Prevalence of lifetime substances use among students in Ethiopia: a systematic review and meta-analysis

Background The use of substances is a growing concern in Ethiopia, and their impacts on younger generation have been a concern of different professionals. Even though students are at high-risk of substance abuse, there is lack of comprehensive evidence for policy decision on substance use among students. Therefore, the aim of this systematic review and meta-analysis was to estimate the prevalence of common substances among students in Ethiopia. Method A comprehensive literature searches were done from biomedical databases: PubMed/Medline, African Journal Online, HINARI, Science Direct, and Google Scholar for article published until Dcember 31, 2017, and Addis Ababa Uiversity’s electronic library search of unpublished thesis and dissertations. Two authors autonomously selected studies, extracted data, and evaluated quality of studies. The prevalence of lifetime substances use was estimated using the random effects model. Q and I2 statistics were computed to measure the extents of heterogeneity. Results A total 676 study articles were identified from electronic databases, and 28 of them were included in meta-analysis. The analysis revealed that the lifetime prevalence of any substance use was 52.5% (95% CI 42.4–62.4%), khat 24.7% (95% CI 21.8–27.7%), alcohol 46.2% (95% CI 40.3–52.2%), and smoking cigarette 14.7% (95% CI 11.3–18.5%). Significant heterogeneity was observed but there was no significant publication bias. The lifetime prevalence of khat, alcohol, and cigarette smoking among high school vs university students was 22.5% (95% CI 15.2–30.7%) vs 25.1% (95% CI 21.9–28.5%), 41.4% (95% CI 22.1–62.1%) vs 47.8% (95% CI 39.9–55.7%), and 21.5% (95% CI 12.6–32.1%) vs 12.9% (95% CI 10.1–16.0%), respectively. Conclusion This meta-analysis highlighted the extent of lifetime prevalence of any substance, khat, alcohol, and cigarettes smoking among students in Ethiopia. Significant percent of high school students have exposed to substances. Policy makers should devise and implement strictly binding regulation to curb widespread of substances around educational institution premises at national level. Priority should be given to intervention strategies that help delay first use of substance to prevent problems later in life. Besides, the problem warrants regular national-level educational institutions based studies focusing on the magnitude, trajectory, and consequences of substance use among students. Systematic review registration: PROSPERO CRD42018082635


Background
Psychoactive substances act on the central nervous system and change the activities of the brain. These substances have wide range of effects, including short-term changes in perceptions, mood, consciousness, and behaviors [1]. Substances such as alcohol, khat, and tobacco are widely used [2,3], and they are leading causes of human sufferings and become important public health and socioeconomic issue globally [4][5][6]. Current trends showed that the use of psychoactive substances have considerably increased predominantly in developing countries [6].
In 2012, 5.9% of all global deaths and 5.1% of disabilityadjusted life year were attributable to alcohol consumption [7]. Roerecke et al. reported that alcohol per capita consumption in 15 years and above in sub-Saharan Africa is higher than the global consumption rate [8]. Similarly, a quarter of the world population smoke cigarettes, and it was leading risk factor for premature death and disability in 2015. Smoking accounts for 11.5% of death and is among five leading risk factors of disability-adjusted lost life years (DALY) in 109 countries and territories in 2015 [9,10]. Studies also showed that chronic khat consumption not only causes severe neurological, psychiatric, cardiovascular, dental, gastrointestinal, and reproductive dysfunction [11][12][13][14][15][16], but it also has adverse socioeconomic effects affecting other aspects of life [17].
Evidence showed that poor socioeconomic condition increases the risk of harmful drug use, and people living in low-income countries disproportionately affected by higher burden of substance-related disability and premature death [18]. It is also worthy to note that rapid economic, social, and cultural changes increased use of various substances in sub-Saharan Africa countries [19].
Like other sub-Saharan countries, Ethiopia is also facing a growing problem of substance use. The use of substances has long been a serious concern for various professionals in Ethiopia due to their adverse impacts on younger generation [20,21]. Therefore, Ethiopia recognizes substance use by young people as a serious health and social problem, and students being a high-risk abusers of substance [22]. Studies revealed that khat, alcohol, and cigarette are commonly abused substances in general population [23][24][25], and they are also widely used substances by high school and university students in Ethiopia [26][27][28][29][30][31][32][33][34].
The use of substance is associated with various health risks. For instance, a study showed that the use of substance is associated with HIV infection and risky sexual behaviors [35]. Additionally, studies revealed that the use of substances poses high risk-taking behaviors among students, which results in economic, social, physical, and health complications [26][27][28][29][30][31][32][33][34]. Other study showed that substance use among students is associated with social phobia, poor academic performance, and the use of multiple substances in lifetime [36]. Furthermore, it has been documented that substance abuse is associated with suicidal attempt [37].
Two meta-analyses done on prevalence of substance use among students have been documented recently [48,49]. Both studies report prevalence substance among university students focusing khat, alcohol, and smoking cigarettes. However, the evidence showed that a vast majority of students initiated using substance before joining university [27,28,32,36,41,46,50,51]. Additionally, the estimated of one meta-analysis [48] did not specify whether prevalence represents lifetime, recent, or current use of substance. Therefore, in order to foreward recommendation for comprehensive intervenation, it is crucial to have prevalence estimates of substance that represents both high school and university students.
A report of Ministry of Education of Ethiopia showed that 3,767,322 students were enrolled in secondary (grade 9-12) school, colleges, TVET, and universities in the 2015/2016 academic year. Of these, 2,421,163 of them were enrolled in secondary education, and 830,287 of them enrolled in higher education [52]. With ever increasing number of students being enrolled in secondary to higher education, it is essential to address the issue of substance use in order to produce productive human power that is free from substance abuse. Therefore, the aim of this review and meta-analysis was to provide comprehensive views of prevalence of different substances practiced by students enrolled in secondary schools and higher educational institutions in Ethiopia for concerned decision makers and to inform administrators to dealing with ever increasing challenges of substances.

Registration
This systematic review has been registered on the International Prospective Register of Systematic Reviews (PROSPERO CRD42018082635).

Search strategy
A comprehensive literature search was done from biomedical databases: PubMed/Medline, HINARI, African Journal Online (AJOL), Science Direct, and Google Scholar. For unpublished studies, master's thesis and PhD dissertation, the official website of Addis Ababa University's electronic library [53] was searched. Additionally, the lists of references of eligible studies were explored to obtain additional studies. All published and unpublished article up to December 31, 2017 were included. The following search terms were used alone or in combination: substance, khat, alcohol, smoking, prevalence, students, university, and Ethiopia. For reporting, PRISMA guideline was used during systematic review [54] (Additional file 1: Table S1).

Inclusion and exclusion
All studies done among secondary school, college, and university students in Ethiopia reporting combined lifetime prevalence of substances, or lifetime prevalence of khat, cigarette smoking, and alcohol consumption reporting in English language were included. The main outcomes of this review and meta-analysis were overall lifetime prevalence of any substance, lifetime prevalence of khat chewing, lifetime prevalence of alcohol consumption, and lifetime prevalence of smoking cigarette. Additionally, studies with cross-sectional design, having response rate ≥ 80%, used probability sampling techniques; reporting quality assurance methods and quality assessment score ≥ 50% were included. Review articles, studies employed non-probability sampling techniques, qualitative studies, studies available only as abstract with unclear outcomes, and studies conducted in non-regular (extension and summer) students were excluded.

Quality assessment and data extraction
The Joanna Briggs Institute Meta-Analysis for Statistics Assessment and Review Instrument (JBI_MAStARI) was used for critical appraisal [55]. The manual contains appraisal checklists. Two reviewers independently assessed articles prior to inclusion in the final review using the checklists. Any disagreement which arose between the reviewers was solved by involving a third reviewer. Data were extracted independently by both authors. For each eligible article or abstract, information about author(s), the study setting, study period, sample size, sampling technique, method of data collection, response rate, age mean/range, substances (khat, alcohol, and tobacco) use measures (lifetime or ever use prevalence), and results were extracted on Microsoft excel 2010.

Data analysis
The analysis of the evidence was based on all studies included in this review in accordance with a PRISMA guidelines. The extracted data were exported to STATA Version 13.0 statistical software package. During the meta-analysis, all selected studies were combined using random effects model [56] to estimate the pooled prevalence of substance use. The Cochran Q test and I 2 statistics were used to test heterogeneity in pooled prevalence estimates. The subgroup data analyses were done using region of study setup, sample size, level of educations, proportion of female students, study year, year of publication, and age of participants. Metaregression analyses were carried out to identify parameters (sample size, year of publication, female proportion, and age of participants) associated with substance use.

Search results
The review identified a total of 676 studies based on literature searches. Of these, 665 articles were from published sources and the remaining 11 were unpublished master's thesis. From the total, 103 duplicated records were excluded and 521 records were excluded after screened by title and abstract. A total of 52 articles were screened for eligibility and quality. From these, 24 articles were excluded with reasons; 9 articles did not meet eligibility criteria, 14 articles failed quality assessment (< 50% score), and 1 article duplicated contents. Finally, 28 articles were included in the final analysis ( Fig. 1).

Lifetime prevalence of any substance
A total of 11 studies reported lifetime prevalence of any substance use (khat, alcohol, or cigarette smoking) [27, 28, 33, 34, 36, 41-43, 51, 57, 68] with a total of 7,909 participants included in meta-analysis. The prevalence ranging from 28.4% in study conducted among Bahir Dar and Gonder University students [41] to 82.7% in study conducted in Mekelle University students [42]. The overall pooled lifetime prevalence of any substance use the use of at least one substance was 52.5% (95% CI 42.4, 62.4%). The analysis revealed substantial heterogeneity across studies with I 2 = 98.8%, p < 0.00 (Fig. 2). However, both Begg's test p < 0.1195 and Egger's test p < 0.1075 showed nonsignificant publication bias.

Discussion
This meta-analysis tried to estimate the pooled lifetime prevalence of most commonly used substances, khat, alcohol, and smoking cigarettes, and the overall prevalence of any substance in students in Ethiopia.
In this meta-analysis, the pooled estimates showed that more than one in two (52%) students involved in the use of at least one substance in their lifetime. The finding was consistent with the results of meta-analysis done on the lifetime prevalence of alcohol consumption among young people in eastern Africa which was 50% [71]. This indicates the use of substance is far more common among students in Ethiopian. It is not surprising to observe high prevalence of substance use when educational institutions surrounded by substance sellers, who even provide their customers with private rooms [72]. Additionally, this could be due to the facts that these substances (khat, alcohol, and smoking) are not controlled, and educational institutions do not have binding law that prevents the use of these substances in Ethiopia. Furthermore, studies [73,74] showed ever increasing partying which is becoming the integral part of culture among students, might explain high lifetime prevalence of substances.
There was significant regional variation with the highest prevalence observed in Tigrai region, although all regions were not represented, including Oromia, the largest region. The differences could be explained by difference in social values attached to different substances [20]. Alcohol drink in Northern and khat use in southwest and Eastern part of Ethiopia are widely accepted. Additionally, the observed difference may be explained by the effect of school environment on healthy behavior, which influenced further by compositional and contextual factors [75,76]. Similarly, there was significant variation of prevalence when studies grouped by year of publication with the highest prevalence observed in conducted 2010-2014, 60%. The use of substances affects wider range of aspects of life. The literature showed that students involved in substance use are at higher risk of developing violent behaviors, risky sexual behaviors, and withdrawal symptoms [20,32,77].
The pooled prevalence showed that 24.7% (95% CI 21.8-27.7%) of students had used khat at least once in Fig. 3 Forest plot of prevalence of lifetime khat consumption among students in Ethiopia their lifetime. This finding was slightly higher than national prevalence of lifetime khat consumption, 19% [78]. The observed difference could be due to difference in study population and study settings. This study showed that khat consumption is spreading at an alarming rate in recent days among students that can be attributed to misconception that khat consumption improves academic performance which in fact disagree with available literatures [36,79]. Even, the results of meta-analysis of various studies revealed that acute or sub-chronic exposure to khat impair short-term memory [80]. The ever increasing khat consumption complemented by absence of laws that regulate the production, distribution, and use in school environment in Ethiopia, and inevitably, there will be challenges ahead in formulating rules and regulation since khat becomes one of the leading commodities for export in recent days [39], and farmers are increasingly abandoning other crops for cultivating khat [81].
Segregation of the study by region and sample size revealed substantial heterogeneity. The pooled estimate based on two studies showed that lifetime prevalence of khat use was highest in Somale Region (n = 1), 33.3%  (95% CI 29.6, 37.3) and followed by Tigrai Region which is non-khat growing region in the country. The widespread use be explained by the fact that the region is predominantly inhabited by Muslim and khat consumption is widely accepted [82]. For the case of the Tigrai Region, the higher prevalence would be explained by the fact that khat consumption is spreading to most major cities from traditional khat growing regions [81]. The prevalence was also highest in studies with sample size greater than 1000 (n = 5), 27.76% (95% CI 22.59, 32.92). This indicates that the estimate was influenced by larger sample size studies. The pooled estimate of lifetime prevalence of alcohol consumption in this meta-analysis was 46.4% (95% CI 38.7-54.2). This finding was slightly lower than national prevalence, 49% [78], and 82% in university students in eastern African countries [71]. The observed difference could be explained by difference in study population and settings. Additionally, the difference could also be attributed to social influence, peer pressure [83], and cultural differences. Most importantly, aggressive commercial promotions of various alcohol brands targeting younger people through major private and government owned media outlets are partly responsible for high consumption of alcohol among students [84,85]. Subgroup analysis by region showed that prevalence of alcohol consumption was significantly different across regions. The variation could be attributed to difference in the social value of alcohol, settings, and study methods [20]. The prevalence significantly varied with sample size that the highest was observed in small sample size studies. However, the overall estimate was not significantly affected by sample size. Similarly, the highest prevalence was observed in studies female proportion 20-30% of sample size. This could be due to difference in risk of substance use between male and female. This could be due to the fact that being male is associated with higher risk of substances use [26,27,29,30,51,57,64,86].
The pooled lifetime prevalence of cigarette smoking in this meta-analysis was 14.7%. The finding was higher than the results of EDHS analysis; 4.1% all forms of tobacco use and 8.1% prevalence in men [87]. However, the prevalence was lower than the result of metaanalysis of studies conducted in Iranian university male students, which was 19.8% [88]. The difference could be explained variation in definition of cigarettes smoking. The latter study did not specify the prevalence as current or lifetime. Additionally, the variation could be attributed to difference in population. This meta-analysis included studies conducted in secondary schools and universities whereas the EDHS conducted among adults, and meta-analysis result from Iran was based on studies done among male university students. Furthermore, cultural and socioeconomic differences might have played role. The observed between group difference in regions in Ethiopia in this meta-analysis was similar with study conducted among adults [87]. Meta-regression analysis in current study showed that age was associated with tobacco smoking, which was consistent with prevalence studies done in Ethiopia [87,89]. However, the review by Wicki, et al. [90] argued that the results on relation of alcohol consumption and the age of the students were inconsistent.
Overall, the results of this meta-analysis showed significant number of students exposed to various substances before joining university. Early exposure to substances at younger age has adverse health and behavioral effects during adulthood [91]. For instance, studies showed that early exposure is associated with risky behaviors and sexualtransmitted diseases, early pregnancy, low educational attainment, alcohol abuse and dependency, and anti-social behavior [92][93][94]. Furthermore, a recent longitudinal study in Finland among substance use discordant twin demonstrated that early exposure to substance disrupts transition into adulthood [95].
The strength of this meta-analysis was the representativeness of the estimates since we strictly followed the PRISMA guideline. The other strength was various substances that are commonly used by students in Ethiopian were comprehensively presented to facilitate accessibility of the evidences for concerned decision makers. Moreover, data extractions were carried out by using comprehensive tools, and two authors independently extracted data to reduce potential risks.
However, some subgroup estimates for prevalence of any substance, khat, alcohol, and tobacco smoking based on a single study do not necessarily reflect the actual context. This limits the generalizability of the finding based on a single study estimate. Even though universities and secondary schools were represented in this study, there were limited studies representing third generation or newly established universities. Inadequate or absence of studies representing public and private colleges also limits the generalizability of the finding to all education settings. Additionally, the current metaanalysis focused on lifetime prevalence of substance use and the results do not show the current substance use status, though large majority of ever substance users are current users. Lastly, we did not pool the estimate for the risk factors because of differences in risk factors across studies. Therefore, future studies should focus on substance use risk factors.

Conclusion
The pooled estimates of this meta-analysis highlighted the extent of lifetime prevalence of any substance, khat, alcohol, and cigarette smoking among students in Ethiopia. The uses of these substances are common in educational institutions and vary with study characteristics such as region, proportion of female students, mean age, and publication year. Therefore, policy makers should devise and implement strictly binding regulation to curb widespread use of substances around educational institution premises at national level. Priority should be given to intervention strategies that help delay first use of substance to prevent problems later in life. Besides, the issue warrants regular national-level educational institutions based studies focusing on the magnitude, trajectory, and consequences of substance use among students.