This is part of a larger study of out-of-school Ethiopian girls and young women aged 15-24, focusing on migration and the transition to different professional roles after migration. . This was a large mixed-methods study that included formative qualitative research and a large-scale quantitative survey. The quantitative study included questions to assess the mental health of those selected for the interview. The majority of survey respondents were rural-urban migrant women, but we also interviewed a small number of rural parents, rural girls and young women, and brokers who help migrants find work. As the issue of mental health was not included in the initial qualitative research, we rely exclusively on the results of the quantitative study for the present study. Similarly, interviews with parents and brokers did not collect information about mental health.
The quantitative study covered six regions of Ethiopia and took place in seven cities: Adama, Addis Ababa, Dessie, Dire Dawe, Harar, Mekelle and Shashemene. These towns were chosen because they were considered to receive a large number of migrants from rural areas. Several categories of migrant girls and young women were interviewed. To be eligible for the study, respondents had to be between the ages of 18 and 24, out of school and had left school at age 16, and had migrated to the city before age 18. However, for domestic workers, we have included an age range of 15-24 to capture the experience of child domestic workers; at the same time, all domestic workers in our sample were over 16 years old. Most respondents were sampled via household lists, followed by random selection of eligible household members. Two additional categories of respondents were purposely sampled: sex workers and bar/café workers.
In each of the cities studied, we identified neighborhoods where young migrant women were known to live, mostly low-income areas. These areas were identified based on consultations with local stakeholders and people knowledgeable about the areas, primarily government officials. We conducted a systematic census of all households and other structures in the area, such as restaurant back rooms or establishments where young women may spend the night. The interviewers went from house to house to list the members of the household or the structure, in order to identify the girls and young women eligible for the study. Eligible women were randomly sampled, using the random number generator available in SPSS v.25.
In order to assess respondents by occupation, we asked (1) What kind of paid work did you do in the past three months? and (2) Of these, what is your main source of cash income and/or payment in kind? The second question was included in cases where respondents were engaged in more than one type of paid work. Sex workers and bar/café workers are more hidden and stigmatized populations. Bar/café workers are often seen as partially engaging in the sex trade, which leads to stigma. These categories of respondents were purposively sampled, visiting places where the targeted respondents were known to congregate (bars, cafes, nightclubs, local breweries and red-light districts) and approaching potential respondents working in the bar/café or working as a professional (the)s of sex for interview. Once contact was established, the interviewers selected respondents based on their eligibility, such as being a migrant to the area and being out of school.
Interviewers were recruited from the study towns to ensure they had the necessary language skills and understood the local culture and community. Our female sample was interviewed only by female interviewers. The interviewer training lasted seven days. Investigators reviewed the questionnaire item by item, reviewed ethical procedures, including informed consent and what to do in the event of adverse events, and engaged in hands-on interviews in pairs, ensuring understanding and adherence to jumping patterns and general administration of the questionnaire. A professional counseling firm was also made available to respondents in case they had negative feedback from the interview or expressed a need for such services.
A structured questionnaire was developed, pre-tested and translated into the local Amharic, Oromiffa and Tigrigna languages. The pretest was conducted in selected study cities and locations outside the study areas. Questionnaires collected information on demographic and socioeconomic characteristics, education, families and social networks, migration, livelihoods, use of job brokers to find work, mental health, marriage , sexual experience, HIV knowledge and behavior, family planning and pregnancy, and use of services. All data was transported from the field and entered in Addis Ababa by trained data entry clerks. Data was converted to SPSS v.25 for analysis. The study received ethics approval from the Institutional Review Board (IRB) of the Population Council and the National Research Ethics Board in Ethiopia.
In order to measure depression symptoms among respondents, we used a modified Patient Health Questionnaire (PHQ-9), asking about symptoms experienced in the two weeks prior to the survey. PHQ-9 was developed to assess depression in primary care settings . Several institutional studies have validated the PHQ-9 in Ethiopia [13, 14]. In this study, respondents were asked the following: “I am going to read you a list of feelings or experiences one may have. I would like you to tell me if you have experienced this in the past two weeks: 1) Little interest or pleasure in doing daily activities, 2) Feeling down, depressed or hopeless, 3) Feeling tired or having low energy, etc. If the respondent answered “yes” to any item, they were asked a follow-up question about the frequency of the experience: occasionally, over several days or daily. Three statements used in the standard PHQ-9 battery of questions have been separated into two items. Indeed, when pre-testing the questionnaire, respondents found some of the statements confusing, too complex or difficult to understand. Compared to previous studies that validated the PHQ-9 questions, this observed confusion may be due to the young age or low level of education of our study population. The questions separated into two items were (denoted separation a. and b.): 1) a. Difficulty falling asleep/staying asleep or b. sleep too much, 2) a. Lack of appetite or b. overeating, 3) a. Moving or talking so slowly that other people might have noticed or b. being so restless or fidgety and moving around more than usual that other people notice. During analysis, separate items were back-coded so that responses matched the original PHQ-9 and total scores ranged from 0 to 27. Each item was scored from 0 to 3 to reflect the existence and frequency of symptoms. A global score of 0 to 4 on the PHQ-9 was considered no depression, those with a score of 5 to 9 were coded as having moderate depression; those with a score of 10 or more were considered to have severe depression .
We present the percentage of young women in each category (domestic workers, sex workers, waitresses/bar workers, other professional and inactive categories) suffering from moderate or severe depression. We also examine the association between depression and various demographic characteristics as well as the experience of social isolation, migration patterns and violence. Logistic regression was used to model the odds of suffering from moderate or severe depression. We assessed the multicollinearity of the independent variables using ViF, with any value greater than 10 indicating the existence of multicollinearity. We also assessed any potential effect of outliers using Cooks distance. Variables with significant associations in the bivariate analysis were found to be uncorrelated, except for age and marital status which had a weak correlation.
In addition to basic demographic variables such as age, religion, and marital status, we explored the association between depression and aspects of girls’ migration and social life. We included measures of early age at migration (under 15) and of having migrated alone, without accompanying family members or other acquaintances. In addition, many migrant girls and young women use brokers during their migration to help them find jobs. The Council’s research has also shown that some brokers can take advantage of migrant girls, which can lead to sexual abuse and put girls at increased risk of trafficking. . We included a measure of social connections and social isolation through the variable indicating whether or not the respondent reports having friends. Finally, we also included in the model measures of violence such as being beaten in the past three months and having experienced forced or forced sexual initiation. Forced sexual initiation was calculated as the percentage of girls who had their first sexual intercourse by any form of force or coercion, including being physically forced to have sex; their partner uses violence to have sex; being locked in a room to have sex against their will; or their partner not taking “no” for an answer. Finally, because sex workers had much higher levels of depression than other occupational categories, we included a covariate indicating whether the respondent was a sex worker.