Study design and setting
A community-based comparative cross-sectional study was conducted in Gemachis district from 10 July 2020 to 20 August 2020. Gemachis district is one of the 17 districts of West Hararghe that is located in the Oromia Regional State of Ethiopia. The district had a total population of 241685, as of 2020, of which 123259 were women and 10344 were children aged 6-23 months (8). Approximately, the overall women unemployment rate in Ethiopia according to 2016EDHS was 48.8% (11).
The study participants were mothers (employed and unemployed) who had children aged 6-23 months and who had resided in the districts at least for six months at the time of data collection. Mothers who were sick and/or not volunteered to participate were excluded from the study.
Sample size and sampling methods
The sample size was determined using a single population proportion formula by using 95%CI, 5% margin of error, and a 10% non-response rate. Based on the evidence from the previous study, the proportion of complementary feeding practice was 56.5% (14). The sample size for the first objective with single population proportion formula and a design effect of 1.5 was 624. The sample size for the second objective was calculated using Epi-info software version 7(stat Cal) with the following assumptions; 80% power of the study, 1:1 unexposed to exposed ratio, and 95% confidence interval. Based on these assumptions, the sample size for the second objective was found to be 674. Comparing the two samples, therefore, the larger sample size, 674, was considered as the final sample for this study.
The study participants were selected by using a multi-stage sampling technique. The primary sampling units were kebeles (small administrative units) found in the Gemachis district. First, from the 38 rural kebeles of the district, 14 Kebeles that account for 34% of the total kebeles were selected randomly. Then pre-survey case identification to identify households with employed and unemployed mothers of children aged 6-23months was conducted and followed by stratification of mothers by employment status (employed and unemployed). Then from each stratum, 226 employed and 448 unemployed mothers having children aged 6-23month were selected randomly and grouped based on their homogeneity. Finally, the total sample was proportionally allocated to each fourteen selected kebeles according to their number of eligible employed and unemployed mothers having children aged 6-23 months living in the kebeles. Following the proportional allocation of the sample to the selected, household codes were given separately for employed and unemployed mothers. Using household code as a sampling frame, the study participants were finally selected using a systematic random sampling method.
Data Collection Tools and Procedures
A standard and validated tool that was adapted from the Ethiopian Demographic and Health Survey (EDHS) and World Health Organization (WHO) was used to collect the data. The tool was prepared first in English then translated into the local language (Afan Oromo). Then it was re-translated to English by language experts to keep its consistency. The questionnaire was including questions that is used to gather information on socio-demographic and economic variables, maternal and health service-related information. Household wealth index computed using a composite indicator for rural residents by considering properties like livestock ownership, selected household assets, size of agricultural land, and quantity of crop production. Principal component analysis was performed to categorize the household wealth index into lowest, middle, and highest.
The questioners were grouped and arranged so that objectives and all-important variables were addressed. Data were collected for fifteen days using interviewer-administered face-to-face interviews. There were sixty-seven questions to be answered and the interview was taken about 25-30min for one respondent to answer all questions. Two BSc nurses and four health officers were recruited for data collection. One supervisor was assigned to seven kebeles in order to supervise and collect questionnaires on daily basis and check for inconsistencies and incompetence. A three-day training was provided to data collectors and supervisors by the principal investigator before actual data collection was started.
The outcome variable for this study was complementary feeding practice which was classified as appropriate or inappropriate. Appropriate complementary feeding practice defined as IYC feeding practice that fulfills the minimum dietary diversity, the minimum meal frequency, continuing breastfeeding with complementary feeding for 2 years, and timely initiation of complementary feeding at the recommended time of World Health Organization (11, 19). While the infant is cauterized having inappropriate complementary feeding practice when IYC feeding practice does not fulfill even one of the components of appropriate complementary feeding practice (19).
The predictor variables used in this study was maternal age, child age child sex, religion, marital status, family size, number of children ever born, maternal education, maternal employment, paternal education, paternal employment, wealth status, ANC visit, place of delivery, postnatal attendance, institutional delivery, initiation of breast and complementary feeding, decision-making role, the distance of health centers and distance of workplace, any illness in the last 2-week, and inadequate care child breastfeeding status. In addition, minimum meal frequency were defined as feeding of infants and young children (IYC) that fulfills at least 2-3 times complementary feeding within 24 hours for children aged to 6-8 months and 3-4 times for those aged 9 months and above (8). Finally infants was grouped as having minimum dietary diversity when the baby fed from at least four food groups within the past 24hours out of the following seven food groups: grain, legumes, dairy products, egg, meat, fruits, and vegetables (8).
Data Quality Control
Before the actual data collection procedure, a questionnaire was pre-tested on 5% of the sample on a similar population of the district, from the kebeles that were not part of the actual sample. Any ambiguity and unclear questions were modified before the data collection. After data collection, each questionnaire was given a unique code by the principal investigator. Finally, data clearance and double data entry checks were conducted by two individuals to minimize errors.
Data Processing and Analysis
Before analysis, the data were checked manually for completeness and consistency. The data were, then, entered into Epi data version 3.1 and exported to STATA version 14.2 for further cleaning and analysis. Frequency distribution, measures of central tendency, and dispersion used to describe the data.
Chi-square and Student t-tests were used as appropriate to check whether the differences between the two groups (employed and unemployed mothers) on the selected characteristics were statistically significant. A binary logistic regression model was fitted to investigate the association between each factor and outcome variable. All variables with a P-value of ≤ 0.25 in bivariate analysis were checked for multicollinearity, using variance inflation factor (VIF), and included in multivariable logistic regression models and adjusted for a confounding effect. Accordingly, three multivariable logistic regression models were constructed to identify factors associated with the appropriate complementary feeding practice among employed mothers, unemployed, and all mothers. Finally, an adjusted odds ratio with a 95% confidence interval or p-value of less than 0.05 was used to report factors independently associated with appropriate complementary feeding practice.
Ethical clearance was obtained from Haramaya University, College of Health and Medical Sciences, Institutional Health Research Ethics Review Committee (IHRERC). A support letter was obtained from the School of Graduate Studies and submitted to Gemachis district Health Office and kebele administrations so that permission to access the study site was granted from each hierarchy. Study participants were adequately informed about the purpose, method, and anticipated benefit of the study by the assigned data collectors. Written/verbal informed consent was obtained from the study participants. The participants were informed as they have the right to refuse or withdraw from participating in the research without any explanation and they had the right to ask any question at any time during the data collection period. Further, the confidentiality of the study participants was strictly maintained by excluding any personal identifiers from the questionnaire and limiting data access only to individuals who were authorized by the IHRERC