The Prevalence and Severity of Depression, Anxiety and Stress Among Medical Undergraduate Students of Arsi University and Their Association With Substance Use, Southeast Ethiopia.

Background: Mental health problems can negatively impact physical and psychological well-being of junior medical students and predisposes them to many unhealthy behaviors. Objective: We aimed to determine the prevalence and severity of depression, anxiety and stress among medical undergraduate students of Arsi University and their association with substance use. Materials and Methods: Institutional based cross-sectional study was conducted on 265 medical students that were selected by systematic random sampling. Data were collected by pre-tested self-administrative questionnaire and analyzed by SPSS-21 software. Logistic regression analysis were employed and statistical signicance was accepted at p<0.05. Result: In the present study, 5 questionnaires were rejected for incompleteness giving response rate of 98.1%. The current prevalence rate of depression, anxiety, stress, khat chewing, cigarette smoking and alcohol drinking was found to be 52.3%, 60.8%, 40.4%, 21.5%, 15.4% and 33.8% respectively. Depression was signicantly associated with monthly income, residency and alcohol drinking. Anxiety was associated with gender, marital status, educational year, residency and cigarette smoking. Stress was signicantly associated with monthly income, educational year, residency, khat chewing, and drinking alcohol. Conclusion: To sum up, depression, anxiety and stress are common problems among medical students of Arsi University. Monthly income, residency and alcohol drinking were identied as risk factors of both depression and stress. Furthermore, educational year and khat chewing were also risk factors for stress. Finally, identied risk factors of anxiety were gender, marital status, educational year, residency and cigarette smoking. Therefore, counseling and awareness creation are recommended.


Introduction
Stress is the generalized, non-speci c response of the body to any factor that overwhelms, or threatens to overwhelm, the body's compensatory abilities to maintain homeostasis (1). Stress that enhances physical or mental function is considered as eustress (2,3).
Conversely, distress is persistent stress that results in cognitive, behavioral and emotional disturbances like anxiety and depression (4,5). Clinically, anxiety is characterized by intense feeling of dread, accompanied by somatic symptoms that indicate hyperactive autonomic nervous system (6). Whereas, depression manifests as loss of interest or pleasure, sadness, feelings of guilt or low selfworth, disturbed sleep or appetite, extreme tiredness, and poor concentration (7).
There are indications that there is a higher rate of psychological morbidity in medical students than students in other disciplines (27,(51)(52)(53). The study conducted across 10 universities in Hong Kong revealed 21% of students experiencing moderate depression, 41% moderate anxiety, and 27% moderate stress (54). The prevalence of depression, anxiety and stress among medical undergraduates studying in a premier medical institution 51.3%, 66.9% and 53% respectively (17). Studies related to stress in medical education in Arab countries have also con rmed that stress, depression, and anxiety are common among medical students (53,55).
Mental health problems can negatively impact academic achievement, physical health, and psychological well-being of junior medical students (48,56). The excessive amount of stress had led to grave consequences such as tempted to cheat on exams, poor academic performance, di culties in solving interpersonal conflicts, decreased attention, reduced concentration, loss of objectivity, increased incidence of errors, negligence, substance misuse, sleep problems, low self-esteem and other stress-related outcomes (38,57). Infrequent exercise, alcohol drinking, smoking, khat chewing, sleep disorders and eating poorly are also identi ed to be associated with increased stress (58)(59)(60)(61). Conversely, chronic nicotine intake affects the brain reward system function which results in depression and anxiety symptoms (62). Khat use was further associated with more mood disturbances than experienced by those who did not use khat and signi cantly associated with depression, anxiety, and cortisol stress response (63) . Heavy alcohol consumption is also associated with anxiety disorder (64,65). Retrieving knowledge about presence of stress, depression and anxiety is therefore important in itself and if found should be given attention for timely intervention. However, very little is known about depression, anxiety and stress among medical undergraduate students in Ethiopia. Therefore, this study aimed to determine the prevalence and severities of depression, anxiety and stress among medical undergraduate students of Arsi University and their association with substance use.

Methods And Materials
Institutional based cross-sectional design was conducted in Arsi University from January 03 -31, 2019 among 265 sampled undergraduate medical students. After proportional allocation of the calculated sample size by a correction formula to each academic year level (enrollment year), strati ed random sampling was employed. From each stratum after randomly selecting the rst respondent based on their id number, required respondents were chosen by every 4 interval through systematic random sampling. All medical students in Arsi University were eligible to participate with exceptions of severely ill students and that were out of town during the time of data collection. Ethical approval was obtained from Arsi University Ethical Review Committee and list the board's name are Hailu Fekadu, Kelil Haji, Gebi Agero, Taresa Kisi, Hinsermu Bayu and Dr. Abebe Sorsa. Informed consent was secured and participation was totally voluntary. Con dentiality was kept unanimously. Data were collected by self-administered questionnaires that comprises of the following parts.
i. Sociodemographic pro le: includes age, gender, enrollment year, marital status, residence, income, religion and ethnicity.
ii. Depression, Anxiety, Stress Scale -21 (DASS-21): is self-report tools designed to measure the emotional states of depression, anxiety and stress over the last week (66). Each of the three DASS-21 scales contains 7 items. The responses are given on a 4-point Likert scale, ranging from zero if "I strongly disagree" to 3 if "I totally agree". According to DASS-21 scoring instructions, the obtained DASS-21 scores need to be multiplied by 2 to have the nal score and ranges of scores correspond to levels of symptoms, ranging from "normal" to "extremely serious" (Table 1). However, to determine the prevalence of DAS, DASS-21 scores with normal levels of depression, anxiety, and stress were coded as "0" whereas those with mild, moderate, severe, or extremely severe levels were coded as "1." Various studies demonstrated the DASS-21 was found to have strong internal consistency (67,68).
iii. Drug Abuse Surveillance Test (DAST): is a 10-item self-administered tool that provides a quick assessment of drug use problems (69). It has been validated in the varied setting; substance-abuse patients (70), primary care (71), in the workplace (72), and adapted for use with adolescents (73). Furthermore, it is also validated to assess uses of substances such as khat chewing, cigarette smoking, and alcohol drinking in undergraduate medical students (38). In present study, khat chewer is the proportion of individuals who had ever used khat at least once in his/her lifetime. Cigarette smoker is the proportion of individuals who had ever used cigarette at least once in his/her lifetime. Alcohol drinker is the proportion of individuals who had ever used alcohol drinks such as tela, tej, katicala/areke, beer, wine, or other drinks that can cause intoxication at least once in his/her lifetime.
The questionnaire was pretested on 13 randomly selected undergraduate medical students of Hawasa University. After checking the collected data for completeness, it was double entered in to Epi-data version 3.1 and exported into SPSS version 21 for analysis.
Incomplete and inconsistent data were excluded from the analysis. The data were processed by using descriptive analysis, including frequency distribution, cross tabulation and summary measures. Bivariate logistic regression was used to measure the association between independent variables with dependent variables. Multivariate logistic regression analysis was carried out to nd the role of each signi cant variable in determining the relevant sub-scale scores. Statistical signi cance was accepted at <0.05.

Sociodemographic Characteristics
From a total of 265 medical students who received the questionnaire, 260 completed the survey, yielding an overall response rate of 98.1%. The age of the respondents ranged between 18 and 27 with the mean (SD) of 22.03 (+2.62) years. In the present study, most respondents were male (63.1%), single in marital status (75.4%), living in campus (91.5%) and had monthly income of < 700 ETB (52.7%)( Table 2). Regarding their academic year, religion and ethnicity, 54 (20.8%) were from sixth year, 130 (50%) were Orthodox believers and 139 (53.9%) were Oromo respectively.

Prevalence of Depression, Anxiety and Stress
The present study demonstrated strong internal consistency between depression and anxiety (r = 0.494, p<0.001), depression and stress (r = 0.456, p<0.001), stress and anxiety (r = 0.420, p<0.001). Furthermore, proportion of respondents detected to have depression, anxiety and stress symptoms were 52.3%, 60.8% and 40.4% respectively. Finally, percentage of respondents who had extremely severe symptoms of depression, anxiety and stress were 6.2%, 16.2% and 2.3% respectively ( Table 3).

Prevalence of Substance Use
In the present study, the overall prevalence of khat chewing, cigarette smoking and alcohol drinking was found to be 21.5%, 15.4% and 33.8% respectively. Moreover, 32.1% of respondents chewed khat for 2 -4 years, 35.0% smoked cigarette for 1 -12 months and 55.7% drank alcohol for > 4 years. Lastly, 44.6% of khat chewers and 75% of alcohol drinkers were occasional users while 32.5% of cigarette smoker smoked cigarette once a week ( Table 4).

Association of Socio-Demographic Characteristics and Use of Substance with Depression Level
In a binary logistic regression analysis of the present study, depression had no statistical association with age, educational year, religion and ethnicity (Table 5). Conversely, it had signi cant association with respondents gender; in which males were 0.59 times less likely depressed than females (95% CI: 0.36 -0.99). It was also associated with monthly income, wherein respondents who had monthly income of < 700 ETB were 1.68 times more likely depressed than those who had >700 ETB (95% CI: 1.03 -2.74). Depression was also associated to marital status, in which respondents who had not in relationship (single) were 0.53 times less likely depressed than those in relationship (married)(95%: 0.30 -0.95). It was also associated to residency, wherein non-dormitory living respondents were 10.52 more likely depressed than those living in dormitory (95% CI: 2.41 -46.00).
Concerning to substance use/ behavioral factors/, depressions had statistical association with khat chewing: respondents who chewed khat were 2.07 times more likely depressed than non-chewers (95% CI: 1.11 -3.83). It was also statistically associated with cigarette smoking: respondents who smoke cigarette were 2.42 times more likely depressed than non-smoker (95% CI: 1.17 -5.00). Lastly, it was also associated to alcohol drinking: participants who drink alcohol were 2.01 times more likely depressed than non-drinkers (95% CI:

Association of Socio-Demographic Characteristics and Use of Substance with Anxiety Level
In a binary logistic regression analysis of the present study, anxiety had no signi cant statistical association with monthly income, religion, and ethnicity (Table 6). Contrariwise, anxiety was statistically associated to age of respondents; those ranged between 20 -24 years were 1.92 times less likely anxious than those > 24 years (95% Cl: 0.34 -0.99). It had also statistical association with gender of respondents; males were 0.58 less likely anxious than females (95% Cl: 0.34 -0.99). Similarly, it was associated to marital status; respondents who had not in relationship (single) were 0.52 times less likely anxious than those in relationship (married)(95%: 0.28 -0.96). Anxiety was also associated to educational level of respondents, third year respondents were 4.85 more likely more anxious than sixth year students (95% CI: 1. 93 -12.19). It had also statistical association with residency, non-dormitory living respondents were 15.48 more likely anxious than those living in dormitory (95% CI: 2.05 -117.00).
Concerning to substance use/ behavioral factors/, anxiety had statistical association with khat chewing; respondent who chewed khat were 2.03 times more likely anxious than non-chewers (95% CI: 1.06 -3.91). It was also statistically associated with cigarette smoking; respondents that smoke cigarette were 2.52 times more likely anxious than non-smoker (95% CI: 1.15 -5.55). Finally, anxiety was also associated to alcohol drinking; participants who drink alcohol were 1.75 times more anxious than non-drinkers (95% CI: 1.01 -3.01).

Association of Socio-Demographic Characteristicsand Use of Substance withStress Level
In the present study, when impact of socio-demographic characteristics and use of substance on stress level was evaluated with binary logistic regression analysis, stress had no association with religion and ethnicity (Table 7). However, it had signi cant statistical association with age of respondents; those ranged < 20 years were 2.07 more likely stressed than those > 24 years (95% CI: 1.01 -4.27). It was also associated to gender of respondents with males were 0.57 time less likely stressed than females (95% CI: 0.34 -0.95). Furthermore, it was also associated to monthly income, participants who had monthly income of < 700 ETB were 1.87 times more likely stressed than those who had > 700 ETB (95% CI: 1.13 -3.10). Similarly, it was also associated to marital status, respondents who had not in relationship (single) were 0.55 times less likely stressed than those in relationship (married)(95%: 0.31 -0.96). Likewise, it was also associated to educational level of respondents, in which second year respondents were 4.92 times more likely stressed than those in sixth year (95% CI: 2.09 -11.61). Again, it was also statistically associated to residency, in which nondormitory living respondents were 3.52 more likely stressed than those in dormitory (95% CI: 1. 38 -8.97).
Concerning to substance use/ behavioral factors/, stress had statistical association with khat chewing, respondents who chewed khat were 1.99 times more likely stressed than non-chewers (95% CI: 1.09 -3.61). Furthermore, it was statistically associated with cigarette smoking; respondent who smoke cigarette were 2.02 times more likely stressed than non-smoker (95% CI: 1.02 -3.98). Finally, stress was also associated to alcohol drinking; participants who drink alcohol were 2.10 more likely stressed than non-drinkers (95% CI: 1.24 -3.54).

Predictor Risk Factors For Depression
In the present study, gender, monthly income, marital status, residency, khat chewing, cigarette smoking and alcohol drinking were candidates for multivariate logistic regressions (P-value < 0.05). However, monthly income, residency, and alcohol drinking were identi ed as predictors of the depression (Table 8). For instance, the odd of being depressed was more likely increases with 2.13 times among respondents who had monthly income of < 700 ETB than > 700 ETB (95% CI: 1.24 -3.66). Similarly, the odd of being depressed was also more likely increases with 13.10 times among respondents living non-dormitory than living in dormitory (95% CI: 2.82 -60.70). Finally, the odd of being depressed were markedly increases with 1.68 times among alcohol users than non-users (95% CI: 1.00 -3.08).

Predictor Risk Factors For Anxiety
In the present study, age, sex, marital status, educational year, residency, khat chewing, cigarette smoking and alcohol drinking were candidates for multivariate logistic regressions (P-value < 0.05). However, gender, marital status, educational year, residency and cigarette smoking were found to be predictors of the anxiety (Table 9). For instance, the odd of being anxious was markedly decreases with 0.51 times among male than female participants (95% CI: 0.27 -0.94). The odd of being anxious was also markedly decreases with 0.46 times among participants who were single than those who were married (95% CI: 0.23 -0.94). After rst year, the odd of being anxious were markedly decreased as educational level increased. For instance, the odd of being anxious was more likely increases with 20.43 times in second year students than sixth years (95% CI: 4.40 -94.89). Furthermore, the odd of being anxious was more likely increases with 58.72 times among participants living in non-dormitory than in dormitory (95% CI: 6.33 -544.87). Finally, the odd of being anxious were markedly increases with 2.60 times among cigarette smokers than non-smokers (95% CI: 1.01 -8.41).

Predictor Risk Factors For Stress
In the present study, age, sex, monthly income, marital status, educational year, residency, khat chewing, drinking alcohols and cigarette smoking were candidate for multivariate logistic regressions (P-value < 0.05). However, monthly income, educational year, residency, khat chewing and drinking alcohol were found to be predictors of the stress (Table 10). The odd of being stressed was markedly increases with 2.21 times among participants who had monthly income of < 700 ETB than those who had above 700 ETB (95% CI: 1.08 -4.51). Similarly, the odd of being stressed were markedly increases with 3.05 times in second year participants than sixth years (95% CI: 1.05 -12.47). The odd of being stressed was also more likely increases with 4.82 times among respondent living in non-dormitory than in dormitory (95% CI: 1.61 -14.46).
Also, the odd of being stressed was more likely increases with 1.90 times among khat chewers than non-chewers (95% CI: 1.02 -5.01).
Finally, the odd of being stressed was more likely increases with 1.84 times among alcohol users than non-users (95% CI: 1.01 -3.42).

Discussion
Severe mental health problems interfere with individuals' emotional, cognitive, as well as social abilities that can lead to underemployment and reduced productivity (74,75). Depression, anxiety, and stress are now the major mental health problems that cause disability globally, and no one is immune to these problems (76,77). Therefore, the main aim of this study was to determine the prevalence and severity of depression, anxiety and stress among medical undergraduate students of Arsi University and their association with substance use.
A medical school is a tertiary educational institution that involve a di cult examination system and year-long courses over a long duration (8,25,52,(78)(79)(80)(81). Actually, medical academic atmosphere promotes competition among learners rather than co-operation (82).
Furthermore, medical students are overloaded with a tremendous amount of information with limited time for internalization, new study environment with obligations to succeed especially during preclinical encounters (25,83,84). This greater degree of workload creates feeling of distress and disappointments that predisposes students to have di culties in solving problems, impaired judgments; absenteeism from class lesson and break their mental stability (38,41,(85)(86)(87).
In the present study, the overall prevalence of depression, anxiety and stress symptoms were 52.3%, 60.8% and 40.4% respectively.
Actually, our present nding is almost similar to the prevalence reported by Basudan et al (88), Kulsoom and Afsar (89), Inam et al (45) and Iqbal et al (17). Conversely, it is higher than study report of Shamsuddin et al (90) (54). This difference could be due to difference in cultural perception of stressful factors, economic burden, very high tuition fees, lack of family support, and higher or lower 'readiness' to report different complaint. In our present setup, academic counselling is not a common practice which may be also a contributor.
In the present study, proportion of respondents who had extremely severe symptoms of depression, anxiety and stress were 6.2%, 16.2% and 2.3% respectively. This prevalence was similar with nding of Patil et al (94). However, inconsistent with study result of Iqbal et al (17), Gan et al (95) and Al-Ani Radeef and Ghazi (96). Discrepancies stemming from methodology and type of questionnaire used, could account for this high prevalence obtained by the aforementioned authors. The other possible reasons for the variability could be due to certain differences in the curricula, teaching facilities, quali cation and experience of the instructors, and levels of care given to the students.
In this cross-sectional study we correlated socio-demographic risk factors and substance use (as independent variables) with the prevalence of depression, anxiety and stress level amongst the undergraduate medical students of Arsi University. In the present study, those students living in non-dormitory have considerably higher degree of depression, anxiety and stress as compared to students living within dorm. It can be inferred from the above data that living in dormitory came out to be a protective. This nding is in line with study reported by Rab et al (40) and Shendarkar (97). However, it is inconsistent with study report of Kunwar et al (98) and Liaqat et al (99).
This unequivocal distribution tells us that probably depression, anxiety and stress are associated with a multitude of factors such as poor dormitory conditions, more economic stress, distance from the family, less structured environment, and problems dealing with roommates, which might be different or parallel in the two study groups under consideration.
In the present study, medical students who had lower monthly income were more likely depressed and stressed as compared to students who had higher monthly income. Despite the fact that food and dormitory services are provided to the students by the university, students need money for excursions, to print hand outs, to buy dressings, and other basic necessities (38). This indicates that nancial constraints could be an additional source of depression, anxiety and stress besides academic stressors (100)(101)(102)(103)(104). However, other studies have noted no difference (105,106). Again, geographical, racial and sample size differences may possibly account for these different results.
In the present study, there was a signi cant association between students' academic year and level of anxiety and stress. There are many associated studies that implies stress and stress-related illness was highest in 2 nd year medical students as compared to other academic year (38,45,84,107). The possible reason could be the amount and complexity of the material to be learned in the second year with progressive assessments of anatomy, physiology, and biochemistry that they have to pass to join the next higher level. Additional supportive evidence is senior students may developed skills of how to manage stress and stress-related illness than students in the early years (108).
In the present study, higher scores of anxiety and stress were associated with female and married students. Genetic, psychosocial, and hormonal factors have been adduced as possible reason for this differences (109,110). This is in line with previous literature which suggests that female medical students have higher rates of symptoms of depression, anxiety, and stress (111)(112)(113). The ndings can also be contextualized with literature among the general population, which suggest that women are more likely to suffer from depression and anxiety than men (114). However, our nding is inconsistent with result nding of Koochaki et al (105) that noted no gender difference in level of stress among Iranian medical students. The possible explanation for this difference may be setup difference, technological difference, and curricular approach. Also, these studies have used different measuring tool.
In the present study, the prevalence of khat chewing was found to be 21.5%. And, the level of stress was higher among khat chewers than nonchewers students. The possible reason is that khat stimulates adrenocortical function. This nding suggests that khat chewers show diversi ed psychological symptoms including stress, anxiety, depression, and emotional instability; this is due to its content of cathinone and cathine, which causes an increase in the release of cortisol, norepinephrine, and dopamine from presynaptic storage site (sympathomimetic effect) in the body of chewers (115,116). Consequently, the respondents experience psychostimulatory effects such as excitement and talkativeness initially. Then, they develop excessive worry, depressed mood, and tension (115,116). Supporting evidence was reported by Al'Absi et al (117), that the level of cortisol in saliva samples was higher among khat chewers than nonchewers. The other possible justi cation is the socioeconomic problems caused due to increased demand of money to buy khat.
In the present study, the prevalence of cigarette smoking among medical students was 15.4%. And, the level of anxiety was higher among cigarette smokers than non-smokers students. As reported by Parrott (118), smokers often mistakenly report that cigarettes help relieves feelings of stress, but the fact is it does not alleviate stress. However, the stress levels of adult smokers are slightly higher than those of nonsmokers (38). The level of stress was increased as they develop regular patterns of smoking and cessation of smoking reduces stress. Furthermore, chronic nicotine intake affects the brain reward system function which results in depression and anxiety symptoms (62).
In the present study, the prevalence of alcohol consumption among medical students was 33.8%, which is consistent with studies reported by Kalayu et al (119). However, it was a bit higher than study reported by Seipone et al (120), because studies have used different measuring tool. In the present study, alcohol users were 1.68 times more likely to have depression and 1.84 times more likely to have stress than non-user students. Alcohol drinking predisposes students to mental illness (38). A related nding was reported by Britton (121) that there is a positive relationship between stress and alcohol use. Brady and Sonne (122) have described that people under chronic stress drinks more alcohol and eat less nutritious foods than nonstressed individuals. Alcohol largely affects the brain and the endocrine system. Alcohol is both a sedative and a depressant that affects the central nervous system. Alcohol changes levels of serotonin and other neurotransmitters in the brain. This can make stress and stress related-illness worse (123,124). Furthermore, alcohol actually increases the stress response by stimulating production of stress hormones. Alcohol directly activates the hypothalamus-pituitary-adrenal axis to produce excess cortisol (38).

Conclusion
Healthy medical students are likely to become healthy doctors. To sum up, overall prevalence rate of depression, anxiety and stress is alarmingly high among Arsi University medical students. Monthly income, residency and alcohol drinking were identi ed as risk factors of both depression and stress. In addition to after mention factors educational year and khat chewing were identi ed as risk factors stress. However, gender, marital status, educational year, residency and cigarette smoking were identi ed as risk factors of anxiety.
Besides stress reduction interventions, implementation of structured orientation program that addresses issues like expectations for each phase, how students are going to be evaluated, how to cope and how to get through each phase smoothly were recommended. In addition to awareness creation about the adverse effect of substance use, establishing student counseling center in the campus with quali ed staff is also highly recommended. Family or close friend problem (recent death or accident), distance from family, frequency of money sent and being rst from home to go far were not assessed.

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

ETHICAL ISSUE
Ethical approval was obtained from Arsi University Ethical Review Committee. Written informed consent was taken from the students during data collection. The con dentiality was kept anonymous.