Prevalence of mental distress and associated factors among university students in Ethiopia: a meta-analysis

Abstract Background Mental distress is an important public health problem and is becoming common health problems among university students. Aims This study aimed to provide a pooled prevalence of mental distress and associated factors among university students in Ethiopia. Method We systematically searched PubMed, EMBASE and PsycINFO databases. A further search was performed at Google Scholar search engine for additional studies. All observational studies reporting the prevalence of mental distress and/or associated factors among university students in Ethiopia were included. Pooled prevalence with 95% confidence interval (95% CI) were calculated using random effects and quality effects models. Subgroup and sensitivity analyses were performed. Heterogeneity between studies and evidence of publication bias were assessed. Results The pooled prevalence of mental distress was 35% (95% CI; 28%–43%). Being female, participating in religious programmes, having close friends, experiencing financial distress, alcohol use, khat use, conflict with friends, lack of interest in their field of study and a family history of mental illness were factors associated with mental distress among students. We found significant heterogeneity, but no evidence of publication bias. Conclusions More than one third of university students in Ethiopia have suffered with mental distress. The finding provides evidence that university students are at risk population for mental health problems and suggests the need for early intervention to prevent severe mental illness.

In Ethiopia, mental and substance use disorders are common public health problems responsible for about 1897 disability adjusted life years (DALYs) per 100,000 population (Institute for Health Metrics & Evaluation [IHME], 2016). Studies have shown that mental distress is prevalent among university students in Ethiopia, ranging from 21.6% to 63.1% (Alem, Araya, Melaku, Wendimagegn, & Abdulahi, 2005;Busi et al., 2016;Byrd et al., 2014;Dachew et al., 2015;Dessie, Ebrahim, & Awoke, 2013;Getachew & Tekle, 2018;Haile, Alemu, & Habtewold, 2017;Kerebih, Ajaeb, & Hailesilassie, 2017;Melese et al., 2016;Tariku, Zerihun, Bisrat, Adissu, & Jini, 2017;Tesfaye, 2009). Although several individual studies have reported the prevalence of mental distress and associated factors among university students in Ethiopia, to our knowledge, there is no published systematic review and meta-analysis that shows pooled estimates of mental distress and its associated factors. Having a pooled prevalence of mental distress and identifying the associated factors would help policy-makers and programme managers in developing evidence-based mental health promotion and disease prevention programmes. Therefore, the objective of this systematic review and meta-analysis was to review the existing literature, with the aim of quantifying the burden of CMDs and identifying factors associated with CMDs among university students in Ethiopia.

Methods
This meta-analysis was conducted in accordance with the preferred reporting items for systematic review and metaanalysis (PRISMA) (Moher et al., 2015). The study protocol was prospectively registered in an International Prospective Register of Systematic Reviews (PROSPER) (Registration Number: CRD42017067223, http://www.crd.york.ac. uk/PROSPERO.

Data sources and search strategies
We searched PubMed, EMBASE and PsycINFO electronic databases and Google Scholar search engine with no publication year restriction. The search comprises both Medical Subject Headings (MeSH) and free text words (title and abstract word searches). We used the following search terms: "Mental distress" OR "Common mental disorders" OR "Mental illness" OR "Psychological distress" OR "Stress" AND "University students" OR "College students" AND "Ethiopia." In addition, the reference lists of included studies were manually searched for additional eligible articles.

Study selection and eligibility criteria
All observational studies reporting the prevalence of mental distress and/or factors associated with mental distress among university students will be included. An article will be included if it meets the following criteria: (1) conducted solely or partly among university and/or college students, (2) reported the prevalence of mental distress using standardized instruments or questionnaires (such as the Beck's Depression Inventory, Patient Health Questionnaire -9, Selfreporting Questionnaire-20 or clinical interviews) and (3) published in English. Conference abstracts, letters to editors, review and commentary articles were excluded. Two investigators (BAD and MAW) assessed the eligibility of each study independently and disagreements were resolved by discussion.

Data extraction
A standardised pre-piloted form was used to extract data from the included studies. Extracted information includes first author's last name, year of publication, study location, sample size, prevalence of mental distress, estimates for risk or protective factors examined in each study, and ascertainment of outcome.

Quality assessment
Two review authors (BAD and MAW) independently assessed the quality of all included studies using Newcastle-Ottawa quality assessment tool adapted for cross-sectional studies (Anon, n.d.) and any disagreements were resolved through discussion. The tool takes into account the selection of participants, comparability and assessment of outcome. A study can be given a maximum of one star for each numbered item within the selection and exposure categories and a maximum of two stars can be given for comparability. Finally, the results are summarised in three categories as good quality (3 or 4 stars in selection domain and 1 or 2 stars in comparability domain and 2 or 3 stars in outcome/ exposure domain), fair quality (2 stars in selection domain and 1 or 2 stars in comparability domain and 2 or 3 stars in outcome/exposure domain) and poor quality (0 or 1 star in selection domain or 0 stars in comparability domain or 0 or 1 stars in outcome/exposure domain). Finally, the quality of each study was categorized as good, fair or poor quality. We have used quality scores for quantitative analysis (Supporting Information Table 1).

Data synthesis
We conducted random-effects meta-analyses with 95% confidence interval (95% CI) to obtain the pooled prevalence of metal distress and the pooled odds ratios of identified factors associated with mental distress (Berkey, Hoaglin, Mosteller, & Colditz, 1995). Quality-effects meta-analysis was also performed to examine how the quality of each study changed the pooled estimate compared with the results from random-effects meta-analysis (Doi & Thalib, 2008). This analysis incorporates the quality score of each study in the calculation of the study weight, which is a robust and innovative technique to help minimize the estimator variance and account for subjectivity in quality assessment (Doi & Thalib, 2008). Heterogeneity between the studies was assessed using both Cochran's Q statistic and the I 2 statistics. I 2 value greater than 50% were considered as indicative of substantial heterogeneity (Higgins & Thompson, 2002). Evidence of publication bias was assessed using Egger's test (p < 0.05) (Egger, Davey Smith, Schneider, & Minder, 1997) and visual inspection of the symmetry in funnel plots (Liu, 2011).
Subgroup analyses were undertaken by sample size, year of publication, study quality and screening tool as possible sources of heterogeneity between studies. Sensitivity analyses were performed by excluding each study one by one and calculating a pooled estimate for the reminder of the studies. All statistical analyses were performed using MetaXL version 5.3 and STATA14 Metaprop package (StataCorp, 2015).

Results
After duplicates removal, 246 articles were identified and of them, 148 were excluded during initial assessment as their titles were found to be irrelevant. We further screened the abstracts of the remaining 98 studies and excluded 64 of these as they did not fulfil the eligibility criteria. The remaining (n ¼ 34) full text papers were screened for relevance and 11 papers were found to be eligible for metaanalysis ( Figure 1). Table 1 shows the characteristics of studies included in the meta-analysis. All of the included studies were cross-sectional and published between 2005 and 2018. The participant ages ranged from 16 to 35 years. The sample sizes of included studies ranged from 240 to 2817, with the response rate of 80-100%. Most studies used stratified random sampling techniques to recruit the study participants. Of the included studies, eight studies used the Self-Reporting Questionnaire (SRQ-20) to assess mental distress. SRQ-20 is a screening tool having 20 item questions, originally developed by World Health Organization (WHO) designed to indicate mental distress (World Health Organization, 1994). The tool is adopted from WHO and was validated in low and middle-income countries including Ethiopia; and found Included Eligibility Screening IdenƟficaƟon Figure 1. Flow diagram of studies included in meta-analysis.

Study characteristics
to be highly sensitive and specific (Ali, Ryan, & De Silva, 2016;Lund et al., 2011;Youngmann, Zilber, Workneh, & Giel, 2008). A cut-off point of eight and above was taken by most of the studies (n ¼ 6). Seven of 11 studies were good in quality studies. The average quality score of included studies was 7 (Supplementray Table 1).

Factors associated with mental distress
In this study, we identified factors associated with mental distress among students. Only those factors that were assessed by at least two studies were considered for meta-analysis. Being female, participating in religious programmes, having close friends, experiencing financial distress, alcohol use, khat use, conflict with friends, lack of interest in their field of study and a family history of mental illness were factors associated with mental distress among students (Table 3).

Discussion
This is the first meta-analysis that determined the pooled prevalence of mental distress and associated factors among university students in Ethiopia. Our finding suggests that more than one third of university students have experienced mental distress. The prevalence of mental distress found in this study is much higher than that found in the general population of Ethiopia (11.7%-17.7%) (Alem, Kebede, Woldesemiat, Jacobsson, & Kullgren, 1999;Gelaye et al., 2012;Kebede, Alem, & Rashid, 1999). The finding provides evidence that university students are at risk population for mental health problems. This higher level of distress may have short and long-term influence on students' achievement. Students are less likely to perform well at university when suffering from mental illness (Kessler et al., 1995; Sijtsema et al., 2014). Studies revealed that up to 5% of college dropout has been associated with mental illness (Kessler et al., 1995). Moreover, mental illness has been shown to be associated with higher risk of substance use (Swendsen et al., 2010) and suicidality (Dvorak et al., 2013;Garlow et al., 2008;Izadinia et al., 2010). This finding is also higher compared to similar studies conducted in East African countries, where a prevalence of 16.2%-19.8% of mental distress was reported (Hersi et al., 2017;Ovuga, Boardman, & Wasserman, 2006). In Ethiopia, mental disorders are not considered as life-threatening problems, and mental health services are not given due priority and the needs of people for mental health care are not met (Ayano et al., 2017). Although it is only partially implemented, a National Mental Health Strategy has been available in Ethiopia since 2012 (World Health Organization, 2014). However, neither the National Mental Health Strategy nor the Health Sector Transformation Plan mentioned students as vulnerable groups for mental illness (Federal Democratic Republic of Ethiopia Ministry of Health, 2012Health, , 2015. Our finding, therefore, highlights that university students are a vulnerable group who require special consideration when developing mental health services. Our findings showed that students who participated in religious practice, regardless of which religion, had a lower risk of mental distress compared with those who were not. Different studies have highlighted the protective role of spirituality in preventing depression, anxiety and substance use disorders and promoting well-being and quality of life (Medlock et al., 2017;Tusa & Burgholzer, 2013;Unterrainer, Lewis, & Fink, 2014). It is also important to note the possibility of reverse causation, as depressed students may stop going to worship. Similarly, students who had close friends were less likely to have mental distress than those who had not. As college life can be stressful, experiencing strong, intimate friendships with a high degree of attachment and support may help in promoting mental health during this period. A recent longitudinal study conducted in the United States of America found that close friendships increase self-worth and decrease anxiety and depressive symptoms (Narr, Allen, Tan, & Loeb, 2019). On the other hand, our finding suggests that conflict with friends were found to increase the risk of mental distress by two folds (Pooled OR ¼ 2.0, 95%CI: 1.6-2.7).
We found that females reported higher levels of mental distress than male students, which is consistent with the existing evidence that a greater proportion of females report common mental health problems than males (Tedstone   Note. Only those factors that were assessed by at least two studies were considered for meta-analysis. Doherty & Kartalova-O'Doherty, 2010). This could be due to the affective nature of their response to stressors, as women have lower self-esteem and sense of control than men, as well as gender-based violence and other biological factors (Eaton et al., 2012;Tedstone Doherty & Kartalova-O'Doherty, 2010). Our study also found that financial hardship was another factor associated with mental distress in students. Students who reported financial stress had a greater risk of mental distress than those who did not. The finding is consistent with the existing evidence where poor mental health has been consistently linked with the experience of financial hardship (Cvetkovski, Reavley, & Jorm, 2012;Kiely, Leach, Olesen, & Butterworth, 2015). Our review found that the risk of mental distress was 70% higher in students who chewed khat as compared to those did not. Khat is a flowering plant which is used as a stimulant in many parts of Africa and the Arabian Peninsula (Warfa et al., 2007). It is a commonly used psychoactive substance among students in Ethiopia, especially during examination periods (Gebrehanna, Berhane, & Worku, 2014). Many studies in Ethiopia and abroad reported a significant association between khat use and various mental disorders (Damena, Mossie, & Tesfaye, 2011;Gebrehanna et al., 2014;Odenwald et al., 2005;Warfa et al., 2007). In line with other studies, alcohol use, lack of interest in their field of study and a family history of mental illness were also other factors associated with mental distress in students (Rohrer, Rohland, Denison, Pierce, & Rasmussen, 2007).
The strengths of this meta-analysis are that we included all studies without time restrictions. Furthermore, we used standardized quality assessment tool with most (n ¼ 7) studies rated good in quality. In addition, almost all studies (n ¼ 10) include in in the analyses used standardized screening questionnaires to assess the prevalence of mental distress and eight of eleven studies used SRQ-20, the tool that has been validated in Ethiopia (Youngmann et al., 2008). However, this meta-analysis study had some important limitations. First, we found a high level of between study heterogeneity, but to address this we used a random effects model as is recommended in situations with high level of between study heterogeneity (Ades, Lu, & Higgins, 2005). In addition, we applied a quality effects model, which allowed us to overcome some problems with a traditional random effects model (Doi & Thalib, 2008). Second, although the quality of all included studies was assessed using the Newcastle-Ottawa quality assessment tool, methodological appraisal remains a subjective exercise. For this reason, to minimize bias in the review process, two reviewers assessed the quality of all included studies independently any disagreements were resolved through discussion. Third, unlike a fixed effects model which produce unbiased estimates, the random effects model introduce bias in the pooled estimates (Kinney & Dunson, 2007). Fourth, although the funnel plot did not show asymmetry on visual inspection and Egger's test also showed no evidence of publication bias, we may not fully exclude the possibility of publication bias as this meta-analysis considers only published studies. Fifth, there is a possibility of reporting and response set bias as all data were self-reported. Finally, the cross-sectional nature of the study does not confirm definitive cause and effect relationships.

Conclusion
The prevalence of mental distress among university students in Ethiopia was found to be high. Being female, participating in religious programmes, having close friends, experiencing financial distress, alcohol use, khat use, conflict with friends, lack of interest in their field of study and a family history of mental illness were factors associated with mental distress among students. Prospective cohort studies would be useful to confirm the observed associations. Qualitative studies are needed to further assess the contributory factors leading to mental distress among university students. Our findings suggest the need for early intervention to prevent severe mental illness in students. We encouraged students to seek formal as well as informal help, usually from friends, family and religious leaders, and to implement self-care strategies to take care and control of their own wellbeing.