity may reflect the more consistent effects of mixed interventions across different settings. Although the test for subgroup differences based on the type of intervention yielded a chi-square statistic of 1.23 with 1 degree of freedom, the associated p-value of 0.27 was not statistically significant, suggesting that the difference between mental and mixed interventions might not be as pronounced or that the sample size was insufficient to detect a significant difference.
Fig. 5Sub-group analysis by type of intervention
Figure 6 illustrates the subgroup analysis by faculty group (working in the “health” education field versus those in “other” fields), with forest plots displaying individual study effects and the pooled SMD for each faculty group. This categorization revealed that faculty from different education fields were a significant source of heterogeneity compared to those from health specialties. For faculty working in other sectors, the I² was 88% with a p-value of less than 0.001, and the pooled SMD was − 1.48 (95% CI: -3.19 to 0.24), implying a potential effectiveness of interventions in this subgroup, though again with a wide confidence interval suggesting variability in the effect size. Meta-regression was applied on quantitative predictors (age and duration of intervention), collectively both variables explained only 3.96% of the variation, the beta-coefficients were not significant (p = 0.949 and p = 0.128) respectively.
Fig. 6Sub-group analysis by faculty group
Multiple methods were used to evaluate the possibility of publication bias in our review. To investigate this, we generated a funnel plot and examined it for symmetry, a common practice to check for potential publication bias. As illustrated through Fig. 7, the funnel plot appeared symmetrical, indicating an absence of significant publication bias in the studies we included in our meta-analysis. This initial visual evaluation was supported by the quantitative evidence from Eggers’ regression, a statistical test to detect asymmetry in a funnel plot. The Eggers’ regression produced a p-value of 0.2454, exceeding the customary significance level of 0.05, thereby providing no statistical indication of publication bias.
Fig. 7Funnel plot assessing publication bias
To further reinforce our assessment of publication bias, we applied the Trim and Fill method, utilizing the Rosenthal approach. This method is aimed at estimating and amending for potentially missing studies in the meta-analysis, especially those small in size and effect, which might not have been published or could have been missed. The method accomplishes this by imputing hypothetical missing studies to balance those on the opposite side of the funnel plot, thus striving for a symmetric distribution of studies around the pooled effect size. Figure 8 reveals that, according to the Trim and Fill method, the funnel plot would become symmetrical with the addition of one study. This amendment led us to a revised pooled SMD of -0.621 with a 95% confidence interval ranging from − 2.4985 to 1.2561, maintaining the significant impact of the intervention on faculty mental health.
Fig. 8Trim and Fill method assessing publication bias
A multi-model inference was used to ascertain the relative significance of different predictors that may affect the mental health of faculty members (Fig. 9). This was deemed advantageous since it permits the evaluation of various models, each combining different predictors, to identify which constellation of factors best accounts for the variations observed within the data, and is especially useful in contexts where there is uncertainty regarding the relevance and impact of different predictors. To this end, we developed and examined a total of 32 distinct models, integrating various predictors. We employed the Akaike Information Criterion corrected for small sample sizes (AICc) to measure the fit of each model. The AICc is widely used in model selection processes for its ability to balance the model’s fit with its complexity, penalizing models with an excessive number of parameters, which could lead to overfitting. The model with the lowest AICc value, which in this case was 44.8, was identified as the one providing the most efficient explanation of the data, using the least number of parameters.
Fig. 9Multi-modal inference for predictor importance
“Region” emerged as the most influential with a score of 0.65, underscoring the strong link between the geographical location of faculty members and their mental well-being. This prominence may stem from variations in work culture, support structures, and environmental stressors that are inherent to different regions. Following closely, the “Duration of Intervention” held a score of 0.55, presenting itself as a substantial factor, though slightly less influential than “Region.” This predictor signals the significance of the temporal extent to which faculty members are exposed to interventions aimed at improving mental health. The “Faculty Group” predictor, with a score of 0.45, the “Type of Intervention” with 0.40, and “Age” with 0.35, though important, had a comparatively lesser impact on the mental health outcomes. Notably, every predictor we examined crossed the threshold of 0.5, as indicated by the vertical line in our figure, marking them as influential factors in our study. Moreover, when the significance of quantitative moderators was tested using meta-regression, age (p = 0.9491) and duration of intervention (p = 0.1284) were not statistically significant.
The high prevalence of mental health issues among academics has been supported by evidence and is often attributed to substantial changes in university systems alongside faculty struggles with individual, interpersonal, and institutional factors [45, 65]. Compared to other professions, university academics experience less job satisfaction and lower psychological health, due to large student cohorts, heavy workload, long working hours, and poor work–life balance [66]. However, and although some studies have included faculty as participants in interventions targeting a wider population [67, 68], there is paucity of high-quality consolidated literature that documents interventions effective in improving the mental health of this population [69]. Against this background, a need for a systematic analysis of existent literature and a search of the most relevant interventions is well justified. According to the results of this meta-analysis, interventions targeting mental health of faculty were overall significant, with a marked positive change in mental health indicators measured after the intervention compared to before. Other key findings identified the region where the study was conducted as a significant predictor for evaluating the effectiveness of faculty mental health interventions.
While interventions were overall significant, a deeper analysis revealed that not all of them documented similar effectiveness. For instance, the use of leadership development by Brewer and Colleagues [52] and of Fitbit devices by Garcia and Colleagues [53] resulted in less significant effects compared by behavioral coaching reported by Ogbuanya and Colleagues [57]. Also, the dependence on yoga, arts, and crafting [54, 55] produced positive, yet uncertain interventional consequences. The limited number of studies fitting the inclusion criteria makes it challenging to identify the most suitable or effective intervention. In this context, a key implication drawn from this meta-analysis suggests the need for increased emphasis on trying out various interventions with faculty. It is important to investigate the impact of successful interventions on different groups of faculty in various settings and to systematically document evidence. The recommendations from such evidence would offer a more robust background on how mental health of this population should be addressed. Additionally, besides main mental outcomes, it is crucial to thoroughly examine the impact of specific interventions on faculty additional mental health-related factors such as job insecurity [70, 71], work-life balance [72], stigma and disclosure [73–75]. These aspects were not directly addressed in the articles included in this meta-analysis, and carry implications for future research. This approach would facilitate replication and establishment of robust data, ultimately contributing to the identification of best practices for enhancing faculty mental health.
As an overall result of the meta-analysis, faculty exposed to interventions of a mental structure only, like leadership courses [52] and digital mental health enhancing tools [56] compared to those who received interventions of a more mixed structure mental-physical structure (55, 58) did not show a significant difference in effectiveness of these two intervention categories. As the CIs for mental health interventions were wider than those for mixed interventions, the latter may appear more consistent and can prove better than mental interventions alone. While evidence is available regarding the effect of exercise in improving mental health outcomes [76–78], with 5 studies in this meta-analysis corresponding to each intervention category, this result should be remain precursive, warranting further analysis. The conclusions drawn here should be, therefore, interpreted with caution, to avoid missing the value and effect of some types of interventions at the expense of others.
Age of faculty in the included studies was not a significant indicator of intervention effectiveness. As such, to attempt to improve the mental health and well-being of faculty at different ages and consequently different stages of their academic career would be beneficial. Further, the meta-analysis showed that faculty working in different specialties as compared to health specialties showed more effective results. Previous evidence shows that faculty working in health education face high levels of stress, burnout, and depression [79–81]. Given the inherent characteristics of their work, healthcare professionals, are at risk of emotional stress, negative mental consequences, and burnout, causing challenges that affect their daily practice [82, 83]. Despite providing excellent care to patients, healthcare professionals frequently struggle with prioritizing their own self-care, becoming at higher risk of mental health issues when compared to the general adult population [84, 85], and this was worsened by the COVID-19 pandemic [86, 87]. In light of such evidence, it may be anticipated that the inherent nature of work and exposures of health faculty may make them less responsive to mental health interventions compared to faculty in other fields. Such finding may be an implication for attempting to mitigate mental issues in faculty from all specialties, although those in health education may need more tailored, specific, or focused interventions. Again, and with two studies only spelling out faculty from health specialties, and other included studies not specifying the faculty group, it would be intriguing to investigate this detail when more similar studies become available.
The included studies in this meta-analysis described interventions of duration between one and 14 weeks. Nevertheless, the duration of the intervention was not particularly indicative of effectiveness, although for the study with longest duration conducted by Ogbuanya and Colleagues [57], the effects of a rational emotive behavior coaching intervention was sustained at 3 months follow-up. While the duration of the intervention could be variable, it is warranting to explore whether efficacy can be sustained at different time points post-intervention. This might assist institutions at identifying the best approaches, regardless of duration, that would result in a sustainable, long-lived effect on the mental well-being of their faculty.
An interesting finding of the meta-analysis was that Africa stands as a region of difference where interventions were particularly effective. Looking into the two included studies from Africa [57, 61] shows that both used interventions of rational emotive behavior training for 10 or 14 weeks. Also, both were done in Nigeria in related university departments and on faculty who teach in technical or vocational workshops, although carried by independent research groups. Rational emotive behavioral therapy (REBT) is an evidence-based approach aiming at management of workplace related stress and improvement of emotional and behavioral reactions, through training individuals to think more logically in face of hardships [88]. As such, the focus of REBT is changing irrational beliefs into rational ones, with the purpose of shifting dysfunctional emotions and maladaptive responses into functional and adaptive ones [89]. None of the other eight included studies used REBT, making this intervention stand out as an apparently effective one. However, such result should be interpreted cautiously, as it involves faculty from a specific domain and institution, and might not apply for other populations with different background and culture. Nevertheless, and in light of evidence peripheral to Africa showing effectiveness of REBT on well-being among students [90, 91] and populations outside the academic circle like athletes [92, 93], nurses [94], and even patients [95, 96], academic institutions may consider this approach within broader initiatives to improve mental health of their faculty. For a general adaptability of REBT or its applicability in a wider context, conducting more focused research on this intervention may lead to a better understanding of its effect on metal health outcomes in faculty, perhaps adding more value and generalizability to the results reported here.
The strength of this meta-analysis is attributed to the careful selection of studies deemed to be of high quality, as shown by results of risk of bias assessment, adherence to PRISMA guidelines and to the inclusion/exclusion criteria. The review brought forward the general scarcity of data regarding how interventions affect the mental health and well-being of academics, where only 10 relevant articles were included after an extensive search strategy. This reveals the gap of knowledge in this regard, and calls for better and more focused research into interventional methodologies in the academic community. The pooled results of this systematic review and meta-analysis are in conformity with other researchers’ earlier findings, and such agreement with previous evidence indicates a good measure of these results’ validity. For instance, results are parallel with a previous systematic review concluding strong evidence of academics’ high levels of stress and low levels of well-being, while experiences employed to deal with their job demands are essentially unknown [97]. Also, our results resonate with a former systematic review that reported inadequate information on measuring the outcome of various management-led mental well-being strategies for faculty, and called for further research with more dynamic study designs and establishing routine mental health assessment, as imminent measures to improve the mental well-being of academic staff [45].
This meta-synthesis possesses certain restraining limitations. The search strategy implemented included a restriction to the English language, resulting in the inclusion of predominantly Western, English-speaking participants in the majority of the selected papers, with the exception of Malaysia and Nigeria. Consequently, the conclusions drawn from this review might not accurately represent the perspectives and experiences of researchers engaged in universities worldwide, and those from various cultural backgrounds. Moreover, mental health and well-being are sophisticated concepts, and we recognize that the set of terms associated with these constructs, which we incorporated into the search strategy to identify pertinent research papers, may have not been exhaustive. Overlooking a more comprehensive list of constructs related to mental health and well-being may have resulted in exclusion of additional relevant papers that contribute to better understanding of the effectiveness of interventions. Furthermore, the evidence included in this review may have its own shortcomings as well. Looking into the included studies, 4 were randomized controlled trials and 6 were pre- and post-analyses, perhaps indicating a gap in evidence-based literature. Another limitation is that two selected studies involved technical college teachers and electronics workshop instructors. Although these two studies were done at colleges, the stress of academics in tertiary institutions and research-oriented universities may be different, perhaps affecting the generalizability of our results, and calling for vigilant conclusions. Also, studies that we included in the review and meta-synthesis focused only on faculty, without considering the whole academic environment, which may include important factors to consider among students and administrative staff. Those remain attractive populations to study in the context of a wholistic approach to improve mental health in academia. Additionally, to further investigate the sources of heterogeneity, more advanced methods such as network or structure meta-analysis may be used, and this remains tempting to explore.
The current findings of this review hold key implications and recommendations. First, in light of few studies fitting the search criteria, further investigation of faculty mental health would accumulate a larger body of evidence and support in building interventions with proven effectiveness. Second, as specific factors related to the academic career affect mental health of faculty, it would be useful to study interventions targeted towards the specific challenges of the academic environment, thus producing tailored approaches. Moreover, considerations of different factors such as age, gender, sociodemographic categories, country, specialty, and others, may be beneficial to get an overall portrait of which interventions are more efficient in each faculty group.
In conclusion, our meta-analysis revealed the overall positive impact of interventions on faculty mental health, with observed heterogeneity in effects underscoring the complexity of the issue. Faculty individual responses to different mental health interventions vary widely, and numerous factors may influence how they engage with and benefit from such mediations. Tailoring interventions to the specific needs and characteristics of the target faculty group and the pertinent environment is essential, emphasizing the importance of further research to refine and expand strategies, training, and resources that support the well-being of faculty members. To this end, the organizational culture at universities needs to largely prioritize faculty mental health, and should nurture a supportive environment with meticulous practices and guidelines that foster more positive responses to mental health interventions.
Below is the link to the electronic supplementary material.
Supplementary Material Supplementary file 1 PRISMA 2020 checklist for this study.
Supplementary Material Supplementary file 2 PRISMA 2020 checklist for abstract.