All methods were carried out in accordance with relevant guidelines and regulations.
Study setting, design and population
A community-based cross-sectional study was conducted from May 01–31, 2019 in Addis Ketema sub-city, one of the ten sub-cities of Addis Ababa. It is a densely populated sub-city consisting of ten districts covering an area of only 8.64 km2 with a total population estimated at 320,000 as of 2017. The total number of OVC due to HIV and other causes were 20,655 in 2017 in the sub-city. OVC is a child who is at high risk of lacking adequate care and protection due to parental death, disease, disaster or acute poverty [3, 11].
All OVC aged 6 to 59 months, who had been living in the study area for a period of at least 6 months were eligible for inclusion and only the youngest child was selected per household. OVC who were severely ill or disabled were excluded due to the difficulty of obtaining accurate measurements. Also, OVC who were not found in three appointed interviews were excluded from the study and replaced by the next eligible OVC. Furthermore, children whose caretakers have a hearing difficulty preventing them from being interviewed were excluded.
Sampling procedures
Sample size for the prevalence of stunting, wasting and underweight was determined by a single population proportion formula by taking prevalence of stunting (35.1%), the most common form of undernutrition which was taken from a study done among OVC in Hawassa town in 2016 [15], Z α/2 = 1.96, 5% margin of error, 10% non-response rate and 1.5 design effect. The calculated sample size was 584. The sample size for the analytical component (factors associated with stunting) was determined by a two population proportion formula by taking 80% power, 95% confidence interval (CI), 5% margin of error, proportion of stunting among literate care givers and proportion of stunting among illiterate care givers from the study conducted in Gondar [16]. The sample size calculated from this formula was 190. Thus, we took the largest possible sample size (n = 584) in our study.
A multi-stage sampling technique was employed to recruit study participants. In the first stage, out of the ten districts, three of them were selected randomly. Then, the total sample size was distributed to each district proportionally to the total households of the district. Social workers living in the community who are collaborating closely with the sub-city identified the households at risk wherein OVC reside in their respective villages of the selected districts. In order to facilitate the data collection, one social worker accompanied each data collector to the OVC households until the sample size was completed. In households with more than one child aged 6 to 59 months, only the youngest OVC was selected. Many of the households received financial support as part of the safety net program. However, none of the children received any form of nutritional support or supplement.
Data collection procedure
Data were collected by using a structured standardized questionnaire (supplementary material 1 attached) and anthropometric measurements performed by the data collecting team with identical scales, measuring boards and MUAC tapes. The questionnaire was adapted from various sources including UNICEF [15, 17]. It has several contents including socio-demographic characteristics, housing and sanitation, feeding practices and dietary diversity, morbidity variables and household food insecurity. It was initially prepared in English and translated into Amharic language for data collection. Back translation of the questionnaire into English was carried out by an independent translator to check for the consistency of the translation. Moreover, the contents of the questionnaire was checked for cultural appropriateness and its content validity by senior experts (Nutrition and Public Health experts). Pre-test was also conducted in a district not included in our study to check for any inconsistencies and modified accordingly.
Data collection was facilitated by nurses and facilitators who were social workers familiar with the OVC in selected households. Fieldworkers were given training by the principal investigator on the objectives and methodology of the study, the contents of the questionnaire, the confidentiality of responses, the use of instruments and standard procedure of anthropometric measurement.
Height was measured in standing position for children ≥2 years and length was measured in recumbent position in children < 2 years. The child was barefooted and free of head wear. For measuring height the child was helped onto the baseboard with feet slightly apart. The back of the head, shoulder blades, buttocks, calves and heels were touching the vertical board. The assistant held the child’s knees and ankles. With the child’s chin held between thumb and forefinger and eyes facing directly forward, the interviewer pulled the headboard down to rest firmly on top of the child’s and read to the nearest completed 0.1 cm [18, 19]. For measuring length the child was placed on its back. The assistant standing opposite the tape held the child’s head against the headboard. The child’s eyes were looking straight up. The interviewer standing on the side of the measuring tape held down the child’s knees with the left hand and moved the footboard with the right hand flat against the soles. The measurement was read and recorded to the nearest completed 0.1 cm [19].
Weight was measured with the child lightly dressed on a standard scale and recorded to the nearest 0.1 kg. For children < 2 years of age, the caretaker was first weighed alone and again holding the undressed child. The difference between the two readings equalled the weight of the child. The scale was calibrated immediately before each session [19].
MUAC is the circumference of the undressed left upper-arm measured at the mid-point between the shoulder tip and elbow in children with a height > 65 cm. The interviewers bent the arm of the child at the elbow and identified and marked the olecranon and acromion processes as well as the midpoint between the two landmarks with a pen. Then, the arm was straightened and hung down the side of the body. The tape was placed around the arm at the marked mid-point at correct tape tension and the circumference was read to the nearest 0.1 cm and repeated twice to ensure accuracy. Colour coding indicates nutritional status [19].
Ethical considerations
Ethical clearance was obtained from the Institutional Review Board (IRB) of Myungsung Medical College. The participants were informed about the objective of the study and written informed consent was gained from the caretakers. Illiterate caretakers were asked to sign the consent form after it was read to them by the interviewer. Moreover, the participants were at no risk of serious harm and had the right to decline participation or withdraw at any time during the interview. Caretakers of acutely malnourished children were urged to seek health care in a nearby facility. The information collected in the study will be treated confidentially and anonymity guaranteed by the principal investigator.
Data management and analysis
Age was documented in completed months. If the caretaker was unsure of the child’s day of birth, the 15th day of the month was used and if the month of birth was unknown the midpoint of the year was used [15]. Food security was assessed using the Household Food Insecurity Access Scale (HFIAS) specifically adapted by the USAID Food and Nutrition Technical Assistance (FANTA) project for use in developing countries as a measure of the degree of food insecurity in the household in the past four weeks. Households were considered food-secure if they scored less than 17 and food-insecure if they scored ≥17 points [17]. Dietary Diversity Scores were calculated by adding the number of food groups consumed in the household over the 24-h recall period and graded as low (≤3) and high (≥ 4) based on the WHO designation of minimum dietary diversity if four or more food groups consumed in the last 24 h [15]. The prevalence of undernutrition was assessed by calculating the percentages of children who are stunted, wasted or underweight using ENA SMART based on the WHO − 2 Z-score cut-off and summarized by percentage and the respective 95% confidence interval. Bivariate and multi-variable binary logistic regression analyses were carried out to identify factors associated with stunting. Those variables with ρ-value < 0.25 in the bivariate analysis were considered for further multi-variable analysis and ρ-values of less than 0.05 were taken as a cut-off point for determining the significant association of independent variables with stunting. Odds ratio (OR) with 95% confidence interval was calculated to determine the strength of associations. Multicollinearity was checked by the multicollinearity diagnostics (Variance Inflation Factor (VIF) and the tolerance test). Goodness of model fitness was assessed by Hosmer Lemeshew goodness of fit test.