Study design, period and set-up
This was an institutional based study conducted from June 1st to August 1st, 2017. The study was conducted among pregnant women of Gedeo zone attending ANC service at rural health centers. Gedeo is one of the 13 zones found in Southern Nation Nationalities and Peoples of Region (SNNPR), Southern Ethiopia. Dilla is the zonal town of Gedeo zone located at 359 km South East far from the capital city of Ethiopia (Addis Ababa). The zone has administrational classification of six districts and two administrations. Within Gedeo zone, there are about 38 health centers providing service for about 41,733 pregnant women expected to initiate the ANC service each year.
Sample size and sampling technique
We have used a single population proportion formula to calculate the minimum sample size required for this study. Assumptions made for the sample size calculation were 50% prevalence of Khat use, 5% margin of error and 95% confidence interval (CI). As we have used multi stage sampling technique, a design effect of 2 was also considered. Finally, the total sample size was 845 by adding a 10% non-response rate.
Two-stages sampling technique was used to recruit the required study participants. In the first stage, four districts and one city administration were randomly selected from a total of six districts and two city administrations found in Gedeo zone. In the second stage, a total of eight rural health centers were also randomly selected from the pre-selected four districts and one city administration. Then, the calculated sample size was proportionally allocated to each of the eight randomly selected health centers. Finally, systematic random sampling technique was used to address individual participants from each selected health center using a sampling interval (K). The sampling interval (K) was determined by dividing the total number of eligible women attending ANC service (age greater or equal to 18 years and without serious medical illness) during the data collection period to the proportionally allocated number of samples to each health center. To identify the first participant, we used lottery method between one and K. After addressing the first participant, K value was added to recruit the next candidate until the proposed sample size was addressed.
Data collection instruments and operational definitions
Data were collected using an interviewer administered questionnaire/interview guide. The interview guide was developed for this study after reviewing different related literatures (Additional file
1). It was first prepared in English and translated to the commonly spoken languages in the study area (Amharic and Gedeufa). Back translation to English was also done to check its accuracy. Pretest of the Amharic and Gedeufa version of the questionnaire was conducted among 43 pregnant women attending ANC service at Chuko health center. The purpose of the pretest was to check the understandability of the questionnaire, ability of the questionnaire to address proposed objectives and appropriateness of expressions for the context of the study area. Based on the pretest results, minor modification was done on the contents of the questionnaire, and some culturally sensitive expressions were rewritten by replacing appropriate statements.
The questionnaire was organized in to six sub-sections. The first section of the questionnaire had socio-demographic variables including age in years, marital status, ethnicity, religion, residency and income level. In the second part, obstetric and gynecological related factors (parity, pregnancy status, abortion history and numbers of ANC visit) were directly recorded from the women’s chart (medical record) in a confidential manner.
The third part was the Oslo-3 Social Support Scale which was used to assess level of social support. Oslo Social Support Scale is an internationally validated tool used to measure the level of social support both in clinical settings and community studies. It has three likert scale questions with five possible response options for the two questions and four options for one question [
20]. The total sum score of Oslo-3 Social Support Scale ranges from 0 to 14. Based on the total sum scores of the scale, level of social support was categorized into three levels (poor = “3–8”, moderate = “9–11”, strong = “12–14”) [
21].
In the fourth section, Self Reporting Questionnaire (SRQ-20) was used to measure the level of mental distress [
22]. SRQ-20 is a widely used instrument for epidemiological studies in clinical and community settings in Ethiopia and other African countries. The tool is validated in Ethiopian context among perinatal women with good accuracy showing a sensitivity and specificity of 0.86 and 0.76, respectively [
23]. It has 20 “Yes” or “No” questions to be used both as self administered and interviewer administered. Its items were scored as “0” (no) and “1” (yes), and the sum score ranges 0–20. Accordingly, pregnant women were considered as screened positive for mental distress if the total sum score of SRQ-20 was 7 or above [
24].
The WHO structured questionnaire of Alcohol Use Disorder Identification Test (AUDIT) was the fifth part of the questionnaire used to assess the level of alcohol consumption. AUDIT is a screening tool used to asses alcohol consumption with levels of “no alcohol use”, “social use”, “harmful drinking”, “hazardous drinking” and “alcohol dependency” with ascending order of alcohol use severity [
25]. AUDIT is a cross culturally validated instrument which can be used in different countries with sensitivity and specificity of 0.90 and 0.80, respectively [
26,
27]. In the current study, AUDIT had good accuracy and internal consistency as evidenced by a Cronbach alpha = 0.96.
The final section of the questionnaire had questions used to assess the outcome variable (Khat use during current pregnancy). Khat use was assessed based on questions developed from different literatures [
28‐
30]. These questions were supposed to address the duration and the numbers of days pregnant women were chewing Khat per week during their current pregnancy. Individuals with current Khat use history were also asked follow-up questions to mention possible reasons that motivate them to initiate Khat use during their current pregnancy, and they identified more than a single reason. Finally, they were asked about the single most important reason why they initiate to chew Khat currently.
Data collection procedures
Five data collectors (clinical nurses) and two supervisors (MSc level public health professionals) were participated in the data collection after attending two days of training regarding the contents of the questionnaire and data collection procedures. Prior to the data collection, written consent was obtained from each participant after providing a brief explanation on the scope and objects of the study. Respondents were also informed as they have the right to refuse or withdraw their participation at any time they want, and no any harm could be imposed towards them due to the withdrawal. Data collectors first evaluated the eligibility of each respondent to participate in the study, and only eligible individuals were invited to be interviewed. Pregnant women whose age was equal or greater than 18 years were included where as women with serious illness and difficulty of communication were excluded from the study. During the data collection, personal identifiers like name and phone numbers of respondents never had been recorded. The collected data was also kept confidential and used only for the purpose of the study.
Data analysis
The completeness and consistency of the collected data was checked first. The data then, entered to Epi-info version 3.5 (software) and exported to a Statistical Package for Social Science (SPSS-version-20) for analysis. We used descriptive statistics to explain study participants in relation to different characteristics and to measure the outcome variable (magnitude of Khat use). These descriptive results were presented using texts, tables and figures. To identify factors associated with Khat use during current pregnancy, bivariable and multi variable analysis were computed. Variables with P-values of less than 0.25 in bivariable regression were considered as candidates for multivariable analysis and entered together to the final model. In the multivariable analysis, variables with P-values of less than 0.05 were considered as statistically significant correlates of Khat use and the strength of the association was measured by adjusted odds ratio (AOR) with corresponding 95% CI.