Study setting and sample
The study took place in a Demographic Surveillance Site (DSS) in Gilgel Gibe Dam area in Southwestern Ethiopia. The Gilgel Gibe DSS is located at about 260 km southwest of Addis Ababa (the capital), and is used to collect vital events data through an update of multiple times in a year. The DSS area has a population of over 55,000, a crude birth rate of 35 per 1000, and a population growth rate of about 2.7% per annum by 2012b. Data collection at the site is done by Jimma University.
Data for the present study comes from a baseline survey conducted as part of a population-based cohort study in which pregnant women were identified and followed to examine factors that influence birth outcome. The outcome variable for the present analysis is maternal depression during pregnancy. All pregnant women in their 2nd and 3rd trimester living in the eleven kebelesc (Villages) in Gilgel Gibe DSS area were targeted for participation. Six hundred twenty seven pregnant women were identified from the DSS registration and from the records of Health Extension Workers who work in each village. A baseline survey was conducted from June to July 2012 on 627 pregnant women.
A structured questionnaire was developed and administered to all study participants (study tool attached as Additional file
1). The questionnaire was first developed in English and then translated and back translated to Oromo – a local language spoken in the study area. Ten trained female interviewers with a minimum of diploma-level education collected the data. They had five days of training on how to administer the questionnaire, practice interviewing and role-plays, and how to deal with ethical issues. After the training, interviewers undertook a pilot study, and information from the pilot study was used to finalize the questionnaire. Data on depressive symptoms, pregnancy intention, social support and other explanatory variables were gathered using an interviewer-administered structured questionnaire. All study participants were interviewed at their home in private area. Ethical approval was obtained from the College of Health Sciences, Addis Ababa University. Moreover, support letters were obtained from regional, zonal and district health offices and kebele (village) administrations were informed about the study. Participants were asked for informed consent, and participation in the study was fully voluntary.
Measurements
Depressive symptoms were measured using the Edinburgh Postnatal Depression Scale (EPDS), which describes depression as cognitive and affective features that last for at least one week, including the inability to laugh, the inability to look forward to things with enjoyment, blaming oneself unnecessarily, anxiety or worry, being scared or panicky, the inability to cope, difficulty sleeping, feeling sad or miserable, crying, and thoughts of harming oneself [
42]. The EPDS is widely used and has been validated for use during pregnancy in different countries and settings [
43‐
47], including urban and rural Ethiopia [
45,
48]. A psychiatrist checked the translation and back translation of the depression questions. Moreover, considering the difficulty in understanding items 1 to 3 of the depression scale (EPDS), we used examples suggested and applied by Hanlon and colleagues in their validation study in Butajira (Central Ethiopia) [
45]. The items were scored on a scale of 0–3, allowing a total score ranging from 0 to 30. The internal consistency of the EPDS was tested using Cronbach alpha and was found to be 0.85. Like other previous studies that used EPDS cut of point of 13 and above [
14,
17], we used a cut of point of 13 and above on the scale to identify women with depressive symptoms. No other measures of mental health problems, such as anxiety and stress, were collected in this study.
The key independent variable was pregnancy intention. Women were asked to recall their feelings at the time they became pregnant: “At the time you became pregnant, did you want to become pregnant then, did you want to wait until later, or did you not want to have any (more) children at all?” The responses were categorized as (1) wanted then “wanted” (2) wanted to happen later “mistimed” and (3) did not want at all “unwanted”.
Social support was measured using the Maternity Social Support Scale (MSSS) developed by Webster and colleagues [
49]. The scale contains six items and includes questions on family support, friendship network, help from spouse, conflict with spouse, feeling controlled by spouse, and feeling unloved by spouse. Each item was measured on a five-point Likert scale and a total score of 30 was possible. We classified social support in to three categories; high social support (for scores 24–30), medium social support (18–23) and low social support (below 18) categories. The internal consistency of the scale was tested using Cronbach’s alpha and was found to be 0.74.
We also considered several other explanatory variables based on previous studies including age (coded as 15–24, 25–34 and 35–49 years), women’s education (none, primary and secondary and above), occupation (housewife, farmer and employed in other services), wealth index, parity, history of miscarriage or stillbirth, perceived work burden during pregnancy, intimate partner violence, and household food security. The wealth index was computed from ownership of the following household assets: radio, television, electricity, toilet, farm land, and animals such as cattle, sheep, and goats. Principal Component Analysis (PCA) was conducted and the resulting index was divided into three categories representing poor, middle and wealthy.
Women were asked whether they have experienced a miscarriage or stillbirth in their lifetime to measure any experience of pregnancy loss. All women participating in the study were asked about the type of physical work they did in the seven days before the survey, and whether they perceived that the work was ‘difficult’, ‘moderate’ or ‘easy’ for them. Intimate partner violence was measured by asking women whether they have ever been beaten during the current pregnancy by their husbands or partners.
Household food insecurity was measured with a six- item scale based on previously validated measures in developing countries. Women were asked whether because food ran out or money was not enough to buy food, in the last 3 months, they: (1) worried about running out of food, (2) ran out of food, (3) reduced the variety of food for their children, (4) did not have enough food to give their children to eat, (5) spent the whole day without food, and (6) or anyone else in the household had ever had to ask others for food or money to buy food. For each item, ‘yes’ was coded with “1” and no coded as “0” and a summative index of food insecurity was created. Households were classified as ‘food-insecure’ if the respondent answered affirmatively to two or more of the six household food security questions. The scale had an internal consistency (Cronback’s alpha) of 0.85.